CpG-island promoters drive transcription of human telomeres.
ABSTRACT The longstanding dogma that telomeres, the heterochromatic extremities of linear eukaryotic chromosomes, are transcriptionally silent was overturned by the discovery that DNA-dependent RNA polymerase II (RNAPII) transcribes telomeric DNA into telomeric repeat-containing RNA (TERRA). Here, we show that CpG dinucleotide-rich DNA islands, shared among multiple human chromosome ends, promote transcription of TERRA molecules. TERRA promoters sustain cellular expression of reporter genes, are located immediately upstream of TERRA transcription start sites, and are bound by active RNAPII in vivo. Finally, the identified promoter CpG dinucleotides are methylated in vivo, and cytosine methylation negatively regulates TERRA abundance. The existence of subtelomeric promoters, driving TERRA transcription from independent chromosome ends, supports the idea that TERRA exerts fundamental functions in the context of telomere biology.
- SourceAvailable from: sciencedirect.com[show abstract] [hide abstract]
ABSTRACT: Telomeres are protective structures present at the ends of linear chromosomes and consist of simple repeating-DNA sequences and specialized proteins [1, 2]. Integrity of the telomeres is important in maintaining genome stability[1-6]. RNA interference(RNAi) involves short double-stranded RNA (21-23 nucleotides long), termed short interference RNA(siRNA), resulting in the downregulation of genes with cognate sequences [7-9]. During transient siRNA-induced RNAi in mouse fibroblast cultures, we found significant reversible changes related to the telomeres. Telomeres acquired distinct heterochromatin features. There were increased bindings of Argonaute-1 (AGO1), telomeric repeat-binding factor 1(TERF1), and heterochromatin protein 1beta (HP1beta) on the telomeres. Histone H3 (lysine 9) was hypermethylated at the telomeres. The chromosome ends also were associated with an unidentified RNA. During RNAi, expression of a transgene inserted adjacent to the telomere was downregulated. In addition, the concentration of a group of heterogeneous high-molecular-weight RNA containing telomeric repeat sequences was increased, and this RNA formed a small number of transient, discrete nuclear foci. Our findings suggest that telomeres participate actively in the siRNA-induced RNAi process. These responses of telomeres to the RNAi process might partially account for the off-target effects of RNAi.Current Biology 03/2008; 18(3):183-7. · 9.49 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: The genomes of prokaryotes and eukaryotic organelles are usually circular as are most plasmids and viral genomes. In contrast, the nuclear genomes of eukaryotes are organized on linear chromosomes, which require mechanisms to protect and replicate DNA ends. Eukaryotes navigate these problems with the advent of telomeres, protective nucleoprotein complexes at the ends of linear chromosomes, and telomerase, the enzyme that maintains the DNA in these structures. Mammalian telomeres contain a specific protein complex, shelterin, that functions to protect chromosome ends from all aspects of the DNA damage response and regulates telomere maintenance by telomerase. Recent experiments, discussed here, have revealed how shelterin represses the ATM and ATR kinase signaling pathways and hides chromosome ends from nonhomologous end joining and homology-directed repair.Annual Review of Genetics 09/2008; 42:301-34. · 17.44 Impact Factor
- Genes & Development 06/2006; 20(9):1050-6. · 12.44 Impact Factor
CpG-island promoters drive transcription
of human telomeres
SOLOMON G. NERGADZE,1,3BENJAMIN O. FARNUNG,2,3HARRY WISCHNEWSKI,2LELA KHORIAULI,1
VALERIO VITELLI,1RAGHAV CHAWLA,2ELENA GIULOTTO,1and CLAUS M. AZZALIN2
1Dipartimento di Genetica e Microbiologia Adriano Buzzati-Traverso, Universita ` di Pavia, 2700 Pavia, Italy
2Institute of Biochemistry, Eidgeno ¨ssische Technische Hochschule Zu ¨rich (ETHZ), CH-8093 Zu ¨rich, Switzerland
The longstanding dogma that telomeres, the heterochromatic extremities of linear eukaryotic chromosomes, are transcription-
ally silent was overturned by the discovery that DNA-dependent RNA polymerase II (RNAPII) transcribes telomeric DNA into
telomeric repeat-containing RNA (TERRA). Here, we show that CpG dinucleotide-rich DNA islands, shared among multiple
human chromosome ends, promote transcription of TERRA molecules. TERRA promoters sustain cellular expression of reporter
genes, are located immediately upstream of TERRA transcription start sites, and are bound by active RNAPII in vivo. Finally, the
identified promoter CpG dinucleotides are methylated in vivo, and cytosine methylation negatively regulates TERRA
abundance. The existence of subtelomeric promoters, driving TERRA transcription from independent chromosome ends,
supports the idea that TERRA exerts fundamental functions in the context of telomere biology.
Keywords: telomeres; TERRA; promoters; transcription; CpG methylation
Telomeres are heterochromatic DNA-protein complexes
located at the end of linear eukaryotic chromosomes.
Telomeres are essential to assure genome stability by allow-
ing cells to distinguish between intrachromosomal DNA
double-stranded breaks and natural chromosome ends,
thereby preventing inappropriate DNA repair events (Palm
and de Lange 2008; Xin et al. 2008). Telomeres also set the
life span of normal adult somatic cells (Harley et al. 1990;
Allsopp et al. 1992; Allsopp and Harley 1995). The DNA
component of mammalian telomeres consists of tandem
arrays of duplex 59-TTAGGG-39/39-AATCCC-59 repeats,
with the G-rich strand extending beyond its complement
to form a 39 overhang; the protein component includes
several factors, the most prominent of which are grouped
together under the name ‘‘shelterin,’’ a multiprotein com-
plex involved in telomere length regulation and telomere
protection (Palm and de Lange 2008; Xin et al. 2008).
In contrast to the longstanding idea that telomeres are
transcriptionally silent, we and others recently discovered
that RNA polymerase II (RNAPII) transcribes telomeric
DNA into telomeric repeat-containing RNA (TERRA)
molecules in a variety of eukaryotes including mammals,
zebra fish, and budding yeast (Azzalin et al. 2007; Azzalin
and Lingner 2008; Chawla and Azzalin 2008; Ho et al. 2008;
Luke et al. 2008; Schoeftner and Blasco 2008). As for the
majority of RNAPII products, at least a fraction of TERRA
is polyadenylated, and RNAPII associates with telomeres in
vivo (Azzalin and Lingner 2008; Schoeftner and Blasco
2008). In mammals, TERRA transcripts contain telomeric
59-UUAGGG-39 RNA repeats, range in size from about 100
bases (b) up to more than 9 kilobases (kb), and are detected
exclusively in nuclear cellular fractions (Azzalin et al. 2007;
Schoeftner and Blasco 2008). Mammalian TERRA forms
discrete nuclear foci that localize to telomeres not only in
interphase cells but also in transcriptionally inactive meta-
phase cells (Azzalin et al. 2007; Ho et al. 2008; Schoeftner
and Blasco 2008), suggesting the existence of post-tran-
scriptional mechanisms retaining TERRA at telomeres.
Interestingly, some human suppressors with morphogenetic
defects in genitalia proteins (namely, SMG1, UPF1, and
hEST1A/SMG6), which are effectors of an evolutionarily
3These authors contributed equally to this work.
Reprint requests to: Elena Giulotto, Dipartimento di Genetica e
Microbiologia Adriano Buzzati-Traverso, Universita ` di Pavia, 2700 Pavia,
Italy; e-mail: firstname.lastname@example.org; fax: +39-0382-528496; or Claus M.
Azzalin, Institute of Biochemistry, Eidgeno ¨ssische Technische Hochschule
Zu ¨rich (ETHZ), CH-8093 Zu ¨rich, Switzerland; e-mail: claus.azzalin@bc.
biol.ethz.ch; fax: +41-44-6321298.
Article published online ahead of print. Article and publication date are at
RNA (2009), 15:2186–2194. Published by Cold Spring Harbor Laboratory Press. Copyright ? 2009 RNA Society.
conserved RNA quality control pathway known as non-
sense-mediated mRNA decay (Isken and Maquat 2008),
negatively regulate TERRA localization to telomeres without
affecting TERRA degradation rate or total cellular levels
(Azzalin et al. 2007). Short hairpin RNA-mediated de-
pletion of SMG1, UPF1, and hEST1A/SMG6 leads to sud-
den loss of discrete telomeric tracts (Azzalin et al. 2007),
suggesting that dysfunctional TERRA localization may
hinder telomere integrity.
In humans, transcription of at least a fraction of TERRA
molecules starts within different subtelomeres (the genomic
regions immediately adjacent to telomeres) and proceeds
toward chromosome ends (Azzalin et al. 2007). Neverthe-
less, the molecular details of TERRA biogenesis remain to
be elucidated. Here we report that CpG dinucleotide-rich
DNA island promoters, sharing conserved repetitive DNA
elements, are located on different human chromosome
ends and drive transcription of independent TERRA
molecules. TERRA CpG-island promoters are able to
sustain expression of reporter genes in human cells, are
located immediately upstream of TERRA transcription start
sites on different subtelomeres, and are bound by active
RNAPII in vivo. We also show that DNA methyltransferase
(DNMT) 1- and 3b-mediated cytosine methylation of TERRA
promoters negatively correlates with RNAPII binding to
TERRA promoters and cellular TERRA abundance, sug-
gesting that cytosine methylation represses TERRA pro-
moter transcriptional activity.
RESULTS AND DISCUSSION
Human subtelomeres contain active CpG-island
The pro-terminal DNA sequences associated with the long-
arm telomeres of human chromosomes X/Y (Xq/Yq) and 10
(10q) were isolated nearly 20 years ago and named
TelBam3.4 and TelSau2.0, respectively (Brown et al. 1990).
The two sequences share a conserved repetitive region that
extends for z1.6 kb (nucleotides 2110–3117) and z1.3 kb
(nucleotides 408–1789) until z280 nucleotides (nt) up-
stream of the terminal array in TelBam3.4 and TelSau2.0,
respectively (Supplemental Fig. S1). This conserved region
contains three different repetitive DNA tracts: the most
centromere-proximal tract comprises tandemly repeated 61-
base-pair (bp) units (five repeats in TelBam3.4 versus six
repeats in TelSau2.0); a second, more distal tract comprises
29-bp tandem repeats (nine repeats versus 18 repeats);
a third tract comprises five tandemly repeated 37-bp DNA
units in both sequences (Supplemental Fig. S1). We refer to
the tandem repeat-containing region as ‘‘61-29-37 repeats’’
and to the z280 nt comprised between the last 37-bp repeat
and the telomeric hexamers as ‘‘pre-tel’’ (Supplemental Fig.
S1). BLAST search analysis demonstrated that 61-29-37
repeats are also present at 13 other human subtelomeres
(Supplemental Fig. S2, chromosome arms 1p, 2p, 3q, 4p, 5p,
8p, 9p, 11p, 15q, 16p, 17p, 19p, 21q), although variable
numbers of tandem repeats are observed at different loci.
Five additional subtelomeres containing TelBam3.4-like se-
quences (Supplemental Fig. S2, 3q, 6p, 9q, 12p, 20p) were
previously identified using in situ hybridization (Brown
et al. 1990). One intrachromosomal locus at 2q13 also
contains 61-29-37 repeats (Supplemental Fig. S2) and
corresponds to an ancestral telomere–telomere fusion point
(Ijdo et al. 1991).
61-29-37 repeats possess a remarkably high (67%–86%)
overall content in CpG dinucleotides, with a peak corre-
sponding to the 29- and 37-bp repeat tracts (Supplemental
Fig. S1). High CpG contents are typical of CpG islands that
are found in broad-type promoters of many eukaryotic
RNAPII-transcribed genes (Sandelin et al. 2007). Indeed,
bioinformatic analysis of TelBam3.4 and TelSau2.0 se-
quences predicted the existence of CpG-island promoters
spanning the 29- and 37-bp repeats (Supplemental Fig. S1).
In order to experimentally test the promoter activity of this
region, we generated promoter reporter plasmids where
progressive 59 deletions of a z3.5-kb subtelomeric DNA
tract comprising 61-29-37 repeats were inserted upstream
of a green fluorescence protein (eGFP) reporter gene (Fig.
1A). Fluorescence microscopy inspection of human HeLa
cells transfected with reporter plasmids revealed that a
z1-kb sequence comprising the 29- and 37-bp repeat tracts
was sufficient to induce cellular eGFP expression (Fig.
1B). Quantification of eGFP fluorescence by densitomet-
ric analysis and of eGFP mRNA by quantitative reverse
transcription-polymerase chain reaction (qRT-PCR) showed
that the identified promoter is approximately five times less
efficient than the strong cytomegalovirus (CMV) promoter
(Supplemental Fig. S3). Thus, 61-29-37 subtelomeric repeats
contain sequences with promoter activity. In agreement with
this, active RNAPII is enriched at both 61-29-37 and
telomeric repeats over repetitive Alu sequences in vivo, as
demonstrated by chromatin immunoprecipitation experi-
ments performed on formaldehyde cross-linked chromatin
from U2OS cells using independent antibodies against
phosphorylation-activated RNAPII large subunit (Fig. 2;
Hirose and Ohkuma 2007).
Subtelomeric CpG-island promoters drive
transcription of TERRA molecules
To determine whether TelBam3.4 and TelSau2.0 subtelo-
meres are transcribed in vivo, we performed Northern blot
analysis of nuclear RNA extracted from human telomerase-
negative HLF primary fibroblasts and U2OS tumor cells
as well as telomerase-positive HeLa and HEK293T tumor
cells, using strand-specific DNA probes, which correspond
to TelBam3.4 and TelSau2.0 pre-tel sequences (Fig. 3A).
The TelBam3.4 probe is 100% identical to Xq/Yq pre-tels;
92%–98% identical to 8p, 9p, 15q, and 19p pre-tels; and
Human TERRA promoters
90%–91% identical to 10q and 21q pre-tels. The TelSau2.0
probe is 100% identical to 10q pre-tel sequences and 89%–
95% identical to 8p, 9p, 16p, 15q, 19p, and 21q pre-tels. The
two probes are therefore expected to detect partially over-
lapping families of transcripts originating from a multitude
of chromosome ends. After high-stringency washes, both
probes detected RNA species varying in length from z500 b up
to more than 5 kb in the different cell lines tested (Fig.
3B; Supplemental Fig. S4). The radioactive signals were
completely abolished upon ribonuclease A treatment (data
not shown), confirming that they originated from RNA.
We obtained smeary patterns of hybridization also when we
hybridized the same filters with telomeric probes, although
the telomeric signal extended to molecular weights lower
than 100 b (Fig. 3B; Supplemental Fig. S4). These results
suggest that transcripts originating from TelBam3.4 and
TelSau2.0 subtelomeres may constitute a fraction of TERRA
molecules. It is probable that total cellular TERRA also
comprises transcripts deriving from subtelomeres devoid of
TelBam3.4 and TelSau2.0 sequences, as well as from 59
processing of TelBam3.4 and TelSau2.0 RNA molecules.
Interestingly, while the signal detected with TelBam3.4 and
TelSau2.0 probes remains fairly constant throughout all
tested cell lines, the signal detected using telomeric probes
reveals a two- to eightfold increase in total TERRA levels in
U2OS cells, suggesting that in these cells a substantial
fraction of TERRA transcripts does not contain TelBam3.4
and TelSau2.0 sequences. Although the molecular mecha-
nisms leading to increased TERRA levels in U2OS remain
unclear, a connection might exist between TERRA cellular
expression and the homologous recombination-based pro-
cesses maintaining telomere length in U2OS cells (Royle
et al. 2008).
To verify whether TelBam3.4 and TelSau2.0 subtelo-
meres are indeed transcribed into TERRA molecules, we
employed an RT-PCR protocol previously developed to de-
tect TERRA transcripts (Azzalin et al. 2007). We reverse tran-
scribed nuclear RNA using a telomeric 59-(CCCTAA)5-39
oligonucleotide (Fig. 3A, oC; Supplemental Table S1),
complementary to the 59-(UUAGGG)n-39 sequence pres-
ent in TERRA molecules. We then PCR-amplified the ob-
tained cDNA with unique primer pairs corresponding to
TelBam3.4 and TelSau2.0 pre-tel sequences immediately
adjacent to the telomeric tracts in the genome (Fig. 3A, o1,
o2; Supplemental Table S1). We obtained amplicons
matching in size (Fig. 3C) and sequence (data not shown)
the ones obtained in PCR reactions performed on genomic
DNA, thus proving that the tested subtelomeres are indeed
transcribed and that transcription continues into the
FIGURE 1. Promoter reporter assays. (A) A z3.5-kb DNA fragment
from Xq/Yq subtelomeres was cloned upstream of an eGFP cDNA
(green box). The fragment contains five 61-bp tandem repeats (red
box), ten 29-bp repeats (dark blue box), and five 37-bp repeats (light
blue box). The original reporter plasmid was digested with restriction
enzymes in order to generate plasmids containing serial 59 deletions.
Plasmid names are given on the right. pE: promoter-less eGFP
negative control plasmid; pCMV: cytomegalovirus promoter eGFP
positive control plasmid. Gray lines indicate plasmid backbone
sequences. The arrow points to the reporter plasmid containing the
minimal DNA tract sufficient to induce eGFP expression. (B) Re-
porter plasmids containing the puromycin resistance gene were
transfected into HeLa cells. Puromycin was added 24 h after trans-
fection, and 4 d later cells were fixed and imaged by fluorescence
microscopy to detect eGFP expression (green, left panels) and DAPI-
stained DNA (blue, right panels).
FIGURE 2. RNAPII binds to 61-29-37 repeats in vivo. (A) Formal-
dehyde cross-linked chromatin from U2OS cells was immunoprecip-
itated using antibodies raised against phosphorylated serine 2 (pS2)
and serine 5 (pS5) from the C-terminal repeat of human RNAPII.
Immunoprecipitated DNA was dot-blotted and hybridized with
radioactive DNA probes detecting 61-29-37 repeats. The same blot
was stripped and re-hybridized sequentially to detect repetitive
telomeric (positive control) and Alu sequences. (B) The bar graph
shows the fraction of input DNA immunoprecipitated in the different
samples, after subtraction of the background signal measured for
control reactions performed using only beads. Bars and error bars
represent averages and standard deviations from three independent
Nergadze et al.
RNA, Vol. 15, No. 12
telomeric tract. To reinforce these findings, we annealed
nuclear RNA prepared from telomerase positive HCT116
human cancer cells with a TelSau2.0 DNA oligonucleotide
adjacent to the telomeric tracts in the genome (Fig. 3A, o2;
Supplemental Table S1). The oligonucleotide is 100% and
70% complementary to sequences from TelSau2.0 and
TelBam3.4 pre-tel transcripts, respectively. We then treated
the RNA with RNase H, which digests specifically RNA
molecules engaged in DNA/RNA hybrids, and subjected the
digested RNA to Northern blot analysis with radioactive
telomeric probes. Quantitative analysis of the obtained
hybridization profiles reproducibly revealed that the RNase
H treatment led to a shift of the telomeric hybridization
profile toward lower molecular weights (Fig. 4). Altogether,
these results indicate that TERRA molecules originating
from TelBam3.4 and TelSau2.0 subtelomeres exist in dif-
ferent human cell lines and constitute a substantial fraction
of total TERRA.
We then used rapid amplification of cDNA ends (RACE)
to isolate intact 59 ends of TelBam3.4 transcripts from
cDNA obtained by reverse transcription of HeLa, HLF,
HCT116, and U2OS nuclear RNA with the telomeric oC
oligonucleotide. Our RACE protocol allows amplification
of RNA molecules with a methylated 59 cap only, thereby
preventing detection of uncapped and degraded RNA
species. In all cell lines, we identified one unique transcrip-
tion start site located 27 nt downstream from the last 37-bp
repeat (Fig. 3D) and conserved on chromosomes 8p, 9p,
15q, 16p, and 19p, and Xq/Yq subtelomeres. Although our
extensive analysis of RACE products revealed only one
transcription start site, we cannot exclude the possibility
that alternative—perhaps more rare—start sites might exist
in the same cells. In addition, as described below, usage
of alternative transcription start sites can occur in DNA
methyltransferase-deficient cells. Nevertheless, transcrip-
tion of a fraction of TERRA molecules begins immediately
downstream from 61-29-37 repeats and continues toward
chromosome ends. Because the physical distance between
the identified transcription start site and the telomeric
tract is z250 nt (Fig. 3D), while the length of TelBam3.4
transcripts visualized by Northern blotting varies between
z500 and >5000 b (Fig. 3B), we infer that transcription
can proceed through the telomeric tract for several kilo-
TERRA promoters are methylated at CpG
dinucleotides and cytosine methylation regulates
TERRA cellular levels
DNA methyltransferase (DNMT)-mediated methylation of
cytosines at promoter CpG dinucleotides negatively regu-
lates transcriptional activity (Esteller 2007; Suzuki and Bird
tablished from patients affected by the immunodeficiency,
FIGURE 3. TERRA transcription from TelBam3.4 and TelSau2.0 subtelomeres. (A) Schematic representation of a 61-29-37 repeat-containing
chromosome end. Note that the sketch is not to scale. (B) Nuclear RNA extracted from the indicated cell lines was electrophoresed and hybridized
with radioactive strand-specific probes corresponding to TelBam3.4 pre-tel sequences (black bar in [A]). The same blot was stripped and probed
for total TERRA. Ethidium bromide-stained 28S and 18S rRNA transcripts are shown to confirm equal RNA loading. Standard molecular weights
are on the left in kilobases (kb). (C) Nuclear RNA (r) from the indicated cell lines was reverse transcribed (RT) using oC oligonucleotides, and
cDNA was PCR amplified using o1 and o2 oligonucleotides specific for TelBam3.4 or TelSau2.0 pre-tel sequences (oligonucleotide positions are
indicated by arrows in A). PCR products were run in agarose gels and stained. Control PCR reactions were performed using genomic DNA (d) as
template (t). (D) RACE experiments performed on nuclear RNA from HeLa, U2OS, and HCT116 cells and from HCT116-derived cells knocked-
out for DNA methyltransferases 1 and 3b (double KO, DKO) identified the capitalized adenines as transcription start sites at TelBam3.4
subtelomeres. Numbers indicate nucleotide positions as they appear in the database entry M57752.1.
Human TERRA promoters
centromere instability, and facial anomalies (ICF) syn-
drome are characterized by hypomethylated subtelomeric
DNA and increased TERRA expression, compared to
normal cells from healthy individuals (Yehezkel et al.
2008). Promoter CpG methylation may thus regulate
TERRA expression. We digested HLF, HeLa, and U2OS
genomic DNA with the methylation-sensitive HpaII re-
striction enzyme or with its methylation-insensitive iso-
schizomer MspI and performed Southern blot hybridiza-
tions with radiolabeled DNA probes, corresponding to 61-
29-37 repeats. While MspI digestion generated prominent
hybridization bands ranging in size between 100 and 500 bp,
additional bands of higher molecular weight (up to more
than 23 kb) appeared in HpaII-digested samples (Fig. 5A).
This indicates that 61-29-37 repeats are methylated in vivo.
61-29-37 repeat methylation appears to be nearly complete
in HeLa cells, while a lower degree of methylation is
observed in HLF and U2OS cells, as demonstrated by
increased size heterogeneity of the hybridization bands in
HpaII-digested samples (Fig. 5A). This is consistent with
recent observations showing that telomerase-positive hu-
man cancer cells carry hypermethylated subtelomeric DNA
compared to telomerase-negative cancer or primary human
cells (Vera et al. 2008; Ng et al. 2009; Tilman et al. 2009).
We then analyzed the methylation state of 61-29-37 re-
peats in human HCT116 cells knocked-out for DNMT1
(DNMT1?/?), for DNMT3b (DNMT3b?/?) or for both
(double KO [DKO]) methyltransferases (Rhee et al. 2000;
Rhee et al. 2002). Concomitant disruption of DNMT1 and
DNMT3b genes completely abolished methylation at 61-29-
37 repeat CpG dinucleotides, while only modest changes
were observed in single KO cells (Fig. 5B), implying that
DNMT1 and DNMT3b enzymes cooperatively maintain
DNA methylation at 61-29-37 repeats in HCT116 cells.
Northern blot analysis disclosed dramatically increased
steady-state levels of TelBam3.4, TelSau2.0, and total
TERRA transcripts in DKO cells compared to parental,
DNMT1?/?, and DNMT3b?/?cells (Figs. 4, 6A; Supple-
mental Fig. S5). Similarly, treatments of HeLa cells with the
DNMT inhibitor 5-azacytidine (5aza) induced an approx-
imately threefold increase in TelBam3.4 and TelSau2.0
RNA species (Supplemental Fig. S6). The extra TelBam3.4
and TelSau2.0 transcripts detected in DKO cells (Figs. 4,
6A) are, at least in part, true TERRA molecules for the
following reasons: first, RT-PCR experiments using telo-
meric oC oligonucleotides for reverse transcription pro-
duced TelBam3.4 and TelSau2.0 amplicons matching the
corresponding genomic sequences both in parental and
DKO cells (Supplemental Fig. S5). Second, RACE experi-
ments performed on nuclear RNA prepared from DKO
cells identified the same TelBam3.4 transcription start site
as for parental HCT116 cells (see Fig. 3D, nucleotide 3143)
in 31 out of the 41 analyzed positive RACE plasmid clones
and a second transcription start site within the first 37-bp
repeat (see Fig. 3D, nucleotide 2918) in the remaining 10
clones. Third, RNase H digestion of DKO nuclear RNA,
previously annealed with TelSau2.0 DNA oligonucleotides,
induced a substantial shift of the TERRA hybridization
profile toward lower molecular weights (Fig. 4). In con-
clusion, CpG methylation appears to modulate the cellular
abundance of TERRA (including TERRA transcribed from
TelBam3.4 and TelSau2.0 subtelomeres) as well as the us-
age of independent transcription start sites. We hypothe-
size that CpG methylation represses TERRA promoter
transcriptional activity. Consistent with this hypothe-
sis, ChIP experiments revealed increased levels of active
FIGURE 4. (Legend on next page)
Nergadze et al.
RNA, Vol. 15, No. 12
RNAPII binding to 61-29-37 and telomeric repeat DNA in
DKO cells compared to parental HCT116 cells (Fig. 6B).
However, we cannot exclude the possibility that, in DKO
cells, impaired degradation of TERRA transcripts could
occur, thereby contributing to their increased steady-state
Aberrant cellular accumulation of TERRA transcripts has
been reported in Saccharomyces cerevisiae yeast mutants
knocked out for the 59–39 RNA exonuclease Rat1p, which is
thought to degrade RNA species derived from transcription
occurring past poly(A) cleavage sites in RNAPII-tran-
scribed genes (Rosonina et al. 2006; Luke et al. 2008).
One could therefore speculate that TERRA molecules
originate from inefficient transcription termination of
genes placed subtelomerically. On the contrary, we reveal
that in human cells, CpG-island promoters, embedded
within subtelomeric 61-29-37 repeats, drive transcription
of TERRA molecules possibly from up to 20 different
subtelomeres, suggesting that common regulatory mecha-
nisms control the biogenesis of TERRA transcripts origi-
nating from independent chromosome ends. We were
unable to identify 61-29-37-like repeats at the remaining
26 human subtelomeres, including 11q and Xp/Yp sub-
telomeres, which were previously demonstrated to be
transcribed (Azzalin et al. 2007). It is likely that different
promoter types contribute to the biogenesis of total human
TERRA, although we cannot exclude the possibility that ill-
defined subtelomeric sequences available in the databases
(Riethman et al. 2004; Riethman 2008) might have led us to
underestimate the actual number of human subtelomeres
carrying 61-29-37 repeats.
Our discoveries define human telomeres as components
of integral ‘‘genic’’ units and make TERRA rise above the
transcriptional noise associated with the human genome,
supporting the idea that TERRA might exert important
functions in telomere biology. It has been proposed that
TERRA could repress telomerase activity at chromosome
ends by base-pairing with the template sequence in the
telomerase RNA moiety (Luke et al. 2008; Schoeftner and
Blasco 2008). In this scenario, methylation-mediated tran-
scriptional repression of TERRA CpG-island promoters
may be part of an epigenetic tumor suppressor gene-
silencing program accompanying
(Esteller 2007; Suzuki and Bird 2008).
MATERIALS AND METHODS
DNA sequence analysis
We aligned the TelBam3.4 and TelSau2.0 sequences with the
entire human genome using the basic local alignment search tool
(BLAST) at the National Center for Biotechnology Information
(http://www.ncbi.nlm.nih.gov/blast/Blast.cgi). To analyze CpG
dinucleotide contents and predict CpG islands, we utilized the
CpGPlot/CpGReport at the European molecular biology open soft-
ware suite program (EMBOSS; http://www.ebi.ac.uk/Tools/emboss/
cpgplot/) (Rice et al. 2000).
FIGURE 4. Transcripts from TelSau2.0 and TelBam3.4 subtelomeres
constitute a substantial fraction of total TERRA molecules. (A)
Nuclear RNA from the indicated cell lines was annealed to a DNA
oligonucleotide 100% and 70% complementary to TelSau2.0 and
TelBam3.4 pre-tel transcripts, respectively. The same RNA was also
annealed to a DNA oligonucleotide complementary to GAPDH
mRNA. RNA was incubated with RNase H or left untreated, electro-
phoresed, and hybridized with radioactive strand-specific probes
corresponding to TelSau2.0 pre-tel sequences. The same blot was
stripped and probed sequentially to detect total TERRA, GAPDH
(positive control for RNAse H reaction), and 18S (control for RNase
H specificity and loading control) RNA sequences. Standard molec-
ular weights are shown on the left in kilobases (kb). (B) Charts
showing the distribution of signal intensity for the Northern blots
shown in (A). The asterisks indicate RNase H-induced peaks in the
TelSau2.0 hybridization profile of HCT116 RNA. The peaks are likely
to correspond to TelSau2.0-like pre-tel sequences from which the
telomeric RNA tracts have been removed by the RNase H treatment.
The black and white arrows indicate the lane area used for signal
quantifications. Similar shifts in the telomeric profiles were obtained
when RNA was incubated only with TelSau2.0 oligonucleotides,
omitting the GAPDH oligonucleotides from the reactions (data not
FIGURE 5. Cytosine methylation at 61-29-37 repeats. (A,B) Genomic
DNA extracted from the indicated cell lines was digested with the
methylation-sensitive HpaII restriction enzyme or with its methyla-
tion-insensitive isoschizomer MspI. Digested DNA was electropho-
resed, blotted, and hybridized with radioactive DNA probes detecting
61-29-37 repeats. (B) DNMT1?/?and DNMT3b?/?are HCT116-derived
clonal cell lines knocked-out for DNA methyltransferases 1 and 3b,
respectively. Double KO (DKO–) are HCT116-derived clonal cell lines
knocked-out for both methyltransferases. Standard molecular weights
are shown on the left of each blot in kilobases (kb).
Human TERRA promoters
Since TelBam3.4 and TelSau2.0 plasmids (Brown et al. 1990) were
no longer available, we subcloned a z3-kb BamHI-EcoRI human
genomic DNA fragment from the bacterial artificial chromosome
clone RP11-34P13. This fragment is identical to a portion of the
TelBam3.4 sequence and contains 61-, 29-, and 37-bp repeats. The
pre-tel fragment comprised between the last 37-bp repeat and
the telomeric repeats was obtained by PCR on human genomic
DNA using primers constructed on the TelBam3.4 sequence. We
constructed a plasmid containing the eGFP cDNA, under the
CMV promoter, and the puromycin resistance gene using the
pEGFN1 and pPUR plasmids (Clontech). We then substituted
the CMV promoter with the z3.5-kb TelBam3.4 fragment
and obtained deletions using appropriate restriction enzymes or
substituting the CMV promoter with PCR amplified fragments.
We checked all constructs by sequencing.
Cell lines and tissue culture procedures
We cultured HLF (human lung fibroblasts), HeLa (human
cervical carcinoma), HEK293T (human embryonic kidney),
U2OS (human osteosarcoma), and HCT116 (human colon
carcinoma) cells in high-glucose D-MEM supplemented with
10% fetal calf serum, nonessential amino acids, and penicillin-
streptomycin (Invitrogen). Where indicated, we treated cells with
10 mM 5-azacytidine (Sigma Aldrich). For promoter reporter
assays, we transfected HeLa cells using the Lipofectamine 2000
reagent (Invitrogen). Twenty-four hours after transfection, we
selected positively transfected cells in medium containing 1 mg/mL
puromycin for 4 d.
RNA preparation and analysis
We prepared RNA from whole cells or nuclei-enriched cellular
fractions (Azzalin et al. 2007) using the Nucleospin RNA II kit
(Macherey-Nagel). We treated RNA twice with RNase-free DNase I
(New England Biolabs) to eliminate any DNA contaminations.
For RNase H experiments, we mixed 10 mg of nuclear RNA with
600 pmol of a TelSau2.0 oligonucleotide and 600 pmol of
a GAPDH oligonucleotide (see Supplemental Table S1). We
incubated the RNA/DNA mix at 65°C for 4 min and at room
temperature for 20 min. We then added 1 U of RNase H (New
England Biolabs) to the mix and allowed digestion at 37°C for 1 h.
For Northern blots, we electrophoresed 10–20 mg of RNA in 1.2%
formaldehyde agarose gels and blotted it to nylon membranes. We
hybridized membranes for z18 h in Church buffer containing
32P-labeled probes at 50°C–64°C. The strand-specific telomeric
probe used to detect total TERRA was described previously
(Azzalin et al. 2007). The TelBam3.4 and TelSau2.0 probes were
genomic PCR products obtained with o1 and o2 oligonucleotides
and labeled by primer extension using the o2 oligonucleotide (see
Supplemental Table S1). The b-actin, GAPDH, and 18S rRNA
probes were 59 end-labeled DNA oligonucleotides (see Supple-
mental Table S1). After hybridization, we washed membranes in
0.2–13 SSC, 0.5% SDS at the same temperatures used for
hybridizations. We detected radioactive signals using a Storm
PhosphorImager (GE Healthcare) and quantified them with
Quantity One (Bio-Rad) and with ImageQuant (GE Healthcare)
software. For quantitative RT-PCR experiments, we reverse
transcribed 1 mg of total RNA with random hexamers using the
SUPERSCRIPT III RNase H-reverse transcriptase (Invitrogen).
We PCR amplified the obtained cDNA using eGFP and GAPDH
oligonucleotide pairs (see Supplemental Table S1) for 10 sec at
95°C, 20 sec at 60°C, and 20 sec at 72°C (45 cycles) using the
LightCycler 480 SYBR Green I master mix and instrument
(Roche). The RT-PCR-based approach to amplify chromosome-
specific TERRA molecules was previously described (Azzalin et al.
2007). We performed 59 RACE experiments with the FirstChoice
RLM-RACE Kit (Ambion) using the oC oligonucleotide for
reverse transcription and the TelBam3.4 gene-specific oligonucle-
otides oR1 and oR2 for PCR (see Supplemental Table S1).
FIGURE 6. DNMT1 and DNMT3b enzymes cooperatively repress
TERRA cellular levels. (A) Nuclear RNA was extracted from the
indicated cell lines, electrophoresed, blotted, and hybridized using
strand-specific TelBam3.4 or TelSau2.0 radioactive probes. The same
blots were stripped and re-probed sequentially for total TERRA and
b-actin. Molecular weights of 28S (4.8-kb) and 18S (1.9-kb) rRNA are
shown on the left. (B) Formaldehyde cross-linked chromatin from the
indicated cell lines was immunoprecipitated using antibodies raised
against phosphorylated serine 2 (pS2) and serine 5 (pS5) from the
C-terminal repeat of human RNAPII. Immunoprecipitated DNA was
dot-blotted and hybridized sequentially with radioactive DNA probes
detecting 61-29-37 and telomeric repeats. Graphs show the fraction of
input DNA immunoprecipitated in the different samples, after sub-
traction of the background signal measured for control reactions
performed using only beads. Bars and error bars represent averages
and standard deviations from three independent immunoprecipita-
Nergadze et al.
RNA, Vol. 15, No. 12
We cloned RACE products into the pDRIVE vector (Qiagen)
and sequenced 18 to 41 independent colony plasmids for each cell
DNA methylation analysis
We prepared genomic DNA with the Wizard genomic DNA kit
(Promega) and digested it with HpaII (CpG methylation-sensi-
tive) or MspI (CpG methylation-insensitive) restriction enzymes
(New England Biolabs). After electrophoresis in 1.2% agarose gels,
we denatured DNA, transferred it to nylon membranes, and
hybridized it for z18 h at 64°C with32P-labeled probes generated
by random primer labeling of a z1-kb DNA fragment comprising
61-29-37 repeats. We performed post-hybridization stringency
washes in 0.23 SSC, 0.5% SDS at 64°C, and detected and
quantified radioactive signals as for Northern blots.
We cross-linked cells in 1% formaldehyde for 30 min at room
temperature. We resuspended cell pellets in 1% SDS, 10 mM
EDTA, and 50 mM Tris-HCl (pH 8); sonicated them using
a Bioruptor (Diagenode); and diluted extracts in 150 mM NaCl,
20 mM Tris-HCl (pH 8), 1% Triton X-100, and 2 mM EDTA. We
performed immunoprecipitations using rabbit polyclonal anti-
bodies raised against phosphorylated serine S2 or serine S5 (A300-
654A and A300-655A, Bethyl Laboratories) from the human
RNAPII C-terminal repeat. After isolating immunocomplexes
using protein A and G beads and purifying immunoprecipitated
DNA with the Wizard SV gel and PCR cleanup system (Promega),
we dot-blotted DNA onto nylon membranes and hybridized it
with32P-labeled 61-29-37 repeat probes as for DNA methylation
analysis. After signal detection, we stripped the filters and
hybridized them sequentially with
telomeric and Alu repeat (see Supplemental Table S1) sequences.
32P-labeled probes detecting
Supplemental material can be found at http://www.rnajournal.org.
We are grateful to William R.A. Brown for helpful discussions,
and to Raffaella Santoro, Bert Vogelstein, Sergio Comincini, and
Fiorenzo Peverali for reagents and instruments. The laboratory of
E.G. is supported by Fondazione Cariplo (2008-2507), Ministero
dell’ Universita ` e della Ricerca (PRIN-2006), and Regione Lombardia
Progetto Biomedicina. The laboratory of C.M.A. is supported by the
Swiss National Science Foundation (3100A0-120090 and PP00P3-
123356), ETH Zu ¨rich (ETH-15 08-1), and Fondazione Cariplo
(2008-2507). R.C. is a Swiss National Science Foundation fellow
Received May 22, 2009; accepted September 3, 2009.
Allsopp RC, Harley CB. 1995. Evidence for a critical telomere length
in senescent human fibroblasts. Exp Cell Res 219: 130–136.
Allsopp RC, Vaziri H, Patterson C, Goldstein S, Younglai EV,
Futcher AB, Greider CW, Harley CB. 1992. Telomere length
predicts replicative capacity of human fibroblasts. Proc Natl Acad
Sci 89: 10114–10118.
Azzalin CM, Lingner J. 2008. Telomeres: The silence is broken. Cell
Cycle 7: 1161–1165.
Azzalin CM, Reichenbach P, Khoriauli L, Giulotto E, Lingner J. 2007.
Telomeric repeat containing RNA and RNA surveillance factors at
mammalian chromosome ends. Science 318: 798–801.
Brown WR, MacKinnon PJ, Villasante A, Spurr N, Buckle VJ,
Dobson MJ. 1990. Structure and polymorphism of human
telomere-associated DNA. Cell 63: 119–132.
Chawla R, Azzalin CM. 2008. The telomeric transcriptome and
SMG proteins at the crossroads. Cytogenet Genome Res 122: 194–
Esteller M. 2007. Cancer epigenomics: DNA methylomes and histone-
modification maps. Nat Rev Genet 8: 286–298.
Harley CB, Futcher AB, Greider CW. 1990. Telomeres shorten during
ageing of human fibroblasts. Nature 345: 458–460.
Hirose Y, Ohkuma Y. 2007. Phosphorylation of the C-terminal
domain of RNA polymerase II plays central roles in the inte-
grated events of eucaryotic gene expression. J Biochem 141: 601–
Ho CY, Murnane JP, Yeung AK, Ng HK, Lo AW. 2008. Telomeres
acquire distinct heterochromatin characteristics during siRNA-
induced RNA interference in mouse cells. Curr Biol 18: 183–
Ijdo JW, Baldini A, Ward DC, Reeders ST, Wells RA. 1991. Origin of
human chromosome 2: An ancestral telomere–telomere fusion.
Proc Natl Acad Sci 88: 9051–9055.
Isken O, Maquat LE. 2008. The multiple lives of NMD factors:
Balancing roles in gene and genome regulation. Nat Rev Genet 9:
Luke B, Panza A, Redon S, Iglesias N, Li Z, Lingner J. 2008. The Rat1p
59 to 39 exonuclease degrades telomeric repeat-containing RNA
and promotes telomere elongation in Saccharomyces cerevisiae. Mol
Cell 32: 465–477.
Ng LJ, Cropley JE, Pickett HA, Reddel RR, Suter CM. 2009.
Telomerase activity is associated with an increase in DNA
methylation at the proximal subtelomere and a reduction in
telomeric transcription. Nucleic Acids Res 37: 1152–1159.
Palm W, de Lange T. 2008. How shelterin protects mammalian
telomeres. Annu Rev Genet 42: 301–334.
Rhee I, Jair KW, Yen RW, Lengauer C, Herman JG, Kinzler KW,
Vogelstein B, Baylin SB, Schuebel KE. 2000. CpG methylation is
maintained in human cancer cells lacking DNMT1. Nature 404:
Rhee I, Bachman KE, Park BH, Jair KW, Yen RW, Schuebel KE,
Cui H, Feinberg AP, Lengauer C, Kinzler KW, et al. 2002. DNMT1
and DNMT3b cooperate to silence genes in human cancer cells.
Nature 416: 552–556.
Rice P, Longden I, Bleasby A. 2000. EMBOSS: The European
molecular biology open software suite. Trends Genet 16: 276–
Riethman H. 2008. Human telomere structure and biology. Annu Rev
Genomics Hum Genet 9: 1–19.
Riethman H, Ambrosini A, Castaneda C, Finklestein J, Hu XL,
Mudunuri U, Paul S, Wei J. 2004. Mapping and initial analysis
of human subtelomeric sequence assemblies. Genome Res 14: 18–
Rosonina E, Kaneko S, Manley JL. 2006. Terminating the transcript:
Breaking up is hard to do. Genes & Dev 20: 1050–1056.
Royle NJ, Foxon J, Jeyapalan JN, Mendez-Bermudez A, Novo CL,
Williams J, Cotton VE. 2008. Telomere length maintenance—an
ALTernative mechanism. Cytogenet Genome Res 122: 281–291.
Sandelin A, Carninci P, Lenhard B, Ponjavic J, Hayashizaki Y,
Hume DA. 2007. Mammalian RNA polymerase II core pro-
moters: Insights from genome-wide studies. Nat Rev Genet 8: 424–
Human TERRA promoters
Schoeftner S, Blasco MA. 2008. Developmentally regulated transcrip-
tion of mammalian telomeres by DNA-dependent RNA poly-
merase II. Nat Cell Biol 10: 228–236.
Suzuki MM, Bird A. 2008. DNA methylation landscapes: Provocative
insights from epigenomics. Nat Rev Genet 9: 465–476.
Tilman G, Loriot A, Van Beneden A, Arnoult N, Londono-Vallejo JA,
De Smet C, Decottignies A. 2009. Subtelomeric DNA hypo-
methylation is not required for telomeric sister chromatid ex-
changes in ALT cells. Oncogene 28: 1682–1693.
Vera E, Canela A, Fraga MF, Esteller M, Blasco MA. 2008. Epigenetic
regulation of telomeres in human cancer. Oncogene 27: 6817–6833.
Xin H, Liu D, Songyang Z. 2008. The telosome/shelterin complex and
its functions. Genome Biol 9: 232.
Yehezkel S, Segev Y, Viegas-Pequignot E, Skorecki K, Selig S. 2008.
Hypomethylation of subtelomeric regions in ICF syndrome
is associated with abnormally short telomeres and enhanced
transcription from telomeric regions. Hum Mol Genet 17: 2776–
Nergadze et al.
RNA, Vol. 15, No. 12