An efficient method for genome-wide
polyadenylation site mapping and RNA
Stefan Wilkening1, Vicent Pelechano1, Aino I. Ja ¨rvelin1, Manu M. Tekkedil1,
Simon Anders1, Vladimir Benes2and Lars M. Steinmetz1,*
1Genome Biology Unit, European Molecular Biology Laboratory and2Genomics Core Facility,
European Molecular Biology Laboratory, Meyerhofstr. 1, 69117 Heidelberg, Germany
Received September 21, 2012; Revised October 23, 2012; Accepted November 3, 2012
The use of alternative poly(A) sites is common and
affects the post-transcriptional fate of mRNA,
including its stability, subcellular localization and
translation. Here, we present a method to identify
poly(A) sites in a genome-wide and strand-specific
manner. This method, termed 30T-fill, initially fills in
the poly(A) stretch with unlabeled dTTPs, allowing
sequencing to start directly after the poly(A) tail
into the 30-untranslated regions (UTR). Our com-
parative analysis demonstrates that it outperforms
existing protocols in quality and throughput and
accurately quantifies RNA levels as only one read
is produced from each transcript. We use this
method to characterize the diversity of polya-
denylation in Saccharomyces cerevisiae, showing
that alternative RNA molecules are present even in
a genetically identical cell population. Finally, we
observe that overlap of convergent 30-UTRs is
frequent but sharply limited by coding regions, sug-
gesting factors that restrict compression of the
The 30-untranslated regions (UTRs) of mRNAs, located
directly after the stop codon, harbor signals for transcript
stability, localization and translational control (reviewed
in (1) and (2)). The control of mRNA expression by 30-
UTRs is mediated by trans-acting factors, including
RNA-binding proteins and
which interact with cis-regulatory elements within the
30-UTRs (3). More than half of mammalian genes
produce transcripts with 30-UTRs of different lengths (4).
Shorter 30-UTRs have been associated with proliferation
and transformation, which is partially explained by the
exclusion of miRNA-binding sites in the 30-UTRs of
proto-oncogenes (5,6). Shortening of 30-UTRs is also
observed at early stages of development in mice (7), flies
(8) and worms (9) as well as during reprogramming of
stem cells (10). Longer 30-UTRs, on the other hand, tend
to be more frequent in differentiated cells (11,12).
Unicellular eukaryotes such as yeast have also been
shown to use different poly(A) sites in response to stress
conditions (13). To understand
underlying alternative poly(A) site usage and its functional
significance, there is a need for tools that accurately and
efficiently map these sites on a genome-wide scale.
With the emergence of next-generation sequencing
technologies, it has become possible to sequence entire
transcriptomes within days. However, standard mRNA
sequencing (RNA-Seq) is inefficient for poly(A) site
mapping: only a small fraction of sequencing reads
contain poly(A) tails, making it difficult to distinguish
transcript end isoforms. Therefore, protocols have been
developed to enrich for transcript ends of poly(A) RNA
prior to RNA-Seq and/or to identify poly(A) sites (13–20).
However, efficient poly(A) site identification remains
technologically challenging because in many protocols, a
substantial fraction of reads either does not reach the
poly(A) site or is of low quality and must be discarded.
In protocols where sequencing proceeds from within the
transcript toward the poly(A) tail (13,14,16,20), the
readout strongly depends on a stringent size selection to
ensure that each read contains enough of the 30-UTR to
map to the genome while still reaching the poly(A)
tail. Other Illumina-based protocols (17,19), in which
sequencing starts at the 30-end and reads through the
(Supplementary FigureS1). This
is likely due
*To whom correspondence should be addressed. Tel: +49 6221 387389; Fax: +49 6221 387518; Email: firstname.lastname@example.org
The authors wish it to be known that, in their opinion, the first three authors should be regarded as joint First Authors.
Nucleic Acids Research, 2013, 1–8
? The Author(s) 2013. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/3.0/), which
permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact
Nucleic Acids Research Advance Access published January 7, 2013
by guest on January 12, 2013
desynchronization of base incorporation caused by poly-
merase slippage during clustering and sequencing or
mispriming of the sequencing oligo. To overcome this,
the poly(A) tail can be shortened (18). However, even a
few remaining nucleotides of the poly(A) tail can com-
promise the delineation of clusters by the sequencing
software (18). Furthermore, performing the poly(A)
shortening directly on the RNA could be damaging.
Therefore, we have developed a poly(A) site mapping
protocol that circumvents reading through the poly(A)
tail, termed ‘30T-fill’. In this protocol, the poly(A)
stretch is filled in with dTTPs before the sequencing
reaction starts (Figure 1). This enables sequencing to
start immediately after the poly(A) tail. After assessing
the accuracyof this approach
comparing it with alternative methods, we analyzed
poly(A) sites in the yeast transcriptome and discovered
mRNA isoforms is the rule rather than the exception. In
addition, we provide new insight into the genome-wide
organization of poly(A) sites in Saccharomyces cerevisiae,
showing that antisense poly(A) sites are sharply limited by
the presence of the coding region, but not by the length of
MATERIALS AND METHODS
Yeast RNA isolation
We grew S. cerevisiae strain SLS045 (S288c background)
(21) to mid-log phase (OD ?1) using either YPD (1%
yeast extract, 2% peptone and 1% glucose) or YPGal
(1% yeast extract, 2% peptone and 1% galactose). Total
RNA was isolated by a standard hot phenol method and
treated with RNase-free DNaseI using Turbo DNA-free
kit (Ambion). To each 600mg of DNaseI-treated total
RNA, 1.36ng pGIBS-LYS, 3.6ng pGIBS-PHE and
10.7ng pGIBS-THR polyadenylated in vitro transcripts
(IVTs) were added as external controls (ATCC 87482,
87483 and 87484, respectively).
For the results presented, 10mg of total RNA was used as
starting material. This amount could be reduced to 500ng
without a significant loss in quality (results not shown).
The RNA was fragmented by incubating the samples at
80?C for 5min in the presence of RNA fragmentation
buffer (40mM Tris-acetate, pH 8.1, 100mM KOAc and
30mM MgOAc). The fragmented RNA was purified using
1.5? Ampure XP Beads (Beckman Coulter Genomics)
and eluted in 12.8ml elution buffer (EB) (10mM Tris–
HCl, pH 8). For retrotranscription, 11.2ml of the eluted
RNA was mixedwith1ml
P5_dT16VN (1mM; Supplementary Table S3) and 1ml of
10mM dNTPs. The samples were incubated at 65?C for
5min and transferred to ice. Four microliters of 5?
first-strand buffer (Invitrogen), 2ml DTT 0.1M, 0.32ml
actinomycin D (1.25mg/ml) and 0.5ml RNasin plus
RNase inhibitor (Promega) were added to each sample,
and samples were incubated at 42?C for 2min to
minimize possible mispriming. Following this, 0.5ml
Superscript II reverse transcriptase (200U/ml; Invitrogen)
was used for retrotranscription (Figure 1a). The reaction
was performed at 42?C for 50min and inactivated at 70?C
for 15min. The samples were purified using 1.5? of
Ampure XP beads and eluted in 40ml EB. For producing
the second cDNA strand, 40ml of sample was mixed with
5ml of 10? DNA polymerase buffer (Fermentas), 2.5ml of
dNTPs (10mM), 0.5ml of RNaseH (5U/ml; NEB) and 2ml
of DNA polymerase I (10U/ml; Fermentas). The samples
were incubated at 16?C for 2.5h, purified with 0.8?
Ampure XP beads and eluted in 20ml EB. Twenty micro-
liters of Dynabeads M-280 Streptavidin (Invitrogen) were
washed two times with 200ml 1? B&W buffer (5mM Tris–
HCl, pH 7.5, 0.5mM ethylenediaminetetraacetic acid
(EDTA) and 1M NaCl) and resuspended in 20ml of 2?
B&W buffer. Twenty microliters of the double-stranded
cDNA sample was bound to the 20ml of Dynabeads by
mixing them for 15min at 25?C. The beads were washed
twice with 200ml of 1? B&W buffer, once with 200ml EB
and resuspended in 21.25ml EB. 2.5ml end repair buffer
and 1.25ml end repair enzyme mix (NEBNext DNA
Sample Prep Master Mix Set 1, NEB) were added and
the samples incubated at 20?C for 30min. The beads
were washed twice with 200ml of 1? B&W buffer, once
with 200ml EB and resuspended in 21ml EB. 2.5ml dA
tailing buffer (10? NEBuffer 2 from NEB and 0.2mM
dATP) and 1.5ml Klenow Fragment (30!50exo–) 5U/ml
Figure 1. 30T-fill method overview. (a) Schematic of the 30T-fill protocol. Fragmented RNA is reverse transcribed with an oligo(dT) primer coupled
to adapter PE 1.0 (orange) and biotin (B). After second strand synthesis, fragments are captured on beads and the barcoded adapter PE 2.0 (brown)
is ligated to the bound fragment. (b) On the cluster station, the poly(A) tail is filled in with complementary, unlabeled dTTPs. Sequencing starts
directly at the end of the 30-UTR (‘sequence 1’ in panel (a)). For multiplexing, the barcode is read by paired-end sequencing (‘sequence 2’).
2 Nucleic AcidsResearch, 2013
by guest on January 12, 2013
(NEB) were added and the samples incubated at 37?C for
30min. The beads were washed twice with 200ml 1? B&W
buffer, once with 200ml EB and resuspended in 8ml EB.
12.5ml 2? Quick ligation buffer (NEB), 2ml P7_T1_Mpx
linker (2.5mM; Supplementary Table S3) and 2ml T4
DNA ligase were added (2000U/ml; NEB). The samples
were incubated while shaking for 15min at 20?C to ligate
the adapters. The beads were washed four times with
200ml 1? B&W buffer, once with 200ml EB and resus-
pended in 50ml EB. Enrichment polymerase chain
reaction (PCR) was performed using 24ml of beads, 25ml
Phusion Master Mix 2? (NEB) and 0.5ml each of oligos
PE1.0 and PE2.0 (10mM; Illumina). The PCR program
was 30s at 96?C, 18 cycles of (10s at 96?C, 10s at 65?C
and 10s at 72?C) and 5min at 72?C. The PCR product
was purified with 1.8? Ampure XP Beads. 300bp libraries
were size selected using e-Gel 2% SizeSelect (Invitrogen).
The 30T-fill reaction
The final libraries were loaded into the cluster station
(cBot, Illumina) and the priming buffer was exchanged
for the T-fill solution: 101ml water, 20ml Phusion buffer
HF (5?) (NEB), 3ml dTTPs (10mM), 0.8ml genomic
DNA Sequencing primer V2 (100mM) (Illumina), 3ml
non-hot start Phusion polymerase (2U/ml, NEB) and
2ml Taq polymerase E (5U/ml; Genaxxon). The addition
of the latter Taq polymerase was crucial for a complete
T-fill. During the 5min incubation at 60?C (standard cBot
step for primer annealing), the primer aligns and the
poly(A) stretch is filled in with dTTPs (Figure 1b). After
clustering, the samples were sequenced on a HiSeq 2000
Read pre-processing and alignment
The python package HTSeq (Anders,S. http://www-
huber.embl.de/users/anders/HTSeq/) was used for pro-
cessing of the sequencing data. De-multiplexed samples
were trimmed to remove the poly(A) tail and adapter se-
quences before alignment. Read alignment was performed
using the GSNAP aligner (22). As a reference genome, we
used the S. cerevisiae S288c genome (SGD R63) combined
with the three IVTs from Bacillus subtilis used as spike-in
controls in this study. Reads with low mapping quality
(<30) or which were composed of >80% As or Ts were
filtered out. We also filtered out potential internal priming
events by discarding reads mapping to a genomic region
with eight or more As downstream, or high A/T content
(27 out of 30 preceding bases).
Assessing signal enrichment at transcript ends
The S. cerevisiae transcriptome annotation of Xu et al.
(21) was updated to reflect the genome sequence version
used. Aligned reads from all tested protocols were mapped
to the 7272 full-length annotated transcript features of this
annotation, with either a +200bp extension downstream
of the transcriptional termination site (TTS) to account
for poly(A) sites extending beyond current annotation or
a ±200bp extension from TTS. Reads with ambiguous
assignment (0.8% of assigned reads for 30T-fill) were
assigned to the transcript with the closest proximal 30-end.
Estimating poly(A) site calling accuracy using IVTs
Poly(A) site calls were defined as the read base closest to
the poly(A) tail (e.g. the first base in the 30T-fill protocol).
The accuracy of poly(A) site calling was calculated as the
fraction of reads supporting the known poly(A) site, of
total non-filtered reads mapping to each IVT.
Distribution of poly(A) sites within transcripts
Poly(A) sites were uniquely mapped to the updated tran-
scriptome annotation from Xu et al. (21) with a+200bp
extension downstream of the TTS. Within these features,
poly(A) site counts were grouped to those assigning to
UTRs, coding regions and extension regions for genes.
Only verified, monocistronic genes were considered for
Definition and analysis of high-confidence poly(A) sites
Poly(A) site calls from all the samples were merged to
create a common poly(A) site map, such that non-zero
counts on consecutive bases were merged. High-confi-
dence poly(A) sites were defined as sites detected in all
three replicates of a given condition and with a
minimum of 10 supporting reads in each. For estimating
the number of poly(A) sites per gene, we further selected
only poly(A) sites accounting for at least 10% of the total
expression of that gene. The alternative presence of 14
RNA-binding protein target sequences, obtained from
Supplementary Data S1 from Riordan et al. (23), was
determined for this set of poly(A) sites. The fraction of
reads accounted for by the major 30-isoform for each gene
was calculated for all genes with multiple high-confidence
poly(A) sites, and the presence of antisense expression was
computed using all high-confidence poly(A) sites.
30T-fill and RNA-Seq reads were assigned to the
full-length transcript annotation of Xu et al. (21) as
described above. Differential expression calling was per-
formed using the R/Bioconductor package DESeq (24).
As our goal was to compare the technical suitability of
30T-fill and RNA-Seq for expression quantification, only
technical replicates were used. Expression levels of poly(A)
isoforms detected by 30T-fill were computed by comparing
counts supporting poly(A) sites in the merged poly(A) site
map by different replicates. Changes in poly(A) site
influenced by the growth medium (glucose versus galact-
ose) were assessed by extension of a statistical method
developed for differential exon usage (25).
Comparison to alternative methods
To assess the performance of our 30T-fill approach, we
conducted an extensive comparison with alternative
Figures S2 and S3 and Supplementary Table S1).
In addition to a strand-specific RNA-Seq protocol (26),
with very few reads that map poly(A) sites (Figure 2), we
evaluated four poly(A)-mapping methods, including one
Nucleic Acids Research, 20133
by guest on January 12, 2013
RNA-Seq following enrichment of poly(A)-containing
RNA fragments (as in (13)), one which sequences into
the poly(A) tail called ‘30Internal’ (27), similar to (14,16),
one which cleaves off the poly(A) tail called ‘30BpmI’
(similar principle to Jan et al. (18), but cleavage is per-
formed on cDNA rather than RNA) and the 30T-fill
method which fills up the poly(A) stretch before
sequencing. Details of the compared methods can be
found in Supplementary Methods and Supplementary
Figures S2 and S3. Our 30T-fill protocol (Figure 1)
avoids delicateRNA processing
transcribing the RNA directly after fragmentation.
Binding the fragments to magnetic beads early in the
protocol allows for automation and easy scaling up to
sequencing proceeds from the 50-direction into the
poly(A) tail, this protocol offers several advantages: size
selection is not critical, no trimming is necessary and both
forward and reverse reads are informative. Importantly,
the 30T-fill method is the most easily performed of all
protocols developed thus far. The 30T-fill protocol is also
the most efficient of all the protocols we tested, with all
mapped reads identifying poly(A) sites with high confi-
dence (74% of raw reads, Supplementary Table S1).
Accuracy of poly(A) site mapping
To estimate the accuracy of the 30T-fill protocol for
mapping the precise poly(A) sites, three IVTs (1–2kb
each) from B. subtilis with a DNA-encoded poly(A) tail
were spiked into the starting RNA material. On average,
83% of 30T-fill reads mapping to these transcripts
identified the exact poly(A) site and 97% within a
window of ±2nt. For the Helicos technology, which
applies a similar approach to fill the poly(A) tail (15),
but used much shorter IVTs (40nt), 65.6% of the IVT
reads were reported to map the exact poly(A) site and
Compared with other methods, the proportion of 30T-fill
reads that uniquely mapped to the yeast 30-UTRs was
highest in the T-fill method, which confirms its specificity
for identifying the correct poly(A) sites.
To benchmark the RNA quantification efficiency of our
approach, we compared the results obtained with our
30T-fill method to those from standard, strand-specific
RNA-Seq. The technical reproducibility of the 30T-fill
method is high and at similar levels to RNA-Seq
(Figure 3a and b). Importantly, 30T-fill and RNA-Seq
(Figure 3c). Thus, the 30T-fill approach can also be used
to identify genes that are differentially expressed between
conditions (Supplementary Figure S4). However, in
contrast to RNA-Seq experiments, in which more reads
are produced from long transcripts, 30T-fill only produces
one read per transcript, resulting in a size-independent
representation of transcript abundance (Figure 3d and e).
This represents a major advantage over RNA-Seq, in
which short transcripts might be underrepresented when
assessing differential expression. Most importantly, a
poly(A) mapping approach allows for discrimination of
transcript end isoforms, which are mostly overlooked in
RNA-Seq. The number of reads with the same poly(A)
Figure 2. Example reads from 30T-fill, RNA-Seq and tiling array (21) in yeast grown in rich glucose media. Only a small fraction of the tiling array
and total RNA-Seq signals arises from 30-ends of transcripts. With the 30T-fill protocol many alternative (green), internal (purple) and overlapping
poly(A) sites (orange) were detected. Visualization was performed with the Integrative Genomics Viewer (IGV) (28).
4 Nucleic AcidsResearch, 2013
by guest on January 12, 2013
position also correlated well between two technical repli-
cates of the T-fill method (Figure 3f).
Poly(A) site organization in the yeast transcriptome
Having established the accuracy of our method, we
collectively analyzed RNA quantification and alternative
30-ends to investigate the nature of poly(A) site organiza-
tion in the yeast transcriptome (excerpt in Figure 2).
We detected 22512 different transcript end isoforms
with high confidence (see ‘Materials and Methods’
section for definition) present in glucose medium alone.
Approximately 43% of all transcripts have multiple
poly(A) sites, and ?700 genes express at least two tran-
script isoforms where the poly(A) sites are at least 100bp
apart. Overall, the major 30-isoform for each gene only
accounts for 67.5% of the reads, indicating that alterna-
tive mRNA molecules are present even in a genetically
identical cell population. In fact, we observe that 13%
of genes have differential isoforms with alternative
presence of RNA-binding protein target sequences (23)
and thus potentially varying posttranscriptional fate.
Although glucose and galactose conditions had remark-
ably similar global poly(A) site profiles, we detected 187
transcripts with differences in poly(A) site usage between
these conditions (Supplementary Table S2). Furthermore,
3.3% of the reads were detected within coding sequences,
indicating internal polyadenylation that could affect the
length of the protein product.
The detected poly(A) sites for coding transcripts
(Figure 4a) yielded well-defined 30-ends that are in good
agreement with the existing transcriptome annotation (21).
A similar pattern was observed for stable unannotated
transcripts (SUTs) (Supplementary Figure S5a), whereas
poly(A) sites for cryptic unstable transcripts (21) were
much less defined (Supplementary Figure S5b), which is
in accordance with their alternative termination pathway
(29). Forty percent of transcripts have antisense tran-
scripts that terminate within their annotated 30-UTRs.
This is more than the 12% (30) and 14% (15) previously
reported and likely due to the high coverage in our
dataset. We found that the antisense poly(A) site signal
peaks ?100bp upstream from the annotated transcription
termination site for coding transcripts (Figure 4a) and
SUTs (Supplementary Figure S5a). Interestingly, the
signal of these antisense peaks declines sharply after the
sense open reading frame (ORF) stop codon (Figure 4b).
This decline, together with the observation that poly(A)
sites accumulate directly upstream of the transcription
Figure 3. Comparison between 30T-fill and RNA-Seq. Reproducibility of transcript quantification was high between technical replicates of 30T-fill (a)
and RNA-Seq (b), as well as between the two methods (c). (d) Since 30T-fill does not display the length bias of RNA-Seq, 30T-fill captures more
shorter transcripts, providing a size-independent estimate of their abundance. (e) The length distribution of differentially expressed genes (in glucose
versus galactose media) in 30T-fill (red) is closer to the global transcript length (dashed line) than transcript length in RNA-Seq (blue). 2441 and 3401
genes were differentially expressed for 30T-fill and RNA-Seq, respectively (adjusted P<0.1). (f) A high reproducibility in mapping individual poly(A)
sites was observed between technical replicates of 30T-fill samples. Data from panels a–c and f correspond to cells grown in galactose.
Nucleic Acids Research, 20135
by guest on January 12, 2013
start site (Figure 4a), suggests that the promoter as well as
the coding sequence limit further compaction of the yeast
As only a very small fraction of reads obtained from a
strand-specific RNA-Seq run contain the poly(A) site
(Figure 2), several methods that aim to identify the
poly(A) sites have been developed, yet to date no com-
parative analysis of these protocols has been performed.
In this study, we tested and compared four methods for
poly(A) site mapping. Being among the most accurate
methods in calling the exact poly(A) site and having
the highest number of reads in annotated 30-UTRs, the
30T-fill outperforms currently available protocols in both
simplicity and efficiency. It avoids delicate RNA process-
ing and is flexible in library size and read length. In
addition to its capacity to distinguish and quantify alter-
native poly(A) site usage, it enables quantification of
RNA expression in a length-independent manner as
only one read is produced per transcript. This method
can easily be extended to non-polyadenylated RNAs (e.g.
bacterial RNA, eukaryotic non-polyadenylated RNAs)
by artificial RNA polyadenylation. The T-fill step has
the potential to improve poly(A) sequencing on other
sequencing platforms, as slippage during poly(A) ampli-
fication is likely to occur on any platform that requires
an amplification step prior to sequencing (e.g. 454, Solid,
Ion Torrent). Also, single cell poly(A) sequencing (31)
where efficiency is crucial for the outcome would
greatly profit from our protocol. One limitation of
(Helicos) (15) (instrument no longer commercially avail-
able) is that it requires amplification of cDNA prior to
sequencing. To circumvent this step, it would in principle
be possible to combine our method with ‘FRT-seq’ (32)
and synthesize the cDNA directly on the flow cell.
In addition to presenting a new poly(A) sequencing
method, our analysis of poly(A) site usage in S. cerevisiae
expands the current view of yeast transcriptome architec-
ture and indicates factors that limit the compression of the
yeast genome. This method constitutes a powerful tool for
future studies of fundamental biological processes, such as
Number of poly(A) sites
Sense poly(A) site
Number of poly(A) sites
Sense poly(A) site
Number of poly(A) sites
Sense poly(A) site
Number of poly(A) sites
Figure 4. Accumulation of poly(A) sites relative to annotated gene features. Poly(A) site positions on sense (upper panels in black) and antisense
strands (lower panels in gray) relative to annotated (a) transcription start sites (left), transcription termination sites (right), (b) ORF starts (left) and
ORF ends (right). Arrows indicate the following: sharp drop-off in poly(A) site abundance of upstream genes just before ORF promoters/tran-
scription start sites (a, left); peak of convergent 30-UTR overlap at 100bp upstream of transcription termination sites (a, right) and sharp drop in
overlap of antisense 30-UTRs beyond ORF stop codons (b, right).
6 Nucleic AcidsResearch, 2013
by guest on January 12, 2013
comparisons of different developmental or disease stages
in higher eukaryotes, in which it can be used to clarify the
role of differential poly(A) site usage and expression level
in regulating cellular function.
Raw and processed data are available from the Gene
Expression Omnibus (GEO, http://www.ncbi.nlm.nih.
gov/geo/, accession number GSE40110) and at our web-
site, http://steinmetzlab.embl.de/polyAmapping/, brows-
able (IGV) files are provided.
Supplementary Data are available at NAR Online:
Supplementary Tables 1–3, Supplementary Figures 1–5
and Supplementary Methods.
We would like to thank Raeka Aiyar, Julien Gagneur,
Nicolas Delhomme, Louise-Amelie Schmitt and Wu Wei
for helpful discussions and EMBL Genomics Core
Facility for technical support.
University of Luxembourg—Institute for Systems Biology
(to L.M.S.) and an EMBO fellowship (to V.P.). Funding
for open access charge: Deutsche Forschungsgemeinschaft
Conflict of interest statement. None declared.
1. Di Giammartino,D.C., Nishida,K. and Manley,J.L. (2011)
Mechanisms and consequences of alternative polyadenylation.
Mol. Cell, 43, 853–866.
2. Lutz,C.S. and Moreira,A. (2011) Alternative mRNA
polyadenylation in eukaryotes: an effective regulator of gene
expression. Wiley Interdiscip. Rev. RNA, 2, 22–31.
3. Kuersten,S. and Goodwin,E.B. (2003) The power of the 30UTR:
translational control and development. Nat. Rev. Genet., 4,
4. Tian,B., Hu,J., Zhang,H. and Lutz,C.S. (2005) A large-scale
analysis of mRNA polyadenylation of human and mouse genes.
Nucleic Acids Res., 33, 201–212.
5. Sandberg,R., Neilson,J.R., Sarma,A., Sharp,P.A. and Burge,C.B.
(2008) Proliferating cells express mRNAs with shortened 30
untranslated regions and fewer microRNA target sites. Science,
6. Mayr,C. and Bartel,D.P. (2009) Widespread shortening of
30UTRs by alternative cleavage and polyadenylation activates
oncogenes in cancer cells. Cell, 138, 673–684.
7. Ji,Z., Lee,J.Y., Pan,Z., Jiang,B. and Tian,B. (2009) Progressive
lengthening of 30untranslated regions of mRNAs by alternative
polyadenylation during mouse embryonic development. Proc. Natl
Acad. Sci. USA, 106, 7028–7033.
8. Thomsen,S., Azzam,G., Kaschula,R., Williams,L.S. and
Alonso,C.R. (2010) Developmental RNA processing of 30UTRs in
Hox mRNAs as a context-dependent mechanism modulating
visibility to microRNAs. Development, 137, 2951–2960.
9. Mangone,M., Manoharan,A.P., Thierry-Mieg,D., Thierry-Mieg,J.,
Han,T., Mackowiak,S.D., Mis,E., Zegar,C., Gutwein,M.R.,
Khivansara,V. et al. (2010) The landscape of C. elegans 30UTRs.
Science, 329, 432–435.
10. Ji,Z. and Tian,B. (2009) Reprogramming of 30untranslated
regions of mRNAs by alternative polyadenylation in generation
of pluripotent stem cells from different cell types. PLoS One, 4,
11. Licatalosi,D.D., Mele,A., Fak,J.J., Ule,J., Kayikci,M., Chi,S.W.,
Clark,T.A., Schweitzer,A.C., Blume,J.E., Wang,X. et al. (2008)
HITS-CLIP yields genome-wide insights into brain alternative
RNA processing. Nature, 456, 464–469.
12. An,J.J., Gharami,K., Liao,G.Y., Woo,N.H., Lau,A.G.,
Vanevski,F., Torre,E.R., Jones,K.R., Feng,Y., Lu,B. et al. (2008)
Distinct role of long 30UTR BDNF mRNA in spine morphology
and synaptic plasticity in hippocampal neurons. Cell, 134, 175–187.
13. Yoon,O.K. and Brem,R.B. (2010) Noncanonical transcript forms
in yeast and their regulation during environmental stress. RNA,
14. Fox-Walsh,K., Davis-Turak,J., Zhou,Y., Li,H. and Fu,X.D.
(2011) A multiplex RNA-seq strategy to profile poly(A+) RNA:
application to analysis of transcription response and 30end
formation. Genomics, 98, 266–271.
15. Ozsolak,F., Kapranov,P., Foissac,S., Kim,S.W., Fishilevich,E.,
Monaghan,A.P., John,B. and Milos,P.M. (2010) Comprehensive
polyadenylation site maps in yeast and human reveal pervasive
alternative polyadenylation. Cell, 143, 1018–1029.
16. Beck,A.H., Weng,Z., Witten,D.M., Zhu,S., Foley,J.W.,
Lacroute,P., Smith,C.L., Tibshirani,R., van de Rijn,M., Sidow,A.
et al. (2010) 30-end sequencing for expression quantification
(3SEQ) from archival tumor samples. PLoS One, 5, e8768.
17. Shepard,P.J., Choi,E.A., Lu,J., Flanagan,L.A., Hertel,K.J. and
Shi,Y. (2011) Complex and dynamic landscape of RNA
polyadenylation revealed by PAS-Seq. RNA, 17, 761–772.
18. Jan,C.H., Friedman,R.C., Ruby,J.G. and Bartel,D.P. (2011)
Formation, regulation and evolution of Caenorhabditis elegans
30UTRs. Nature, 469, 97–101.
19. Derti,A., Garrett-Engele,P., Macisaac,K.D., Stevens,R.C.,
Sriram,S., Chen,R., Rohl,C.A., Johnson,J.M. and Babak,T. (2012)
A quantitative atlas of polyadenylation in five mammals. Genome
Res., 22, 1173–1183.
20. Elkon,R., Drost,J., van Haaften,G., Jenal,M., Schrier,M.,
Vrielink,J.A. and Agami,R. (2012) E2F mediates enhanced
alternative polyadenylation in proliferation. Genome Biol., 13, R59.
21. Xu,Z., Wei,W., Gagneur,J., Perocchi,F., Clauder-Munster,S.,
Camblong,J., Guffanti,E., Stutz,F., Huber,W. and Steinmetz,L.M.
(2009) Bidirectional promoters generate pervasive transcription in
yeast. Nature, 457, 1033–1037.
22. Wu,T.D. and Nacu,S. (2010) Fast and SNP-tolerant detection of
complex variants and splicing in short reads. Bioinformatics, 26,
23. Riordan,D.P., Herschlag,D. and Brown,P.O. (2011) Identification
of RNA recognition elements in the Saccharomyces cerevisiae
transcriptome. Nucleic Acids Res., 39, 1501–1509.
24. Anders,S. and Huber,W. (2010) Differential expression analysis
for sequence count data. Genome Biol., 11, R106.
25. Anders,S., Reyes,A. and Huber,W. (2012) Detecting differential
usage of exons from RNA-seq data. Genome Res., 22,
26. Parkhomchuk,D., Borodina,T., Amstislavskiy,V., Banaru,M.,
Hallen,L., Krobitsch,S., Lehrach,H. and Soldatov,A. (2009)
Transcriptome analysis by strand-specific sequencing of
complementary DNA. Nucleic Acids Res., 37, e123.
27. Pelechano,V., Wilkening,S., Ja ¨ rvelin,A.I., Tekkedil,M.M. and
Steinmetz,L.M. (2012) Genome wide Analyses of Chromatin and
Transcripts. In: Carl Wu,C.D.A. (ed.), Nucleosomes, Histones &
Chromatin, Part B, Vol. 513. Academic Press, San Diego, CA,
28. Robinson,J.T., Thorvaldsdottir,H., Winckler,W., Guttman,M.,
Lander,E.S., Getz,G. and Mesirov,J.P. (2011) Integrative
genomics viewer. Nat. Biotechnol., 29, 24–26.
Nucleic Acids Research, 20137
by guest on January 12, 2013
29. Arigo,J.T., Eyler,D.E., Carroll,K.L. and Corden,J.L. (2006) Download full-text
Termination of cryptic unstable transcripts is directed by yeast
RNA-binding proteins Nrd1 and Nab3. Mol. Cell, 23, 841–851.
30. Nagalakshmi,U., Wang,Z., Waern,K., Shou,C., Raha,D.,
Gerstein,M. and Snyder,M. (2008) The transcriptional landscape
of the yeast genome defined by RNA sequencing. Science, 320,
31. Hashimshony,T., Wagner,F., Sher,N. and Yanai,I. (2012)
CEL-Seq: single-cell RNA-Seq by multiplexed linear amplification.
Cell Rep., 2, 666–673.
32. Mamanova,L. and Turner,D.J. (2011) Low-bias, strand-specific
transcriptome Illumina sequencing by on-flowcell reverse
transcription (FRT-seq). Nat. Protoc., 6, 1736–1747.
8 Nucleic AcidsResearch, 2013
by guest on January 12, 2013