Purification and mass spectrometric identification of human TERT complex components
To characterize the size of the telomerase holoenzyme complex, we performed glycerol gradient sedimentation analyses using extracts from HeLa cells and extracts from HeLa cells stably expressing Flag-TERT from a retroviral promoter. Endogenous telomerase activity, measured by telomerase repeat amplification protocol (TRAP), and endogenous TERC, assayed by northern blot, co-sedimented in a size range consistent with previous estimates of 1-2MDa () (Schnapp et al., 1998
; Xin et al., 2007
). Flag-TERT similarly sedimented as a large complex, suggesting that retrovirally expressed TERT associates stably with other factors that have potential relevance for telomerase function. To identify novel protein components of the telomerase complex, we purified TERT protein complexes using a modified TAP tag approach. TERT was fused at its N terminus with a dual affinity tag consisting of a protein A moiety and three HA epitopes separated by a TEV protease cleavage site (AH3). To determine whether our tagged TERT protein was functional, we expressed AH3-TERT by retroviral transduction in primary human fibroblasts that lack TERT expression and senesce after many population doublings due to progressive telomere shortening. Tagged TERT reconstituted telomerase activity, lengthened telomeres, and immortalized human fibroblasts, indicating that AH3-TERT is fully active at telomere (Figure S1
Pontin and reptin co-purify with TERT through dual affinity purification and are components of a large TERT complex
For telomerase complex purifications, HeLa S3 cells expressing AH3-TERT were grown in suspension cultures. In optimizing conditions for extracting TERT protein, we found that a detergent-based lysis procedure solubilized approximately 75% of AH3-TERT (referred to here as “S100” extract) and salt extraction of the nuclear pellet liberated the remaining 25% of AH3-TERT (data not shown). To ensure a thorough analysis of TERT-associated proteins, TERT complexes were purified from both S100 extracts (n=2) and nuclear extracts (n=2). AH3-TERT was purified on rabbit IgG resin, eluted specifically with TEV protease, captured again with anti-HA resin, then eluted and analyzed by SDS-PAGE (Figure S1E,F
). After staining with Coomassie blue, protein bands were excised, digested and analyzed by nanoflow reversed-phase LC-MS/MS (see Materials and Methods). Mass spectrometric analysis of the 127kDa band identified between 23 and 57 unique peptides matching the TERT open reading frame (). In addition, at least six co-purifying polypeptides were enriched in each TERT purification compared to negative controls ().
Mass spectrometric analysis of the 50kDa band, one of the most prominent TERT-associated proteins by SDS-PAGE, revealed that it comprised two independent proteins, the related ATPases pontin and reptin. Pontin and reptin peptides were recovered in each of four independent experiments, spanning more than 50% of the pontin and reptin open reading frames (). To study their role in the telomerase complex, we generated highly specific polyclonal antibodies to pontin and reptin. We found that pontin and reptin participate in large complexes, which overlap the size distributions of both endogenous telomerase and Flag-TERT by glycerol gradient sedimentation (). Analysis of extracts from Flag-TERT cells fractionated by glycerol gradient sedimentation revealed that endogenous pontin and reptin were stably associated with Flag-TERT (). These results show that endogenous pontin and reptin interact with Flag-TERT in a large molecular weight complex. Based on these data, we investigated a potential role for pontin and reptin in telomerase function using biochemical and genetic approaches.
Pontin and reptin interact with endogenous TERT protein
Pontin and reptin are well conserved AAA+ ATPases (for A
ssociated with various cellular a
ctivities) and have been identified in chromatin remodeling complexes, as co-factors for the transcriptional regulators c-myc and β-catenin and as proteins that interact with small nucleolar RNA (snoRNA) complexes (Newman et al., 2000
; Gallant, 2007
). Pontin and reptin physically interact and are thought to function together, although there is some evidence that pontin and reptin can act independently or in an opposing fashion (Rottbauer et al., 2002
; Kim et al., 2005
). Pontin-specific shRNA sequences not only depleted pontin, but also co-depleted reptin (). Similarly, shRNA sequences directed against reptin co-depleted pontin, indicating that these proteins mutually depend on their interaction for stability. We also found that pontin and reptin biochemically co-depleted one another in co-transfection assays (Fig. S2A
). Furthermore, HA-TERT interacted with Flag-pontin, but not with Flag-reptin in co-transfection experiments (Fig. S2B, S2C
). Interestingly, the addition of pontin facilitated interaction between Flag-reptin and HA-TERT by co-immunoprecipitation, indicating that reptin is recruited into a TERT complex through bridging pontin molecules (Fig. S2B
). Association of endogenous pontin and reptin with Flag-TERT stably expressed in HeLa cells was resistant to treatment with DNase I, ethidium bromide, and RNase A, indicating that these interactions are not mediated by co-purifying nucleic acids (Figure S2D
). Together, these data show that pontin and reptin are interdependent and are recruited into TERT complexes through an association between TERT and pontin.
Pontin and reptin interact with endogenous TERT and TERC
We reasoned that endogenous TERT should be detectable in purified complexes of pontin and reptin. However, reliable detection of the endogenous TERT protein by western blot has proven difficult due to both the low abundance of TERT and a dearth of antibodies that recognize TERT protein (Wu et al., 2006
). We sought to address these significant technical hurdles by developing quantitative methods for immunoprecipitating pontin and reptin and by enhancing our ability to detect TERT through generation of high affinity polyclonal antibodies. In attempting to overexpress pontin, we noticed that Flag-pontin under-accumulated relative to the endogenous protein (, compare lanes 1 and 5). However, expression of an shRNA-resistant form of Flag-pontin followed by transduction of an shRNA retrovirus targeting the endogenous pontin protein resulted in accumulation of Flag-pontin to endogenous levels (, compare lanes 1 and 3). Reptin was substituted with Flag-reptin by an analogous strategy. (). We refer to these cell lines as Flag-pontin+shRNA
, respectively. Substituting pontin and reptin with tagged versions allows for quantitative immunoprecipitation using well-characterized monoclonal antibodies directed against the tag, a strategy routinely employed in yeast.
To study the endogenous TERT protein, we generated rabbit polyclonal antibodies directed against a bacterially expressed TERT polypeptide, followed by affinity purification of anti-TERT antibodies on the cognate antigen (see Materials and Methods). In extensive testing, these anti-TERT antibodies readily recognized stably expressed TERT by western blot and immunofluorescence, and specifically immunoprecipitated both TERC by northern blot and telomerase activity by TRAP assay (Figure S4
). To determine if endogenous TERT associates with endogenous levels of pontin and reptin, extracts from Flag-pontin+shRNA
cells were immunoprecipitated with anti-Flag resin and assayed by western blot with anti-TERT antibodies. Anti-TERT antibodies reproducibly detected a polypeptide consistent with TERT's molecular mass of 127 kDa in pontin and reptin immunoprecipitates, but not in negative controls (). Anti-TERT immunoreactive bands were significantly diminished by two TERT-specific shRNAs and were recognized by affinity purified antibodies from two independent rabbits (). As an independent measure of the association of endogenous telomerase with pontin and reptin, we found that TERC was specifically associated with pontin and reptin by immunoprecipitation-northern blot and this interaction was largely TERT-dependent (). Together, these data show that pontin and reptin interact with endogenous TERT and TERC, the catalytic core of telomerase.
Essential role for pontin and reptin in TERC accumulation
Based on these data showing that TERT associates with pontin and reptin at the endogenous level, we investigated how loss of pontin and reptin affected telomerase activity using pontin shRNA that depletes both pontin and reptin. Whole cell lysates prepared from HeLa cells transduced with shRNA vectors targeting pontin were analyzed for telomerase activity by TRAP assay. Treatment with pontin shRNA severely diminished telomerase activity to 10-20% of wild-type levels (). To understand why telomerase activity was so significantly reduced following pontin and reptin depletion, we assessed TERC levels by northern blot. Strikingly, three different pontin shRNA sequences markedly reduced TERC RNA to approximately 25% the level in negative controls (). Pontin and reptin were previously shown to be required for accumulation of the human U3 small nucleolar RNA (snoRNA) (Watkins et al., 2004
). As a control, we confirmed that pontin and reptin depletion reduced U3 snoRNA levels, but did not affect the amount of U1, a small nuclear RNA involved in mRNA splicing ().
Pontin and reptin are required for telomerase activity and for TERC accumulation
To further address the specificity of the pontin knockdown effect on TERC, we asked if TERC levels were rescued by retroviral transduction with an shRNA-resistant Flag-pontin. Pontin knockdown had no effect on TERC levels in HeLa cells expressing shRNA-resistant Flag-pontin (, lane 6), demonstrating that loss of TERC in cells treated with pontin shRNA is not due to off-target effects. Correspondingly, we found that U3 levels were also restored. The ability to rescue the defect in TERC levels allowed us to ask if pontin's ATPase activity is required for TERC accumulation. TERC levels were significantly reduced in HeLa cells expressing shRNA-resistant Flag-pontinD302N following treatment with pontin shRNA, indicating that pontin ATPase activity is required to maintain wild-type levels of TERC (, lane 7). To understand if pontin depletion affects only a pool of TERC molecules that are free of TERT or if the TERC associated with TERT is also reduced, we measured the amount of TERC bound by TERT in cells treated with pontin shRNA. Immunoprecipitates of endogenous TERT in HeLa cells or ectopically expressed TERT in HeLa-Flag-TERT cells contained significantly less TERC in cells treated with pontin shRNA compared to negative controls (). Together, these results show that pontin and reptin are essential for telomerase activity and for TERC accumulation through a mechanism that requires ATPase function.
Pontin and reptin interact with dyskerin, forming a complex required for telomerase RNP assembly
The loss of TERC that occurs with depletion of pontin and reptin was particularly striking because it is reminiscent of the reduction in TERC levels seen in those dyskeratosis congenita patients with mutations in dyskerin. Dyskerin binds TERC at its 3′ H/ACA motif, a structural element required for TERC stability. With mutation or inactivation of dyskerin, TERC levels diminish, resulting in decreased telomerase activity (Mitchell et al., 1999b
). The similar requirements for pontin, reptin and dyskerin in maintenance of TERC levels suggested a functional relationship among these proteins. To determine whether dyskerin interacts with pontin or reptin, we first performed co-transfection assays. HA-dyskerin bound both Flag-pontin and Flag-reptin, but not the negative control Flag-BAF57 in co-transfection assays (). Furthermore, when transfected alone, Flag-dyskerin efficiently complexed with endogenous pontin and reptin (). To understand the interaction between dyskerin and telomerase, TERC and/or TERT were co-expressed with Flag-dyskerin. Neither exogenous TERT nor exogenous TERC altered the amount of pontin or reptin associated with Flag-dyskerin. Co-expression of HA-TERT with Flag-dyskerin resulted in a minor amount of HA-TERT in Flag-dyskerin immunoprecipitates. However, co-expression of TERC and TERT dramatically enhanced the amount of TERT associated with dyskerin, consistent with recruitment of TERT into dyskerin complexes through their common interaction with TERC () (Mitchell et al., 1999b
). Consistent with these findings, immunoprecipitation of stably expressed Flag-TERT pulled down endogenous dyskerin in an RNase A-sensitive manner (, lanes 11, 12).
Pontin and reptin interact with dyskerin and are required for dyskerin accumulation
To understand the nature of this complex at the endogenous level, we used Flag-pontin+shRNA cells and Flag-reptin+shRNA cells to immunoprecipitate pontin and reptin complexes (). Endogenous dyskerin was readily detected in pontin and reptin complexes, indicating that dyskerin interacts with both pontin and reptin at the endogenous level. We further noted that the pontin-reptin-dyskerin association was not sensitive to RNase A treatment, indicating that these contacts are not mediated by co-purifying TERC. Thus, TERC connects TERT to dyskerin, while pontin and reptin bind both TERT and dyskerin through protein-protein contacts.
To study the potential interdependence of dyskerin with pontin and reptin, we depleted each protein using RNA interference and assessed protein levels by western blot. Dyskerin was depleted to varying degrees with three independent shRNA sequences, but no reciprocal effect on levels of either pontin or reptin was detected (, lanes 3, 4). In contrast, knockdown of pontin and reptin led to a significant reduction in steady-state dyskerin levels (, lanes 8-10). Thus, dyskerin depends in part on pontin and reptin for expression at wild-type levels, highlighting the critical importance of pontin and reptin in dyskerin function.
Pontin interacts directly with dyskerin and TERT
Our data show that pontin and reptin associate with both known protein constituents of telomerase, TERT and dyskerin. To better characterize the nature of these interactions, we mapped the domain of TERT that mediates the interaction with pontin and reptin. We assessed binding of endogenous pontin and reptin by immunoprecipitating a series of Flag-tagged amino terminal or carboxy terminal deletion fragments of TERT in 293T cells (). Although fragments containing the C-terminal domains of TERT did not co-immunoprecipitate pontin and reptin, extending these fragments into the RT domain conferred pontin and reptin binding ability (, IP lanes 6-11). Similarly, extending the N-terminal fragment into the RT domain enabled a more efficient interaction with pontin and reptin (, IP lanes 2-4). Together, these results implicate the central RT domain in binding pontin and reptin.
Pontin interacts directly with dyskerin and TERT
Recombinant TERT has been extensively studied using rabbit reticulocyte lysates (RRL) as an expression system. Remarkably, we found that Flag-TERT as well as Flag-dyskerin expressed in RRL co-immunoprecipitated rabbit pontin intrinsic to the lysate, whereas negative controls Flag-GFP and Flag-BAF57 did not (). Expression of a subset of the deletion mutants of TERT in RRL showed the same requirement for the RT domain in binding pontin seen in transfected cells ().
To study interactions in the absence of other eukaryotic proteins, we expressed MBP-dyskerin, two MBP-TERT fragments and Flag-(His)6-pontin in bacteria (). Purified Flag-(His)6-pontin was incubated with immobilized MBP-dyskerin or MBP-TERT in a pull-down assay and bound pontin was assayed by Flag western blot. Recombinant pontin bound recombinant MBP-dyskerin and both MBP-TERT proteins, indicating that pontin interacts directly with both dyskerin and TERT (). Furthermore, the fact that pontin bound MBP-TERT-D provides additional evidence that the RT domain of TERT mediates the interaction with pontin.
To understand which region of pontin mediates interaction with dyskerin, we expressed pontin domains based on the X-ray crystal structure of pontin (Matias et al., 2006
). Domains I and III interact to form the hexamer ring, whereas domain II projects downward from the plane of the ring. Expression of Flag-tagged pontin fragments () with HA-dyskerin in 293T cells revealed that pontin-C and pontin-D fragments bound HA-dyskerin like full-length pontin. In contrast, pontin-B comprising domain II did not bind HA-dyskerin (). Together, these data show that pontin directly interacts with dyskerin and with TERT's RT domain, and suggest that the interaction surface of pontin required for binding dyskerin resides in its C-terminal domain, the region containing domain III and the Walker B motif.
Pontin and reptin interact with TERT in S phase
Dynamic regulation of TERT during the cell cycle: the TERT-pontin-reptin complex peaks in S phase
Although telomerase lengthens short telomeres preferentially during S phase of the cell cycle in yeast (Marcand et al., 2000
), and human telomerase localizes to telomeres during S phase (Jady et al., 2006
; Tomlinson et al., 2006
), it is unclear what mechanisms ensure that human telomerase acts on telomere ends in S phase. To determine whether pontin and reptin contribute to cell cycle control of endogenous telomerase, we synchronized Flag-reptin+shRNA
cells at the G1/S transition using double thymidine blockade. Cells were released and harvested at two hour intervals, and synchrony was monitored by measuring DNA content by propidium iodide staining and flow cytometry ().
Immunoprecipitation of Flag-reptin showed that neither the total reptin pool nor the amount of pontin or dyskerin bound to reptin varied through the cell cycle. In marked contrast, the amount of TERT associated with reptin peaked in S phase. The amount of TERT bound to reptin was three fold higher in S phase than in G2, M or G1 (). To control for potential confounding effects from thymidine blockade, we repeated the synchronization experiment and carried the HeLa-Flag-reptin+shRNA
cells through two cell cycles. We found that the association of reptin with endogenous TERT peaked in consecutive S phases, closely matching the pattern of expression of the S-phase marker PCNA (). Endogenous TERT also preferentially associated with pontin in Flag-pontin+shRNA
cells and the S-phase specific interaction among TERT, pontin and reptin was also seen in experiments employing a thymidine-aphidicolin protocol for synchronization (Figure S5
). These data provide evidence for dynamic regulation of telomerase during the cell cycle and indicate that TERT's association with pontin and reptin peaks in S phase, which may reflect cell cycle regulation of total TERT protein and/or assembly of telomerase in the phase of the cell cycle during which it must act on telomeres.
TERT exists in at least two complexes
Our observations that 1) pontin and reptin are required for telomerase biogenesis and 2) the pontin-reptin-TERT complex peaks during each S phase led us to hypothesize that TERT complexes are dynamic in nature and that pontin and reptin may be involved in cyclical telomerase assembly. We asked to what extent pontin and reptin associate with “active” telomerase particles as measured using the TRAP assay. Flag antibody immunoprecipitation of lysates from Flag-pontin+shRNA, Flag-reptin+shRNA or Flag-dyskerin+shRNA cells, in which pontin, reptin, or dyskerin was replaced by a Flag-tagged version at the endogenous level, quantitatively depleted each Flag-tagged protein from the lysate (). TRAP assays performed on lysates pre- and post-immunodepletion demonstrated that, whereas immunoprecipitation of dyskerin or over-expressed Flag-TERT depleted TRAP activity from the lysate, immunoprecipitation of pontin and reptin did not reduce the overall level of TRAP activity in the extract (). Accordingly, TRAP assays performed on the anti-Flag immunoprecipitates showed that dyskerin and Flag-TERT brought down robust TRAP activity, whereas pontin and reptin were associated with a small, but reproducible amount of activity (). Remarkably, however, analysis of the immunoprecipitates by western blot for endogenous TERT showed that pontin and reptin were associated with at least as much TERT protein as that bound by dyskerin (). Therefore, our data indicate that pontin and reptin associate with a significant population of TERT molecules that do not yield high level TRAP activity. This marked discordance between TERT protein and catalytic activity in vitro suggests specific models for understanding how telomerase is assembled in human cancer cells (see Discussion). Together, these data establish pontin and reptin as both TERT-interacting proteins and dyskerin-interacting proteins and show that pontin and reptin are required for assembly of a core telomerase complex, including TERT, TERC and dyskerin.
TERT exists in multiple telomerase complexes