Observation and quantification of protein production in single living cells
171
Observation and quantification of protein production in single living cells by Ibrahim Kays Integrated Program in Neuroscience McGill University, Montréal November, 2016 A thesis submitted to McGill University in partial fulfilment of the requirements of the degree of Doctor of Philosophy © Ibrahim Kays, 2016
Transcript of Observation and quantification of protein production in single living cells
Observation and quantification of protein production
in single living cells
by
Ibrahim Kays
Integrated Program in Neuroscience
McGill University, Montréal
November, 2016
A thesis submitted to McGill University in partial fulfilment of the requirements of
the degree of Doctor of Philosophy
© Ibrahim Kays, 2016
ii
Abstract
Accurate quantification of protein production is fundamental to understanding basic
molecular and cellular processes. Dysregulation of protein levels can harm cells and lead to
diseases such as cancer and neurodegenerative diseases. However, little is known about how
protein production in a cell changes over time and in response to external factors. The current
assays used to quantify protein production are invasive, time consuming, and have poor
resolution. As a result, researchers have turned to mRNA expression as a measure for protein
abundance, although this has been demonstrated to be inaccurate.
To address these issues, my thesis explores new tools and techniques I developed to
monitor protein production in single living cells. By simultaneously examining the levels of
mRNA and protein of a gene from a single cell, I describe a system used to determine how
individual cells vary in their transcriptional and translational landscapes, and demonstrate the
low predictive power of mRNA levels over protein abundance.
My second approach to understand protein production is aimed at directly observing
protein synthesis in living cells. I describe the generation of animals used for imaging protein
production in single cells in real time. I also describe a system that uses the reconstitution of split
GFP as a spatial and temporal quantitative marker of local protein synthesis. I used single-cell
quantitative imaging, electrophysiology and immunocytochemistry to demonstrate that proteins
produced with split GFP reporters function properly, and that their level of production correlates
with the intensity of the reconstituted GFP signal.
The experiments presented in this thesis demonstrate tools I developed to probe protein
production with high spatial and temporal resolution. The tools and reagents are accessible to a
wide range of researchers and the assays provide high accuracy and reliability. Protein analysis
in single cells can reveal unprecedented insight into the dynamics of the gene expression.
iii
Résumé
La quantification précise de la production de protéines est un outil fondamental pour
comprendre les processus cellulaires et moléculaires de base. La dérégulation des niveaux de
protéines peut endommager les cellules et mener à des maladies telles que le cancer et les
maladies neurodégénératives. Cependant nous en savons très peu sur la façon dont la production
de protéine change dans une cellule avec le temps ou en réponse à des facteurs externes. Les
méthodes actuelles de quantification de la production de protéines sont invasives, prennent
beaucoup de temps et ont une mauvaise résolution. En conséquence les chercheurs se sont
tournés vers l'expression de l'ARNm afin de mesurer l'abondance de protéines, bien qu'il ait été
démontré que cette méthode est imprécise.
En réponse à ce problème, mon travail de thèse explore de nouveaux outils et techniques
que j'ai développés afin de mesurer la production de protéines dans une cellule vivante. En
examinant simultanément les niveaux d'ARNm et de protéines pour un gène dans une cellule
unique, je décris un système qui peut être utilisé pour déterminer les variations de cellule à
cellule dans les paysages transcriptionel et traductionel, et je démontre la capacité faible du
niveau l'ARNm à prédire l'abondance de protéines.
Ma seconde approche visant à comprendre la production de protéines et dirigée
directement vers l'observation de la synthèse de protéines dans des cellules vivantes. Je décris
l’élaboration d'animaux utilisés pour capturer la production de protéines dans une cellule unique
en temps réel. Je décris aussi un système qui utilise la reconstitution d'une protéine fluorescente
verte (PFV) découpée comme marqueur quantitatif spatial et temporel de la synthèse locale de
protéines. J'ai utilisé de l'imagerie quantitative de cellules uniques, de l'électrophysiologie et de
l'immunocytochimie afin de démontrer que les protéines produites avec comme reporter la PFV
découpée fonctionnent correctement, et que leur niveau de production sont en corrélation avec
l'intensité du signal de la PFV reconstituée.
Les expériences présentées dans cette thèse démontrent les outils que j'ai développés afin
d'examiner la production de protéines avec une haute résolution spatiale et temporelle. Les outils
et réactifs chimiques sont accessibles à une grande variété de chercheurs, et cette méthode
fournit une haute précision et fiabilité. L'analyse de protéines dans des cellules uniques peut
révéler une connaissance approfondie sans précédent de la dynamique de l'expression génique.
iv
Acknowledgements
My experience in Dr. Brian Chen’s laboratory has been nothing short of transformative. I
was an undergraduate with no research experience when Brian provided me an opportunity to
join his team. He patiently guided me through this journey, helping develop my skills both in the
lab and outside along the way. My gratitude also goes to Drs. Don Van Meyel, David Stellwagen
and Keith Murai for invaluable advice and guidance, as well as to members of my advisory
committee Drs. Artur Kania and Hiroshi Tsuda.
I would like to thank all my colleagues at the Centre for Research in Neurosciences. A
special thank you to Drs. Tiago Ferreira, Emily Peco, Todd Farmer and Haider Al-Timimi for
hallway chats that taught me more lessons than seminars. To Sejal Davla for constantly sharing
her expertise in everything from fly food to Indian food. Many thanks to Chris Salmon, Benny
Kacerovsky and Dr. Gael Quesseveur for help with mouse work. I will always cherish the
memories of the breakfasts I had every Thursday with Hunter Shaw, charmer of the sixth floor
volunteers.
I am indebted to Dr. Chiu-An Lo, who watched and helped me grow since my undergraduate
years, and together with Tsung-Jung Lin taught me most of what I know about molecular
biology. A special thank you to Dr. Farida Emran, the jane of all trades who also coached and
helped me with every aspect of my graduate work.
Last but certainly not least, to my guide and co-conspirator Dr. Vedrana Cvetkovska, you
have helped realize this work in more ways than you know. I dedicate this thesis to my parents
Dima and Anwar, my sister Yasmina and my pug Winston, for your unconditional love, support
and prodding.
v
Contribution of authors
This thesis is presented in thesis-based format in accordance with McGill University
Graduate and Postdoctoral Studies guidelines. It comprises original work from one published
manuscript, one submitted manuscript and one currently in preparation.
The work in this thesis is based on the Protein Quantification Ratioing (PQR) technique
that I co-developed with Dr. Chiu-An Lo, published as:
Lo C*, Kays I*, Emran F, Lin T-J, Cvetkovska V, Chen BE. Quantification of Protein Levels in
Single Living Cells. Cell Reports. 2015;13(11):2634-2644. doi:10.1016/j.celrep.2015.11.048.
Brian E. Chen designed the experiments and supervised the project. Chiu-An Lo, Ibrahim Kays,
Farida Emran, Tsung-Jung Lin, Vedrana Cvetkovska and Brian E. Chen performed experiments
and analyzed the data. Chiu-An Lo, Ibrahim Kays, and Brian E. Chen wrote the manuscript.
The published technique, to which I contributed 4 years of my graduate work, constitutes Dr.
Lo’s PhD thesis work, obtained under Dr. Brian Chen in 2016, and its development and
validation are outside of the scope of my thesis. In this thesis I use our published technique,
described throughout the thesis and in detail in Chapter 1, as a stepping stone for the
development of novel systems and techniques. As per the McGill University Graduate and
Postdoctoral Studies guidelines, I have obtained written consent from Dr. Lo to describe the
technique and include it as a resource for my work.
A modified version of Chapter 2 has been submitted for publication as:
Kays I, and Chen BE. Protein and RNA quantification in single cells, submitted
vi
Ibrahim Kays and Brian Chen designed the experiments. Ibrahim Kays collected and analyzed all
the data. Ibrahim Kays wrote the manuscript. Brian Chen supervised the study.
A modified version of Chapter 3 is in preparation for publication as:
Kays I, and Chen BE. Direct observation of local protein synthesis in vivo, submitted
Ibrahim Kays and Brian Chen designed the experiments. Ibrahim Kays collected and analyzed all
the data. Ibrahim Kays drafted the manuscript. Brian Chen supervised the study.
vii
Table of contents
Abstract .......................................................................................................................................... ii
Résumé .......................................................................................................................................... iii
Acknowledgements ...................................................................................................................... iv
Contribution of authors ................................................................................................................ v
Table of contents ......................................................................................................................... vii
List of figures ................................................................................................................................. x
List of abbreviations .................................................................................................................... xi
Chapter I - Current state of quantification of protein production .......................................... 1
1.1 Introduction ........................................................................................................................... 2
1.2 Protein structure .................................................................................................................... 3
1.2.1 The primary structure of a protein is its linear chain ...................................................... 4
1.2.2 Secondary structures interconnect and stabilize protein residues .................................. 5
1.2.3 Protein folding and maturation are prerequisite to function ........................................... 6
1.3 Protein function ................................................................................................................... 11
1.3.1 Regulation of protein function ...................................................................................... 11
1.3.2 Regulated protein synthesis and degradation ............................................................... 12
1.3.3 Protein phosphorylation ................................................................................................ 13
1.3.4 Regulated translation of localized mRNAs .................................................................. 14
1.3.5 Local translation of mRNA shapes development ......................................................... 16
1.3.6 Local translation of mRNA in neurons ......................................................................... 18
1.4 Quantification of gene expression and protein levels ......................................................... 20
1.4.1 Quantification of mRNA levels .................................................................................... 22
1.4.2 Quantification of protein levels .................................................................................... 23
1.4.3 mRNA levels as proxy for protein abundance .............................................................. 25
1.5 Fluorescence-based single cell resolution protein quantification ....................................... 26
1.5.1 Quantification of protein levels in single living cells ................................................... 27
1.5.2 Protein production reporters must be carefully chosen ................................................ 29
1.5.3 Approaches to visualizing locally translated proteins .................................................. 30
1.5.4 Requirements for a local protein synthesis reporter ..................................................... 34
1.5.5 Split GFPs are indicators of protein interaction ........................................................... 35
1.6 Figures ................................................................................................................................. 37
viii
1.7 Thesis introduction .............................................................................................................. 40
Chapter II - Quantification of mRNA and protein levels in single cells ................................ 41
2.1 Relation to overall thesis ..................................................................................................... 41
2.2 Introduction ......................................................................................................................... 42
2.3 Experimental design and detailed protocol ......................................................................... 44
2.3.1 Materials ....................................................................................................................... 45
2.3.2 Gene editing using CRISPR-Cas9 ................................................................................ 47
2.3.3 Single-cell protein level quantification ......................................................................... 49
2.3.4 Total RNA extraction ................................................................................................... 49
2.3.5 Reverse-transcription .................................................................................................... 50
2.3.6 Real-time polymerase chain reaction ........................................................................... 51
2.3.7 Calculation of absolute mRNA transcript number ....................................................... 52
2.3.8 Readout of amplification .............................................................................................. 53
2.3.9 Assay controls............................................................................................................... 54
2.3.10 Image acquisition and analysis ................................................................................... 55
2.4 Results ................................................................................................................................. 56
2.5 Discussion ........................................................................................................................... 59
2.6 Conclusion ........................................................................................................................... 62
2.7 Figures ................................................................................................................................. 64
Chapter III - A system for direct observation of subcellular protein translation in single
living cells. .................................................................................................................................... 79
3.1 Relation to overall project ................................................................................................... 79
3.2 Introduction ......................................................................................................................... 80
3.3 Materials and Methods ........................................................................................................ 83
3.3.1 Protein Quantification Reporter constructs .................................................................. 83
3.3.2 Split GFP DNA constructs ........................................................................................... 83
3.3.3 GFP1-10 protein production and extraction ................................................................. 84
3.3.4 GFP 11 peptides............................................................................................................ 85
3.3.5 Cell culture ................................................................................................................... 85
3.3.6 In vitro protein reconstitution ....................................................................................... 86
3.3.7 Endoplasmic reticulum and ribosome staining ............................................................. 86
3.3.8 Electrophysiology ......................................................................................................... 87
3.3.9 Image acquisition and analysis ..................................................................................... 87
ix
3.3.10 Statistical analysis....................................................................................................... 88
3.4 Results ................................................................................................................................. 89
3.4.1 GFP11 and GFP1-10 reconstitute spontaneously in vitro ............................................ 89
3.4.2 GFP reconstitution in vitro occurs at millisecond timescales ....................................... 90
3.4.3 GFP11 detects GFP1-10 in living cells ........................................................................ 91
3.4.4 GFP reconstitution can report sites of protein translation ............................................ 92
3.4.5 Proteins co-translated with GFP11 reporters function properly ................................... 94
3.4.6 GFP reconstitution can quantitatively readout protein translation ............................... 95
3.5 Discussion and conclusions ................................................................................................. 96
3.5.1 GFP1-10 fluorophore maturation ................................................................................. 97
3.6 Figures ............................................................................................................................... 100
Chapter IV - Applications and future directions of protein quantification using PQR ..... 109
4.1 Relevance to overall project .............................................................................................. 109
4.2 Applications of optical protein quantification using PQR ................................................ 110
4.2.1 Dynamic observation of protein synthesis in vivo ..................................................... 110
4.2.2 Optical normalization of protein production in vivo .................................................. 112
4.3 Split GFP as a quantitative marker of local protein synthesis in vivo .............................. 114
4.3.1 Generation of animals constitutively expressing GFP1-10 ............................................ 115
4.3.2 Local translation of Gurken protein in Drosophila oocytes ........................................... 117
4.3.3 Detection of local protein translation in living neurons ................................................. 121
4.4 Detection of local protein synthesis using PQR photoconvertible reporters. ................... 123
4.5 Figures ............................................................................................................................... 128
Chapter V - Thesis directions and conclusions ...................................................................... 137
References .................................................................................................................................. 142
x
List of figures
Figure 1.1 Brightfield image of dissected whole ovarioles ......................................................... 37
Figure 1.2 PQR reporters allow quantification of protein production from cells. ....................... 39
Figure 2.1 Workflow of protein and mRNA measurement from the same cell. .......................... 64
Figure 2.2 Protein and mRNA measurement for multiple genes in a single cell. ........................ 66
Figure 2.3 Insertion of PQR-XFP reporters into the endogenous genomic loci of IgK and Rpl13a
using CRISPRs.............................................................................................................................. 68
Figure 2.4 Validation of CRISPR-mediated insertion of PQR-GFP in the endogenous IgK locus.
....................................................................................................................................................... 69
Figure 2.5 Illustration of the important steps and typical equipment used in the protocol. ......... 70
Figure 2.6 Titration of starting input cDNA volume. .................................................................. 72
Figure 2.7 Standard curve of serially diluted known amount of Rpl13a target. .......................... 73
Figure 2.8 Contamination of RNA sample quantification by genomic DNA can be assessed
using no-RT control reaction. ....................................................................................................... 74
Figure 2.9 Endogenous RNA and protein quantification from single cells. ................................ 75
Figure 2.10 Protein and mRNA relationships between multiple genes in single cells. ............... 76
Table 2.1 Sequences of primers and probes used in this protocol. .............................................. 78
Figure 3.1 Stoichiometric production of GFP11 reporters using PQR. ..................................... 100
Figure 3.2 In vitro characterization of the split GFP reconstitution reaction. ........................... 101
Figure 3.3 Reconstitution of split GFP occurs on the order of milliseconds in vitro. ............... 102
Figure 3.4 Split GFP reporters can be expressed using PQRs and the reconstitution of GFP
marks the presence GFP1-10 protein. ......................................................................................... 103
Figure 3.5 Split GFP reconstitution occurs at sites of active protein translation. ...................... 106
Figure 3.6 Co-translation of GFP11 reporters using PQR preserves the protein of interest’s
localization and function. ............................................................................................................ 107
Figure 4.1 PQR reporters are inserted in-frame into endogenous genes. .................................. 128
Figure 4.2 PQR constructs injected into mouse embryos result in red fluorescent pronuclei. .. 129
Figure 4.3 SplitGFP as a protein translation reporter. ............................................................... 131
Figure 4.4 GFP1-10 is expressed at high levels in transgenic animals. ..................................... 132
Figure 4.5 GFP11 can detect Gurken local translation in oocytes. ............................................ 134
Figure 4.6 Novel split fluorescent reporters exhibit more efficient reconstitution. ................... 136
xi
List of abbreviations
22c10 IgK-secreting mouse hybridoma cell line
2A Self-cleaving 2A peptide
3D Three dimensional
a.u. Arbitrary units
ActB Beta-actin
AMPA α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid
Arc Activity-regulated cytoskeleton-associated protein
ATP Adenosine triphosphate
BDNF Brain-derived neurotrophic factor
BFP blue fluorescent protein
bp Basepair
BSA Bovine serum albumin
cAMP Cyclic adenosine monophosphate
CCD Charge-coupled device
cDNA Complementary deoxyribonucleic acid
CHYSEL Cis-acting Hydrolase Element
CO2 Carbon dioxide
CRISPR Clustered regularly interspersed palindromic repeats
Ct Cycle threshold
ddH2O Double distilled water
DNA Deoxyribonucleic acid
DSB Double-strand break
E. coli Escherichia coli
ELISA Enzyme-linked immunosorbent assay
ER Endoplasmic reticulum
FACS Fluorescence-assisted cell sorting
FMRP Fragile X mental retardation protein
FP Fluorescent protein
GFP (sfGFP) Green fluorescent protein (superfolder GFP)
Gria1/GluR1 Glutamate ionotropic receptor AMPA type subunit 1
Grk Gurken
gRNA (sgRNA) Guide RNA
HEK293/T Human embryonic kidney 293 (Transformed)
hnRNP-R Heterogeneous nuclear ribonucleoprotein R
Ig Immunoglobulin
IgK Immunoglobulin light chain kappa
IHC Immunohistochemistry
IRES Internal ribosomal entry site
I-V Current-voltage
kDa Kilodalton
LTP Long-term potentiation
MAP1b Microtubule-associated protein 1b
MAP2 Microtubule-associated protein 2
mGluR1 Metabotropic glutamate receptor 1
mGRASP Mammalian GFP reconstitution across synaptic partners
xii
mRNA Messenger ribonucleic acid
NGF Nerve growth factor
nls Nuclear localization signal
nols Nucleolar localization signal
OMP Orotidine 5’phosphate decarboxylase
OPT GFP11-OPT, optimized
pCAG Promoter from Cytomegalovirus, beta-Actin and beta Globin genes
PCR Polymerase chain reaction
PEST Sequences containing Proline, Glutamic acid, Serine and Threonine
pJFRC Janelia Farms Research Center promoter
PQR Protein quantification reporter
PSD95 Postsynaptic density protein 95
qPCR Quantitative real time PCR
R2 Pearson’s linear regression correlation coefficient of determination
RFP Red fluorescent protein
RhoA Ras homolog family member A
RNA Ribonucleic acid
RNase Ribonuclease
ROI Region of interest
RPL13A Human ribosomal protein L13A
Rpl13a Mouse ribosomal protein L13a
RT Reverse-transcription
RT-qPCR Reverse-transcription real-time polymerase chain reaction
SDS-PAGE Sodium dodecyl sulfate-polyacrylamide gel electrophoresis
sec Second(s)
SMN1 Survival of motor neuron 1
UAS Upstream activator sequence
UTR Untranslated region
VegT VegT protein
XFP Generic fluorescent protein (any color)
αCaMKII Alpha Ca2+/Calmodulin-dependent protein kinase II
1
Chapter I - Current state of quantification of protein
production
2
1.1 Introduction
Proteins execute almost all fundamental cellular functions. They are the products of
genetic information and the molecular actors that affect change within a cell. To begin and
continue, life requires information and the ability to control this information. A hallmark of
organisms is the exquisite control that has evolved to ensure that biological processes, including
protein production, only occur when and where they are meant to.
Most chemical reactions in living systems are catalyzed by proteins. Proteins lower the
activation energy of cellular reactions, allowing them to proceed at faster rates. A classic and
illustrative example is the Orotidine 5’ phosphate (OMP) decarboxylase, which has been shown
to accelerate uncatalyzed reactions by a factor of 1017 (Radzicka & Wolfenden, 1995). To put
this into perspective, OMP decarboxylase does in 18 milliseconds what would otherwise require
78 million years to spontaneously occur (Callahan & Miller, 2007). Since all cellular function
involves defined chemical reactions, all cellular functions directly involve or require proteins.
Many proteins can be grouped into a few broad functional categories. Structural proteins give
cells shape and form highways within cells that direct the intracellular transport and localization
of various molecules and organelles. Regulatory proteins act as sensors or signals that orchestrate
and fine-tune cellular processes. Signaling proteins relay information from one cellular location
to another. Membrane transport proteins permit the flow of ions and molecules across cellular
membranes. Finally, protein enzymes form and break covalent bonds, allowing chemical
reactions to occur (Alberts et al., 2002; Lodish et al., 2008).
Proteins are the result of translation of messenger RNA by ribosomes, which are large
and complex molecular machines that link amino acids to form peptides and proteins according
to genomic instructions encoded by the mRNA sequence. The rate of synthesis of a protein is
3
dependent on the rate of transcription of the gene into mRNA, the steady-state abundance of the
mRNA within the cell, and the rate of translation or rate at which the mRNA is converted into
protein (Lodish et al., 2008). However, despite the widespread use of mRNA as a proxy measure
for protein abundance, tools that instead directly assay protein production unarguably offer more
insight into the regulation of the proteome, and its consequences.
In this thesis, I describe my efforts to understand when and where proteins are produced
by developing tools to monitor protein production in single cells. In Chapter 2, using a novel
technique I present results that demonstrate that mRNA levels do not necessarily correlate with
protein production, particularly in single cells. Therefore, Chapters 3 and 4 describe tools I have
developed to observe protein production in single living cells with high spatial and temporal
resolution. The work presented in this thesis provides a framework and resources that will allow
for monitoring of global and local protein synthesis in vivo.
1.2 Protein structure
Louis Sullivan is perhaps best known among scientists as the architect who proposed the
dicta “form follows function” (L. H. Sullivan, 1896), which is true for most man made structures.
However, in evolutionary and protein science, the reverse is true (Pauwels & Tompa, 2016). Key
to the elucidation of how proteins execute specific functions is the concept that protein function
is derived from three-dimensional (3D) structure. Therefore, a prerequisite to understanding and
manipulating protein function is understanding the structure of proteins. One of the best
examples illustrating this approach is with the green fluorescent protein (GFP) (Prasher et al.,
1992; Ward et al., 1980) .
GFP is arguably the most widely studied and known fluorescent protein. The fact that it is
genetically encoded and can autocatalytically develop the green fluorescent signal make it a tool
4
that has helped advance our understanding of several areas of biology in a revolutionary way,
allowing us to literally observe the interior of living cells and visualize cellular processes as they
occur. Since its discovery and cloning from the jellyfish Aequoria victoria, GFP has been the
subject of many experiments aimed at improving many of its properties including stability and
fluorescence emission.
The common variants of GFP used today offer a number of significant improvements
over the original wild type version, which was dim in brightness, unstable at 37°C, and had two
excitation peaks with the dominant one in the ultra-violet range (Prasher et al., 1992; Tsien,
1998). Over the years many mutations have been introduced and hundreds of variants of GFP are
now available for a wide range of applications (Kent et al., 2008; Tsien, 1998; Zimmer, 2002).
Understanding the structure of GFP was crucial to enable the rational design of mutations that
predictably altered its properties. For example, the elucidation of the GFP crystal structure in
1996 allowed researchers to design new versions of the protein in a wider range of colors,
brightness and stability (Remington, 2011; Tsien, 1998). In this thesis, I developed tools to
monitor protein translation in single living cells using fluorescent reporters, such as GFP. The
well-known history and properties of GFP allow us to exploit different characteristics to develop
new assay-tailored tools using GFP. In the following section I will describe aspects of protein
structure in the context of GFP.
1.2.1 The primary structure of a protein is its linear chain
The primary structure of a protein is the simple linear arrangement or sequence of its
amino acid constituents. At this fundamental level, proteins are constructed by the
polymerization of 20 different types of amino acid building blocks. Amino acids have an amine
(N) group and a carboxyl (C-O) group on either end, and peptide bond formation between the
5
amine group of one amino acid and the carboxyl group of another amino acid results in the net
release of one water molecule and their linkage via a peptide bond.
The average molecular weight of an amino acid in a protein, given average relative
abundance, is ~ 113 Da (Daltons). Therefore, the number of residues in a protein can be
estimated from the molecular weight of the protein, and vice versa. The Dalton mass unit is used
to report the size of a protein, with 1 Dalton = 1 atomic mass unit (or molecular weight) (Alberts
et al., 2002; Lodish et al., 2008). For example, GFP is a relatively small 238 amino acid protein
that has a molecular mass of 26.9 kDa, in contrast to large immunoglobulin antibody molecules
which are on average 150 kDa.
1.2.2 Secondary structures interconnect and stabilize protein residues
Protein secondary structures are stable spatial arrangements of segments or domains of a
protein chain. They are formed and linked together by non-covalent (mainly hydrogen) bonds
between backbone oxygen and hydrogen atoms. Repeating secondary structures are often found
along the protein chain, and a single protein may contain several types of secondary structures in
several parts of the polypeptide chain, depending on the amino acid sequence in that region. The
main types of secondary structures are the alpha helix, the beta sheet and the beta turn.
In an average protein, roughly 60% of the chain forms alpha helices and beta sheets.
Within an alpha helix, the tight spiralling of residues serves to hold portions of the protein
backbone in a rigid, rod-like cylindrical structure. Alpha helices are usually abundant in proteins
present in cell membranes, where their transmembrane domain is largely a straight alpha helix
that traverses the plasma membrane.
Beta sheets are composed of laterally packed beta strands, which in turn are composed of
5- to 8-residue stretches of fully extended polypeptide segments. The core of many proteins is
6
composed of regions rich in beta sheets, and these regions can form either between adjacent
domains along the polypeptide chain or between domains far apart within the sequence that can
fold back on themselves and form dense beta sheets. Both types are held together by extensive
hydrogen bonding and this results in highly rigid structures.
The GFP molecule is barrel shaped, consisting of 11 beta strands and a central coaxial
helix, with the chromophore forming from the central helix (Ormö et al., 1996). The beta sheets
shield the chromophore from the outer environment and this is important as removal of
individual beta sheets results in loss of fluorescence emission (Chapters 3 and 4). In addition, the
abnormal oligomerization of protein domains into beta sheet-rich structures is associated with
several pathological states, as epitomized by the implication of oligomerized amyloid beta
protein as one cause of Alzheimer’s disease (Nelson et al., 2005).
The transition of proteins from their linear string of residues into the three-dimensional
world of molecules enables proteins to have functions. The addition of salt bridges, disulfide and
hydrogen bonds, as well as the tight packing of side chains lock protein domains into place,
giving soluble proteins such as antibodies and GFP a compact globular three-dimensional
structure. The folding of a straight protein chain into the tertiary and quaternary structures
observed in living cells is a prerequisite and crucial step to “switch on” the function of proteins.
1.2.3 Protein folding and maturation are prerequisite to function
Thousands of types of proteins exist in every organism (Dobson, 2001). Following
synthesis by the ribosome, each protein must fold into the particular conformational shape
dictated by its sequence, to carry out its function. The Levinthal paradox is a thought experiment
that proposed it would take longer than the age of the universe for even a short polypeptide to go
7
through all possible conformations before arriving at the correct (lowest energy) structure
(Levinthal, 1969). Yet in the cell, protein folding takes place in seconds (Piana et al., 2013).
In Chapters 2 and 3, I use GFP as a reporter of antibody production by co-expressing it
with the kappa light chain of mouse antibodies. In the next sections I describe folding and
maturation of GFP and mammalian immunoglobulin antibodies.
1.2.3a Green fluorescent protein
Almost all cellular proteins require some folding before they can mature and begin to
function. GFP is nonfluorescent as it leaves the ribosome and only begins to emit fluorescence
when proper 3D structure is achieved, allowing the chromophore to form. The chromophore, p-
hydroxybenzylideneimidazolinone, is the source of fluorescence emission and is formed from the
spontaneous cyclization and oxidation of residues serine 56 (S65), tyrosine 66 (Y66) and glycine
67 (G67), by essentially a nucleophilic attack at the carbonyl carbon of S65 by the amide
nitrogen of G67 followed by dehydration. In the presence of molecular oxygen, Y66 and the new
imidazolinone group conjugate, forming the mature excitable chromophore (Cody et al., 1993;
Prasher et al., 1992). GFP exhibits green (509 nm) fluorescence when excited with blue (488 nm)
light (Prasher et al., 1992). At a rate of 6 amino acids/second (Ingolia et al., 2011), it takes
roughly 40 seconds for one molecule of GFP to be translated in generic mammalian cytoplasm
and several minutes for folding to be completed (Shaner et al., 2008). Therefore, crucial to
chromophore maturation and fluorescence emission is the proper folding of the protein to near
native conformation. Protein maturation always proceeds protein folding and it is important to
distinguish these two processes. It is also important to note, unlike most proteins that require
molecular chaperones to fold, no enzymes or cofactors are needed for GFP maturation to occur,
except molecular oxygen. Dependence of maturation on oxygen was established based on the
8
simple finding that GFP fluorescence does not develop in the absence of atmospheric oxygen
(Cubitt et al., 1995; Heim et al., 1995; Inouye & Tsuji, 1994). GFP that has been produced in
anaerobic conditions is non-fluorescent. However, the protein emits fluorescence that develops
exponentially after air is introduced. This process was measured to take 95 minutes and found to
be unaffected by either the concentration of starting GFP or the presence or absence of various
tested cellular factors (Heim et al., 1995).
1.2.3b Immunoglobulin antibodies
In the immune systems of vertebrates, four polypeptide molecules, two identical heavy
chains and two light chains constitute the basic structural unit of an antibody molecule. Five
classes of mammalian antibodies exist, each with its own class of heavy chain and either of two
classes of light chains: kappa (κ) and lambda (λ). Examination kappa chain amino acid
sequences has shown the C-terminal region consists of nearly identical residues, while the
variable N-terminal half consists of a relatively constant framework region and three small
hypervariable loops that provide the structural basis for the diversity of antigen-binding sites. A
pair of 7-9 antiparallel beta strands forms an 80 amino acid barrel-like structure termed the
immunoglobulin (Ig) domain, which forms the basis for protein-protein or protein-ligand
interactions (Janeway et al., 2001; Lodish et al., 2008).
The diversity in antigen binding sites is the result of extensive genomic rearrangement
that occurs at immunoglobulin gene loci. In mice, the total number of immunoglobulin kappa
(IgK) genes is 180 (174 variable genes, 5 joining genes and 1 constant gene). Recombinase
enzymes recognize recombination signal sequences which choose and join one variable gene in
frame with one joining gene and the common constant exon. Such mechanisms which are
essential to generate large diversities in antibody specificity, necessarily increase the probability
9
of producing proteins that fold improperly or cannot assemble. Instability in protein structure can
hinder the transport and secretion of antibodies in addition to restricting the interaction of
antibodies with signaling molecules, all of which compromise immune responses. It is therefore
not surprising that some of the earliest endoplasmic reticulum folding enzyme substrates were
immunoglobulin molecules (Haas & Wabl, 1983; Rao et al., 1976). Moreover, many protein
quality control mechanisms that ensure only correctly assembled proteins are retained, were
originally identified by their association to Ig antibody chain production. (Feige et al., 2010).
Using molecular dynamics simulations, it is now generally agreed that small biases
towards native-like states during protein folding lead to stable intermediary transition states that
reduce the conformational search to obtain realistic folding times (Martinez et al., 1998; D. C.
Sullivan & Kuntz, 2002). Following folding, quality control mechanisms ensure that only
properly folded and assembled proteins remain, and misfolded proteins are sent for degradation
and recycling in the proteasome (Lodish et al., 2008).
1.2.3c Measurement of protein folding
It is not trivial to measure the time it takes for a protein to fold and mature within a cell,
primarily because it is near impossible to establish a “time zero” moment when the protein is
translated. A complex orchestration of modifications and interactions follow the synthesis of
proteins in cells: for example, many proteins begin folding as the nascent peptide chain exits the
ribosome tunnel (Holtkamp et al., 2015). Many other proteins, such as proinsulin, are produced
in an inactive form and require posttranslational cleavage by a protease to generate the active
form of the protein (Orci et al., 1986). However, certain aspects of folding and maturation can be
utilized to untangle the two processes. Characteristic protein functions and properties such as the
emission of fluorescence in fluorescent proteins can be used, with caution, as reporters of protein
10
maturation (Craggs, 2009). For example, since the rate limiting step of GFP maturation is the
final oxidation step, one way maturation rates of different GFPs were determined in vitro and in
vivo was by manipulating the levels of available molecular oxygen in the assay, either by using
anaerobic in vitro protein synthesis (Iizuka et al., 2011) or by growing bacteria in anaerobic
conditions (Hansen et al., 2016; K. P. Scott et al., 1998). During synthesis the protein begins to
fold, and the maturation process begins with cyclization and dehydration, but halts at the rate
limiting oxidation step. Measuring the fluorescence recovery after oxygen admission effectively
measures the rate of the last and longest step of chromophore maturation (Iizuka et al., 2011).
GFP folding rates are usually measured by first completely denaturing the protein
rendering it unfolded using denaturants such as urea, then measuring the time it takes for
fluorescence to develop after washing the urea off (Pédelacq et al., 2006). It should be kept in
mind that prior to protein unfolding the chromophore is already in its mature form, and when the
protein is unfolded the chromophore remains in a cyclized state (Waldo et al., 1999; Ward et al.,
1980). Therefore, the maturation of the GFP chromophore is a permanent indicator that it had
once folded properly to its native state (Reid & Flynn, 1997). This eliminates the need for the
unfolded protein to go through the maturation step during subsequent refolding, therefore
measuring the fluorescence recovery effectively measures the rate of the folding step, unaffected
by maturation (Waldo et al., 1999).
There are several reasons why researchers require fluorescent proteins to have stable and
characterized folding and maturation times. One of the earliest and most common uses of GFP in
cells is its fusion to a protein to investigate that protein’s localization (Htun et al., 1996; Marshall
et al., 1995; Rizzuto et al., 1995). Early unstable versions of GFP posed several problems in
fusion protein experiments as improperly folded proteins aggregate into inclusion bodies. Protein
11
aggregates can be toxic to cellular health, and render the spatial visualization of proteins or
quantification of fluorescence intensity ambiguous (Kopito, 2000). Moreover, assays requiring
tagging of low abundance proteins, or those with short half-lives, require reporters that have high
spatiotemporal resolution. Therefore, reporters that present any time delay between their
production and the emission of fluorescence can preclude accuracy in determining when and
where the protein is produced.
1.3 Protein function
The biological properties of a protein determine the type of interaction or binding a
protein undergoes. The regulation and modification of such properties by cells is a direct way to
regulate protein function. All proteins interact with other molecules, be it small molecules,
nucleic acids or other proteins. For example, antibodies bind to antigens and mark them for
phagocytosis and destruction, cell surface receptors bind other protein ligands and transduce
signals, and ion channels bind ions and metals. The ability for a protein to bind its ligand with
high selectivity and affinity depends on the degree of formation of weak non-covalent bonds
such as hydrogen bonds, van der Waals attraction and ionic bonds, which are in turn dictated by
how closely the ligand and protein surfaces fit together, analogous to a lock and key concept
(Alberts et al., 2002; Lodish et al., 2008).
1.3.1 Regulation of protein function
Many proteins have critical cellular functions such as the catalysis of master metabolic
reactions that have widespread downstream effects, or the critical regulation of cell division to
prevent abnormal cell proliferation as seen in many cancers. Therefore, the existence of different
regulatory mechanisms that coordinate protein activity is key to proper cell function. As such,
12
the regulation of the function or activity of a protein is achieved at several levels during a
protein’s life span within the cell. In general, there are at least three ways in which cells can
regulate the activity of proteins (Lodish et al., 2008). One of the first levels of control over
protein activity is at the level of regulation of gene expression. Upregulation and downregulation
of mRNA expression and translation, in addition to protein degradation alter the steady-state
level of the protein, and by proxy the level of activity that results in the cell. A second level of
regulation of protein activity is using molecules that are either irreversibly added to the structure
of the protein, or that bind and dissociate acting as molecular on and off switches that control
protein activity. A third way by which cells can regulate protein activity is by localizing and
concentrating proteins to subcellular compartments to spatially restrict their activity. Similarly,
cells can produce point sources or gradients of another molecule or cofactor required for the
activity of the protein, thereby resulting in a spatial gradient of activation or suppression of
protein activity (Alberts et al., 2002; Lodish et al., 2008).
1.3.2 Regulated protein synthesis and degradation
Intracellular and extracellular cues often work in concert to regulate gene expression. For
example, increased neuronal firing stimulates the expression of immediate-early gene mRNAs
such as c-fos and Arc and their subsequent conversion to protein within minutes (Gissel et al.,
1997; Na et al., 2016). Such markers of external stimuli are labile, and must be recycled quickly
to prevent the constitutive activation of signaling and response mechanisms.
The stabilities and half-lives of cellular proteins vary widely. Proteins are constantly
degraded and replaced with newly synthesized copies, and this turnover process ensures a
constant supply of new functional protein to replace non-functional and damaged (sometimes
toxic) species. Turnover studies across organisms have shown that protein half-life can vary
13
many orders of magnitude within cells (Toyama & Hetzer, 2013). For example, examination of
turnover rates in budding yeast (cell cycle of 1.5 hours) revealed that mean protein half-life
under normal conditions is ~43 minutes (Belle et al., 2006), and in dividing human cells (cell
cycle of 24 hours), this figure increases to 0.5-35 hours (Cambridge et al., 2011). Although
protein half-life within the cell can vary from minutes to days, the turnover of proteins seems to
correlate with their function or subcellular localization (Toyama & Hetzer, 2013). For example,
endoplasmic reticulum and mitochondrial proteins on average have longer half-lives than other
cellular proteins (Price et al., 2010). In contrast, eye lens, tooth enamel and tooth dentine protein
half-lives are on the order of decades (Helfman & Bada, 1975, 1976; Masters et al., 1977).
Accelerated turnover of proteins is often the result of signals or modifications that are
added to the protein to increase its rate of degradation. For example, sequences rich in proline
(P), glutamic acid (E), serine (S) and threonine (T) (PEST) are associated with proteins that have
a short intracellular half-life such as the neuronal activity-regulated protein Arc (t1/2 ~ 37 mins),
whose expression marks short-term increased neuronal activity. In other cases, phosphorylation
can mark a protein for rapid turnover and degradation. For example, the production and
degradation of circadian clock protein such as period and timeless, are tightly regulated via
phosphorylation, and the resulting accelerated turnover ensures constant levels of circadian
proteins are never achieved, and instead a cycle is preserved (Kwon et al., 2006). Prior to
degradation, many proteins require the addition of ubiquitin groups, which is catalyzed by a
number of ubiquitin modification enzymes. The number of ubiquitin tags a protein contains
usually correlates with the level of recognition and degradation by the 26S proteasome (Smalle
& Vierstra, 2004).
1.3.3 Protein phosphorylation
14
The phosphorylation of proteins imparts negative charges, which result in subtle
conformational changes in protein shape that affect ligand binding or other functions of the
protein. Protein kinases catalyze phosphorylation of specific residues (tyrosine or serine and
threonine kinases) and this reaction is reversed by protein phosphatases, which hydrolyze and
release phosphate groups from modified residues. The combined action of protein kinases and
phosphatases is an important regulatory mechanism that can act as a reversible switch to increase
or decrease the activity of proteins, as such it is highly conserved across prokaryotes and
eukaryotes (Alberts et al., 2002). All classes of proteins including structural proteins, signaling
proteins, enzymes, membrane channels and scaffolds are known to be regulated by
phosphorylation. At any given time, 1/3 of the proteome is thought to be phosphorylated,
resulting in many proteins having more than one phosphate group (Lodish et al., 2008).
Through similar effects, reversible protein phosphorylation is known to control the
structure, localization and therefore function of a host of eukaryotic proteins. The simple
addition and removal of phosphate groups in response to signals is an elegant mechanism used to
ensure adequate protein activity, relay of signals between organelles or as a check-point for
molecular processes (Johnson, 2009).
1.3.4 Regulated translation of localized mRNAs
The localization and regulated translation of mRNA is an example of a cellular regulatory
mechanism used to spatially and temporally restrict gene expression to discrete cellular sites
(Martin & Ephrussi, 2010). Some of the best studied examples of locally translated mRNAs
include those whose protein products must be spatially sequestered in order to play specialized
roles within defined subcellular compartments (Ables, 2015; Martin & Ephrussi, 2010). For
example, the induction of mesodermal and endodermal cell fates in the Xenopus oocyte vegetal
15
pole is regulated by the local translation of VegT mRNA (King et al., 2005). Similarly, the
specification of asymmetric cell division in solely daughter cells of budding yeast is a result of
localized translation of ASH1 transcriptional repressor mRNA in the daughter cell (Paquin &
Chartrand, 2008). In addition, large-scale analyses have revealed several other genes whose
mRNAs are spatially sequestered and locally translated. For example, high-throughput in situ
hybridization of 3,000 mRNA transcripts in Drosophila has found that more than 70% of the
assayed transcripts were present in spatially distinct patterns (Lecuyer et al., 2007), which
indicates that the localization of mRNAs to subcellular compartments is more prevalent than
previously thought (S. Kim et al., 2010).
There are many advantages of regulating gene expression via local mRNA translation.
First, it is a mechanism that allows spatial restriction of gene expression within a cell’s
cytoplasm. Second, it provides cells with fine temporal resolution to control translation of
already spatially restricted transcripts, using local cues that can stimulate initiation of translation
on site. Third, almost every step involved in the production and transport of a protein to sites
where it is needed, requires expensive cellular reagents. Therefore, it is much more economical
for cells to locally produce many copies of a protein from the same few mRNA molecules, rather
than shuttle individual proteins from the cell soma to where they are required (Martin &
Ephrussi, 2010)
The targeting of mRNAs to their cellular destination involves cis-acting elements in the
RNA sequence. Most often, these localization elements are found in the 3’UTR, although some
transcripts such as the sensorin gene mRNA contain localization elements in their 5’UTR that
allow them to be targeted to dendrites and synapses (Dan Ohtan Wang et al., 2009). In addition,
some mRNA localization elements are found within the coding sequence, such as in the gurken
16
mRNA in Drosophila (Lasko, 2012). Several studies have shown that the processing of pre-
mRNA in the nucleus is required for its cytoplasmic localization by interacting with RNA
binding proteins that recognize localization and translational regulation signals within the
sequence (Giorgi & Moore, 2007). The association of RNAs with RNA binding proteins forms
ribonucleoprotein complexes which in many cases are part of a bigger RNA transport granule, a
structure transported by motor proteins that use the cell cytoskeleton to deliver the RNA to its
final destination (Martin & Ephrussi, 2010). During mRNA transport, mechanisms are in place to
repress the translation of RNA until the granule is anchored at its destination, where additional
mechanisms ensure it is translated at the right time (Besse & Ephrussi, 2008).
1.3.5 Local translation of mRNA shapes development
The local translation of mRNAs in the Drosophila oocyte is one of the earliest and best
studied examples of how the localization of mRNA is used as a mechanism for translational
regulation. During egg development, germ cell specification and embryonic axis patterning are
established via molecular asymmetries created by position-dependent regulation of the
translation of mRNAs deposited maternally into the oocyte (Richter & Lasko, 2011). The
location-dependent activation of translation is also coupled with mechanisms that localize and
concentrate specific mRNA transcripts in areas where the corresponding protein will be
produced. Such mechanisms ensure that proteins are present in their highest concentrations
where they are required, as opposed to where their presence would be deleterious.
The Drosophila ovary consists of ovarioles that harbor oocytes and provide the
microenvironment necessary for the development of the oocyte and the synthesis of maternally
deposited DNAs and RNAs that will be required post-fertilization for the proper development of
the egg axis and shell (Ables, 2015). At the anterior-most end of each ovariole is the germarium,
17
a structure which houses the primordial germ cells from which oocytes develop (Figure 1.1).
These germline stem cells undergo several rounds of asymmetric cell division to produce the
germline cyst, a cluster of 16 germ cells interconnected by intracellular ring canal bridges.
Fifteen of these cells will be destined to become nurse cells and the remaining cell becomes the
oocyte (Figure 1.1). Nurse cells produce several mRNAs, such as the gurken (grk) mRNA which
is transported into the oocyte where it colocalizes with the oocyte nucleus. The 1.7 kb grk
mRNA encodes the Gurken protein which is a ligand for Torpedo/EGF receptor, a receptor
located on the inner surface of follicle cells that envelop the oocyte. The analysis of grk mRNA
transcripts has revealed a conserved RNA stem loop element within the grk coding region that
forms the signal for dynein-dependent grk mRNA transport and localization to the oocyte
nucleus (Van De Bor et al., 2005).
During the early stages of oogenesis (stage 6 onwards) the nucleus is located at the posterior
end of the oocyte and grk mRNA accumulates there, however, during later stages of
development (stages 9,10 and 10B) the nucleus moves to the anterodorsal corner of the oocyte,
where the grk mRNA follows to create the characteristic crescent pattern of localization between
the apical surface of the nucleus and the surrounding cortex (Richter & Lasko, 2011). Localized
translation of Gurken protein at the anterodorsal corner creates a local source and molecular
gradient of Gurken signalling such that the highest EGF-R/Torpedo signalling occurs in
neighbouring anterodorsal follicle cells which initiates cell fate specification locally (Gavis,
1995). The anterior-posterior specification of polarity in the oocyte arises from the movement of
the oocyte to the posterior of the egg chamber prior to stage 6. This precedes the dorsoventral
specification of the oocyte axis at stage 8 which is mediated by the movement of the grk mRNA
from the anterior cortex of the oocyte to the anterodorsal corner.
18
The local translation of Gurken in Drosophila oocytes is therefore a defined system in which
the spatial and temporal characteristics of Gurken mRNA translation are known and invariable.
1.3.6 Local translation of mRNA in neurons
Animal development and behavior are shaped by wiring and activity of neural circuits.
During the formation and development of circuits, neurons elaborate processes that can extend
great distances before forming synaptic contacts with their partners. Axonal and dendritic
subcellular compartments must integrate a wide array of molecular cues whose correct spatial
and temporal processing is critical for correct patterning and circuit formation. Stimulus-induced
changes in the structure and function of these compartments are vital to the formation and
plasticity of neural circuits (Kandel, 2001). The ever-changing demands of growing axons and
dendrites raise the question of how gene expression can be spatially restricted within a neuron.
The localized translation of mRNA provides one solution, and translation of localized transcripts
within different subcellular neuronal compartments has been observed (Jung et al., 2012; Dan
Ohtan Wang et al., 2010). For example, local translation of proteins can occur in axonal growth
cones during axon guidance and circuit formation (Lin & Holt, 2008), in dendritic spines during
learning and memory formation (Dan Ohtan Wang et al., 2009) and in axons during injury-
related regeneration (Willis & Twiss, 2006). In addition, local protein translation rates have been
shown to vary throughout development. For example, in rat hippocampal neurons, radiolabeled
amino acid incorporation analysis showed that rates of local protein synthesis in dendrites rise
early in development and peak during synaptogenesis, but decline by adult stages (Steward et al.,
1998).
Studies of activity-regulated genes were among the first to show that synaptic activity
regulates dendritic mRNA localization, and in a transcript-specific manner. For example,
19
induction of LTP in rat hippocampus has been shown to result in increased localization of
αCaMKII and MAP2 mRNA to granule cell dendrites in vivo (Roberts et al., 1998). Similarly,
newly synthesized Arc mRNA specifically localizes to and accumulates in activated synapses of
the rat dentate gyrus (Steward et al., 1998).
Other types of cues may also signal mRNA transport and local protein synthesis. For
example, the regulation of axon growth cone architecture requires the local translation of
cytoskeletal proteins such as beta-actin and cofilin, in addition to signaling proteins such as
RhoA and MAP1b, demonstrating an important role for axonal protein synthesis in the
development and maintenance of neuronal function (Hengst & Jaffrey, 2007). During
development of mouse dorsal root ganglion neurons, axonal mRNA translation is responsible for
retrograde signaling that regulates transcription of genes in the nucleus. Specifically, cAMP
response binding element (CREB) local translation in axons via NGF signaling is required for
somatic CRE-dependent transcription and subsequent neuronal survival mediated by nerve
growth factor (NGF) (Cox et al., 2008; Sharon A Swanger & Bassell, 2011). This indicates that
the distal production of proteins has important implications on the expression of genes in the
nucleus and thus can result in cell-wide changes. Moreover, the involvement of neuronal survival
factors such as NGF, cAMP and BDNF suggests implications of these processes in neuronal and
developmental diseases.
Indeed, dysregulation in mRNA localization and local protein synthesis in neurons has
been observed in several neurological diseases. For example, spinal muscular atrophy (SMA) is a
degenerative disease that results in motor neuron death and muscle atrophy. It is caused by
mutations in survival of motor neuron1 protein (SMN1). The observation of SMN1 localization
to RNA granules in axons suggested a possible role in mRNA transport. In fact, SMN has been
20
shown to interact with axonal RNA binding proteins, such as hnRNP-R (Mourelatos et al., 2001).
In addition, SMN knockdown in cultured mouse motor neurons has been shown to result in a
reduction of axonal beta-actin mRNA levels, which was correlated with stunted axonal growth
and decreased size of growth cones, recapitulating SMN1-deficiency phenotypes (Glinka et al.,
2010).
Loss of fragile X mental retardation protein (FMRP) causes fragile X syndrome, a
developmental disorder associated with intellectual, behavioral and physical delays, and the most
common inherited form of cognitive deficiency (Bagni et al., 2012). FMRP is known to have
dual functions in regulating dendritic mRNA transport and local protein translation in an
activity-dependent manner (Besse & Ephrussi, 2008). Many FMRP targets are mRNAs encoding
synaptic proteins that include receptors, signaling molecules and cytoskeletal proteins such as
MAP1b and PSD95, in addition to its own mRNA (Bassell & Warren, 2008; Martin & Ephrussi,
2010; S. A. Swanger & Bassell, 2013). Metabotrobic glutamate receptor (mGluR) signaling is a
major regulator of FMRP-mediated local protein synthesis, and mGluR activation induces the
transport of FMRP into dendrites, where it associates with polyribosomes and inhibits protein
translation (Martin & Ephrussi, 2010).
The local translation of mRNAs in neurons is therefore a fundamental mechanism by
which gene expression can be spatially and temporally regulated. Detecting the location and time
of protein synthesis with high resolution can provide insight into the correlation between
localized protein production and cellular changes.
1.4 Quantification of gene expression and protein levels
The plethora of cellular mechanisms regulating mRNA and protein levels presents
unambiguous evidence of the importance of the careful balance of protein activity. The central
21
dogma of biology states that proteins are translated from mRNA molecules that get transcribed
from DNA. Biological identity and function are the result of complex interplay between the four
fundamental processes involved in gene expression: transcription, mRNA degradation, protein
translation and degradation, and their independent interactions with the environment. Each step
is governed by distinct regulatory mechanisms that ensure a careful balance of mRNA and
protein levels is present to properly meet constantly changing cellular needs. For example,
dendritic arbor complexity of Drosophila larval body wall neurons is dependent on the careful
regulation of the level of the transcription factor Cut, that activates genes involved in the growth
and stabilization of neuronal branches. The more Cut protein a particular neuron expresses, the
more complex its dendritic arbor will be (Grueber et al., 2003; Lo, Kays et al., 2015). Similarly,
cyclic changes in the concentration of mRNA and proteins of genes such as period, clock and tim
in the Drosophila lateral neurons drive circadian rhythms required for essential behaviors such as
locomotor activity and eclosion during metamorphosis (Benito et al., 2007). In human disease,
the quantification of mRNA and protein levels has helped develop clinical and diagnostic
markers that routinely aid the discovery of early stages of diseases such as in cancer (Kishikawa
et al., 2015; Krishna Prasad et al., 2013). For example, many human cancers are characterized by
the upregulation in expression of known oncogenes such as myc and akt2 (Prelich, 2012) and
downregulation of tumour suppressor genes such as p53 and PTEN in tumour cells (Shain &
Pollack, 2013). Dysregulation of protein levels is also a hallmark feature of many
neurodegenerative diseases as described in the previous section.
Copy numbers of mRNAs and proteins normally fluctuate over time and vary from cell to
cell, mostly as a response to various environmental and intracellular cues, and in part due to
stochastic molecular events during gene expression. In addition, post-translational mechanisms
22
can upregulate or downregulate protein activity, which in turn can alter the expression of many
other genes. Therefore, to understand the influence of genes on cellular phenotypes, it is
important to understand how different levels of gene products affect normal and abnormal
cellular states, which requires the development of specific tools for quantifying gene expression.
These gene expression assay tools can be broken down into two main categories: those that
quantify changes in mRNA levels and those that quantify protein levels.
1.4.1 Quantification of mRNA levels
Until recently, it was common practice to use mRNA levels as proxy measurements for
protein levels, primarily because of difficulties in estimating protein abundances on a large scale
(Greenbaum et al., 2003; Maier et al., 2009). Classical hybridization-based methods such as the
Northern blot, which use RNA probes complementary to target sequences, allowed the steady
state detection and quantification of select mRNA transcripts. More recent approaches use arrays
of thousands of RNA or cDNA probes complimentary to target mRNAs, which allows the
unbiased detection of mRNAs present in the sample. However, probe-based hybridization assays
require the a priori requirement to know the sequence of the targeted mRNAs to design the
hybridization probes, which hinders the detection of new transcripts. In addition, probe-based
assayed suffer from cross-hybridization artifacts that result from non-specific sequence
homology between unrelated transcripts, which can result in detection inaccuracies. More recent
RNA sequencing technologies assay all RNA species in the sample in an unbiased approach,
enabling the detection and quantification of previously unidentified transcripts.
With the advent of PCR in the early 1980s came unprecedented sensitivity and efficiency
in the exponential amplification of targets (Mullis et al., 1986; Saiki et al., 1985). The
development of real-time PCR (qPCR) in 1993 represented a milestone in PCR becoming a
23
quantitative assay (Higuchi et al., 1993). The addition of a reverse-transcription (RT) step prior
to qPCR produced the RT-qPCR assay which provided the ability to detect and amplify mRNA
transcripts. The amplification step allowed detection of relatively low-abundance mRNA
transcripts, which overcame sensitivity issues faced with traditional hybridization-based methods
such as the Northern blot (Alwine et al., 1977). RT-qPCR now represents a workhorse of mRNA
expression studies due to its high sensitivity, high reproducibility, broad dynamic range and ease
of use (S. A. Bustin, 2000; Ding & Cantor, 2004).
Measuring mRNA levels has come a long way (Kavanagh & Baker, 2009; Ozsolak &
Milos, 2011; Sage et al., 2015; Vogel & Marcotte, 2012). However, it should be kept in mind
that mRNA levels only reflect which genes have been transcribed, and give no insight into the
dynamics of the proteome. Therefore, the interpretation of mRNA levels as a proxy measure for
protein levels is purely correlative. This suggests that directly measuring protein levels instead is
an approach that is likely to yield data that can be interpreted as causative of observed cellular
changes.
1.4.2 Quantification of protein levels
Numerous analytical assays have been developed to measure protein levels. They can be
broken down into a few main categories: dye-based and colorimetric absorbance measurements
for the quantification of bulk levels of protein within a solution, antibody-based approaches for
the detection of single proteins and mass spectrometry approaches for the unbiased detection and
quantification of any protein species within a sample.
Bulk protein measurements using colorimetric assays are very sensitive assays that can
detect down to 20 nanograms of protein (Fowler, 1996). However, they are extremely sensitive
to reaction-quenching substances that commonly contaminate protein sample preparations, which
24
can result in erroneous quantification of protein levels. In bulk assays the protein source is often
the total protein fraction from populations of cells or dissected tissue, which inevitably includes
unwanted and unpredictable cell types, resulting in heterogeneity and poor cellular resolution.
The detection of proteins using antibodies is perhaps the most common approach used by
research and clinical labs. Techniques such as the quantitative immunoblot and the enzyme-
linked immunosorbent assay require the use of an antibody specific to the protein of interest,
which is then detected using chemical reactions or fluorescence to provide a “snapshot in time”
of the level of the protein of interest. Crucial to the success of these techniques is the availability
and quality of the antibody used, which often determines how consistent and accurate the
observed results are. A major disadvantage that prevents the use of these tools in vivo is that they
are inherently invasive, requiring the lysis of cells to release their protein content. Moreover,
assays requiring multi-antibody incubation, or high resolution separation of proteins, are time-
consuming, and can often require up to three days from start to finish (Baker, 2015). These
issues are complicated when using polyclonal antibodies, which display high variability in their
binding, and poor signal to noise ratios during quantification (Marx, 2013). Furthermore, the
phosphorylation state of the target protein, access of the antibody to the protein epitope as well
as inherent variability in the affinity and avidity of antibodies, often contribute to issues of poor
reproducibility (Baker, 2015).
In this thesis, I use genetically encoded fluorescent proteins to track protein production in
living cells. Similarly, genetically encoded fluorescent protein fusions have enabled the direct
observation of intracellular proteins, circumventing the need for secondary treatments or agents.
The fusion of a fluorescent protein to a protein of interest is routinely used to track the
production, trafficking and localization of the fused protein of interest. As discussed previously,
25
critical to the function of any protein is its 3D conformation which is dependent on the careful
folding of the amino acid sequence. Therefore, the fusion of fluorescent proteins, even via
peptide linkers, can be expected to interfere to some extent with protein function. For example,
fused fluorescent proteins have been shown to interfere with protein function via steric
hindrance, or by changing the fused protein’s half-life (Snapp, 2005). In addition, tracking
secreted proteins, low abundance proteins or proteins with punctate distributions can be difficult,
as non-geometric and open cellular spaces require relatively large local concentrations of
fluorescent proteins for them to be observed with standard fluorescence microscopy methods. A
typical excited FP can emit hundreds to thousands of photons before entering its dark state
(Kubitscheck et al., 2000). However this transition is a random event with a fixed half-life,
indicating that some fluorophores will enter the dark state before enough photons are captured by
the detector. Therefore, assuming the brightness of the fluorophore is constant, reliable detection
above noise levels requires either a larger concentration of single fluorescent proteins or more
sensitive detectors in order to overcome shot noise of photon movement and electronic noise in
the detectors when operating at low detection ranges (Bagshaw & Cherny, 2006). Nevertheless,
it is noteworthy that advances in microscopy and detector technologies have enabled the
detection of single fluorescent molecules (G.-W. Li & Xie, 2011; Pinaud & Dahan, 2011;
Tatavarty et al., 2012), however access to such equipment by common research labs is yet to
become mainstream.
1.4.3 mRNA levels as proxy for protein abundance
Despite significant improvements in technologies used to quantify proteins, detection and
measurement of protein production from cells is still laborious and invasive (Walker, 2009). As a
result, researchers have turned to correlations between mRNA and protein levels to estimate
26
protein abundances from quantitative mRNA data, which are easier to collect (Greenbaum et al.,
2003). However, combined analysis of proteomic and transcriptomic data from populations of
cells has recently revealed an unexpectedly poor correlation between mRNA and protein levels,
hovering around 40% explanatory power (Greenbaum et al., 2003; Maier et al., 2009;
Schwanhäusser et al., 2011).
Quantitative mRNA analyses have and continue to demonstrate their value and
usefulness in biological discovery. However, the poor correlation between mRNA and protein
levels suggests that understanding cellular phenotypes using transcript levels as a proxy for
protein abundances is inaccurate, particularly at the single cell level.
1.5 Fluorescence-based single cell resolution protein quantification
Most of the common tools used to quantify the levels of mRNA and proteins lack single
cell resolution. High detection limits, limited sensitivity and low signal-to-noise ratios force
researchers to increase the amount of starting material used in any assay. To overcome these
problems, populations of cells, tissues or whole animals are routinely used to prepare mRNA and
protein samples and consequently the collected data are averaged across thousands of individual
cells.
Individual cells are inherently heterogeneous entities; for example, neighboring neurons
and glia in the brain have been shown to express as little as 65% of the same genes (Schubert,
2011). Moreover, immune cells that are commonly grouped by surface expression of markers
such as CD1 and CD4, have also been found to express entirely different sets of genes and can
have widely varying responses to vaccines and therapy (Flatz et al., 2011). This means assays
performed on large numbers of cells result in population-averaged measurements that miss
important individual cellular differences. Therefore results obtained from a population sample
27
cannot be translated into the single-cell scale (Li & Xie, 2011). In addition, the fact that proteins
are what exert change within a cell, suggests we should be directly examining the production of
proteins and their interactions to understand the molecular processes that govern cellular
function. Assaying individual cells overcomes issues of cellular heterogeneity, and directly
assaying protein levels overcomes the significant weakness in mRNA abundance prediction
power. It is therefore intuitive and logical to conclude that the tools most likely to yield the most
informative results would be tools that directly assay protein production from single cells.
1.5.1 Quantification of protein levels in single living cells
I have recently contributed to the development of a technique to quantify protein
production from single cells in vivo (Lo, Kays et al., 2015). This Protein Quantification Ratioing
(PQR) technique uses a genetic tag that produces a stoichiometric ratio of a fluorescent protein
reporter and the protein of interest during protein translation (Figure 1.2a). The fluorescence
intensity is proportional to the number of molecules of protein of interest produced and is thus
used to determine the level of protein production within the cell (Figure 1.2b). Some RNA
viruses have sequences encoding for short polypeptides ~20 residues in length called CHYSEL
polypeptides (also known as “2A” and “2A-like” peptides, collectively) (de Felipe et al., 2006).
During protein translation, the interaction of the nascent CHYSEL peptide with residues in the
ribosome exit tunnel causes a conformational change which restricts peptide bond formation
between the last two residues of the CHYSEL peptide. This does not terminate translation, but
instead causes the ribosome to autonomously skip and continue translation starting from the last
CHYSEL residue, effectively resulting in the production of two separate functional proteins
(Figure 1.2a).
28
This unique mechanism can be exploited to allow multiple proteins to be produced from a
single polycistronic RNA strand to produce stoichiometric amounts of upstream and downstream
proteins, and this has been demonstrated in vitro and in vivo using high-throughput analyses (J.
H. Kim et al., 2011; Radcliffe & Mitrophanous, 2004; Szymczak et al., 2004; Szymczak-
Workman et al., 2012). We modified and tested different CHYSEL peptides for efficient and
stoichiometric separation of the upstream and downstream proteins and identified different
optimized sequences for use in mammalian and insect systems; we have collectively called these
DNA constructs Protein Quantification Reporters (PQRs). To demonstrate that PQRs can be
used to accurately and reliably quantify protein production, we validated the stoichiometric ratio
and linear relationship between different genes at the single cell level using quantitative imaging
and electrophysiology and found that PQRs allow for equimolar separation between different
proteins and this can be used to correlate changes in protein production with cellular phenotypes
in living cells (Figure 1.2b).
The bicistronic co-expression of a fluorescent reporter with a protein of interest using
PQR for tracking protein production has enormous advantages. Compared to internal ribosomal
entry sites (Pelletier & Sonenberg, 1988), the small size (20-30 amino acids) and low complexity
of the PQR peptide sequence allow it to be easily cloned anywhere along a DNA construct, and
this minimizes any extraneous amino acids that are fused to the proteins of interest. Its self-
processing mechanism is autonomous and insensitive to variabilities of cofactors, enzymes or
global differences in cellular states (J. H. Kim et al., 2011; Radcliffe & Mitrophanous, 2004).
Perhaps most importantly is the fact that the well-characterized and stoichiometric expression of
proteins separated by PQR-type linkers allows consistent and high level expression of proteins
from multicistronic constructs. In addition, since the signal is genetically encoded and is
29
fluorescence-based, quantification of protein production can be done non-invasively over time,
in a rapid and straightforward manner. Moreover, the fluorescence intensity of the reporter is
directly proportional to the level of production of the protein of interest over a wide dynamic
range of quantification and thus the fluorescence output of a cell is a measure of the production
of the protein of interest in that cell (Figure 1.2b). The fluorescence imaging of fluorescent
reporters such as GFP can be done at single cell resolution with very high sensitivity, using
standard fluorescence microscopy techniques which allows the detection of low abundance
proteins and small differences in cellular protein production rates with high spatial and temporal
resolution. Quantification of protein production using PQR reporters is also an approach that is
robust and insensitive to agents that routinely interfere with common protein assays and this
facilitates the observation of protein production over time in single living cells.
1.5.2 Protein production reporters must be carefully chosen
Wild-type GFP tends to misfold and aggregate when expressed in cells, which limits
many of its applications including its use as a reporter of gene expression. To address these
issues, GFP has been subjected to a series of mutations aimed at targeting residues that directly
result in changes in folding and maturation properties. The mutation of a phenylalanine at
residue 64 to a leucine (F64L) introduced EGFP, a more stable protein at 37°C which tends to
aggregate less. A serine to threonine mutation at residue 65 shifted the spectrum of GFP and
resulted in a single excitation peak at 484nm and an emission peak at 507nm (Heim et al., 1995).
This mutation made the use of GFP in living systems more amenable as blue light wavelengths
instead of the more damaging ultraviolet wavelengths could now be used to visualize the protein.
Early efforts by the lab of Willem Stemmer attempted to improve the brightness of native GFP
expressed in bacteria using DNA shuffling, a technique that utilized recombination of in vitro
30
homologous sequences, which they pioneered for the screening and propagation of beneficial
mutations (Crameri et al., 1996). Three cycles of DNA shuffling resulted in a variant known as
“GFP cycle 3” which had a 45-fold increase in fluorescence compared to the standard
commercial GFP at the time (Crameri et al., 1996). This improvement was thought to be most
likely due to a reduction in protein aggregation and improved solubility as a result of several
hydrophobic residues being mutated to hydrophilic residues in GFP cycle 3. However, the
folding and maturation rates were not changed (~95 minutes in bacteria). Using the same
technique, a fast folding GFP bearing six additional mutations was identified and termed
superfolder GFP. Superfolder GFP is one of the most commonly used variants due to its high
folding efficiency, fast folding rate and high denaturant resistance (Pédelacq et al., 2006).
In contrast, many efforts attempted to destabilize GFP, and instead accelerate its turnover
rate to produce variants with half-lives as short as 2 hours (Li et al., 1998). The rapid turnover of
destabilized GFPs has allowed its use in studies requiring reporters with especially short half-
lives, such as in circadian rhythm (Hastings, 2005) or gene expression studies (Li et al., 1998). In
addition, the rapid turnover results in less accumulation of GFP in cells which leads to lower
toxicity at high levels (Li et al., 1998), and decreased fluorescence intensity saturation during
fluorescence imaging enabling longer cellular expression times and longer imaging durations.
1.5.3 Approaches to visualizing locally translated proteins
The detection of new protein synthesis in localized cellular compartments provides
powerful spatial and temporal resolution that allows the correlation of protein production and
development of cellular phenotypes. Genetically encoded fluorescent proteins such as GFP can
be used in fusion constructs, similar to Figure 1.2a, and the development of green fluorescence in
subcellular compartments is used as a marker of new protein translation. This seemingly
31
unambiguous approach, however, raises a critical concern for the need to verify that the observed
new green signal is truly made locally instead of being produced at the soma, and then
transported to the local site. The majority of fluorescent proteins require on average several tens
of minutes from the moment they are translated for proper folding, maturation and for the
eventual fluorescence signal to develop (Shaner et al., 2005). During this time, the non-
fluorescent protein molecule can freely diffuse or be transported within the cell. In addition,
increased ribosomal translation rates at perinuclear regions can result in saturation of the
detection range from somatically translated GFP signals, impeding the detection of locally
synthesized signals (S. Kim et al., 2010).
In this thesis, I am interested in developing tools that define the location, time and rate of
translation of proteins, with the goal to examine local synthesis of proteins in neuronal
subcellular compartments. In the following sections I describe how the discovery of mRNAs in
neuronal subcellular compartments led to efforts to visualize local protein synthesis in neurons.
1.5.3a Identification of localized mRNAs in neurons
The effort to develop tools to detect newly synthesized proteins in neurons started with
the discovery that mRNAs were found to localize to discrete dendritic and axonal compartments
(Ochs et al., 1969; Tobias & Koenig, 1975). Such findings provided an attractive model to
answer how long lasting changes in structure and function of subcellular compartments such as
growth cones and dendrites can persist and be spatially restricted in highly polarized cells such
as neurons. However, this also posed a problem as it raised the need to restrict the study of
mRNA localization and protein translation specifically in those compartments.
Early experiments aimed at separating neuronal processes from cell somas, either
biochemically using fractionation (Krichevsky & Kosik, 2001) or by a variety of physical ways
32
to isolate processes and somas (Kim et al., 2010) and then profiling the RNA species present in
the sample. In such assays, the purity of the sample is critical as the amount of mRNA transcript
present in neuronal processes is orders of magnitude less than in cell bodies. Any somatic
contamination can therefore overwhelm the ability to detect process-localized transcripts. Once
separated from lysate, the RNA is profiled either by T7 RNA amplification (J Eberwine et al.,
2001; Miyashiro et al., 1994), hybridization to known probes (Eberwine et al., 2001) or by
quantitative real time PCR amplification of transcripts (Zheng et al., 2001). Such approaches
have led to the identification of over 400 dendritically localized transcripts and over 150
synaptically-enriched transcripts such as MAP2 and αCaMKII (Eberwine et al., 2001; Tian et al.,
1999). Similar methods were used to isolate axonally-localized mRNAs, such as CREB and
RhoA mRNA from rat sensory neurons (Cox et al., 2008; Zheng et al., 2001). Process-localized
transcripts can then be used as starting material for the generation of cDNA libraries of locally
translated genes (Moccia et al., 2003). These assays approach the problem in an unbiased
manner, however they required rigorous controls to rule out the contamination of somatically-
derived transcripts. For example, in a study examining the contribution of contaminating somatic
mRNAs in process-localized mRNA preparations, the presence of false-positive hits due to
somatic contamination was found to be minimal (S. Kim et al., 2010; Poon et al., 2006).
However, surprisingly little overlap was seen when comparing dendritically localized rodent
hippocampal mRNAs identified using independent approaches. Such results raise a crucial need
to independently confirm the localization of identified transcripts using different approaches (S.
Kim et al., 2010). One way to confirm the localization of mRNA transcripts identified using
amplification is in situ hybridization (ISH) which uses labeled RNA or DNA probes that
hybridize to specific mRNAs, which can then be imaged directly in the case of fluorescent ISH
33
(FISH) or revealed using secondary agents such as enzymes (van de Corput et al., 1998) or
antibodies (Poon et al., 2006). FISH approaches have been used to identify the localization of
transcripts with great success in a variety of tissues such as cultured hippocampal neurons (Lyles
et al., 2006), whole-mount Aplysia CNS (Dan Ohtan Wang et al., 2009) and sections of Aplysia
ganglia (Poon et al., 2006). However, ISH only provides a snapshot in time of mRNA
localization and thus cannot provide much information into the dynamics of mRNA localization
or translation of new protein. Therefore, new methods for the dynamic imaging of mRNA
translation are needed.
One important limitation of visualization methods is that they provide information that is
correlative in nature. Although these methods fill a critical gap in our understanding of activity-
dependent and neuromodulatory control of local translation, they do not reveal whether the
presence of mRNA transcripts contributes to active local protein translation. The presence of an
mRNA at dendrites does not necessarily indicate it is actively being translated (Na et al., 2016).
However, the presence of an mRNA at distal sites does suggest it is a process-localized mRNA
used to provide a rapid local supply of its protein product. Assays that instead directly monitor
protein production from localized mRNAs can determine the moment a protein is made with
higher spatial and temporal resolution than visualization methods. This enables correlating local
cellular changes with the moment a protein is produced locally. Below, I describe alternative
approaches that have been applied specifically with this goal in mind.
1.5.3b Direct detection of local protein synthesis using photoconvertible reporters
A significant portion of our understanding of local protein synthesis has been the result of
using genetically encoded fluorescent protein reporters such as GFP, RFP and Dendra2 (Kim et
al., 2010; Martin & Ephrussi, 2010). Genetically encoded fluorescent proteins fused to proteins
34
of interest, or placed between the 5’ and 3’ untranslated regions of a target mRNA result in the
production of new fluorescent signals each time the mRNA of interest is translated. Genetically
encoded fluorescent proteins eliminate the need to use exogenous detection agents, are
minimally toxic to cellular health and emit fluorescence with high signal to noise ratios. This
enables the non-invasive long term imaging of living tissues. In neurons, the most successful
approaches took advantage of an interesting class of fluorescent proteins known as
photoconvertible fluorescent proteins.
Kaede and Dendra2 are examples of photoconvertible fluorescent proteins that normally
emit green fluorescence, but can be permanently photoconverted to emit red fluorescence by
ultra-violet range (350-400nm) illumination. The photoconversion property of Kaede originates
from the Histidine 62-Tyrosine 63-Glycine 64 tripeptide motif that forms a green fluorescence
emitting chromophore analogous to that in GFP. Upon exposure to UV radiation, a cleavage of
the amide bond at the histidine 62 residue results in the formation of a new spectrally distinct red
chromophore (Ando et al., 2002; Dittrich et al., 2005).
The permanent photoconversion of fluorescent proteins is an interesting property that can
be exploited to examine the production of new proteins by simply photoconverting all
preexisting protein from green to red and then monitoring new translation as the development of
new green signal. While this approach provides several advantages over using standard
fluorescent proteins, the issue of whether the protein is truly locally synthesized or has diffused
into the site in its immature nonfluorescent state, is not addressed with photoconvertible
fluorescent proteins. These drawbacks must be considered and accounted for when using
genetically encoded fluorescent proteins for examining local protein translation events.
1.5.4 Requirements for a local protein synthesis reporter
35
An ideal local protein synthesis reporter must meet certain performance criteria for it to
be able to accurately and reliably detect local protein translation events in vivo. For instance,
introducing the reporter into the animal, and the collection of data should be as minimally
invasive as possible. Non-invasive approaches enable long-term and repeated tracking of
changes in a single cell within a living animal, effectively allowing the examination of cellular
function in its natural, dynamic and unperturbed context. Local protein synthesis reporters must
also have high temporal resolution with a signal that develops immediately after the moment of
protein translation to overcome issues of maturation time and diffusion, faced by most standard
fluorescent proteins. This signal will ideally also be quantifiable with high spatial resolution, to
enable the quantitative comparison of protein production levels in different subcellular
compartments within the same cell or between different cells.
1.5.5 Split GFPs are indicators of protein interaction
Bimolecular fluorescence complementation using GFP has emerged as a simple solution
and has been repeatedly used in protein and other macromolecular interaction studies (Kerppola,
2006). Split GFP is a GFP molecule that has been split into two nonfluorescent polypeptides that
compose the entire sequence of GFP. By fusing both halves of GFP to interacting proteins, the
close proximity and ensuing physical interaction of the partners results in the spontaneous
reassembly and emission of fluorescence from the reconstituted GFP molecule. However, despite
its widespread use for in vivo imaging, little is known about the mechanism of reconstitution
(Kent et al., 2008).
The most popular split GFP partners are GFP1-10 and GFP11, resulting from a separation
of the peptide chain between beta barrels 10 and 11. The GFP reconstitution across synaptic
partners (GRASP) techniques uses the fusion of the split GFP fragments to synaptic membranes
36
in distinct neuronal populations (Kim et al., 2011). If two neurons from these populations, each
expressing one of the fragments, interact and form a synapse, then the GFP molecule can
reconstitute and emit fluorescence. The green fluorescent signal thus marks the location synapses
along the cell within tissues. GRASP has been used to map cell contacts and neuronal
connections in Drosophila (Gordon & Scott, 2009) and C. elegans (Feinberg et al., 2008).
37
1.6 Figures
Figure 1.1 Brightfield image of dissected whole ovarioles
Representative image of dissected ovarioles containing stage 5-6 and 9-10B oocytes. During
oogenesis, a cluster of 16 germ cells interconnected by intracellular ring canal bridges, form the
germline cyst. Fifteen of these cells will become nurse cells and the remaining germ cell
becomes the oocyte (outlined). After fertilization, the oocyte chamber (outlined) grows in size to
support the development of the embryo.
38
(Lo, Kays et al., 2015)
39
Figure 1.2 PQR reporters allow quantification of protein production from cells.
(a) Insertion of a Protein Quantification Reporter (PQR) between a fluorescent reporter (GFP)
and a gene of interest creates a polycistronic mRNA for co-transcription and co-translation of
GFP and the gene of interest. The PQR construct allows for one molecule of GFP to be
synthesized for every one protein of interest synthesized. Because of the fluorescence output of
GFP is directly proportional to the concentration of GFP, then the fluorescence intensity of the
cell is used to quantify the level of production of the protein of interest. (b) Because the
fluorescence output of GFP is directly proportional to its concentrations (top panel), then by
using a protein Quantification Reporter sequence to produce a stoichiometric ratio between GFP
and the protein of interest (middle panel), the fluorescence intensity of GFP can be used as a
measure of the production of the protein of interest (bottom panel) (Lo, Kays et al., 2015).
40
1.7 Thesis introduction
The aim of this dissertation is to develop techniques and reagents that will be used to
explore the process of protein translation in vivo. Since proteins are the last molecular effectors
of change in a cell, and are responsible for the multitude of phenotypes we observe in cells, I am
particularly interested in knowing when, where and how much protein is being produced in the
context of a living animal. Expanding the molecular toolbox by which we can observe cellular
processes is key for our understanding of basic cellular processes. In Chapter 2, I demonstrate
combined protein and RNA measurement from single cells, in a system I have developed to
examine gene expression changes and how they relate to protein levels. In Chapter 3, I discuss
the development and validation of a novel technique to mark protein translation events at
subcellular resolution in living cells. Chapter 4 explores the applications of our technique to
quantifying protein translation events, and where I describe key proof of principle experiments
and reagents required for imaging protein translation in single cells in vivo. The results of this
work contribute to the tools we currently use examine protein production, with specific
advantages in spatial and temporal resolution and quantification accuracy.
41
Chapter II - Quantification of mRNA and protein
levels in single cells
2.1 Relation to overall thesis
Cells are inherently heterogeneous, and the current tools we use to quantify protein levels
miss important individual cellular differences that are key to understanding cell and molecular
biology. In addition, mRNA is routinely used as a proxy measurement for protein levels,
although this has repeatedly been demonstrated to be inaccurate. Assaying mRNA and protein
expression from single cells can therefore bring a new layer of information on the regulation of
protein expression. To address these issues, I have developed a system to combine mRNA and
protein quantification from the same cell. Single cells edited to co-produce PQR reporters each
time an endogenous protein of interest is translated are first imaged to determine the level of
translation of the protein of interest. The same cell is then lysed and the absolute mRNA levels
encoding the endogenous protein of interest are quantified. In this approach, mRNA and protein
production can be quantified simultaneously for multiple genes, providing valuable metrics that
can be used to: 1) Understand the regulation in expression of either the mRNA or protein of a
particular gene. 2) Correlate changes in mRNA levels with changes in protein levels of a gene. 3)
Reveal insight into the coregulation of multiple genes within the same cell, which can be used to
establish links between the transcriptional and translational regulation of gene products. This
chapter presents an extension of the PQR technique as part of my overall goal to develop
techniques to understand protein production at single cell resolution, by looking at concurrent
changes in mRNA production.
42
2.2 Introduction
Understanding the relationship between genes and phenotypes is a central component of
biology. Profiling the expression of genes with respect to messenger RNA (mRNA) or protein
products has advanced our fundamental understanding of cell biology, such as the maintenance
of cell structure (Gronowicz et al., 1992), progression of cell cycle (Ly et al., 2014), neural
development (Cáceres & Nilson, 2005), as well as how abnormalities in gene expression can lead
to disease states (Wong et al., 2001). Several analytic methods have been developed to profile
gene expression at the level of mRNA or protein. For example, microarrays and deep sequencing
technologies can identify mRNA from tens of thousands of different genes, whereas the
expression and abundance of proteins can be assayed using spectrometry, chromatography and
immunoassays (Greenbaum et al., 2003; Maier et al., 2009).
The current tools to assay mRNA and protein expression are generally performed on
large pools of cells or tissues (Wu & Singh, 2012a). Large numbers of cells increase the
reliability of results and overcome issues of low sensitivity and low signal to noise ratio of
common assays. However, it is becoming increasingly clear that even neighboring cells within
tissues or pools of clonal cells are not homogenous (Schubert, 2011). Cellular dynamics, such as
differing transcriptional and translational states, cell signaling, cell cycle, development, and other
molecular processes in addition to stochasticity together produce cellular heterogeneity, even in
immortalized clonal cell lines (Stockholm et al., 2007). Therefore, assays performed on large
numbers of cells cannot be translated into the single-cell scale (Schubert, 2011; Wu & Singh,
2012b). As described in Chapter 1, individual cells are inherently heterogeneous, therefore
analysis of single cells is required to identify molecular mechanisms that are masked by the
population average.
43
The traditional and more convenient quantification of mRNA levels as a proxy for
protein abundance implicitly assumes that changes in mRNA levels directly correspond to
changes in protein levels, and thus has biological relevance. This has repeatedly been shown to
be inaccurate, as more global analyses of transcriptomics and proteomics are revealing
surprisingly poor correlations in mRNA-protein level concordance (Greenbaum et al., 2003;
Gunawardana et al., 2015; Maier et al., 2009; Schwanhäusser et al., 2011). These findings create
a concern when making inferences only from mRNA data. The utility of mRNA expression
studies is clear, particularly for understanding mechanisms of transcription and in the continued
development of sensitive preclinical disease markers (Abrahamsen et al., 2004; Mikaelian et al.,
2013). However, the interpretation of quantitative mRNA results with respect to cellular
phenotypes is merely correlative. Therefore, protein expression directly needs to be examined to
understand causative change within cells. Taken together, this suggests that assaying mRNA and
protein expression together from a single cell overcomes cellular heterogeneity issues, and
protein data can be used to validate mRNA findings. In addition, the availability of mRNA data
from the same cell adds another level at which the regulation and correlation of mRNAs and
proteins can be examined. Examining these dynamics allows understanding of how changes in
mRNA levels contribute to the level of protein that ultimately generates cellular phenotypes.
We have previously developed a technique that allows for quantification of protein levels
in single cells in vivo (Lo, Kays et al., 2015). This Protein Quantification Ratioing (PQR)
technique uses a genetic tag that produces a stoichiometric ratio of a fluorescent protein reporter
and the protein of interest during protein translation. The fluorescence intensity is proportional
to the number of molecules of protein of interest produced and is used to determine the relative
protein amount within the cell (Figure 1.1a). Using genome editing tools, PQR constructs can be
44
inserted into any endogenous genomic locus to quantify endogenous protein levels in single
living cells (Lo, Kays et al., 2015).
This chapter describes a fast and accessible way for researchers to quantify the
expression of a gene of interest by measuring both its absolute mRNA transcript numbers and
protein production levels in the same cell (Figure 2.1). This tool can be used by any researcher
with access to a standard fluorescence microscope and a real-time PCR thermocycler and can
easily be applied to assay single cells obtained from various sources such as dissociated animal
and tissue samples, as well as FACS sorted and patient-derived stem cells. In addition, this
protocol is amenable to various cell types such as neurons, immune cells, immortalized cell lines
and dissociated animal or plant cells, if they are amenable to common genetic and biochemical
assays. By first measuring the level of protein translation using a PQR fluorescent reporter and
then quantifying the absolute mRNA copy number of that gene using RT-qPCR (Nolan et al.,
2006), the transcriptional and translational expression of a gene can be accurately determined at
the single cell level (Figure 2.1). This approach can be applied to examine the expression of
multiple genes at once, which results in an overarching view of the regulation of transcript and
protein expression of multiple genes in the same cell (Figure 2.2).
2.3 Experimental design and detailed protocol
This protocol describes a system for quantification of gene expression at the mRNA and
protein level from a single cell. We chose the mouse monoclonal antibody cell line 22c10 as a
source of single cells (Fujita et al., 1982). 22c10 hybridoma cells are large, easy to manipulate
and secrete high levels of antibody containing kappa-isotype immunoglobulin light chains (IgK).
Antibodies are composed of two heavy chains and two light chains. The genomic locus encoding
the kappa light chain contains a constant region that is not affected by genomic rearrangements
45
that occur following immune responses, and is thus common to all light chains produced from
the IgK locus (Janeway et al., 2001; Lodish et al., 2008). The insertion of a PQR-RFP reporter
within the constant region of the kappa light chain ensures the production of a molecule of RFP
with each kappa light chain translated, at high levels (Figure 2.3a). In contrast, we chose the
Rpl13a gene which encodes for Ribosomal Protein L13A and is stably expressed in all
eukaryotic cells at consistent levels. Due to its consistent expression across tissues, Rpl13a is
routinely used as a reference “housekeeping” gene for normalization of quantitative mRNA and
protein measurements (Curtis et al., 2010; Gubern et al., 2009; Mane et al., 2008). The
quantification of Rpl13a, by insertion of a PQR-GFPnols reporter at the end of its coding
sequence (Figure 2.3b), can therefore be used as a measure of a cell’s individual translational
status (Lo, Kays et al., 2015). In the following section, I present the protocol we use to take
advantage of the differences between IgK and Rpl13a expression, to highlight how this technique
can be used to profile the transcriptional and translational landscape of multiple genes
simultaneously at single cell resolution (Figure 2.2).
2.3.1 Materials
Reagents
Caution: Trizol reagent and chloroform are toxic and cause harm to skin and airways. Protect
skin and eyes and handle stock bottles in a fume hood.
• Serum free culture medium, e.g. H-Cell (Wisent, cat. no. 000-035-CL)
• Nuclease free water (Wisent, cat. no. 809-115-CL)
• Nuclease-free Glycoblue (Thermo-Fisher, cat. no. AM9515)
• RNase OUT (40 U/µl) (Thermo-Fisher, cat. no. 10777019)
• Superscript IV Reverse Transcriptase 200 U/µl (Thermo-Fisher, cat. no. 18090050)
• Taqman Fast Advanced 2X Mastermix (Thermo-Fisher, cat. no. 4444558)
46
• Trizol reagent (Thermo-Fisher, cat. no.15596018)
• Chloroform (Sigma, cat. no. 288306)
• Isopropanol (Sigma, cat. no. I9516)
• 75% ethanol prepared with nuclease-free water
• Gene-specific reverse transcription primers (Integrated DNA Technologies)
• Real-time polymerase primers and probes (Integrated DNA Technologies)
Equipment
Caution: Pipetting by mouth can be hazardous. Appropriate tubing filters must be used to
protect against accidental liquid or particulate inhalation. However, in our experience the
microliter volumes handled using a mouth pipette in this protocol are too small to pose any risk
of crossing the filter or the length of tubing, and the aspirated solutions are not generally
harmful. Nevertheless, a 1 ml syringe can be used instead of the mouth pipette, but much less
control over pipetting is achieved.
• Aspirator assembly unit for mouth pipette (Sigma, cat. no. A5177)
• 0.22 µm tubing filter (Millipore, cat. no. SLGP033NB)
• Teflon-coated patterned glass slides, 10-15 mm diameter spots (EMS, cat. no. 63426-06)
• 96-well qPCR microplates (Thermo-Fisher, cat. no.4346907)
• High optical clarity sealing film (Sarstedt, cat. no. 95.1999)
• 8-strip PCR tubes (Diamed, cat. no. DIATEC420-1402)
• Nuclease-free 1.5 mL microfuge tubes (Thermo-Fisher, cat. no.AM12400)
• 1.5 x 1.1 mm O.D/I.D borosilicate glass capillaries without filament (Sutter Instruments, cat.
no. B150-110-10)
• Glass micropipette puller e.g. P80/PC (Sutter Instruments)
• Microscope with excitation source and appropriate fluorescence emission filters (For example
(excitation/emission): GFP (488nm/515nm), RFP(543nm/594nm)
• Imaging camera e.g. Axiocam MRm CCD (Zeiss)
• Real-Time PCR System, e.g. StepOnePlus (Applied Biosciences)
47
2.3.2 Gene editing using CRISPR-Cas9
In order to target fluorescent PQR reporter constructs into endogenous genomic loci for
protein quantification, we used Clustered Regularly Interspaced Short Palindromic Repeats-Cas9
(CRISPR-Cas9) genome editing, which uses an RNA-guided Cas9 endonuclease that can be
targeted to induce double-strand breaks (DSB) in any known DNA sequence with pinpoint
sequence accuracy (Jinek et al., 2012). DNA DSBs can be repaired by the cellular machinery
using the homologous recombination pathway, which can be exploited to insert exogenous DNA
sequences into genomic loci (Mao et al., 2008).
To insert PQR-XFP reporters into the endogenous IgK or Rpl13a locus, we designed and
generated several CRISPR/Cas9 targeting vectors that cut the endogenous loci directly upstream
of the stop codon (Figure 2.3a). Co-transfection of IgK or RPL13A-specific repair plasmids into
22c10 cells results in a DSB, which stimulates the DNA repair machinery to recombine the PQR-
XFP fragment in-frame at the end of the coding sequence of the endogenous locus. When
successful integration of the PQR-XFP construct occurs, mRNAs transcribed from the
endogenous RPL13A or IgK locus will be slightly longer in size (~850 extra nucleotides), but
will preserve the 5’ and 3’ untranslated regions (UTRs) and native coding sequence, which
contain elements required for proper mRNA export, localization, stability and translation (Figure
2.3b). For every one molecule of endogenous RPL13A or IgK protein produced, there will be
one corresponding molecule of fluorescent reporter co-translated. Since the reporter and the
protein of interest are translated from one open reading frame, the fluorescence intensity of the
reporter (i.e. brightness of the cell) can be used as a measure of how much protein is being
produced (Figure 2.1) (Lo, Kays et al., 2015).
48
The repair templates used to repair the genomic DSB consist of a PQR-XFP construct
placed between two 1.0 kB-long homology arms specific to genomic IgK or RPL13A. The
homology arms lack any promoters or transcriptional activators, which prevents expression of
the PQR-XFP until in-frame genomic integration within an active coding gene. The arms align
the repair plasmid to the endogenous locus via molecular homology, and the DNA repair
machinery recombines the exogenous PQR-XFP into the break region, directly upstream of the
endogenous stop codon (Figure 2.3b).
We screened several CRISPR targeting vectors and repair templates for optimal nuclease
and recombination activity. Validation of the correct integration of PQR-GFP into the IgK locus
was achieved by genomic DNA extraction six days post-transfection and genotyping using
primers that lie outside and within the homology arms of the repair template (Figure 2.4a). The
5’ and 3’ ends of the endogenous IgK locus are probed using PCR with these two sets of primers
(Figure 2.4c). The PCR products are then digested using restriction enzymes specific for sites
within the PQR construct. In parallel, the edited genomic locus is sequenced to identify the PQR
and genomic junctions. The same procedure is repeated to validate the insertion of PQR-XFP
into a second locus (Lo, Kays et al., 2015).
To generate 22c10 cells carrying PQR reporters at both the endogenous Rpl13a and IgK
loci, we first established a stable line with a PQR-GFPnols at the Rpl13a locus, and then targeted
the IgK locus with a PQR-RFP reporter. Our preliminary experiments using an IgK-PQR-GFP
edited line (Figure 2.4) showed that the fluorescence intensity of a RFP-based red Rpl13a
reporter was too dim to be observed without a camera. High levels of GFP (from IgK), in
addition to overlap in the spectra of GFP and RFP led to high background fluorescence levels in
the red channel, posing a burden on the manual isolation of fluorescent cells and quantification
49
of fluorescence intensities. We therefore swapped the reporters and chose to include a nucleolar
(nols) targeting sequence to the GFP in order to sequester the GFP to nucleoli in Rpl13a-edited
cells, a cellular organelle 100 times smaller than the cytoplasm in volume (M. S. Scott et al.,
2010). Sequestering fluorescent proteins into small cellular organelles increases their local
concentration which results in a brighter signal, facilitating fluorescence quantification and cell
manipulation. Indeed, we found that using a green reporter for Rpl13a and a red reporter for IgK
produced cells that displayed clear green and red fluorescent signals (Figure 2.1b).
2.3.3 Single-cell protein level quantification
To determine the level of production of IgK or Rpl13a protein within a cell, genome
edited 22c10 cells carrying red and green fluorescent PQR reporters, respectively, were used
(Figure 2.3). Immediately prior to the experiment, a concentrated single cell suspension (> 5x105
cells/ml) was prepared and placed on ice to delay any changes in cell metabolism or gene
expression. Using freshly pulled wide-bore micropipettes (Figure 2.5a), a single red and green
fluorescent cell was pipetted (Figure 2.5c) into a droplet of culture medium on Teflon-coated
glass slide (Figure 2.5d). A high magnification objective was dipped into the drop of culture
medium, forming a meniscus around the lens of the microscope (Figure 2.5d), and the cell was
brought into focus to image the fluorescence intensity of the cell (Figure 2.5e). It is important to
use excitation and acquisition settings that have been previously determined by imaging control
cells to establish the range of minimum and maximum fluorescence levels that can be reasonably
expected from the cell line used to minimize imaging cells that are too dim or those that saturate
acquisition settings.
2.3.4 Total RNA extraction
50
After quantitative fluorescence imaging, the cell was immediately transferred using a new
clean micropipette to a microfuge tube prefilled with 200 µL of the phenolic reagent, Trizol
(Figure 2.5f). It is important to change the micropipette after each sample to avoid cross
contamination by nucleic acids or cells that may have adhered to the inside surface of the
micropipette glass. The cell is immediately lysed once in the phenolic solution and its nucleic
acid and protein content are separated into a top aqueous layer containing nucleic acids and a
bottom organic layer containing protein and cellular debris. One volume of chloroform was then
added to further eliminate contaminating phenol via a second phase separation and dilution of the
phenol. The total RNA dissolved in the aqueous layer was then recovered by precipitation with
isopropanol and subsequent centrifugation at ≥ 12,000g. Glycogen is an inert polysaccharide that
is often used as a carrier to increase RNA recovery by co-precipitating with nucleic acids,
effectively trapping precipitated RNA molecules in large sugar precipitate complexes. Glycoblue
is glycogen that has been conjugated to a blue dye, and addition of 5 µg of Glycoblue prior to
isopropanol improved RNA recovery without compromising subsequent transcription and
amplification steps. The RNA pellet obtained after centrifugation was washed twice with 70%
ethanol that has been prepared with RNase-free water, and then dried at room temperature for ~5
minutes, or until most of the ethanol was evaporated. The pellet was resuspended in 6 µL of
RNase-free water and this total volume was used as input RNA for the subsequent reverse-
transcription (RT) reaction.
2.3.5 Reverse-transcription
Superscript IV was chosen as a reverse transcriptase. It is a modified Moloney murine
leukemia virus (M-MLV) reverse transcriptase, a variant of thermostable reverse transcriptases
(Das et al., 2004). Thermostability allows RT reactions to be carried out at high temperatures at
51
which RNA is linear and secondary structures cannot form. Superscript IV has also been shown
to outperform other commercial enzymes in terms of reproducibility in the presence of inhibitors
commonly present in RNA samples such as carryover phenol, isothiocyanates and ethanol
(Suslov & Steindler, 2005), which pose a concern particularly in this protocol. To produce
complementary DNA (cDNA) from mRNA isolated from single cells, a mixture of gene-specific
primers (final concentration 0.1 µM) was used to specifically reverse transcribe edited IgK and
Rpl13a mRNA transcripts in a 15 µL RT reaction containing 30 units of RNase inhibitor and 150
units of Superscript IV enzyme. Gene-specific RT primers have been shown to be the most
sensitive and specific approach to convert target mRNA into cDNA, and produce less variability
compared to oligo-dT or random hexamer primers (Stephen A Bustin & Nolan, 2004). The
reaction was incubated at 55°C for 50 minutes to allow the RT reaction to proceed.
2.3.6 Real-time polymerase chain reaction
By titrating the amount of input cDNA comparing 1 µL, 3 µL and 6 µL, we found that
using 6 µL of cDNA from the RT reaction is optimal for obtaining high quality and reproducible
amplification without significant interference from the RT reaction mixture (Figure 2.6a). The
rationale was to determine the minimum amount of cDNA that could be used in the qPCR
reaction and the advantages are twofold: First, to reduce the total amount of glycerol, a common
component in RT and qPCR enzyme mixtures that is inhibitory beyond 20-30% v/v (Yasukawa
et al., 2010). Second, to spare as much RT mixture as possible in order to repeat failed
amplifications or to run replicate samples during optimization and control reactions. The
TaqMan Fast Advanced mastermix (ThermoFisher), which contains the AmpliTaq fast DNA
polymerase, was used to amplify the IgK and Rpl13a cDNA molecules. 20 µL reactions were
prepared and loaded into a 96-well optical plate sealed with high-optical clarity sealing film; the
52
plate was briefly centrifuged and loaded into the thermal cycler which has been preprogrammed
to define the cycling parameters and well assignments. Thresholding of amplification and
calculation of cycle threshold (Ct) values were performed automatically using the StepOnePlus
software. Ct values are inversely proportional to starting target concentration, with low Cts being
indicative of high concentrations of input DNA.
2.3.7 Calculation of absolute mRNA transcript number
qPCR standard curves are plots of obtained Ct values versus series of diluted known
standards that are used to convert Ct data into quantity values. Analysis of standard curves
provides a substantial amount of information about the qPCR assay. In addition to providing a
calibrator sample of known amounts for the absolute or relative quantification of mRNA
concentrations from unknown test samples, variability in the reverse-transcription reaction and
general interexperiment variability resulting from running multiple samples on different plates
can be accounted for and corrected (Nolan et al., 2006; Stahlberg & Bengtsson, 2010). The
inclusion of a standard curve with each experiment is an important control to normalize
quantitative measurements and to ensure the accurate comparison of data points.
The standard curve was generated by plotting Ct (y-axis) against the logarithm of
template quantity (x-axis) (Figure 2.7). Ct values obtained from single cell experiments could
then be compared to this curve to determine the quantity of starting nucleic acid material. A
linear dynamic range was normally seen over at least 6 logs of dilutions, which essentially
defines the working range of accurate quantification of the assay (Larionov et al., 2005).
In this single cell assay, standard amounts of DNA covering a range of approximately 10
to 107 copies were prepared to minimize any extrapolation outside of the predicted dynamic
53
range (Figure 2.7) (Ståhlberg & Kubista, 2014; Svec et al., 2013). It is ideal to amplify the serial
dilutions of standards in duplicate to determine the reproducibility of the assay. In addition, the
template used to generate the standard curve should accurately reflect the complexity of
experimental samples, ideally by using a total RNA preparation from the investigated cells or
using a known amount of pure cDNA, plasmid DNA or synthetic oligonucleotides containing the
target of interest (Dhanasekaran et al., 2010). Synthetic oligonucleotides containing the assayed
target region are easy, fast and inexpensive to commercially synthesize. In addition, commercial
synthetic oligonucleotides are usually highly purified samples of known concentration.
2.3.8 Readout of amplification
The qPCR assay uses the TaqMan chemistry principle for the detection of amplified
transcripts. The fluorescent signal originates from a 20bp DNA probe that contains a conjugated
fluorophore and a chemical quencher that hybridizes to the amplified target region (Holland et
al., 1991). Exonuclease activity of the passing DNA polymerase releases the quencher from the
probe. The otherwise quenched fluorophore can now change conformation and emit
fluorescence, which is then detected by the real-time PCR thermal cycler detectors. The
fluorescence intensity of the reaction during each cycle of the reaction is measured by the
machine and an automatic thresholding algorithm detects the cycle threshold (Ct), the cycle at
which the fluorescence intensity and its rate of increase reach a significant level above the noise
(S. Zhao & Fernald, 2005). Cts are inversely proportional to starting DNA concentration (Figure
2.6, Figure 2.7) and thus to convert a Ct value into number of mRNA molecules, the use of a
standard curve is required (Nolan et al., 2006).
54
In addition to the commonly used positive and negative qPCR amplification controls, it is
important to regularly include additional controls to monitor any drift in technical or
experimental variability.
2.3.9 Assay controls
The qPCR assay used must be optimized to produce the most sensitive and reliable
amplification (Tichopad et al., 2009). For the PCR reaction to produce twice as much product
during each cycle, the amplification of product must be as close to 100% efficient as possible, as
this results in the most sensitive and reliable quantification (Taylor et al., 2010). The efficiency
of the assay is most easily determined by using the equation efficiency = 10(-1/slope). The slope is
obtained from standard curves generated by amplifying serial dilutions of target template under
different primer and probe concentrations. The slope of our standard curve was calculated to be -
3.7, resulting in 86% efficiency (Figure 2.7).
Variability in the instrument and detectors, reaction setup, and stability of reagents can
contribute to erroneous quantification. The slope and y-axis of the standard curve can be used as
metrics to monitor and adjust for variability. Therefore, for the highest quantification accuracy, a
standard curve should be prepared and included on the same plate each time experimental
samples are assayed. Moreover, it is important to use the minimum amount of cDNA that
produces the most sensitive and highest quality amplification, to reduce any inhibitory effects
caused by components in the RT reaction mixture (Opel et al., 2010).
The reproducibility of a qPCR assay is reflected by the variability between technical
replicate reactions, as determined by the R2 of the best fit line of standard curve data. R2 values
of 0.98 or higher are indicative of a stable and reproducible assay (Nolan et al., 2006). Our
standard curve resulted in R2=0.99 (p<0.05) (Figure 2.7).
55
This protocol uses phenol-based purification of RNA which biochemically separates
away genomic DNA, however in some cases, genomic DNA may contaminate the obtained RNA
sample. Genomic DNA controls are thus crucial to determine the contribution of contaminating
genomic DNA to the observed amplification (Stephen A. Bustin et al., 2009). Although in the
case of single cells contamination by genomic DNA has minimal effect (2 extra copies of the
target in a diploid cell), some sensitive applications require maximum quantification accuracy.
Since this protocol requires that the entire RNA content of the cell be used in the RT reaction, it
is useful to randomly allocate some samples for genomic DNA contamination controls. The
contribution of contaminating genomic DNA to the observed single cell Ct can be determined by
performing a no-RT control reaction, where the reaction is assembled but the RT enzyme is
omitted (Figure 2.8). Any observed amplification can then be attributed to genomic DNA that
has been carried over during RNA purification. The contribution of genomic DNA from single
cells using this protocol was found to be negligible (in > 90% of randomly tested samples)
demonstrating that genomic DNA can be reliably separated from the aqueous RNA during Trizol
purification (Figures 2.8a-b). It is however useful when first performing this protocol or when
using different cell types to dedicate some samples for performing no-RT controls to determine
the contribution of contaminating genomic DNA. qPCR results are generally acceptable if the
cycle thresholds of genomic DNA control samples are at least 5 Ct larger than those of test
samples (S. Bustin & Nolan, 2004; Nolan et al., 2006). However, we found that most samples
contain no detectable genomic DNA, even when the PCR was programmed to run for 50 cycles
(data not shown).
2.3.10 Image acquisition and analysis
56
Fluorescence and brightfield microscopy was performed using a Zeiss AxioScope A1. All
images were acquired at 1388 x 1040 pixels using a 40× water objective, N.A. 1.0
(epifluorescence) (Figure 2.5d). Fluorescence emission was detected using a charge-coupled
device (CCD) camera (MRm). All image acquisition parameters were fixed for each imaging
channel for exposure time, excitation intensity and gain. Cells that were dimmer or brighter than
the fixed initial acquisition dynamic range were not included for analysis.
Images were selected for analysis based on identification of single healthy
(morphologically) cells and low background. Fluorescence pixel intensities were measured in
several random regions of interest (ROIs) within the cell cytoplasm (IgK) or nucleolus (Rpl13a)
using ImageJ. Average pixel intensities were calculated from five ROIs of 10x10 pixels for
measurements within the cytoplasm and 5x5 pixels for measurements within nucleoli (Figure
2.1b). All signal intensities were background subtracted from the average of three 10x10 pixel
ROIs surrounding the cell.
2.4 Results
The PQR fluorescent reporters are translated stoichiometrically each time Rpl13a or IgK
protein molecules are produced. The fluorescence intensity of the cell (in arbitrary units) is
proportional to the concentration of fluorescent reporter which is proportional to the production
rate of the protein of interest. Therefore, the fluorescence intensity or brightness of the cell can
be used as a measure of the amount of production of the protein of interest (Lo, Kays et al.,
2015). Similarly, the cycle threshold of the real time amplification reaction is proportional to the
concentration of input nucleic acid, and this is used to quantify the relative or absolute levels of
the target mRNA (Nolan et al., 2006). Measuring the fluorescence intensity of the cell and Ct
values from the qPCR amplification can therefore be used to assay gene expression of any
57
gene(s) from the same cell and determine the variation at both the transcriptional and
translational levels.
To determine the levels of IgK and Rpl13a protein in single 22c10 cells, we used
CRISPR-Cas9 to generate a cell line carrying a PQR-RFP and PQR-GFPnols insertion at the
endogenous IgK and Rpl13a genomic loci, respectively. Single cells fluorescent in both the red
and green channels were chosen and isolated to derive single cell clones. The relative levels of
IgK and Rpl13a protein production were determined by quantifying the nucleolar GFP
fluorescence and cytoplasmic RFP fluorescence intensities from a single cell, respectively. The
same cell was then lysed and the absolute abundance of IgK and Rpl13a mRNA transcripts were
determined using quantitative real time PCR (Figure 2.1).
Using quantitative imaging, we found that Rpl13a protein levels, as measured by
nucleolar GFP fluorescence intensity, ranged from 91 to 842 a.u. (arbitrary units) with an
average of 364 ± 205 (mean ± SD; n=26). The frequency distribution of Rpl13a protein levels
showed a peak at 150 a.u, and was broad and asymmetrical (Figure 2.9a). Specifically, over 60%
of 22c10 cells had GFP intensities of less than 500 a.u., and the remaining cells distributed in a
short tail of higher protein levels. There appeared to be only one peak in the distribution of
Rpl13a protein levels. In contrast, cytoplasmic RFP fluorescence intensities ranged from 936 to
3,263 a.u, with an average of 2,004 ± 619 a.u (n=24). The peak value was around 2,000 a.u., and
the overall shape of the distribution was more normal than that of Rpl13a (Figure 2.9b).
Using a standard curve of serially diluted known amounts of IgK and Rpl13a synthetic
oligonucleotide DNA templates, we determined the absolute number of IgK and Rpl13a mRNA
molecules (Figure 2.9c-d). We found that the absolute number of Rpl13a mRNA transcripts in
single 22c10 cells ranged from 287 to 3,023 transcripts, with an average of 1,156 mRNA
58
transcripts, giving the distribution of measured Rpl13a mRNA numbers a sharp peak at 960
transcripts (n=17) (Figure 2.9c). Similarly, absolute IgK mRNA numbers were found to range
from 2,420 to 159,000 transcripts, with an average of 63,835 IgK mRNA transcripts per cell
(n=25) (Figure 2.9d). This is expected as 22c10 cells are designed and selected to express high
levels of kappa light chain (Li et al., 2010; Lo & Gillies, 1991), while Rpl13a is a non-essential
housekeeping gene (Zhou et al., 2015).
To determine the relationship between gene expression at the transcriptional and
translational levels, we analyzed the linear correlations between mRNA and protein levels of IgK
and Rpl13a and found interesting differences that shed light on the variety of mRNA to protein
relationships that exists for different genes. Pearson’s linear correlations were calculated by
fitting the data to a simple linear regression model, with the coefficient of determination, R2. We
tested the null hypothesis that the variables were independent of each other and that the true R2
value was 0. Rpl13a mRNA levels did not correlate with Rpl13a protein levels, with a Pearson’s
correlation coefficient of 0.1 (p>0.05, n=17) (Figure 2.10a and e). In contrast, IgK mRNA levels
were a better predictor of IgK protein levels, with a correlation coefficient of 0.44 (p<0.05,
n=24) (Figure 2.10b and e), consistent with the generally reported power of predicting protein
abundances from mRNA levels (Greenbaum et al., 2003; Maier et al., 2009; Schwanhäusser et
al., 2011).
To determine whether changes in Rpl13a protein can predict changes in IgK protein, the
red and green fluorescence intensities in single 22c10 cells were compared. We found no
significant linear correlation (R2=0.14, p>0.05, n=24) which indicates that the levels of Rpl13a,
do not correlate with changes in the levels of IgK protein (Figure 2.10c), consistent with
Rpl13a’s role as a housekeeping protein. In contrast, to determine whether the expression of IgK
59
and Rpl13a mRNA might be linked, we compared absolute IgK and Rpl13a transcript numbers.
We found that Rpl13a transcript numbers were a better predictor of IgK mRNA levels (R2=0.3,
p<0.05, n=17) (Figure 2.10d), compared to protein levels. Taken together, these results
demonstrate that Rpl13a mRNA is likely to vary more in a single cell with other mRNAs under
normal conditions, suggesting that Rpl13a is more stable as a housekeeping normalization
control at the protein level.
In order to determine whether the regulation of Rpl13a mRNA and protein levels is
preserved across species, we generated human embryonic kidney 293 (HEK293) cells carrying a
PQR-GFP insertion at the endogenous human RPL13Aa (hRPL13A) locus. Using the same
approach, we found that hRPL13A protein levels ranged from 60 to 240 a.u. (compared to 91 to
842 a.u for mouse Rpl13a), and absolute hRPL13A mRNA transcript numbers ranged from 11 to
192 (compared to 287 to 3,023 mouse Rpl13a transcripts), resulting in a similarly poor inverse
correlation between hRpl13a mRNA and protein levels (R2=0.12, p>0.05, n=25) (Figure 2.10f).
These results demonstrate similar mRNA-protein relationships for RPL13a between the human
and mouse genomes.
2.5 Discussion
The genome-wide correlation between mRNA and protein levels in cell populations is
recognized to be poor, with mRNA predicting protein levels with only 40% power
(Schwanhäusser et al., 2011). This is typically attributed to regulatory mechanisms that can
independently affect each step of gene expression. Measuring and correlating the levels of IgK
and Rpl13a mRNA and protein from single 22c10 cells shows exactly how different genes can
have different mRNA-protein relationship dynamics: IgK mRNA level was a better predictor of
IgK protein levels (R2=0.4, p<0.05, n=24), compared to Rpl13a mRNA to protein levels
60
(R2=0.1, p>0.05, n=17) (Figure 2.10e). In addition, Rpl13a mRNA was more likely to co-vary
with other mRNAs, compared to Rpl13a protein variation (Figures 2.10c and e).
This protocol also allows the examination of relationships between the expression of
different genes and their products. Since this technique yields two metrics per gene, assaying
multiple genes allows the co-correlation of mRNA and protein levels across different genes. This
can be used to study the regulation and differential expression of mRNAs and proteins of
multiple genes at the single cell level. One analogous example is the use of housekeeping genes
for the normalization of quantitative mRNA and protein measurements. In this case changes in
IgK mRNA or protein levels may be normalized to the mRNA and proteins levels of Rpl13a, a
commonly used housekeeping gene for normalization of expression. For example, this knockin
22c10 cell line can be treated with drugs and then screened for cells that exclusively upregulate
or downregulate the expression of IgK antibodies, without affecting normal cellular
housekeeping functions, as would be assayed using Rpl13a. Similarly, this technique can be used
to examine whether different genes are co-regulated within a single cell, under different
treatment or disease conditions. Interestingly, our results revealed a large variability in Rpl13a
transcript numbers (mean ± S.D = 1156 ± 655), making it a poor normalization standard for
mRNA quantitation in 22c10 cells. While Rpl13a has been shown to be more stable than other
commonly used reference genes such as Gapdh or Hprt (Ling et al., 2011), our results show that
such universal conclusions cannot be drawn across all cell types and studies performed at the
single cell level.
The protocol is limited by the time required by the experimenter to collect and assay the
individual cells. The protocol can be improved by faster collection and imaging of cells using
automated or microfluidic chambers, and using non-precipitation approaches to isolate the total
61
RNA. However, in its present form the protocol requires 5 days to perform CRISPR experiments
and initiate clonal selection of cells carrying PQR reporters at any endogenous genomic locus
and ~4 h from start of imaging to qPCR amplification. In addition, the protocol does not require
uncommon reagents or equipment and is accessible to a wide range of researchers. Because of
the small quantity of starting template, certain low copy-number genes may fall below the
detection range of the reverse-transcriptase or PCR polymerase enzymes, and the success and
quality of the amplification depends on the optimization of the primer and probe conditions, in
addition to the quality of the template. With 85% efficiency primers, our results show that the
average amplification failure rate is around 2-3 in 10 cells, and is target-dependent (9 failed
Rpl13a reactions vs 0 failed IgK reactions). Such results can be explained by variability in the
RT step between IgK and Rpl13a RT primers and by the substantially lower number of Rpl13a
transcripts (mean= 1,156 transcripts) as compared to IgK (mean= 61,282 transcripts). The failure
rate may be improved with more efficient primers, and this further increases the reproducibility
of the assay.
The PQR technique is a translational reporter that marks when proteins are produced
within cells, however the differences in maturation and turnover rates between the PQR reporters
and the assayed genes may in some cases hinder accurate protein quantification. While we used
our most optimized PQR sequences that produce equimolar products and result in virtually no
fusion product (Lo, Kays et al., 2015), it cannot be guaranteed that within the cell the levels of
the reporter and protein of interest are equimolar at all times. Differences in folding, maturation
time and the inherent turnover rate of proteins can act differently on the reporter and protein of
interest. This intuitively suggests that protein quantification using PQR provides a relative
measure of protein production at the time of imaging, and it should be noted that PQR does not
62
measure absolute protein levels (Lo, Kays et al., 2015). However, several recent quantitative
protein studies indicate that effects such as protein stability, half-life and degradation account for
a little over 5% of the variation in protein abundance (Li et al., 2014; Schwanhäusser et al.,
2011). Using pulse-chase labelling of newly synthesized proteins, the abundance of a protein in a
given cell was determined to be mainly dictated by the abundance of its mRNA and rate of its
translation, accounting for 40% and 50 % of the variance, respectively (Schwanhäusser et al.,
2011). Since the protein of interest and the reporter are produced from the same mRNA molecule
by the same ribosome, the main source of variability between their levels is the difference in
their turnover rates. This means that the fluorescence intensity of the reporter can be reliably
used to readout the production of the protein of interest with an acceptable error margin.
Differences in turnover rate and stability merely change the slope of the linear relationship
between fluorescence intensity and protein production, as we have extensively demonstrated
using experiments and simulations in Lo, Kays et al., 2015.
2.6 Conclusion
Analysis of gene expression at the transcriptional and translational level in single cells
provides maximal resolution into the dynamics of gene expression. Combining novel techniques
with recent advances in genome editing has opened the door for the interrogation of the
endogenous genome, transcriptome and proteome. Tools such as this combined RNA and protein
measurement approach will reveal previously hidden molecular mechanisms that govern the
expression of proteins and phenotypes. For example, insertion of spectrally different PQR
reporters into different alleles of the same gene can be used to dissect out the allelic contribution
and balance in the expression of a gene (unpublished). In addition, insertion of PQR reporters
into endogenous disease loci can result in a reporter that can be used to readout mRNA and
63
protein responses to drugs or therapies (unpublished). Finally, this protocol can easily be applied
by any researcher who wishes to examine how well they can use the mRNA levels of their gene
of interest as a proxy for protein abundance. Differential mRNA expression studies are still faster
and more economical to perform than protein expression studies, and have a higher multiplex
potential, and with this technique our current arsenal of mRNA markers can be reduced to the
most informative and predictive transcripts.
64
2.7 Figures
Figure 2.1 Workflow of protein and mRNA measurement from the same cell.
65
(a) Insertion of a PQR reporter between a fluorescent reporter (RFP) and a gene of interest
results in the stoichiometric cotranslation of a fluorescent reporter which can be used to quantify
endogenous protein production levels. PQR constructs allow for one molecule of RFP to be
produced for each molecule of protein of interest translated. Since the fluorescence output of
GFP is directly proportional to its concentration, the fluorescence intensity of the cell can be
used to determine the level of production of the protein of interest. (b) Quantitative fluorescence
intensity measurement from regions of interest within the cell can be used to quantify the
production level of the upstream protein. Following imaging, the same cell was lysed for total
RNA extraction and absolute mRNA abundance quantification using single cell qPCR.
66
Figure 2.2 Protein and mRNA measurement for multiple genes in a single cell.
PQR constructs carrying an RFP and a GFPnols reporter were inserted into the endogenous loci of
IgK and Rpl13a, respectively, in 22c10 cells. The transcript numbers and protein levels of both
genes from the same cell were obtained using single cell qPCR and quantitative imaging.
67
a
68
b
Figure 2.3 Insertion of PQR-XFP reporters into the endogenous genomic loci of IgK
and Rpl13a using CRISPRs.
(a) Workflow and timeline of a typical CRISPR targeting experiment to insert PQR-XFP
reporters into any endogenous locus. (b) The fluorescent reporters (approximately 800 base pairs
in size) are inserted in-frame immediately upstream of the stop codon, preserving the
endogenous 5’ and 3’ untranslated regions and the native coding sequence.
69
Figure 2.4 Validation of CRISPR-mediated insertion of PQR-GFP in the
endogenous IgK locus.
(a-b) Insertion of a PQR-GFP reporter into the endogenous IgK locus in 22c10 cells results in
green fluorescent hybridomas. (c) Successful genomic integration is verified with PCR primers
that lie within and outside the PQR insertion (Primer sets A and B for 5’ and 3’ end verification,
respectively).
a
b c
70
Figure 2.5 Illustration of the important steps and typical equipment used in the
protocol.
(a) An aspirator assembly is used to manipulate cells into individual 100uL drops of culture
medium on a Teflon coated, spotted glass slide. DNA and RNA-free freshly pulled micropipettes
are used for transferring cells. The glass micropipette tip may be broken slightly on the bottom of
the slide in order to create a tip diameter wide enough for unperturbed cell flow (inset). (b) PQR-
edited 22c10 express fluorescent reporters each time a protein of interest is translated. The cell
suspension is prepared at a low enough density to permit easy manual manipulation of single
cells. (c) Representative examples of single 22c10 cells near the broken micropipette tip. Note
the large diameter of the micropipette tip (2-3 times the size of one cell). Arrowheads point to
single cells. (d) The microscope objective is dipped into the drop of culture medium and a liquid
71
meniscus is formed around the diameter of the lens. (e) A single fluorescent 22c10 cell is
brought into focus and its fluorescence intensity is measured. (f) Following imaging, the cell is
immediately transferred using a new micropipette directly into Trizol solution. The micropipette
tip may be broken inside the tube to fully expel the contents of the pipette into the solution.
72
Sam
ple 1
Sam
ple 2
Sam
ple 3
26
28
30
32
34 1 l
3 l
6 l
cDNA input
Cycle
th
resh
old
Figure 2.6 Titration of starting input cDNA volume.
The volume of input cDNA is titrated to find the optimal volume that produces the most sensitive
amplification without exceeding 30% of the total reaction volume. Cycle threshold is inversely
proportional to starting template amount and thus in all three samples above, 6uL of starting
cDNA results in high amplification sensitivity without obvious inhibitory effects. n=3 cells.
73
Figure 2.7 Standard curve of serially diluted known amounts of Rpl13a target.
Absolute mRNA abundances are determined by using standard curves which are generated by
running a serial dilution of a known quantity of DNA that contains the qPCR assay target. The
logarithm of DNA quantity is plotted against cycle threshold (Ct) values and the slope and y-
intercept are used to convert Ct values into defined quantity values (number of
molecules/quantity). Ct values obtained from single cell experiments can then be compared to
this curve, in order to interpolate and determine the absolute abundance of mRNA from single
cells. Standard curves are prepared and quantified with each experiment. The slope of the
standard curve above is -3.7, producing an assay with 86% efficiency, and the R2 of the linear
regression was 0.99 (p<0.05).
0 2 4 6 8 100
10
20
30
40
50
Log DNA quantity (ng)
Ct
74
Amplification plot
Figure 2.8 Contamination of RNA sample quantification by genomic DNA can be
assessed using no-RT control reaction.
Reverse transcription reactions in which the reverse transcriptase enzyme has been omitted can
be used as template for qPCR to quantify the contamination of the RNA sample by carryover
genomic DNA. (a) Raw amplification plot from no-RT control reactions. No above-threshold
amplification can be observed. Using Trizol purification, genomic DNA is efficiently removed
and no contamination is observed when no-RT control reactions are amplified using 1, 3 or 6 µl
starting volume. (b) Quantification of data from (a), n=3 cells.
Sam
ple 1
Sam
ple 2
Sam
ple 3
-1
0
11 l
3 l
6 l
cDNA Input
Am
plificati
on
b
a
75
010
020
030
040
050
060
070
080
090
0
0
2
4
6
F(GFP) a.u.
Fre
qu
en
cy
700
900
1100
1300
1500
1700
1900
2100
2300
2500
2700
2900
3100
3300
0
2
4
6
F(RFP) a.u.
Fre
qu
en
cy
200
400
600
800
1000
1200
1400
1600
1800
0
1
2
3
4
5
RPL13a number of transcripts
Fre
qu
en
cy
Figure 2.9 Endogenous RNA and protein quantification from single cells.
(a) GFP fluorescence intensity (in arbitrary units a.u.) from single genome-edited 22c10 cells
shows a moderate distribution, n=26 cells. (b) RFP fluorescence intensities were higher and
clustered more, indicating consistently high level IgK production, n=24 cells. (c) Frequency
distributions of absolute Rpl13a mRNA numbers shows moderate expression of the Rpl13a gene,
n=17 cells. (d) Absolute IgK mRNA levels were relatively higher, indicating high expression of
the IgK gene compared to Rpl13a, n=24 cells.
a b
c d
5000
2500
0
4500
0
6500
0
8500
0
1050
00
1250
00
1450
00
1650
00
0
2
4
6
IgK number of transcripts
Fre
qu
en
cy
76
Figure 2.10 Protein and mRNA relationships between multiple genes in single cells.
(a) Rpl13a mRNA levels did not correlate with green fluorescence intensities indicating Rpl13a
mRNA is a weak predictor of Rpl13a protein production (R2=0.0, p>0.05, n=17). (b) IgK
mRNA levels had a moderate correlation with red fluorescence intensities (R2=0.44, p<0.05,
77
n=24). (c-d) IgK and Rpl13a exhibit nonsignificant correlations at both the protein level
(R2=0.14, p>0.05, n=24), and at the mRNA level (R2=0.06, p>0.05, n=17). (e) Examination of
mRNA and protein levels of multiple genes simultaneously from the same cell allows the co-
correlation of mRNA and protein levels. Black lines connect data obtained from the same cell
(n=17). (f) Human RPL13A (hRPL13A) mRNA levels were also a poor predictor of red
fluorescent intensities (R2=0.12, p>0.05, n=25), indicating a similar kind of relationship
between RPL13A mRNA and protein across the mouse and human genomes.
78
Forward Primer Reverse Primer RT Primer qPCR Probe
Mouse Rpl13a TCCCTCCACCC
TATGACAAG
GTCACTGCCT
GGTACTTCC
GCAGCCCTGCT
ACTCATTTTC
AGACTAAAA
TTCGTCGCTC
CGCTTCC
Human RPL13A TGTTTGACGGC
ATCCCAC
CTGTCACTGC
CTGGTACTTC
CTGCTGGCCAC
ATTTTATGTC
CTTCAGACGC
ACGACCTTG
AGGG
IgK AGTGGAAGATT
GATGGCAGTG
CTGTCTTTGCT
GTCCTGATCA
GGTGGATTTCA
GGGCAACTA
ACAAAATGG
CGTCCTGAAC
AGTTGG
Mouse Rpl13a
outside of homology
arms
CGGGTTGCTAA
CCTGGAATA
CAGTCTCCAT
CAAGGGGAAA
IgK outside of
homology arms
GGGGGAAAGG
CTGCTCATAA
TAACTGGGGG
AAGGGACACT
Table 2.1 Sequences of primers and probes used in this protocol.
79
Chapter III - A system for direct observation of
subcellular protein translation in single living cells.
3.1 Relation to overall project
The localization of mRNA and its local translation is an elegant regulatory mechanism to
restrict the expression of genes to subcellular compartments. While local mRNAs in distal
cellular sites can be visualized, the current tools to detect local protein synthesis events are
invasive and lacking in spatial and temporal resolution. To address this issue, I modified our
original PQR technique (Lo, Kays et al., 2015) to develop a fluorescence-based technique to
directly observe local protein synthesis events in single living cells. This chapter describes the
technique and validation experiments used to demonstrate its use as a quantitative marker of
local protein translation events.
80
3.2 Introduction
Understanding protein dynamics in vivo requires accurate and sensitive tools that can
detect spatiotemporal changes in protein translation at subcellular resolution. The local
translation of new proteins is a widely used cellular mechanism to restrict gene products to
specific regions of a cell or animal (Besse & Ephrussi, 2008; S. Kim et al., 2010; Lin & Holt,
2008; Martin & Ephrussi, 2010; Sharon A Swanger & Bassell, 2011; Dan Ohtan Wang et al.,
2010). mRNAs of locally translated genes are transported from the cell soma to discrete distal
cellular sites, and their localized translation only at those sites ensures local delivery of cargo,
cell fate specification and local signal responses. For example, the local synthesis of proteins
during development plays a crucial role in the establishment of cell fate (Cáceres & Nilson,
2005). In the nervous system, the dynamics of the local protein translation of the immediate early
gene Arc have been imaged in real time in dendrites of neurons expressing a luciferase-based
reporter. Local translation of Arc in neurons was induced by glutamate and could be detected
within 15 seconds in dendrites but not spines, supporting a model where stalled ribosomes at
dendrites are reactivated following glutamate stimulation to rapidly produce Arc protein (Na et
al., 2016). While the luciferase reporter that was used in the study provides high temporal
resolution, a drawback of this approach was the requirement for a substrate for the luciferase
enzyme to emit light. The concentration of this substrate, provided in the extracellular medium,
must be maintained throughout the experiment in order to produce accurate and consistent
results, and several instability issues have been reported (Craig et al., 1991; Morse & Tannous,
2012). Therefore, maintaining controlled levels of substrates and ensuring its constant access to
the enzyme are unpredictable which can present variability that biases results.
81
Genetically encoded fluorescent protein fusions have enabled the direct observation of
intracellular proteins, however protein fusions have been shown to affect protein properties and
localization (Palmer & Freeman, 2004) and folding and function of both the protein of interest
and the reporter (H. L. Zhao et al., 2008). In addition, the constitutive fluorescence of the
reporters may saturate measurement ranges over time, making the detection of small changes in
signal intensity difficult. Internal ribosomal entry sites (IRES) are sequences that initiate
translation of mRNA in a 5’ cap-independent process (Pelletier & Sonenberg, 1988). Bicistronic
expression of using IRES placed between a gene of interest and a fluorescent reporter can
address some of the issues encountered with fluorescent protein fusions. However, IRES
sequences can be 500 bp long and IRES-based translation is not stoichiometric and has been
shown to strongly favor the translation of the upstream gene (Houdebine & Attal, 1999;
Mizuguchi et al., 2000) making them unamenable for quantitative coexpression in vivo.
Our protein quantification technique uses the stoichiometric separation and production of
a fluorescent reporter molecules each time a molecule of protein of interest is translated (Lo,
Kays et al., 2015). Therefore, the fluorescence intensity or the brightness of the cell can be used
as a proxy measure for how much protein of interest is being translated. This technique is an in
vivo protein translational reporter technique that takes advantage of a viral genetic tag that
triggers a ribosomal skipping mechanism during translation, resulting in the stoichiometric
production of two proteins. One drawback of using standard PQR reporters for the detection of
the small and transient local protein synthesis events is the fact that standard fluorescent
reporters such as GFP, RFP and YFP require on the order of tens of minutes to properly fold,
mature and emit fluorescence signals (Snapp, 2005). Molecular diffusion forces can move
82
proteins great distances in the cytoplasm prior to the emission of fluorescence which precludes
any accurate spatial or temporal determination of protein synthesis (S. Kim et al., 2010)
In an effort to enhance PQR reporters to be able to detect and quantify local protein
synthesis events, I have investigated whether we could use the split GFP reconstitution system
for the detection of local translation at single cell resolution, in vivo. The split GFP system is
based on bimolecular fluorescence complementation of proteins, which takes advantage of the
fact that the GFP protein can be split into two nonfluorescent complementary parts that can
spontaneously and non-covalently reassemble to form the full fluorescent protein (Kerppola,
2006). Splitting the GFP protein at the 213th residue between the 10th and 11th barrels to make
GFP1-10 and a 15 amino acid GFP11 peptide results in split GFP partners that do not require any
cofactors or enzymes for reconstitution (Cabantous, Terwilliger, et al., 2005). Placing the
sequence of a GFP11 peptide downstream of our PQR reporter would result in the translation of
a GFP11 peptide for every molecule of protein of interest produced. In the presence of GFP1-10,
the splitGFP parts can recombine and emit green fluorescence, which marks the time and
location of the translation of the protein of interest (Figure 3.1). Key to this reconstitution is
timing, the complementation and fluorescence signal must develop before the GFP protein can
diffuse too far away from the site where GFP11 is translated. The signal must develop at the
moment and location of translation for this system to offer advantages over using standard
fluorescent protein fusion or coexpression.
Using the split GFP approach to monitor protein translation events offers a number of
advantages over probe or antibody-based approaches. For example, the reporters are genetically
encoded and the signal is fluorescence based which minimizes the invasiveness associated with
detecting protein translation. In addition, the fast reconstitution of the reporter immediately after
83
translation would provide exceptional spatial and temporal resolution that can be used to localize
protein translation evens in subcellular compartments such as the endoplasmic reticulum or
neuronal dendrites. Such an approach would open the door to the direct observation of local
protein synthesis in neurons, which is key to understanding the distal processes that both
maintain cellular homeostasis and mediate plasticity.
3.3 Materials and Methods
3.3.1 Protein Quantification Reporter constructs
For work in mammalian systems, the DNA sequence chosen for the PQR peptide was
GGAAGCGGAGCGACGAATTTTAGTCTACTGAAACAAGCGGGAGACGTGGAGGAAA
ACCCTGGACCT. For work in Drosophila, the DNA sequence for the PQR peptide was
GGAAGCGGAGAAGGTCGTGGTAGTCTACTAACGTGTGGTGACGTCGAGGAAAATCC
TGGACCT (Lo, Kays et al., 2015). Sequences were generated using gene synthesis (BioBasic)
and cloned into pCAG or a modified version of pCFD3 for mammalian or Drosophila work,
respectively. Plasmids containing mammalian and insect PQR reporters are now available from
Addgene. GFP, RFP, and BFP constructs were based on superfolderGFP, TagRFP-T, and
mTagBFP2, respectively. SuperfolderGFP and TagRFP-T were chosen for their relatively fast
maturation times, 6 min and 100 min, and photostability, respectively (Pédelacq et al., 2006;
Shaner et al., 2008). ShakerGFP cDNA (R. Blunck, Université de Montréal) was a kind gift, and
ShakerRFP was generated by swapping out the GFP fluorophore in the original construct for a
Tag-RFP-T fluorophore. All other plasmids were obtained through Addgene (Cambridge, USA).
3.3.2 Split GFP DNA constructs
84
Residues 1-213 of GFP, corresponding to the first 10 beta barrels of GFP (GFP1-10),
were amplified and cloned from evolved superfolderGFP, according to published split GFP data
(Cabantous, Terwilliger, et al., 2005; Kim et al., 2011). GFP11 was generated by amplifying and
cloning the last 15 residues of GFP into pCAG. To stoichiometrically co-express GFP1-10 or
GFP11 with other proteins of interest, a PQR peptide was added in-frame upstream or
downstream of the GFP1-10 or GFP 11 sequence depending on the desired orientation to
produce reporter constructs that stoichiometrically express either GFP1-10 or GFP11
downstream of a gene of interest. For extracellular membrane-bound expression of GFP1-10, the
complete Neuroligin-1 signal sequence, in addition to portions of the Neuroligin-1 extracellular,
transmembrane, and intracellular anchoring domains were fused to the N-terminus of GFP1-10
(J. Kim et al., 2011). For electrophysiology experiments, a PQR-GFP11 reporter was placed
downstream of the ShakerRFP coding sequence to generate ShakerRFP-PQR-GFP11, and
cloned into pCAG.
3.3.3 GFP1-10 protein production and extraction
DNA encoding GFP1-10 protein was transformed and expressed under the control of an
arabinose inducible promoter in Escherichia coli strain BL21(DE3) (New England BioLabs).
Cells were grown in LB medium to an initial O.D of 0.2, at which point induction of protein
production was initiated with 0.2% L-Arabinose and cells were further grown at 37°C for an
additional 16 hours with shaking at 225 rpm to encourage inclusion body formation. Cultures
were harvested using centrifugation and GFP1-10 was purified from inclusion bodies by
resuspension with TNG buffer [100 mM Tris-HCl (pH 7.4), 150 mM NaCl, 10% glycerol
vol/vol] containing 0.5 mg/ml lysozyme, 50 units of DNase I. The lysate was then incubated at
37°C for 25 min. Crude lysates containing GFP1-10-rich inclusion bodies were separated using
85
centrifugation at 16,000g at 4°C. Inclusion bodies were lysed using B-Per (Thermo-Fisher) and
sonication (as above) and GFP1-10 protein was collected and filtered using a 0.22 µm filter
before concentration with 10,000 molecular weight cutoff columns.
3.3.4 GFP 11 peptides
Variants of the GFP11 peptide were chemically synthesized with >75% purity
(Genscript). The amino acid sequences of the GFP11 peptides were: GFP11v1:
RDHMVLHEYVNAAGIT, GFP11v2: RDHMVLLEFVTAAGIT and GFP11v3:
RDHMVLHEFVTAAGIT (see Table 1 for full sequence list). Lyophilized peptides were
resuspended in water to > 10 mg/ml and frozen at -20°C. For extracellular GFP reconstitution in
HEK293 cells, GFP11 peptide was dissolved into the culture medium at a final concentration of
50 µM and cells were returned to a 37°C incubator for 2 hours before live imaging.
3.3.5 Cell culture
HEK293 cells were cultured at 37°C under 5% CO2 in Dulbecco's Modified Eagle
Medium, supplemented with 10% fetal bovine serum (Wisent), or for Drosophila melanogaster
S2 cells, at 25°C in Ex-Cell 420 Medium (Sigma-Aldrich). Media were supplemented with 100
units/mL penicillin (Thermo-Fisher) and 100 μg/mL streptomycin (Thermo-Fisher). Mammalian
cells were transfected with 3.5 µg of plasmid DNA in 35 mm dishes using Lipofectamine 3000
(Thermo-Fisher), or Transit-Insect (Mirus) for Drosophila cells. For extracellular GFP
fluorescence reconstitution, HEK293 cells were transfected with constructs expressing GFP1-10
tagged to the transmembrane and extracellular domains of the cell surface molecule Neuroligin-1
and incubated for 24-36 hours. Cells displaying GFP1-10 on the extracellular side of the cell
membrane were incubated in culture medium containing 50 µM GFP11 peptide for 3 hours at
86
37°C before live imaging. GFP1-10 protein was allowed to accumulate for 24 hours before
transfection of GFP11 constructs.
3.3.6 In vitro protein reconstitution
In vitro fluorescence complementation was performed by mixing purified GFP1-10
protein and chemically synthesized GFP11 peptides and the fluorescence intensity of the reaction
was collected with a StepOnePlus real-time thermal cycler (Thermo-Fisher). Briefly, 3 mM
GFP1-10 in TNG buffer or PBS (varied pH) was added to wells of a 96-well microplate coated
with 1 mM bovine serum albumin (BSA) and allowed to equilibrate for 60 seconds. GFP11
peptide was added according to different final peptide concentrations and the microplate was
immediately loaded into the fluorescence reader. The fluorescence intensity was measured every
10 seconds for 45 minutes at 32°C or 37°C with excitation at 495 nm and emission set at 520
nm. The fluorescence intensity was normalized to the initial fluorescence intensity to express
relative fluorescence increase upon fluorescence reconstitution. Standard curves for GFP
fluorescence measurements were generated by either using reconstituted GFP or GFP purified
from E.coli using GFP-specific chromatography columns (Bio-Rad). GFP protein concentration
was determined using the Bradford assay and absorbance readings at 280nm with a NanoDrop
2000 (Thermo-Fisher). Samples were serially diluted (1:10 or 1:5) and 10 µl samples were
imaged to reduce any non-linear fluorescence excitation effects.
3.3.7 Endoplasmic reticulum and ribosome staining
To visualize endoplasmic reticula (ER), HEK293 cells were transfected with split GFP
PQR reporter constructs and stained (live or fixed) with the ER and ribosome-specific stain
Cytopainter (Abcam). Stained cells were imaged in the green and red channels to examine the
87
co-localization of reconstituted GFP and red ER signals. Colocalization of green and red signals
was determined by calculating the Pearson’s and Mander’s correlation coefficients for
overlapping green and red pixel intensities. Individual z-planes were background subtracted and
thresholded to remove the lowest and highest pixel intensities. Ten ROIs comprising cells and
excluding background and nuclear regions were used for analysis. Both Pearson and Mander’s
colocolization coefficients were independently obtained and cross-validated using Coloc2
(ImageJ) and BioImageXD (Kankaanpää et al., 2012).
3.3.8 Electrophysiology
Standard whole cell voltage clamp was used to record potassium currents from HEK293
cells. Cells were maintained at 25°C in extracellular solution containing 140 mM NaCl, 10 mM
CaCl2, 7.5 mM KCl, 10 mM HEPES, and 10 mM glucose at pH 7.4, 319 mOsm during
recordings. Patch electrodes were pulled from standard wall borosilicate glass (BF150-86-10,
Sutter instruments) with 3–5 MΩ resistances. The intracellular pipette solution was 120 mM
KCl, 2 mM MgCl2, 1 mM CaCl2, 2 mM EGTA, 20 mM HEPES, and 20 mM sucrose at pH 7.23,
326 mOsm. Whole cell currents were low pass filtered at 10 kHz and measured using an
Axopatch 200B amplifier (Axon instruments), and recorded using a DigiData 1200 with
pClamp9 software (Molecular Devices). Cells were held at -80 mV and then given +20 mV steps
of 45 ms. To accurately compare I-V curves and current data across cells and experiments, the
steady-state current was divided by the membrane capacitance (mean Cm=15 pF, n=3), and
current density (pA/pF) was used for comparisons. Consistent cell capacitance, and membrane
and access resistances were verified before and after recordings.
3.3.9 Image acquisition and analysis
88
Fluorescence and brightfield microscopy was performed using a Zeiss AxioScope A1. All
images were acquired at 1388 x 1040 pixels using a 40× water objective, N.A. 1.0
(epifluorescence). Fluorescence emission was detected using a charge-coupled device (CCD)
camera (MRm). 488 nm blue light was used to excite GFP, and 515 nm emission light was
collected. Similarly, 543 nm light was used to excite TagRFP-T and 594nm emission light was
collected. All image acquisition parameters were fixed for each imaging channel for exposure
time, excitation intensity and gain. Cells that were dimmer or brighter than the fixed initial
acquisition dynamic range were not included for analysis. Time-series images were collected
using an open-shutter video configuration in ZenLite (Zeiss). Images were acquired every 167
milliseconds with exposure times of 260 milliseconds.
Images were selected for analysis based on identification of healthy cells and low background.
Fluorescence pixel intensities were measured in several random regions of interest (ROIs) within
the target cellular region using a custom written program in MatLab (MathWorks) or ImageJ.
Average pixel intensities were calculated from five ROIs of 7x7 pixels for measurements within
the cytoplasm, perinucleus, and nucleus, and 3x3 pixels for measurements within the plasma
membrane. All signal intensities were background subtracted from the average of three ROIs
immediately surrounding the cell. For time-series image analysis, background was considered as
the region immediately adjacent (<15 µm) to the perinucleus, or cytoplasm. Kinetic increases in
fluorescence from timelapse or video data were plotted was Ft/F0.
3.3.10 Statistical analysis
Pearson’s linear correlations were calculated by fitting the data to a simple linear
regression model, with the coefficient of determination, R2. Kinetic reconstitution traces were fit
with a simple one-site binding model with the coefficient of determination R2 using Prism
89
(GraphPad). We tested the null hypothesis that the variables were independent of each other and
that the true R2 value was 0 for both linear and nonlinear models.
3.4 Results
3.4.1 GFP11 and GFP1-10 reconstitute spontaneously in vitro
To initially characterize the reconstitution of GFP1-10 and GFP11, we performed in vitro
fluorescence reconstitution with diluted GFP11 peptides. Diluted GFP11 peptides prepared over
a 100-fold range of molar ratios covering 0.1 pmol to 50 nmol were added to 3 mM of a crude
bacterial inclusion body lysate containing overexpressed GFP1-10 (Cabantous & Waldo, 2006;
Huang & Bystroff, 2009). The resulting fluorescence intensity of the reaction was collected over
20 minutes at 37°C (Figure 3.2a). The relative fluorescence intensity of the reaction was
normalized to the initial fluorescence intensity to express relative fluorescence increase upon
complementation. We found that the fluorescence intensity of the reaction was directly
dependent on the input amount of the split GFP components (Figure 3.2b), and the optimal ratio
of GFP1-10 to GFP11 that produced the most efficient reconstitution was 2:1 (Figures 3.2a-b).
The relative fluorescence increase was plotted against GFP11 peptide concentrations and fit to a
simple one-site binding model to determine the dissociation constant (kd) of the GFP11 peptide.
Consistent with previous results, we found that kd was = 0.48 ± 0.12 nmol (R2=0.95, p<0.05)
(Figure 3.2b) (Do & Boxer, 2011; Huang & Bystroff, 2009).Therefore, GFP1-10 must exist in
molar excess of GFP11 for reconstitution to be efficient.
To investigate the dependence of GFP reconstitution on temperature and pH, we
performed in vitro complementation under different temperature and pH conditions. GFP11
peptides (2 µmol) were mixed with an excess of GFP1-10 (1 mg/ml) in a 100 µL reaction
volume and the reconstitution was incubated at 32°C or 37°C in TNG or PBS buffered to pH 4,
90
pH 7 or pH 8.5 for 55 minutes. The resulting green fluorescence intensity was monitored every 5
minutes for 1 hour (Figures 3.2 c-d). Consistent with previous results, the complementation of
GFP was found to be severely reduced in acidic conditions (pH< 7) and lower temperatures and
is more efficient at pH >7 and 37°C (Patterson et al., 1997). These results demonstrate that the
reconstitution of GFP occurs efficiently at physiological conditions. Next, we sought to examine
the kinetics with which GFP reconstitution occurs in vitro.
3.4.2 GFP reconstitution in vitro occurs at millisecond timescales
For split GFP to be useful as a fast and reliable marker of local protein translation events,
the GFP parts must reconstitute immediately after translation. To characterize the kinetics of in
vitro complementation between GFP1-10 and GFP11 at a higher temporal resolution, we mixed
freshly purified GFP1-10 and GFP11 protein at 2:1 ratio and incubated the mixture in a
fluorescence reader at 37°C. The fluorescence intensity of the reaction was measured every 10
seconds for 20 minutes. The fluorescence intensity of the reaction rose logarithmically at a sharp
rate in the first 200 seconds and began to saturate after 5 minutes (Figure 3.3a), compared to the
GFP1-10 alone control. The relative fluorescence increase against time was fit to a simple one-
phase association model to determine the rate constant (k) of GFP11 peptide binding, resulting in
a value of ~ 7 ± 0.0001 x 10-3 s-1 and a half time of 95 seconds (R2=0.99, p<0.001) (Figure 3.3a),
consistent with previous results (Huang & Bystroff, 2009).
It is interesting to note that the initial fluorescence intensity of the most efficient
complementation reaction at t=0 is already almost four times that of the GFP1-10 alone control
(3.8 vs 0.7 x105 a.u.) (Figure 3.3b), indicating that fully formed and fluorescent GFP is present in
the reaction tube by the time it is loaded into the machine and before measurements are begun. It
takes roughly 30-45 seconds to pipet the GFP11 peptide into the reaction plate and load it into
91
the machine and in that time, at room temperature the complementation reaction is already taking
place and GFP fluorescence can be observed. Therefore, to determine the speed of the
complementation reaction with the available equipment, we sought to determine how many
molecules of GFP are being reconstituted per unit time. Using a serial dilution of known
quantities of GFP protein, we measured the fluorescence intensity to generate a standard curve of
fluorescence intensities versus amount of protein, covering a six-fold range of fluorescence
levels and GFP protein amounts (Figure 3.3c). Fitting the standard curve data to a linear fit
model results in an equation which essentially assigns a fluorescence value for any given
quantity of GFP protein (R2=0.99, p<0.0001, n=6). Given that 1 ng of GFP contains 2x1010
molecules, the rate of fluorescence increase per unit time obtained from the reaction
(fluorescence increase (a.u) / millisecond = 8.616) can thus be converted to molecules per unit
time, to obtain ~ 71x106 reconstituted molecules per millisecond. This result demonstrates the
fast reconstitution of split GFP and suggests that following protein translation, newly produced
GFP11 peptides can reconstitute with GFP1-10 on the order of milliseconds.
3.4.3 GFP11 detects GFP1-10 in living cells
To determine whether GFP11 can detect and recombine with GFP1-10 in living cells, we
expressed GFP1-10 on the extracellular surface of HEK293 cells. Portions of the synaptic protein
Neuroligin-1 were used to target GFP1-10 and anchor it on the extracellular surface of the cell
membrane (J. Kim et al., 2011). Cells expressing extracellular GFP1-10 were incubated with 50
µM synthetic GFP11 peptide for 2 hours and then imaged. Fluorescence live imaging shows
clear and distinct green fluorescent HEK293 cells after addition of GFP11 peptide (Figure 3.4b).
GFP was specifically detected at the membranes of HEK293 cell (Figure 3.4b, arrows). The
emission of fluorescence was solely triggered by the addition of GFP11 peptides, and no
92
fluorescence was observed from control untransfected cells, or when no GFP11 peptide was
added (Figure 3.4a).
We tested the intracellular complementation of split GFP by transiently transfecting
HEK293 cells with separate plasmids encoding cytoplasmic GFP1-10 and GFP11. We observed
green fluorescence as early as 7 hours after transfection and cells were homogenously filled with
fluorescent GFP 24-48 post transfection (Figure 3.4c). To test the effect of placing split GFP
reporters upstream or downstream of PQR constructs, we expressed GFP1-10 and GFP11
reporters separated by PQRs in different orders. Fluorescence imaging of live cells shows
homogenous green fluorescence throughout the cytoplasm, irrespective of whether the split GFP
reporter was placed before or after the PQR (Figure 3.4d). These results indicate that split GFP
reporters can be coexpressed with any protein of interest using PQR, and this does not affect the
reconstitution of fluorescence.
3.4.4 GFP reconstitution can report sites of protein translation
To better understand the sources of green fluorescence signals within the cell, we
expressed RFPnols-PQR-GFP11 in HEK293 cells (Figures 3.5b-c). The nucleolar localization of
RFP is a useful marker to highlight the size and boundary of the nucleus, enabling easy
identification of perinuclear regions where most protein translation occurs (M. S. Scott et al.,
2010). In the absence of GFP1-10, cells were nonfluorescent in the green channel but showed
clear red fluorescent nucleoli (not shown). In the presence of GFP1-10, the green fluorescence
signal was concentrated and clustered at perinuclear regions, often forming spots and halos
immediately adjacent to the nuclear boundary (Figures 3.5b-f). Similar patterns were observed
when a DNA construct encoding ShakerRFP-PQR-GFP11 was co-transfected with GFP1-10
(Figure 3.6). ShakerRFP is a membrane-bound potassium channel that is translated and
93
processed in the ER and Golgi apparatus before being transported to the plasma membrane.
Transfection of either the nucleolar RFP or membrane-bound ShakerRFP constructs showed the
correct localization of the upstream protein, indicating the co-expression of GFP11 reporter does
not perturb the translation or localization of upstream proteins (Figures 3.5 and 3.6).
To confirm that the local concentration of green fluorescence might correspond to sites of
active protein translation, we stained cells expressing GFP1-10 and GFP11 PQR constructs with
a red stain specific for endoplasmic reticula and ribosomes. High-resolution imaging of live and
fixed cells revealed high spatial overlap of the green and red fluorescence signals (Figures 3.5d-
f), indicating GFP fluorescence overlaps with endoplasmic reticulum and ribosomal signals
(Figures 3.5d-g). Mander and Pearson’s correlation coefficients were calculated for individual z-
slices to validate our results. Green and red pixel intensities were confirmed to colocalize;
specifically 71% of above-threshold GFP signals colocalized with 82% of above-threshold red
ER and ribosome stain signals (thresholded Mander’s (tM) coefficients tM1=0.71, tM2=0.82,
Pearson’s R2=0.84, Costes p-value=1 (i=100), n=10 ROIs) (Figure 3.5g). These results suggest
that the reconstitution of GFP is taking place at sites of active protein translation.
We further examined the dynamics of the fluorescent signal originating at perinuclear
sites, compared to regions within the cytoplasm to determine whether we could detect new GFP
synthesis. HEK293 cells transfected with split GFP components were imaged and the
fluorescence intensity within several regions of interest was quantified over time. We chose to
begin our imaging sessions 18 hours post transfections to coincide with high rates of translation
of transfected plasmids. Using quantitative imaging we found that the intensity of fluorescent
signals originating from perinuclear sites rose while cytoplasmic fluorescent signals did not over
the span of ~ 2 minutes (Figures 3.5g-h). Fluorescent signals from both perinuclear and
94
cytoplasmic regions showed an initial small decrease in fluorescence intensity, which we
attributed to photobleaching. The subsequent fluorescence increase at perinuclear sites was
consistently larger compared to cytoplasmic regions, even in cells that displayed more initial
photobleaching at perinuclear regions, indicating new GFP protein synthesis and reconstitution
(n=6) (Figure 3.5i). These results demonstrate that splitGFP can be used to observe the localized
synthesis of proteins and confirm that regions stained with the endoplasmic reticulum and
ribosome marker correspond to sites of new protein production.
3.4.5 Proteins co-translated with GFP11 reporters function properly
To verify that proteins produced with a PQR and GFP11 reporter can function properly,
we expressed the ShakerRFP ion channel separated from a GFP11 reporter by a PQR peptide
(ShakerRFP-PQR-GFP11) in HEK293 cells (Figure 3.6a). ShakerRFP is a large tetrameric
potassium channel that has a RFP molecule embedded in its N-terminal domain. The RFP
fluorophore acts as a tethered ball and chain which prevents the channel from inactivating,
allowing it to pass current as long as depolarization is maintained. As with most ion channels,
upon translation they are processed in the secretory pathway and packaged in vesicles to be sent
to the membrane, where they are inserted for periods of hours to days (Narahashi, 1988).
Fluorescence imaging of live HEK293 cells expressing ShakerRFP-PQR-GFP11 alone showed
distinct red fluorescent cell membranes, indicating the membranous distribution of the channel
(Figure 3.6a). Using whole-cell patch clamp, +20 mV voltage steps were applied and the steady
state current passed by the channel at each voltage step was used to generate an current-voltage
(I-V) relationship curve (Figure 3.6b). I-V curves are useful metrics that relate the current passed
across a membrane, when a voltage is applied across it. Since the current is determined by the
conductances present in the membrane, I-V curves serve as a diagnostic for proper channel
95
expression and function within a membrane (Tester, 1997). To normalize for variability in
transfection and uptake of plasmids by HEK293, current is divided by cell capacitance to give
current density (pA/pF), which is more appropriate when comparing I-V curves (Figure 3.6b). In
HEK293 cells in which ShakerRFP has been transiently overexpressed at very high levels, the
main conductance in the membrane is the Shaker potassium channel. Endogenous voltage-gated
calcium and potassium channels have been reported in the non-neuronal HEK293 cell line, but
the peak current is less than 300pA (He & Soderlund, 2010; Yu & Kerchner, 1998). Thus, the
contribution of endogenous HEK293 currents to the nanoampere range of currents observed in
cells with overexpressed channels can be considered negligible. The addition of a PQR-GFP11
reporter downstream of the coding sequence of ShakerRFP did not appear to affect the
translation or function of ShakerRFP, as determined by live imaging and electrophysiology.
HEK293 cells transiently transfected with ShakerRFP-PQR-GFP11 DNA showed the same
membrane pattern of red fluorescence as cells expressing ShakerRFP alone (Figure 3.6a). I-V
curves generated from cells transfected with ShakerRFP-PQR-GFP11 showed no significant
difference from those transfected with ShakerRFP alone, in terms of slope or reversal potential
(p<0.05 for both). In addition, the measured reversal potential (~ -73 mV) (Figure 3.6b),
corresponded to the potassium ion reversal potential predicted using the Nernst equation to
within 10% error. These results indicate that the use of GFP11 as a reporter of protein does not
perturb protein localization, function, or the general health of the cell in which they are
expressed, which suggests that it is amenable for expressing upstream proteins in vivo.
3.4.6 GFP reconstitution can quantitatively readout protein translation
Protein synthesis of a PQR-GFP11 reporter requires approximately 2 seconds (6 amino
acids/second) (Kramer et al., 2009; Ross & Orlowski, 1982). In the presence of GFP1-10, the
96
split proteins recombine in milliseconds to emit a green fluorescence signal to indicate the time
and location of protein translation within the cell. To determine whether the green signal
produced by GFP reconstitution can quantitatively readout the levels of translation of the
upstream protein, we took advantage of the RFP fluorophore that is embedded in the ShakerRFP
molecule. We co-expressed ShakerRFP-PQR-GFP11 and GFP1-10 in HEK293 cells and
verified that the fluorescent signals observed in the green and red channels are linearly
proportional in intensity (Figures 3.6a,c). Quantitative fluorescence imaging revealed a linear
correlation of the green and red fluorescence intensities (R2=0.71, p<0.001, n=35), indicating
the translation and reconstitution of the GFP11 can quantitatively readout the translation of an
upstream protein (Figure 3.6c). Therefore, the fluorescence intensity of the signal is directly
proportional to the level of translation of GFP11, which in turn is proportional to the translation
of the upstream protein of interest (Figure 3.1).
3.5 Discussion and conclusions
The reconstitution of GFP occurs on the order of milliseconds and split GFP partners can
be stoichiometrically co-expressed with proteins of interest using PQR. This is exploited as a fast
marker to read out protein translation events by using the GFP signal emitted after reconstitution
to mark the time, location and level of protein translation within a living cell. The rationale
behind this approach is to take advantage of the fast translation and reconstitution of the GFP11
peptide. The 15-amino acid GFP11 peptide requires just over 3 seconds to be translated. This is
crucial as once it is produced, it requires no further processing or folding and is free to
immediately recombine with GFP1-10 that has been previously expressed, allowed to fold and is
primed for reconstitution. The concentration of GFP1-10 is rate limiting factor in this approach,
97
and so GFP1-10 must be present in high concentrations in the cell as a critical prerequisite for
the accuracy and reliability of fast protein production quantification.
The GFP11 peptide requires ~ 3 seconds between the initiation of GFP11 translation and
the reconstitution and emission of fluorescence, indicating the moment of protein translation
within milliseconds for GFP11 and with a 3 second delay from the moment of translation of the
upstream protein. Whether or not the upstream protein diffuses away from the site of translation,
translation of GFP11 will mark the original site of mRNA translation, unless the RNA-bound
ribosome diffuses away. Ribosome diffusion in cytoplasm is 0.04 µm2/sec (Bakshi et al., 2012),
and thus the 3 second delays in detecting the initial protein translation even will produce a spatial
error of ~ 350 nm. Our results suggest that the kinetics and efficiency of the reconstitution
reaction can be improved, as different variants of the GFP11 peptide produced different rates of
reconstitution (Figure 3.2c). Minor differences in the solubility, charge and size of the GFP11
peptide can affect the rate and efficiency of reconstitution, and ultimately the properties of the
reconstituted protein. Therefore, screening for GFP11 peptides that result in the most sensitive
reconstitution will certainly improve this technique in the context of monitoring local protein
translation events. Split GFP components that fail to reconstitute, or reconstitute but fail to
fluoresce can affect the spatial, temporal and quantitative accuracy of the GFP11 reporter. While
we did not observe such problems in our experiments, their occurrence is impossible to predict or
estimate in vivo. However, the finding that different variants of GFP11 reconstitute differently
suggests that undiscovered GFP11 peptide variants may possibly outperform currently available
ones. Nonetheless, I am constantly screening for new GFP11 variants that reconstitute faster and
more efficiently, and this is discussed in the next chapter.
3.5.1 GFP1-10 fluorophore maturation
98
It is not known whether the fluorophore is GFP1-10 can mature when GFP1-10 is
produced separately from GFP11 (Cabantous, Pédelacq, et al., 2005; Chudakov et al., 2010;
Huang & Bystroff, 2009; Kaddoum et al., 2010; Kerppola, 2006). Unfolding and refolding
experiments have shown that unfolded GFP that was once native in structure preserved the
maturation of the fluorophore, which explains the fast recovery of fluorescence upon refolding of
the protein. However, in this case the incomplete structure of GFP (GFP1-10) is produced
directly from the gene, and it is not clear whether this precludes the maturation of the
fluorophore. The folding and maturation of the fluorophore occurs at residues 65, 66 and 67
which are over 150 residues away from the site where GFP11 is split, suggesting at least that
splitting the protein does not physically interfere with these residues. However, if the
fluorophore could mature without GFP11, then it would yet GFP1-10 is nonfluorescent, and
remains so until it interacts with GFP11. Residue Glu222, located in the C-terminal domain of
GFP and part of the GFP11 fragment is known to be required for the switching of the
fluorophore between protonated forms, which affects its stability (Kent et al., 2008).
Furthermore, Ser205, Thr203 and His148 are required for fluorophore stabilization (Heim et al.,
1995), but the question of whether a non-stabilized form of a mature fluorophore can form has
not been answered. In addition, the presence of water molecules inside the barrel structure leads
to quenching of fluorescence (Ormö et al., 1996). This raises the idea that an uncomplimented
GFP1-10 could have a mature yet unstable and quenched fluorophore, which quickly switches
conformations and emits fluorescence upon reconstitution with GFP11. This model is supported
by my data, as fluorescence emission began within seconds after introducing GFP11 in vitro.
Given that the maturation of the GFP fluorophore is known to require on average tens of minutes
(Iizuka et al., 2011; Shaner et al., 2008), this suggests that the GFP1-10 fluorophore is in some
99
transition state between folding and fluorophore maturation. Therefore, reconstitution with
GFP11 quickly allows the new molecular conditions to favor the full maturation and emission of
fluorescence from the chromophore.
Finally, the goal in developing this technique is to observe local protein synthesis in vivo,
particularly in neuronal cells. In the next chapter I describe the experimental design of the
strategy to apply the technique in the context of a living animal. The reagents that are currently
being developed are described and some preliminary results are presented.
100
3.6 Figures
Figure 3.1 Stoichiometric production of GFP11 reporters using PQR.
Insertion of a GFP11 reporter downstream of a PQR tag results in slightly longer mRNAs being
transcribed (extra 100 bases). During translation, a ribosomal skipping mechanism results in the
stoichiometric production of a GFP11 molecule each time a molecule of protein of interest is
translated. In the presence of GFP1-10, the splitGFP parts can recombine and emit green
fluorescence, which marks the time and location of the translation of the protein of interest.
101
Figure 3.2 In vitro characterization of the split GFP reconstitution reaction.
(a) Fluorescence reconstitution kinetic traces for amounts of GFP11 that were mixed with an
excess of GFP1-10 in vitro, and incubated at 37°C. (b) Sensitivity of split GFP reconstitution in
vitro. The fluorescence intensity of the reconstitution reaction is linearly dependent on input
GFP11 concentration. (c) Fluorescence reconstitution of split GFP occurs more efficiently at
37°C and different variants of GFP11 result in different reconstitution rates and efficiencies.
20pmol of GFP11 peptide were mixed with an excess of crude GFP1-10 protein and the resulting
fluorescence intensity was collected at regular intervals for 55 minutes. (d) In vitro reconstitution
of GFP was tested under different buffer and pH conditions at 37°C and was most efficient in
TNG buffer at pH 8.5 (n=4).
102
Figure 3.3 Reconstitution of split GFP occurs on the order of milliseconds in vitro.
(a) Kinetic trace of split GFP reconstitution. GFP1-10 and GFP11 were mixed at a 2:1 molar
ratio in TNG buffer and the fluorescence intensity was recorded over time. The fluorescence
intensity of the reaction rose logarithmically and began to saturate after 5 minutes. The
fluorescence increase against time was fit to a one-site model and the rate constant k was
determined to be 0.007 ± 0.0001 s-1, with a halftime of 95 seconds (R2=0.99, p<0.001). (b) The
first 50 seconds from (a) are shown to demonstrate that a portion of splitGFP was already
reconstituted before the earliest timepoint could be acquired. (c) Standard curve of serially
diluted GFP showing the linear dependence of fluorescence intensity on GFP amount (R2=0.997,
p<0.05).
103
Figure 3.4 Split GFP reporters can be expressed using PQRs and the reconstitution
of GFP marks the presence GFP1-10 protein.
(a-b) Extracellular membrane-tagged GFP1-10 was expressed in HEK293 cells. Cells display a
green membranous fluorescent signal upon addition of GFP11 peptide into the culture medium.
(c) Split GFP components can be expressed cytosolically and recombine upon translation to
produce fluorescent GFP. (d) SplitGFP reporters efficiently recombine to produce GFP when
104
expressed using PQRs placed upstream or downstream. Scale bars are 30 µm in (b) and 20 µm in
(c) and (d).
105
106
Figure 3.5 Split GFP reconstitution occurs at sites of active protein translation.
(a-c) SplitGFP reconstitution occurs at perinuclear sites. RFPnols was included in (b-c) to
highlight the nuclear boundary. (d-f) Cells expressing splitGFP constructs were fixed and stained
with a red marker for endoplasmic reticula and ribosomes (Cytopainter). High overlap between
green and red fluorescent signals was observed, particularly at perinuclear regions. (g)
Quantification of colocalization of green and red pixel intensities from ten ROIs confirmed that
green and red signals colocalized above chance (p<0.05). (h) Time-series normalized
fluorescence intensity analysis at perinuclear and cytoplasmic regions of interest in a cell
expressing splitGFP constructs. Perinuclear regions showed increasing levels of fluorescence
over time, compared to cytoplasmic levels. (i) Traces of fluorescence intensity over time from
perinuclear (top panel, n=6) or cytoplasmic (bottom panel, n=5) regions of interest from several
cells. Scale bars are 50 µm in (a-c) and 10 µm in (d-f).
107
Figure 3.6 Co-translation of GFP11 reporters using PQR preserves the protein of
interest’s localization and function.
108
(a) Co-expression of a GFP11 reporter does not affect the membrane localization of ShakerRFP
in HEK293 cells. (b) Representative I-V curve and voltage step protocol used to characterize the
potassium conductance in cells transfected with ShakerRFP-PQR-GFP11. (c) In the presence of
GFP1-10, GFP11 reporters co-produced with ShakerRFP reconstitute and form fluorescent GFP
(image). The green and red fluorescence intensities with linearly correlated, indicating the level
of GFP11 production, and thus reconstitution, is proportional to the level of production of
ShakerRFP (Pearson’s R2=0.72, p<0.05, n=35). Scale bars are 20 µm in (a) and 10 µm in (c).
109
Chapter IV - Applications and future directions of
protein quantification using PQR
4.1 Relevance to overall project
My goal in this part of the thesis is to develop proof of principle experiments to
demonstrate ways to monitor and quantify the translation of proteins in the living animal. The
specific questions to be addressed with these experiments are: 1) how does endogenous protein
production dynamically change in single cells in vivo? and 2) can I use the reconstitution of GFP
as a reliable and quantitative detector of subcellular local protein translation in single cells in
vivo? In this chapter I describe strategies and reagents that are being developed to visualize
protein translation in vivo, at both cellular and subcellular resolution. This chapter presents DNA
constructs, mice and flies currently being generated to demonstrate the applicability of the
techniques in vivo and provide useful resources for the research community.
110
4.2 Applications of optical protein quantification using PQR
4.2.1 Dynamic observation of protein synthesis in vivo
Antibodies are produced and secreted by the immune system in response to pathogen and
toxin exposure as part of the body’s immune response. In vertebrate immune systems, two
identical heavy chains and two light chains constitute the basic structural unit of an
immunoglobulin antibody molecule. There exist five classes of antibodies, each with its own
class of heavy chain and either class of light chains: kappa (κ) and lambda (λ). The detection and
measurement of antibody production has been crucial for the diagnosis and tracking of immune
responses as well as the progression or regression of immune-mediated diseases. Testing
therapeutic agents, particularly those of protein origin, can elicit harmful antibody responses in
the host against the agent, which are disadvantageous (De Groot & Scott, 2007). Therefore, the
immunogenicity of therapeutic proteins and agents remains a concern and monitoring levels of
antibody production is an important indicator in treatment of disease.
Currently, the detection of proteins is mostly performed using antibody-based approaches
such as the ELISA assay or the immunoblot (western blot). To develop a system to optically
monitor dynamic endogenous protein translation in single cells in vivo (question 1), we will
create animals using genome editing that express fluorescent PQR reporters as real time
indicators of protein translation. The first animal that we are generating carries a PQR-GFP
insertion at the endogenous IgK locus. Its main purpose is to allow for in vivo examination of
IgK antibody production in real time, at single cell resolution. B cells throughout the mouse body
will co-produce a molecule of GFP for every molecule of IgK light chain (Figure 4.1). B cells
from different immune regions in the body such as the spleen, lymph nodes, and bone marrow
differ in their reactivity to different stresses and thus they also differ in the amount of antibody
111
they produce (Leandro et al., 2013; Nutt et al., 2015). This mouse will serve as a platform to 1)
optically localize and quantify local immune responses in real time in vivo, 2) easily screen
agents or therapies that increase or decrease the levels of antibody production by simply
monitoring changes in fluorescence intensity and 3) allow the fast and easy identification,
extraction and enrichment of B cells that express varying levels of kappa isotype antibodies from
anywhere in the mouse body.
In order to generate mice carrying PQR reporters at the endogenous IgK locus, validated
CRISPR sgRNAs, Cas9 protein and PQR repair templates from Chapter 2 were injected into the
pronuclei of mouse embryos. We used a complex of Cas9 protein and synthetic RNA CRISPR
instead of using DNA or mRNA encoding the Cas9 and sgRNA, as this has been shown to
significantly decrease DNA toxicity, low expression levels, and issues of nucleic degradation by
nucleases (Liang et al., 2015).
Since immunoglobulin loci can undergo several forms of genomic rearrangements that
render the architecture of the genomic locus unpredictable, inserting the PQR reporters into the
end of the constant region of the light chain locus will ensure that the variety of kappa-type
antibodies, which all share a common constant region, are stoichiometrically co-produced with a
fluorescent reporter (Figure 4.1b). In addition, this approach preserves the endogenous IgK 5’
and 3’UTR, which may contain critical regulatory elements that influence the transcription,
splicing, translation and localization of IgK mRNA.
The resulting founder mice are bred and checked for the integration of the PQR reporter
into the germline by extracting genomic DNA from tail or ear clips followed by genotyping
using primers that lie within and outside the recombined fragment and by sequencing and
112
digestion-verification of the product. Positive founders are then raised and backcrossed to
determine germline transmission and eliminate any potential off-target CRISPR effects.
4.2.2 Optical normalization of protein production in vivo
The expression of nearly half of genes in the mouse genome, including circadian genes
has been shown to occur in an oscillatory manner in many types of tissues throughout the body
(Zhang et al., 2014). In contrast, stochastic “rush hours” of gene expression are frequently being
reported in gene expression atlases (Lein et al., 2007; J. Z. Li et al., 2013; Panda et al., 2002;
Zhang et al., 2014). Thus, expression of a gene can either be stochastic occurring in bursts or
rhythmic, exhibiting patterns.
In any quantitative measurement of change, particularly in vivo, it is crucial to ascertain
whether the observed changes are due to a specific effect, or a non-specific. This also applies to
the quantification of mRNA and protein levels. To address this issue, reference, or
“housekeeping”, gene expression is often used in quantitative mRNA and protein experiments to
normalize for observed changes across conditions and experiments. For example, PQRs inserted
at endogenous housekeeping loci such as RPL13A allow for the fluorescence from those loci to
be used as a normalization signal for measurements taken across cells and experiments. In vivo,
the complex crowding and depth of tissues introduce optical aberrations such as scattering of
excitation and emission light as well as an exponential drop in light penetration power with
increased tissue depth. Insertion of PQR reporters into the endogenous RPL13A locus in a mouse
will result in the production of a fluorescence-based normalization signal in all cells of all tissues
in that mouse. This enables the normalization of any protein measurements imaged in a second
channel (i.e., a second gene of interest), to changes in the levels of RPL13A in that same cell or
tissue. This permits the comparison of normalized data within and between experiments. In
113
addition, this mouse will allow for in vivo dynamic monitoring of Rpl13a production levels at
single cell resolution, which is particularly informative when examining global cellular states in
normal and disease conditions (Poddar et al., 2013). For example, quantifying the Rpl13a
fluorescence in a second channel can be used as a measure of an individual cell’s transcriptional
and translational levels (Chapter 2). Thus, any change in IgK protein production can be
normalized to the cell’s global protein production status, reflecting the true net increase in IgK
production.
The same approach and techniques described in previous chapters were used to insert a
PQR-RFPnols construct into the endogenous Rpl13a locus in mouse embryos. First, a double-
strand break (DSB) at the end of the Rpl13a coding sequence was induced by a guided Cas9
nuclease. PQR-RFPnols repair templates with arms homologous to Rpl13a were used as a
template for homologous recombination resulting in an edited locus of the form: Rpl13a-PQR-
RFPnols (Figure 4.2).
Preliminary results from the injection of a PQR-RFP reporter and Rpl13a-specific
CRISPR/Cas9 into mouse pronuclei showed that the validated CRISPRs from Chapter 2 can
efficiently and repeatedly produce red-fluorescent 2-cell, 4-cell and blastocyst-stage embryos
(Figure 4.2). The implantation of injected embryos into carrier mothers is ongoing and correct
genomic editing of the Rpl13a locus is verified by PCR genotyping and sequencing of the locus
as previously described.
Crossing the IgK knock-in mice with the Rpl13a knock-in mice will produce an F1
generation with PQR-GFP and PQR-RFPnols insertions at the endogenous Igk and Rpl13a loci,
respectively. Cells from this animal will stoichiometrically produce GFP and RFP under the
control of the endogenous genomic IgK and Rpl13A loci, respectively. Every cell in this animal
114
will have red fluorescent nucleoli, with immune cells expressing IgK being green fluorescent.
Using the green to red ratio of fluorescence, the relative protein levels of IgK can be determined
by normalizing to the levels of Rpl13a protein, and this can be used to account for optical
artifacts inherent to in vivo imaging and differences in global transcriptional and translational
states of the cell (Figure 4.2b).
A common limitation of using CRISPRs for genome editing is their tendency to induce
off-target effects that can introduce variability and confound experimental outcomes (Cho et al.,
2014). CRISPRs have been shown to induce DSBs and subsequently mutations at loci with more
than 5-nucleotide differences in the sequence of the target (Fu et al., 2013). Excessive off-target
effects can induce large chromosomal rearrangements that can be toxic to cells, in addition to the
misregulation and potential activation of deleterious genes such as oncogenes (Cho et al., 2014).
Homologous integration of the repair template at non-homologous sites is unlikely to occur due
to extremely low recombination efficiency (10-4 to 10-6 events per basepair per generation)
(Brown et al., 2011). Genome edited mice can easily be backcrossed to the parental strain to
remove almost all the genetic background except the PQR insertion. Backcrossing mice to
parental background at least twice ensures more than 75% of the genetic background has been
isogenized to the parental strain. Therefore, off-target effects pose a bigger problem for in vitro
work.
4.3 Split GFP as a quantitative marker of local protein synthesis in vivo
In Chapter 3, I describe a strategy we developed to use the reconstitution of GFP as a
reliable and quantitative indicator of protein translation in single living cells (Figure 4.3). For
this part of the thesis I aim to create and test different DNA constructs that will be used to
115
generate animals that express split GFP components for the quantitative imaging of local protein
synthesis events in vivo.
The local synthesis of proteins occurs as transient small magnitude events in spatially
restricted compartments far from the cell soma. We can take advantage of the small size and
rapid translation time of GFP11 peptides and place those downstream of PQR sequences. This
allows the generation of a protein synthesis reporter that requires ~ 3 seconds to be translated,
does not require any folding or post translational modification and is thus free to reconstitute
with the larger GFP1-10 fragment and emit green fluorescence. As demonstrated in Chapter 3,
the reconstitution of GFP occurs at extremely high temporal resolution, and this can be taken
advantage of to use GFP11 as a PQR reporter of local protein translation. By generating animals
that express separate split GFP reporters under control of a gene of interest, green fluorescent
signals observed in the F1 generation would be indicative of GFP reconstitution and thus a
protein translation event.
The concentration of GFP1-10 in this assay is a rate-limiting step. To fully exploit that
GFP11 can reconstitute with GFP1-10 in milliseconds, high levels of GFP1-10 are required
throughout the cell, particularly in small distal sites such as neuronal dendrites and axonal tips.
In the next section, I discuss our strategy to generate animals that stably express high levels of
GFP1-10 ubiquitously, and how they will be used as a reagent in our approach to visualize local
protein synthesis events.
4.3.1 Generation of animals constitutively expressing GFP1-10
To produce a stable and high level source of GFP1-10 expressing cells in vivo, we
generated transgenic animals that expresses GFP1-10 under the control of the ubiquitous actin
promoter. This transgenic animal has many uses: First, cell from any tissue can be used as a
116
source of high levels of GFP1-10 protein. Second, primary cell cultures where homogenous and
endogenous GFP1-10 expression is required can easily be generated from any tissue of the
animal. Third, by expressing GFP1-10 in one animal and simply crossing it to another expressing
a GFP11 reporter, fluorescence reconstitution is allowed to occur. This means these animals can
be repeatedly used in future local translation experiments using split GFP, without having to
modify their genotype. In other words, to screen for locally translated candidate mRNAs,
“Protein of interest-PQR-GFP11” animals can be crossed to the same GFP1-10 animal. This
makes the GFP1-10 animal a versatile and practical reagent for my work as well as the research
community.
There typically exist just over 1 million copies of Actin monomers per mammalian cell
typically at concentrations up to 95 µM in the cytoplasm (Luby-Phelps, 2000). This constitutes
roughly 1% of the total proteome, making Actin one of the most abundant proteins in any cell
containing a cytoskeleton (Lodish et al., 2008). In addition, the high cellular levels of Actin are
directly dependent on the long half-life of the actin mRNA (Dormoy-Raclet et al., 2007)
Therefore, driving the expression of GFP1-10 expression using the actin promoter in vivo can
reasonably be predicted to produce high levels of GFP1-10 high levels in ubiquitous cell types
throughout the life of the animal (Qin et al., 2010; Quitschke et al., 1989). To achieve a two-fold
increase in fluorescence intensity over cellular background autofluorescence, it was shown that
EGFP must be expressed at 200 nM, which translates to roughly 10,000 diffuse molecules in a
typical cytoplasm (Patterson et al., 1997; Snapp, 2005). However, fewer molecules may be
detected if they are spatially localized into a small organelle such as into the nucleolus (See
Chapter 2).
117
Using standard transgenic practices, we generated a fly that expresses GFP1-10 driven by
the fly actin (ActB) promoter. To verify the expression of GFP1-10 protein in this fly, we first
performed an immunoblot using adult fly head extracts. Using an anti-GFP antibody, we found
that the predicted 24 kDa GFP1-10 protein is correctly translated and runs at the expected size on
an SDS-PAGE gel (Figure 4.4a). To examine whether GFP1-10 is expressed at high levels in this
fly, we performed immunohistochemistry using the same anti-GFP antibody on larval brain
samples. We confirmed that nonfluorescent GFP1-10 is expressed and present at high levels as
shown by its accumulation in both neuronal and glial cell bodies and neurite projections of the
adult fly brain (Figure 4.4b) and larval ventral nerve cord (Figure 4.4c). From this evidence, we
conclude that the ActB promoter can drive the high-level expression of nonfluorescent GFP1-10
in flies, and this is particularly obvious in the somas and projections of cells in the fly brain.
To enable the translation of such experiments to mammalian systems, we are generating
transgenic mice that express GFP1-10 under the control of the beta actin promoter, a constitutive
and high expression promoter (Gunning et al., 1987; Qin et al., 2010). Mouse pups from the first
round of injection are now born and the correct insertion of GFP1-10 has been verified by
genotyping and Sanger sequencing of PCR products from genomic DNA.
4.3.2 Local translation of Gurken protein in Drosophila oocytes
The first step we took to demonstrate the applicability of the technique in vivo is to
identify a model or system best suited for the unambiguous identification of local translation.
The rationale is to take advantage of a known and characterized system in which the local
translation of a specific gene is known to occur. The local translation of mRNAs in the
Drosophila oocyte is one of the earliest and best studied examples of how the localization of
mRNA is used as a mechanism for translational regulation (Cáceres & Nilson, 2005; Driever &
118
Nüsslein-Volhard, 1988; Martin & Ephrussi, 2010). During egg development, germ cell
specification and embryonic axis patterning are established via molecular asymmetries created
by position-dependent regulation of the translation of mRNAs deposited maternally into the
oocyte (Richter & Lasko, 2011)
The 1.7 kb grk mRNA encodes the Gurken protein, a torpedo/EGF receptor ligand, a
receptor located on the inner surface of follicle cells that envelop the oocyte. A conserved RNA
stem loop element within the grk coding region forms the signal for dynein-dependent grk
mRNA transport and localization (Van De Bor et al., 2005). Additional elements in the 5’ and
3’UTRs ensure proper translational regulation (Saunders & Cohen, 1999). Gurken local
translation at the anterodorsal corner establishes a molecular EGF signaling gradient such that
the highest signalling occurs in neighbouring anterodorsal follicle, initiating cell fate
specification (Figure 4.5a) (Nilson & Schüpbach, 1999).
To observe the local translation of gurken transcripts within the Drosophila oocyte, we
generated a transgenic fly that expresses Grk-PQR-GFP11 under the control of the
transcriptional activator, Gal4 (Figure 4.5a). The Gal4 activator binds to elements in the
upstream activating sequence (UAS), a sequence placed upstream of the construct which allows
the recruitment of the transcription machinery and the subsequent initiation of transcription. The
spatially restricted expression of Gal4 allows the transcription of UAS transgenes only in those
tissues (Brand & Perrimon, 1993; Duffy, 2002). Widely available tissue-specific Gal4 lines such
as nanos-Gal4 allow the spatial restriction of Gal4 expression to germline cells (Rørth, 1998),
which results in the restricted expression of Grk-PQR-GFP11 specifically in those Gal4-
expressing cells. To generate the DNA construct for microinjection into embryos, the entire
1.7kb Grk cDNA including the coding sequence and the 5’ and 3’UTRs were used to flank a
119
Drosophila codon-optimized PQR-GFP11 reporter (Figure 4.5a). Our reason for including the
complete transcript sequence was to preserve all potential regulatory elements of the native Grk
mRNA, ensuring the correct processing, localization, and translational regulation of transcripts
generated from the integrated transgene.
The fundamental rationale behind the experiment is to cross the GFP1-10 and Grk-PQR-
GFP11 transgenic flies and observe Gurken local translation using GFP reconstitution in the F1
generation. F1 females bear the genotype ;GFP1-10/UAS-Grk-PQR-GFP11; nosGal4/+ and this
means that oocytes produced from these females contain high levels of GFP1-10 and nosGal4,
which will drive the expression of the Grk-PQR-GFP11 transgene in the oocyte. Fluorescence
imaging of Stage 9 and 10 oocytes showed a crescent shaped green signal that always localized
to the corner of the oocyte chamber where the nucleus was located (Figures 4.5b-c, n=6). The
observed Gurken translation pattern was consistent with the characteristic Grk signal observed
using immunofluorescence or in situ hybridization studies (Cáceres & Nilson, 2005; Jaramillo et
al., 2008; MacDougall et al., 2003). Imaging of control genotypes in which a component of the
split GFP system is omitted, or if the expression of UAS-Grk-PQR-GFP11 is repressed by
removing Gal4, showed complete absence of a green signal that localizes to the anterodorsal
corner in all assayed control oocytes (Figures 4.5d-e, n>25). This experiment demonstrates that
GFP11 reporters can be translated and can reconstitute with GFP1-10 with high spatial
resolution, and this can be used to monitor the local synthesis of proteins in vivo.
A major problem encountered in fluorescence imaging of Drosophila oocytes is the high
level of autofluorescence seen in several emission channels (Figures 4.4 and 4.5) (Boulina et al.,
2013; Parton et al., 2010). The level of autofluorescence in these tissues can easily obscure the
observation of true fluorescence signals, if the signal intensity is low (Mavrakis et al., 2008). For
120
example, the oocyte chamber contains a yolk which harbours factors necessary to sustain the
development of the oocyte and embryo. The concentration of macromolecular factors in the
cytoplasm can therefore reach concentrations as high as 400 mg/ml (Guigas et al., 2007), which
in addition to dense packing of tissue has been long known to cause nonspecific autofluorescence
emission (Mavrakis et al., 2008; H.-W. Wang et al., 1998). The high autofluorescence seen when
imaging Drosophila oocytes may explain the perceived low fluorescence intensity of the
observed GFP signals in the Gurken local translation experiments. Another possible explanation
is inefficient split GFP translation or reconstitution in an in vivo context. While it is difficult to
directly and quantitatively test reconstitution efficiency in vivo, the reconstitution and translation
of split GFP components can be examined in vitro. For example, in our constant effort to develop
split reporters can reconstitute faster and more efficiently, we recently identified a novel GFP11
variant, GFP11-OPT, that seems to exhibit superior reconstitution efficiency when expressed in
cultured Drosophila S2 cells (Figure 4.6a). The reconstitution efficiency is assessed by
examining how early the fluorescence can be seen from cells, and the characteristics of the signal
including relative brightness and spatial distribution within the cell. Once an adequate candidate
is identified, such as GFP11-OPT, the peptide is commercially synthesized and the reconstitution
kinetics are measured and quantitatively compared to the currently available GFP11 variants.
It is therefore reasonable to consider that using GFP11-OPT instead of the current
reporter in the Grk-PQR-GFP11 transgenic fly and future animals, might produce more reliable
fluorescence signals that will enable unambiguous identification of local protein translation
events. Weakly expressed transcripts, or imaging sites with high autofluorescence can preclude
the high contrast detection of signals. Therefore, it is beneficial to have faster and more efficient
reconstitution reporters.
121
4.3.3 Detection of local protein translation in living neurons
My ultimate goal in developing this technique is to develop tools to visualize and
quantify local protein translation events in living neuronal subcellular compartments. Ionotropic
and metabotropic neurotransmitter receptors are some of the most abundant proteins present at
synapses and they mediate most of neuronal communication in the brain. Glutamate is the most
abundant excitatory neurotransmitter in the vertebrate nervous system, and glutamate receptor
signaling is the brain’s main excitatory signaling mechanism. The AMPA (α-amino-3-hydroxy-
5-methyl-4-isoxazolepropionic acid) receptor is an ionotropic transmembrane receptor for
glutamate that in addition to mediating synaptic transmission, has been implicated as a major
player in forms of synaptic plasticity that underlie learning and memory, cognition, long term
potentiation and synaptic scaling (Henley & Wilkinson, 2013; Ibata et al., 2008; Willard &
Koochekpour, 2013). AMPA receptors are composed of highly conserved subunits encoded by
genes termed GluR1-4 (Gria1-4), that combine to form tetramers that are trafficked into and out
of synapses (Willard & Koochekpour, 2013). AMPA receptor trafficking is mediated by the
interaction of subunit-specific proteins in addition to various post-translational modifications that
occur at their C-termini, and is known to be highly dynamic (Anggono & Huganir, 2012). The
number of AMPA receptors present at synapses and spines is directly dependent on the rates of
endocytosis and exocytosis at the post-synaptic membrane. Increases in synaptic activity result in
enhanced GluR1-containing AMPA receptor exocytosis and insertion, while increased receptor
endocytosis rates have been associated with synaptic long-term depression (Kessels & Malinow,
2009).
As mentioned in Chapter 1, some of the earliest dendrite-targeted mRNAs that were
identified included BDNF, Arc and αCaMKII. AMPA receptor subunit mRNAs have also been
122
found to localize to dendrites and spines, which suggests that AMPA receptor local translation
regulates the local abundance and composition of the synaptic receptor pool (Grooms et al.,
2006; Ju et al., 2004). Endogenous GluR1 mRNAs have been reported to be localized to
proximal and distal dendrites of rat hippocampal neurons (Grooms et al., 2006). In addition,
transfected tagged GluR1 and GluR2 subunits have been observed to be synthesized in dendritic
compartments, independently of protein synthesis in the cell soma (Ju et al., 2004).
The local translation of AMPA receptors in neurons is therefore an attractive system to
test our local protein synthesis reporters in vivo. Specifically, we can take advantage of the
characterized expression patterns of GluR1 AMPA receptor subunits and the extensive literature
covering the localization of GluR1 mRNA and its regulation by neuronal activity. This enables
the design proof of principle experiments that would allow the unambiguous detection of GluR1
local protein synthesis in living neurons.
The design and rationale of the experiment to visualize GluR1 local translation in
neurons is similar to that used to observe Gurken local translation in the fly oocyte. A genome
edited animal expressing GluR1-PQR-GFP11 is crossed to an animal expressing GFP1-10 and
fluorescence imaging of GFP in neurons is used as a spatial and temporal marker of local GluR1
production. To generate an animal that expresses a GFP11 reporter each time a GluR1 subunit is
translated, we used CRISPR genome editing to insert a PQR-GFP11 reporter at the end of the
coding sequence of the endogenous GluR1 locus (Gria1) in mouse pronuclei. Every cell that
expresses GluR1 subunits will proportionally co-produce GFP11 reporters. In the presence of
GFP1-10 protein, the split GFP components can reconstitute and emit fluorescence. The
fluorescent signal is thus used to mark the time and location of protein translation and the
fluorescence intensity can be used as a quantitative measure of the level of GluR1 production.
123
Injections of Cas9 protein, repair oligonucleotides and gRNAs targeted against the GluR1
locus have been performed and the resulting mice are currently being tested for the correct
insertion of PQR-GFP11 into the GluR1 locus. A simple validation experiment would entail
crossing the transgenic GFP1-10 and genome edited GluR1-PQR-GFP11 mice and imaging in
the F1 generation brain for the development of green fluorescence signals in distal neurites. To
validate any observed signals and to determine that they in fact reflect known GluR1 dynamics,
we can take advantage of characterized scenarios in which GluR1 is known to be locally
produced. For example, it has been shown that potentiation of hippocampal synaptic currents
using pharmacological agents, or by dopamine receptor activation increases the surface
expression of GluR1 subunits in mice (Doyle & Kiebler, 2011; Ju et al., 2004; Smith et al.,
2005). GluR1 mRNA has also been shown to localize to distal dendrites of rodent cortical and
hippocampal neurons (Chen, Onisko, & Napoli, 2008; Doyle & Kiebler, 2011; Grooms et al.,
2006; Muddashetty et al., 2007).
4.4 Detection of local protein synthesis using PQR photoconvertible reporters.
The use of fluorescence proteins to detect local protein synthesis events has been
hindered by the long folding and maturation times of common fluorescent proteins (S. Kim et al.,
2010; Shaner et al., 2005). Immature fluorescent proteins can diffuse up to 50 µm2/ sec in their
non-fluorescent state producing a high temporal and spatial discrepancy between the original site
of protein synthesis, and the site where the fluorescent signal is observed. To address this issue,
photoconvertible fluorescent proteins have been used to track local protein translation in cells
(Kim et al., 2010; Wang et al., 2009).
Kaede is a 28kDa photoconvertible fluorescent protein that exists as a 116 kDa
homotetramer in vivo (Ando et al., 2002). Kaede normally emits green (~518 nm) fluorescence
124
but can be permanently photoconverted to emit red (~580 nm) fluorescence. Along with Dendra2
and mEos2, it is one of the commonly used photoactivatable fluorescent proteins to study local
protein translation, particularly in neurons (S. Kim et al., 2010). By photoconverting areas of the
cell from green to red, the development of new green fluorescence is associated to the translation
of new protein. However, a simple back-of-the-envelope calculation of the diffusion time and
distance of homotetrameric Kaede in typical mammalian cytoplasm shows that the protein can
diffuse upwards of 60 microns in just 1 minute. Although the maturation time of Kaede has not
yet been accurately measured, we can assume for the sake of argument that Kaede has a
maturation time identical to one of the fastest measured fluorescence protein Venus, with a
measured maturation time of 2 minutes (Nagai et al., 2002; Tatavarty et al., 2012). In 2 minutes
nonfluorescent Kaede can diffuse 90 microns; easily traversing the length of an average
mammalian neuronal soma (20 microns (García-López et al., 2006)). In addition, some of the
largest dendritic spines have volumes of 1 µm3 (Nimchinsky et al., 2002) and are spaced 1 µm
apart in more extensively branched neurons. Using slow-maturing fluorescent proteins therefore
poses spatial and temporal constraints and the examples above serve to show that in such
experiments, where Kaede is seen is likely not exactly where it was made.
Using split-photoconvertible fluorescent proteins is the most suitable approach to detect
local protein translation repeatedly, and if optimized can result in highly quantitative spatial and
temporal measurements of protein production over time. A splitKaede or splitDendra2 system
would work similar to splitGFP reconstitution, where the Dendra2 protein would be split into
two non-fluorescent parts of unequal sizes (Figure 4.6b). The larger, slower folding portion part
would be expressed at high levels in all parts of the cell, similar to GFP1-10, and the remaining
polypeptide would be placed downstream of a gene of interest, separated by a PQR. When the
125
protein of interest is translated, Dendra2 reconstitution would occur in milliseconds and mark the
exact time, location and rate of protein translation. The photoconvertible property therefore acts
a reset switch, overcoming signal saturation issues and would be used to define a time zero for
monitoring new translation events, as seen by development of new green fluorescence. Such as
system would combine the numerous advantages of co-expression of reporters using PQR, the
fast reconstitution of a fluorescent protein as a marker of translation, and an arbitrary time zero at
which measurements can be started.
Splitting a fluorescent protein into two nonfluorescent parts that can spontaneously
reassemble with no co-factors or enzymes presents several problems. In order for a split FP
system to work properly in the context of local protein synthesis detection, it should satisfy some
basic criteria. First and most important, each protein half must not exhibit any activity on its
own. Second, the affinities of the split parts to each other must be sufficiently high to ensure
proper interaction, and third the reconstituted FP must have an easily quantifiable readout. While
most fluorescent proteins in theory can be split into interacting parts, finding regions along the
protein structure that result in partners that conserve the above criteria is less straightforward. If
split FP parts that exist in near native conformation can be generated, then their binding affinity
could be made significantly higher (Huang & Bystroff, 2009). This is counterproductive for
traditional protein interaction studies using split fluorescent proteins such as mGRASP, since the
reporter could reconstitute without the interaction of the assayed proteins. However, in the
context of developing a fast protein translation sensor, increased binding affinity means reduced
delay in reconstitution and fluorescence emission after translation of the reporter. The original
split GFP developed by Cabantous et al. relied solely on the binding of the 11th beta barrel for
fluorescence reconstitution to occur, however, the slow maturation of the chromophore led to the
126
resulting fluorescence requiring several hours to reach a plateau (Cabantous, Terwilliger, et al.,
2005). This problem stimulated efforts to find variants with shorter chromophore maturation
times, and using DNA shuffling (Stemmer, 1994), the Waldo group developed GFP1-10
optimum (GFP1-10 OPT) which had 80-fold better reconstitution efficiency compared to the
original split GFP (Cabantous, Pédelacq, et al., 2005).
Experiments to split Dendra2 into non-fluorescent and spontaneously interacting partners
that can be used as local protein translation reporters are in progress and we have successfully
split the Dendra2 molecule into two nonfluorescent parts, which reconstitute to form a green
fluorescent protein that can be permanently photoconverted to emit red fluorescence using
ultraviolet light (Figure 4.6). High homology between the Dendra2 and GFP 3D crystal
structures allowed us to take advantage of the wealth of information we have on splitting and
folding GFP, and applied it to Dendra2. Similar to GFP, Dendra2 is composed of beta turn
secondary structures and beta sheets that envelop the chromophore in a barrel-like structure. This
allowed the comparison of the Dendra2 and GFP individual protein domains in parallel, which
offered locations along the sequence where the protein can in theory be split to generate
reconstitution partner candidates. Specifically, we split Dendra2 between the 190th and 191st
residue (Figure 4.6b) and verified that they could spontaneously reconstitute and emit green (~
507 nm) fluorescence by expressing the larger and smaller Dendra2 fragments in HEK293 cells
(Figure 4.6c).
Upon exposure to a flash of ultraviolet (~350 nm) light, we found that reconstituted
Dendra2 could be permanently photoconverted to emit red (~573 nm) fluorescence (Figure 4.6c).
By quantifying the levels of green and red fluorescence over time, we observed that immediately
following photoconversion, red fluorescence intensities rose and steadily declined over time,
127
while green fluorescence intensities dropped and then steadily rose over tens of minutes (Figure
4.6c). These results indicate that reconstituted green Dendra2 was permanently photoconverted,
and rising green fluorescence intensities within minutes post photoconversion suggest that new
Dendra 11 synthesis and reconstitution events could be observed immediately after
photoconversion. Importantly, no increases in fluorescence intensity were observed in the red
channel post photoconversion (Figure 4.6c), which suggests that unreconstituted Dendra1-10
does not photoconvert in the absence of Dendra11, and still emits green fluorescence when
reconstituted. This allows split Dendra2 to be used repetitively as a reset switch to constantly
allow observation of new protein synthesis events in the green channel.
128
4.5 Figures
a
b
Figure 4.1: PQR reporters are inserted in-frame into endogenous genes.
(a) An edited IgK locus showing the location of the insertion of a PQR-GFP reporter. The kappa
immunoglobulin locus is composed of 174 variable genes, 5 joining genes and 1 constant gene
that rearrange to form the mature light chain. One variable gene is joined to one joining gene and
both are joined to the constant gene. (b) The reporter is inserted at the end of the coding
sequence of the constant region, to avoid disruption of the open reading frame and result in
stoichiometric coexpression of GFP with each molecule of IgK.
129
a
b
Figure 4.2: PQR constructs injected into mouse embryos result in red fluorescent
pronuclei.
(a) Preliminary results showing mouse pronuclei that were co-injected with Rpl13a-specific
CRISPR and repair plasmids. Insertion of PQR-RFPnols reporters into the endogenous RPL13A
locus was achieved and results in fluorescent embryos. Injections and implantation of red-
fluorescent embryos are ongoing in collaboration with Mitra Cowan and the Goodman Cancer
research centre. Scale bar is 100 µm. (b) Hypothetical results obtained from infection of a
Rpl13a-PQR-RFPnols/IgK-PQR-GFP double knock-in mouse. Green indicates antibody response,
red indicate housekeeping protein production. Changes in IgK production between cells or
130
experiments can be normalized to Rpl13a production levels in those same cells, to reflect the net
increase in IgK protein production.
131
Figure 4.3: SplitGFP as a protein translation reporter.
Insertion of a PQR sequence between a gene of interest and a GFP11 reporter results in slightly
longer mRNAs being transcribed (< 100 bases), and stoichiometric translation of GFP11. Each
time a molecule of protein of interest is translated, a molecule of GFP11 is co-produced. In the
presence of GFP1-10, the split GFP parts reconstitute to form fluorescent GFP.
132
Figure 4.4: GFP1-10 is expressed at high levels in transgenic animals.
(a) Immunoblot performed using α-GFP antibody on whole fly and brain and head extracts. The
24 kDa predicted GFP1-10 protein is properly expressed in flies and runs at the expected size.
(b-c) Immunohistochemistry done in collaboration with Sejal Davla, showing ubiquitous high
level expression of GFP1-10 protein in the fly adult brain (b) and larval ventral cord (c). GFP1-
10 can clearly be seen in cell somas (arrows) and projections (arrowheads). Anti-GFP antibody
(cyan) was used to probe for GFP1-10 (cyan), RFP marks projections, nc82 is a marker of
synapses. Scale bar is 50 µm in (b) and 75 µm in (c).
133
134
Figure 4.5: GFP11 can detect Gurken local translation in oocytes.
Representative images of local Gurken translation in Drosophila oocytes. (a) The grk transcript
localizes to the anterior dorsal corner of the oocyte near the nucleus, where its translation is
initiated. (b-c) Translation of Grk-PQR-GFP11 shows green fluorescence (arrows) that is always
associated with the anterodorsal nucleus (star) in stage 9 (b) and stage 10 (c). (d-e) Control
animals in which GFP1-10 (d) or Gal4 (e) are not expressed show complete lack of the
characteristic signal that associates with the oocyte nucleus (n>25). Anterior is left, dorsal is up.
Scale bar is 10 µm.
135
136
Figure 4.6: Novel split fluorescent reporters exhibit more efficient reconstitution.
(a) GFP11 reporters can be improved to produce more efficient GFP reconstitution in cultured
S2 cells. (b) Illustration showing an example of splitting Dendra2 into two non-fluorescent
reconstitution partners. The protein is split into a larger, slower folding non-fluorescent part and
the remaining smaller polypeptide. The two are separately nonfluorescent but spontaneously
reconstitute to form the native fluorescent Dendra2 structure and emit fluorescence. In the
example above, Dendra2 is split between the 190th and 191st to generate the first split
photoconvertible Dendra2 reconstitution partners.
137
Chapter V - Thesis directions and conclusions
138
Proteins execute essential functions in cells. Dysregulation in protein production is a
hallmark feature of many diseases. Tremendous progress has been made towards understanding
when and where gene products are produced, yet we still know little about the complete picture
that is how the expression of genes, and their regulation contribute to the development of cellular
phenotypes. In this thesis, I have expanded the capabilities of our PQR technique by developing
alternative and new ways to examine the expression of genes and subsequent production of
proteins. To pave the way for understanding how proteins execute their functions in cells, I have
taken a two-pronged approach to examine the process of gene expression. By simultaneously
examining the levels of RNA and protein of a gene from a single cell, I have developed a system
to determine how individual cells vary in their transcriptional and translational landscapes within
a population.
My second approach to understand protein production aimed at directly observing the
process of new protein creation. The animals that will be generated to demonstrate endogenous
protein production measurement in vivo, the IgK and Rpl13a knockin mice, are useful not only as
proof of principle experiments in my work, but will also be of use to the wider research
community. Rpl13a is a well-established reference gene whose expression is used to reflect both
the transcriptional and translational status of cells. Therefore, the Rpl13a knockin mouse
provides researchers quantifying the expression of their gene of interest, from isolated cells or in
vivo, to normalize any changes observed to the general transcriptional and translational status of
the cell. The mouse represents a resource that complements, instead of replacing current
approaches used to study the production of proteins. The mouse is amenable to both antibody-
and fluorescence-based measurement and normalization of protein levels, which opens the door
139
for longer in vivo imaging of the production of proteins in addition to independent means of
quantifying and validating collected data.
The IgK mouse is a translational reporter of immune function. An increasing number of studies
now track immunological responses at the whole-body level in living mice, and this has revealed
a number of previously unknown mechanisms underlying variability in adaptive immune
responses (Nair-Gill et al., 2008), tissue transplantation (Donahue et al., 2015), infection (Prado
et al., 2015; Santangelo et al., 2015) and cancer growth (Edinger et al., 2003). Expression of
genetically encoded fluorescent reporters from endogenous antibody loci allows the optical
tracking of native immune responses at the microscopic and macroscopic levels, without the
need to laboriously label cells ex vivo and reintroduce them into the animal, one of the common
ways of examining immune responses in a living animal (Germain et al., 2006). Moreover,
genetically tagging antibody chains with fluorescent reporters opens the door for multicolor
imaging of immune responses, which allows the easy identification, screening and isolation of B
cells that express varying levels of antibodies of specific types produced against specific
responses. Ultimately, this mouse is most useful for the screening and development of
therapeutics that modulate immune responses. For example, the development of an immune
response against a pathogen in the IgK mouse can be optically monitored over time by observing
changes in GFP fluorescence intensity from B cells in the lymphatic system (Figure 4.2b).
Similarly, the progression of the infection or disease and efficacy of therapy can be optically
monitored in that same mouse, providing unprecedented resolution into the mechanisms that
mediate heterogeneous individual immune responses (Satija et al., 2014).
The concept of local protein translation in neurons has been examined over decades, and
only now are we beginning to unravel its mechanisms and understand its impact. My strategy to
140
use split fluorescent proteins as reporters of small protein translation events, is particularly well
suited to examine the occurrence of these events in neuronal cells. The fast reconstitution time,
linear dependence of the brightness of the signal on the concentration of reporter and
stoichiometric production of the genetically encoded reporter makes the technique amenable to
long-term in vivo imaging of protein production in neuronal subcellular compartments, which
will reveal unprecedented insight into the dynamics of gene expression response far from the cell
nucleus.
In continuing our progression of developing the PQR technique, we have found that local
protein translation reporters could be enhanced by optimizing their sequences for organism-
tailored split reporters. These reporters are more efficient at reconstitution in terms of speed,
brightness and efficiency (Figure 4.6a). Finally, our efforts to create split photoconvertible
fluorescent proteins may offer a complete shift in experiments examining protein production in
living cells. By photoconverting cells or subcellular compartments to a different spectrum, the
dynamics of protein production can be assayed from the same cell over the life of the animal,
with no issues of signal saturation or small dynamic range, in multiple channels. This type of
extended information obtained from single cells is extremely valuable. We know that cellular
heterogeneity could explain a substantial portion of the variability seen in assays conducted at
the tissue or population level (J Eberwine et al., 2001; Schubert, 2011). Upgraded reporters bring
increased spatial and temporal accuracy in determining protein levels, and so by ever-improving
reporters we can improve the quality and value of data that is collected. Similar to the efforts that
led to brighter and more stable GFPs, or the constant adaptation and expansion of optogenetic
techniques (Deisseroth, 2015; Remington, 2011; Tsien, 1998), I believe that the story of our PQR
141
technique has just begun. Protein analysis in single cells is certain to keep improving and open
new ways to understand cell biology.
142
References
Ables, E. T. (2015). Drosophila Oocytes as a Model for Understanding Meiosis: An Educational
Primer to Accompany “Corolla Is a Novel Protein That Contributes to the Architecture of
the Synaptonemal Complex of Drosophila.” Genetics, 199(1).
Abrahamsen, H. N., Nexo, E., Steiniche, T., Hamilton-Dutoit, S. J., & Sorensen, B. S. (2004).
Quantification of Melanoma mRNA Markers in Sentinel Nodes. The Journal of Molecular
Diagnostics, 6(3), 253–259. https://doi.org/10.1016/S1525-1578(10)60518-1
Alberts, B., Johnson, A., Lewis, J., Raff, M., Roberts, K., & Walter, P. (2002). Molecular
Biology of the Cell, Fourth Edition. Molecular Biology. https://doi.org/citeulike-article-
id:691434
Alwine, J. C., Kemp, D. J., & Stark, G. R. (1977). Method for detection of specific RNAs in
agarose gels by transfer to diazobenzyloxymethyl-paper and hybridization with DNA
probes. Proceedings of the National Academy of Sciences of the United States of America,
74(12), 5350–4. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/414220
Ando, R., Hama, H., Yamamoto-Hino, M., Mizuno, H., & Miyawaki, A. (2002). An optical
marker based on the UV-induced green-to-red photoconversion of a fluorescent protein.
Proceedings of the National Academy of Sciences of the United States of America, 99(20),
12651–6. https://doi.org/10.1073/pnas.202320599
Anggono, V., & Huganir, R. L. (2012). Regulation of AMPA receptor trafficking and synaptic
plasticity. Current Opinion in Neurobiology, 22(3), 461–9.
https://doi.org/10.1016/j.conb.2011.12.006
Bagni, C., Tassone, F., Neri, G., Hagerman, R., Martin, J., Bell, J., … Narayanan, U. (2012).
Fragile X syndrome: causes, diagnosis, mechanisms, and therapeutics. The Journal of
Clinical Investigation, 122(12), 4314–22. https://doi.org/10.1172/JCI63141
Bagshaw, C. R., & Cherny, D. (2006). Blinking fluorophores: what do they tell us about protein
dynamics? Biochemical Society Transactions, 34(Pt 5), 979–982.
https://doi.org/10.1042/BST0340979
Baker, M. (2015). Reproducibility crisis: Blame it on the antibodies. Nature, 521(7552), 274–
276. https://doi.org/10.1038/521274a
Bakshi, S., Siryaporn, A., Goulian, M., & Weisshaar, J. C. (2012). Superresolution imaging of
ribosomes and RNA polymerase in live Escherichia coli cells. Molecular Microbiology,
85(1), 21–38. https://doi.org/10.1111/j.1365-2958.2012.08081.x
Bassell, G. J., & Warren, S. T. (2008). Fragile X Syndrome: Loss of Local mRNA Regulation
Alters Synaptic Development and Function. Neuron.
https://doi.org/10.1016/j.neuron.2008.10.004
Belle, A., Tanay, A., Bitincka, L., Shamir, R., & O’Shea, E. K. (2006). Quantification of protein
half-lives in the budding yeast proteome. Proceedings of the National Academy of Sciences
of the United States of America, 103(35), 13004–9.
https://doi.org/10.1073/pnas.0605420103
Benito, J., Zheng, H., Ng, F. S., & Hardin, P. E. (2007). Transcriptional feedback loop
regulation, function, and ontogeny in Drosophila. Cold Spring Harbor Symposia on
Quantitative Biology, 72, 437–44. https://doi.org/10.1101/sqb.2007.72.009
Besse, F., & Ephrussi, A. (2008). Translational control of localized mRNAs: restricting protein
synthesis in space and time. Nature Reviews Molecular Cell Biology, 9(12), 971–980.
https://doi.org/10.1038/nrm2548
Boulina, M., Samarajeewa, H., Baker, J. D., Kim, M. D., & Chiba, A. (2013). Live imaging of
143
multicolor-labeled cells in Drosophila. Development, 140(7).
Brand, A. H., & Perrimon, N. (1993). Targeted gene expression as a means of altering cell fates
and generating dominant phenotypes. Development (Cambridge, England), 118(2), 401–15.
Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/8223268
Brown, A. D., Claybon, A. B., & Bishop, A. J. R. (2011). A conditional mouse model for
measuring the frequency of homologous recombination events in vivo in the absence of
essential genes. Molecular and Cellular Biology, 31(17), 3593–3602.
https://doi.org/10.1128/MCB.00848-10
Bustin, S. A. (2000). Absolute quantification of mrna using real-time reverse transcription
polymerase chain reaction assays. Journal of Molecular Endocrinology.
https://doi.org/JME00927 [pii]
Bustin, S. A., Benes, V., Garson, J. A., Hellemans, J., Huggett, J., Kubista, M., … Wittwer, C. T.
(2009). The MIQE guidelines:Minimum Information for publication of quantitative real-
time PCR experiments. Clinical Chemistry, 55(4), 611–622.
https://doi.org/10.1373/clinchem.2008.112797
Bustin, S. A., & Nolan, T. (2004). Pitfalls of quantitative real-time reverse-transcription
polymerase chain reaction. Journal of Biomolecular Techniques : JBT, 15(3), 155–66.
https://doi.org/15/3/155 [pii]
Bustin, S., & Nolan, T. (2004, September). Pitfalls of quantitative real- time reverse-transcription
polymerase chain reaction. Journal of Biomolecular Techniques. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/15331581
Cabantous, S., Pédelacq, J. D., Mark, B. L., Naranjo, C., Terwilliger, T. C., & Waldo, G. S.
(2005). Recent advances in GFP folding reporter and split-GFP solubility reporter
technologies. Application to improving the folding and solubility of recalcitrant proteins
from Mycobacterium tuberculosis. Journal of Structural and Functional Genomics, 6(2–3),
113–119. https://doi.org/10.1007/s10969-005-5247-5
Cabantous, S., Terwilliger, T. C., & Waldo, G. S. (2005). Protein tagging and detection with
engineered self-assembling fragments of green fluorescent protein. Nature Biotechnology,
23(1), 102–7. https://doi.org/10.1038/nbt1044
Cabantous, S., & Waldo, G. S. (2006). In vivo and in vitro protein solubility assays using split
GFP. Nature Methods, 3(10), 845–854. https://doi.org/10.1038/nmeth932
Cáceres, L., & Nilson, L. A. (2005). Production of gurken in the nurse cells is sufficient for axis
determination in the Drosophila oocyte. Development (Cambridge, England), 132(10),
2345–53. https://doi.org/10.1242/dev.01820
Callahan, B. P., & Miller, B. G. (2007). OMP decarboxylase—An enigma persists. Bioorganic
Chemistry, 35(6), 465–469. https://doi.org/10.1016/j.bioorg.2007.07.004
Cambridge, S. B., Gnad, F., Nguyen, C., Bermejo, J. L., Krüger, M., & Mann, M. (2011).
Systems-wide proteomic analysis in mammalian cells reveals conserved, functional protein
turnover. Journal of Proteome Research, 10(12), 5275–84.
https://doi.org/10.1021/pr101183k
Chen, N., Onisko, B., & Napoli, J. L. (2008). The nuclear transcription factor RARalpha
associates with neuronal RNA granules and suppresses translation. The Journal of
Biological Chemistry, 283(30), 20841–7. https://doi.org/10.1074/jbc.M802314200
Cho, S. W., Kim, S., Kim, Y., Kweon, J., Kim, H. S., Bae, S., & Kim, J. S. (2014). Analysis of
off-target effects of CRISPR/Cas-derived RNA-guided endonucleases and nickases.
Genome Research, 24(1), 132–141. https://doi.org/10.1101/gr.162339.113
144
Chudakov, D. M., Matz, M. V., Lukyanov, S., & Lukyanov, K. A. (2010). Fluorescent Proteins
and Their Applications in Imaging Living Cells and Tissues. Physiological Reviews, 90(3).
Cody, C. W., Prasher, D. C., Westler, W. M., Prendergast, F. G., & Ward, W. W. (1993).
Chemical structure of the hexapeptide chromophore of the Aequorea green-fluorescent
protein. Biochemistry, 32(1979), 1212–1218. https://doi.org/10.1021/bi00056a003
Cox, L. J., Hengst, U., Gurskaya, N. G., Lukyanov, K. A., & Jaffrey, S. R. (2008). Intra-axonal
translation and retrograde trafficking of CREB promotes neuronal survival. Nature Cell
Biology, 10(2), 149–59. https://doi.org/10.1038/ncb1677
Craggs, T. D. (2009). Green fluorescent protein: structure, folding and chromophore maturation.
Chemical Society Reviews, 38(10), 2865–75. https://doi.org/10.1039/b903641p
Craig, F. F., Simmonds, A. C., Watmore, D., McCapra, F., & White, M. R. (1991). Membrane-
permeable luciferin esters for assay of firefly luciferase in live intact cells. The Biochemical
Journal, 276 ( Pt 3, 637–41. Retrieved from
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1151052&tool=pmcentrez&ren
dertype=abstract
Crameri, A., Whitehorn, E. A., Tate, E., & Stemmer, W. P. C. (1996). Improved green
fluorescent protein by molecular evolution using DNA shuffling. Nature Biotechnology,
14(3), 315–319. https://doi.org/10.1038/nbt0396-315
Cubitt, a B., Heim, R., Adams, S. R., Boyd, a E., Gross, L. a, & Tsien, R. Y. (1995).
Understanding, improving and using green fluorescent proteins. Trends in Biochemical
Sciences, 20(11), 448–55. https://doi.org/10.1016/S0968-0004(00)89099-4
Curtis, K. M., Gomez, L. A., Rios, C., Garbayo, E., Raval, A. P., Perez-Pinzon, M. A., …
Horwitz, E. (2010). EF1alpha and RPL13a represent normalization genes suitable for RT-
qPCR analysis of bone marrow derived mesenchymal stem cells. BMC Molecular Biology,
11(1), 61. https://doi.org/10.1186/1471-2199-11-61
Das, D., Georgiadis, M. M., Avidan, O., Loya, S., Tonjes, R. ., Sevilya, Z., … Steitz, T. . (2004).
The crystal structure of the monomeric reverse transcriptase from Moloney murine
leukemia virus. Structure (London, England : 1993), 12(5), 819–29.
https://doi.org/10.1016/j.str.2004.02.032
de Felipe, P., Luke, G. a, Hughes, L. E., Gani, D., Halpin, C., & Ryan, M. D. (2006). E unum
pluribus: multiple proteins from a self-processing polyprotein. Trends in Biotechnology,
24(2), 68–75. https://doi.org/10.1016/j.tibtech.2005.12.006
De Groot, A. S., & Scott, D. W. (2007). Immunogenicity of protein therapeutics. Trends in
Immunology. https://doi.org/10.1016/j.it.2007.07.011
Deisseroth, K. (2015). Optogenetics: 10 years of microbial opsins in neuroscience. Nature
Neuroscience, 18(9), 1213–1225. https://doi.org/10.1038/nn.4091
Dhanasekaran, S., Doherty, T. M., Kenneth, J., & TB Trials Study Group. (2010). Comparison of
different standards for real-time PCR-based absolute quantification. Journal of
Immunological Methods, 354(1–2), 34–9. https://doi.org/10.1016/j.jim.2010.01.004
Ding, C., & Cantor, C. R. (2004). Quantitative analysis of nucleic acids--the last few years of
progress. Journal of Biochemistry and Molecular Biology, 37(1), 1–10.
https://doi.org/<ARTICLE_ID IdType=doi> [pii]
Dittrich, P. S., Schäfer, S. P., & Schwille, P. (2005). Characterization of the photoconversion on
reaction of the fluorescent protein Kaede on the single-molecule level. Biophysical Journal,
89(5), 3446–55. https://doi.org/10.1529/biophysj.105.061713
Do, K., & Boxer, S. G. (2011). Thermodynamics, kinetics, and photochemistry of β-strand
145
association and dissociation in a split-GFP system. Journal of the American Chemical
Society, 133(45), 18078–81. https://doi.org/10.1021/ja207985w
Dobson, C. M. (2001). The structural basis of protein folding and its links with human disease.
Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences,
356(1406), 133–45. https://doi.org/10.1098/rstb.2000.0758
Donahue, R. E., Srinivasula, S., Uchida, N., Kim, I., St. Claire, A., Duralde, G., … Di Mascio,
M. (2015). Whole-body imaging of the immune system reveals discordance in CD4+
lymphoid tissue recovery after transplantation in rhesus macaques. Blood.
Dormoy-Raclet, V., Ménard, I., Clair, E., Kurban, G., Mazroui, R., Di Marco, S., … Gallouzi, I.-
E. (2007). The RNA-binding protein HuR promotes cell migration and cell invasion by
stabilizing the beta-actin mRNA in a U-rich-element-dependent manner. Molecular and
Cellular Biology, 27(15), 5365–5380. https://doi.org/10.1128/MCB.00113-07
Doyle, M., & Kiebler, M. A. (2011). Mechanisms of dendritic mRNA transport and its role in
synaptic tagging. The EMBO Journal, 30(17), 3540–52.
https://doi.org/10.1038/emboj.2011.278
Driever, W., & Nüsslein-Volhard, C. (1988). A gradient of bicoid protein in Drosophila
embryos. Cell, 54(1), 83–93. https://doi.org/10.1016/0092-8674(88)90182-1
Duffy, J. B. (2002). GAL4 system in Drosophila: a fly geneticist’s Swiss army knife. Genesis
(New York, N.Y. : 2000), 34(1–2), 1–15. https://doi.org/10.1002/gene.10150
Eberwine, J., Kacharmina, E. J., Andrews, C., Miyashiro, K., McIntosh, T., Becker, K., …
Marciano, P. (2001). mRNA Expression Analysis of Tissue Sections and Single Cells. The
Journal of Neuroscience, 21(21), 8310–8314.
Eberwine, J., Miyashiro, K., Kacharmina, J. E., & Job, C. (2001). Local translation of classes of
mRNAs that are targeted to neuronal dendrites. Proceedings of the National Academy of
Sciences of the United States of America, 98(13), 7080–5.
https://doi.org/10.1073/pnas.121146698
Edinger, M., Cao, Y.-A., Verneris, M. R., Bachmann, M. H., Contag, C. H., & Negrin, R. S.
(2003). Revealing lymphoma growth and the efficacy of immune cell therapies using in
vivo bioluminescence imaging. Blood, 101(2).
Feige, M. J., Hendershot, L. M., & Buchner, J. (2010). How antibodies fold. Trends in
Biochemical Sciences. https://doi.org/10.1016/j.tibs.2009.11.005
Feinberg, E. H., VanHoven, M. K., Bendesky, A., Wang, G., Fetter, R. D., Shen, K., &
Bargmann, C. I. (2008). GFP Reconstitution Across Synaptic Partners (GRASP) Defines
Cell Contacts and Synapses in Living Nervous Systems. Neuron, 57(3), 353–363.
https://doi.org/10.1016/j.neuron.2007.11.030
Flatz, L., Roychoudhuri, R., Honda, M., Filali-Mouhim, A., Goulet, J. P., Kettaf, N., … Nabel,
G. J. (2011). Single-cell gene-expression profiling reveals qualitatively distinct CD8 T cells
elicited by different gene-based vaccines. Proc Natl Acad Sci U S A, 108(14), 5724–5729.
https://doi.org/10.1073/pnas.1013084108
Fowler, S. J. (1996). Protein Staining and Immunodetection Using Colloidal Gold. In The
Protein Protocols Handbook (pp. 275–287). Totowa, NJ: Humana Press.
https://doi.org/10.1007/978-1-60327-259-9_44
Fu, Y., Foden, J. a, Khayter, C., Maeder, M. L., Reyon, D., Joung, J. K., & Sander, J. D. (2013).
High-frequency off-target mutagenesis induced by CRISPR-Cas nucleases in human cells.
Nature Biotechnology, 31(9), 822–6. https://doi.org/10.1038/nbt.2623
Fujita, S. C., Zipursky, S. L., Benzer, S., Ferrús, A., & Shotwell, S. L. (1982). Monoclonal
146
antibodies against the Drosophila nervous system. Proceedings of the National Academy of
Sciences of the United States of America, 79(24), 7929–33. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/6818557
García-López, P., García-Marín, V., & Freire, M. (2006). Three-dimensional reconstruction and
quantitative study of a pyramidal cell of a Cajal histological preparation. Journal of
Neuroscience, 26(44), 11249–11252. https://doi.org/10.1523/JNEUROSCI.3543-06.2006
Gavis, E. R. (1995). Pattern formation. Gurken meets torpedo for the first time. Current
Biology : CB, 5(11), 1252–4. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/8574581
Germain, R. N., Miller, M. J., Dustin, M. L., & Nussenzweig, M. C. (2006). Dynamic imaging of
the immune system: progress, pitfalls and promise. Nat Rev Immunol, 6(7), 497–507.
https://doi.org/10.1038/nri1884
Giorgi, C., & Moore, M. J. (2007). The nuclear nurture and cytoplasmic nature of localized
mRNPs. Seminars in Cell and Developmental Biology.
https://doi.org/10.1016/j.semcdb.2007.01.002
Gissel, C., Doutheil, J., & Paschen, W. (1997). Temporal analysis of changes in neuronal c-fos
mRNA levels induced by depletion of endoplasmic reticulum calcium stores: effect of
clamping cytoplasmic calcium activity at resting levels. Journal of Neurochemistry, 69(6),
2538–45. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9375687
Glinka, M., Herrmann, T., Funk, N., Havlicek, S., Rossoll, W., Winkler, C., & Sendtner, M.
(2010). The heterogeneous nuclear ribonucleoprotein-R is necessary for axonal β-actin
mRNA translocation in spinal motor neurons. Human Molecular Genetics, 19(10), 1951–
1966. https://doi.org/10.1093/hmg/ddq073
Gordon, M. D., & Scott, K. (2009). Motor Control in a Drosophila Taste Circuit. Neuron, 61(3),
373–384. https://doi.org/10.1016/j.neuron.2008.12.033
Greenbaum, D., Colangelo, C., Williams, K., & Gerstein, M. (2003). Comparing protein
abundance and mRNA expression levels on a genomic scale. Genome Biology, 4(9), 117.
Gronowicz, G., Hadjimichael, J., Richards, D., Cerami, A., & Rossomando, E. F. (1992).
Correlation between tumor necrosis factor-alpha (TNF-alpha)-induced cytoskeletal changes
and human collagenase gene induction. Journal of Periodontal Research, 27(6), 562–8.
Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/1334144
Grooms, S. Y., Noh, K.-M., Regis, R., Bassell, G. J., Bryan, M. K., Carroll, R. C., & Zukin, R. S.
(2006). Activity bidirectionally regulates AMPA receptor mRNA abundance in dendrites of
hippocampal neurons. The Journal of Neuroscience : The Official Journal of the Society for
Neuroscience, 26(32), 8339–51. https://doi.org/10.1523/JNEUROSCI.0472-06.2006
Grueber, W. B., Jan, L. Y., Jan, Y. N., & Francisco, S. (2003). Different Levels of the
Homeodomain Protein Cut Regulate Distinct Dendrite Branching Patterns of Drosophila
Multidendritic Neurons, 112, 805–818.
Gubern, C., Hurtado, O., Rodríguez, R., Morales, J. R., Romera, V. G., Moro, M. A., …
Davalos, A. (2009). Validation of housekeeping genes for quantitative real-time PCR in in-
vivo and in-vitro models of cerebral ischaemia. BMC Molecular Biology, 10(1), 57.
https://doi.org/10.1186/1471-2199-10-57
Guigas, G., Kalla, C., & Weiss, M. (2007). Probing the nanoscale viscoelasticity of intracellular
fluids in living cells. Biophysical Journal, 93(1), 316–23.
https://doi.org/10.1529/biophysj.106.099267
Gunawardana, Y., Fujiwara, S., Takeda, A., Woo, J., Woelk, C., & Niranjan, M. (2015). Outlier
147
detection at the transcriptome-proteome interface. Bioinformatics (Oxford, England),
31(15), 2530–6. https://doi.org/10.1093/bioinformatics/btv182
Gunning, P., Leavittt, J., Muscat, G., Ngt, S.-Y., & Kedes, L. (1987). A human f3-actin
expression vector system directs high-level accumulation of antisense transcripts.
Biochemistry, 84, 4831–4835.
Haas, I. G., & Wabl, M. (1983). Immunoglobulin heavy chain binding protein. Nature,
306(5941), 387–9. https://doi.org/10.1038/306387a0
Hansen, M. C., Palmer, R. J., Udsen, C., White, D. C., Molin, S., & Hansen, M. C. (2016).
Assessment of GFP fluorescence in cells of Streptococcus gordonii under conditions of low
pH and low oxygen concentration. Microbiology, 2337(147), 23–1383.
Hastings, M. H. (2005). Circadian biology: Fibroblast clocks keep ticking. Current Biology.
https://doi.org/10.1016/j.cub.2004.12.012
He, B., & Soderlund, D. M. (2010). Human embryonic kidney (HEK293) cells express
endogenous voltage-gated sodium currents and Nav1.7 sodium channels. Neuroscience
Letters, 469(2), 268–272. https://doi.org/10.1016/j.neulet.2009.12.012
Heim, R., Cubitt, a B., & Tsien, R. Y. (1995). Improved green fluorescence. Nature.
https://doi.org/10.1038/373663b0
Helfman, P. M., & Bada, J. L. (1975). Aspartic acid racemization in tooth enamel from living
humans. Proceedings of the National Academy of Sciences of the United States of America,
72(8), 2891–4. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/1059082
Helfman, P. M., & Bada, J. L. (1976). Aspartic acid racemisation in dentine as a measure of
ageing. Nature, 262(5566), 279–281. https://doi.org/10.1038/262279b0
Hengst, U., & Jaffrey, S. R. (2007). Function and translational regulation of mRNA in
developing axons. Seminars in Cell and Developmental Biology.
https://doi.org/10.1016/j.semcdb.2007.01.003
Henley, J. M., & Wilkinson, K. A. (2013). AMPA receptor trafficking and the mechanisms
underlying synaptic plasticity and cognitive aging. Dialogues in Clinical Neuroscience,
15(1), 11–27. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/23576886
Higuchi, R., Fockler, C., Dollinger, G., & Watson, R. (1993). Kinetic PCR analysis: real-time
monitoring of DNA amplification reactions. Bio/technology.
https://doi.org/10.1038/nbt0993-1026
Holland, P. M., Abramson, R. D., Watson, R., & Gelfand, D. H. (1991). Detection of specific
polymerase chain reaction product by utilizing the 5’----3’ exonuclease activity of Thermus
aquaticus DNA polymerase. Proceedings of the National Academy of Sciences of the United
States of America, 88(16), 7276–80. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/1871133
Holtkamp, W., Kokic, G., Jäger, M., Mittelstaet, J., Komar, A. A., & Rodnina, M. V. (2015).
Cotranslational protein folding on the ribosome monitored in real time. Science (New York,
N.Y.), 350(6264), 1104–7. https://doi.org/10.1126/science.aad0344
Houdebine, L. M., & Attal, J. (1999). Internal ribosome entry sites (IRESs): reality and use.
Transgenic Research, 8(3), 157–77. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/10478488
Htun, H., Barsony, J., Renyi, I., Gould, D. L., & Hager, G. L. (1996). Visualization of
glucocorticoid receptor translocation and intranuclear organization in living cells with a
green fluorescent protein chimera. Proceedings of the National Academy of Sciences of the
United States of America, 93(10), 4845–50. Retrieved from
148
http://www.ncbi.nlm.nih.gov/pubmed/8643491
Huang, Y., & Bystroff, C. (2009). Complementation and Reconstitution of Fluorescence from
Circularly Permuted and Truncated Green Fluorescent Protein †. Biochemistry, 48(5), 929–
940. https://doi.org/10.1021/bi802027g
Ibata, K., Sun, Q., & Turrigiano, G. G. (2008). Rapid Synaptic Scaling Induced by Changes in
Postsynaptic Firing. Neuron, 57(6), 819–826. https://doi.org/10.1016/j.neuron.2008.02.031
Iizuka, R., Yamagishi-Shirasaki, M., & Funatsu, T. (2011). Kinetic study of de novo
chromophore maturation of fluorescent proteins. Analytical Biochemistry, 414(2), 173–178.
https://doi.org/10.1016/j.ab.2011.03.036
Ingolia, N. T., Lareau, L. F., & Weissman, J. S. (2011). Ribosome profiling of mouse embryonic
stem cells reveals the complexity and dynamics of mammalian proteomes. Cell, 147(4),
789–802. https://doi.org/10.1016/j.cell.2011.10.002
Inouye, S., & Tsuji, F. I. (1994). Evidence for redox forms of the Aequorea green fluorescent
protein. FEBS Letters, 351(2), 211–214. https://doi.org/10.1016/0014-5793(94)00859-0
Janeway, C. A., Travers, P., Walport, M., & Shlomchik, M. (2001). Immunobiology: The
Immune System In Health And Disease. Immuno Biology 5. https://doi.org/10.1111/j.1467-
2494.1995.tb00120.x
Jaramillo, A. M., Weil, T. T., Goodhouse, J., Gavis, E. R., & Schupbach, T. (2008). The
dynamics of fluorescently labeled endogenous gurken mRNA in Drosophila. Journal of Cell
Science, 121(Pt 6), 887–94. https://doi.org/10.1242/jcs.019091
Jinek, M., Chylinski, K., Fonfara, I., Hauer, M., Doudna, J. A., & Charpentier, E. (2012). A
Programmable Dual-RNA – Guided DNA Endonuclease in Adaptive Bacterial Immunity.
Science, 337(August), 816–822. https://doi.org/10.1126/science.1225829
Johnson, L. N. (2009). The regulation of protein phosphorylation. Biochemical Society
Transactions, 37(Pt 4), 627–41. https://doi.org/10.1042/BST0370627
Ju, W., Morishita, W., Tsui, J., Gaietta, G., Deerinck, T. J., Adams, S. R., … Malenka, R. C.
(2004). Activity-dependent regulation of dendritic synthesis and trafficking of AMPA
receptors. Nature Neuroscience, 7(3), 244–53. https://doi.org/10.1038/nn1189
Jung, H., Yoon, B. C., & Holt, C. E. (2012). Axonal mRNA localization and local protein
synthesis in nervous system assembly, maintenance and repair. Nature Reviews.
Neuroscience, 13(5), 308–24. https://doi.org/10.1038/nrn3210
Kaddoum, L., Magdeleine, E., Waldo, G. S., Joly, E., & Cabantous, S. (2010). One-step split
GFP staining for sensitive protein detection and localization in mammalian cells.
BioTechniques, 49(4), 727–736. https://doi.org/10.2144/000113512
Kandel, E. R. (2001). The Molecular Biology of Memory Storage: A Dialogue between Genes
and Synapses. Science, 294(5544), 1030–1038. https://doi.org/10.1126/science.1067020
Kankaanpää, P., Paavolainen, L., Tiitta, S., Karjalainen, M., Päivärinne, J., Nieminen, J., …
White, D. J. (2012). BioImageXD: an open, general-purpose and high-throughput image-
processing platform. Nature Methods, 9(7), 683–689. https://doi.org/10.1038/nmeth.2047
Kavanagh, I. C., & Baker, S. C. (2009). Advances in Nucleic Acid Detection and Quantification.
Biochemical Society Transactions, 37(2).
Kent, K. P., Childs, W., & Boxer, S. G. (2008). Deconstructing Green Fluorescent Protein.
Journal of the American Chemical Society, 130(30), 9664–9665.
https://doi.org/10.1021/ja803782x
Kerppola, T. K. (2006). Visualization of molecular interactions by fluorescence
complementation. Nature Reviews Molecular Cell Biology, 7(6), 449–456.
149
https://doi.org/10.1038/nrm1929
Kessels, H. W., & Malinow, R. (2009). Synaptic AMPA receptor plasticity and behavior.
Neuron, 61(3), 340–50. https://doi.org/10.1016/j.neuron.2009.01.015
Kim, J. H., Lee, S.-R., Li, L.-H., Park, H.-J., Park, J.-H., Lee, K. Y., … Choi, S.-Y. (2011). High
cleavage efficiency of a 2A peptide derived from porcine teschovirus-1 in human cell lines,
zebrafish and mice. PloS One, 6(4), e18556. https://doi.org/10.1371/journal.pone.0018556
Kim, J., Zhao, T., Petralia, R. S., Yu, Y., Peng, H., Myers, E., & Magee, J. C. (2011). mGRASP
enables mapping mammalian synaptic connectivity with light microscopy. Nature Methods,
9(1), 96–102. https://doi.org/10.1038/nmeth.1784
Kim, S., Chen, P., & Martin, K. (2010). Visualizing mRNA Trafficking and Local Translation
Within Individual Neurons. Society for Neuroscience, 35. Retrieved from
http://www.sfn.org/siteobjects/published/0000BDF20016F63800FD712C30FA42DD/64CF
FFC781B594C25468FF82134E35AD/file/Short Course Book II.pdf#page=37
King, M. Lou, Messitt, T. J., & Mowry, K. L. (2005). Putting RNAs in the right place at the right
time: RNA localization in the frog oocyte. Biology of the Cell / under the Auspices of the
European Cell Biology Organization, 97, 19–33. https://doi.org/10.1042/BC20040067
Kishikawa, T., Otsuka, M., Ohno, M., Yoshikawa, T., Takata, A., & Koike, K. (2015).
Circulating RNAs as new biomarkers for detecting pancreatic cancer. World Journal of
Gastroenterology, 21(28), 8527–40. https://doi.org/10.3748/wjg.v21.i28.8527
Kopito, R. R. (2000). Aggresomes, inclusion bodies and protein aggregation. Trends in Cell
Biology. https://doi.org/10.1016/S0962-8924(00)01852-3
Kramer, G., Boehringer, D., Ban, N., & Bukau, B. (2009). The ribosome as a platform for co-
translational processing, folding and targeting of newly synthesized proteins. Nature
Structural & Molecular Biology, 16(6), 589–597. https://doi.org/10.1038/nsmb.1614
Krichevsky, A. M., & Kosik, K. S. (2001). Neuronal RNA Granules: A Link between RNA
Localization and Stimulation-Dependent Translation. Neuron, 32(4), 683–696.
https://doi.org/10.1016/S0896-6273(01)00508-6
Krishna Prasad, R. B., Sharma, A., & Babu, H. M. (2013). An insight into salivary markers in
oral cancer. Dental Research Journal, 10(3), 287–95. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/24019794
Kubitscheck, U., Kückmann, O., Kues, T., & Peters, R. (2000). Imaging and Tracking of Single
GFP Molecules in Solution. Biophysical Journal, 78(4), 2170–2179.
https://doi.org/10.1016/S0006-3495(00)76764-6
Kwon, I., Lee, J., Chang, S. H., Jung, N. C., Lee, B. J., Son, G. H., … Lee, K. H. (2006).
BMAL1 shuttling controls transactivation and degradation of the CLOCK/BMAL1
heterodimer. Molecular and Cellular Biology, 26(19), 7318–30.
https://doi.org/10.1128/MCB.00337-06
Larionov, A., Krause, A., & Miller, W. (2005). A standard curve based method for relative real
time PCR data processing. BMC Bioinformatics, 6, 62. https://doi.org/10.1186/1471-2105-
6-62
Lasko, P. (2012). mRNA localization and translational control in Drosophila oogenesis. Cold
Spring Harbor Perspectives in Biology, 4(10). https://doi.org/10.1101/cshperspect.a012294
Leandro, M. J., Weill, J., Weller, S., Reynaud, C., Lammers, A., Porto, A. de, … Nicoletti, A.
(2013). B-cell subpopulations in humans and their differential susceptibility to depletion
with anti-CD20 monoclonal antibodies. Arthritis Research & Therapy, 15(Suppl 1), S3.
https://doi.org/10.1186/ar3908
150
Lecuyer, E., Yoshida, H., Parthasarathy, N., Alm, C., Babak, T., Cerovina, T., … Krause, H. M.
(2007). Global Analysis of mRNA Localization Reveals a Prominent Role in Organizing
Cellular Architecture and Function. Cell, 131(1), 174–187.
https://doi.org/10.1016/j.cell.2007.08.003
Lein, E. S., Hawrylycz, M. J., Ao, N., Ayres, M., Bensinger, A., Bernard, A., … Jones, A. R.
(2007). Genome-wide atlas of gene expression in the adult mouse brain. Nature, 445(7124),
168–176. https://doi.org/10.1038/nature05453
Levinthal, C. (1969). How to fold graciously. Mössbauer Spectroscopy in Biological Systems
Proceedings, 24(41), 22–24. https://doi.org/citeulike-article-id:380320
Li, F., Vijayasankaran, N., Shen, A. (Yijuan), Kiss, R., & Amanullah, A. (2010). Cell culture
processes for monoclonal antibody production. mAbs, 2(5), 466–479.
https://doi.org/10.4161/mabs.2.5.12720
Li, G.-W., & Xie, X. S. (2011). Central dogma at the single-molecule level in living cells.
Nature, 475(7356), 308–15. https://doi.org/10.1038/nature10315
Li, J. J., Bickel, P. J., & Biggin, M. D. (2014). System wide analyses have underestimated
protein abundances and the importance of transcription in mammals. PeerJ, 2, e270.
https://doi.org/10.7717/peerj.270
Li, J. Z., Bunney, B. G., Meng, F., Hagenauer, M. H., Walsh, D. M., Vawter, M. P., … Bunney,
W. E. (2013). Circadian patterns of gene expression in the human brain and disruption in
major depressive disorder. Proceedings of the National Academy of Sciences of the United
States of America, 110(24), 9950–5. https://doi.org/10.1073/pnas.1305814110
Li, X., Zhao, X., Fang, Y., Jiang, X., Duong, T., Fan, C., … Kain, S. R. (1998). Generation of
destabilized green fluorescent protein as a transcription reporter. Journal of Biological
Chemistry, 273(52), 34970–34975. https://doi.org/10.1074/jbc.273.52.34970
Liang, X., Potter, J., Kumar, S., Zou, Y., Quintanilla, R., Sridharan, M., … Chesnut, J. D. (2015).
Rapid and highly efficient mammalian cell engineering via Cas9 protein transfection.
Journal of Biotechnology, 208, 44–53. https://doi.org/10.1016/j.jbiotec.2015.04.024
Lin, A. C., & Holt, C. E. (2008). Function and regulation of local axonal translation. Current
Opinion in Neurobiology. https://doi.org/10.1016/j.conb.2008.05.004
Ling, D., Salvaterra, P. M., White, K., Salvaterra, P., & Lewis, J. (2011). Robust RT-qPCR Data
Normalization: Validation and Selection of Internal Reference Genes during Post-
Experimental Data Analysis. PLoS ONE, 6(3), e17762.
https://doi.org/10.1371/journal.pone.0017762
Lo, C.-A., Kays, I., Emran, F., Lin, T.-J., Cvetkovska, V., & Chen, B. E. (2015). Quantification
of Protein Levels in Single Living Cells. Cell Reports, 13(11), 2634–2644.
https://doi.org/10.1016/j.celrep.2015.11.048
Lo, K. M., & Gillies, S. D. (1991). High level expression of human proteins in murine
hybridoma cells: induction by methotrexate in the absence of gene amplification.
Biochimica et Biophysica Acta, 1088(2), 217–24. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/1672095
Lodish, H., Berk, A., Kaiser, C. A., Matsudaira, P., Baltimore, D., Bretscher, A., … Matsudaira,
P. (2008). Molecular Cell Biology. Book (Vol. 5). https://doi.org/10.1016/S1470-
8175(01)00023-6
Luby-Phelps, K. (2000). Cytoarchitecture and physical properties of cytoplasm: volume,
viscosity, diffusion, intracellular surface area. International Review of Cytology, 192, 189–
221. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10553280
151
Ly, T., Ahmad, Y., Shlien, A., Soroka, D., Mills, A., Emanuele, M. J., … Lamond, A. I. (2014).
A proteomic chronology of gene expression through the cell cycle in human myeloid
leukemia cells. eLife, 3, e01630. https://doi.org/10.7554/elife.01630
Lyles, V., Zhao, Y., & Martin, K. C. (2006). Synapse formation and mRNA localization in
cultured Aplysia neurons. Neuron, 49(3), 349–356.
https://doi.org/10.1016/j.neuron.2005.12.029
MacDougall, N., Clark, A., MacDougall, E., & Davis, I. (2003). Drosophila gurken (TGFα)
mRNA localizes as particles that move within the oocyte in two dynein-dependent steps.
Developmental Cell, 4(3), 307–319. https://doi.org/10.1016/S1534-5807(03)00058-3
Maier, T., Güell, M., & Serrano, L. (2009). Correlation of mRNA and protein in complex
biological samples. FEBS Letters, 583(24), 3966–73.
https://doi.org/10.1016/j.febslet.2009.10.036
Mane, V. P., Heuer, M. A., Hillyer, P., Navarro, M. B., & Rabin, R. L. (2008). Systematic
method for determining an ideal housekeeping gene for real-time PCR analysis. Journal of
Biomolecular Techniques : JBT, 19(5), 342–7.
Mao, Z., Bozzella, M., Seluanov, A., & Gorbunova, V. (2008). DNA repair by nonhomologous
end joining and homologous recombination during cell cycle in human cells. Cell Cycle
(Georgetown, Tex.), 7(18), 2902–6. https://doi.org/10.2964/jsik.kuni0223
Marshall, J., Molloy, R., Moss, G. W., Howe, J. R., & Hughes, T. E. (1995). The jellyfish green
fluorescent protein: a new tool for studying ion channel expression and function. Neuron,
14(2), 211–5. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/7531985
Martin, K. C., & Ephrussi, A. (2010). mRNA Localization : Gene Expression in the Spatial
Dimension. Cell, 136(4), 1–21. https://doi.org/10.1016/j.cell.2009.01.044.mRNA
Martinez, J. C., Pisabarro, M. T., & Serrano, L. (1998). Obligatory steps in protein folding and
the conformational diversity of the transition state. Nature Structural Biology, 5(8), 721–
729. https://doi.org/10.1038/1418
Marx, V. (2013). Finding the right antibody for the job. Nature Methods, 10(8), 703–707.
https://doi.org/10.1038/nmeth.2570
Masters, P. M., Bada, J. L., & Zigler Jr., J. S. (1977). Aspartic acid racemisation in the human
lens during ageing and in cataract formation. Nature, 268(5615), 71–73.
https://doi.org/10.1038/268071a0
Mavrakis, M., Rikhy, R., Lilly, M., Lippincott-Schwartz, J., Mavrakis, M., Rikhy, R., …
Lippincott‐ Schwartz, J. (2008). Fluorescence Imaging Techniques for Studying
Drosophila Embryo Development. In Current Protocols in Cell Biology (p. 4.18.1-4.18.43).
Hoboken, NJ, USA: John Wiley & Sons, Inc.
https://doi.org/10.1002/0471143030.cb0418s39
Mikaelian, I., Scicchitano, M., Mendes, O., Thomas, R. A., & Leroy, B. E. (2013). Frontiers in
preclinical safety biomarkers: microRNAs and messenger RNAs. Toxicologic Pathology,
41(1), 18–31. https://doi.org/10.1177/0192623312448939
Miyashiro, K., Dichter, M., & Eberwine, J. (1994). On the nature and differential distribution of
mRNAs in hippocampal neurites: implications for neuronal functioning. Proceedings of the
National Academy of Sciences of the United States of America, 91(23), 10800–4.
https://doi.org/10.1073/pnas.91.23.10800
Mizuguchi, H., Xu, Z., Ishii-Watabe, A., Uchida, E., & Hayakawa, T. (2000). IRES-dependent
second gene expression is significantly lower than cap-dependent first gene expression in a
bicistronic vector. Molecular Therapy : The Journal of the American Society of Gene
152
Therapy, 1(4), 376–82. https://doi.org/10.1006/mthe.2000.0050
Moccia, R., Chen, D., Lyles, V., Kapuya, E., E, Y., Kalachikov, S., … Martin, K. C. (2003). An
unbiased cDNA library prepared from isolated Aplysia sensory neuron processes is
enriched for cytoskeletal and translational mRNAs. The Journal of Neuroscience : The
Official Journal of the Society for Neuroscience, 23(28), 9409–9417.
https://doi.org/23/28/9409 [pii]
Morse, D., & Tannous, B. A. (2012). A water-soluble coelenterazine for sensitive in vivo
imaging of coelenterate luciferases. Molecular Therapy : The Journal of the American
Society of Gene Therapy, 20(4), 692–3. https://doi.org/10.1038/mt.2012.38
Mourelatos, Z., Abel, L., Yong, J., Kataoka, N., & Dreyfuss, G. (2001). SMN interacts with a
novel family of hnRNP and spliceosomal proteins. The EMBO Journal, 20(19), 5443–52.
https://doi.org/10.1093/emboj/20.19.5443
Muddashetty, R. S., Kelić, S., Gross, C., Xu, M., & Bassell, G. J. (2007). Dysregulated
metabotropic glutamate receptor-dependent translation of AMPA receptor and postsynaptic
density-95 mRNAs at synapses in a mouse model of fragile X syndrome. The Journal of
Neuroscience : The Official Journal of the Society for Neuroscience, 27(20), 5338–5348.
https://doi.org/10.1523/JNEUROSCI.0937-07.2007
Mullis, K., Faloona, F., Scharf, S., Saiki, R., Horn, G., & Erlich, H. (1986). Specific enzymatic
amplification of DNA in vitro: The polymerase chain reaction. Cold Spring Harbor
Symposia on Quantitative Biology, 51(1), 263–273.
https://doi.org/10.1101/SQB.1986.051.01.032
Na, Y., Park, S., Lee, C., Kim, D., Park, J. M., Sockanathan, S., … Worley, P. F. (2016). Real-
Time Imaging Reveals Properties of Glutamate-Induced Arc/Arg 3.1 Translation in
Neuronal Dendrites. Neuron, 91(3), 561–573. https://doi.org/10.1016/j.neuron.2016.06.017
Nagai, T., Ibata, K., Park, E. S., Kubota, M., Mikoshiba, K., & Miyawaki, A. (2002). A variant
of yellow fluorescent protein with fast and efficient maturation for cell-biological
applications. Nature Biotechnology, 20(1), 87–90. https://doi.org/10.1038/nbt0102-87
Nair-Gill, E. D., Shu, C. J., Radu, C. G., & Witte, O. N. (2008). Non-invasive imaging of
adaptive immunity using positron emission tomography. Immunological Reviews, 221, 214–
28. https://doi.org/10.1111/j.1600-065X.2008.00585.x
Narahashi, T. (1988). Ion Channels : Volume 1. Springer US.
Nelson, R., Sawaya, M. R., Balbirnie, M., Madsen, A. Ø., Riekel, C., Grothe, R., & Eisenberg,
D. (2005). Structure of the cross-β spine of amyloid-like fibrils. Nature Cell Biology,
435(7043), 773–778. https://doi.org/10.1038/nature03680
Nilson, L. A., & Schüpbach, T. (1999). EGF receptor signaling in Drosophila oogenesis. Current
Topics in Developmental Biology, 44, 203–43. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/9891881
Nimchinsky, E. A., Sabatini, B. L., & Svoboda, K. (2002). Structure and function of dendritic
spines. Annual Review of Physiology, 64, 313–53.
https://doi.org/10.1146/annurev.physiol.64.081501.160008
Nolan, T., Hands, R. E., & Bustin, S. A. (2006). Quantification of mRNA using real-time RT-
PCR. Nature Protocols, 1(3), 1559–82. https://doi.org/10.1038/nprot.2006.236
Nutt, S. L., Hodgkin, P. D., Tarlinton, D. M., & Corcoran, L. M. (2015). The generation of
antibody-secreting plasma cells. Nature Reviews. Immunology, 15(3), 160–171.
https://doi.org/10.1038/nri3795
Ochs, S., Sabri, M. I., & Johnson, J. (1969). Fast transport system of materials in mammalian
153
nerve fibers. Science (New York, N.Y.), 163(3868), 686–7. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/5762934
Opel, K. L., Chung, D., & McCord, B. R. (2010). A Study of PCR Inhibition Mechanisms Using
Real Time PCR. Journal of Forensic Sciences, 55(1), 25–33. https://doi.org/10.1111/j.1556-
4029.2009.01245.x
Orci, L., Ravazzola, M., Amherdt, M., Madsen, O., Perrelet, A., Vassalli, J. D., & Anderson, R.
G. (1986). Conversion of proinsulin to insulin occurs coordinately with acidification of
maturing secretory vesicles. The Journal of Cell Biology, 103(6 Pt 1), 2273–81.
https://doi.org/10.1083/JCB.103.6.2273
Ormö, M., Cubitt, A. B., Kallio, K., Gross, L. A., Tsien, R. Y., Remington, S. J., … YANG, F.
(1996). Crystal structure of the Aequorea victoria green fluorescent protein. Science (New
York, N.Y.), 273(5280), 1392–5. https://doi.org/10.1126/science.273.5280.1392
Ozsolak, F., & Milos, P. M. (2011). RNA sequencing: advances, challenges and opportunities.
Nature Reviews. Genetics, 12(2), 87–98. https://doi.org/10.1038/nrg2934
Palmer, E., & Freeman, T. (2004). Investigation into the use of C- and N-terminal GFP fusion
proteins for subcellular localization studies using reverse transfection microarrays.
Comparative and Functional Genomics, 5(4), 342–353. https://doi.org/10.1002/cfg.405
Panda, S., Antoch, M. P., Miller, B. H., Su, A. I., Schook, A. B., Straume, M., … Hogenesch, J.
B. (2002). Coordinated transcription of key pathways in the mouse by the circadian clock.
Cell, 109(3), 307–20. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12015981
Paquin, N., & Chartrand, P. (2008). Local regulation of mRNA translation: new insights from the
bud. Trends in Cell Biology. https://doi.org/10.1016/j.tcb.2007.12.004
Parton, R. M., Vallés, A. M., Dobbie, I. M., & Davis, I. (2010). Collection and mounting of
Drosophila embryos for imaging. Cold Spring Harbor Protocols, 2010(4), pdb.prot5403.
https://doi.org/10.1101/pdb.prot5403
Patterson, G. H., Knobel, S. M., Sharif, W. D., Kain, S. R., & Piston, D. W. (1997). Use of the
green fluorescent protein and its mutants in quantitative fluorescence microscopy.
Biophysical Journal, 73(5), 2782–90. https://doi.org/10.1016/S0006-3495(97)78307-3
Pauwels, K., & Tompa, P. (2016). Editorial: Function and Flexibility: Friend or Foe? Frontiers
in Molecular Biosciences, 3. https://doi.org/10.3389/fmolb.2016.00031
Pédelacq, J.-D., Cabantous, S., Tran, T., Terwilliger, T. C., & Waldo, G. S. (2006). Engineering
and characterization of a superfolder green fluorescent protein. Nature Biotechnology,
24(1), 79–88. https://doi.org/10.1038/nbt1172
Pelletier, J., & Sonenberg, N. (1988). Internal initiation of translation of eukaryotic mRNA
directed by a sequence derived from poliovirus RNA. Nature, 334(6180), 320–325.
https://doi.org/10.1038/334320a0
Piana, S., Lindorff-Larsen, K., & Shaw, D. E. (2013). Atomic-level description of ubiquitin
folding. Proceedings of the National Academy of Sciences of the United States of America,
110(15), 5915–20. https://doi.org/10.1073/pnas.1218321110
Pinaud, F., & Dahan, M. (2011). Targeting and imaging single biomolecules in living cells by
complementation-activated light microscopy with split-fluorescent proteins. Proc Natl Acad
Sci U S A, 108(24), E201-10. https://doi.org/10.1073/pnas.1101929108
Poddar, D., Basu, A., Baldwin, W. M., Kondratov, R. V, Barik, S., & Mazumder, B. (2013). An
extraribosomal function of ribosomal protein L13a in macrophages resolves inflammation.
Journal of Immunology (Baltimore, Md. : 1950), 190(7), 3600–12.
https://doi.org/10.4049/jimmunol.1201933
154
Poon, M. M., Choi, S.-H., Jamieson, C. a M., Geschwind, D. H., & Martin, K. C. (2006).
Identification of process-localized mRNAs from cultured rodent hippocampal neurons. The
Journal of Neuroscience : The Official Journal of the Society for Neuroscience, 26(51),
13390–13399. https://doi.org/10.1523/JNEUROSCI.3432-06.2006
Prado, M., Eickel, N., De Niz, M., Heitmann, A., Agop-Nersesian, C., Wacker, R., … Heussler,
V. T. (2015). Long-term live imaging reveals cytosolic immune responses of host
hepatocytes against Plasmodium infection and parasite escape mechanisms. Autophagy,
11(9), 1561–79. https://doi.org/10.1080/15548627.2015.1067361
Prasher, D. C., Eckenrode, V. K., Ward, W. W., Prendergast, F. G., & Cormier, M. J. (1992).
Primary structure of the Aequorea victoria green-fluorescent protein. Gene, 111(2), 229–
233. https://doi.org/10.1016/0378-1119(92)90691-H
Prelich, G. (2012). Gene overexpression: uses, mechanisms, and interpretation. Genetics, 190(3),
841–54. https://doi.org/10.1534/genetics.111.136911
Price, J. C., Guan, S., Burlingame, A., Prusiner, S. B., & Ghaemmaghami, S. (2010). Analysis of
proteome dynamics in the mouse brain. Proceedings of the National Academy of Sciences of
the United States of America, 107(32), 14508–13. https://doi.org/10.1073/pnas.1006551107
Qin, J. Y., Zhang, L., Clift, K. L., Hulur, I., Xiang, A. P., Ren, B.-Z., & Lahn, B. T. (2010).
Systematic Comparison of Constitutive Promoters and the Doxycycline-Inducible Promoter.
PLoS ONE, 5(5), e10611. https://doi.org/10.1371/journal.pone.0010611
Quitschke, W. W., Lin, Z. Y., DePonti-Zilli, L., & Paterson, B. M. (1989). The beta actin
promoter. High levels of transcription depend upon a CCAAT binding factor. The Journal
of Biological Chemistry, 264(16), 9539–46. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/2722849
Radcliffe, P. A., & Mitrophanous, K. A. (2004). Multiple gene products from a single vector:
“self-cleaving” 2A peptides. Gene Therapy, 11(23), 1673–1674.
https://doi.org/10.1038/sj.gt.3302361
Radzicka, A., & Wolfenden, R. (1995). A proficient enzyme. Science, 267(5194).
Rao, A. K., Garver, F., & Mendicino, J. (1976). Biosynthesis of the carbohydrate units of
immunoglobulins. 1. Purification and properties of galactosyltransferases from swine
mesentary lymph nodes. Biochemistry, 15(23), 5001–5009.
Reid, B. G., & Flynn, G. C. (1997). Chromophore formation in green fluorescent protein.
Biochemistry, 36(22), 6786–6791. https://doi.org/10.1021/bi970281w
Remington, S. J. (2011). Green fluorescent protein: a perspective. Protein Science : A
Publication of the Protein Society, 20(9), 1509–19. https://doi.org/10.1002/pro.684
Richter, J. D., & Lasko, P. (2011). Translational control in oocyte development. Cold Spring
Harbor Perspectives in Biology. https://doi.org/10.1101/cshperspect.a002758
Rizzuto, R., Brini, M., Pizzo, P., Murgia, M., & Pozzan, T. (1995). Chimeric green fluorescent
protein as a tool for visualizing subcellular organelles in living cells. Current Biology : CB,
5(6), 635–42. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/7552174
Roberts, L. A., Large, C. H., Higgins, M. J., Stone, T. W., O’Shaughnessy, C. T., & Morris, B. J.
(1998). Increased expression of dendritic mRNA following the induction of long-term
potentiation. Molecular Brain Research, 56(1–2), 38–44. https://doi.org/10.1016/S0169-
328X(98)00026-6
Rørth, P. (1998). Gal4 in the Drosophila female germline. Mechanisms of Development, 78(1–2),
113–118. https://doi.org/10.1016/S0925-4773(98)00157-9
Ross, J. F., & Orlowski, M. (1982). Growth-rate-dependent adjustment of ribosome function in
155
chemostat-grown cells of the fungus Mucor racemosus. Journal of Bacteriology, 149(2),
650–3. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/6799491
Sage, A. T., Besant, J. D., Mahmoudian, L., Poudineh, M., Bai, X., Zamel, R., … Kelley, S. O.
(2015). Fractal circuit sensors enable rapid quantification of biomarkers for donor lung
assessment for transplantation. Science Advances, 1(7).
Saiki, R. K., Scharf, S., Faloona, F., Mullis, K. B., Horn, G. T., Erlich, H. a, & Arnheim, N.
(1985). Enzymatic amplification of beta-globin genomic sequences and restriction site
analysis for diagnosis of sickle cell anemia. Science (New York, N.Y.), 230(4732), 1350–
1354. https://doi.org/10.1126/science.2999980
Santangelo, P. J., Rogers, K. a, Zurla, C., Blanchard, E. L., Gumber, S., Strait, K., … Villinger,
F. (2015). Whole-body immunoPET reveals active SIV dynamics in viremic and
antiretroviral therapy-treated macaques. Nature Methods, 12(5), 427–32.
https://doi.org/10.1038/nmeth.3320
Satija, R., Shalek, A. K., Raj, A., al., et, Germain, R. N., Littman, D. R., … al., et. (2014).
Heterogeneity in immune responses: from populations to single cells. Trends in
Immunology, 35(5), 219–29. https://doi.org/10.1016/j.it.2014.03.004
Saunders, C., & Cohen, R. S. (1999). The role of oocyte transcription, the 5’UTR, and translation
repression and derepression in Drosophila gurken mRNA and protein localization.
Molecular Cell, 3(1), 43–54. https://doi.org/10.1016/S1097-2765(00)80173-2
Schubert, C. (2011). Single-cell analysis: The deepest differences. Nature, 480(7375), 133–137.
https://doi.org/10.1038/480133a
Schwanhäusser, B., Busse, D., Li, N., Dittmar, G., Schuchhardt, J., Wolf, J., … Selbach, M.
(2011). Global quantification of mammalian gene expression control. Nature, 473(7347),
337–342. https://doi.org/10.1038/nature10098
Scott, K. P., Mercer, D. K., Glover, L. A., & Flint, H. J. (1998). The green fluorescent protein as
a visible marker for lactic acid bacteria in complex ecosystems. FEMS Microbiology
Ecology, 26(3), 219–230. https://doi.org/10.1016/S0168-6496(98)00037-3
Scott, M. S., Boisvert, F. M., McDowall, M. D., Lamond, A. I., & Barton, G. J. (2010).
Characterization and prediction of protein nucleolar localization sequences. Nucleic Acids
Research, 38(21), 7388–7399. https://doi.org/10.1093/nar/gkq653
Shain, A. H., & Pollack, J. R. (2013). The spectrum of SWI/SNF mutations, ubiquitous in human
cancers. PloS One, 8(1), e55119. https://doi.org/10.1371/journal.pone.0055119
Shaner, N. C., Lin, M. Z., McKeown, M. R., Steinbach, P. A., Hazelwood, K. L., Davidson, M.
W., & Tsien, R. Y. (2008). Improving the photostability of bright monomeric orange and
red fluorescent proteins. Nature Methods, 5(6), 545–51. https://doi.org/10.1038/nmeth.1209
Shaner, N. C., Steinbach, P. A., & Tsien, R. Y. (2005). A guide to choosing fluorescent proteins.
Nature Methods, 2(12), 905–909. https://doi.org/10.1038/nmeth819
Smalle, J., & Vierstra, R. D. (2004). The ubiquitin 26S proteasome proteolytic pathway. Annual
Review of Plant Biology, 55, 555–90.
https://doi.org/10.1146/annurev.arplant.55.031903.141801
Smith, W. B., Starck, S. R., Roberts, R. W., & Schuman, E. M. (2005). Dopaminergic
stimulation of local protein synthesis enhances surface expression of GluR1 and synaptic
transmission in hippocampal neurons. Neuron, 45(5), 765–79.
https://doi.org/10.1016/j.neuron.2005.01.015
Snapp, E. (2005). Design and use of fluorescent fusion proteins in cell biology. Current
Protocols in Cell Biology / Editorial Board, Juan S. Bonifacino ... [et Al.], Chapter 21, Unit
156
21.4. https://doi.org/10.1002/0471143030.cb2104s27
Stahlberg, A., & Bengtsson, M. (2010). Single-cell gene expression profiling using reverse
transcription quantitative real-time PCR. Methods, 50(4), 282–288.
https://doi.org/10.1016/j.ymeth.2010.01.002
Ståhlberg, A., & Kubista, M. (2014). The workflow of single-cell expression profiling using
quantitative real-time PCR. Expert Review of Molecular Diagnostics, 14(3), 323–331.
https://doi.org/10.1586/14737159.2014.901154
Stemmer, W. P. (1994). DNA shuffling by random fragmentation and reassembly: in vitro
recombination for molecular evolution. Proceedings of the National Academy of Sciences of
the United States of America, 91(22), 10747–10751.
https://doi.org/10.1073/pnas.91.22.10747
Steward, O., Wallace, C. S., Lyford, G. L., & Worley, P. F. (1998). Synaptic activation causes
the mRNA for the leg Arc to localize selectively near activated postsynaptic sites on
dendrites. Neuron, 21(4), 741–751. https://doi.org/10.1016/S0896-6273(00)80591-7
Stockholm, D., Benchaouir, R., Picot, J., Rameau, P., Neildez, T. M. A., Landini, G., … Paldi,
A. (2007). The Origin of Phenotypic Heterogeneity in a Clonal Cell Population In Vitro.
PLoS ONE, 2(4), e394. https://doi.org/10.1371/journal.pone.0000394
Sullivan, D. C., & Kuntz, I. D. (2002). Protein Folding as Biased Conformational Diffusion. The
Journal of Physical Chemistry B, 106(12), 3255–3262. https://doi.org/10.1021/jp012911g
Sullivan, L. H. (1896). The tall office building artistically considered. Lippincott’s Magazine.
Retrieved from http://ocw.mit.edu/courses/architecture/4-205-analysis-of-contemporary-
architecture-fall-2009/readings/MIT4_205F09_Sullivan.pdf
Suslov, O., & Steindler, D. A. (2005). PCR inhibition by reverse transcriptase leads to an
overestimation of amplification efficiency. Nucleic Acids Research, 33(20), e181.
https://doi.org/10.1093/nar/gni176
Svec, D., Andersson, D., Pekny, M., Sjöback, R., Kubista, M., & Ståhlberg, A. (2013). Direct
cell lysis for single-cell gene expression profiling. Frontiers in Oncology, 3(November),
274. https://doi.org/10.3389/fonc.2013.00274
Swanger, S. A., & Bassell, G. J. (2011). Making and breaking synapses through local mRNA
regulation. Current Opinion in Genetics & Development, 21(4), 414–21.
https://doi.org/10.1016/j.gde.2011.04.002
Swanger, S. A., & Bassell, G. J. (2013). Dendritic protein synthesis in the normal and diseased
brain. Neuroscience, 232(404), 106–127.
https://doi.org/10.1016/j.neuroscience.2012.12.003
Szymczak, A. L., Workman, C. J., Wang, Y., Vignali, K. M., Dilioglou, S., Vanin, E. F., &
Vignali, D. a a. (2004). Correction of multi-gene deficiency in vivo using a single “self-
cleaving” 2A peptide-based retroviral vector. Nature Biotechnology, 22(5), 589–94.
https://doi.org/10.1038/nbt957
Szymczak-Workman, A. L., Vignali, K. M., & Vignali, D. A. A. (2012). Design and
Construction of 2A Peptide-Linked Multicistronic Vectors. Cold Spring Harbor Protocols,
2012(2), pdb.ip067876-ip067876. https://doi.org/10.1101/pdb.ip067876
Tatavarty, V., Ifrim, M. F., Levin, M., Korza, G., Barbarese, E., Yu, J., & Carson, J. H. (2012).
Single-molecule imaging of translational output from individual RNA granules in neurons.
Molecular Biology of the Cell, 23(5), 918–29. https://doi.org/10.1091/mbc.E11-07-0622
Taylor, S., Wakem, M., Dijkman, G., Alsarraj, M., & Nguyen, M. (2010). A practical approach
to RT-qPCR—Publishing data that conform to the MIQE guidelines. Methods, 50(4), S1–
157
S5. https://doi.org/10.1016/j.ymeth.2010.01.005
Tester, M. (1997). Techniques for studying ion channels: an introduction. Journal of
Experimental Botany, 48 Spec No, 353–9. https://doi.org/10.1093/jxb/48.Special_Issue.353
Tian, Q. B., Nakayama, K., Okano, A., & Suzuki, T. (1999). Identification of mRNAs localizing
in the postsynaptic region. Molecular Brain Research, 72(2), 147–157.
https://doi.org/10.1016/S0169-328X(99)00214-4
Tichopad, A., Kitchen, R., Riedmaier, I., Becker, C., Ståhlberg, A., & Kubista, M. (2009).
Design and Optimization of Reverse-Transcription Quantitative PCR Experiments. Clinical
Chemistry, 55(10).
Tobias, G. S., & Koenig, E. (1975). Influence of nerve cell body and neurolemma cell on local
axonal protein synthesis following neurotomy. Experimental Neurology, 49(1 Pt 1), 235–45.
Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/52557
Toyama, B. H., & Hetzer, M. W. (2013). Protein homeostasis: live long, won’t prosper. Nature
Reviews. Molecular Cell Biology, 14(1), 55–61. https://doi.org/10.1038/nrm3496
Tsien, R. Y. (1998). The green fluorescent protein. Annual Review of Biochemistry, 67, 509–44.
https://doi.org/10.1146/annurev.biochem.67.1.509
Van De Bor, V., Hartswood, E., Jones, C., Finnegan, D., & Davis, I. (2005). gurken and the I
factor retrotransposon RNAs share common localization signals and machinery.
Developmental Cell, 9(1), 51–62. https://doi.org/10.1016/j.devcel.2005.04.012
van de Corput, M. P., Dirks, R. W., van Gijlswijk, R. P., van de Rijke, F. M., & Raap, A. K.
(1998). Fluorescence in situ hybridization using horseradish peroxidase-labeled
oligodeoxynucleotides and tyramide signal amplification for sensitive DNA and mRNA
detection. Histochemistry and Cell Biology, 110(4), 431–7. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/9792422
Vogel, C., & Marcotte, E. M. (2012). Insights into the regulation of protein abundance from
proteomic and transcriptomic analyses. Nature Reviews. Genetics, 13(4), 227–32.
https://doi.org/10.1038/nrg3185
Waldo, G. S., Standish, B. M., Berendzen, J., & Terwilliger, T. C. (1999). Rapid protein-folding
assay using green fluorescent protein. Nature Biotechnology, 17(7), 691–695.
https://doi.org/10.1038/10904
Walker, J. M. (2009). The Protein Protocols Handbook. Humana Press.
https://doi.org/10.1007/978-1-59745-198-7_26
Wang, D. O., Kim, S. M., Zhao, Y., Hwang, H., Miura, S. K., Sossin, W. S., & Martin, K. C.
(2009). Synapse- and Stimulus-Specific Local Translation During Long-Term Neuronal
Plasticity. Science, 324(5934), 1536–1540. https://doi.org/10.1126/science.1173205
Wang, D. O., Kim, S. M., Zhao, Y., Hwang, H., Miura, S. K., Sossin, W. S., & Martin, K. C.
(2009). Synapse- and stimulus-specific local translation during long-term neuronal
plasticity. Science (New York, N.Y.), 324(5934), 1536–1540.
https://doi.org/10.1126/science.1173205
Wang, D. O., Martin, K. C., & Zukin, R. S. (2010). Spatially restricting gene expression by local
translation at synapses. Trends in Neurosciences, 33(4), 173–182.
https://doi.org/10.1016/j.tins.2010.01.005
Wang, H.-W., Willis, J., Sivak, M. V., & Izatt, J. A. (1998). Correlation of autofluorescence to
proliferation and inflammatory indices in human premalignant colonic tissues. In
Biomedical Optical Spectroscopy and Diagnostics / Therapeutic Laser Applications (p.
BTuA4). Washington, D.C.: OSA. https://doi.org/10.1364/BOSD.1998.BTuA4
158
Ward, W. W., Cody, C. W., Hart, R. C., & Cormier, M. J. (1980). Spectrophotometric Identity of
the Energy Transfer Chromophores in Renilla and Aequorea Green-Fluorescent Proteins.
Photochemistry and Photobiology, 31(6), 611–615. https://doi.org/10.1111/j.1751-
1097.1980.tb03755.x
Willard, S. S., & Koochekpour, S. (2013). Glutamate, glutamate receptors, and downstream
signaling pathways. International Journal of Biological Sciences, 9(9), 948–59.
https://doi.org/10.7150/ijbs.6426
Willis, D. E., & Twiss, J. L. (2006). The evolving roles of axonally synthesized proteins in
regeneration. Current Opinion in Neurobiology. https://doi.org/10.1016/j.conb.2006.01.002
Wong, I. H., Yeo, W., Chan, A. T., & Johnson, P. J. (2001). Quantitative correlation of
cytokeratin 19 mRNA level in peripheral blood with disease stage and metastasis in breast
cancer patients: potential prognostic implications. International Journal of Oncology, 18(3),
633–8. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11179498
Wu, M., & Singh, A. K. (2012a). Single-cell protein analysis. Current Opinion in Biotechnology,
23(1), 83–88. https://doi.org/10.1016/j.copbio.2011.11.023
Wu, M., & Singh, A. K. (2012b). Single-cell protein analysis. Current Opinion in Biotechnology,
23(1), 83–8. https://doi.org/10.1016/j.copbio.2011.11.023
Yasukawa, K., Konishi, A., & Inouye, K. (2010). Effects of organic solvents on the reverse
transcription reaction catalyzed by reverse transcriptases from avian myeloblastosis virus
and Moloney murine leukemia virus. Bioscience, Biotechnology, and Biochemistry, 74(9),
1925–30. https://doi.org/10.1271/bbb.100337
Yu, S. P., & Kerchner, G. A. (1998). Endogenous voltage-gated potassium channels in human
embryonic kidney (HEK293) cells. Journal of Neuroscience Research, 52(5), 612–7.
https://doi.org/10.1002/(SICI)1097-4547(19980601)52:5<612::AID-JNR13>3.0.CO;2-3
[pii]
Zhang, R., Lahens, N. F., Ballance, H. I., Hughes, M. E., & Hogenesch, J. B. (2014). A circadian
gene expression atlas in mammals: implications for biology and medicine. Proceedings of
the National Academy of Sciences of the United States of America, 111(45), 16219–24.
https://doi.org/10.1073/pnas.1408886111
Zhao, H. L., Yao, X. Q., Xue, C., Wang, Y., Xiong, X. H., & Liu, Z. M. (2008). Increasing the
homogeneity, stability and activity of human serum albumin and interferon-alpha2b fusion
protein by linker engineering. Protein Expression and Purification, 61(1), 73–7.
https://doi.org/10.1016/j.pep.2008.04.013
Zhao, S., & Fernald, R. D. (2005). Comprehensive algorithm for quantitative real-time
polymerase chain reaction. Journal of Computational Biology : A Journal of Computational
Molecular Cell Biology, 12(8), 1047–64. https://doi.org/10.1089/cmb.2005.12.1047
Zheng, J. Q., Kelly, T. K., Chang, B., Ryazantsev, S., Rajasekaran, a K., Martin, K. C., & Twiss,
J. L. (2001). A functional role for intra-axonal protein synthesis during axonal regeneration
from adult sensory neurons. The Journal of Neuroscience : The Official Journal of the
Society for Neuroscience, 21(23), 9291–9303. https://doi.org/21/23/9291 [pii]
Zhou, X., Liao, W. J., Liao, J. M., Liao, P., & Lu, H. (2015). Ribosomal proteins: Functions
beyond the ribosome. Journal of Molecular Cell Biology, 7(2), 92–104.
https://doi.org/10.1093/jmcb/mjv014
Zimmer, M. (2002). Green Fluorescent Protein (GFP): Applications, Structure, and Related
Photophysical Behavior. Chemical Reviews, 102(3), 759–782.
https://doi.org/10.1021/cr010142r
159
Fin
