A high content imaging flow cytometry approach to study...

9
A high content imaging flow cytometry approach to study mitochondria in T cells: MitoTracker Green FM dye concentration optimization Namrata Gautam a , Shvetha Sankaran a , John A. Yason b , Kevin S.W. Tan b , Nicholas R.J. Gascoigne a,a Department of Microbiology and Immunology, Yong Loo Lin School of Medicine, and Immunology Programme, National University of Singapore, Singapore b Department of Microbiology and Immunology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore article info Article history: Received 8 August 2017 Received in revised form 3 November 2017 Accepted 27 November 2017 Available online 2 December 2017 Keywords: T cells Mitochondria MitoTracker Green FM Imaging flow cytometry (IFC) Amnis ImageStream Raw Max Pixel abstract Mitochondria, the powerhouse of the cell, are known to remodel their membrane structures through the process of fusion or fission. Studies have indicated that T cells adopt different energy metabolic pheno- types, namely oxidative phosphorylation and glycolysis depending on whether they are naïve, effector and memory T cells. It has recently been shown that changes in mitochondrial morphology dictate T cell fate via regulation of their metabolism. Our keen interest in T cell function and metabolism led us to explore and establish a method to study mitochondria in live T cells through a novel high content approach called Imaging Flow Cytometry (IFC). The focus of our current study was on developing a protocol to standardize the concentration of MitoTracker Green FM dye to observe mitochondria in live T cells using IFC. We began the study by using widefield microscopy to confirm the localisation of MitoTracker Green FM labelled mitochondria in live T cells. This was followed by testing various concentrations of the dye to achieve a similar labelling pattern using IFC while eliminating false positive or negative staining. The optimization of the method used to label the mitochondria by IFC for analysis included standardisation of a number of important parameters such as dye concentration, voltage, fluorescence intensity values for acquisition and processing. IFC could potentially be a powerful method to study T cells in a relatively high throughput manner. Ó 2017 Elsevier Inc. All rights reserved. 1. Introduction 1.1. T cells and their functional subsets T cells are an important subset of the adaptive branch of the immune system. They arise from lymphoid precursors in the bone marrow, which travel through the blood and enter the thymus [1]. The thymus, classified as a primary lymphoid organ, is the organ where T cells develop, and become educated and selected to express a useful antigen-recognition repertoire. After selection and final differentiation, they migrate to the periphery. This selec- tion is important to make the T cells capable of mounting an immune response against non-self-antigen fragments presented by self Major Histocompatibility Complex (MHC) proteins, whereas they must remain non-responsive to self-antigens. In this way, they refrain from the development of autoimmunity. Following this selection, T cells migrate from the thymus to the periphery [2,3]. The three known major functional subsets of T cells are Naïve (T N ), Effector (T E ) and Memory (T M ). When a T N cell residing in sec- ondary lymphoid organs such as lymph nodes, encounters its cog- nate antigen and receives an adequate amount of co-stimulatory signal, it transforms into an activated T cell. The T N undergoes extensive proliferation termed as ‘‘clonal expansion”, ultimately differentiating into T E cells. Upon clearance of antigen or infection, almost the entire population of these short-lived T E cells is lost during the contraction phase of the response. Another small subset of T cells derived from T N cells differentiates into long lived T M cells, which helps in mediating a quicker and enhanced immune response upon subsequent exposure to the same antigen [4,5]. 1.2. Immunity and metabolism: OXPHOS/glycolysis in T cells Recently a multitude of studies point towards a connection between immunity and metabolism. There is an increasing evi- dence that T cells are able to perform their respective functions only by virtue of their variable, function-specific energy require- ments [6]. The two known energy generation pathways adopted https://doi.org/10.1016/j.ymeth.2017.11.015 1046-2023/Ó 2017 Elsevier Inc. All rights reserved. Corresponding author at: Department of Microbiology and Immunology, National University of Singapore, Block MD4 level 3, 5 Science Drive 2, Singapore 117545, Singapore. E-mail address: [email protected] (N.R.J. Gascoigne). Methods 134–135 (2018) 11–19 Contents lists available at ScienceDirect Methods journal homepage: www.elsevier.com/locate/ymeth

Transcript of A high content imaging flow cytometry approach to study...

Methods 134–135 (2018) 11–19

Contents lists available at ScienceDirect

Methods

journal homepage: www.elsevier .com/locate /ymeth

A high content imaging flow cytometry approach to study mitochondriain T cells: MitoTracker Green FM dye concentration optimization

https://doi.org/10.1016/j.ymeth.2017.11.0151046-2023/� 2017 Elsevier Inc. All rights reserved.

⇑ Corresponding author at: Department of Microbiology and Immunology,National University of Singapore, Block MD4 level 3, 5 Science Drive 2, Singapore117545, Singapore.

E-mail address: [email protected] (N.R.J. Gascoigne).

Namrata Gautam a, Shvetha Sankaran a, John A. Yason b, Kevin S.W. Tan b, Nicholas R.J. Gascoigne a,⇑aDepartment of Microbiology and Immunology, Yong Loo Lin School of Medicine, and Immunology Programme, National University of Singapore, SingaporebDepartment of Microbiology and Immunology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore

a r t i c l e i n f o a b s t r a c t

Article history:Received 8 August 2017Received in revised form 3 November 2017Accepted 27 November 2017Available online 2 December 2017

Keywords:T cellsMitochondriaMitoTracker Green FMImaging flow cytometry (IFC)Amnis ImageStreamRaw Max Pixel

Mitochondria, the powerhouse of the cell, are known to remodel their membrane structures through theprocess of fusion or fission. Studies have indicated that T cells adopt different energy metabolic pheno-types, namely oxidative phosphorylation and glycolysis depending on whether they are naïve, effectorand memory T cells. It has recently been shown that changes in mitochondrial morphology dictate T cellfate via regulation of their metabolism. Our keen interest in T cell function and metabolism led us toexplore and establish a method to study mitochondria in live T cells through a novel high contentapproach called Imaging Flow Cytometry (IFC).The focus of our current study was on developing a protocol to standardize the concentration of

MitoTracker Green FM dye to observe mitochondria in live T cells using IFC. We began the study by usingwidefield microscopy to confirm the localisation of MitoTracker Green FM labelled mitochondria in live Tcells. This was followed by testing various concentrations of the dye to achieve a similar labelling patternusing IFC while eliminating false positive or negative staining. The optimization of the method used tolabel the mitochondria by IFC for analysis included standardisation of a number of important parameterssuch as dye concentration, voltage, fluorescence intensity values for acquisition and processing. IFC couldpotentially be a powerful method to study T cells in a relatively high throughput manner.

� 2017 Elsevier Inc. All rights reserved.

1. Introduction

1.1. T cells and their functional subsets

T cells are an important subset of the adaptive branch of theimmune system. They arise from lymphoid precursors in the bonemarrow, which travel through the blood and enter the thymus [1].The thymus, classified as a primary lymphoid organ, is the organwhere T cells develop, and become educated and selected toexpress a useful antigen-recognition repertoire. After selectionand final differentiation, they migrate to the periphery. This selec-tion is important to make the T cells capable of mounting animmune response against non-self-antigen fragments presentedby self Major Histocompatibility Complex (MHC) proteins, whereasthey must remain non-responsive to self-antigens. In this way,they refrain from the development of autoimmunity. Following

this selection, T cells migrate from the thymus to the periphery[2,3].

The three known major functional subsets of T cells are Naïve(TN), Effector (TE) and Memory (TM). When a TN cell residing in sec-ondary lymphoid organs such as lymph nodes, encounters its cog-nate antigen and receives an adequate amount of co-stimulatorysignal, it transforms into an activated T cell. The TN undergoesextensive proliferation termed as ‘‘clonal expansion”, ultimatelydifferentiating into TE cells. Upon clearance of antigen or infection,almost the entire population of these short-lived TE cells is lostduring the contraction phase of the response. Another small subsetof T cells derived from TN cells differentiates into long lived TMcells, which helps in mediating a quicker and enhanced immuneresponse upon subsequent exposure to the same antigen [4,5].

1.2. Immunity and metabolism: OXPHOS/glycolysis in T cells

Recently a multitude of studies point towards a connectionbetween immunity and metabolism. There is an increasing evi-dence that T cells are able to perform their respective functionsonly by virtue of their variable, function-specific energy require-ments [6]. The two known energy generation pathways adopted

12 N. Gautam et al. /Methods 134–135 (2018) 11–19

by T cells are Oxidative phosphorylation (OXPHOS) and aerobicglycolysis [7].

It has been shown that quiescent T cells such as TN and TM cellsprimarily use OXPHOS, a highly efficient (generating 36 ATP mole-cules from one glucose molecule) catabolic pathway. In contrast, TEcells adopt aerobic glycolysis, a less efficient (generating 2 ATPmolecules from one glucose molecule) anabolic pathway as theirkey energy pathway [7]. This switch to aerobic glycolysis, alsoknown as the Warburg effect, is a characteristic feature of highlyproliferative tumor cells [8]. Thus, modulation of the T cells’ meta-bolic pathways and needs facilitates their ability to perform thefunctions of the particular subset of T cells [9].

1.3. Mitochondria and immunity: Powerhouse of T cells

Studies have shown that mitochondria, double membranousorganelles, are critical regulators of immunity and play essentialroles in T cells [10]. Mitochondria are known as the powerhouseof the cell because it is the site where OXPHOS occurs. In T cells,as in other cells, energy substrates like glucose, fatty acids and glu-tamine get converted into products which participate in the tricar-boxylic acid (TCA) cycle localised in the mitochondrial matrix. Thiscycle generates two electron carriers namely NADH and FADH2,which serve as electron donors for the electron transport chain(ETC) comprising of four complexes present in the inner mitochon-drial membrane. The sequential transfer of electrons builds up aproton gradient due to simultaneous pumping of protons fromthe matrix into the inter membrane space. The proton gradient isthus utilised to fuel the production of ATP from ADP and inorganicphosphate [11,12]. Moreover, mitochondria aid in antigen-inducedT cell activation via generation of reactive oxygen species (ROS)[13], regulation of TM metabolism via fatty acid synthesis [14], Tcell aging, autoimmunity [12], maintenance of calcium homeosta-sis [15], regulation of programmed cell death [16] and anti-viralimmunity [17].

1.4. Mitochondrial morphology: Indicator of OXPHOS/glycolysis in Tcells

Mitochondria possess the intrinsic property to continuouslychange shape by undergoing fission or fusion. This is tightly linkedto their ability to perform their functions [18]. A recent study indi-cates that changes in the mitochondrial morphology dictate T cellfate via regulation of their metabolism [19]. Naïve T cells show a‘‘fused” morphology with relatively low density, but TM cellsundergo mitochondrial fusion resulting in mitochondria with elon-gated tubular morphology facilitating efficient OXPHOS [19]. Onthe other hand, TE cells undergo mitochondrial fission and havepunctate mitochondrial morphology resulting in inefficientOXPHOS. The punctate mitochondria adopt glycolysis as theirmajor pathway of energy production [19].

1.5. MitoTracker probes

The advent of fluorescent dyes has revolutionised the study ofvarious organelles under different experimental conditions, asthese dyes are able to directly label the organelles allowing themto be visualised under a microscope. Fluorescent dyes used to labelmitochondria include the MitoTracker probes, which are availableconjugated to a wide variety of fluorochromes. Although they havedistinct properties that allow the study of the functions, morphol-ogy and location of mitochondria, they share a common stainingprinciple [20,21].

MitoTracker probes contain a mildly thiol-reactive chemicalgroup which mediates selective binding to mitochondria [22,23].Another advantage of these probes is the ease with which mito-

chondrial staining is achieved, as they can diffuse non-energetically across the plasma membrane.

1.6. MitoTracker Green FM dye

MitoTracker Green FM dye has bright green fluorescence andthree specific advantages which make it an appropriate choice tovisualize mitochondria in live T cells [24]. Firstly, it is a highlymitochondria-selective dye and a suitable chemical tool to studythe mitochondrial mass due to its insensitivity to changes in mito-chondrial membrane potential. [25]. Due to its insensitivity tomembrane potential, this dye can be used to label the total mito-chondrial pool (both healthy and unhealthy mitochondria). Sec-ondly, it can produce a bright mitochondria specific signal evenat low concentrations due to its photo-stability. Lastly, the brightsignal and selectivity for mitochondria can be attributed to theability of the dye to fluoresce only in the lipid environment foundwithin the mitochondria and not in aqueous solution, whichgreatly decreases the background, making it an ideal choice forimaging purposes [24].

It is important to note that although our study discusses the useof only MitoTracker Green FM to visualize mitochondria in T cells,other bulk mitochondrial stains like MitoTracker Red and Deep Redcan also be used in this procedure. Both MitoTracker Red and DeepRed are used to label live mitochondria and this labelling is depen-dent on membrane potential (unlike labelling using MitoTrackerGreen FM) [22]. Therefore, only the mitochondria which arehealthy and thus able to maintain the membrane polarity to gener-ate ATP during oxidative phosphorylation are labelled. An interest-ing set of experiments using this protocol could be performed bycombining MitoTracker Red/MitoTracker Deep Red and Mito-Tracker Green FM to assess the different physiological states ofmitochondria versus total mitochondria in T cells subjected to var-ious experimental conditions [26,27].

1.7. High content imaging flow cytometry (Amnis): A novel approachto study T cell mitochondria

Conventionally, microscopy has been used for imaging singlecells or small populations of cells while flow cytometry has beenused for multi-wavelength fluorescence analysis of millions ofcells. Recently, the development of a novel technique termed Imag-ing Flow cytometry (IFC) has transformed the world of highthroughput imaging [28]. IFC, as its name suggests, is an amalga-mation of microscopy and flow cytometry which allows the userto obtain images of thousands of cells in a single-cell suspension.

We were keen to utilise this novel approach to study the mito-chondria in T cells for the following reasons. IFC analysis of mito-chondria in T cells can be performed in an unbiased, highcontent manner. In contrast, microscopy can potentially be limitedby the number of cells that can be analysed. IFC allows us to detectthe fluorophore-tagged molecule on the cell surface as well asinternal to the cell, which cannot be achieved by flow cytometry[29]. Moreover, it allows automated computational analysis of vastamounts of data (typically, thousands of images) using >85 variedmorphological characteristics that can be user defined and can beapplied to both brightfield and fluorescent channels. It also pro-vides the user a unique ability to design a combination of masksor formulas to study the parameters of their choice [29].

In this article, we discuss the methods used to optimize the con-centration of MitoTracker Green FM dye to observe mitochondriain live T cells using IFC. First, the MitoTracker Green FM labelledmitochondria in live T cells were visualised by widefield micro-scopy to confirm their localisation. Next, the following two param-eters were strictly monitored to optimize the IFC protocol: a) thatthe mitochondrial localisation from the IFC was similar to what

N. Gautam et al. /Methods 134–135 (2018) 11–19 13

was observed with microscopy, and b) that the concentration ofdye used did not report false positive or false negative staining.

2. Methodology

2.1. Mice

Wild type OT-I TCR transgenic mice were used for all theexperiments. These mice exclusively express OT-I TCR whichspecifically recognizes ovalbumin (OVA) residue 257–264 pre-sented by mouse MHCI H-2 Kb. Therefore, OT-I TCR transgenicmice generate MHCI-restricted, OVA-specific, CD8 + T cells (OT-IT cells). OTI T cells are powerful tool to study peripheral CD8 Tcells [30]. The OTI mice were obtained from Jackson (Jax strainnumber 003831), and were maintained in restricted flora (RF)facilities at Comparative Medicine, NUS under IACUC approvalB13-04470 and BR17-04780.

2.2. T cell activation

To activate CD8 T cells, spleens were isolated from OT-I TCRtransgenic mice and a single cell suspension of splenocytes wasprepared. Red blood cells were then lysed using 1X Ack(Ammonium-Chloride-Potassium) lysis buffer (1.5 M NH4Cl, 100mM KHCO3, 1 mM EDTA, pH 7.2) and splenocytes were stimulatedusing 250 nM OVA peptide plus 10U/ml recombinant human IL-2(Peprotech) for 2 days at 37 �C in complete RPMI medium (500ml RPMI (HyClone), 10% fetal bovine serum (HyClone), 10 U/mlpenicillin/streptomycin (gibco), 2 mM L-glutamine (gibco), 10 mMHEPES (HyClone), b-mercaptoethanol (sigma) and 1 mM sodiumpyruvate (gibco)).

2.3. Controls and samples for IFC

After two days of stimulation, the following types of sampleswere prepared:

1. Unstained sample: to check for autofluorescence.2. Single color control samples: to be used for compensation

(a) Hoechst 33,342, (b) CD8 APC, (c) MitoTracker Green FM(5 nM, 50 nM, 100 nM).

3. Fluorescence minus one (FMO) sample: samples stained withCD8 APC and Hoechst 33,342 but not MitoTracker Green FM.This was used to minimize the possibility of false positive sig-nals arising from dead cells or nonspecific binding.

4. Three-color stained samples: T cells were stained with Hoechst33,342, CD8 APC and MitoTracker Green FM. Three sampleswere prepared with similar concentrations of the former twobut varying concentrations of MitoTracker Green FM: 5 nM,50 nM or 100 nM.

Table 1Description of dyes. The table presents the manufacturer details, target organelle/molecul

Stain Manufacturer Details Target Organelle/Molecule

Hoechst 33,342 Invitrogen (Catalog No.H3570)

Nucleus

APC Rat Anti Mouse CD8aClone 53–6.7 (RUO)

BD Biosciences (Catalog No.5,61,093)

CD8 co-receptor onCD8 T cells

MitoTracker Green FM Thermo Fischer Scientific(Catalog No. M7514)

Mitochondria

2.4. T cell staining protocol

2.4.1. Imaging flow cytometry (IFC) sample preparation1 � 106 stimulated CD8 T cells in a 96-well plate were first

stained with 100ll of CD8 APC at 1:300 dilution in flow cytometrystaining buffer (0.5% BSA in 1X DPBS (HyClone)) for 30 mins on ice.To remove excess dye, T cells were then washed with 200 ll ofFACS Wash Buffer, followed by centrifugation at 300g for 5 minsat 4 �C. Further, T cells were stained using 125 ll of the desiredconcentration (5 nM, 50 nM, and 100 nM) of MitoTracker GreenFM dye diluted in pre-warmed RPMI without serum for 35 minsat 37 �C (Table 1). After 25 mins Hoechst 33,342 (1:125 dilutionof 1 mg/ml solution) was added to MitoTracker Green FM stainingsolution to stain the nucleus. Finally, T cells were washed oncewith 1X DPBS, and resuspended in 50 ll of 1X DPBS. These werethen transferred into a 1.5 ml tube for acquisition using AmnisImageStream.

It should be noted that the amount of the dye must be propor-tionate to the number of cells used. Therefore, this protocol is rec-ommended for a cell density of 0.5–1.5 � 106 T cells. For cellnumbers >1.5 � 106, dye concentration should be optimised.

2.4.2. Widefield microscopy sample preparationThe protocol for sample preparation for microscopy was similar

to that for IFC with some changes. For microscopy, half a millioncells were stained with 200 nM of MitoTracker Green FM dye (asrecommended by the manufacture’s protocol). Cells were washedtwice with 1X DPBS followed by resuspension in 100 ll of DPBS.Finally, the cells were seeded in an 8-well Lab-Tek II chamber(Fisher Scientific, USA) and imaged. Images were acquired on anOlympus IX83 inverted microscope with a 100� oil immersion lenswith the appropriate filters (DAPI, FITC and Cy5 for Hoechst 33,342,MitoTracker Green FM and CD8-APC, respectively). CellSens soft-ware was used for acquisition.

2.5. Acquisition using INSPIRE on ImageStream X MKII imaging flowcytometer (ISX)

2.5.1. Acquisition template: Gating strategy, laser, channel and voltageset up

The standardization of MitoTracker Green FM dye was per-formed using ImageStream X MKII imaging flow cytometer (ISX)(AMNIS, EMD Millipore). Firstly, the ISX was calibrated usingASSIST software followed by generation of an acquisition templatecomprising of the following dot plots (Fig. 1): (i) Raw Max Pixel ofMitoTracker Green FM (channel 2) versus Hoechst 33,342 (channel1), (ii) Raw Max Pixel of CD8 APC (channel 5) versus Hoechst33,342 (channel 1), (iii) Aspect ratio versus area on brightfield(channel 4) wherein gating was done to select ‘‘single cells” withan aspect ratio between 0.8 and 1.0 and area between 100 and300 (equivalent to 25 lm2�75 lm2),and (iv) An intensity plot of

e and dilution for the various dyes used in this study.

Stock Dilution Used

1 mg/ml (a sub stock of 1 mg/ml wasprepared by diluting the suppliedstock of 10 mg/ml in water)

1:125 of 1 mg/ml (in serum- freeRPMI)

0.2 mg/ml 1: 300 (in flow cytometrystaining buffer)

1 mM (Add 74.4 ll DMSO andthoroughly vortex)

200 nM 100 nM 50 nM 5 nM(Diluted in pre-warmed RPMIwithout serum)

Fig. 1. Gating strategy for acquisition. (i), (ii), (iii) and (iv) depict the sequential gating strategy adopted during acquisition using INSPIRE software on ISX. (i) and (ii) RawMaxPixel dot plots of MitoTracker Green FM (channel 2) versus Hoechst 33342 (channel 1) and CD8 APC (channel 5) versus Hoechst 33342 (channel 1). (iii) Gating of single cellsbased on a dot plot of aspect ratio versus area on brightfield (channel 4). (iv) Gating of single cells as CD8 and MitoTracker Green FM positive (CD8 and Mito positive) based onthe intensity dot plot of CD8 APC (channel 5) and MitoTracker Green FM (channel 2).

14 N. Gautam et al. /Methods 134–135 (2018) 11–19

CD8 APC (channel 5) versus MitoTracker Green FM (channel 2)wherein gating was done for ‘‘CD8 and MitoTracker Green FM dou-ble positive cells”.

The laser voltages used for acquiring different fluorophoreswere then adjusted (as described in Table 2) by first acquiringthe three-colored stain ensuring that the Raw Max Pixels of theirrespective channels remained well within the detection capacity(below 4094 Raw Max Pixel) of ISX. Samples were acquired atlow speed, high sensitivity and 60� magnification (0.3 lm/pixel)to achieve high resolution.

2.5.2. Sample acquisitionFirst, the three-color stained samples with varying concentra-

tions of MitoTracker Green FM were acquired in increasing order

Table 2Channel and laser voltage settings during acquisition. The table shows the acquisition set

Stain Laser Voltage Exc

Hoechst 33,342 357 nm 1 mW 350CD8 APC 642 nm 150 mW 650MitoTracker Green FM 488 nm Varying Voltages 490Brightfield – – –Side Scatter (SSC) – – –

of their concentrations. The voltage of the 488 nm laser used forMitoTracker Green FM was changed according to the three differ-ent concentrations used (discussed later in Section 3.2). The lasers,brightfield and SSC were switched on while acquiring the three-color stained samples and 5000 of the desired ‘‘CD8 and Mito-Tracker Green FM double positive cells” were collected (Fig. 1).After that, a thorough wash with 10% bleach followed by 1X PBSwas done to avoid any contamination while acquiring the single-color controls. For single color controls, all the lasers wereswitched on and brightfield and SSC were turned off. As the bright-field was switched off, the initial gate for ‘‘single cells” based onbrightfield was no longer valid and therefore, only 500 events ofnon-gated cells i.e. ‘‘All” were collected (Fig. 1). We ensured thatsingle color controls for Hoechst 33,342 and CD8 APC were

tings for the lasers with details about the voltage and channels used.

itation wavelength Emission wavelength Channel Acquired

nm 461 nm Channel 1nm 660 nm Channel 5nm 516 nm Channel 2

– Channel 4– Channel 6

N. Gautam et al. /Methods 134–135 (2018) 11–19 15

acquired at all the different voltages of MitoTracker Green FM dyeused at their respective concentrations. Also, a thorough wash with10% bleach followed by 1X PBS was done after acquisition of eachsingle color control.

2.6. Analysis of mitochondria in T cells using IDEAS software Version6.1

The raw images, i.e. images without compensation acquired onISX using INSPIRE application are stored as .rif (Raw Image File)files. The .rif files were then converted into .cif (CompensatedImage File) using IDEAS Software Version 6.1 by applying a com-pensation matrix created as a .ctm file comprising of all the singlecolor controls. Technically, three different compensation matriceswere created comprising of Hoechst 33,342-only stained controland CD8 APC-only stained control acquired at similar voltagesacross all the three matrices, combined with three MitoTrackerGreen FM-only stained samples (5 nM, 50 nM and 100 nM),acquired at varying voltages. The analysis was then performed on.cif files using the ‘‘Begin Analysis” wizard. As a part of this wizard,a histogram was created based on the gradient RMS of brightfield(channel 4) (Fig. 2). The cells in focus were then gated as ‘‘focused”by clicking different bins across the histogram and simultaneouslymonitoring them in the image gallery. Generally, cells having gra-dient RMS > 40 are chosen as they are expected to be in focus. The‘‘focused” cells were further gated as ‘‘single” cells using thebrightfield channel (channel 4) such that cells with a higher aspectratio (>0.8) and intermediate area (100–300, equivalent to 25lm2�75 lm2) were gated as single cells as mentioned in Sec-tion 2.5.1. The intermediate area eliminated dead cells or beadswhereas the aspect ratio eliminated doublets for analysis. ‘‘CD8and MitoTracker Green double positive cells” were gated basedon their intensity dot plots. The file resulting from analysis wasstored as .daf (Data Analysis File). The analysis template was savedand used for analysis of other data sets.

3. Results and discussion

3.1. Mitochondria are located around the nucleus in T cells

Mitochondria in T cells were visualised using 200 nM of Mito-Tracker Green FM dye through widefield microscopy (Fig. 3). Weobserved that the mitochondria are located around the nucleusin stimulated T cells. Microscopy confirmed that the mitochondriaare restricted to a part of the cytoplasm and do not occupy theentire cytoplasm.

Fig. 2. Gating strategy for analysis. (i), (ii) and (iii) Sequential gating strategy adopted whratio versus area based on brightfield (channel 4). (ii) Selection of single cells using a dot psingle cells as CD8 and Mitotracker positive (CD8 and Mito positive) based on the inten

3.2. Voltage adjustments for concentration optimization ofMitoTracker Green FM

This section describes the rationale behind testing differentlaser voltages for MitoTracker Green FM (channel 2) acquisition.Firstly, the highest voltage of 200 mW was applied across all theconcentrations followed by a gradual decrease where required toget a Raw Max Pixel Value below saturation level (4094).

For 5 nM, at the highest voltage of 200 mW the Raw Max Pixelvalue was observed to be <300. This was too low a value to be con-sidered as a positive signal. For 50 nM, at the voltage of 150 mWthe Raw Max Pixel yield was in the range of 500–2000, whichwas considered to be a good range for a positive signal. On theother hand, at 100 nM and voltage of 15mW the Raw Max Pixelwas >4094 suggesting signal saturation (Fig. 4A). Hence, 50 nM ofMitoTracker Green FM was concluded to be the optimum concen-tration to label stimulated T cells.

To further validate the optimum dye concentration, FMO andunstained control samples acquired at 150 mW were analysedand compared. Results clearly indicated that stimulated T cells dis-played autofluorescence in channel 2 (MitoTracker Green FM) andchannel 5 (CD8 APC) reflected by their Raw Max Pixel values of upto 200 (Fig. 4B). On the other hand, there was negligible autofluo-rescence detected in channel 1 (Hoechst 33,342) as seen by its lowRaw Max Pixel value of �40 (Fig. 4B). FMO for MitoTracker Greenwas used to identify true positively stained cells in channel 2(Mitotracker Green FM). This allowed us to eliminate signals fromnonspecific binding of dye or cell debris in channel 2, and the RawMax Pixel value was up to 200 for the FMO. Also, the intensity his-tograms for both unstained and FMO were reflective of their RawMax Pixel values and appeared in the range of 1e3-1e5 while theirpeak intensity fell between 1e3-1e4 (Table 3, Fig. 4C). Therefore, weconcluded that MitoTracker Green FM stained T cells with RawMax Pixel values >500 (upper limit-below 4094) and the peakintensity beyond 1e4 can be gated as ‘‘true positively stained” cells.

The conclusion drawn from unstained and FMO controlsstrongly indicated that 5 nM is not the optimum concentrationbecause despite increasing the voltage to 200 mW the observedRaw Pixel value was <300 (Fig. 4A). Although the Raw Max Pixelof 300 was higher than unstained and FMO controls (�200), theintensity range showed an overlap with that of unstained andFMO samples (1e3-1e5) (Table 3, Fig. 4C). This is indicated by theblue line in Fig. 4C which signifies the cut off for negatively stainedcells with respect to the FMO control. Therefore, this concentrationwas not considered optimum because it was incapable of accu-rately discriminating the positively stained cells from the negativeones. For 50 nM, a voltage of 150 mW yielded a Raw Max Pixel

ile analysing on IDEAS 6.1 software. (i) Gating of focused cells using a histogram oflot of aspect ratio versus area based on brightfield (channel 4). (iii) Further gating ofsity dot plot of CD8 APC (channel 5) and MitoTracker Green FM (channel 2).

Fig. 3. Mitochondrial location in CD8 T cells post activation. Location of live mitochondria in wild type OTI CD8 T cells activated by OVA and IL-2 using widefield microscopyat 100X. The four panels depict nucleus in blue (Hoechst 33342), surface CD8 co-receptor in red (CD8 APC), mitochondria in green (MitoTracker Green FM), and an overlap ofthese. Scale bar, 5 lm.

16 N. Gautam et al. /Methods 134–135 (2018) 11–19

range of 500–2000 (Fig. 4A). This included all the different popula-tions of T cells with varying intensities of mitochondrial staininghighlighted by its intensity range 1e4-1e6 (Table 3, Fig. 4C). Thiswas the optimum concentration as it could precisely discriminatebetween the positively and negatively stained cells. For 100 nM,despite reducing the voltage to 15 mW the Raw Max pixel was sat-urated (i.e. 4094, Fig. 4B) with the intensity range of 1e5-1e6 andpeak intensity falling between these values. Thus, it resulted in sat-uration of the signal (Table 3, Fig. 4C).

In this study, we have compared RawMax Pixel of different con-centrations of MitoTracker Green FM with its respective FMO as itwas a one dye: one laser scenario, i.e. MitoTracker Green FM wascontrolled by a single laser (488 nm). An important thing to con-sider would be to use Max Pixel rather than Raw Max Pixel underconditions where two dyes are excited by a single laser or/and abright and variable dye such as PE is included in the staining panel.This is because Max Pixel is compensated, and hence this parame-

ter must then be compared with FMO to optimize the working con-centration of a dye.

3.3. Optimum concentration of MitoTracker Green FM to visualizemitochondria in T cells

The optimum concentration 50 nM of MitoTracker Green FMwas further validated by analysing the images obtained by IFC.To determine the localisation of the mitochondria in the imagesgenerated using IFC, they were compared with those obtained fromwidefield microscopy. Images for IFC were handpicked from theCD8/MitoTracker Green FM double positive population for eachindividual concentration of MitoTracker Green FM.

Images of mitochondria in T cells generated using 5 nM concen-tration at 60� resulted in an extremely weak signal in >95% of thecells that was insufficient to capture the mitochondrial structure(Fig. 5). Therefore, it was not possible to assess the mitochondrial

Fig. 4. Comparison of RawMax Pixel and intensity for different concentrations of MitoTracker Green FM with respect to unstained and FMO controls. RawMax Pixel dot plotsof: A. MitoTracker Green FM (channel 2) versus Hoechst 33342 (channel 1) of 5 nM, 50 nM and 100 nM. B. MitoTracker Green FM (channel 2) versus Hoechst 33342 (channel1) and CD8 APC (channel 5) versus Hoechst 33342 (channel 1) of unstained and FMO controls. C. MitoTracker Green FM (channel 2) versus Hoechst 33342 (channel 1) andintensity histograms of MitoTracker Green FM (channel 2) of unstained, FMO, 5 nM, 50 nM and 100 nM samples. The blue line represents the threshold used for gating thetrue positively stained cells.

Table 3Raw Max Pixel and intensity range of controls and samples. This table represents RawMax Pixel range and intensity range of unstained control and FMO control acquired ata voltage of 150 mW by the 488 nm laser (channel 2, MitoTracker Green FM). It alsotabulates the Raw Max Pixel and intensity range of different concentrationsof MitoTracker Green FM acquired at different voltages by the by 488 nm laser(channel 2, MitoTracker Green FM).

Concentration Laser Voltage Raw Max PixelRange

IntensityRange

Unstained 150 mW Upto 200 1e3-1e5

FMO (MitoTrackerGreen FM)

150 mW Upto 200 1e3-1e5

5 nM 200 mW (Max) Below 300 1e3-1e5

50 nM 150 mW 500–2500 1e4-1e6

100 nM 15 mW 1000–4000 (saturated) 1e5-1e6

N. Gautam et al. /Methods 134–135 (2018) 11–19 17

structure with 5 nM MitoTracker Green FM as evidenced by thelow Raw Pixel values and intensity range as discussed inSection 3.2.

For 100 nM, �60–70% of the images from the IFC showed a sat-urated signal even after compensation (Fig. 5). Also, in theseimages the mitochondrial signal was seen to occupy the entire sur-face of T cells, which is in contrast to what was observed withmicroscopy. Additionally, the saturated signal was reflective of sat-urated Raw Max Pixel and a high intensity range as discussed inSection 3.2.

Hence, 50 nM of MitoTracker Green FM was found to be theoptimal concentration to label the mitochondria in T cells. Itallowed the visualisation of the correct mitochondrial location(around the nucleus) in 85–90% of the cells and captured thelow, intermediate, and high intensity-stained mitochondria(Fig. 5). The images shown here are best representatives of opti-mally stained mitochondria in T cells, shown by sharp, focussedimages displaying the mitochondria around the nucleus asobserved with microscopy. Moreover, it also presented the appro-priate range of Raw Max Pixel (500–2000) and intensity (1e4-1e6)as discussed in Section 3.2.

Fig. 5. Concentration optimization of MitoTracker Green FM. Compensated images of OTI CD8 T cells obtained by imaging flow cytometry using 5 nM, 50 nM and 100 nM ofMitoTrakcer Green FM post stimulation with OVA peptide and rhIL-2 for 2 days. Nucleus in purple (Hoechst 33342, channel 1), mitochondria in green (MitoTracker Green FM,channel 2), CD8 co-receptor in red (CD8 APC, channel 5) and a composite view showing all stains together (merge). Scale bar, 10 lm.

18 N. Gautam et al. /Methods 134–135 (2018) 11–19

We therefore conclude that 1 � 106 T cells stained using 50 nMof MitoTracker Green FM acquired by 488 nm laser at a voltage of150mW observed in channel 2 are the best parameters to studymitochondria in T cells using ImageStream X MKII.

4. Conclusion

This protocol would allow an immunologist to study the T cellmitochondria in a high content manner, which can further beextended to assess the morphological features of mitochondria inT cells. The format can also be used to investigate changes in T cellmitochondria with drug treatments or to study the roles of variousgenes involved in T cell mitochondrial function using animalknockout models.

5. Points to remember

1) After reconstituting MitoTracker Green FM dye with DMSO(according to manufacturer’s protocol), stock aliquots of 1ll should be made to avoid freeze-thaw cycles [24].

2) It is highly recommended to use RPMI without serum todilute MitoTracker Green FM dye for preparing workingsolutions as serum may oxidise some of the reduced ver-sions of mitotracker probes causing reduced labelling [24].

3) It is important to prewarm the serum free RPMI at 37 �Cwhile preparing working solutions to achieve proper stain-ing since the cells are incubated at this temperature [24].

4) Standardization of optimal concentration of dye is crucial toavoid non-specific staining [24].

5) It is very important to test various cell-to-dye ratios todetermine the concentration for optimal staining.

6) When multi-color staining panels where two dyes areexcited by a single laser, or when a bright dye is used, onemust use Max pixel to compare with FMO to standardizethe concentration of the dyes used.

7) It is extremely important to include unstained and FMOsamples as controls to avoid false positive signals. Similarly,a threshold for positive signal should be set up during stan-dardization for IFC.

Acknowledgements

We sincerely acknowledge the technical support provided bythe Amnis team members, Elsie Lip Chuen Ho, Geok Choo Ng andMay Ling Leong at Department of Microbiology and Immunology,NUS. This research is supported by the Singapore Ministry ofHealth’s National Medical Research Council under grant CBRG15-may017, and by Singapore Ministry of Education’s grant 2014-T2-1-136.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at https://doi.org/10.1016/j.ymeth.2017.11.015.

References

[1] E. Zhao, H. Xu, L. Wang, I. Kryczek, K. Wu, Y. Hu, G. Wang, W. Zou, Bone marrowand the control of immunity, Cell Mol. Immunol. 9 (1) (2012) 11–19.

N. Gautam et al. /Methods 134–135 (2018) 11–19 19

[2] T.K. Starr, S.C. Jameson, K.A. Hogquist, Positive and negative selection of T cells,Annu. Rev. Immunol. 21 (2003) 139–176.

[3] Y. Takahama, Journey through the thymus: stromal guides for T-celldevelopment and selection, Nat. Rev. Immunol. 6 (2) (2006) 127–135.

[4] J.T. Chang, E.J. Wherry, A.W. Goldrath, Molecular regulation of effector andmemory T cell differentiation, Nat. Immunol. 15 (12) (2014) 1104–1115.

[5] S.M. Kaech, E.J. Wherry, R. Ahmed, Effector and memory T-cell differentiation:implications for vaccine development, Nat. Rev. Immunol. 2 (4) (2002) 251–262.

[6] M.D. Buck, D. O’Sullivan, E.L. Pearce, T cell metabolism drives immunity, J. Exp.Med. 212 (9) (2015) 1345–1360.

[7] E.L. Pearce, Metabolism in T cell activation and differentiation, Curr. Opin.Immunol. 22 (3) (2010) 314–320.

[8] V.A. Gerriets, J.C. Rathmell, Metabolic pathways in T cell fate and function,Trends Immunol. 33 (4) (2012) 168–173.

[9] C.J. Fox, P.S. Hammerman, C.B. Thompson, Fuel feeds function: energymetabolism and the T-cell response, Nat. Rev. Immunol. 5 (11) (2005) 844–852.

[10] E.L. Mills, B. Kelly, L.A.J. O’Neill, Mitochondria are the powerhouses ofimmunity, Nat. Immunol. 18 (5) (2017) 488–498.

[11] S.E. Weinberg, L.A. Sena, N.S. Chandel, Mitochondria in the regulation of innateand adaptive immunity, Immunity 42 (3) (2015) 406–417.

[12] T. Chao, H. Wang, P.C. Ho, Mitochondrial Control and Guidance of CellularActivities of T Cells, Front. Immunol. 8 (2017) 473.

[13] L.A. Sena, S. Li, A. Jairaman, M. Prakriya, T. Ezponda, D.A. Hildeman, C.R. Wang,P.T. Schumacker, J.D. Licht, H. Perlman, P.J. Bryce, N.S. Chandel, Mitochondriaare required for antigen-specific T cell activation through reactive oxygenspecies signaling, Immunity 38 (2) (2013) 225–236.

[14] G.J. van der Windt, B. Everts, C.H. Chang, J.D. Curtis, T.C. Freitas, E. Amiel, E.J.Pearce, E.L. Pearce, Mitochondrial respiratory capacity is a critical regulator ofCD8+ T cell memory development, Immunity 36 (1) (2012) 68–78.

[15] R. Uzhachenko, A. Shanker, G. Dupont, Computational properties ofmitochondria in T cell activation and fate, Open Biol. 6 (11) (2016).

[16] R.J. Youle, M. Karbowski, Mitochondrial fission in apoptosis, Nat. Rev. Mol. CellBiol. 6 (8) (2005) 657–663.

[17] J. Nunnari, A. Suomalainen, Mitochondria: in sickness and in health, Cell 148(6) (2012) 1145–1159.

[18] J.R. Friedman, J. Nunnari, Mitochondrial form and function, Nature 505 (7483)(2014) 335–343.

[19] M.D. Buck, D. O’Sullivan, R.I. Klein Geltink, J.D. Curtis, C.H. Chang, D.E. Sanin, J.Qiu, O. Kretz, D. Braas, G.J. van der Windt, Q. Chen, S.C. Huang, C.M. O’Neill, B.T.Edelson, E.J. Pearce, H. Sesaki, T.B. Huber, A.S. Rambold, E.L. Pearce,Mitochondrial Dynamics Controls T Cell Fate through MetabolicProgramming, Cell 166 (1) (2016) 63–76.

[20] A. Csiszar, N. Labinskyy, J.T. Pinto, P. Ballabh, H. Zhang, G. Losonczy, K. Pearson,R. de Cabo, P. Pacher, C. Zhang, Z. Ungvari, Resveratrol induces mitochondrialbiogenesis in endothelial cells, Am. J. Physiol. Heart Circ. Physiol. 297 (1)(2009) H13–H20.

[21] A.Y. Abramov, T.K. Smulders-Srinivasan, D.M. Kirby, R. Acin-Perez, J.A.Enriquez, R.N. Lightowlers, M.R. Duchen, D.M. Turnbull, Mechanism ofneurodegeneration of neurons with mitochondrial DNA mutations, Brain 133(Pt 3) (2010) 797–807.

[22] M. Poot, Y.Z. Zhang, J.A. Kramer, K.S. Wells, L.J. Jones, D.K. Hanzel, A.G. Lugade,V.L. Singer, R.P. Haugland, Analysis of mitochondrial morphology and functionwith novel fixable fluorescent stains, J. Histochem. Cytochem. Official J.Histochem. Soc. 44 (12) (1996) 1363–1372.

[23] A.D. Presley, K.M. Fuller, E.A. Arriaga, MitoTracker Green labeling ofmitochondrial proteins and their subsequent analysis by capillaryelectrophoresis with laser-induced fluorescence detection, J. Chromatogr. B793 (1) (2003) 141–150.

[24] MitoTracker� Mitochondrion-Selective Probes - Thermo Fisher Scientific.[25] W. Pendergrass, N. Wolf, M. Poot, Efficacy of MitoTracker Green and

CMXrosamine to measure changes in mitochondrial membrane potentials inliving cells and tissues, Cytometry A 61 (2) (2004) 162–169.

[26] S.P. Elmore, Y. Nishimura, T. Qian, B. Herman, J.J. Lemasters, Discrimination ofdepolarized from polarized mitochondria by confocal fluorescence resonanceenergy transfer, Arch. Biochem. Biophys. 422 (2) (2004) 145–152.

[27] T. Simsek, F. Kocabas, J. Zheng, R.J. Deberardinis, A.I. Mahmoud, E.N. Olson, J.W.Schneider, C.C. Zhang, H.A. Sadek, The distinct metabolic profile ofhematopoietic stem cells reflects their location in a hypoxic niche, Cell StemCell 7 (3) (2010) 380–390.

[28] D. Basiji, M.R. O’Gorman, Imaging flow cytometry, J. Immunol. Methods 423(2015) 1–2.

[29] I.A. Vorobjev, N.S. Barteneva, Quantitative functional morphology by imagingflow cytometry, Methods Mol. Biol. 1389 (2016) 3–11.

[30] K.A. Hogquist, S.C. Jameson, W.R. Heath, J.L. Howard, M.J. Bevan, F.R. Carbone, Tcell receptor antagonist peptides induce positive selection, Cell 76 (1) (1994)17–27.