E2F mediates enhanced alternative polyadenylation in proliferation
- Ran Elkon†1,
- Jarno Drost†1,
- Gijs van Haaften†1,
- Mathias Jenal1,
- Mariette Schrier1,
- Joachim AF Oude Vrielink1 and
- Reuven Agami1, 2Email author
© Elkon et al.; licensee BioMed Central Ltd. 2012
Received: 16 February 2012
Accepted: 2 July 2012
Published: 2 July 2012
The majority of mammalian genes contain multiple poly(A) sites in their 3' UTRs. Alternative cleavage and polyadenylation are emerging as an important layer of gene regulation as they generate transcript isoforms that differ in their 3' UTRs, thereby modulating genes' response to 3' UTR-mediated regulation. Enhanced cleavage at 3' UTR proximal poly(A) sites resulting in global 3' UTR shortening was recently linked to proliferation and cancer. However, mechanisms that regulate this enhanced alternative polyadenylation are unknown.
Here, we explored, on a transcriptome-wide scale, alternative polyadenylation events associated with cellular proliferation and neoplastic transformation. We applied a deep-sequencing technique for identification and quantification of poly(A) sites to two human cellular models, each examined under proliferative, arrested and transformed states. In both cell systems we observed global 3' UTR shortening associated with proliferation, a link that was markedly stronger than the association with transformation. Furthermore, we found that proliferation is also associated with enhanced cleavage at intronic poly(A) sites. Last, we found that the expression level of the set of genes that encode for 3'-end processing proteins is globally elevated in proliferation, and that E2F transcription factors contribute to this regulation.
Our results comprehensively identify alternative polyadenylation events associated with cellular proliferation and transformation, and demonstrate that the enhanced alternative polyadenylation in proliferative conditions results not only in global 3' UTR shortening but also in enhanced premature cleavage in introns. Our results also indicate that E2F-mediated co-transcriptional regulation of 3'-end processing genes is one of the mechanisms that links enhanced alternative polyadenylation to proliferation.
Cleavage and polyadenylation are required for the maturation of most mRNA transcripts . The pre-mRNA is cleaved approximately 10 to 30 nucleotides downstream of the polyadenylation signal (PAS) and an untemplated poly(A) tail is added to the transcript. The canonical PAS is AAUAAA, which appears in about 50% of the cleavage sites (CSs). More than 10 PAS variants have been documented, the most prevalent of which is AUUAAA [2, 3]. In addition, various upstream and downstream auxiliary elements were found to regulate the efficiency of poly(A) site usage, prominent among them U-rich and GU-rich elements downstream of the cleavage sites . Recently, using proteomics analysis, more than 90 proteins were shown to consist or physically interact with the pre-mRNA 3'-end processing machinery in human cells . Important units in this machinery are the cleavage and polyadenylation specific factor (CPSF) complex, which interacts with the PAS, and the cleavage stimulation factor (CstF) complex, which interacts with the U/GU-rich downstream elements [5, 6].
Recent studies have demonstrated that alternative cleavage and alternative polyadenylation (APA) are much more prevalent than was previously appreciated, and that they involve more than 50% of mammalian genes [7–9]. The number of genes reported to contain alternative cleavage and polyadenylation sites in their 3' UTR is steadily increasing, mainly due to the maturation of deep-sequencing techniques specifically adapted for mapping of transcript 3'-ends [7, 8, 10, 11]. 3' UTRs carry high regulatory potential as they serve as a flexible platform for interactions between target transcripts and RNA-binding proteins and regulatory RNAs (mainly microRNAs (miRNAs)) [12, 13]. Those interactions play major roles in regulating transcript stability, localization and translational efficiency [12, 13]. APA modulates these modes of gene regulation by generating gene isoforms that differ in their 3' UTRs.
The functional significance of APA is largely unknown and is just beginning to emerge (recent reviewed in ). Recent reports linked APA to proliferation , development [8, 16, 17] and cancer . Sandberg et al.  reported a general link between proliferation and 3' UTR shortening in multiple cell types and tissues. Mechanisms that underlie this broad 3' UTR shortening associated with increased proliferation are currently unknown. Mayr and Bartel  linked APA also to cancer and suggested that cellular transformation is also associated with broad 3' UTR shortening, over and above the association expected from their proliferative state. However, as these two studies were based on expression array data, whose capacity to study APA is limited, comprehensive analysis of APA during cellular proliferation and transformation transitions is still lacking.
Accumulating evidence indicates a dynamic interplay between polyadenylation and splicing. It was reported that approximately 20% of human genes have polyadenylation sites within their introns, most of which result in altered protein products . A dynamic competition between the cleavage and splicing machineries was implied by the fact that intronic cleavage sites were more prevalent in larger introns with weaker 5' splice site signal. An additional insight into the interplay between these two processes was provided by a recent study that observed that knocking-down U1 small nuclear ribonucleoprotein (U1 snRNP) resulted in extensive induction of cleavage at numerous cryptypic polyadenylation sites within introns, suggesting that, under normal conditions, this factor protects against premature intronic cleavage .
In this study we set out to dissect and compare, on a transcriptomic-scale, APA events associated with proliferation and transformation. We applied a deep-sequencing method for mapping and quantification of polyadenylation sites to two human cellular systems, primary fibroblast BJ cells and non-transformed mammary epithelial MCF10A cells, each of them examined under proliferative, arrested and transformed states. In both cell lines, the proliferative state was strongly associated with enhanced APA, resulting not only in global 3' UTR shortening but also in enhanced premature cleavage at intronic polyadenylation sites. The link between 3' UTR shortening and proliferation was markedly stronger than its association with transformation. Furthermore, our results pinpoint E2F-mediated transcriptional co-regulation of genes that encode 3'-end processing proteins as one of the mechanisms underlying the association between enhanced APA and proliferation.
Transcriptome-wide characterization of polyadenylation sites using 3'-Seq
Next, we examined the overlap between the CSs identified in our experiment and the annotated sites recorded in polyA-DB . While 40% of the 3' UTR CSs in our dataset overlapped polyA-DB records, only 3% of the intronic CSs did so. We observed that the set of 3' UTR CSs identified in our dataset that did not overlap polyA-db records were significantly weaker than those that did overlap, and that intronic CSs were, as a set, significantly weaker than 3' UTR sites (Figure 1d), showing the improved sensitivity of 3'-Seq in identification of weaker poly(A) sites.
Association between APA and proliferation in primary fibroblast BJ cells
3'-Seq allows not only the identification of poly(A) sites, but also the quantification of gene expression levels (similar to standard RNA-Seq, the number of reads that align to a transcript provides an estimate for its expression level). Gene Ontology (GO) analysis showed that the set of genes that were repressed in the transition of BJ cells from the proliferative to confluent state were highly enriched for cell-cycle related functional categories, and the set of genes that were induced in this state were enriched for developmental and adhesion processes (Table S2a in Additional file 1). To obtain a quantitative measure of the proliferative status of the examined cells, we calculated for each condition a proliferation index, based on the relative expression levels of cell-cycle-related genes (Materials and methods). The proliferation indices calculated for the proliferating and confluent BJ cells reflected the strong decrease in proliferation in the transition to the confluent state (Figure 2d).
Next, we examined whether the change in CS usage was associated with any alterations in expression level. Under the assumptions that 3' UTRs mainly serve as a platform for miRNA-mediated regulation, and that miRNAs generally restrict target transcript levels, we hypothesized that 3' UTR shortening would result in increased expression of the affected transcripts due to removal of miRNA target sites. However, we did not observe that changes in poly(A) CS usage resulted in a significant effect on transcript level (Figure S2a in Additional file 1). We therefore conclude that the proliferative state of primary human BJ cells is associated with a widespread increase in the usage of proximal poly(A) CSs, resulting in 3' UTR shortening in hundreds of transcripts, but this induction of APA does not seem to have a significant global effect on the expression level of the targeted transcripts.
Association between APA and proliferation in non-transformed epithelial MCF10A cells
Subsequently, we wished to generalize the results we obtained with BJ cells using a second cellular model, human non-transformed MCF10A epithelial cells, which were arrested by serum deprivation. In this dataset we identified 8,835 transcripts, of which 1,681 contained multiple CSs in their 3' UTRs. Here too, the proliferative state was associated with increased cleavage at proximal 3' UTR CSs, reflected by the higher value of the proximal PUI measure (Figure 2e). Furthermore, among the 1,681 transcripts with multiple CSs, 788 showed significant shift (P < 0.001) in CS usage in the comparison between serum-deprived and control MCF10A cells. Significantly, proliferation was strongly associated with a broad shortening of 3' UTRs: 590 transcripts (75%) showed enhanced usage of proximal CSs in the proliferating cells compared to the serum-starved ones (P = 2.5 × 10-46, binomial distribution tail) (Figure 2f). Gene expression analysis and calculation of the proliferation index demonstrated the strong decrease in proliferation that resulted from the deprivation of serum (Figure 2g; Table S2b in Additional file 1). Similar to BJ cells, changes in APA did not affect mRNA expression levels (Figure S2b in Additional file 1), reinforcing the conclusion that the strong APA modulation associated with proliferation does not have a major global effect on mRNA levels. Still, it is possible that only few genes show marked change in mRNA levels following APA, or that APA impacts global gene expression at regulatory layers downstream of mRNA production and stability (for example, mRNA localization and translation efficiency).
Proliferation is associated with enhanced cleavage at intronic poly(A) sites
APA events during cellular transformation
Analyzing APA associated with transformation of MCF10A cells led to a similar conclusion. We transformed MCF10A cells by ectopic induction of oncogenic RASG12V (which is sufficient, without any further genetic manipulation, for efficient transformation of MCF10A ), and harvested them at time points 0, 2 and 8 days following RASG12V induction. Figure S3 in Additional file 1 shows the efficient induction of RASG12V and its associated induction of transformation phenotype as measured by anchorage-independent growth of cells in soft agar. The intersection between the two time points after RAS induction identified 221 transcripts that showed significant shift in CS usage that was consistent in its direction (only 19 changes occurred in opposite directions). Out of these 221 transcripts, 139 (63%) showed a shift towards proximal CSs (Figure 5d,e). A slight decrease in the proliferation index was observed in RASG12V-transformed MCF-10A cells (Figure 5f). Altogether, this cellular model also demonstrates that the 3' UTR shortening effect, which is associated with the transition from the arrested to proliferating state, is stronger than the 3' UTR shortening effect associated with the transition from the proliferative to transformed state.
E2F transcriptional regulation of 3'-end processing genes
The co-repression observed for the set of transcripts encoding the proteins of the 3'-end processing machinery suggested their transcriptional co-regulation. Hence, we next searched their promoter regions for over-represented cis-regulatory motifs. Interestingly, the top-scoring motif detected by de novo motif analysis matched the binding signature of the E2F family (Figure 6b), which is one of the master regulators of the transcriptional network associated with cell cycle progression . As expected, the expression levels of E2F1 and E2F2 significantly decreased in the transition from proliferative to confluent state (by 3- and 7.5-fold, respectively; Figure 6c). To further support the regulation of the 3'-end processing machinery by E2F, we searched for publicly available datasets that profiled E2F-chromatin interactions on a genomic scale. ChIP-Seq analyses were recently applied to the E2F1  and E2F4  members of this family. While E2F4 mainly acts as transcriptional repressor, to a large extent it shares its binding signature and target genes with E2F1-3, which are transcriptional activators . We found that many promoters of key 3'-end processing genes were reported to be bound by E2F1 and/or E2F4 (Figure 6d). To demonstrate more directly a role for E2F in the regulation of 3'-end processing genes, we cloned promoter regions of eight core components of the cleavage machinery upstream of a luciferase gene in the pGL3 reporter construct, and examined the effect of knocking-down E2F1 on promoter activity. The promoter region of MCM2, a potent E2F target gene , was used as a positive control, and a reporter containing an artificial p53 promoter was used as negative ones. In six out of the eight 3'-end processing gene promoters examined, silencing E2F1 resulted in a statistically significant reduction in promoter activity (Figure 6e), and in the other two cases reduction in promoter activity was also observed, but did not reach the statistical cutoff of 0.05. (The effect of E2F1-kd was mild on all the targets examined, including the positive control, MCM2, probably due to redundancy between E2F members and residual E2F1 activity remaining after knocking it down.) Next, we examined the effect of knocking down E2F1 and E2F2 on the expression of three core cleavage factors: CSTF2, CSTF3 and CPSF2. In all these cases we confirmed that knocking down E2F resulted in deceased levels of the cleavage factors (Figure 6f). Last, we selected three target genes that showed highly significant induction of cleavage at 3' UTR proximal CSs in proliferation, and examined the effect of knocking down E2F1 and E2F2 on the relative usage of the proximal and distal sites of these transcripts. Reducing the levels of E2F indeed resulted in weakened cleavage at the proximal sites of these transcripts (Figure 6g).
Taken together, our results indicate that E2F-mediated co-regulation of genes that function in recognition and cleavage of poly(A) sites contributes to the link between increased cellular proliferation and enhanced usage of proximal CSs (Figure 6h).
Previous studies that reported the association between APA and proliferation, transformation and development were mainly based on microarray measurements [15–18]. Consequently, these analyses were limited to those 3' UTRs that were covered by array probes both downstream and upstream of proximal CSs (a setting that allows per gene comparison between the expression level of the short and long 3' UTR isoforms), to pre-defined poly(A) sites inferred from EST databases, and to 3' UTRs with exactly two poly(A) sites. Here, we applied 3'-Seq, a deep-sequencing technique specifically designed for identification and quantification of poly(A) sites. Applying this technique to human cellular systems in proliferative and arrested states enabled us to examine APA events associated with proliferation at a much improved scale and sensitivity. Our analysis identified hundreds of previously uncharacterized poly(A) sites and dozens of novel APA events that occur during the transitions between the proliferative, arrested and transformed states. Moreover, beyond confirming on a broader scale and with improved resolution the global 3' UTR shortening associated with cellular proliferation, the two very different cellular models used in our study allowed the identification of a core set of 216 transcripts subjected to this mode of regulation. The statistical test that identifies changes in CS usage is inherently less sensitive to lowly expressed transcripts (covered by lower number of sequenced reads; Figure S5 in Additional file 1), and therefore it is very likely that many more transcripts undergo 3' UTR shortening in proliferation. In agreement, random partition of our dataset into subsets of increasing size showed that the number of identified APA events in our dataset was not reaching saturation (data not shown). Hence, the identification of APA events on lowly expressed genes requires increase in sequencing depth.
The mechanism underlying the enhanced APA associated with proliferation is unknown. Previous reports observed that embryonic development is associated with progressive lengthening of 3' UTRs, which is correlated with decreased levels of the set of genes encoding the 3'-end processing machinery [17, 29]. This observation indicated that 3' UTR lengthening is likely caused by weakening of mRNA polyadenylation activity. Similarly, our analysis identified elevated expression of the set of genes encoding the 3'-end processing machinery in proliferating BJ cells compared to arrested ones. The collective induction of this set of genes could explain the extensive 3' UTR shortening observed in proliferation. Mechanistically, the co-induction of these genes in proliferation suggested their co-transcriptional regulation. In accordance, we identified significant over-representation of the E2F binding motif in the promoters of this set of genes. Additionally, we noted a physical interaction of members of the E2F family with many promoters in this set of genes. We demonstrated empirically a role for E2F1 in regulating key genes in this set, and showed that reducing E2F levels decreased cleavage at proximal 3' UTR CSs of selected target transcripts. Thus, our results suggest that elevated levels of E2F members in proliferative states lead to transcriptional activation of key 3'-end processing genes, resulting in enhanced 3' cleavage and polyadenylation activity, which leads to widespread 3' UTR shortening.
Furthermore, our results show that the enhanced APA associated with proliferation results not only in global 3' UTR shortening but also in increased rates of premature cleavage at intronic poly(A) sites. This observation also indicates augmentation of mRNA polyadenylation activity in proliferation. A previous report suggested a dynamic competition between the splicing and polyadenylation apparatuses , and recently a mechanism that suppresses premature cleavage and polyadenylation from cryptic polyadenylation signals located in introns was documented . Our results suggest that E2F-mediated elevation in expression of 3'-end processing genes in proliferation causes augmented cleavage at both intronic and 3' UTR proximal CSs. Yet, it is clear that many more factors are involved in APA regulation. Using a RNA interference screen, we recently identified PABPN1 as an APA-regulator . We found that knocking-down PABPN1 resulted in global enhancement of cleavage at proximal 3' UTR CSs, reminiscent of the effect observed in the transition from the arrested to proliferative state. We are currently studying the possible role for PABPN1 in regulating APA in proliferation.
3' UTR shortening was previously associated with both proliferation  and oncogenic transformation , linking APA to human cancer. However, the relative contribution of the enhanced proliferative capacity and the transformed phenotype to the global shortening of 3' UTRs had not previously been empirically assessed. We employed a very controlled setup using two very different cellular systems manipulated to change their status from the proliferative to arrested or transformed states. Both our systems, BJ and MCF10A cells, led to the same observation: increased usage of proximal poly(A) sites was more strongly associated with proliferation than with transformation. Recently, Fu et al.  applied a deep-sequencing method to compare APA between two breast cancer cell lines, MCF7 and MB231, and the non-transformed MCF10A cells. They found opposing tendencies in the change in 3' UTR length: while MCF7 cells showed mainly 3' UTR shortening compared to MCF10A cells, MB231 cells showed extensive 3' UTR lengthening, further indicating the complexity of APA regulation and demonstrating its dependence on cellular context beyond the effect of proliferative and transformation states.
3' UTRs carry high regulatory potential by serving as major docking platforms for RNA-binding proteins and miRNAs. As miRNAs are believed to play mainly a repressive role, it could have been expected that the global loss of miRNA target sites stemming from the extensive shortening of 3' UTRs in proliferation would result in increased expression of the affected transcripts. However, in the two cellular systems that we examined we did not find any global effect for the 3' UTR shortening/lengthening on expression level of the target genes. Similar to our observation, the global 3' UTR shortening detected in response to activation of immune cells  did not result in a significant effect on transcript expression level. It is possible that the enhanced APA elicits strong changes in mRNA levels of only a few genes and/or a more significant effect on translation and mRNA subcellular localization of the target genes, as 3' UTRs play major roles in these two layers of gene regulation as well.
It is also relevant to note that our conclusion on global 3' UTR shortening in proliferation is based on the observation that, in these conditions, the levels of proximal poly(A) sites were globally elevated relative to the level of distal ones. Yet, such a global increase in the relative levels of proximal poly(A) sites could, in principle, stem not only from enhanced APA, but also from some widespread mRNA destabilization that preferentially affects the longer isoforms (for example, global induction of miRNA activity in proliferation compared to arrested cells). A model that argues for a global destabilization of longer isoforms as a major contributor to the relative increase in the level of proximal CSs would predict that the transcripts that undergo 3' UTR shortening should show overall decreased expression. However, no evidence for such an effect was observed in our data. Furthermore, the elevated level of the set of genes encoding 3'-end processing factors and the enhanced cleavage at intronic poly(A) sites observed in the proliferative states strongly argue that enhanced polyadenylation and cleavage activity, and not stability modulation of the target transcripts, indeed comprise the underlying mechanism for the relative increase in the levels of the proximal CSs.
Our knowledge on the modulation of APA in various biological conditions is just beginning to grow, and this layer of gene regulation is anticipated to be involved in many processes relevant to human disease. Yet, the functional significance of APA is still unclear. What is the objective of the program that leads to the extensive 3' UTR shortening in cellular proliferation? And on the other hand, what is the biological significance of the broad 3' UTR lengthening associated with differentiation? These are still open questions that call for further investigations.
Materials and methods
Primary BJ foreskin fibroblasts containing expression constructs for the ecotropic receptor and the human telomerase gene (hTERT), BJ-EHT, were either maintained in a cycling state or contact inhibited by growing them to confluence in DMEM (Gibco, Bleiswijk, The Netherlands) supplemented with 10% fetal calf serum and antibiotics. BJ-EHT/p53kd/p16kd/RASGV12ER cells expressing shRNA constructs targeting p53 and p16INK4A and 4-hydroxy-tamoxifen (4-OHT)-inducible oncogenic HRASG12V were cultured for 3 days in the presence of 100 nM 4-OHT to transform the cells. These cell lines are described in detail in . BJ-EHT cells were transfected in a final concentration of 50 nM small interfering RNAs targeting E2F1 (targeting sequence 5′-GGCCCGAUCGAUGUUUUCC-3′) and E2F2 (targeting sequence 5′-GACUCGGUAUGACACUUCG-3′) using Dharmafect reagent (Dharmacon, Lafayette, Colorado, USA).
Non-transformed mammary epithelial MCF10A cells were either cultured in DMEM:F12 Ham's medium (Sigma, Zwijndrecht, The Netherlands) 1:1, supplemented with 10% fetal calf serum (Gibco), insulin (10 μg ml-1; Sigma), hydrocortisone (0.5 μg ml-1) and epidermal growth factor (20 ng ml-1; Peprotech, New Jersey, USA) or serum starved with DMEM:F12 containing no supplements for 48 hours. To transform the MCF10A cells, we transduced them with a retroviral vector expressing RASG12VER, and cultured them for 2 and 8 days in the presence of 100 nM 4-OHT.
U2OS cells were cultured in DMEM (Gibco) supplemented with 10% fetal calf serum and antibiotics.
Construction of 3'-Seq libraries
Schematic representation of the 3'-Seq method is provided in Figure S1a in Additional file 1. Briefly, total RNA was extracted from either BJ or MCF-10A cells described above using the RNeasy Mini kit from Qiagen (Venlo, The Netherlands). mRNA was isolated with the Oligotex mRNA kit from Qiagen and 500 to 600 ng of the isolated mRNA was heat fragmented for 5 minutes at 70°C. The fragmented mRNA was then converted to single-stranded cDNA using SuperScript III reverse transcriptase (Invitrogen, Bleiswijk, The Netherlands) and a P7-t25-vn oligo-dT primer (5'-CAAGCAGAAGACGGCATACGAGATTTTTTTTTTTTTTTTTTTTTTTTTVN-3') according to the manufacturer's instructions. This was followed by second-strand cDNA synthesis (Invitrogen) and end-repair using T4 DNA polymerase, T4 polynucleotide kinase (NEBNext End Repair Enzyme Module, New England Biolabs, New England, USA) and Klenow DNA polymerase (Invitrogen). After purification with the QIAquick PCR Purification Kit (Qiagen), 1 µl of 45 µM annealed P5-splinkerette (forward, 5'-ATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT-3'; reverse, 5'-AGATCGGAAGAGCGTCGTGTAGGGTTTTTTTTTTCAAAAAAA-3') was ligated overnight at 16°C to the cDNA. The splikerette-ligated cDNA was then size-selected for approximately 220 bp long cDNA fragments using the E-Gel iBase Power System from Invitrogen. The final 3'-Seq libraries were then generated by PCR amplification of the linker-ligated cDNA with P5 and P7 primers. After a last purification step the 3'-Seq libraries were sequenced on the Illumina Genome Analyzer (BJ proliferation, confluent and RAS-transformed samples) or the Illumina HiSeq system (all MCF10A samples and technical repeats of the BJ proliferating and confluent samples).
Analysis of 3'-Seq data
Sequenced reads were aligned against the human genome (hg18) using Bowtie . Up to two mismatches were allowed in the reads' seed region (the reads' first 28 nucleotides). To allow the alignment of reads that span poly(A) cleavage sites and therefore contain the start of the untemplated poly-A tail, Bowtie's -e parameter was increased to tolerate mismatches in all bases after the seed region. Only uniquely mapped reads were used in subsequent analysis. Numbers of total and uniquely mapped reads are provided in Table S1 in Additional file 1. Wig files that summarize read coverage along the chromosomes (normalized to 10 million mapped reads) and raw sequence files are available at the Gene Eexpression Omnibus (accession number GSE33592). To map reads to genes and genomic regions (for example, 3' UTRs, introns, coding sequences, and so on), gene coordinates and annotations were extracted from the human Ensembl-Gene table of the UCSC browser . To cover novel cleavage sites located downstream of current transcript 3'-end annotations, we extended the 3' UTR of each transcript by 1,000 bp.
Mapping and characterization of poly(A) cleavage sites
In order to identify poly(A) CSs we searched for uniquely mapped reads that contained untemplated stretches of As (Figure S1b in Additional file 1). To reduce false calls that stem from priming of the oligo-dT primer to internal A-rich regions within transcripts, we required that, to support a CS call, reads should contain a stretch of at least eight As and at least five of the first eight As in the stretch should mis-match the corresponding bases on the transcript reference sequence (Figure S1c in Additional file 1). The location of the cleavage was taken as the location where the untemplated A stretch started. Since the location of the CSs often fluctuated around a major site, for each gene and sample we calculated a 'poly(A) CS profile', which recorded the number of reads supporting a cleavage at each position along the transcript (Figure S1e in Additional file 1). We considered the local maxima of these CS profiles as the CS locations, and required spacing of at least 50 nucleotides between consecutive CSs (in case of lower spacing between CSs, the stronger, that is, the one supported by a higher number of reads, was chosen). Only CSs supported by at least ten reads were considered in subsequent analyses.
Differential usage of poly(A) cleavage sites
A decrease in <CS_J> indicates 3' UTR shortening in the corresponding condition and vice versa.
Poly(A) site usage index
where E j is the level of the peak associated with the j-th cleavage site, and <E> is the geometric mean of the levels of all the peaks associated with the N cleavage sites. (By definition, the proximal PUI of a transcript in a certain condition is PUI j = 1 .) In each condition, we calculated the distribution of proximal PUIs, and took its median as a global measure for the usage of proximal CSs in the corresponding samples. Global 3' UTR shortening in a certain condition is reflected by increase in its median proximal PUI.
Gene expression analysis
As an estimate for expression level in each sample we took for each gene the total number of reads mapping to its 3' UTR. Quantile normalization and lower bound of expression level of five reads (to avoid division by 0) were applied before comparing expression levels between samples. GO enrichment analysis was carried out using GOrilla . All expression analyses, except the one presented in Figure S4a in Additional file 1, were based on the 3'-Seq data. The analysis shown in Figure S4a in Additional file 1 is based on standard RNA-Seq applied to BJ cells either grown under normal proliferative conditions or deprived of serum for 48 hours. RNA-Seq was carried out using standard protocols and used Illumina's HiSeq platform. Reads were mapped to the canonical set of human transcripts, and expression levels were estimated using RPKM (reads per kilobase of exon model per million mapped reads) measures.
De novomotif discovery
Sequences flanking ±50 nucleotides relative to the poly(A) CSs were extracted from the genome using Galaxy . De novo motif discovery analysis was done using AMADEUS , and was applied to either CS regions or to the set of promoters of 94 genes encoding proteins consisting or physically interacting with the 3'-processing machinery . Promoter regions were taken as -1,000 to +200 bp relative to the annotated transcription start sites. The set of promoters of all annotated human genes served as a background set for the analysis.
As a measure that reflects the proliferation status of cells under different conditions, we defined the proliferation index in each condition as the (log2) ratio between the median expression level of cell-cycle-related genes (genes assigned to the GO 'cell-cycle' functional category) and the overall median expression level in that condition.
Promoter activity assays
U2OS cells were co-transfected with 450 ng of pRS or pRS-E2F1KD in combination with 50 ng luciferase reporter and 5 ng renilla plasmid using polyethylenimine. Three days after transfection, luciferase assays were performed in accordance with the manufacturer's instructions (Dual Luciferase system; Promega, Utrecht, The Netherlands).
qRT-PCR and 3'-end qRT-PCR
Total RNA was extracted using Trizol reagent (Invitrogen). cDNA was synthesized using Superscript III reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Quantitative RT-PCR was performed with Light Cycler 480 Sybr Green I Master mix (Roche Applied Science, Almere, The Netherlands) and the Chromo 4 Real Time PCR Detector (Bio-Rad Laboratories, Providence, Rhode Island, USA). Primer sequences were: CSTF2, forward 5'-ATCCTTGCCTGCGAATGTCC-3', reverse 5'-GGGTGGTCCTCGGCTCTC-3'; CSTF3, forward 5'-TGAAGTGGATAGAAAACCAGAATACCC-3', reverse 5'-TGGAAACAGATAGGAGGAGGGAGA-3'; CPSF2, forward 5'-CTTTGGAACCCTTGCCACCT-3', reverse 5'-TGCGGACTGCTACTTGATTGTTG-3'; E2F2, forward 5'-ACTGGAAGTGCCCGACAGGA-3', reverse 5'-GGTGGAGGTAGAGGGGAGAGG-3'.
For 3'-end qRT-PCR, 2.5 μg of total RNA was heat fragmented with fragmentation buffer (Ambion, Bleiswijk, The Netherlands) for 5 minutes at 70°C. The fragmented total RNA was precipitated and the first strand of cDNA was synthesized as described above using a P7-t25-vn oligo-dT primer (5'-AAGCAGAAGACGGCATACGAGATTTTTTTTTTTTTTTTTTTTTTTTTVN-3'). qRT-PCR was performed as described above using P7-primer and the following gene-specific forward primers: β-actin distal forward 5'-CAGCCAGGGCTTACCTGT-3'; RNF220 intron forward 5'-TTTGGGTGGGGAAATGGAAT-3'; RNF220 distal forward 5'-CCTGTGGTGTGATGCTGTGTCT-3'; FAM70B intron forward 5'-CTCTGGCTCTGGGCTTCCC-3'; FAM70B distal forward 5'-GTTGATGCCCCCTGTGTTTG-3'; FAM100B proximal forward 5'-CAGGAGTTTTTCAGGCAAGTTTTTC-3'; FAM100B distal forward 5'-AGTGGAGAGCCTGCCTTTGG-3'; PTGS1 proximal forward 5'-GGCAAGGAAGTGGGGTGTTC-3'; PTGS1 distal forward 5'-CCTGCTAGTCTGCCCTATGGATTT-3'; TMEM119 proximal forward 5'-CCCTGGCAACATTGTGAGACC-3'; TMEM119 distal forward 5'-TCTCCCCCATCCCTCCATCT-3'.
cleavage and polyadenylation specific factor
Dulbecco's modified Eagle's medium
polymerase chain reaction
- poly(A) CS:
polyadenylation cleavage site
poly(A)-site usage index
short hairpin RNA
We thank all members of the Agami laboratory for technical help and discussions. We thank Arno Velds and Ron Kerkhoven at the NKI Central genomics facility for deep sequencing our samples. This work was supported by ERC (European Research Council), KWF (Dutch Cancer Foundation), and VICI-NWO (Nederlandse Organisatie voor Wetenschappelijk Onderzoek) to RA, SNF (The Swiss National Science Foundation) to MJ, and VENI-NWO to GvH.
- Colgan DF, Manley JL: Mechanism and regulation of mRNA polyadenylation. Genes Dev. 1997, 11: 2755-2766. 10.1101/gad.11.21.2755.PubMedView ArticleGoogle Scholar
- Hu J, Lutz CS, Wilusz J, Tian B: Bioinformatic identification of candidate cis-regulatory elements involved in human mRNA polyadenylation. RNA. 2005, 11: 1485-1493. 10.1261/rna.2107305.PubMedPubMed CentralView ArticleGoogle Scholar
- Beaudoing E, Freier S, Wyatt JR, Claverie JM, Gautheret D: Patterns of variant polyadenylation signal usage in human genes. Genome Res. 2000, 10: 1001-1010. 10.1101/gr.10.7.1001.PubMedPubMed CentralView ArticleGoogle Scholar
- Shi Y, Di Giammartino DC, Taylor D, Sarkeshik A, Rice WJ, Yates JR, Frank J, Manley JL: Molecular architecture of the human pre-mRNA 3' processing complex. Mol Cell. 2009, 33: 365-376. 10.1016/j.molcel.2008.12.028.PubMedPubMed CentralView ArticleGoogle Scholar
- Neilson JR, Sandberg R: Heterogeneity in mammalian RNA 3' end formation. Exp Cell Res. 2010, 316: 1357-1364. 10.1016/j.yexcr.2010.02.040.PubMedPubMed CentralView ArticleGoogle Scholar
- Lutz CS: Alternative polyadenylation: a twist on mRNA 3' end formation. ACS Chem Biol. 2008, 3: 609-617. 10.1021/cb800138w.PubMedView ArticleGoogle Scholar
- Ozsolak F, Kapranov P, Foissac S, Kim SW, Fishilevich E, Monaghan AP, John B, Milos PM: Comprehensive polyadenylation site maps in yeast and human reveal pervasive alternative polyadenylation. Cell. 2010, 143: 1018-1029. 10.1016/j.cell.2010.11.020.PubMedPubMed CentralView ArticleGoogle Scholar
- Shepard PJ, Choi EA, Lu J, Flanagan LA, Hertel KJ, Shi Y: Complex and dynamic landscape of RNA polyadenylation revealed by PAS-Seq. RNA. 2011, 17: 761-772. 10.1261/rna.2581711.PubMedPubMed CentralView ArticleGoogle Scholar
- Tian B, Hu J, Zhang H, Lutz CS: A large-scale analysis of mRNA polyadenylation of human and mouse genes. Nucleic Acids Res. 2005, 33: 201-212. 10.1093/nar/gki158.PubMedPubMed CentralView ArticleGoogle Scholar
- Jan CH, Friedman RC, Ruby JG, Bartel DP: Formation, regulation and evolution of Caenorhabditis elegans 3'UTRs. Nature. 2011, 469: 97-101. 10.1038/nature09616.PubMedPubMed CentralView ArticleGoogle Scholar
- Fu Y, Sun Y, Li Y, Li J, Rao X, Chen C, Xu A: Differential genome-wide profiling of tandem 3' UTRs among human breast cancer and normal cells by high-throughput sequencing. Genome Res. 2011, 21: 741-747. 10.1101/gr.115295.110.PubMedPubMed CentralView ArticleGoogle Scholar
- Fabian MR, Sonenberg N, Filipowicz W: Regulation of mRNA translation and stability by microRNAs. Annu Rev Biochem. 2010, 79: 351-379. 10.1146/annurev-biochem-060308-103103.PubMedView ArticleGoogle Scholar
- Andreassi C, Riccio A: To localize or not to localize: mRNA fate is in 3'UTR ends. Trends Cell Biol. 2009, 19: 465-474. 10.1016/j.tcb.2009.06.001.PubMedView ArticleGoogle Scholar
- Di Giammartino DC, Nishida K, Manley JL: Mechanisms and consequences of alternative polyadenylation. Mol Cell. 2011, 43: 853-866. 10.1016/j.molcel.2011.08.017.PubMedPubMed CentralView ArticleGoogle Scholar
- Sandberg R, Neilson JR, Sarma A, Sharp PA, Burge CB: Proliferating cells express mRNAs with shortened 3' untranslated regions and fewer microRNA target sites. Science. 2008, 320: 1643-1647. 10.1126/science.1155390.PubMedPubMed CentralView ArticleGoogle Scholar
- Ji Z, Lee JY, Pan Z, Jiang B, Tian B: Progressive lengthening of 3' untranslated regions of mRNAs by alternative polyadenylation during mouse embryonic development. Proc Natl Acad Sci USA. 2009, 106: 7028-7033. 10.1073/pnas.0900028106.PubMedPubMed CentralView ArticleGoogle Scholar
- Ji Z, Tian B: Reprogramming of 3' untranslated regions of mRNAs by alternative polyadenylation in generation of pluripotent stem cells from different cell types. PLoS One. 2009, 4: e8419-10.1371/journal.pone.0008419.PubMedPubMed CentralView ArticleGoogle Scholar
- Mayr C, Bartel DP: Widespread shortening of 3'UTRs by alternative cleavage and polyadenylation activates oncogenes in cancer cells. Cell. 2009, 138: 673-684. 10.1016/j.cell.2009.06.016.PubMedPubMed CentralView ArticleGoogle Scholar
- Tian BPZ, Lee JY: Widespread mRNA polyadenylation events in introns indicate dynamic interplay between polyadenylation and splicing. Genome Res. 2007, 17: 156-165. 10.1101/gr.5532707.PubMedPubMed CentralView ArticleGoogle Scholar
- Kaida D, Berg MG, Younis I, Kasim M, Singh LN, Wan L, Dreyfuss G: U1 snRNP protects pre-mRNAs from premature cleavage and polyadenylation. Nature. 2010, 468: 664-668. 10.1038/nature09479.PubMedPubMed CentralView ArticleGoogle Scholar
- Beck AH, Weng Z, Witten DM, Zhu S, Foley JW, Lacroute P, Smith CL, Tibshirani R, van de Rijn M, Sidow A, West RB: 3'-end sequencing for expression quantification (3SEQ) from archival tumor samples. PLoS One. 2010, 5: e8768-10.1371/journal.pone.0008768.PubMedPubMed CentralView ArticleGoogle Scholar
- Lee JY, Yeh I, Park JY, Tian B: PolyA_DB 2: mRNA polyadenylation sites in vertebrate genes. Nucleic Acids Res. 2007, 35: D165-168. 10.1093/nar/gkl870.PubMedPubMed CentralView ArticleGoogle Scholar
- Voorhoeve PM, Agami R: The tumor-suppressive functions of the human INK4A locus. Cancer Cell. 2003, 4: 311-319. 10.1016/S1535-6108(03)00223-X.PubMedView ArticleGoogle Scholar
- Drost J, Mantovani F, Tocco F, Elkon R, Comel A, Holstege H, Kerkhoven R, Jonkers J, Voorhoeve PM, Agami R, Del Sal G: BRD7 is a candidate tumour suppressor gene required for p53 function. Nat Cell Biol. 2010, 12: 380-389. 10.1038/ncb2038.PubMedView ArticleGoogle Scholar
- DeGregori J, Johnson DG: Distinct and overlapping roles for E2F family members in transcription, proliferation and apoptosis. Curr Mol Med. 2006, 6: 739-748.PubMedGoogle Scholar
- Cao AR, Rabinovich R, Xu M, Xu X, Jin VX, Farnham PJ: Genome-wide analysis of transcription factor E2F1 mutant proteins reveals that N- and C-terminal protein interaction domains do not participate in targeting E2F1 to the human genome. J Biol Chem. 2011, 286: 11985-11996. 10.1074/jbc.M110.217158.PubMedPubMed CentralView ArticleGoogle Scholar
- Lee BK, Bhinge AA, Iyer VR: Wide-ranging functions of E2F4 in transcriptional activation and repression revealed by genome-wide analysis. Nucleic Acids Res. 2011, 39: 3558-3573. 10.1093/nar/gkq1313.PubMedPubMed CentralView ArticleGoogle Scholar
- Leone G, DeGregori J, Yan Z, Jakoi L, Ishida S, Williams RS, Nevins JR: E2F3 activity is regulated during the cell cycle and is required for the induction of S phase. Genes Dev. 1998, 12: 2120-2130. 10.1101/gad.12.14.2120.PubMedPubMed CentralView ArticleGoogle Scholar
- Ji Z, Lee JY, Pan Z, Jiang B, Tian B: Progressive lengthening of 3' untranslated regions of mRNAs by alternative polyadenylation during mouse embryonic development. Proc Natl Acad Sci USA. 2009, 106: 7028-7033. 10.1073/pnas.0900028106.PubMedPubMed CentralView ArticleGoogle Scholar
- Jenal M, Elkon R, Loayza-Puch F, van Haaften G, Kuhn U, Menzies FM, Vrielink JA, Bos AJ, Drost J, Rooijers K, Rubinsztein DC, Agami R: The poly(a)-binding protein nuclear 1 suppresses alternative cleavage and polyadenylation sites. Cell. 2012, 149: 538-553. 10.1016/j.cell.2012.03.022.PubMedView ArticleGoogle Scholar
- Langmead B, Trapnell C, Pop M, Salzberg SL: Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 2009, 10: R25-10.1186/gb-2009-10-3-r25.PubMedPubMed CentralView ArticleGoogle Scholar
- Rhead B, Karolchik D, Kuhn RM, Hinrichs AS, Zweig AS, Fujita PA, Diekhans M, Smith KE, Rosenbloom KR, Raney BJ, Pohl A, Pheasant M, Meyer LR, Learned K, Hsu F, Hillman-Jackson J, Harte RA, Giardine B, Dreszer TR, Clawson H, Barber GP, Haussler D, Kent WJ: The UCSC Genome Browser database: update 2010. Nucleic Acids Res. 2010, 38: D613-619. 10.1093/nar/gkp939.PubMedPubMed CentralView ArticleGoogle Scholar
- Eden E, Navon R, Steinfeld I, Lipson D, Yakhini Z: GOrilla: a tool for discovery and visualization of enriched GO terms in ranked gene lists. BMC Bioinformatics. 2009, 10: 48-10.1186/1471-2105-10-48.PubMedPubMed CentralView ArticleGoogle Scholar
- Woollard PM: Asking complex questions of the genome without programming. Methods Mol Biol. 2010, 628: 39-52. 10.1007/978-1-60327-367-1_3.PubMedView ArticleGoogle Scholar
- Linhart C, Halperin Y, Shamir R: Transcription factor and microRNA motif discovery: the Amadeus platform and a compendium of metazoan target sets. Genome Res. 2008, 18: 1180-1189. 10.1101/gr.076117.108.PubMedPubMed CentralView ArticleGoogle Scholar
- Elkon R, Vesterman R, Amit N, Ulitsky I, Zohar I, Weisz M, Mass G, Orlev N, Sternberg G, Blekhman R, Assa J, Shiloh Y, Shamir R: SPIKE - a database, visualization and analysis tool of cellular signaling pathways. BMC Bioinformatics. 2008, 9: 110-10.1186/1471-2105-9-110.PubMedPubMed CentralView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.