Volume 143, Issue 6, Pages 1018-1029 (December 2010) Comprehensive Polyadenylation Site Maps in Yeast and Human Reveal Pervasive Alternative Polyadenylation  Fatih Ozsolak, Philipp Kapranov, Sylvain Foissac, Sang Woo Kim, Elane Fishilevich, A. Paula Monaghan, Bino John, Patrice M. Milos  Cell  Volume 143, Issue 6, Pages 1018-1029 (December 2010) DOI: 10.1016/j.cell.2010.11.020 Copyright © 2010 Elsevier Inc. Terms and Conditions

Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 1 Polyadenylation Site Detection in Yeast and Human (A) The blue and black panels show the DRS reads emanating from transcripts in the + and – direction, respectively. The major peaks in the blue panel correspond to the 13 polyadenylation sites at locations 722690, 722692, 722695, 722710, 722716, 722718, 722723, 722726, 722746, 722750, 722752, 722775, and 722777 previously identified for HIS3 (Mahadevan et al., 1997) using 3′ RACE-PCR. (B) Zoomed-in view of (A). y axis was reduced from 0–300 scale to 0–50. x axis was reduced from 722,500–722,900 scale to 722,660–722,740. All “end tags” identified by Nagalakshmi et al. (2008) in this region are also shown (y axis for these tags is on the scale of 0–5). Arrows mark the sites identified by Mahadevan et al. (1997) in the region shown. (C and D) Overview (B) and a zoomed-in view (C) of reads mapping to UGT2B4 3′ annotated ends. Multiple potential polyadenylation sites are evident in panel C (see also Figure S2 and Table S1). Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 2 Characteristics of Polyadenylation Sites in Yeast and Human (A and B) Y-axes indicate the fraction of DRS reads aligning at x-distances (in 10 bp bins) relative to the annotated 3′ ends of yeast ORFs (A) and the annotated 3′ ends of human UCSC genes (B). (C and D) ADD2 (C) and BBOX1 (D) polyadenylation sites in human liver and brain. The polyadenylation sites identified (indicated as A1, A2 ,and A3) for both genes agree well with previous findings (Costessi et al., 2006; Rigault et al., 2006) (see also Figure S3 and Table S3). Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 3 Polyadenylation Motif Analyses Panels (A), (C), (E), and (G) indicate human motif elements identified. TTTTTTTTT (B), AWTAAA (D), CCAGSCTGG (F), and RGYRYRGTGG (H) distance distribution are shown in respective panels. Human categories were defined as sites that are within 5 nucleotides of annotated 3′ ends of known human genes in sense orientation (category 1), in the last exon and 1 kb downstream of annotated 3′ ends of human known genes in sense orientation (category 2), located anywhere within the transcripts in sense orientation except in categories 1 and 2 (category 3), antisense to genes (category 4) and in intergenic regions (category 5). In distance plots, y axis indicates the fraction of motifs (in percentages) at x-distances relative to the polyadenylation location (at base location 101) in each category. X-distances were calculated between the polyadenylation location identified with DRS and the first base immediately before the motif element. In panels B, F, and H, only the categories 3, 4, and 5 representing genic and intergenic sites were shown, because less than 10% (250–350) of these motifs were in categories 1 and 2, and not in sufficient numbers to be plotted in the graphs. Absolute numbers of motif counts for these latter three panels across all five human categories were provided in Figures S6A–S6C (see also Figure S4 and Table S5). Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 4 The Nucleotide Composition Surrounding Polyadenylation Cleavage Locations in Humans (A–E) Category descriptions were provided in Figure 3. y axis indicates the nucleotide composition (in percentages) at x-locations relative to the cleavage positions (at base location 101). Dark blue (diamond), blue (rectangle), green (triangle), and red (cross) lines indicate T, G, C, and A nucleotides, respectively. Polyadenylation locations in C3-5 differ from those in C1-2, and exhibit elevated T and A content in 40–50 nt upstream of polyadenylation cleavage positions (see also Figure S5 and Table S6). Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 5 Distance Distribution of Yeast EE (TAYRTA) Motif across Human Categories y axis indicates the fraction of motifs (in percentages) at x-distances relative to the cleavage positions (at base location 101) in each category. X-distances were calculated between the cleavage location identified with DRS and the first base immediately before the motif element. Human category descriptions were provided in Figure 3 legend. The enrichment of the EE motif immediately upstream of the cleavage sites in human categories 3, 4, and 5, but not in categories 1 and 2, is in parallel to the upstream human T-enrichment pattern shown in Figure 4 (see also Figure S6). Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S1 Characteristics of the DRS Approach, Related to Table 1 The data visualization was performed with the Integrated Genome Browser (http://www.affymetrix.com/partners_programs/programs/developer/tools/download_igb.affx). All coordinates are in the SGD/sacCer2 and NCBIv36 versions of the yeast and human genome throughout the manuscript. (A) Analysis of 40-mer oligoribonucleotide template DRS data revealed that the majority of sequencing initiation sites is at the last (3′) two bases of the template. Position 40 (P-40) signifies the 3′ end of the 40-mer oligoribonucleotide template. The resolution of the approach is thus ±2 nt. Further discussions are provided in the Extended Experimental Procedures. (B) Zoomed-in view of Figure 1C. Out of the five major clusters shown in Figure 1D, the zoomed-in view focuses on the dominant cluster at the annotated 3′ end of UGT2B4 gene (y axis is on the scale of 0–4000 reads). (C) Verification of novel unannotated polyadenylation regions with cloning and RACE in human. Four sites were validated: (1) intergenic chr11:62768208-62768259, (2) intergenic chr10:60431185-60431236, (3) genic chr5:42842861-42842913 and (4) intergenic chr12:131689811-131689863. The site shown is genic chr5:42842861-42842913. In addition, our data showed high agreement with the polyadenylation regions identified previously for 11 yeast genes (Duvel et al., 1999; Heidmann et al., 1992; Hoopes et al., 2000; Sparks and Dieckmann, 1998). Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S2 DRS Reads Mapping to Intergenic Regions Tend to Associate with Potential Regulatory Regions in S. cerevisiae, Related to Figure 1 (A) Read distribution around DNase I hypersensitive sites (DHSs). Distances are measured between the 5′ end of the divergent reads and the center of the annotated DHSs. A consistent enrichment can be observed in both reads from the “+” strand and reads from the “-“ strand, at around +22bp from the footprint centers. This enrichment of reads around the DNAse I footprints is statistically significant (p value = 8.041e-07, non-parametric Two-sample Kolmogorov-Smirnov test, see distributions in panel B). (B) Distribution of the number of reads at each position within two distinct classes of genomic regions: the footprint region, defined as all positions not further than 50bp away from the DHS centers (N = 2 x 50 = 100 nt), and the footprint flanks, which comprise the next 150 positions on each side of the DHS regions (N = 2X150 = 300 nt). Positions refer to the 5′ ends of the aligned reads. Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S3 Genomic Positioning of Polyadenylation Events, Related to Figure 2 (A) Zoomed-in view of Figure 2B to highlight lower frequency polyadenylation locations. (B–E) Sense and antisense novel transcription in yeast (B and C) and human (D and E). Panel-D shows a known human sense/antisense pair FGF2/FGF2-AS (Macfarlane et al., 2010). Panel-E exemplifies novel antisense transcription in the DAB1 locus that is also supported by a spliced AceView annotation. The scales of the DRS read counts are 1 to 50 for both strands on panels B and C, 1 to 10 for both strands on panels D and E. Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S4 Polyadenylation Site Definition, Related to Figure 3 In this graph, each point of a curve represents the sensitivity/specificity values of a set of polyadenylation sites produced by applying a given maxgap (indicated by the color) and sensitivity/specificity threshold (ranging from 1 to 50%). At high threshold values, the stringency increases while the number of predictions and the sensitivity drop (upper left corner). Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S5 Motif and Sequence Characteristics of Polyadenylation in Yeast, Related to Figure 4 (A and B) Distance distributions of PE (AAWAAA) and EE (TAYRTA) motif are shown in panels A and B, respectively, in yeast. y axis indicates the fraction of motifs (in percentages) at x-distances relative to the cleavage positions (at base location 101) in each category. X-distances were calculated between the cleavage location identified with DRS and the first base immediately before the motif element. Yeast category descriptions were provided in panels C-F below. (C–F) The nucleotide composition surrounding polyadenylation cleavage locations in yeast. Nucleotide compositions surrounding yeast cleavage sites that are within 200 nucleotides downstream of 3′ ends of yeast ORFs in sense orientation (category 1, C), located within ORFs in sense orientation (category 2, D), located within ORFs in antisense orientation (category 3, E), and in intergenic regions at least 1kb away from known 3′ ends of yeast ORFs (category 4, F) are shown. Categories 1, 2/3 and 4 are mutually exclusive; categories 2 and 3 are not. y axis indicates the nucleotide composition (in percentages) at x-locations relative to the cleavage positions (at base location 101). Dark blue (diamond), blue (rectangle), green (triangle), and red (cross) lines indicate T,G,C, and A nucleotides, respectively. A T-rich region is evident in all categories in 40-50 nt upstream regions of cleavage positions. Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S6 Novel Motif Characteristics of Polyadenylation Regions, Related to Figure 5 (A–C) The figures represent the distribution of motifs in respective Figure 3B (TTTTTTTTT, S6A), Figure 3F (CCAGSCTGG, S6B), and Figure 3H (RGYRYRGTGG, S6C), except that y axis here represent the absolute number of motifs at each nucleotide location. Cell 2010 143, 1018-1029DOI: (10.1016/j.cell.2010.11.020) Copyright © 2010 Elsevier Inc. Terms and Conditions