Nielsen H. RNA: Methods and Protocols
Подождите немного. Документ загружается.
4 Nielsen
prokaryotes), the functional transcripts are co-linear with the
genome. In other organisms (typical of eukaryotes), transcripts
are made as larger precursors that are subsequently processed into
smaller mature transcripts. The understanding of the complexity
of the transcriptome and the view on the transcriptional land-
scape is rapidly changing these years. The central observation is
that most, if not all, of the genome is transcribed such that mul-
tiple, overlapping transcripts derived from both strands of DNA
are being produced throughout the genome. This realization is
driven by application of new experimental methods and large-
scale bioinformatics analysis of gene expression. The new view
calls for a revision of central concepts in molecular biology, such
as the gene, the promoter (driving transcription initiation), and
the transcription unit. An introduction to the new discoveries is
provided in the following.
In the traditional approach to studying gene expr ession, a
complex RNA sample was fractionated by gel electrophoresis,
blotted onto a membrane, and analyzed by hybridization analysis
using labeled probes representing individual genes. This yielded
relatively little information and was much improved with the
introduction of array analysis that can be thought of as an “inver-
sion” of the experimental strategy. Here, large collections of
probes fixed to a membrane or a chip are hybridized with labeled,
complex RNA. This type of array analysis is biased in the sense
that the analysis is limited by the selection of probes on the
membrane or chip. These would typically be chosen to repre-
sent already known transcripts. To overcome this, an unbiased
way of interrogating the transcriptional complexity was intro-
duced with the tiling array analysis in which overlapping oligonu-
cleotides covering an entire region of the genome is bound on
the chip. Another unbiased method that was introduced more
recently is deep sequencing in which an RNA sample is copied
into DNA molecules that are sequenced in parallel. New sequenc-
ing technologies allows for the analysis of millions of parallel
sequencing reads. The breakthrough in understanding of the
transcriptional complexity came first from application of tiling
arrays to the mouse and human genome (1, 2). A rough calcula-
tion of the transcriptional complexity prior to this would be that
10–20% of the genome (one DNA strand only) was transcribed
into discrete RNA molecules. This was b ased on an estimated
21–23,000 protein coding genes that require transcription of
1.2% of the genome for specifying the amino acids and consider-
ing that mRNAs initially are made as intron-containing precursors
that are processed into mature mRNAs. However, the tiling array
studies showed that >90% of both strands of the genome are tran-
scribed into multiple, o verlapping RNA molecules. Furthermore,
this was based on analysis of RNA from selected cells and did
not take into account the complexity of transcripts when all types
The Transcriptional Landscape 5
of cells at all stages in development were considered. The major-
ity of the transcriptional output is non-polyadenylated transcripts
derived from what was previously considered intergenic regions.
The nuclear complexity exceeds that of the cytosolic compart-
ment but the new transcripts are not confined to the nucleus
and both polyadenylated and non-polyadenylated unannotated
transcripts are found in both compartments (3, 4). The most
updated annotation of the human genome is probably provided
through the ENCODE project (Encyclopedia of DNA elements)
(http://www.genome.gov/10005107).
Much of the newly discovered transcriptional complexity is
low abundance transcripts with steady-state levels of less than 1–
10 copies per cell (5). Transcripts at the lower end of this range
could in principle be due to transcriptional “noise” resulting from
stochastic transcription. However, care should be taken not to
confuse steady-state levels with transcriptional activity and disre-
gard low abundance transcripts. Within cells, RNA surveillance
systems rapidly remove some transcripts. In cells that have been
crippled in their RNA surveillance systems (e.g., by elimination of
nuclear exosome activity), many transcripts have increased steady-
state levels that are comparable to those of known functional
RNAs. In fact there are examples of regulatory systems that are
dependent on rapid turnover of regulatory RNAs (6, 7). In sup-
port of the idea that many RNAs are transcribed at high frequency
but rapidly turn ed over, analysis of localization of RNA poly-
merase II by chromatin precipitation experiments show similar
occupancy at these sites and those of RNAs with high steady-state
levels. One other concern is that most of the transcriptional com-
plexity is poorly conserved in sequence when human and mouse is
compared ( 2, 8, 9). A classical criterion for functional importance
is conservation at the sequence or RNA structure level. However,
there is an increasing emphasis on lineage-specific traits, not least
as part of the search of what defines a human.
It is now widely accepted that the genome is pervasively
transcribed. This raises questions on the functions of the tran-
scripts that are pr oduced, most of which are not understood
at present. First, transcription per se rather than the RNA
being produced could be functional in many cases. Transcrip-
tion impacts the structure of chromatin and could be required
to keep the chromatin accessible to regulatory factors. Sec-
ond, many new functional RNAs have indeed been identi-
fied. The current estimate of RNA genes is in the range 4–
5,000 compared to 21–23,000 protein coding genes and is
rapidly increasing. New families of functional RNA species and
RNA structural domains are annotated in databases such as
Rfam (http://www.sanger.ac.uk/Software/Rfam/)(10)and
described through the RNA WikiProject (11). An interim term
“ncRNA” is frequently used to classify these RNAs simply to state
6 Nielsen
that they are non-protein coding RNAs. The term is not very
useful when applied both to small RNAs with distinct functions
such as the modification guide RNAs or the regulatory miRNAs
and to long RNA molecules that are transcribed by RNA poly-
merase II, capped, polyadenylated, and spliced but not encod-
ing a protein. Incidentally, the discovery of the latter group of
RNAs is one of the most important new discoveries. In the days
of describing gene expression by construction of cDNA libraries
of expressed sequence tags (EST libraries), many cDNAs were
found not to have an open reading frame for protein coding. The
cDNAs were made by priming at the polyA tail with oligodT,
and it was believed that these cDNAs were copied from mRNA
with very long 3
UTRs (untranslated regions). When methods
became available for synthesis of full length cDNAs, it became
clear that these RNAs were devoid of open reading frames. Sur-
prisingly, this class of RNAs may constitute half of the RNA poly-
merase II output (1, 12, 13). Only few of these have been thor-
oughly analyzed for function. As an example, NRON is an alter-
natively spliced RNA that is found as transcripts ranging in size
from 0.8 to 3.7 kb. It is bound to 11 different proteins and the
complex functions as a specific regulator of transcription factor
NFAT nuclear trafficking (14). There are other examples of ncR-
NAs w ith similar functions and it is quite possible, given the num-
ber of such RNAs, that this is a general and as yet unanticipated
layer of gene regulation. A third type of function is related to
the abundance of sense–antisense pairs of transcripts. Depending
on the method applied, the occurrence of antisense transcripts
in mammalian genomes have been estimated from a few percent
to 72% of the corresponding sense transcripts (12, 15). Some
early studies based on cDNA probes may have suffered from the
propensity of reverse transcriptase to synthesize a second strand
cDNA using the first strand as primer and template by a loop-back
mechanism. However, this mechanism can be inhibited by actino-
mycin D (16) and some data sets have been corrected accordingly.
The extent of antisense regulation and the mechanisms (e.g., tran-
scriptional interference or production of double-stranded RNA)
remains to be e xplored but there are several well-documented
cases of functional antisense RNAs (17–19). Finally, a fourth type
of unanticipated function of transcripts is related to the recent dis-
covery of new promoter-associated transcripts. Their description
requires a brief introduction to the eukaryotic RNA polymerase
II promoter.
2. The Promoter
The traditional view of transcription is that it is driven from
a promoter that recruits the RNA polymerase directly or indi-
rectly through the action of transcription factors and furthermore
The Transcriptional Landscape 7
Fig. 1.1. a Two main types of promoters in the human genome. The sharp type of
promoters (10–20%) initiates transcription at a well-defined site and typically has TATA-
boxes and initiator (INR) elements. The broad type of promoters (>50%) initiates over a
few hundred base pairs and are typically characterized by the presence of CpG-islands.
b A close-up of the promoter region showing short bi-directional transcripts (TSS-a)
originating from the nucleosome-free region (NFR) comprising the transcription start site
(
broken arrow
). Unstable promoter proximal transcripts (PROMPTS) in both orientations
and peaking in the region –500 to –2,500 are revealed in cells that are deficient in RNA
turnover.
determines the direction of transcription. The prototype of pro-
moters is one that harbors a TATA-box located approximately 30
base pairs upstream of the start site and initiates transcription at
a well-defined site (sharp type of promoters; see Fig. 1.1a). This
turns out not to be representative of promoters in the human
genome – even in protein coding genes. First of all, only 10–20%
of these genes have a TATA-box. The remainder of the genes
recruits the polymerase in a TATA-box independent manner, pri-
marily (>50%) in regions rich in 5
CpG sequences (20). Sec-
ond, transcription initiation does not take place at a specific site
in most genes but is scattered over several hundred base pairs.
This phenomenon was revealed by genome-wide cap analysis of
gene expression (CAGE) (21). In this method, RNAs carrying an
m
7
G-cap, the hallmark of a 5
end representing transcription initi-
ation, are affinity purified and used as templates for making small
DNA fragments that are sequenced and mapped on the genome.
From this type of analysis it appeared that both TATA-box
containing and TATA-less genes used both types of initiation
but that the scattered initiation (broad type of promoters) was
8 Nielsen
predominantly associated with the TATA-less promoters. An even
bigger surprise was recently revealed from several studies of pro-
moter associated transcripts (see Fig. 1.1b). One study based on
deep sequencing of human RNAs smaller than 200 nt disclosed
classes of RNA mapping to the 5
(promoter-associated small
RNAs; PASRs) and 3
ends (terminator-associated small RNAs;
TASRs) of transcription units (22, 23). The sites of origin coin-
cide with regions that frequently are nucleosome free and the
PASR align with CAGE tags indicating that they are capped.
Another sequencing study of murine small RNAs r eported a
class of uncapped 20–90 nt TSS-a RNAs (transcription start site-
associated RNAs) (24). These RNAs flank the transcription star t
site with peaks of antisense at –250, and sense at +50, respectively.
These transcripts are like PASR and TASR sufficiently abundant
to be detected by northern blotting. Transcripts in both direc-
tions co-localized with chromatin markers of transcription ini-
tiation (RNA polymerase II and H3K4-trimethylated histones)
but a marker of transcription elongation (H3K79-dimethylated
histones) was only found downstream of the transcription start
site, i.e. in association with synthesis of sense transcripts. Finally, a
study of nascent transcripts in human cells found promoter proxi-
mal transcripts (nuclear run-on RNAs; NRO-RNAs) flanking the
transcription start site in 30% of the genes (25). The peaks of anti-
sense and sense transcripts mapped to the same positions as in the
above mentioned mouse study. The three studies together shows
that transcription initiation at the eukaryotic RNA polymerase
II promoter is bi-directional. The initiation mechanism and the
functional implications of this phenomenon are not known. It is
also not known why synthesis of long transcripts eventually takes
place in the sense direction only and whether this phenomenon is
related to conversion of polymerases from pausing to processive
mode as seen in the regulation of many genes. A completely unan-
ticipated class of transcripts was recently found by tiling array anal-
ysis of exosome-depleted human cells, which are cells that are defi-
cient in RNA turnover (26). These promoter upstream transcripts
(PROMPTS) are variable length, polyadenylated transcripts that
are transcribed from both strands and show overlapping distri-
bution in the upstream –500 to –2,500 region. They are corre-
lated with active genes and their presence is dependent on the
nearby promoter. Based on their instability, they may be related
to cryptic unstable transcripts (CUTs) in yeast. These are pre-
dominantly found as divergent transcripts from promoter regions
of bona fide genes and are believed to have regulatory functions
(6, 7). The promoter-associated transcripts show that transcrip-
tion initiation is a complex phenomenon. This may not come as
a surprise given that promoter elements and transcription factor
binding sequences are short sequences that occur frequently in
the genome. Thus, transcription initiation can be viewed a dual
The Transcriptional Landscape 9
task of recruitment of the RNA polymerase and suppression
of initiation at illegitimate start sites, primarily by formation of
inhibitory chromatin structure (27).
3. The
Transcription Unit
The transcription unit has been a useful concept because meth-
ods are known for mapping the ends of a transcript. The site of
polyadenylation is used to define the 3
end of transcription units
because the actual site of termination of transcription generally is
unknown and considered to be of little interest. The clear defini-
tion of a transcription unit contrasts the definition of a gene that
originally was introduced as a physical entity corresponding to an
observable phenotype. In many genes, distant sequence elements
influence the expression and thus the phenotype in subtle ways
and make it virtually impossible to delineate the gene as a physi-
cal entity. In the traditional view, transcription units have distinct
boundaries. Although this is still the case in a strict sense, the pic-
ture is blurred by the observation from genome-wide studies that
transcripts from a genomic region generally have multiple 5
and
3
ends. In a specialized type of cloning procedure, short tags rep-
resenting the extreme 5
and 3
ends of a transcript are joined as
“ditags” and sequenced. Analysis of such ditags revealed that the
large majority of transcriptional units have alternative transcrip-
tion start sites and polyadenylation sites with an average of 1.32
5
start/3
end and 1.83 3
end/5
start (1). This result can to a
large extent be explained by the existence of broad type promot-
ers (see Fig. 1.1a). Adding to the picture of 5
end heterogene-
ity is the observation from the ENCODE project that >65% of
the genes were alternatively transcribed from previously unnoted
upstream promoters that on the average were located more than
100 kb upstream (2, 28). These transcripts typically were spliced
to incorporate upstream exons that extend the 5
UTR and thus
the regulatory potential of the mRNA but could also add pr otein-
coding exons to the mRNA. At the 3
end, a recent mouse study
showed that the 3
UTRs of mRNAs tend to be longer due to
alternative polyadenylation at later stages in development (29).
Longer 3
UTRs increase the potential for post-transcriptional
regulation and the observation emphasizes the dynamical nature
of the transcription unit. Transcription termination is generally
not considered important in gene expression studies. It is worth
to note that a genome-wide study of nascent transcripts showed
that the RNA polymerase on average travels 10 kb beyond the
polyadenylation site (30). This additional RNA is normally turned
over rapidly, but at least the termination-associated transcription
10 Nielsen
activity has the potential to influence the expression of nearby
genes by transcriptional interference.
The transcriptional activity of a hypothetical region of the
human genome encoding a protein is depicted in Fig. 1.2.To
simplify the picture, the transcription unit is drawn with a sharp
type of promoter (see Fig. 1.1a). The figure is drawn to scale
to represent a unit that is average in most respects. The basic
primary transcript is composed of seven exons (average 200 nt)
interrupted by six introns. Overall, protein coding exons and
introns make up 1.2 and 30%, respectively, of the genome and
this ratio was used in the figure. In the default mature mRNA,
the 5
UTR will be 150 nt and the 3
UTR 520 nt with a
post-transcriptionally added polyA tail of 150–250 residues. The
resulting 2.3-kb mRNA encodes a protein of 476 amino acids.
Using this basic transcription unit as the starting point several
variant and additional transcripts are noted from the region:
Fig. 1.2. Transcription map of a generalized region of the human genome. Regular protein coding mRNAs are tran-
scribed from the main promoter in the region (
broken arrow
) or from a far upstream promoter belonging to a different
transcriptional region. Some transcripts are 3
extended due to suppression of the default polyadenylation site and use of
alternative sites. Small transcripts are generated in both orientations from the 5
end (TSS-a, PASR, NRO RNA) and 3
end
(TASR) and overlapping, unstable transcripts are found upstream of the main promoter (PROMPTS). Small nucleolar RNAs
(snoRNAs) and microRNAs (miRNAs) are made by processing of intron RNA. Non-coding RNAs (ncRNAs) are produced in
both sense and antisense orientations.
– Variant transcripts are initiated from far upstr eam promot-
ers that in some cases are associated with other transcription
units. These transcripts results in incorporation of additional
exons derived from upstream transcription units or intergenic
region into the mature mRNA.
– Other variant transcripts ar e extended to downstream
polyadenylation sites by suppression of the default site. These
transcripts have longer 3
UTR with additional potential for
post-transcriptional regulation.
– Unstable transcripts (PROMPTS) that are dependent on
the main promoter in the region are generated from both
The Transcriptional Landscape 11
the forward and the reverse strand in the upstream region
between –500 and –2,500. The function of these is currently
unknown.
– Flanking the transcription start site, short transcripts are
made in both directions (PASR, TSS-a, and NRO RNAs).
To simplify the picture, three different types of observations
described above are merged into a single class of transcripts in
the figure although it is unclear whether these observations
are of the same phenomenon. In any event, these transcripts
are intimately associated with the mechanism of transcription
initiation in a way that is currently unclear. Some transcripts
in the sense orientation may be paused transcripts that are
involved in a regulatory mechanism in which the polymerase
is shifted from a paused to a processive mode.
– SnoRNAs (small nucleolar RNAs) of 60–300 nt ar e made
from within introns in protein coding genes. These are stable
transcripts made by processing of the intron RNA. They are
chiefly of two types (box C/D methylation guides and box
H/ACA pseudouridylation guides) that specify sites of mod-
ification of other RNAs, mainly ribosomal RNA. In other
eukaryotes, these RNAs are predominantly made from inde-
pendent genes or polycistronic transcripts.
– Many miRNAs (micro RNAs) ar e similarly made by process-
ing of intron RNA in humans. The mature miRNAs are 21–
23 nt RNAs that acts as translational repressors targeting a
substantial fraction of the mRNA population (current esti-
mates range from 30 to 90%).
– NcRNAs (Non-coding RNAs) are transcribed from both
the forward and the reverse strand and from different start
sites. Those transcribed f rom the reverse strand are the main
source of antisense RNAs found abundantly throughout the
genome and has a potential to regulate the expr ession of the
protein coding transcripts made from the forward strand in
various ways.
– TASRs (terminator-associated short R NAs) are f ound in
both orientations transcribed from a region corresponding
to the polyadenylation site and also found to be relatively
devoid of nucleosomes. These RNAs are preferentially asso-
ciated with active genes and their function is unknown.
Figure 1.2 describes a generalized picture of transcription
of a protein coding region of the human genome. One of the
fascinating things of our time is that similar information relat-
ing to a gene of specific interest is publicly accessible in genome
browsers such as the UCSC (http://genome.ucsc.edu/)and
Ensembl (http://www.ensembl.org/index.html) browsers.
Furthermore, these databases have links for each entry to many
12 Nielsen
other types of biological and medical information. Thus, it is
possible to get an overview of a gene of interest that may be suf-
ficient to generate relevant hypotheses simply by browsing the
databases.
Complicated as it may seem at present, the full story of the
transcriptional complexity of the human genome has probably not
yet been told. Interesting new information is likely to come from
studies of co-transcriptional events and analyses on how tran-
scripts are used in modifying DNA and chromatin. Due to its
ability to fold into intricate conformations, RNA has the ability to
bind specifically to small molecule ligands, including metabolites
and to catalyze b iochemical reactions. This is well known fr om
riboswitches and ribozymes found in other systems (31). There
are currently no known riboswitches described from the human
genome and the ribozymes are only represented by the funda-
mental catalytic activities in the ribosome, the spliceosome (not
definitively proven), and RNase P and a few sporadic small cleav-
age ribozyme of uncertain biological function (32). However, it is
noteworthy that bioinformatics analyses reveals hundreds of thou-
sands of sequences that appears to be conserved at the RNA struc-
ture level indicating that much more is to be found in relation to
functional RNA molecules (33, 34). What is clear is that RNA
is taking over the central stage in our efforts to understand the
complex structure and expression of genetic information. This
is challenging the gene as the basic operational unit in molecu-
lar biology. Just like alternative splicing makes it is impossible to
decide whether a given sequence belongs to an exon or an intron
at the DNA level without reference to the mature mRNA tran-
script, it is generally not possible to assign a DNA sequence to a
single transcript and thus to a single phenotype. It has been sug-
gested that it is time to consider the individual transcript as the
operational unit in place of the gene (5). This would be in concert
with the evolutionary view on genetic information. Or, in other
words, the RNA World is still around!
References
1. Carninci, P., Kasukawa, T., Katayama, S.,
Gough, J., Frith, M. C., Maeda, N., Oyama,
R., Ravasi, T., Lenhard, B., Wells, C.,
et al. (2005) The transcriptional landscape
of the mammalian genome. Science 309,
1559–1563.
2. Birney, E., Stamatoyannopoulos, J. A.,
Dutta, A., Guigo, R., Gingeras, T. R., Mar-
gulies, E. H., Weng, Z., Snyder, M., Der-
mitzakis, E. T., Thurman, R. E., et al.
(2007) Identification and analysis of func-
tional elements in 1% of the human genome
by the ENCODE pilot project. Nature 447,
799–816.
3. Cheng, J., Kapranov, P., Drenkow, J., Dike,
S., Brubaker, S., Patel, S., Long, J., Stern,
D., Tammana, H., Helt, G., et al. (2005)
Transcriptional maps of 10 human chro-
mosomes at 5-nucleotide resolution. Science
308, 1149–1154.
4. Kiyosawa, H., Yamanaka, I., Osato,
N., Kondo, S. and Hayashizaki, Y.
(2003) Antisense transcripts with FAN-
TOM2 clone set and their implications
The Transcriptional Landscape 13
for gene regulation. Genome Res 13,
1324–1334.
5. Gingeras, T. R. (2007) Origin of pheno-
types: genes and transcripts. Genome Res 17,
682–690.
6. Neil, H., Malabat, C., Ubenton-Carafa, Y.,
Xu, Z., Steinmetz, L. M. and Jacquier, A.
(2009) Widespread bidirectional promoters
are the major source of cr yptic transcripts in
yeast. Nature 457, 1038–1042.
7. 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.
8. Bertone, P., Stolc, V., Royce, T. E.,
Rozowsky, J. S., Urban, A. E., Zhu, X., Rinn,
J. L., Tongprasit, W., Samanta, M., Weiss-
man, S., et al. (2004) Global identification of
human transcribed sequences with genome
tiling arrays. Science 306, 2242–2246.
9. Kampa, D., Cheng, J., Kapranov, P.,
Yamanaka, M., Brubaker, S., Cawley, S.,
Drenkow, J., Piccolboni, A., Bekiranov, S.,
Helt, G., et al. (2004) Novel RNAs identi-
fied fro m an in-depth analysis of the tran-
scriptome of human chr omosomes 21 and
22. Genome Res 14, 331–342.
10. Gardner, P. P., Daub, J., Tate, J. G.,
Nawrocki, E. P., Kolbe, D. L., Lindgreen,
S., Wilkinson, A. C., Finn, R. D., Griffiths-
Jones, S., Eddy, S. R., et al. (2009) Rfam:
updates to the RNA families database.
Nucleic Acids Res 37, D136–D140.
11. Daub, J., Gardner, P. P., Tate, J., Ram-
skold, D., Manske, M., Scott, W. G., Wein-
berg, Z., Griffiths-Jones, S. and Bateman,
A. (2008) The RNA WikiPr oject: commu-
nity annotation of RNA families. RNA 14,
2462–2464.
12. Katayama, S., Tomaru, Y., Kasukawa, T.,
Waki, K., Nakanishi, M., Nakamura, M.,
Nishida, H., Yap, C. C., Suzuki, M., Kawai,
J., et al. (2005) Antisense transcription in
the mammalian transcriptome. Science 309,
1564–1566.
13. Claverie, J. M. (2005) Fewer genes, more
noncoding RNA. Science 309, 1529–1530.
14. W illingham, A. T., Orth, A. P., Batalov, S.,
Peters, E. C., Wen, B. G., Za-Blanc, P.,
Hogenesch, J. B. and Schultz, P. G. (2005) A
strategy for probing the function of noncod-
ing RNAs finds a repressor of NFAT. Science
309, 1570–1573.
15. He, Y., Vogelstein, B., Velculescu, V. E.,
Papadopoulos, N. and Kinzler, K. W. (2008)
The antisense transcriptomes of human cells.
Science 322, 1855–1857.
16. Perocchi, F., Xu, Z., Clauder-Munster, S. and
Steinmetz, L. M. (2007) Antisense artifacts
in transcriptome microarray experiments are
resolved by actinomycin D. Nucleic Acids Res
35, e128.
17. Krystal, G. W., Armstrong, B. C. and Battey,
J. F. (1990) N-myc mRNA forms an RNA-
RNA duplex with endogenous antisense tran-
scripts. MolCellBiol10, 4180–4191.
18. Thrash-Bingham, C. A. and Tartof, K. D.
(1999) aHIF: a natural antisense transcript
overexpressed in human renal cancer and
during hypoxia. J Natl Cancer Inst 91,
143–151.
19. Hongay, C. F., Grisafi, P. L., Galitski, T. and
Fink, G. R. (2006) Antisense transcription
controls cell fate in Saccharomyces cerevisiae.
Cell 127, 735–745.
20. Sandelin, A., Carninci, P., Lenhard, B., Pon-
javic, J., Hayashizaki, Y. and Hume, D. A.
(2007) Mammalian RNA polymerase II core
promoters: insights from genome-wide stud-
ies. Nat Rev Genet 8, 424–436.
21. Kodzius, R., Kojima, M., Nishiyori, H .,
Nakamura, M., Fukuda, S., Tagami, M.,
Sasaki, D., Imamura, K., Kai, C., Harbers,
M., et al. (2006) CAGE: cap analysis of gene
expression. Nat Methods 3, 211–222.
22. Kapranov, P., Cheng, J., Dike, S., Nix, D. A.,
Duttagupta, R., Willingham, A. T., Stadler, P.
F.,Hertel,J.,Hackermuller,J.,Hofacker,I.
L., et al. (2007) RNA maps reveal new RNA
classes and a possible function for pervasive
transcription. Science 316, 1484–1488.
23. Affymetrix/Cold Spring Harbor Laboratory
ENCODE Transcriptome Project (2009)
Post-transcriptional processing generates a
diversity of 5
-modified long and short
RNAs. Nature 457, 1028–1032.
24. Seila, A. C., Calabrese, J. M., Levine, S. S.,
Yeo, G. W., Rahl, P. B., Flynn, R. A., Young,
R. A. and Sharp, P. A. (2008) Divergent tran-
scription from active promoters. Science 322,
1849–1851.
25. Core, L. J. and Lis, J. T. (2008) Transcrip-
tion regulation through promoter-proximal
pausing of RNA polymerase II. Science 319,
1791–1792.
26. Preker, P., Nielsen, J., Kammler, S., Lykke-
Andersen, S., Christensen, M. S., Mapen-
dano, C. K., Schierup, M. H. and Jensen, T.
H. (2008) RNA exosome depletion reveals
transcription upstream of active human pro-
moters. Science 322, 1851–1854.
27. Buratowski, S. (2008) Transcription. Gene
expression–where to start? Science 322,
1804–1805.
28. Gerstein, M. B., Bruce, C., Rozowsky, J.
S., Zheng, D., Du, J., Korbel, J. O.,