|Home | About | Journals | Submit | Contact Us | Français|
Telomere-repeat encoding RNA (referred to as TERRA) has been identified as a potential component of yeast and mammalian telomeres. We show here that TERRA RNA interacts with several telomere-associated proteins, including telomere repeat factors 1 (TRF1) and 2 (TRF2), subunits of the Origin Recognition Complex (ORC), heterochromatin protein 1 (HP1), histone H3 trimethyl K9 (H3 K9me3), and members of the DNA damage sensing pathway. siRNA depletion of TERRA caused an increase in telomere dysfunction-induced foci (TIF), aberrations in metaphase telomeres, and a loss of histone H3 K9me3 and ORC at telomere repeat DNA. Previous studies found that TRF2 amino-terminal GAR domain recruited ORC to telomeres. We now show that TERRA RNA can interact directly with the TRF2 GAR and ORC1 to form a stable ternary complex. We conclude that TERRA facilitates TRF2 interaction with ORC and plays a central role in telomere structural maintenance and heterochromatin formation.
Telomeres are the nucleoprotein structures that protect the ends of linear chromosomes (Blackburn et al., 2006; Palm and de Lange, 2008). Dynamic changes in telomere length and structure play key roles in cellular division, replicative aging, and genome stability. Mammalian telomere DNA consists of double stranded TTAGGG repeats that end in a single stranded 3′ overhang and can form complex higher order-structures that include a lariat like T-loop or a single stranded G-quadruplex. The DNA is tightly associated with a protein core, referred to as the Shelterin complex, as well as with surrounding chromatin and chromosome regulatory factors (Blasco, 2007; de Lange, 2005). The major DNA binding proteins that comprise the Shelterin complex include the double strand-specific factors TRF1 and TRF2, and the single strand binding protein POT1. TRF1, TRF2, and POT1 have non-redundant functions at telomeres that are essential for telomere structural maintenance and genome stability.
In addition to Shelterins, several other chromatin-associated proteins play an important role in telomere regulation (Blasco, 2007). The G9a histone methyltransferase and the Rb tumor suppressor protein have been implicated in the maintenance of high levels of H3 K9 methylation at telomeres (Garcia-Cao et al., 2004; Gonzalo et al., 2006). Heterochromatin protein 1 (HP1) can bind H3 methyl K9 at telomeres and facilitate telomere heterochromatin formation (Garcia-Cao et al., 2004; Perrini et al., 2004). Recent studies from our lab have shown that the Origin Recognition Complex (ORC) can localize to telomeres through an interaction with TRF2 (Deng et al., 2007). ORC bound to the amino-terminal basic (GAR) domain of TRF2, and the loss of ORC binding caused an increase in telomere circles and dysfunctional telomeres in metaphase chromosome spreads. ORC is a multifunctional heteromeric hexamer that is essential for initiation of bi-directional DNA replication (Bell, 2002; Prasanth et al., 2004a; Sasaki and Gilbert, 2007). ORC has also been implicated in transcriptional silencing, heterochromatin formation, centrosome function, and sister chromatid cohesion (Prasanth et al., 2004b; Rusche et al., 2003; Shimada and Gasser, 2007). Although ORC can interact with HP1, it is not known whether ORC functions in the assembly or maintenance of heterochromatin at mammalian telomeres (Pak et al., 1997; Shareef et al., 2003; Shareef et al., 2001).
Recent studies have revealed that a non-coding RNA containing UUAGGG repeats, referred to as TERRA, is expressed from telomeres in multiple species, including human and yeast (Azzalin et al., 2007; Luke et al., 2008; Schoeftner and Blasco, 2008). TERRA RNA can associate with telomere repeats, but the mechanism underlying this localization is not known. Several members of the nonsense mediated mRNA decay (NMD) factors regulate TERRA localization and telomere length homeostasis, but the mechanism appears to be distinct from the canonical NMD pathway (Azzalin et al., 2007). TERRA levels can be regulated by TRF1 through a mechanism involving an interaction between TRF1 and RNA polymerase II (Schoeftner and Blasco, 2008). TERRA RNA levels are developmentally regulated and inversely correlate with telomerase activity during stem cell differentiation (Schoeftner and Blasco, 2008). Other non-coding RNAs, like XIST and PIWI, have been implicated in gene regulation and chromatin organization (Buhler and Moazed, 2007; Wutz and Gribnau, 2007). The precise function of TERRA RNA in telomere regulation and the mechanisms regulating its localization to telomeres remains largely unknown. In this work, we provide evidence that TERRA RNA is recruited to telomeres through direct interactions with core Shelterin components, and that TERRA facilitates ORC recruitment and heterochromatin formation at telomeres.
TERRA RNA associates with telomere foci, but its precise function and mechanism of localization at telomeres is largely unknown. To identify proteins that interact with TERRA RNA, we used an unbiased RNA affinity assay with Raji cell nuclear extract and three different RNA oligonucleotide substrates (Fig. 1). Biotinylated RNA oligonucleotide substrates were coupled to streptavidin magnetic beads and incubated with nuclear extracts. Silver staining analysis of affinity purified protein eluates revealed a striking number of proteins that bound specifically to TERRA oligonucleotide sequence (UUAGGG)8, but not to control oligonucleotides (CACUGA)8 or (CCCUAA)8 (Fig. 1A). UUAGGG-specific binding proteins were identified by LC/MS/MS and then validated by Western blot for several known species (Fig. 1B). Among the known shelterin proteins, TRF2 and TRF1 were significantly enriched in the UUAGGG-associated proteins. We also identified telomerase-associated Dyskerin and Est1A by LC/MS/MS, and confirmed Dyskerin by Western blot (Fig. 1C). Origin Recognition Complex subunits ORC1, ORC2, and ORC4 were also enriched in the UUAGGG-associated proteins, while several other replication proteins, including MCMs and PCNA, were not detected. Surprisingly, the EBV encoded viral protein EBNA1 was detected by LC/MS/MS and was confirmed by Western blot. This was notable since our previous studies implicated EBNA1 in an RNA-dependent recruitment of ORC (Norseen et al., 2008) and an association of TRF2 with EBNA1 (Atanasiu et al., 2006). In addition to telomere repeat binding factors and ORC, LC/MS/MS revealed that numerous hnRNP proteins, the CpG methyl-binding protein (meCP2), and several DNA damage recognition proteins, including DNA PKcs, PARP1, Blm, topoisomerases I and II, were enriched among the UUAGGG-associated proteins (Fig. 1C).
TERRA RNA is known to associate with telomere repeat DNA foci (Azzalin et al., 2007). To determine if any of the known shelterin proteins may be involved in this process, we tested whether they interact with TERRA RNA in vivo using an RNA-immunoprecipitation (IP) assay (Fig. 2). To eliminate concerns for variations in antibody efficiency or accessibility, we used FLAG-tagged expression vectors for TRF1, TRF2, hRap1, POT1, and TPP1 (FLAG-TIN2 was not expressed to detectable levels and therefore not included in this analysis). The Shelterin components were expressed transiently in HCT116 cells and then assayed by RNA IP for TERRA binding (Fig. 2A). We found that FLAG-TRF1 and FLAG-TRF2 bound TERRA RNA, while FLAG-hRap1, FLAG-POT1, and FLAG-TPP1 did not recruit detectable amounts of TERRA. No C-rich RNA or control GAPDH RNA was precipitated in these IP assays (Fig. 2A, lower panels). The expression levels of input and immunoprecipitated (IP) proteins is shown in Fig. 2B, which indicates that all proteins were recovered by IP to similar levels. The TERRA RNA detection was shown to be insensitive to DNase I treatment (Fig. 2C and S1A) and sensitive to RNase A and RNase H treatment (Fig. 2D and E). These findings suggest that TRF1 and TRF2 interact more directly with TERRA RNA. However, we can not rule out that FLAG-epitopes or ectopic expression prevents detection of TERRA interaction with other components of the Shelterin complex.
To determine if TERRA interacts directly with endogenous TRF1 or TRF2 in vivo, we performed RNA ChIP assays with antibodies to TRF1, TRF2, hRap1, or control IgG (Fig. 2 D and E). These antibodies have been used successfully in ChIP assays for isolation of telomere repeat DNA(Deng et al., 2007). We found that antibodies to TRF1 and TRF2 ChIP the G-rich TERRA RNA (Fig. 2D, top panel), but do not ChIP control GAPDH (middle panel) or C-rich telomere-containing RNA (lower panel) in HCT116 cells (Fig. 2D), as well as in U2OS cells (Fig. 2E). Although TERRA RNA levels were several fold higher in U2OS cells, the percentage of total TERRA bound by TRF1 and TRF2 were higher in HCT116 cells compared to U2OS (Fig. 2F). These studies suggest that TRF1 and TRF2 associate directly with TERRA RNA in vivo in multiple cell types.
To better understand the function of TRF2 interaction with TERRA, we attempted to identify which domains of TRF2 were responsible for TERRA binding (Fig. 3). Several TRF2 deletion mutations were expressed as FLAG-tagged proteins in HCT116 cells and used for RNA IP assays (Fig. 3A). All of the FLAG-TRF2 proteins could be efficiently expressed and immunoprecipitated with anti-FLAG antibody (Fig. 3B, lower panel). Expression of full length TRF2 (FL) bound ~15% of the total TERRA RNA in the lysates. Deletions of TRF2 lacking the amino terminal domain (aa 2-45) reduced TERRA binding to ~2% of input (e.g. (45-500), (90-500), (245-500)) (Fig. 3A). On the other hand, TRF2 deletions lacking the carboxy-terminal myb/SANT domain (aa 454-500), completely abrogated TERRA RNA binding (e.g (2-250), (2-454), (45-454), (90-454)). All of the RNA recovered from the FLAG-IPs were shown to be sensitive to RNase A, and did not cross react with GAPDH or (TTAGGG)6 probes (Fig. 3A, lower panels). TRF2 mutants lacking the amino terminal GAR domain that were compromised for TERRA RNA binding could still bind telomere DNA in vivo (Fig. S1B), indicating that these mutants were expressed and localized properly in cells.
To determine if any TRF2 domains could bind to TERRA RNA in vitro, we generated and purified GST-TRF2 fusion proteins and assayed their ability to bind TERRA-containing oligonucleotide RNA using an RNA EMSA (Fig. 3C-E). GST-TRF2 fusion proteins were expressed and purified to near homogeneity (Fig. 3E). We found that GST-TRF2 bound efficiently to TERRA RNA oligonucleotide, while GST-alone had no detectable binding activity. GST-TRF2 (2-90), and GST-TRF2 (13-47) bound TERRA RNA with comparable efficiency to full-length TRF2, while GST-TRF2 (45-454) and (45-250) had substantially reduced binding activity (Fig. 3C). GST-TRF2 (454-500), and (245-454), which contain the myb DNA binding domain, also bound TERRA RNA. In contrast, none of the GST-TRF2 fusion protein bound the C-rich probe (Fig. 3D).
The nucleic acid binding specificity of TRF2 GAR and Myb domains were examined by oligonucleotide competitor challenge using EMSA (Fig. 3F). Both TRF2 GAR (aa 2-90) and myb domain (454-500) bound the UUAGGG RNA probe (TelrG) in the absence of competitor. Addition of a 20 fold molar excess of CCCUAA RNA oligonucleotide (TelrC) had no effect, while addition the TelrG oligonucleotide reduced binding of both TRF2 domains to ~10% of unchallenged levels. Addition of duplex RNA oligonucleotide (TelrG/TelrC) inhibited both TRF2 domains similar to the single strand TelrG. DNA-RNA duplexes containing TelrG (TelrG/TeldC) also inhibited binding of both TRF2 domains, while a duplex containing the DNA TeldG and the RNA TelrC was much less inhibitory. Addition of 100-fold molar excess of poly dIdC or poly dGdC inhibited TRF2 myb domain binding more effectively than the GAR domain (Fig 3F, lower panel). Similarly, DNA oligonucleotides TeldG, TeldC, or duplex TeldG/TeldC inhibited the myb domain more effectively than the GAR domain. Taken together, these data suggest TRF2 can bind TERRA RNA through multiple domains, but the amino terminal GAR domain (aa 13-47) has greater specificity for RNA than the carboxy-terminal myb domain (Fig. 3F).
The amino terminus of TRF2 has been implicated in ORC recruitment and in the stable maintenance of telomere T-loop structure (Deng et al., 2007; Wang et al., 2004). Our mapping studies also indicate that TRF2 amino terminal GAR domain (13-47) can bind TERRA RNA in vitro and is required for stable association of TERRA RNA in vivo (Fig. 3). We therefore assayed TERRA localization in cells expressing TRF2ΔB (aa 45-500) relative to full length TRF2. TERRA RNA was detected by RNA FISH with G-strand specific PNA probes and assayed for colocalization with FLAG-tagged TRF2 protein (Fig. S2). As expected, TERRA RNA colocalized with a high percentage of telomere foci marked by FLAG-TRF2 full-length protein (Fig. S2A). In contrast, TERRA RNA was more diffuse and formed fewer telomere foci in cells expressing FLAG-tagged TRF2ΔB (Fig. S2A and B). RNase A treatment completely eliminated the TERRA RNA signal, indicating that the FISH signal reflected RNA, and not single stranded DNA (Fig.S2C).
We also noted that TERRA RNA formed aberrant, bridge-like structures in U2OS cells expressing TRF2ΔB (Fig. S3). These RNA structures colocalized with FLAG-TRF2ΔB. The structures appear as stretched fibers between recently divided nuclei, similar to lagging chromosomes in mitotic figures, although they did not stain positive for DNA with DAPI. RNase A treatment eliminated the appearance of these structures and the TERRA RNA signal (Fig. S3B). While the precise molecular basis for these bridges is not yet clear, these findings suggest that TRF2 basic domain is required for the proper localization of TERRA RNA to telomere foci.
To explore the potential function of TERRA RNA, we used siRNA to deplete endogenous levels of TERRA. Two different siRNA were found to partially deplete TERRA RNA levels to ~40% of control levels and with no detectable effect on GAPDH mRNA levels, as measured by RNA dot blot and real-time PCR (Figs. 4A-C). Northern blot analysis revealed that siRNA caused a significant reduction in TERRA abundance, as well as in TERRA length (Fig. 4D). siRNA depletion also caused a significant loss of telomere-associated TERRA RNA foci (Fig. S4). The effect of TERRA RNA depletion on cellular proliferation was first measured by cell number and viability after transfection of siRNA (Fig. 4E). We found that siTERRA-1 and -2 caused a ~2.4 fold decrease in viable cell number at 96 hrs post-transfection. Cell cycle profiles suggested that siTERRA reduced G2 population and increased the subG1 population relative to siRNA control transfected cells (Fig. 4F). This also correlated with an ~5 fold increase in telomere dysfunction-induced foci (TIF) based on the increase colocalization of 53BP1 with TRF2 foci in siTERRA-depleted cells (Fig. 4G and H). A similar increase in γ-H2AX colocalization with TRF2 foci was observed in siTERRA depleted cells (Fig. S5).
The effects of TERRA RNA depletion on telomere DNA structure was also examined by telomere restriction fragment length analysis (Fig. S6) and by metaphase FISH (Fig. 5). TERRA RNA depletion caused an ~1.4 fold increase in telomere repeat signal intensity, but no significant change in average telomere length (Fig. S6A and B, left panels). A similar increase in telomere restriction fragment signal intensity was observed when TERRA RNA was depleted by two different siRNAs and by an anti-sense phosphorothioate oligonucleotide, suggesting that the effect was independent of the method of RNA depletion (Fig. S6B and C). More dramatic telomere aberrations were observed by metaphase chromosome FISH (Fig. 5A-C). We found that HCT116 cells treated with siTERRA resulted in a significant increase in telomere defects, including telomere free ends (TFE), telomere doublets (TD), telomere double minutes (TDMs) and chromatid duplications (CD).
To investigate the potential mechanism through which TERRA RNA depletion causes telomere dysfunction, we examined the changes in telomere-associated chromatin. Previous studies from our lab found that ORC localizes to telomeres through an interaction with TRF2 (Deng et al., 2007). Both ORC and TERRA RNA have been implicated in telomere-associated heterochromatin formation (Schoeftner and Blasco, 2008). We therefore tested whether TERRA RNA depletion led to a change in ORC binding or in histone modifications at telomere repeats (Fig. 5D-E, and Fig. S10). We found that siRNA depletion of TERRA RNA led to a ~46% loss of ORC2 binding and ~72% loss of H3 dimethyl K9 at telomere repeats, while no significant changes were observed in acetylated H3 (AcH3). TERRA depletion did not alter TRF2 binding and had no detectable effect on H3 K9 methylation at Alu repeats (Fig. 5E, lower panel).
To further explore the role of TERRA in telomere heterochromatin formation, we used RNA-ChIP assay to determine if TERRA was in close proximity to any known heterochromatic proteins or histone modifications (Fig. 6). We assayed TRF2, acetylated histone H3 (H3Ac), histone H3 dimethyl K4 (H3 K4me2), H3 dimethyl K9 (H3 K9me2), H3 trimethyl K9 (H3 K9 me3), HP1α, and ORC1 for interaction with TERRA in RNA-ChIP. As expected, TERRA was enriched in TRF2 RNA-ChIP. Remarkably, TERRA was also highly enriched in H3 K9me3 RNA-ChIP. TERRA was also detected in HP1α, and ORC1 RNA-ChIP, but not in control IgG or H3Ac. These findings suggest that TERRA RNA is in close association to heterochromatin-associated proteins and modified histones. To determine which, if any, isoform of HP1 interacts with TERRA, we compared the ability of FLAG-tagged HP1α, β, and γ to IP TERRA RNA (Fig. 6C-E). We found that HP1α and β bound TERRA RNA, while control vector and HP1γ did not bind TERRA in the RNA-IP assay. This suggests that TERRA may associate with specific types of heterochromatin that form at telomeres.
ORC and HP1α have been implicated in heterochromatin formation and may be part of the mechanism through which TERRA facilitates heterochromatin formation at telomeres (Leatherwood and Vas, 2003). We therefore tested whether ORC2 depletion causes a similar loss of histone H3 K9 methylation at telomeres as did TERRA depletion. ORC was depleted using previously validated siRNA against ORC2 (Deng et al., 2007) (Fig. 7A). We found that ORC2 depletion caused a 58% loss of H3K9 methylation at telomere RNA, with no detectable effect at control Alu repeats and no alteration of H3 K4 methylation (Fig. 7B). This indicates that ORC subunits also contribute to the enrichment of H3 K9me3 and heterochromatin formation at telomeres.
Previous studies have shown that TRF2 can recruit ORC through its amino terminal GAR domain (Atanasiu et al., 2006; Deng et al., 2007; Tatsumi et al., 2008). To determine if TERRA RNA contributes to ORC recruitment by TRF2, we first tested whether the interaction was sensitive to RNase A treatment (Fig. 7C). GST, GST-TRF2 (full length), and GST-TRF2 (2-90) were incubated with nuclear extracts and then treated with RNase A or DNase I or mock digestion. Recruitment of ORC subunits was monitored by Western blot. We found that ORC1, ORC2, and ORC4 were specifically recruited to GST-TRF2 and GST-TRF2 (2-90), but not to GST control proteins. We found that RNase A, but not DNase I treatment, reduced ORC1 binding to GST-TRF2 and GST-TRF2 (2-90). RNase A treatment did not cause a loss of ORC2 or ORC4 in these experiments. This may be due to additional interactions between TRF2 and ORC subunits that are not RNA-dependent. The activity of the DNase I and RNase A were verified on pure DNA or RNA substrates in parallel reactions (Fig. S7). These findings indicate that ORC1 binding to TRF2 is RNase A sensitive.
Previous studies indicated that ORC1 (aa 200-511) bound to TRF2 (aa 13-47) (Deng et al., 2007). We therefore tested whether ORC1 (aa 200-511) could interact with TRF2 (13-47) and TERRA RNA oligonucleotide to form a stable trimeric complex (Fig. 7E and S8). As shown in Fig. 3C, GST-TRF2 (13-47) bound specifically to the (UUAGGG)6 probe in EMSA. ORC1 was expressed and purified in three hexa-histidine tagged fragments (1-200, 200-511, or 511-861) (Fig. 7D). We found that ORC1 (1-200) and (512-861) bound weakly to TERRA by itself, but did not enhance the binding or alter the mobility of GST-TRF2 (13-47) bound to TERRA (Fig. 7E). In contrast, addition of ORC1 (200-511) strongly enhanced GST-TRF2 (13-47) binding to TERRA and produced a modest shift in the bound complex (O-T), as well as a weaker, slower migrating species (O-T*). GST-ORC1 fragments were also capable of stimulating GST-TRF2 (2-90) binding to TERRA and produced a similar alteration in the mobility of the bound complex (Fig. S8). The TRF2-ORC1 complex formed on TERRA RNA, and on a related G-rich RNA capable of forming a G-quadruplex structure, but did not form on several other G-rich and C-rich RNA oligonucleotides incapable of forming G-quadruplex structures (Fig. S9). Taken together, these data suggest that TERRA can mediate interactions between TRF2 and ORC1, and that this interaction is important for H3 K9me3 and HP1-associated heterochromatin formation at telomeres (Fig. 7F).
Abnormal regulation of telomere length, structure, and replication are hallmarks of cellular ageing and cancer. In this work, we have found that the TERRA RNA is required for telomere structural stability and heterochromatin formation. Our data indicates that TERRA is a structural component of the telomere nulceoprotein complex, and is bound to the telomere through physical interactions with Shelterin components TRF1 and TRF2 (Figs. (Figs.11--3).3). TERRA RNA depletion caused telomere dysfunction, including an increase in metaphase chromosome aberrations, and an increase in 53BP1 and γH2AX-associated telomere foci (Figs. (Figs.44 and and5).5). Our work focused on the role of TRF2 recruitment of TERRA RNA through its amino terminal GAR domain. This domain has been implicated in ORC recruitment and telomere T-loop stability. Our findings indicate that TERRA RNA can interact with both TRF2 GAR domain and the region of ORC1 that associates with TRF2 GAR, and suggests that TERRA mediates and stabilizes this interaction (Fig. 7). Furthermore, TERRA RNA was associated with several heterochromatin marks, including H3 K9me3 and HP1 isoforms known to accumulate at constitutive heterochromatin (Fig. 6). We propose that TERRA functions to facilitate heterochromatin formation at telomeres, in part, through its direct interactions with ORC1 and TRF2 GAR domain.
Localization of TERRA at telomeres is likely to be mediated by interactions with one or more components of the Shelterin complex. We found that at least three domains of TRF2 could interact with TERRA RNA, including the amino-terminal basic domain (GAR). This interaction was demonstrated in vivo using RNA ChIP (Fig. 2) and in vitro using RNA-affinity purification and EMSA (Figs. (Figs.11 and and3).3). While the TRF2 myb domain could also interact with TERRA, this interaction was competed away with duplex DNA containing telomere repeats (Fig. 3F). The TRF2 GAR domain has been implicated in the recruitment of ORC and also in the binding to DNA junctions that form at T-loops (Atanasiu et al., 2006; Fouche et al., 2006). The GAR-domain resembles the RGG-domains commonly observed in RNA-binding proteins. The RGG-domain of the fragile X mental retardation protein (FMRP) has been shown to bind to G-quadruplex RNA (Darnell et al., 2004). Telomere G-rich DNA and RNA can form G-quadruplex structures, and G-quadruplexes have been implicated in telomere and telomerase regulation (Burger et al., 2005; Xu et al., 2008). Our data indicates that the TRF2 GAR has a relatively high affinity for the G-rich RNA capable of forming G-quadruplex structures, suggesting that the recognition depends more on RNA structure than a specific sequence. This structure-specific binding of the TRF2 GAR domain may also explain its role in binding DNA junctions. Thus, TRF2 GAR may stabilize several complex DNA and RNA structures that form at telomeres.
Our work implicates TERRA in mediating interactions between TRF2 and ORC1. RNA affinity purification revealed that both TRF2 and ORC1 could be recruited specifically to TERRA RNA oligonucleotides (Fig. 1B). We also show that TRF2 recruitment of ORC1 is sensitive to RNase A treatment, but not to DNase I treatment (Fig. 7C). In addition, we demonstrate that purified ORC1 (aa200-511) can bind weakly to TERRA and enhance TRF2 binding to TERRA oligonucleotides in EMSA (Fig. 7E and S8). We propose TERRA partially mediates and stabilizes the interaction between TRF2 and ORC1 at telomeres (Fig. 7F). RNA-dependent recruitment of ORC has been observed at the EBV origin of plasmid replication (OriP), which is bound by TRF2 and by the viral encoded protein EBNA1 (Atanasiu et al., 2006; Norseen et al., 2008). EBNA1 contains several RGG-domains, which are known to bind RNA and to confer replication and plasmid segregation activity (Lindner and Sugden, 2007; Snudden et al., 1994). EBNA1 was found among the TERRA RNA affinity bound proteins (Fig. 1), but it is not known whether TERRA RNA mediates EBNA1 functions at OriP. Although TERRA localizes primarily to telomere repeats, it is possible that TERRA may also function at chromosomal locations other than telomeres. Indeed, non-telomeric TERRA RNA was found to partially colocalize with Xist RNA (Schoeftner and Blasco, 2008) .
The functional significance of TERRA was investigated by siRNA and anti-sense depletion of TERRA RNA (Figs. (Figs.4,4, ,5,5, S4, S5, and S6). Although the TERRA depletion was only partial (~40% of control), it caused a significant decrease in average TERRA RNA length (Fig. 4D). TERRA depletion and shortening caused several structural aberrations at telomeres, which also correlated with an increase in 53BP1 and γH2AX positive TIFs. Several striking structural aberrations were observed by metaphase FISH, including telomere free ends and telomere foci duplications (Fig. 5A-C). The telomere duplications may partially explain the observed increase in telomere signal intensity detected by Southern blot analysis of restriction fragments (Fig. S6). Given that ORC plays an essential role in DNA replication initiation, we speculate that that TERRA regulates aspects of telomere DNA replication, perhaps by direct interactions with ORC1.
Our findings most clearly implicate TERRA in telomere heterochromatin formation and maintenance. We observed that TERRA siRNA depletion caused a loss of H3 K9me3 and ORC association with telomere repeat DNA (Fig. 5D-E, and S10). A similar loss of H3 K9me3 was observed with siRNA depletion of ORC2 (Fig. 7B). These findings lead us to propose that TERRA and ORC cooperate to maintain telomeric heterochromatin. ORC has also been implicated in heterochromatin formation and is known to associate with HP1 in mammalian cells (Prasanth et al., 2004b). TERRA RNA was found to be closely associated with H3 K9me3 and HP1α and β by RNA-IP assay (Fig. 6). H3 K9me3, HP1α, and β, are well-established components of constitutive heterochromatin and are likely to be stabilized at telomeres by their association with TERRA RNA (Blasco, 2007). Other studies have implicated non-coding RNAs, including XIST and PIWI, in HP1 recruitment and heterochromatin formation at other genetic loci (Wutz and Gribnau, 2007; Yin and Lin, 2007). We suggest that TERRA RNA facilitates heterochromatin formation at telomeres by stabilizing multiple interactions between heterochromatin factors at telomere repeats. Our data indicates that the TRF2-TERRA-ORC1 interactions are critical for maintaining heterochromatin proteins (HP1) and histone marks (H3 K9me3) at telomeres. We conclude that TERRA is a structural component of the Shelterin complex that helps maintain heterochromatin and DNA stability at telomere repeats.
The on-target siRNA were synthesized and purified from Dharmacon and include the following target sequences: siTERRA-1 (5′-NNAGGGUUAGGGUUAGGGUUA-3′), siTERRA-2 (5′-NNGGGUUAGGGUUAGGGUUAG-3′), siORC2 (5′-GAAGAAACCUCCUAUGAGAUU-3′), and control (Dharmacon nontargeting siRNA #D-001810-01). TERRA or control antisense phosphorothioate was synthesized and purified from Integrated DNA Technologies (IDT) and include sequences 5′-A*A*C*CCTAACCCTAACCCT*A*A*C-3′ or 5′-A*A*C*AATAGACATACAGACAT*A*A*C-3′, respectively.
DNA ChIP assays were performed as described previously (Deng et al., 2002). RNA ChIP assays were performed as described (http://www.epigenome-noe.net/researchtools/protocol.php?protid=28) with minor modifications. Briefly, 1 × 107 cells were cross-linked for 10 mins by the addition of formaldehyde to a final concentration of 1%. Glycine was added to a final concentration of 125mM to quench cross-linking, and the cells were collected at 1500 rpm for 5 mins. The cell pellet was washed with cold PBS, resuspended in 500 μl of Buffer A (5mM PIPES, pH 8.0, 85mM KCl, 0.5% NP40, protease inhibitors cocktail, 50U/ml SUPERase•in), and placed on ice for 10 minutes. The nuclei was collected by centrifugation at 5000 rpm for 5 mins at 4 °C, washed in Buffer A without NP-40, and resuspended in 500 μl of Buffer B (1% SDS, 10mM EDTA, 50mM Tris-HCl pH 8.1, protease inhibitors cocktail, 50U/ml SUPERase•in) on ice for 10 mins. The lysates were sonicated, cleared by centrifugation to remove insoluble materials, and diluted 10 fold into IP Buffer (0.01% SDS, 1.1% Triton X-100, 1.2mM EDTA, 16.7mM Tris pH 8.1, 167mM NaCl, protease inhibitors cocktail, 50U/ml SUPERase•in) for IP reaction (~2 × 106 cells per IP) at 4 °C overnight. Each immune complex was washed five times (1 ml wash, 10 mins each) in ChIP related wash buffer at 4 °C, eluted by addition of 200μl Elution buffer (10mM Tris, pH 8.0, 5mM EDTA, 1% SDS, 50U/ml SUPERase.in) at 70 °C for 20 mins, and the elutes were placed at 70 °C for 1-2 hrs to reverse cross-linking. RNA was isolated with Trizol reagent, and resuspended in 50μl of DEPC-treated water followed by DNaseI treatment and further analysis using the method described.
Biotinylated RNA probes were synthesized and purified from IDT, and include following sequences: 5′-/Bio/-(UUAGGG)8, 5′-/Bio/-(CCCUAA)8, and control 5′-/Bio/-(CACUGA)8. Raji cell nuclear extracts binding with biotinylated RNA bound to streptavidin was essentially the same as binding with biotinylated DNA as described previously (Deng et al., 2002), except that 50U/ml SUPERase.in (Ambion) was added in the reaction throughout the procedures. Briefly, nuclear extracts in D150 buffer (20mM HEPES pH 7.9, 20% glycerol, 0.2mM EDTA, 150mM NaCl, 0.05% NP-40, 1mM PMSF, and 10mM mercaptoethanol) supplemented with protease inhibitor cocktails (Sigma) and SUPERase.in (Ambion) were precleared with control (CACUGA)8 RNA-coupled Streptavidin beads (Dynal) twice for 30 mins each with rotation at 4 °C. The cleared nuclear extracts were further incubated with (UUAGGG)8 or ((CCCUAA)8 RNA-coupled Streptavidin beads for 1 hr with rotation at 4 °C. The bounded materials were washed 5 times with D150 buffer, eluted with 100 μl 1 × B&W Buffer (5 mM Tris pH 7.5, 1 M NaCl, 0.5 mM EDTA) for 15 mins at 4 °C, and elutes were concentrated by TCA precipitation prior to SDS-PAGE and Western Blotting analysis.
RNA oligonucleotides were radiolabeled with γ-ATP and T4 polynucleotide kinase and incubated with purified GST-fusion proteins or hexa-histidine tagged proteins as indicated. Protein (~200 ng) and RNA (~1 fmol) complexes were formed in a 10 μl reaction buffer containing 20 mM HEPES (pH 7.9), 18% glycerol, 1 mM EDTA, 5 mM β-mercapthoethanol, 0.05%NP40, 2.5 mM MgCl2, 100 μg/ml BSA, and 100 mg/ml tRNA at 25°C for 30 min and then loaded directly onto 1.5% horizontal agarose gels containing 0.5 × TBE, essentially as described previously (Atanasiu et al., 2005).
TIF assay was performed as described (Dimitrova and de Lange, 2006) with some modifications. Briefly, cells grown on coverslips were fixed for 15 min in 2% paraformaldehyde at RT, followed by 15 min in 100% methanol at −20 °C. After rehydration in PBS for 5 min, cells were incubated for 30-60 min in blocking solution (1mg/ml BSA, 3% fetal bovine serum, 0.1% Triton X-100, 1 mM EDTA in PBS) before immuno-staining. Primary antibodies were prepared in blocking solution as following dilutions: 53BP1 (1:40), γH2AX (1:50), and rabbit polyclonal TRF2 (1:1000). Nuclei were counterstained with 0.1 μg/ml DAPI in blocking solution and slides were mounted with VectorShield (Vector Laboratories, Inc). Cells with five or more 53BP1 or γH2AX foci colocalizing with TRF2 foci were scored as TIF positive.
We thank Songyang Zhou (Baylor College of Medicine, TX) for kindly providing the POT1 and TPP1 plasmids; HongZhuang Peng and Frank Rauscher III (The Wistar Institute, PA) for providing FLAG-tagged HP1α, HP1β, and HP1γ expression plasmids; Andy Sydner (The Wistar Institute) for 53BP1 antibody and helpful comments. We acknowledge Ruchira Ranaweera for her assistance and the contribution of the Wistar Cancer Center Core Facilities. This work was funded by grants from NIH (CA093606) to PML, a Leukemia Lymphoma Society Special Fellow Award to Z.D., and the UPENN Training Grant in Tumor Virology (T32 AI07324-17) to J.N.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Additional Methods are included in Supplmental Materials.