High-Throughput RNAi Screening Reveals Novel Regulators of Telomerase

Telomerase is a ribo-nucleoprotein complex essential for the maintenance of chromosome structure and function [reviewed in (1)]. Telomerase adds de novo telomeric repeats to chromosome termini, thereby ensuring the “capping” and protection of chromosome ends and limiting termini erosion due the end replication problem (1). In normal human cells, telomerase activity (TA) is tightly regulated and is restricted primarily to germline cells. As a consequence, telomeres shorten at each round of cell division until they reach a critical length, which triggers replicative senescence (1). It has been proposed that replicative senescence acts as a barrier to unrestricted cell proliferation and tumourigenesis (1). In keeping with this model, most human immortal cell lines and tumor cells bypass replicative senescence by activating mechanisms to maintain telomeres (1). This occurs via activation of telomerase, or alternatively via a second mechanism based on DNA recombination, ALT [alternative lengthening of telomeres, (1)].

The minimal telomerase enzyme encompasses a catalytic subunit (hTERT), an RNA-dependent DNA polymerase with reverse transcriptase activity, and an RNA subunit (hTR), which functions as template for the addition of telomeric repeats to chromosome termini (1). TA is tightly regulated in human cells and this regulation involves a number of protein kinases (2). For example, hTERT is phosphorylated in vitro by the kinase AKT (3, 4). Conversely, c-ABL kinase has been identified as a negative regulator of TA (5). Members of the protein kinase C (PKC) family also phosphorylate hTERT in vitro, induce TA, and stimulate hTERT promoter activity (6, 7). The kinase SRC acts as a negative regulator of TA by phosphorylating hTERT in response to oxidative stress and inducing its export from the nucleus (8). Several other kinases, such as ERK1/2 and JNK, have also been linked to the regulation of TA but the mechanisms underlying these effects are still not clear (9, 10).

The almost universal presence of telomerase in human cancers and its absence in most normal tissues make telomerase an attractive therapeutic target. Indeed several compounds that inhibit TA have been developed; however, the clinical results with these drugs to date have been inconclusive (11). As an alternative to direct telomerase inhibition, inhibition of proteins that modulate TA could be exploited in cancer therapy. This could be achieved by targeting transcription factors that mediate the expression of either telomerase subunit, but this is pharmacologically challenging. Alternatively, targeting protein kinases that modulate TA could also be used. Protein kinases are important mediators of signal transduction and are involved in a wide variety of biological processes. Moreover, the relative ease by which catalytic inhibitors of kinases can be developed makes kinases attractive drug targets. To comprehensively profile the contribution of kinases to telomerase regulation, we conducted a high-throughput RNA interference screen (HTS). We used a library of siRNAs targeting kinases and kinase-associated genes and measured the effects of kinase silencing on TA. This approach identified a number of new modulators of TA, whose inhibition could be exploited as a means to target telomere maintenance.

HeLa, CAL51, MCF7, MCF10A, and MCF12A cells were obtained from American Type Culture Collection. Human foreskin BJ fibroblasts were kindly provided by Dr. Silvia Bacchetti (IRE, Rome, Italy). The protein kinase siRNA library (siARRAY) was obtained in 10 96-well plates from Dharmacon.

pCMV5-HA-ERK8 and pCMV5-HA-ERK8-D154A (HA-ERK8-KD) containing, respectively, HA-tagged wild-type and kinase-dead mutant ERK8 were kindly provided by Prof. Cohen (University of Dundee, Scotland, UK; ref 12). pcDNA3-Flag-hTERT was kindly provided by Prof. Chantal Autexier (McGill University, Montreal, Canada). The siERK8-04–resistant cDNA constructs were generated by site-directed mutagenesis using the Quickchange Kit (Stratagene) according to the manufacturer's instructions. The sequence in ERK8 targeted by siERK8-oligo 4 (siERK8-04) is the following: CCUAUGGCAUUGUGUGGAAUU. Cell transfection, drug treatment, lentiviral infection, and gene silencing validation methods are described in the Supplementary Methods section.

BX795 was obtained from Axon Medchem and dissolved in dimethyl sulfoxide (DMSO). Ro318220 was obtained from Calbiochem. XAV939 was obtained from Maybridge and dissolved in DMSO.

HeLa cells were plated in duplicate 96-well plates and transfected with the siRNA library (final concentration 100 nmol/L). After 5 days, cell viability was assessed using CellTiterGlo Luminescent Cell Viability Assay (CTG; Promega) in plate 1 and whole cell extracts were prepared from plate 2. For the analysis of cell viability, the effect of each SMARTpool was calculated by dividing the mean luminescence of each siRNA-transfected well by the mean luminescence of the siCtrl-transfected wells in each plate.

To determine the overall assay quality and reproducibility, Z scores were calculated using the surviving fractions from all 10 siRNA plates centered on the median and the standard deviation estimated from all the plates. Z′-factors between negative control (siCtrl) and positive control [sihTERT (siRNA targeting hTERT)] were calculated as reported (13), using Z′ ≥ 0.5 as a threshold of acceptable dynamic range.

We optimized a 96-well–based method of whole cell extracts suitable for the TRAP assay [Telomere Repeat Amplification Protocol; (14)]. Briefly, cells in 96-well plates were scraped in a NP-40–based lysis buffer (15), incubated in ice for 15 minutes, transferred to a Multiwell Filter Plate containing a 0.2 μm Biolnert Filter (AcroPrep, VWR), centrifugated for 10 minutes at 4°C, and snap frozen in liquid nitrogen. After quantifying the total protein concentration, 0.1 μg of whole cell extracts were used for SYBR Green-based qTRAP according to reference (16). Standard curves were calculated for all reactions with serial dilutions of HeLa extracts. As a negative control, we used HeLa extracts heat-inactivated for 10 minutes at 85°C and a lysis buffer only control. Relative telomerase activity (RTA) was calculated using the average C_t of the 3 replicas for each sample with the following formula RTA = 10 (C_t sample-Y_int)/slope, where Y_int, the intercept on the y-axis and the slope were derived from the standard curve (16). The RTA for each siRNA was normalized to RTA of the siCtrl cells.

For immunofluorescence, cells were treated with hypotonic solution for 5 minutes (0.25% Triton, 20 mmol/L Hepes, 50 mmol/L NaCl, 3 mmol/L MgCl₂ in H₂0), fixed in 4% paraformaldehyde, permeabilized with 0.25% Triton X-100 for 15 minutes, treated with ice-cold methanol for 20 minutes, and blocked in PBS with 5% FBS.

Telomere dysfunction–induced DNA damage foci (TIF) were detected by the colocalization of 53BP1 (BP13, 1/200, Upstate Cell Signaling) and telomeric-repeat binding factor (TRF1; 1/300, Abcam). Nuclei were counterstained with 4′,6-diamidino 2 phenylindole (DAPI) and cells were analyzed using a confocal fluorescence microscope. At least 50 cells for each experiment in at least 3 independent experiments were scored. Statistical differences were analyzed with the unpaired t test, using the online GraphPad QuickCalcs software.

Metaphase spreads were prepared as previously reported (17) and hybridized with a telomeric FITC-labeled probe (DAKO), counterstained with DAPI and processed according to the manufacturer's instructions. Fluorescent signals were visualized with a confocal fluorescent microscope.

Genomic DNA was prepared using the QIAamp DNA Blood Minikit (Qiagen). Telomere length was measured using a real-time PCR–based method, qTRF (18). For each population the TRF length was normalized to the length of sh-NS cells and differences between controls and ERK8-silenced populations were calculated using the unpaired t test. For differences in TRF over time, the TRF length of each population was normalized to that of the early population doubling and statistical differences between early and late time-points were analyzed with the paired t test.

To identify novel determinants of TA in human cells, we conducted a HTS in a telomerase-positive human tumor cell line, HeLa (Fig. 1A). We used a human SMARTpool siRNA kinase library. In this library, each well contained 4 distinct siRNA species (SMARTpool) targeting a single gene. HeLa cells were plated in duplicate 96-well plates and transfected with the siRNA library. After 5 days, the effect of each SMARTpool on cell viability was assessed in 1 set of the replica plates (plate 1), using the CTG assay. The second set of plates was used to measure TA (plate 2). To estimate the effect of each SMARTpool on TA, we used the telomerase-specific TRAP assay (14), a PCR-based method that measures the ability of telomerase to add telomeric repeats to a substrate. To use this method in a high-throughput format, we used a real-time PCR–based version of the TRAP assay, qTRAP (16), which allows to estimate TA in real-time via fluorescence measurements, rather than other similar, but less precise methods that rely on the quantification of electrophoresed PCR products. Whole cell extracts from siRNA-transfected cells were used in the qTRAP assay and the TA for each SMARTpool was normalized to that in cells transfected with a nontargeting control siRNA (siCtrl). To generate whole cell lysates, we used a 96-well–based filter method that was compatible with the qTRAP assay (see Material and Methods section). Cell number, siRNA concentration, and duration of the screen were optimized to obtain less than 10% to 15% reduction in viability in cells transfected with siCtrl compared with mock-transfected (transfection reagent present, no siRNA) and untransfected cells. The screening conditions were also titrated to ensure high-efficiency transfection, as measured by the ability of siRNA targeting the essential gene, PLK1 (polo-like kinase 1), to cause more than 80% reduction in cell viability compared with siCtrl-transfected cells. As a positive control for the effect on TA, we used a siRNA targeting the catalytic subunit of telomerase, hTERT, whose silencing reduced TA (see Supplementary Fig. S1B and Fig. 2A).

Following optimization, the screen was conducted twice. Z scores for viability between the replica screens were highly reproducible (r² = 0.7783; Fig. 1B). Moreover, the dynamic range of the screen was estimated using the performance of the negative (siCtrl) and positive (sihTERT) controls used in the TRAP assay and was acceptable in both replicas with Z′ = 0.5 in each replica (19).

To account for the differing effects of each SMARTpool on cell viability and to minimize nonspecific effects on TA due to significant cell death, only SMARTpools that caused less than 50% reduction in viability compared with control-transfected cells were analyzed by qTRAP. To further increase the specificity of the screen, we set stringent threshold values for TA in candidate selection; those kinases whose inhibition reduced TA below 35% of TA in the siCtrl cells were considered potential positive regulators, whereas those that increased TA above 150% of the controls were considered potential negative regulators. Of all the siRNAs analyzed by qTRAP (480 siRNAs), about 1/5 were hits in both replica screens using these criteria (109 kinases out of 480 analyzed; see Supplementary Table S1).

Internal validation of the screen was provided by the observation that siRNAs targeting SRC (v-SRC sarcoma viral oncogene homolog), a negative regulator of telomerase (8), significantly increased TA (see Supplementary Fig. S1A). In a similar fashion, siRNAs targeting PRKCQ (protein kinase C theta), a positive regulator of telomerase (6), significantly decreased TA (see Supplementary Fig. S1A).

Annotation of the hits from the screen identified 2 main groups modulating TA: (i) mitotic kinases involved in the spindle formation and the mitotic spindle checkpoint, such as aurora kinase B (AURKB), NEK [NIMA (never in mitosis)-related kinase] family members, CCRK (cell-cycle–related kinase), and (ii) MAP kinase super-family members, including members of the ERK1/ERK2 pathway as well as members of the JUNK/p38 MAPK pathway, such as ERK1/2, MAP3K4, MAP2K7, MAPKAPK5, and ERK8 (MAPK15).

Of the novel modulators of TA identified, we selected a subset of 26 (see Supplementary Tables S1, genes in bold; and S2) for further analysis. Since the main aim of this work was to identify novel modulators of telomerase and ultimately novel targets for cancer therapy, we selected a number of kinases not previously implicated in telomerase regulation. Candidates were also selected where several members of the same sub-family were represented among the hits, perhaps supporting their involvement in regulating telomerase. Among the mitotic kinases selected were AURKB, NEK2, and NEK7. AURKB is an important regulator of mitosis involved in chromosome alignment and spindle assembly checkpoint, is frequently overexpressed in human cancers and has been pursued as a target for anticancer therapies (20–22). AURKB has also been shown to modulate telomerase via Survivin (23). NEK2 and NEK7 belong to the NEK family of kinases, of which 5 members were present among the hits. There is now substantial evidence that NEK2, NEK6, NEK7, and NEK9 regulate mitotic progression and mitotic spindle organization (24). Furthermore, high expression levels of NEK kinases have also been found in human cancers, such as breast and bladder cancers (25, 26) and NEK kinases are now being pursued as potential therapeutic targets (27, 28). Members of the MAPK superfamily were also selected given their diverse functions in a wide variety of biological processes (9, 29). Other kinases previously unconnected to telomere maintenance that were further analyzed include ERK8, a new member of the MAPK family whose expression is upregulated in colorectal and other cancers (30, 31; see Supplementary Table S2). As controls we analyzed the kinases PRKCQ and SRC (2).

Using this selected gene subset, we first repeated the original screen assay to confirm the initial results and filter out aberrant observations common in HTS (see Supplementary Fig. S1B). Of the 26 original hits, 13 showed a similar effect to the original HTS and were selected for further analysis. To validate the specificity of the observed phenotypes and minimize “off-target” effects, we repeated the HTS assay using each individual siRNA present in the SMARTpools targeting the selected hits (32). For most of the kinases analyzed at least 2 of 4 siRNAs within the SMARTPools modulated TA, replicating the effects observed in the HTS (Fig. 1C and D). To confirm that the reduction in TA was associated with silencing of the target kinases, we carried out qPCR to quantify mRNA levels or western blotting to estimate protein expression of the target genes (see Supplementary Fig. S2).

To eliminate the possibility of identifying effects particular to HeLa cells, we assessed the ability of siRNAs to modulate TA in additional cell lines. We selected 2 telomerase-positive cell lines, CAL51 and MCF7, which are amenable to high-efficiency siRNA transfection. The majority of siRNAs had comparable effects on TA in both MCF7 and CAL51 cells as those observed in HeLa cells (Fig. 1E and F).

TA is tightly regulated in human cells and several proteins have been implicated in its regulation (2). We assessed whether the kinases identified in the screen modified the transcription of hTERT or hTR (33). We found that silencing of MAPKAPK5 or ERK8 significantly reduced hTERT mRNA levels compared with siCtrl-transfected cells (Fig. 2A), effects that correlated with a marked decrease in TA in the qTRAP (see Supplementary Fig. S1). Conversely, knockdown of SRC resulted in a significant induction of hTERT mRNA, consistent with the upregulation of TA in the qTRAP assay and also a proposed model whereby SRC acts as a negative regulator of telomerase. Interestingly, siRNAs targeting PRKCQ, which has been proposed to activate the promoter activity of hTERT in T lymphocytes (6), did not significantly affect hTERT mRNA levels. However, the use of different cell line models could explain this difference. For the other kinases analyzed we found that silencing did not significantly affect hTERT mRNA levels (Fig. 2A), suggesting that these genes most likely regulated TA via some other mechanism. We also measured the levels of hTR upon silencing of the selected kinases by qPCR and found that none of the siRNAs analyzed significantly affected hTR levels (see Supplementary Fig. S3).

One of the cellular consequences of telomerase inhibition is telomere dysfunction due to uncapped telomeres, which result in the formation of DNA damage–induced foci localized at the telomeres [TIFs (34, 35)]. TIFs can be identified by the colocalization of a telomeric-specific protein, such as TRF1, and a DNA damage-associated protein, such as 53BP1.

To investigate whether modulation of the selected kinases caused telomere disturbances, we measured the presence of TIFs by monitoring the colocalization of TRF1 and 53BP1 by immunofluorescence. We found that silencing of kinases that elicited a reduction in TA also caused an increase in the percentage of cells containing TIFs, whereas silencing of kinases that had no effect on TA did not (Fig. 2B and D), supporting the hypothesis that perturbing TA via kinase modulation could result in telomere dysfunction. Specifically, targeting of ERK8 or PRKCQ resulted in a high percentage of cells containing multiple TIFs (2–5 TIFs/cell; 23% and 18% of cells with TIFs in siERK8- and siPRKCQ-transfected cells vs. 7% in siCtrl-cells; P < 0.001). siRNAs targeting of AURKB, PNKP, NEK2, MAPKAPK5, and MAP2K7 induced more moderate yet significant increases in cells containing multiple TIFs (each effect P < 0.05 compared with siCtrl-cells), whereas MAPK1, NEK7, and MAP3K4 silencing increased the percentage of cells containing 1 or 2 large TIFs (11.2, 15, and 14.4% vs. 7%; P < 0.05). Of the kinases analyzed, only MAPK3 knockdown did not significantly increase TIFs. A similar increase in the number of cells containing TIFs was also observed using multiple individual siRNAs for ERK8 and MAPKAPK5 (see Supplementary Fig. S4A).

ERK8 is a rather poorly characterized MAPK, but does appear to be regulated by DNA damage and some activated oncogenes (31, 36, 37). Recently, it has been shown that ERK8 promotes the neoplastic transformation and progression of colorectal cancer (30). In addition, ERK8 expression appears to be elevated in various types of human cancers, including lung, pancreatic, and ovarian tumors (31, 38). ERK8 knockdown had the strongest effect among the kinases analyzed on the formation of TIFs and a dramatic effect on TA, which suggested that ERK8 might be a specific and potent regulator of telomerase. Initially we noted that inhibition of ERK8 in 2 telomerase-negative cell lines did not induce TIFs, in contrast to the observations made in telomerase-positive cells (see Supplementary Fig. S4B), providing some support to the hypothesis that the effects of ERK8 on telomere function were mediated via an effect on telomerase. Subsequently, we investigated whether the catalytic activity of ERK8 was required for modulation of TA. To achieve this, we used a combination of siRNA-mediated targeting of the endogenous ERK8 gene together with recapitulation of ERK8 expression. We introduced siRNA-insensitive HA-ERK8 or HA-ERK8-KD cDNA constructs into cells in which ERK8 levels were reduced by siRNA (see Supplementary Fig. S5) and measured viability and TA. Reintroduction of wild-type ERK8 reconstituted TA to levels almost comparable to siCtrl-transfected cells, whereas reconstitution of the kinase-dead mutant ERK8 did not (Fig. 3A), suggesting that ERK8 kinase activity is essential for TA regulation. Moreover, introduction of the kinase-dead ERK8 in siCtrl-transfected cells reduced TA, reminiscent of a dominant-negative effect and supporting the hypothesis that ERK8 activity is essential for TA regulation. Furthermore, introduction of hTERT cDNA construct restored TA in ERK8-silenced cells, suggesting that the effect of ERK8 on TA acts via hTERT (Fig. 3A).

The most well-established function of telomerase is telomere elongation. To further assess the phenotypic outcome of ERK8 modulation of telomerase, we investigated whether prolonged ERK8 silencing could affect telomere length over time. To examine this, we infected MCF7 cells with 4 lentiviruses expressing sh-RNAs targeting ERK8 or a control, nontargeting sh-RNA (sh-NS). After puromycin selection, we assessed ERK8 silencing and TA. Of the 4 sh-RNA used, E8-72 did not affect ERK8 levels or any of the other parameters analyzed, whereas 2 sh-RNAs, E8-1294 and E8-1846, elicited a significant reduction in ERK8 expression, correlative reduction in TA and a mild decrease in hTERT mRNA levels (Fig. 4A and B, and data not shown). Moreover E8-1294 and E8-1846 also induced TIFs (19% and 22% of cells with TIFs in sh-E8-1294 and sh-E8-1846-infected cells vs. 8% in sh-NS cells, P < 0.01; Fig. 4C and D). The increase in TIF frequency was restored to basal levels by the introduction of a wild-type but not a kinase-dead ERK8 cDNA, confirming that these effects were specific for ERK8 kinase activity, similar to the effect on TA in HeLa cells (Fig. 3B). A final sh-RNA construct, E8-486, induced a marked albeit not significant reduction of ERK8 levels, but this did not result in any effect on TA; we assume that the silencing of ERK8 with this sh-RNA did not reach a critical threshold required to affect telomerase function.

To determine the effect of ERK8 modulation on telomere length, we used the qTRF assay (18). Compared with sh-NS infected cells, populations with significant ERK8 silencing (sh-E8-1294 and sh-E8-1846) had shorter telomeres (Fig. 4E). More importantly, whereas sh-NS infected populations and populations in which ERK8 silencing was not profound maintained stable telomere lengths over time, sh-E8-1294 and sh-E8-1846-infected cells showed significant telomere shortening (Fig. 4F), indicating that inhibiting telomerase via ERK8 silencing affected telomere maintenance.

Finally, we examined the presence of telomere fusions in cells with ERK8 silencing, but were unable to detect major differences between the control and siERK8-transfected cells (see Supplementary Fig. S6). Nevertheless, the ability of ERK8 silencing to elicit a reduction of TA, an increase in TIF frequency and a reduction in telomere length together confirmed the status of ERK8 as a telomerase modulating enzyme.

We also investigated the potential for recapitulating the phenotypes observed using silencing of ERK8 with small molecule inhibitors. While clinical inhibitors of ERK8 do not yet exist, 2 experimental inhibitors have been described that inhibit the kinase activity of ERK8 with some level of specificity: BX795 and Ro318220 (12, 39). BX795 was originally described as a PDK1 inhibitor, but also inhibits the catalytic activity of ERK8 with similar potency to PDK1 inhibition (39). Ro318220 was developed as a PKC inhibitor, but also inhibits auto-phosphorylation of ERK8 at similar concentrations used to inhibit PKC (12).

We treated HeLa cells with BX795 or Ro318220 for 24 hours before testing the effects on viability, TA, and TIFs formation. At the concentrations used, neither compound affected cell viability but both compounds caused a significant reduction of TA compared with vehicle-treated cells (Fig. 5A). Moreover, both compounds elicited an increase in the frequency of cells containing TIFs compared with vehicle alone, suggesting that chemical inhibition of the catalytic activity of ERK8 mirrored the phenotypes observed with siRNA (Fig. 5B and C).

Given the therapeutic potential of modifying ERK8 activity in tumor cells, we also analyzed the effects of ERK8 inhibition in nontumor cells. For this purpose, we treated 2 telomerase-positive noncancerous cell lines, MCF10A and MCF12, as well as a primary cell line, BJ, with BX795 and Ro318220 (Fig. 5D and E). We found that treatment with either inhibitor did not impair cell viability in normal cells, giving further support for the clinical potential of ERK8 inhibition.

The main limitation of targeting telomerase as an anticancer approach is the lag time between initial telomerase inhibition and crisis, which could limit efficacy and foster the development of drug resistance (40–42). As such combinatorial therapeutic approaches that involve telomere dysfunction might be more successful. We investigated whether inhibition of TA-modulating kinases could synergize with agents that targeted telomere maintenance via a separate mechanism. We selected a tankyrase inhibitor, XAV939, which has been shown to inhibit tankyrase 1 and 2 (43). Tankyrase is a shelterin component (44), whose inhibition enhances the effects of telomerase inhibition (45, 46). As expected XAV939 did not significantly affect TA, but caused an increase in cells containing TIFs (see Supplementary Fig. S7), in keeping with the role of tankyrase in controlling telomere maintenance (46).

We transfected HeLa cells with siRNAs targeting ERK8, treated them with XAV939 and measured clonogenic survival. Silencing of ERK8 dramatically increased sensitivity to XAV939 (Fig. 6A; Table 1), suggesting that modulation of telomerase function via inhibition of ERK8 could synergize with anticancer therapeutics that impact on other aspects of telomere maintenance. To give further support to these results, we combined XAV939 with small molecule ERK8 inhibitors. We found that the combination of XAV939 with either BX795 or Ro318220 was significantly synergistic with combination indexes (CI) below 0.7 (47; Fig. 6B and C; Table 2). These results confirmed our observations with the siRNAs and suggested that combining inhibition of targets that regulate telomere maintenance via different mechanisms (i.e., ERK8 that modulates TA and tankyrase which affects, at least in part, telomerase access to telomeres) could have a synergistic effect on cell viability.

To understand whether silencing of the other kinases identified in the screen would also synergize with XAV939 treatment, we transfected HeLa cells with siRNA targeting these kinases and then assessed XAV939 sensitivity (Fig. 7). Of the genes tested, silencing of AURKB and NEK7 strongly increased sensitivity to XAV939 in a similar fashion to ERK8, whereas MAPK1 knockdown had a significant, although less marked, effect (Fig. 7A; Table 3). These results highlighted the potential of our screen to identify novel telomerase modulators for targeting telomere maintenance and may support a new rationale for the development of novel telomere-based anticancer approaches.

Telomere maintenance is an essential pre-requisite for unlimited cell proliferation and tumourigenesis. Indeed, more than 85% of human cancers activate telomerase to maintain telomeres (48, 49). Inhibiting telomerase results in telomere shortening and eventually crisis and cell death due to dysfunctional telomeres (40–42). Telomerase is, therefore, an attractive target for the development of cancer therapies. Several compounds that inhibit telomerase have been developed as potential cancer therapeutics, but to date these have had only limited clinical success (11). As an alternative, the inhibition of telomerase regulators could also be used to target telomere maintenance in tumor cells. Here we describe a high-throughput siRNA screen to identify novel modulators of telomerase that could be used as potential therapeutic targets. Given the pharmacological tractability and involvement in a plethora of signaling processes, we selected kinases and functionally profiled their ability to modulate TA. In doing so we identified a number of novel effectors, involving kinases with no previous association with telomere biology. Several kinases that had been previously shown to regulate telomerase were also among the hits, confirming the specificity of the functional read-out of the HTS (6–8). Moreover several of the identified kinases have been found overexpressed in different human cancers and are now being exploited as potential targets for cancer therapeutics (20–22, 25–29). Subsequent validation experiments showed that modulation of some of these proteins resulted in perturbation of telomerase function and elicited phenotypic changes characteristic of telomere dysfunctions, suggesting that the proteins identified in our screen are likely true modifiers of telomerase. Although the molecular mechanisms underlying many of these observations remain to be elucidated, we have showed that silencing of ERK8 or MAPKAPK5 can impair the expression of hTERT mRNA, perhaps explaining at least in part their effect on TA. It is possible that these kinases directly regulate hTERT expression though there is no previous evidence in the literature for their direct involvement in transcriptional regulation; alternatively they might act indirectly by modulating the activity of other factors that in turn regulate hTERT expression (2).

We examined ERK8 in more detail because it had the strongest effect on telomerase function. We showed that inhibition of this rather poorly characterized MAPK kinase, not only affected telomerase function at the telomeres, but also synergized with compounds such as XAV939, which inhibits the telomere-associated binding protein poly (ADP-ribose) polymerase tankyrase. More importantly from a therapeutic point of view and contrary to what we observed in cancer cells, inhibition of ERK8 did not noticeably affect viability or induce any telomere-related phenotype in noncancerous cells or in normal cells. Taken together, these data showed that ERK8 represents a promising target for the development of anticancer treatments based on targeting telomerase.

Importantly, the effects of the combination of kinase inhibition (in this case of ERK8) with other compounds that also target the telomeres (such as XAV939) suggest that this approach may hasten crisis and reduce the potential problems associated with the lag time between telomerase inhibition and eventual cell death. We also noted that XAV939 treatment synergized with silencing of other TA-modifying kinases including AURKB, NEK7, and MAPK1; each of these kinases is pharmacologically tractable and as such expands the range of approaches that could eventually be used in future therapeutic strategies.

We do note that other studies have used RNAi screening with the same siRNA platform to identify telomerase regulators, most notably by looking at modulators of telomerase expression rather than activity (50). One such study identified GSK3B as a regulator of telomerase expression and showed that targeting this protein strongly affected cell proliferation as well as telomere length, providing a link between the Wnt pathway and telomere maintenance (50). Our results complement and extend this previous study by analyzing the effects of modulating kinase activity on TA and ultimately telomerase function at the telomeres.

In summary, our data identify several novel regulators of TA that modulate telomere maintenance. More importantly, modulation of telomerase functions by these regulators can be used to sensitize tumor cells to other anticancer therapeutics that impact at least in part on telomere maintenance. Our data have important clinical implications and suggest a new rationale for the development of novel mechanism-based anticancer approaches.

No potential conflicts of interest were disclosed

The authors thank members of the Gene Function Laboratory for discussion and Prof. Philip Cohen, Dr. Silvia Bacchetti, and Prof. Chantal Autexier for the provision of reagents and helpful discussion.

This work was funded by Breakthrough Breast Cancer and CRUK.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.