Single-Chromosome Transcriptional Profiling Reveals Chromosomal Gene
Total Page：16
File Type：pdf, Size：1020Kb
BRIEF COMMUNICATIONS with 16 short, fluorescently labeled oligonucleotides4 (Fig. 1a Single-chromosome and Supplementary Fig. 1) enabled us to measure whether the gene was actively transcribing5 and, if it was, to identify the three- transcriptional profiling dimensional coordinates of the gene6–8. Control experiments con- firmed that the intron spot marked the site of transcriptionally reveals chromosomal gene active genes (Supplementary Figs. 2–7). Note that even genes considered constitutively active transcribe RNA in ‘bursts’ thought expression regulation to arise from random aspects of the transcriptional process9–12. The overall transcription rate is proportional to the probability of finding such a spot for each gene (Supplementary Discussion). Marshall J Levesque & Arjun Raj To concomitantly measure overall chromosome structure, we designed probes targeting the introns of 20 genes along chromo- We report intron chromosomal expression FISH (iceFISH), some 19 (Supplementary Fig. 1 and Supplementary Table 1). This a multiplex imaging method for measuring gene expression and yielded an average resolution of 3 megabases (Mb) with a minimum chromosome structure simultaneously on single chromosomes. of 360 kilobases (kb), but we were also able to distinguish loci sepa- We find substantial differences in transcriptional frequency rated by just 30 kb (Supplementary Fig. 6). To measure all 20 genes’ between genes on a translocated chromosome and the same transcriptional statuses simultaneously, we labeled each gene’s genes in their normal chromosomal context in the same cell. introns with a particular ‘pseudocolor’, which is a distinct code for Correlations between genes on a single chromosome pointed each gene consisting of either two or three (out of five) spectrally toward a cis chromosome-level transcriptional interaction distinguishable fluorophores13,14 (Fig. 1a). To assign gene identity, spanning 14.3 megabases. we looked for colocalization of two or three spots in each fluores- cence channel (Supplementary Fig. 8). In human foreskin fibro The transcription of a gene’s DNA into RNA is thought to be con- blasts, we could discern two clearly separated chromosomes (Fig. 1b) trolled largely by the interaction of regulatory proteins with DNA 78% of the time (Supplementary Fig. 9). On average, we found sequences proximal to the gene itself. At the same time, genes are 6 ± 2 expressing genes (out of the 20 labeled) per chromosome. organized by the thousands into chromosomes, thus raising the The positions of these genes were more spread out than expected15 Nature America, Inc. All rights reserved. Inc. Nature America, 3 possibility that the structure or organization of chromosomes influ- (Supplementary Fig. 6 and Supplementary Discussion). We found ences transcription1,2; however, little is known about how organi- that using 16 singly colored probes did not change the spot detec- © 201 zation at the chromosome-length scale affects gene expression. tion efficiency (Online Methods), nor did pseudocoloring incur Here we begin to address this question 0 Mb 3,4 Export and with a method based on RNA FISH a b Chr 19 PTBP1 Spliced mRNA Translation called iceFISH that enabled us to generate EEF2 Chr 19 transcriptional profiles of 20 genes simul- Exon Exon DNMT1 Exon Intron Intron ILF3 taneously along individual copies of 5 base colors DHPS human chromosome 19 in single cells. RPL18A UBA52 We took advantage of the fact that introns Introns degraded before typically degrade rapidly after being spliced leaving site of transcription ZNF91 2 colors 3 colors out of nascent RNA. Labeling the intron 20 pseudocolors Cytoplasm Nucleus UBA2 Figure 1 | Unique identification of 20 loci on Chromosome 19 Chromosome 19 EIF3K SUPT5H FBL chromosome (chr) 19 by RNA FISH targeting EGLN2 Nucleus RPS19 introns in human foreskin fibroblasts. TOMM40 MARK4 (a) Pseudocoloring scheme for labeling the site 4 20 m) SLC1A5 µ of transcription by targeting gene introns with ( 2 16 PPP2R1A z 0 12 a series of labeled oligonucleotide probes. x ( m) RPS9 16 µ ZNF444 (b) Transcriptional activity and location of the 12 8 Mature 8 4 20 genes computationally identified from images 4 mRNA y (µm) 0 0 59.13 Mb from each fluorescence channel. Scale bar, 5 µm. Department of Bioengineering, University of Pennsylvania, Philadelphia, Pennsylvania, USA. Correspondence should be addressed to M.J.L. or A.R. ([email protected]). RECEIVED 17 NOVEMBEr 2012; ACCEPTED 18 JAnUArY 2013; PUBLISHED OnLInE 17 FEBrUArY 2013; DOI:10.1038/nMETH.2372 NATURE METHODS | ADVANCE ONLINE PUBLICATION | BRIEF COMMUNICATIONS 0 Mb a significant rate of spot misidentification PTBP1 a HeLa karyotype b EEF2 (Supplementary Fig. 10). We ensured that t(13;19) the cells we analyzed were in the G0/G1 Chr 13 t(6;19) DNMT1 t(6;19) ILF3 q-arm Chr 19 stage of the cell cycle by colabeling cyclin DHPS RPL18A A2 mRNA, which is abundant during UBA52 16 the S, G2 and M phases of the cell cycle Chr 19 ZNF91 Chr 19 Fusion (Supplementary Fig. 11). Chr 6 chromosomes p-arm q-arm By grouping actively transcribing UBA2 Chr 19 EIF3K genes into territories corresponding to t(13;19) Chromosome 19 SUPT5H FBL each chromosome, we constructed tran- EGLN2 RPS19 TOMM40 scriptional profiles showing which of MARK4 Chr 19 SLC1A5 our 20 genes were on or off per chromo q-arm PPP2R1A Chr 6 RPS9 some. Researchers largely believe that ZNF444 p-arm gene transcription depends on both 59.13 Mb chromosome-extrinsic trans factors (such Chr 19 as transcription factors) and local cis fac- c Chr 13 Translocation Centromere t(13;19) 1.0 tors on the DNA (typically within 1 Mb *** t(6;19) n = 50 n = 30 *** of the gene itself). Our method enabled *** *** us to examine the possibility that nonlo- ** *** ** *** *** * ** cal mechanisms at the chromosome scale may also regulate transcription. 0 MZT1 DIAPH3 RPL18AUBA52 ZNF91 UBA2 EIF3K SUPT5H FBL EGLN2 RPS19 TOMM40MARK4 SLC1A5PPP2R1A RPS9 ZNF444 Translocations provide a means to search for such possibilities: although Chr 13 gene Chromosome 19 gene they disrupt the large-scale structure 1.0 of a chromosome, the cell’s trans envi- n = 30 t(6;19) ronment and local cis DNA regulatory | iceFISH spots per chromosome Figure 2 Translocated portions of chromosome (chr) 19 code remain unchanged for most genes display different expression patterns than intact on the translocated chromosome. For chromosomes. (a) Schematic showing chr 19 and its derivatives in our HeLa cells. (b) Computational example, HeLa cells contain two intact 0 PTBP1 EEF2 DNMT1 ILF3 DHPS copies of chromosome 19 and one copy identification of actively transcribing genes revealed the two intact copies and the two translocated pieces that is split into two pieces fused to parts Chromosome 19 gene 17 of chr 19. Scale bar, 5 µm. (c) Comparison of the of other chromosomes : one, denoted transcriptional activity of the genes on the translocated fragments (as measured by frequency of t(6;19), consists of the first 17–20 Mb of observing a transcription site per chr) to the activity of the intact copies of chr 19. We similarly Nature America, Inc. All rights reserved. Inc. Nature America, chromosome 19 fused to part of chromo- analyzed two genes on chr 13 (MZT1 and DIAPH3), as described in Supplementary Figure 15. P values 3 some 6; and the other, denoted t(13;19), for the difference in frequency (using a binomial distribution test) *P < 0.05, **P < 0.01, ***P < 0.001. consists of the remaining 40–43 Mb of © 201 chromosome 19 translocated onto a por- tion of chromosome 13 (Fig. 2a and Supplementary Fig. 12). appear to correlate with differences in spatial chromosome Our iceFISH data recapitulated these genetic rearrange- conformation (Supplementary Figs. 6, 16 and 17). ments (Fig. 2b). We found that most genes on t(13;19) were We next looked for evidence of interactions governing the up to fivefold more transcriptionally active than those on the transcription of genes within a single chromosome by examin- normal copies of chromosome 19 (Fig. 2c and Supplementary ing whether the transcriptional status of one gene in our panel Fig. 13), consistent with the existence of chromosome-specific affected the transcriptional status of another gene on the same transcriptional regulation that the translocation may have dis- chromosome. Such an interaction would manifest itself as a devia- rupted. Intron spot intensities were roughly the same on all the tion from independence, with positive correlations signifying that chromosomes we examined (Supplementary Fig. 14), suggest- the two genes ‘A’ and ‘B’ would have a likelihood greater than ing that transcriptional hyperactivation results from an increased chance to be actively transcribing at the same time on the same probability of a gene being active (Supplementary Discussion). chromosome, and with anticorrelations indicating that transcrip- We also found that the transcriptional frequencies of two genes tion of genes A and B would be mutually exclusive. from chromosome 13 (DIAPH3 and MZT1) were roughly twofold We found that most pairwise interactions on single chromo- higher on t(13;19) than on the normal copies of chromosome 13 somes did not show significant deviations from independence (Fig. 2c and Supplementary Fig. 15), suggesting that this trans- (Fig. 3 and Supplementary Figs. 18 and 19). However, one pair location resulted in hyperactivation of all genes on t(13;19) of genes, RPS19 and ZNF444 (separated by 14.3 Mb), showed an irrespective of location. Meanwhile, transcription of the chro- anticorrelation (R = −0.40 ± 0.08; P = 3.99 × 10−5, Fisher exact mosome 19 genes on t(6;19) was similar to that of the normal test). One explanation for this anticorrelation is fluctuations in a copies (Fig. 2c), suggesting that translocations do not necessar- potential trans-acting factor, such as a transcription factor, that ily lead to transcriptional changes. Reports of such effects18 are activated RPS19 and inactivated ZNF444 in some cells while acti- not widespread because of the averaging effects of most assays.