Chk2 Phosphorylation of Survivin-ΔEx3 Contributes to a DNA Damage–Sensing Checkpoint in Cancer

As a unique member of the inhibitor of apoptosis gene family, survivin has attracted attention for its multiple roles in cell proliferation and cell death (1), and its abundant distribution in every human tumor, compared with normal tissues (2). Disparate oncogenic pathways control the differential expression of survivin in cancer (3), and influence its localization to multiple subcellular compartments (4, 5). Accordingly, transcription of the survivin locus is complex, giving rise to at least 5 mRNA species that encode, in addition to wild-type (WT) survivin, the variants survivin-2B, -3B, -2α, and -ΔEx3 (6, 7). Structurally, survivin-2α and -3B are generated by “read-through” into intron 2 (8), or via inclusion of an alternative exon 3B (9), whereas survivin-2B and -ΔEx3 originate from the insertion of an alternative exon 2B (10), or the skipping of exon 3 (11), respectively.

Elucidating the function(s) of the survivin-spliced variants has been challenging, given their low level of expression in most cells, and the limited availability of isoform-specific reagents. For instance, survivin-2B has been reported to promote apoptosis, in vitro (10, 12). However, low levels of survivin-2B correlate with better survival in acute myeloid leukemia (13), and its silencing in ovarian cancer has been linked to greater sensitivity to taxanes (14). A role of the survivin isoforms in mitosis has been equally controversial, as this function has been proposed in some reports (9, 10), but negated in others (15).

In this study, we took a multidisciplinary approach of genome-wide bioinformatics, analysis of the DNA damage response, and evaluation of primary patient samples to dissect a potential role of survivin-ΔEx3 in cancer (6, 7). We found that survivin-ΔEx3 is a nuclear substrate of the checkpoint kinase, Chk2 (16) in its unique –COOH terminus (6, 7), and that this pathway contributes to a DNA damage–sensing checkpoint in tumor cells (17).

Fourteen cancer-related datasets with a total of 702 samples assayed on GPL5188 (Affymetrix Human Exon 1.0 ST) arrays were examined for expression of survivin-ΔEx3 (6, 7). Of the 14 datasets, 9 compared cancer or cancer-related cells with normal controls, and 5 compared either different cancers or the same cancer at different stages (Supplementary Table S1).

The HuEx-1_0-st Affymetrix microarray platform contains 22 probesets designed to detect sequences derived from 3 isoforms of the survivin locus (Fig. 1A): NM_001168 (survivin), NM_001012270 (survivin-ΔEx3), and NM_001012271 (survivin-2B). Of the 22 probesets, 9 were retained in all 14 datasets and 13 were removed because of expression below background. Probesets 3 and 16 were also removed as their expression profiles were the same as 6 other probesets that targeted the same isoforms. Of the remaining probesets (Fig. 1A), 8 of 9 probes targeted regions that were common to all 3 survivin isoforms. Probeset 9 specifically targets exon 3, which is deleted in survivin-ΔEx3 (6, 7). Specific expression of survivin-ΔEx3 was calculated as the difference between the average expression of the 8 common survivin probesets and probeset 9.

Human lung adenocarcinoma H460, breast adenocarcinoma MDA-431 and MCF-7, glioblastoma LN229, and colorectal adenocarcinoma HCT116 and SW480 cells were obtained from the American Type Culture Collection. HCT116-DR-GFP cells were kindly provided by Dr. S. Powell (Memorial Sloan Kettering Cancer Center, New York, NY). Consistent with editorial guidelines, all cell lines were used within 6 months of receipt from the cell bank. The following antibodies to Chk2 (Santa Cruz), Thr68 phosphorylated Chk2 (Cell Signaling), survivin (Novus Biologicals), p53 (Calbiochem), Ser15-phosphorylated p53 (Cell Signaling), p21 (Calbiochem), Ser139-phosphorylated histone H2AX, that is γH2AX (Millipore), Aurora B (Bethyl Laboratories), Alexa Fluor 488 (Invitrogen), FLAG (Sigma-Aldrich), β-actin (Sigma-Aldrich), COX-IV (Cell Signaling), and RCC1 (Santa Cruz) were used.

Substitution of predicted Chk2 phosphorylation sites Thr79→Ala, Thr127→Ala, and Ser98→Ala in the unique –COOH terminus of survivin-ΔEx3 was carried out using QuikChange Site-Directed Mutagenesis Kit (Stratagene) with oligonucleotides (mutated sequences underlined): 5′-ATGCAAAGGAAACCAGCAATAAGAAGAAAGAAT-3′ (Thr79, ACA→GCA), 5′-TTATTCCCTGGTGCCGCCAGCCTTCCTGTGGGC-3′ (Thr127, ACC→GCC), and 5′-AATCCATGGCAGCCAGGCGCTCGATGGCACGGC-3′ (Ser98, AGC→GCC). Mutant constructs were confirmed by DNA sequencing.

Tumor cell types (10⁵/well) were transfected with FLAG-tagged cDNAs in the presence of lipofectamine 2000 (Invitrogen) and 250 μL Opti-MEM I (Invitrogen) per well (18). In some experiments, HCT116 transfectants were treated with or without etoposide (2.5 μmol/L), immunoprecipitated with an antibody to FLAG (2 μg) plus protein G-Sepharose, and immune complexes were analyzed with an antibody to phosphorylated Ser residues (Upstate-Millipore) by Western blotting. Gene silencing by small interfering RNA (siRNA) was carried out with nontargeting or SMART pool siRNA to Chk2 (Dharmacon). Two survivin-ΔEx3-specific siRNA sequences, 5′-GAAGAAAGAAUUUGAGGAAUU-3′ and 5′-GCAAAGGAAACCAACAAUAUU-3′, were used.

Recombinant GST fusion proteins were expressed in Escherichia coli as described (19).

Tumor cell types (6–7 × 10⁷) were fractionated in mitochondrial and cytosolic extracts as described (20). In some experiments, extracts (5 μg) from HCT116 transfectants were separated on 13% acrylamide:bis-acrylamide, 29:1 gels in the presence of 20 μmol/L Phos-tag acrylamide AAL-107M (Wako) containing 3% MeOH and 40 μmol/L MnCl₂. Pull down experiments with recombinant proteins (10 μg) were carried out as described (19). For protein stability, HCT116 transfectants were treated with 0.1 mg/mL cycloheximide, harvested after 0, 2, 4, and 6 hours, and analyzed by Western blotting.

Recombinant proteins were incubated with human active Chk2 (R&D Systems, 0.1 μg/μL) in kinase buffer (pH 7.2) containing 25 mmol/L MOPS, 12.5 mmol/L β-glycerolphosphate, 25 mmol/L MgCl₂, 5 mmol/L EGTA, 2 mmol/L EDTA, 0.25 mmol/L DTT, and 0.1 mmol/L ATP in the presence or absence of ³²P-γATP (Amersham) for 20 minutes at 30°C, and detection of phosphorylated bands by autoradiography. Recombinant Cdc25 was a control Chk2 substrate.

For analysis of DNA damage, tumor cell types were treated with etoposide (2.5 μmol/L) or camptothecin (CPT, 2.5 μmol/L) for 16 hours at 37°C, followed by 4 or 8 hours recovery. Cells (1 × 10⁶) were fixed in 1% paraformaldehyde for 15 minutes and 70% ethanol for 30 minutes, permeabilized in 0.2% Triton X-100 plus 1% BSA for 10 minutes, and incubated with an antibody to γH2AX for 90 minutes. After addition of Alexa Fluor 488, cells were stained with propidium iodide (PI), and analyzed by multiparametric flow cytometry. Alternatively, cells grown on glass coverslips were processed as above and analyzed by fluorescence microscopy for γH2AX-containing foci formation using a Nikon E600 upright microscope. Fluorescence intensity and number of foci were quantified using Image J software.

A pDR-GFP plasmid that uses 2 modified GFP genes (I-Sce1-GFP and iGFP) to create a recombination reporter was used (21). Stable HCT116-pDR-GFP clones were cotransfected with pCMV-I-SceI alone or with various cDNAs, and analyzed after 3 days by multiparametric flow cytometry (22).

HCT116 transfectants (5 × 10⁵) treated with or without etoposide (1.25 μmol/L) were analyzed for colony formation in semisolid medium, as described (20). Colony size (≤100 or 100–200 μm) was quantified using NIH Image J software.

All patient-related investigations were approved by an Institutional Review Board. Histologically confirmed tumor and matched normal tissue samples from chemo-naive patients with diagnosis of non–small cell lung cancer (n = 18), or prostate cancer (n = 2) were analyzed for differential expression of survivin-ΔEx3 mRNA by RT-PCR using primers 5′-TGGAAGGCTGGGAGCCA (forward) and 5′-AGAAAGAATTTGAGGAAACTGCGGA-3′ (reverse). Relative levels of survivin-ΔEx3 mRNA were quantified using β2-microglobulin (β-2M, Assay-on-Demand chemistry, Hs 99999907_m1) as a reference gene and the 2⁁-DDCt formula.

Anonymous primary human tissue samples representative of different stages of the colorectal adenoma-to-carcinoma transition (hyperplasia, moderate or severe dysplasia, or adenocarcinoma) were analyzed for expression of survivin, Chk2, Thr68-phosphorylated Chk2, or p53, by immunohistochemistry (20). Twenty cases per condition were scored, and a cutoff of 5% positively stained cells was used.

Data were analyzed using the unpaired t test on a GraphPad software package for Windows (Prism 4.0). Differences in survivin-ΔEx3 gene expression in datasets with paired cancer/control samples (datasets 3 and 4) were analyzed using the 2-tailed t test. Additional comparisons were done using the Spearman correlation test. The level of significance was set at P < 0.05.

We used a genome-wide bioinformatics approach to examine the expression of survivin-ΔEx3 in cancer-associated GEO datasets (Supplementary Table S1). A schematic diagram of the genomic organization of the survivin gene, and its relationship to the probesets on the arrays are shown in Fig. 1A. By principal component analysis (Supplementary Fig. S1A), the expression of probeset 9 (deleted in survivin-ΔEx3) differed from that of other 8 probesets common to other survivin isoforms, validating the ability of the microarray platform to distinguish between survivin isoforms that either contain or lack exon 3 (Supplementary Fig. S1A). A hierarchical clustering based on differential probeset expression is shown in Supplementary Fig. S1B. Given the average expression of probesets 4, 5, 13, 14, 15, and 17 (X = cumulative expression of all 3 isoforms) and probeset 9 (Y = expression of survivin and survivin-2B, but not survivin-ΔEx3), we determine that Y < X for the majority of samples (Fig. 1B). Calculated as the difference X − Y and then log₂-transformed, survivin-ΔEx3 mRNA was expressed at significantly higher levels in tumors of the ovary, stomach, brain, lung, and prostate compared with matched normal tissues (Fig. 1C). WT survivin was also differentially expressed in tumor versus normal tissues in the same datasets (Fig. 1; ref. 2).

Survivin-ΔEx3 was expressed at higher levels in grade IV glioblastoma compared with low-grade oligodendrogliomas, metastatic compared with localized prostate cancer, and stage 4 compared with stage 1 neuroblastoma (Fig. 1D; Supplementary Table S2). Survivin-ΔEx3 expression correlated with hallmarks of disease progression, including poor survival (P = 0.01, Spearman correlation), recurrence (P = 6 × 10⁻⁵, Spearman correlation), and advanced Gleason grade (P = 0.0004) in prostate cancer, N-myc amplification in neuroblastoma, Helicobacter pylori infection in gastric cancer, and chromosome 11q13 amplification in head and neck cancer (Supplementary Table S2).

To validate these bioinformatics predictions, quantitative PCR analysis of primary tumor samples and matched normal tissues was next carried out. In these experiments, survivin-ΔEx3 mRNA was consistently overexpressed in cases of non–small cell lung cancer (Fig. 1E, left) and prostate cancer (Fig. 1E, right), compared with matched normal tissues.

Consistent with previous observations (6, 7), endogenous survivin-ΔEx3 mRNA was expressed at lower abundance in MCF-7 cells, compared with WT survivin (Supplementary Fig. S2A). When transfected in HCT116, H460, or MDA-431 tumor cells, FLAG-tagged survivin-ΔEx3 was exclusively localized to nuclei, whereas WT survivin was distributed in nuclear and cytosolic fractions (Fig. 2A; ref. 23). In addition, survivin-ΔEx3 did not localize to mitochondria in transfected tumor cells, as opposed to an abundant pool of WT survivin associated with tumor mitochondria (Fig. 2B). Functionally, treatment of H460 cells with the apoptosis inducer, staurosporine, caused loss of mitochondrial transmembrane potential, in a reaction unaffected by transfection of control plasmid or survivin-ΔEx3 (Fig. 2C). Conversely, a ∼50% knockdown of endogenous survivin-ΔEx3 expression using isoform-specific siRNA (Supplementary Fig. S2A) produced nuclear events, with accumulation of tumor cells with G2/M DNA content, and appearance of a cell population with >4N DNA content (Supplementary Fig. S2B). In contrast, survivin-ΔEx3–specific siRNA did not reduce WT survivin mRNA levels, and a nontargeting siRNA had no effect (Supplementary Fig. S2A and S2B).

By sequence analysis, 3 potential phosphorylation sites matching the consensus for the DNA damage kinase, Chk2 (16) were identified at Thr79, Ser98, and Thr127 in the unique –COOH terminus of survivin-ΔEx3 (Fig. 3A; refs. 6, 7). In contrast, no Chk2 phosphorylation sites were predicted in WT survivin. Accordingly, active Chk2 readily phosphorylated recombinant survivin-ΔEx3, but not WT survivin, in a kinase assay (Fig. 3B). Ala substitution of Thr127 in survivin-ΔEx3 (ΔEx3-Mut1), or double mutation of Thr79 and Ser98 (ΔEx3-Mut2) partially reduced phosphorylation by Chk2 (Fig. 3B). In addition, Ala mutagenesis of all 3 predicted Chk2 phosphorylation sites at Thr127, Thr79, and Ser98 (ΔEx3-Mut3) abolished Chk2 phosphorylation of survivin-ΔEx3 (Fig. 3B).

We next asked whether Chk2 phosphorylated survivin-ΔEx3, in vivo, potentially as part of a DNA damage response in tumor cells (17). First, treatment of HCT116 cells with the DNA damaging agent, etoposide, did not result in electrophoretic mobility changes of transfected FLAG-WT survivin (Fig. 3C). In contrast, etoposide-induced DNA damage caused gel retardation of FLAG-survivin-ΔEx3 in HCT116 transfectants (Fig. 3C). A Chk2 phosphorylation-defective FLAG-survivin-ΔEx3-Mut3 (Fig. 3B) transfected in HCT116 cells did not exhibit gel shift retardation after DNA damage (Fig. 3C), and no changes in electrophoretic mobility were observed for survivin-ΔEx3 or survivin-ΔEx3-Mut3 in untreated cultures (Fig. 3C). Second, survivin-ΔEx3 immune complexes precipitated from etoposide-treated HCT116 cells (Fig. 3D, top) reacted with an antibody to phosphorylated Ser residues, compared with untreated cultures (Fig. 3D, bottom). In contrast, survivin-ΔEx3-Mut3 immunoprecipitates (Fig. 3D, top) did not exhibit changes in phosphorylated Ser reactivity, with or without etoposide (Fig. 3D, bottom).

A potential regulation of survivin-ΔEx3 by Chk2 phosphorylation was next investigated. WT survivin transfected in HCT116 cells remained stable over a 6-hour period (Fig. 3E). In contrast, survivin-ΔEx3 exhibited a much accelerated turnover, with an estimated half-life of ∼1 hour (Fig. 3E). Ablation of Chk2 phosphorylation in survivin-ΔEx3-Mut3 improved protein stability, prolonging its half-life to ∼2 hours (Fig. 3E). Expression of survivin-ΔEx3 or survivin-ΔEx3-Mut3 did not affect endogenous Chk2 levels in HCT116 transfectants (Fig. 3E). In addition, recombinant survivin-ΔEx3 or survivin-ΔEx3-Mut3 did not associate with endogenous Chk2 in pull down experiments (Supplementary Fig. S3A). As control, recombinant survivin, but not survivin-ΔEx3, associated with the mitochondrial chaperone, Hsp60 in tumor cells (Supplementary Fig. 3B; ref. 19).

Etoposide treatment of HCT116 cells resulted in the appearance of punctate nuclear foci containing Ser139-phosphorylated histone H2AX or γH2AX (24), a marker of unrepaired DNA damage (17), by fluorescence microscopy (Fig. 4A and B). Expression of WT survivin or survivin-ΔEx3 did not affect formation of γH2AX-containing foci in response to etoposide (Fig. 4A and B). In contrast, transfection of survivin-ΔEx3-Mut3 inhibited the appearance of γH2AX-containing nuclear foci after DNA damage (Fig. 4A and B). Similar results were obtained by quantification of γH2AX reactivity by multiparametric flow cytometry. Accordingly, HCT116 cells transfected with vector, WT survivin, or survivin-ΔEx3 exhibited increased γH2AX reactivity after DNA damage, in a response that peaked 16 hours after etoposide treatment (Supplementary Fig. S4A), and returned to baseline during a 4- to 8-hour recovery period (Fig. 4C). In contrast, transfection of survivin-ΔEx3-Mut3 suppressed the induction of γH2AX reactivity after DNA damage (Fig. 4C; Supplementary Fig. S4A).

To test the specificity of γH2AX modulation by survivin-ΔEx3, we next treated HCT116 cells with the topoisomerase I inhibitor, camptothecin (CPT), which induces single-strand DNA breaks, as opposed to the double-strand breaks (DSB) caused by etoposide. CPT induced robust and time-dependent γH2AX reactivity in vector transfectants, by multiparametric flow cytometry (Fig. 4D; Supplementary Fig. S4B). Similarly, transfection with WT survivin, survivin-ΔEx3, or survivin-ΔEx3-Mut3 did not affect the magnitude or time-course of γH2AX induction by CPT (Fig. 4D; Supplementary Fig. S4B).

We next asked whether Chk2 phosphorylation of survivin-ΔEx3 affected homologous-directed DSB repair using HCT116 cells stably transfected with a GFP-DR recombination reporter (21, 22). Cotransfection of HCT-pDR-GFP cells with pCMV-I-SceI plus vector resulted in the appearance of GFP⁺ cells, indicative of homologous DSB repair (Fig. 4E; refs. 21, 22). Addition of WT survivin, survivin-ΔEx3, or survivin-ΔEx3-Mut3 to these cells also resulted in quantitatively comparable degree of homologous DSB repair (Fig. 4E). No GFP⁺ cells were detected in the absence of pCMV-I-SceI (Fig. 4E), ruling out spontaneous intrachromosomal DNA conversion (21, 22).

We next asked whether the effect of survivin-ΔEx3 on DSB foci formation reflected changes in total H2AX expression. HCT116 cells transfected with survivin-ΔEx3 or survivin-ΔEx3-Mut3 did not exhibit significant changes in endogenous H2AX levels, in the presence or absence of DNA damage (Fig. 5A). Second, we tested whether this pathway required the DNA damage–sensing Mre11-Rad50-Nsb1 (MRN) complex, which is deficient in HCT116 cells (25). Transfection of survivin-ΔEx3 in colorectal adenocarcinoma SW480 cells, which express the MRN complex, resulted in robust γH2AX induction in response to etoposide, and this response was abolished by expression of survivin-ΔEx3-Mut3 (Fig. 5B).

Finally, we looked at the activation of downstream pathway(s) of DNA damage upon expression of the various survivin isoforms. Transfection of survivin-ΔEx3-Mut3 suppressed γH2AX induction in etoposide-treated HCT116 cells (Fig. 5C), consistent with the data above. However, stabilization of p53, phosphorylation of p53 on Ser15, or upregulation of the cyclin-dependent kinase inhibitor, p21, was indistinguishable in cells expressing survivin-ΔEx3 or survivin-ΔEx3-Mut3 (Fig. 5C). Similarly, no differences were observed in etoposide-induced Thr68 phosphorylation of Chk2 in the presence of the various survivin isoforms (Fig. 5C), and transfection of survivin-ΔEx3 or survivin-ΔEx3-Mut3 did not alter the expression of the chromosomal passenger protein, Aurora B during DNA damage, compared with WT survivin (Supplementary Fig. S5).

In the absence of DNA damage, transfection of survivin-ΔEx3 or survivin-ΔEx3-Mut3 significantly enhanced HCT116 colony formation in soft agar, compared with control cultures (Fig. 5D, left), suggesting a cytoprotective role of both molecules irrespective of DNA damage or Chk2 phosphorylation. Expression of survivin-ΔEx3 still supported HCT116 colony formation after etoposide-induced DNA damage (Fig. 5D, right). In contrast, transfection of survivin-ΔEx3-Mut3 impaired the ability of HCT116 cells to form large colonies (up to 200 μm in diameter) after etoposide treatment (Fig. 5D, right).

As far as potential implications of this pathway for human tumorigenesis, Thr68-phosphorylated, that is active Chk2 became abundantly expressed at the earliest, hyperplastic stages of the colorectal adenoma-to-carcinoma transition, and persisted in advanced disease stages (Supplementary Fig. S6A and S6B). In contrast, active Chk2 was undetectable in normal epithelium, and total Chk2 levels remained unchanged throughout disease progression (Supplementary Fig. S6A and S6B). The expression of active Chk2 correlated with aberrantly increased levels of survivin and p53, which rose in dysplastic lesions and remained elevated throughout all stages of colorectal transformation (Supplementary Fig. S6A and S6B).

In this study, we have shown that survivin-ΔEx3 (6, 7) is a nuclear protein phosphorylated by the checkpoint kinase, Chk2 (16) during DNA damage on 3 novel amino acids in its unique –COOH terminus. Ablation of Chk2 phosphorylation enhanced survivin-ΔEx3 stability in vivo, suppressed γH2AX reactivity specifically in response to DSB, and impaired long-term clonogenic tumor cell survival. When analyzed in human tumors, survivin-ΔEx3 was differentially expressed in tumors compared with normal tissues, correlating with markers of unfavorable prognosis.

Previous studies had suggested a link between survivin and the DNA damage response. This was originally associated with the ability of survivin to counter radiation-induced apoptosis (26), but recent evidence suggested a more direct participation of the survivin network in DNA damage and repair, potentially independent of cytoprotection (27). Accordingly, interference with survivin expression or function resulted in reduced DNA repair, especially in gliomas (28), and therapeutic targeting of survivin with the small molecule transcriptional suppressor, YM155 (29), delayed DSB repair (30). Despite a potential suggestion that survivin may control the expression of the DNA repair protein, Ku70 (31), the molecular basis for a potential direct participation of the survivin network in DNA damage or repair has remained largely unexplored.

The data presented here add a new layer of complexity to these observations, and propose a role of the survivin spliced variant, ΔEx3 (6, 7) in sensing unrepaired DNA damage in tumor cells. This extends the panoply of functions previously associated with survivin-ΔEx3, including nucleolar (32), or mitochondrial control of apoptosis inhibition (33, 34), or endothelial cell migration and invasion (11), implicating Chk2 phosphorylation (16) as a novel regulator of survivin-ΔEx3. As shown here, Chk2 phosphorylation of survivin-ΔEx3 targeted its unique –COOH terminus introduced by the splicing event (6, 7), occurred during DNA damage, in vivo, and contributed to accelerated degradation of survivin-ΔEx3, compared with WT survivin. A role of survivin-ΔEx3 in the DNA damage response had not been previously proposed; however, this is consistent with the exclusive nuclear localization of this survivin isoform (35), which has been postulated to involve a bipartite nuclear localization signal, also introduced in its –COOH terminus by the splicing event (36).

Against this backdrop, Chk2 has long been recognized as a pivotal regulator of genomic integrity in response to DNA damage. Activated via phosphorylation on Thr68 by the DNA damage sensor kinase, ataxia telangiectasia mutated (ATM), active Chk2 in turn phosphorylates panoply of downstream targets coordinating checkpoint activation, cell-cycle arrest, and apoptosis (16). Adding to this scenario, Chk2 phosphorylation of survivin-ΔEx3 was required to induce time-dependent induction of γH2AX, a marker of unrepaired DNA damage (24) important to recruit repair proteins to nuclear foci (37). Further studies are required to fully elucidate how a Chk2-survivin-ΔEx3 signaling axis contributes to a DNA damage response in tumor cells. However, the evidence presented here suggests that this pathway is specific for DSB, does not affect endogenous H2AX levels, is independent of the MRN complex (38), and is not required for homologous-directed DNA repair. Taken together, these findings suggest a model in which Chk2 phosphorylation of survivin-ΔEx3 provides a feedback amplification loop to enhance DSB sensing via optimal induction of γH2AX, which, in turn, may facilitate recruitment of repair proteins, including DNA-PK (39) or Ku70 (31), at sites of DNA damage (27). Given the rapid turnover of Chk2-phosphorylated survivin-ΔEx3, in vivo, this pathway may contribute to the earliest assembly of the repair machinery at DNA foci, and would be expected to confer a survival advantage for tumor cells exposed to genotoxic stimuli. Consistent with this hypothesis, survivin-ΔEx3 was preferentially observed in tumors compared with normal tissues, and correlated with markers of aggressive disease, in vivo (40, 41). Conversely, loss of Chk2 phosphorylation of survivin-ΔEx3 was detrimental for tumor growth, and impaired long-term clonogenic tumor cell survival after DSB.

Aside from its tumor suppressive properties, which are inactivated by mutagenesis in selected tumors (42), Chk2 has been intensely pursued for novel cancer therapeutics (43). Broadly defined as “checkpoint abrogation,” this strategy relies on the premise that pharmacologic inhibition of Chk2 (and Chk1) may prevent cell cycle arrest in response to DNA damage, and thus sensitize tumor cells to apoptosis (44). While validating Chk2 as an important cancer therapeutic target (45), the data presented also anticipate a broader role of this pathway in tumor progression, via exploitation of the survivin network. Mechanistically, this may involve Chk2-mediated release of cytoprotective survivin from its mitochondrial stores during DNA damage (18), coupled with efficient DSB sensing via phosphorylation of survivin-ΔEx3 (this study). More evidence is needed to further credential this model, in vivo, but it is intriguing that both active Chk2 and survivin become expressed at the earliest stages of malignant transformation in humans, for instance during the adenoma-to-carcinoma transition. The early appearance of Chk2 in comparable settings has been interpreted as a barrier to further malignant transformation (46, 47). However, it is also possible that Chk2 exploitation of the survivin network under these conditions may paradoxically contribute to accelerated tumor development, simultaneously promoting resistance to apoptosis (18), and faster control of genotoxic damage (this study). This scenario also reiterates the broad contribution of survivin family proteins to multiple aspects of tumorigenesis (1), and validate therapeutic strategies aimed at globally suppressing aberrant survivin gene transcription in tumors (30).

No potential conflicts of interest were disclosed.

Conception and design: M. Tavecchio, J.C. Ghosh, T. Dohi, D.C. Altieri

Development of methodology: S. Lisanti, J.C. Ghosh, T. Dohi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Lopergolo, M. Tavecchio, V. Vaira, S. Bosari, L.C. Showe

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Tavecchio, S. Lisanti, J.C. Ghosh, T. Dohi, V. Vaira, S. Bosari, D. Delia, A.V. Kossenkov, L.C. Showe

Writing, review, and/or revision of the manuscript: J.C. Ghosh, D. Delia, A.V. Kossenkov, L.C. Showe, D.C. Altieri

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.C. Ghosh, N. Tanigawa

Study supervision: J.C. Ghosh, D.C. Altieri

Carried out experiments: A. Lopergolo

Real-time PCR analysis: A. Faversani

The authors thank Drs. Akiko Miyamoto and Hiroshi Kawasaki for help with the procurement of human tissue samples and Simon Powell for HCT116-DR-GFP cells.

This work was supported by NIH grants CA140043, CA78810, and CA118005. Support for Core Facilities used in this study was provided by Cancer Center Support Grant (CCSG) CA010815 to The Wistar Institute.

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.