Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 12 (1), e1005794
eCollection

PAF Complex Plays Novel Subunit-Specific Roles in Alternative Cleavage and Polyadenylation

Affiliations

PAF Complex Plays Novel Subunit-Specific Roles in Alternative Cleavage and Polyadenylation

Yan Yang et al. PLoS Genet.

Erratum in

Abstract

The PAF complex (Paf1C) has been shown to regulate chromatin modifications, gene transcription, and RNA polymerase II (PolII) elongation. Here, we provide the first genome-wide profiles for the distribution of the entire complex in mammalian cells using chromatin immunoprecipitation and high throughput sequencing. We show that Paf1C is recruited not only to promoters and gene bodies, but also to regions downstream of cleavage/polyadenylation (pA) sites at 3' ends, a profile that sharply contrasted with the yeast complex. Remarkably, we identified novel, subunit-specific links between Paf1C and regulation of alternative cleavage and polyadenylation (APA) and upstream antisense transcription using RNAi coupled with deep sequencing of the 3' ends of transcripts. Moreover, we found that depletion of Paf1C subunits resulted in the accumulation of PolII over gene bodies, which coincided with APA. Depletion of specific Paf1C subunits led to global loss of histone H2B ubiquitylation, although there was little impact of Paf1C depletion on other histone modifications, including tri-methylation of histone H3 on lysines 4 and 36 (H3K4me3 and H3K36me3), previously associated with this complex. Our results provide surprising differences with yeast, while unifying observations that link Paf1C with PolII elongation and RNA processing, and indicate that Paf1C subunits could play roles in controlling transcript length through suppression of PolII accumulation at transcription start site (TSS)-proximal pA sites and regulating pA site choice in 3'UTRs.

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Fig 1
Fig 1. Paf1C subunits, except for Rtf1, are associated in mouse muscle cells but differentially impact histone modifications.
(A) Western blot analysis of expression of Paf1C subunits in myoblasts (MB) and myotubes (MT). Rtf1 and Ctr9 are detectable in the chromatin-bound fraction but not whole cell extracts, as indicated. (B) Co-immunoprecipitation of endogenous Paf1C subunits, except for Rtf1, in the chromatin-bound fraction. IgG, control antibody. (C) siRNA-mediated depletion of the indicated Paf1C subunits in myoblasts have distinct effects on the levels of histone marks. siCtrl, non-specific control siRNA. (D) qChIP analysis of the enrichment of the Paf1C at specific loci in myoblasts and myotubes. The number shown for each locus indicates the midway point within the amplicon in base-pairs from the TSS of a specific gene. Gene desert is a genomic region devoid of genes.
Fig 2
Fig 2. Genome-wide enrichment of Paf1C at both TSS and TES and subunit-specific localization on chromatin.
(A) The average ChIP-seq enrichment profiles of indicated Paf1C subunits show occupancy across the gene with substantial enrichment at the transcription start site (TSS) and transcript end sites (TES). Enrichment (y-axis) is presented as the log2 ratio of RPM between ChIP and input samples in each 50 bp bin. (B) Left, A heatmap representing RNA-seq analysis of myoblasts and myotubes. Right, k-means clustering (k = 10) of ChIP-seq data reveals common and distinct localization of different Paf1C subunits. Spt6, histone modification and PolII data are also shown for comparison. Each row represents a single 4 kb genic region surrounding the TSS. The RNA-seq heatmap and ChIP-seq data are registered according to genes. (C) Cluster analysis of co-occupancy of Paf1C subunits on chromatin regions. Values are Pearson correlation coefficients. Correlations of RPM values encompassing genomic regions (TSS/TES upstream/downstream 1kb) were analyzed. (D) Percentage of overlapping ChIP-seq peaks or target genes of Paf1C subunits. Peaks within 200 bp from one another were considered overlapping. (E) Representative genome browser tracks illustrating Paf1C ChIP-seq data. The y-axis represents normalized reads in reads per million mapped (RPM). MB, myoblasts; MT, myotubes. Gene name and the maximum value of y-axis are indicated.
Fig 3
Fig 3. Depletion of Paf1C subunits leads to genome-wide changes in pA site usage.
(A) Schematic depicting different types of APA sites and method to detect total and full-length transcript abundance. Solid boxes are exons, and open ones are introns. 3’UTR-APA sites are those located in the 3’UTR. CDS-APA sites are those located in introns or exons upstream of the 3’-most exon. (B) Left, normalized number of genes with regulated 3’UTR-APA in each sample as examined by Global Analysis of Alternative Polyadenylation (GAAP, see Materials and Methods for detail). Normalized number is based on (observed value—expected value). Red and blue bars represent genes with shortened 3’UTRs (Sh, proximal pA isoform up-regulated relative to distal pA) and lengthened 3’UTRs (Le, distal pA isoform up-regulated relative to proximal pA), respectively. FDR = 0.05 (Significance Analysis of Alternative Polyadenylation, SAAP, see Materials and Methods for detail) was used to select genes with significant 3’UTR-APA regulation. Only the top two most abundant APA isoforms (based on the number of PASS reads) of each gene were used for this analysis. Error bars show standard deviation based on 20 iterations of bootstrapping. Log2(Sh#/Le#) is log2(ratio) of the number of Sh genes to the number of Le genes. Right, genome browser tracks showing 3’READS data for a selected gene, Zfp260. The maximum value of y-axis for each track is indicated. APA sites are indicated by arrows. (C) Left, normalized number of genes with regulated CDS-APA as examined by GAAP. Red and blue bars represent genes with up-regulated CDS-APA isoforms (UP) and down-regulated isoforms (DN), respectively. All CDS-APA isoforms were combined and compared to all 3’-most exon isoforms combined by SAAP. FDR = 0.05 (SAAP) was used to select genes with significant CDS-APA regulation. Error bars are as described in panel B. Samples were sorted by the total number of genes with CDS-APA changes. Log2(UP#/DN#) is log2(ratio) of the number of UP genes to the number of DN genes. Right, genome browser tracks showing 3’READS data for a selected gene, Pcbp2. The maximum value of y-axis for each track is indicated. APA sites are indicated by arrows. (D) Left, normalized number of genes with regulated uaRNA expression. Red and blue bars represent genes with up-regulated (UP) and down-regulated (DN) uaRNA expression, respectively. Log2(UP#/DN#) is log2(ratio) of the number of UP genes to the number of DN genes. All uaRNAs were combined and compared to all sense strand transcripts whose pAs were beyond 2 kb from the TSS by SAAP. Q-value < 0.05 (SAAP) was used to select genes with a significant uaRNA expression difference. Error bars are as described in panels B and C. Right, metagene plots of uaRNA and sense strand RNA expression in siPaf1, siCdc73, siSki8 and siCtrl samples. Expression is represented by reads per million (RPM, pA site-supporting reads only) at pA positions. (E) RT-qPCR validation of impact on APA after depletion of Paf1, Cdc73, or Ski8. p-value indicates significant difference between Paf1- and Ski8-depleted cells. Negative control genes are indicated. (F) Heatmap showing common and distinct target genes with changes in CDS-APA after depletion of Paf1, Cdc73 or Ski8. Genes were sorted using the siPaf1 data. Samples were clustered using Pearson correlation. (G) Top, Venn diagrams depicting overlap of genes among Paf1C targets as detected by ChIP-seq, genes showing altered levels of expression in RNA-seq (fold-change >1.4, p-value<0.01 (Fisher's exact test)) and genes with changes in APA after depletion of Paf1C subunits (SAAP FDR = 0.05; see Materials and Methods). Right, data are shown in percentage of overlap.
Fig 4
Fig 4. Depletion of Paf1, Cdc73 but not Ski8 leads to prevalent activation of TSS-proximal intronic APA sites.
(A) Up-regulation of intronic APA sites is biased toward the TSS after depletion of Paf1 or Cdc73, but not Ski8. Regions between TSS and the 3'-most splice site were divided into 5 equally-sized bins. “siCtrl" indicates the background distribution of intronic pAs whose usage was detected in control cells. The distribution of up-regulated intronic pAs in knock-down samples were compared to background. The p-values (Fisher’s exact test) for the percentage of pAs in Bin1, the 5’-most bin, for each depletion and control pair are 4x10-28, 4x10-19, and 0.02 for siPaf1, siCdc73, and siSki8, respectively. (B) TSS-proximal intronic pAs are more strongly activated as compared to downstream intronic pAs. +1, +2, middle, -2 and -1 introns are the first, second, middle (not first two or last two), second to the last, and the last introns, respectively. The error bar represents standard error of mean for all analyzed pAs in a given intron group. Expression changes are log2(ratio) of PASS reads in knock-down sample vs. control sample. (C) Top, Venn diagrams depicting overlap of genes with regulated intronic pAs between Paf1- or Cdc73-depletion and U1 inhibition. Bottom, data are shown in percentage of overlap.
Fig 5
Fig 5. Chromatin occupancy of Paf1C correlates with APA regulation after depletion of Paf1C subunits.
(A) The average enrichment profiles of Paf1, Cdc73, and Ski8 were plotted on genes that showed regulated, low-expressed, or unaffected CDS-APA after depletion of Paf1, Cdc73 or Ski8. A region from 3 kb upstream of TSS to 3 kb downstream of TES was analyzed. (B) The average enrichment profiles of Paf1, Cdc73, and Ski8 were plotted over genes that showed shortening or unaffected length of 3’UTR after depletion of Paf1, Cdc73, or Ski8. A region from 2 kb up-or downstream of the two most regulated 3’UTR pA sites was analyzed (A distance between the proximal and distal pAs greater than 1 kb was required).
Fig 6
Fig 6. Paf1, but not Ski8, depletion leads to strong accumulation of PolII downstream of the TSS of genes with up-regulated intronic APA.
(A) The average enrichment profiles of PolII after depletion of Paf1 or Ski8. A region from 3 kb upstream of TSS to 3 kb downstream of TES of all RefSeq genes was analyzed. (B and C) qChIP analysis and genome browser tracks demonstrate increased PolII and CPSF100 occupancy, respectively, immediately downstream of the TSS of genes that showed CDS-APA up-regulation after Paf1 depletion. In (B), qChIP data obtained using two different anti-PolII antibodies were averaged. Genes with CDS-APA unaffected by Paf1 depletion are shown as controls. (D) The ratio of PolII ChIP-seq signals in Paf1 versus control siRNA-treated cells was plotted over genes that exhibited low or undetectable expression or that were unaffected or showed CDS-APA regulation after depletion of Paf1 or Ski8. A region from 3 kb upstream of TSS to 3 kb downstream of TES was analyzed. (E) Cumulative Distribution Function (CDF) curves indicate significant decrease in the levels of gene expression with up-regulated CDS-APA after subunit depletion. The p-values were calculated based on Wilcoxon Rank Sum Test by comparing the up- or down-regulated (red or blue line, respectively) to the unaffected CDS-APA events (gray line). Specifically, in Paf1 knock-down, p = 5x10-20 and p = 8x10-23, respectively; in Cdc73 knock-down, p = 4x10-35 and p = 1x10-4, respectively; and in Ski8 knock-down, p = 5x10-7 and p = 2x10-8, respectively. Median values were indicated as vertical dotted lines. The x-axis represents the log2 ratio of RPKM (reads per kilobase per million) values between knock-down and control samples. (F) The average enrichment profiles of H3K4me3, H3K36me3, and histone H3 density after depletion of Paf1 or Ski8 are shown. Regions from 3 kb upstream of TSS to 3 kb downstream of TES of all RefSeq genes were analyzed.
Fig 7
Fig 7. A model depicting the role of Paf1C in regulation of PolII progression and APA.
Prevalent transcript shortening and increased production of upstream antisense transcripts occurs after depletion of Paf1C subunits. See text for details.

Similar articles

See all similar articles

Cited by 18 articles

See all "Cited by" articles

References

    1. Jaehning JA. The Paf1 complex: platform or player in RNA polymerase II transcription? Biochim Biophys Acta. 2010;1799(5–6):379–88. Epub 2010/01/12. 10.1016/j.bbagrm.2010.01.001 - DOI - PMC - PubMed
    1. Zhu B, Mandal SS, Pham AD, Zheng Y, Erdjument-Bromage H, Batra SK, et al. The human PAF complex coordinates transcription with events downstream of RNA synthesis. Genes Dev. 2005;19(14):1668–73. Epub 2005/07/19. - PMC - PubMed
    1. Kim J, Guermah M, Roeder RG. The human PAF1 complex acts in chromatin transcription elongation both independently and cooperatively with SII/TFIIS. Cell. 2010;140(4):491–503. Epub 2010/02/25. 10.1016/j.cell.2009.12.050 - DOI - PMC - PubMed
    1. Chen Y, Yamaguchi Y, Tsugeno Y, Yamamoto J, Yamada T, Nakamura M, et al. DSIF, the Paf1 complex, and Tat-SF1 have nonredundant, cooperative roles in RNA polymerase II elongation. Genes Dev. 2009;23(23):2765–77. Epub 2009/12/03. 10.1101/gad.1834709 - DOI - PMC - PubMed
    1. Rahl PB, Lin CY, Seila AC, Flynn RA, McCuine S, Burge CB, et al. c-Myc regulates transcriptional pause release. Cell. 2010;141(3):432–45. Epub 2010/05/04. 10.1016/j.cell.2010.03.030 - DOI - PMC - PubMed

Publication types

LinkOut - more resources

Feedback