Corrections

BIOPHYSICS AND COMPUTATIONAL BIOLOGYCorrection for “Structural insights into the histone H1-nucleo-some complex,” by Bing-Rui Zhou, Hanqiao Feng, HidenoriKato, Liang Dai, Yuedong Yang, Yaoqi Zhou, and Yawen Bai,which appeared in issue 48, November 26, 2013, of Proc NatlAcad Sci USA (110:19390–19395; first published November 11,2013; 10.1073/pnas.1314905110).The authors note that, due to a printer’s error, references

41–50 appeared incorrectly. The corrected references follow.

41. Schalch T, Duda S, Sargent DF, Richmond TJ (2005) X-ray structure of a tetranucleo-some and its implications for the chromatin fibre. Nature 436(7047):138–141.

42. Clore GM, Tang C, Iwahara J (2007) Elucidating transient macromolecular interactionsusing paramagnetic relaxation enhancement. Curr Opin Struct Biol 17(5):603–616.

43. Dominguez C, Boelens R, Bonvin AM (2003) HADDOCK: a protein-protein dockingapproach based on biochemical or biophysical information. J Am Chem Soc 125(7):1731–1737.

44. Thakar A, et al. (2009) H2A.Z and H3.3 histone variants affect nucleosome structure:biochemical and biophysical studies. Biochemistry 48(46):10852–10857.

45. Vogler C, et al. (2010) Histone H2A C-terminus regulates chromatin dynamics, re-modeling, and histone H1 binding. PLoS Genet 6(12):e1001234.

46. Wong H, Victor JM, Mozziconacci J (2007) An all-atom model of the chromatin fibercontaining linker histones reveals a versatile structure tuned by the nucleosomal re-peat length. PLoS ONE 2(9):e877.

47. Lee KM, Hayes JJ (1998) Linker DNA and H1-dependent reorganization of histone-DNA interactions within the nucleosome. Biochemistry 37(24):8622–8628.

48. Boulikas T, Wiseman JM, Garrard WT (1980) Points of contact between histone H1and the histone octamer. Proc Natl Acad Sci USA 77(1):127–131.

49. Travers AA, Muyldermans SV (1996) A DNA sequence for positioning chromatosomes.J Mol Biol 257(3):486–491.

50. Goytisolo FA, et al. (1996) Identification of two DNA-binding sites on the globulardomain of histone H5. EMBO J 15(13):3421–3429.

www.pnas.org/cgi/doi/10.1073/pnas.1323266111

DEVELOPMENTAL BIOLOGYCorrection for “Organ-specific function of adhesion G protein-coupled receptor GPR126 is domain-dependent,” by ChinmoyPatra, Machteld J. van Amerongen, Subhajit Ghosh, FilomenaRicciardi, Amna Sajjad, Tatyana Novoyatleva, Amit Mogha,Kelly R. Monk, Christian Mühlfeld, and Felix B. Engel, whichappeared in issue 42, October 15, 2013, of Proc Natl Acad Sci USA(110:16898–16903; first published September 30, 2013; 10.1073/pnas.1304837110).The authors note that on page 16902, right column, third full

paragraph, lines 24–25 “5’-CGGGTTGGACTCAAGACGATAG-3’” should instead appear as “5’-ACAGAATATGAATACCTGA-TACTCC-3’.”

www.pnas.org/cgi/doi/10.1073/pnas.1323830111

PHYSICSCorrection for “Stable three-dimensional metallic carbon with in-terlocking hexagons,” by ShunhongZhang, QianWang, XiaoshuangChen, and Puru Jena, which appeared in issue 47, November 19,2013, of Proc Natl Acad Sci USA (110:18809–18813; first publishedNovember 4, 2013; 10.1073/pnas.1311028110).The authors note: “Our paper unfortunately missed refer-

ence to an earlier suggestion of the T6 structure (43). This workentitled ‘A hypothetical dense 3,4-connected carbon net andrelated B2C and CN2 nets built from 1,4-cyclohexadienoidunits’ by M. J. Bucknum and R. Hoffmann was published inJ Am Chem Soc 116:11456–11464 (1994), where the electronicstructure of a hypothetical 3,4-connected tetragonal allotrope ofcarbon is discussed. The results in this article are consistent withwhat we find. The same group had also suggested a metalliccarbon structure (44) that was published in J Am Chem Soc105:4831–4832 (1983), which we also missed to cite. We thankProf. Hoffmann for bringing these papers to our attention.”The complete references appear below.

43. Bucknum MJ, Hoffmann R (1994) A hypothetical dense 3,4-connected carbon netand related B2C and CN2 nets built from 1,4-cyclohexadienoid units. J Am Chem Soc116(25):11456–11464.

44. Hoffmann R, Hughbanks T, Kertesz M, Bird PH (1983) Hypothetical metallic allotropeof carbon. J Am Chem Soc 105(14):4831–4832.

www.pnas.org/cgi/doi/10.1073/pnas.1323385111

CELL BIOLOGYCorrection for “Visualization of repetitive DNA sequences inhuman chromosomes with transcription activator-like effectors,”by Hanhui Ma, Pablo Reyes-Gutierrez, and Thoru Pederson,which appeared in issue 52, December 24, 2013, of Proc NatlAcad Sci USA (110:21048–21053; first published December 9,2013; 10.1073/pnas.1319097110).The authors note that, due to a printer’s error, references

25–29 appeared incorrectly. The corrected references are:

25. Miyanari Y, Ziegler-Birling C, Torres-Padilla ME (2013) Live visualization of chromatindynamics with fluorescent TALEs. Nat Struct Mol Biol 20(11):1321–1324.

26. Sanjana NE, et al. (2012) A transcription activator-like effector toolbox for genomeengineering. Nat Protoc 7(1):171–192.

27. Ma H, et al. (2012) A highly efficient multifunctional tandem affinity purificationapproach applicable to diverse organisms. Mol Cell Proteomics 11(8):501–511.

28. Uetake Y, et al. (2007) Cell cycle progression and de novo centriole assembly aftercentrosomal removal in untransformed human cells. J Cell Biol 176(2):173–182.

29. Jacobson MR, Pederson T (1997) RNA traffic and localization reported by fluorescencecytochemistry. Analysis of mRNA Formation and Function, ed Richter JD (Academic,New York), pp 341–359.

www.pnas.org/cgi/doi/10.1073/pnas.1323494111

1222 | PNAS | January 21, 2014 | vol. 111 | no. 3 www.pnas.org

Visualization of repetitive DNA sequences in humanchromosomes with transcription activator-like effectorsHanhui Ma1, Pablo Reyes-Gutierrez, and Thoru Pederson1

Program in Cell and Developmental Dynamics, Department of Biochemistry and Molecular Pharmacology, University of Massachusetts Medical School,Worcester, MA 01605

Edited by Joseph G. Gall, Carnegie Institution of Washington, Baltimore, MD, and approved November 13, 2013 (received for review October 9, 2013)

We describe a transcription activator-like effector (TALE)-basedstrategy, termed “TALEColor,” for labeling specific repetitive DNAsequences in human chromosomes. We designed TALEs for thehuman telomeric repeat and fused them with any of numerousfluorescent proteins (FPs). Expression of these TALE–telomere–FPfusion proteins in human osteosarcoma’s (U2OS) cells resulted inbright signals coincident with telomeres. We also designed TALEsfor centromeric sequences unique to certain chromosomes, en-abling us to localize specific human chromosomes in live cells.Meanwhile we generated TALE–FPs in vitro and used them asprobes to detect telomeres in fixed cells. Using human cells withdifferent average telomere lengths, we found that the TALEColorsignals correlated positively with telomere length. In addition,suspension cells were followed by imaging flow cytometry to re-solve cell populations with differing telomere lengths. Thesemethods may have significant potential both for basic chromo-some and genome research as well as in clinical applications.

Transcription activator-like effectors (TALEs) are able torecognize specific DNA sequences based on sequence com-

position of repeating oligopeptide elements (1). Advances inDNA cloning technologies have enabled facile assembly ofTALEs for sequence-specific DNA recognitions as well as fusionof paired nucleases (TALENs) for genome engineering (2). Al-though TALEs and TALENs have rapidly become powerfultools for genome editing and transcription regulation (3), theirintranuclear dynamics of DNA recognition are not well un-derstood because they are typically directed to a single-copysequence, thus limiting cytological studies and applications. Itoccurred to us that at least in cases of tandemly repeated DNAsequences, it should be possible to detect chromosomal sites offluorescent TALE recognition and binding in live cells. By ex-tension, we also considered it likely that fluorescent TALEsmight be used as probes to detect DNA sequences in fixed cellpreparations, as in conventional in situ hybridization but withouta need to denature the DNA because TALEs read the targetsequence in double-stranded form. Here we report the de-velopment of such methods as applied to both human telomeresand centromeric repetitive sequences.

ResultsOur initial purpose in developing the methods to be reportedstemmed from our interest in the relative intranuclear positionsof telomeres and nucleoli in living cells. Our laboratory hadpreviously developed methods to label and track ribosomal RNAout of nucleoli in living cells (4). The genes for ribosomal RNAlie close to telomeres in the short arms of several human chro-mosomes (5) and we pondered how we might label telomeres inlive cells, as we had succeeded in doing for ribosomal RNAtranscripts themselves. One of us (H.M.) considered that be-cause TALEs recognize specific sequences in double-strandedDNA form, live cell applications would be feasible and thata telomere-specific TALE fused to a fluorescent protein mightbe a way to label the ends of chromosomes in live cells. As shownin Fig. 1A, TALEs were designed to recognize either DNAstrand of the telomeric repeat. The TALE polypeptides wereconstructed as DNA plasmids with in-frame fusions to the desired

fluorescent protein, followed by transfection and expression inhuman osteosarcoma’s (U2OS) cells. Fig. 1B shows the results ofan experiment in which the TALEs TelL20 or TelR20 targetingto either strand of the telomere repeats were coexpressed for24 h. Numerous discrete fluorescent foci were observed in in-terphase cells with either of the two TALEs. TALEs recognizespecific DNA sequences in native double-stranded DNA byreading from the major grove. The fact that coexpression ofTALE–fluorescent proteins (FPs) designed for either strand ofthe telomeric repeat resulted in similar patterns of discrete nu-clear foci with the two colors displaying complete spatial co-incidence indicates that both strands of the telomeric repeatare accessible. U2OS cells are aneuploid, with ∼65 chromo-somes (6), and so are expected to have ∼130 telomeres in G1cells and ∼260 in G2 cells. The number of TALE-labeled fociobserved was typically less than 50 per nucleus, indicating eitherthat not all telomeres were being detected or that many labeledsites are out of the focal plane. When serial optical sections wereobtained by confocal microscopy (Fig. S1), the total number oflabeled foci throughout the nucleus was between 50 and 70. It isof interest that in a previous study in which telomeres in liveU2OS cells were labeled with a fluorescent peptide nucleic acid(PNA) probe (7) the number of interphase foci was also less than70, indicating that a similar sized subpopulation of telomeres isavailable to the PNA and TALE probes, notwithstanding howchemically distinct their modes of DNA binding are. It is alsonoteworthy that there was considerable variation in the size/intensity of the TALE-labeled foci, suggesting either that certaintelomeres are very heavily labeled or, alternatively, that sometelomeres are clustered. To conduct time-lapse imaging wegenerated a stable cell line expressing the TelR20-mCherryand tracked dynamic movements of the foci during cell cycle

Significance

Repetitive DNA sequences such as telomeres and centromeresare, like the chromosomes in which they reside, highly dynamicwithin the interphase nucleus, moving about by diffusion. Livecell imaging of these specific chromosomal sites has been lim-ited and the approach presented in this study, based onexpressed transcription activator-like effector (TALE)-fluores-cent proteins, offers new opportunities in basic research onchromosome dynamics. In parallel, the described use of fluo-rescent TALEs as probes with fixed cells presents advantagesover fluorescent in situ hybridization and may find particularapplications in clinical and diagnostic settings.

Author contributions: H.M. and T.P. designed research; H.M. and P.R.-G. performed re-search; H.M. and P.R.-G. contributed new reagents/analytic tools; H.M. and P.R.-G. ana-lyzed data; and H.M. and T.P. wrote the paper.

Conflict of interest statement: H.M. is an inventor on a US patent application filed by theUniversity of Massachusetts.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1To whom correspondence may be addressed. E-mail: hanhui.ma@umassmed.edu orthoru.pederson@umassmed.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1319097110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1319097110 PNAS Early Edition | 1 of 6

CELL

BIOLO

GY

progression (Movie S1 and Fig. S2). Although we have not madea detailed analysis of these movements, their kinetics and spatialparameters were very similar to those previously reported instudies in which telomeres were labeled by other methods inU2OS cells (7, 8) or a human bladder carcinoma cell line (9). Toexamine the specificity of TALEs binding to telomeres, wedesigned various lengths of TALEs (TelR06, TelR09, TelR12,TelR15, and TelR20). As shown in Fig. S3, all of the probes withvarious lengths in our TALE-based strategy presented heretermed “TALEColor” showed similar patterns in interphase nu-clei except TelR06. TALE-TelR06 showed some specific foci butalso a high background throughout the nucleus, suggesting TALEswith six monomers lost the telomere specificity to some degree.Encouraged by these telomere labeling results, we extended

our TALE-based method to detect other tandemly repetitiveDNA sequences, namely the satellite DNA sequences lying at oradjacent to centromeres. Human centromeric DNA consists ofalpha satellites, a tandem repeat family that is commonly studiedin a chromosome-specific manner (10). We designed a plasmidencoding a TALE to recognize a motif specific to the alphasatellite consensus sequence (11, 12), providing a pan-centromereprobe (Pan-Cen) to study concurrently with the telomere-specificTALE in U2OS cells. This resulted in a pattern of discrete nuclearfoci labeled with the Pan-Cen probe that was nonoverlapping withthe pattern of telomere foci (Fig. 2, Top row). This is compatiblewith the well-established finding that telomeres and centromeresare neither coincident nor polarized in most higher eukaryoticcells. We then designed TALEs specific for higher order alphasatellite repeats that are unique to either chromosome 18 (13)or 15 (14) and expressed them in U2OS cells. As shown in Fig.2, Middle and Bottom rows, each of these TALEs labeled a setof discrete foci: five with the Cen18 and six with the Cen15probes, respectively, consistent with the karyotype of U2OS cells,namely, trisomy of chromosomes 18 and 15 (6).Human chromosome 15 is one of five autosomes that carry

a tandem array of repeated genes for ribosomal RNA in theiracrocentric arms (5). We were therefore intrigued to note thejuxtaposition of the centromere 15 and telomere signals nearnucleoli in a number of cases (e.g., Fig. 2, Bottom row, Far Rightpanel). This observation triangulates the centromere, the rDNAarray (nucleolus), and the adjacent telomere in a spatial con-figuration compatible with the close distances among these threesites on chromosome 15. This suggests that the TALE-basedmethod is accurately reading interphase genomic space.Given these results, we wondered if TALE–FPs might also be

used as probes to detect telomeres in fixed cells. We reasonedthat such a method would not require DNA denaturation orpossibly other preconditioning or annealing steps needed in

conventional FISH and thus might offer a shorter turnaroundtime. We constructed plasmids for coupled in vitro transcrip-tion–translation of telomere-specific TALEs fused in-frame withvarious fluorescent proteins, e.g., TALEGreen-TelR15 (Fig. 3A).When this was used as a probe with fixed U2OS cells, ∼50–70discrete fluorescent foci were observed in interphase and alsomitosis (Fig. 3B). To confirm that these signals represent bindingof the TALE to telomeres, we carried out immunostaining forthe telomere-specific protein TRF2 (15), which revealed coloc-alization with the TALE signals in both interphase (Fig. 3C) andmitotic cells (Fig. S4). As with the live cell experiments, thenumber of labeled foci in fixed cells was considerably less thanthe karyotypically predicted number of telomeres. Serial opticalsections acquired by confocal microscopy of the fixed cells (Fig.S5) again revealed from 50 to 70 labeled foci, indicating thatmethanol fixation did not appreciably change the number oftelomeres available to the TALE–FP probe. We also investigatedthe fixed cell method with HeLa cells, which have a similar de-gree of aneuploidy as U2OS cells, and again observed numerousdiscrete fluorescent foci in interphase and also mitosis (Fig. S6).To determine how wide an array of FPs might be applicable to

this method, we constructed a number of additional telomere–TALE–FP plasmids for coupled in vitro transcription–translation,as well as a telomere-specific TALE lacking a fused fluorescentprotein but into which a fluorescent amino acid (green lysine)was incorporated during the translation step. As shown in Fig. 4,the entire spectrum of fluorescent proteins tested (Fig. 4A), aswell as the green lysine-labeled TALE (Fig. 4B), resulted incomparable signals with the same spatial patterns as establishedin the live cell experiments. This indicates that, in this fixed cellversion of the method, none of the fluorescent proteins, linkedat a TALE’s C terminus, interferes with the TALE’s DNA se-quence recognition nor is the fluorescence intensity problemat-ically attenuated by intramolecular folding interactions back intothe TALE. Moreover, the fact that the telomere TALE with aninternally incorporated fluorescent amino acid (Fig. 4B) alsogave the same pattern and with strong signal intensity demon-strates that chemical modification within the TALE polypeptidecan be accommodated.

Fig. 1. Illustration of telomere detection by TALEColor. (A) TALEYellow andTALERed probes were designed to target either strand of the telomere re-peat by fusion of Venus or mCherry at C terminus. (B) U2OS cells werecotransfected with TALEYellow-TelL20 (Center Left) and TALERed-TelR20(Center Right) and labeling was assessed in the live cells 24 h later. The FarLeft panel is the phase-contrast images and the Far Right panel is the two-color overlays, respectively. (Scale bar, 5 μm.)

Fig. 2. Live cell imaging of centromeres and telomeres by TALEColor. U2OScells were cotransfected for 24 h with TALEmCherry-TelR20 to label telo-meres together with one of three TALEs designed to recognize centromericrepeats. (Top row) TALEVenus-PanCen, a TALE predicted to bind all humancentromeres. (Middle row) TALEVenus-Cen18, specific for an α-satellitehigher order repeat on chromosome 18 (D18Z1). (Bottom row) TALEVenus-Cen15, a specific α-satellite higher order repeat on chromosome 15 (D15Z3).Overlay images are shown in the Far Right column. (Scale bar, 5 μm.)

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1319097110 Ma et al.

Although live cell labeling of telomeres (Fig. 1) offers uniqueopportunities in basic cell biology and chromosome research,which we are pursuing, we wanted to explore the fixed cellmethod further. In particular, we asked how this TALE-based,fixed cell telomere detection method could be applied to humancell lines with differing telomere lengths, both to assess the in-terphase patterns of telomeres in these various cell lines and toalso get an initial impression of whether or not the intensity ofeach focal fluorescent signal might be related to the telomerelength. As mentioned earlier, we had so far no sense of whetherwe were labeling telomeres at some low, statistical level or somehigher degree of labeling across the telomeric repeat possiblyapproaching or even reaching target saturation by the TALE.Fig. 5 shows the results of applying a TALE–telomere probe tofixed human cells having different average telomere lengths.U2OS cells have a wide array of telomere lengths, from <3 kbto >50 kb due in part to the operation of the alternative length-ening of telomeres (ALT) pathway (8). The HeLa cell line 1.3 hasaverage telomere length ∼23 kb (16). In contrast, weaker signalswere observed in three other human cell lines known to haveshorter telomeres: HeLa S3 (telomere length 2–10 kb) (17), In-stitute for Medical Research 90 (IMR90) (average length ∼7.5 kb)(18) and retina pigmented epithelium 1 (RPE1) (∼2–12 kb) (19),suggesting that under the constant probe conditions used in thesefixed cell experiments, the signals obtained correlate with averagetelomere length to at least some degree.To further explore the relationship between the TALE probe

signal intensity and the length of telomeres, we set up a proof-of-principle experiment. The 1.3 and S3 HeLa cell lines were usedas samples with longer vs. shorter telomeres (average length ∼23

kb and 2–10 kb, respectively). The two cell lines were coculturedon coverglasses and subjected to TALE labeling (Fig. 6A). Thetelomere signals in HeLa 1.3 were much brighter as can be seenin separate or cocultured cells (Fig. 6A). Imaging flow cytometrywas then used to analyze telomere length by TALE labeling ofsuspension cultures of the two cell lines. DAPI (a blue fluores-cent DNA dye) and DRAQ5 (a far-red fluorescent DNA dye)were used to stain the DNA of HeLa 1.3 and S3, respectively, thecells were then mixed and TALE labeled with TALEGreen-TelR15, followed by FACS with the instrument’s parallel singlecell imaging capability (Materials and Methods). As can be seen inFig. 6B, Top row, Far Left, the two cell populations were clearlyresolved on the basis of their two DNA labels, as expected, witheach population displaying a typical cell cycle distribution in-cluding G1, S, and G2/M phases by DNA contents (Fig. 6B,Center Left in Top row and Center Left in Middle row). TheTALEGreen-TelR15 signals were separated into three pop-ulations (Fig. 6B, Bottom row, Far Left). As can be seen in theoverlay plots in the Center Left in the Bottom row of Fig. 6B, twopopulations having high and moderate telomere labeling wereDAPI positive (HeLa 1.3), whereas a third population havinglow telomere labeling was DRAQ5 positive (HeLa S3 cells),compatible with the known telomere lengths of these two celllines and consistent with the imaging from the coverglass cul-tures (Fig. 6A).We next analyzed the various telomere labeling populations in

each of the two cell lines with respect to the cell cycle (Fig. 6B).We gated the DAPI-positive cells as “R1” (high telomere labeling,shown in light green) and “R2” (moderate telomere labeling,shown as dark green). Meanwhile we gated the DRAR5-positivepopulation as “R3” (low telomere labeling, shown as teal). Theindividual or overlay plots of R1, R2, and R3 are shown in theCenter Right column of Fig. 6B. It can be seen that the high telo-mere labeling population was typified by a greater proportion ofS/G2/M phase cells, whereas the moderate telomere labeledpopulation was enriched in G1 phase cells (Top row, Far Right).The imaging flow cytometer allows us to image each single cell

represented in the above plots. Representative images of cells inthe R1, R2, and R3 populations (Fig. 6B) are shown in Fig. 6C.As can be seen in the Center Left four columns, the R1 pop-ulation cells were DAPI positive (purple, representing HeLa 1.3cells) and displayed high TALEGreen-TelR15 signals (green).The single-cell imaging revealed that the majority of this pop-ulation consisted of mitotic cells. The R2 cell population (CenterRight four columns) were also DAPI positive (and thus wereHeLa 1.3) and had a moderate telomere labeling. This pop-ulation was mostly G1 cells. Meanwhile the R3 cell population(Far Right four columns in Fig. 6C), defined as DRAQ5 positive(red, thus representing HeLa S3 cells) displayed low telomerelabeling and consisted of all cell cycle stages.

DiscussionThe methods we have developed and reported here seize uponthe extraordinarily specific nucleotide sequence recognition ca-pacity of TALEs and exploit, in particular, their unique affinityfor targets in their DNA double-stranded form. We reasonedthat given such sequence specificity and their preferential rec-ognition of targeted sequences in native DNA, the attachment ofa fluorescent protein to a given TALE would produce strongsignals if the targeted sequence were tandemly repeated in thegenome, and if the tethered fluorescent protein did not interferewith DNA sequence recognition in the TALE backbone. Thesehypotheses were borne out in the reported study. We were ableto label the human telomeric repeats, the centromere sequencecommon to all chromosomes, and two centromere repeats spe-cific to two chromosomes. In principle, it should be possible toextend this method to any other tandemly repeated DNA se-quence element in a genome, such as the genes for ribosomalRNA (known to undergo expansion or attrition) or ones impli-cated in human diseases before and after genomic expansion.For example, we are now applying these methods to trinucleotide

Fig. 3. TALE–FPs label telomeres in fixed cells. (A) Diagram of TALEGreen-TelL15. (B) U2OS cells were fixed in 90% methanol and incubated with theprobe. Shown are representative images in an interphase and anaphase cell.(C) After exposing fixed cells to the TALEGreen telomere probe, immunos-taining was carried out with a TRF-2 antibody followed by a TRITC-labeledsecondary antibody. (Top row) Probe imaged in both the green and redchannels. (Middle row) TRF2 immunostaining imaged in both channels.(Bottom row) Probe and TRF2 immunostaining imaged in each channel. TheFar Left column shows phase-contrast images and the Far Right columnshows images in which both the probe and TRF2 merged onto DAPI staining.(Scale bars in A–C, 5 μm.)

Ma et al. PNAS Early Edition | 3 of 6

CELL

BIOLO

GY

repeat expansion diseases (20). The ability of this method to labelspecific human chromosomes also offers unique opportunities todetect aberrant chromosomes, and we are currently labeling andtracking the intranuclear dynamics of all three copies of chro-mosome 21 in human trisomy 21 patient cells (21) in relation tothe territories they explore in these live cell studies.Given the extreme sequence specificity of TALEs (22), one

might ask how far this method can be pushed. Obviously this willdepend on the brightness of the fluor attached to a TALE andthe genomic prevalence of the targeted DNA sequences (downto possibly single-copy genes). We can recall that the first dem-onstration of in situ nucleic acid hybridization involved repeatedDNA sequences (the ribosomal RNA genes) (23) and that themethod’s refinement to detect single-copy DNA sequences tooksome years. Another point to be emphasized is that the TALE-based method reported here docks a protein (the TALE) and itsattached fluorescent protein onto a DNA sequence, so this iscertainly a “cargo” as regards the live cell application of thismethod and this point must be borne in mind when interpretingthe telomere and centromere dynamics observed.Three previous studies have tracked telomeres in live human

cells. In one, a lactose operator array was inserted into the telo-meric repeat and was detected by expressing a GFP lactose re-pressor (8). In contrast, the method reported here does notinvolve a disruption of the telomeric repeat. A second study useda telomere-specific PNA probe (7), whereas another was basedon GFP-tagged TRF1 or TRF2 (9). The live cell version of thepresently reported methods, as applied to telomeres, is notclaimed to be superior to these previous ones except for theavoidance of sequence interruption in the lac method. Thepresent method does obviate the purchase of PNAs or the needfor cloning to insert the lac repressor repeats.The fixed cell variation of our TALE-based method has

a number of key virtues. The preparation of fluorescent TALEsby coupled in vitro transcription–translation (Materials andMethods) is very time efficient compared with the synthesis orcommercial procurement of fluorescent oligonucleotide probesfor conventional FISH. Even more important is the very fast

timescale of the TALE-based protocol. Starting with a coverglassculture, the steps of methanol fixation, rinse, probe incubation,and rinse takes less than 1 h compared with many hours in typicalFISH methods. As H.M. anticipated when first envisioning thismethod, the ability of TALEs to recognize targeted sequences indouble-stranded DNA obviates the need for a DNA denaturationstep and we have also found that nonbound TALEs require onlya single, rapid wash for removal.Our initial studies with human cells with differing telomere

lengths have hinted that the TALE-based signals may be reportingon average telomere length. Without a direct determination ofhow many TALEs bind along the telomeric repeat in any of thehuman cells we studied, we cannot presently assert that themethod is reporting on telomere length in a truly quantitative way(i.e., with a linear relationship between telomere length and signalintensity over a wide range). This notwithstanding, it is clearthat the signal intensities do correlate with the average telomerelengths of the cell lines. This result suggests that this method,with refinement, could have clinical applications in diagnosticsituations where the average telomere length of a cell biopsy isrelevant. A precise analysis of telomere length involves a mo-lecular biology assay (24) but many hospital laboratories do nothave the capability to carry out such techniques. In contrast,many clinical laboratories do have personnel familiar with in situnucleic acid hybridization and our method is really a fore-shortened version of this. It seems plausible to envision that theTALE-based method reported here could be applied to a biopsyand reported back to the operating room within minutes, as-suming that such a preliminary assessment of telomere length inthese excised cells would have value in the subsequent surgery orpatient treatment. Further refinements of the methods reportedhere may advance both basic research in human genomics andclinical applications.Just after this manuscript was submitted, a study appeared in

which TALE–FPs were used to label repeated sequences in bothcultured mouse cells and embryos (25). The applications pre-sented in that important study and the present one are differentbut complementary.

Materials and MethodsConstruction of TALEColor Plasmids. TALEs for TALEColor were assembledusing the TAL effector toolbox (26) obtained from Addgene. The destinationvector for mammalian cell expression was derived from pcDNA4-TO-hygromycin (27) and contains a FLAG tag, the simian virus 40 nuclear lo-calization signal (NLS), and a truncated wild-type TALE backbone from thetoolbox. For the specific telomere and centromere DNA probes used in thisinvestigation, tandem repeats of 34-amino acid (aa) TALE monomers targeting

Fig. 4. Spectral variants of TALEColor probes. (A) TALE-TelR15 probes weredesigned with various fused fluorescent proteins as indicated and applied tofixed U2OS cells. Images were captured in the appropriate channels (Lowerrow). (B) TALE-TelR15 probe with no fused fluorescent protein was producedcarrying internal lysine resides labeled with a green dye (Materials andMethods). The labeling obtained (Upper row) was imaged and comparedwith that with the same TALE carrying fused mCherry (Lower row), with theFar Right column representing the respective images overlaid onto DAPIimages. (Scale bar in A and B, 5 μm.)

Fig. 5. Telomeres compared by TALEColor in variety of human cell lines.U2OS, HeLa 1.3, HeLa S3, IMR90, and RPE1 cells were fixed and incubatedwith TALEGreen-TelR15 (Middle row). All of the images of TALEGreen-TelR15 (Middle row) are scaled to the same. The phase images are shown inTop row and images merged with DAPI are shown in the Bottom row. (Scalebar, 5 μm.)

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1319097110 Ma et al.

6–20 bp in the case of telomeric repeats and 20 bp in the case of the cen-tromeric repeats were inserted into the destination vector to generatepTH-TelL20-mVenus and pTH-TelR20-mCherry for telomeres and pTH-PanCen-mVenus, pTH-Cen18-mVenus, and pTH-Cen15-mVenus for cen-tromeres. To produce TALEColors by in vitro coupled transcription–translation for the fixed cell application the 1-Step Human Coupled InVitro Translation kit (Pierce) was used. TelR15 coding sequences weresubcloned from the mammalian expression plasmid into in vitro trans-lation plasmid pT7CFE1-His and generated pT7CFE1-TelR15-mTagBFP2,pT7CFE1-TelR15-mTFP1, pT7CFE1-TelR15-sfGFP, pT7CFE1-TelR15-YPet, andpT7CFE1-TelR15-mCherry. To produce TelR15 with green lysine in-corporation, TelR15 or TelR15-mCherry were subcloned into the bacterial ex-pression plasmid pET30a to generate pET30a-TelR15 and pET30a-TelR15-mCherry, and these plasmids were then used as DNA templates for coupledtranscription–translation in the TnT T7 Quick Coupled kit (Promega) in the pres-ence of green lysine (Promega). TALEColor plasmids are available at Addgene.

Telomere and Centromere Target Sequences of TALEColors. TALEs were de-signed to target the human telomere repeat (TTAGGG) regions on eitherstrand. The forward telomere target sequence (TelL) was the 20-merTAGGGTTAGGGTTAGGGTTA. The reverse telomere target sequences (TelR)were the 20-mer TAACCCTAACCCTAACCCTA, the 15-mer TAACCCTAACCC-TAA, the 12-mer TAACCCTAACCC, the 9-mer TAACCCTAA, and the 6-merTAACCC. The pan-centromere target sequence, the chromosome 18-specificcentromere target sequence, and the chromosome 15-specific centromeretarget sequence were TAGACAGAAGCATTCTCAGA, TTGAACCACCGTTTT-GAAGG, and TCACTTCAAGATTCTACGGA, respectively.

Cell Culture and Transfection of TALEColors. The U2OS, HeLa 1.3 (11), HeLa S3,and IMR90 cells were cultured at 37 °C in Dulbecco-modified Eagle’s Mini-mum Essential Medium (DMEM; Life Technologies) supplemented with 10%(vol/vol) FBS. RPE1 cells (28) were cultured at 37 °C in DMEM:F12 mediumsupplemented with 10% (vol/vol) FBS. For live imaging, cells were grown onLab-Tek two-well coverglasses in Hepes-buffered DMEM containing 10%(vol/vol) FBS, penicillin (100 units/mL), and streptomycin (100 μg/mL) andthen overlaid with mineral oil. A total of 50 ng of TALEColor plasmids weretransfected using Lipofectamine 2000 (Life Technologies) and the cells wereincubated for another 24 h. The microscope stage incubation chamber wasmaintained at 37 °C as described previously (29). Phase-contrast and fluores-cence microscopy were performed with a Leica DM-IRB inverted microscopeequipped with a mercury arc lamp, a 10-position filter wheel (Sutter In-strument), CFP/YFP/HcRed filter set, GFP/DsRed filter set (Semrock), a CCD cam-era (Photometrics), and MetaMorph acquisition software (Molecular Devices).

DNA Labeling by TALEColors in Fixed Cells. Cells grown on coverslips were fixedin ice-cold methanol for 10 min at −20 °C. All subsequent steps were carriedout at room temperature. The fixed cells were incubated with 2N HCl for5 min and then washed twice with PBS for 5 min each and then incubatedwith a given TALEColor probe as a 1:10 dilution from the coupled in vitrotranscription–translation reaction mixtures for 30 min. The cells were thenwashed once with PBS for 5 min. Coverslips were mounted in ProlongAntifade (Molecular Probes), and images were captured with the fluores-cence microscopy system described above.

Dual Labeling of Telomeres by TRF2 Immunofluorescence and TALEColors. Cellsgrown on coverslips were fixed in ice-cold methanol for 10 min at −20 °C. Allsubsequent steps were carried out at room temperature. Coverslips wereincubated with TRF2 monoclonal antibodies (1:200 dilution; Millipore) inPBS-1% BSA for 1 h and followed by incubation together with TRITC-con-jugated goat anti-mouse secondary antibody (Sigma) and in vitro translatedTALEs: TelR15-sfGFP.

TALEColor Quantification and Single-Cell Imaging by Flow Cytometry. Afterlabeling fixed cells with a given TALEColor probe as described above, theywere trypsinized and centrifuged at 200 × g for 5 min and then washed

Fig. 6. Imaging flow cytometry assessment of average telomere length andintracell population heterogeneity. (A) HeLa 1.3 and HeLa S3 cells culturedeither alone or together and then incubated with TALEGreen-TelR15 andimaged. (Scale bar, 10 μm.) (B) Separate coverglass cultures of HeLa 1.3 andHeLa S3 cells were trypsinized, fixed, and incubated with TALEGreen-TelR15together with DNA staining with DAPI or DRAQ5 for the HeLa 1.3 and S3cells, respectively. The two cell populations were then mixed and imagingflow cytometry was carried out immediately. Single cells were gated by anaspect ratio program in the instrument’s software (Middle row, Far Left).DAPI positive cells (purple) and DRAQ5 positive cells (red) were gated bytheir intensity (Top row, Far Left) and their intensity plots are shown in theindicated panels. The DNA intensity plots of the two cell populations (re-solved out from the mixture of the two cell lines) are shown in the indicatedpanels. The scatter plot of TALEGreen-TelR15 signals in all cells is shown(Bottom row, Far Left). These were sorted into DAPI positive (purple) andDRAQ5 positive (red) populations (Bottom row, Middle). The DAPI positivecells were then sorted into distinct levels of telomere labeling: a high level

(R1, light green, Top row, Far Right) and a moderate level (R2, dark green,Top row, Far Right). DRAQ5 positive cells with their low level of telomerelabeling were sorted in parallel (R3, teal, Middle row, Far Right). (C) Rep-resentative DAPI images for HeLa 1.3 cells not labeled with TALEGreen-TelR15 (Far Left three columns), DAPI positive R1 cells (Center Left fourcolumns), DAPI positive R2 cells (Center Right four columns), DRAQ5 positiveR3 cells (Far Right four columns). BF, brightfield.

Ma et al. PNAS Early Edition | 5 of 6

CELL

BIOLO

GY

once with PBS. The cell concentration was adjusted to 1 × 107/mL in PBSand ice-cold methanol was then added to a final concentration of 90%(vol/vol) with gentle mixing. A total of 106 cells were resuspended in 100 μLof 2N HCl and incubated at 5 min at ambient temperature, then washedthree times with 100 μL PBS (300 × g for 2 min). The cells were resuspendedand DNA was labeled by adding 100 μL PBS containing 1 μg/mL of DAPI orDRAQ5 for 10 min and then washed twice with 100 μL PBS. Imaging flowcytometry was performed in the University of Massachusetts MedicalSchool FACS Core Facility with an Amnis FlowSight imaging cytometer(Amnis). GFP was excited at 488 nm and its emission was collected in a 505-to 560-nm channel; DAPI was excited at 405 nm and its emission collectedusing a 430- to 505-nm filter. DRAQ5 was excited at 642 nm and its emission

collected using a 642- to 740-nm filter. Flow cytometry and quantitativeimaging data were acquired and analyzed by INSPIRE and IDEAS software(Amnis), respectively.

ACKNOWLEDGMENTS.We thank Dr. Karen H. Miga (University of California,Santa Cruz) for generously making available centromeric DNA sequenceinformation on chromosomes 15 and 18 prior to publication, as well asconstructive comments on the manuscript, and Susanne Pechhold and PaulFurcinitti, respectively, in our institution’s Flow Cytometry and Digital LightMicroscopy Cores, for their assistance with the imaging flow cytometry andconfocal microscopy. We also thank Titia de Lange (Rockefeller University)for providing the HeLa 1.3 cell line. This investigation was supported byGrant MCB-1051398 (to T.P.) from the National Science Foundation.

1. Boch J, et al. (2009) Breaking the code of DNA binding specificity of TAL-type III ef-fectors. Science 326(5959):1509–1512.

2. Baker M (2012) Gene-editing nucleases. Nat Methods 9(1):23–26.3. Bogdanove AJ, Voytas DF (2011) TAL effectors: Customizable proteins for DNA tar-

geting. Science 333(6051):1843–1846.4. Politz JC, Tuft RA, Pederson T (2003) Diffusion-based transport of nascent ribosomes

in the nucleus. Mol Biol Cell 14(12):4805–4812.5. Henderson AS, Warburton D, Atwood KC (1972) Location of ribosomal DNA in the

human chromosome complement. Proc Natl Acad Sci USA 69(11):3394–3398.6. Janssen A, Medema RH (2013) Genetic instability: Tipping the balance. Oncogene

32(38):4459–4470.7. Molenaar C, et al. (2003) Visualizing telomere dynamics in living mammalian cells

using PNA probes. EMBO J 22(24):6631–6641.8. Jegou T, et al. (2009) Dynamics of telomeres and promyelocytic leukemia nuclear

bodies in a telomerase-negative human cell line. Mol Biol Cell 20(7):2070–2082.9. Wang X, et al. (2008) Rapid telomere motions in live human cells analyzed by highly

time-resolved microscopy. Epigenetics Chromatin 1(1):4.10. Willard HF, Waye JS (1987) Hierarchical order in chromosome-specific human alpha

satellite DNA. Trends Genet 3(7):192–198.11. Waye JS, Willard HF (1987) Nucleotide sequence heterogeneity of alpha satellite re-

petitive DNA: A survey of alphoid sequences from different human chromosomes.Nucleic Acids Res 15(18):7549–7569.

12. Vissel B, Choo KH (1987) Human alpha satellite DNA—consensus sequence and con-served regions. Nucleic Acids Res 15(16):6751–6752.

13. Alexandrov IA, et al. (1991) Chromosome-specific alpha satellites: Two distinct fami-lies on human chromosome 18. Genomics 11(1):15–23.

14. Choo KH, Earle E, Vissel B, Filby RG (1990) Identification of two distinct subfamiliesof alpha satellite DNA that are highly specific for human chromosome 15. Genomics7(2):143–151.

15. Broccoli D, Smogorzewska A, Chong L, de Lange T (1997) Human telomeres containtwo distinct Myb-related proteins, TRF1 and TRF2. Nat Genet 17(2):231–235.

16. Takai KK, Hooper S, Blackwood S, Gandhi R, de Lange T (2010) In vivo stoichiometryof shelterin components. J Biol Chem 285(2):1457–1467.

17. Bryan TM, Englezou A, Dunham MA, Reddel RR (1998) Telomere length dynamics intelomerase-positive immortal human cell populations. Exp Cell Res 239(2):370–378.

18. Ouellette MM, Aisner DL, Savre-Train I, Wright WE, Shay JW (1999) Telomerase ac-tivity does not always imply telomere maintenance. Biochem Biophys Res Commun254(3):795–803.

19. Bodnar AG, et al. (1998) Extension of life-span by introduction of telomerase intonormal human cells. Science 279(5349):349–352.

20. Mirkin SM (2007) Expandable DNA repeats and human disease. Nature 447(7147):932–940.

21. Antonarakis SE, Lyle R, Dermitzakis ET, Reymond A, Deutsch S (2004) Chromosome 21and Down syndrome: From genomics to pathophysiology. Nat Rev Genet 5(10):725–738.

22. Meckler JF, et al. (2013) Quantitative analysis of TALE-DNA interactions suggestspolarity effects. Nucleic Acids Res 41(7):4118–4128.

23. Gall JG, Pardue ML (1969) Formation and detection of RNA-DNA hybrid molecules incytological preparations. Proc Natl Acad Sci USA 63(2):378–383.

24. Kim NW, et al. (1994) Specific association of human telomerase activity with immortalcells and cancer. Science 266(5193):2011–2015.

25. Sanjana NE, et al. (2012) A transcription activator-like effector toolbox for genomeengineering. Nat Protoc 7(1):171–192.

26. Ma H, et al. (2012) A highly efficient multifunctional tandem affinity purificationapproach applicable to diverse organisms. Mol Cell Proteomics 11(8):501–511.

27. Uetake Y, et al. (2007) Cell cycle progression and de novo centriole assembly aftercentrosomal removal in untransformed human cells. J Cell Biol 176(2):173–182.

28. Jacobson MR, Pederson T (1997) RNA traffic and localization reported by fluorescencecytochemistry. Analysis of mRNA Formation and Function, ed Richter JD (Academic,New York), pp 341–359.

29. Miyanari Y, Ziegler-Birling C, Torres-Padilla ME (2013) Live visualization of chromatindynamics with fluorescent TALEs. Nat Struct Mol Biol 20(11):1321–1324.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1319097110 Ma et al.