In this report we have performed a large-scale shRNA screen to identify a regulatory pathway involving ETV1, ATR and TERT that is preferentially required for proliferation of diverse p53− cancer cells. We found that in p53− cells, TERT
transcription is highly dependent upon ETV1, which functions as a direct transcriptional activator by binding to the TERT
promoter downstream of the transcription start-site. In p53+ cells, ETV1, although present at comparable levels, is not required for TERT
transcription and surprisingly is not bound to the same region of the TERT
promoter. Notably, ectopic TERT
expression restored normal proliferation in p53− cells depleted of ETV1 or ATR ( and Figure S7A
), indicating that the promotion of TERT
expression is an important, but not necessarily the only, mechanism by which ETV1 and ATR maintain proliferation of p53− cells. Consistent with our results, a previous study reporting a requirement for ETV1 in TERT
was primarily based upon experiments performed in 293T cells, which lack p53 activity due to expression of SV40 large T antigen.
The results described above suggest that p53+ cells express a transcription factor that functionally substitutes for ETV1, and that one or more proteins associated with the TERT
promoter in p53+ cells prevent binding of ETV1. Several transcription factors, including SP1 (NP_612482.2), E2F1 (NP_005216.1) and MYC (NP_002458.2), have been previously shown to be associated with the human TERT
promoter (reviewed in 
). To ask whether these factors, or p53 itself, might contribute to the differential regulation of TERT we performed ChIP experiments in p53+ and p53− HCT116 cells. Consistent with previous studies, we found that E2F1 and MYC were associated with the TERT
promoter; binding of E2F1 was modestly increased in p53− HCT116 cells (Figure S15A
), whereas for MYC there was no difference in p53+ and p53− HCT116 cells (Figure S15B
). In p53+ HCT116 cells there was increased binding of SP1 (Figure S15C
) and, most notably, there was substantial binding of p53 to the TERT
promoter (Figure S15D
). Interestingly, a number of previous studies have reported physical and functional interactions between SP1 and p53 (see, for example, 
). Our ChIP results reveal substantial differences between the composition of proteins associated with the TERT
promoter in p53+ and p53− HCT116 cells, which may be related to the differential requirement for ETV1.
Interestingly, in contrast to human cancer cell lines, we found that ATR was not required for TERT
expression in experimentally derived p53− MCF10A cells, an immortalized but non-transformed human cell line (Figure S16A
). In addition, ATR was not required for TERT
expression in p53− mouse embryo fibroblasts (Figure S16B
), consistent with the lack of conservation between the mouse and human TERT
promoter (data not shown). Thus, the requirement of ATR and ETV1 for TERT
expression may be specific to human p53− cancer cell lines.
Several previous studies have reported results that are consistent with the synthetic interaction between p53 and ATR we have described here. For example, p53− cells have been found to be particularly sensitive to pharmacological inhibition of ATR (see, for example, 
). In addition, mice expressing a hypomorphic allele of ATR
have an aging phenotype that is exacerbated in the absence of p53 
. Significantly, mouse embryo fibroblasts containing this hypomorphic ATR
allele have an elongated G2 phase following loss of p53, consistent with our cell cycle results ( and Figure S6B, S6C
). However, a preferential role for ETV1 in p53− cells and its cooperative function with ATR has not been previously described and underscores the power of unbiased, large-scale RNAi-based screens.
Our screening strategy did not emphasize reaching saturation but rather sought to follow-up, by directed experiments, a limited number of candidates isolated in the primary screen. For several reasons, we believe that our screen, like other large-scale shRNA screens (see, for example, 
), did not achieve saturation. For example, a previous siRNA screen identified several factors, in particular the serine/threonine kinase receptor-associated protein UNRIP (also called STRAP; NP_009109.3), whose loss affected proliferation of p53− HCT116 cells more severely than p53+ HCT116 cells 
. However, we did not isolate UNRIP in our primary screen and, conversely, ATR and ETV1 were not isolated in the previous siRNA screen, suggesting that neither screen was truly saturating. Reasons for a failure to reach saturation in this and other large-scale shRNA screens include suboptimal efficacy of some shRNAs 
, unequal representation of shRNAs in the primary screen, and an insufficient depth of deep sequencing. Thus, it is possible that additional factors that act in the ATR-ETV1-TERT pathway, or unrelated pathways preferentially required for proliferation of p53− cells, remain to be identified.
The decreased proliferation of p53− cell lines was first evident within a few days following knockdown of ETV1, ATR or TERT. It therefore seems likely that this reduced proliferation is not a result of replicative senescence due to telomere attrition, which would require many cell divisions. Senescence occurred at much later times (10–14 days) and may be a secondary effect of the proliferation block. We observed that knockdown of ETV1, ATR or TERT resulted in an increased percentage of cells in G2/M ( and Figure S6
). Although senescent cells are generally believed to arrest in G1, it has been found that senescent cells can also arrest in G2/M (see, for example, 
A variety of previous studies have shown that TERT can promote proliferation by multiple mechanisms, several of which are unrelated to telomere length including inhibiting apoptosis 
, regulating cell signaling pathways and/or stimulating expression of diverse growth-promoting genes (see, for example, 
). It seems likely that the decreased proliferation of p53− cells following depletion of ETV1, ATR or TERT involves one of these alternative mechanisms. We have found that p53− cells depleted of ETV1, ATR or TERT have multiple growth defects including increased levels of senescence ( and Figure S5
) and an altered cell cycle ( and Figure S6
). A further understanding of how TERT promotes proliferation of p53− cells is likely to identify new factors that are potential therapeutic targets.