Rescue of Hippo coactivator YAP1 triggers DNA damage-induced apoptosis in hematological cancers.

Oncogene-induced DNA damage elicits genomic instability in epithelial cancer cells, but apoptosis is blocked through inactivation of the tumor suppressor p53. In hematological cancers, the relevance of ongoing DNA damage and the mechanisms by which apoptosis is suppressed are largely unknown. We found pervasive DNA damage in hematologic malignancies, including multiple myeloma, lymphoma and leukemia, which leads to activation of a p53-independent, proapoptotic network centered on nuclear relocalization of ABL1 kinase. Although nuclear ABL1 triggers cell death through its interaction with the Hippo pathway coactivator YAP1 in normal cells, we show that low YAP1 levels prevent nuclear ABL1-induced apoptosis in these hematologic malignancies. YAP1 is under the control of a serine-threonine kinase, STK4. Notably, genetic inactivation of STK4 restores YAP1 levels, triggering cell death in vitro and in vivo. Our data therefore identify a new synthetic-lethal strategy to selectively target cancer cells presenting with endogenous DNA damage and low YAP1 levels.

Introduction

In epithelial cancers, rampant DNA double strand break (DSB) formation followed by activation of DNA damage response (DDR) occurs in both premalignant and malignant conditions. However, in precancerous settings senescence and apoptotic responses, the so-called ‘tumorigenesis barrier’, prevent progression to malignancy[1] until the tumor suppressor TP53 (p53) is inactivated, thereby triggering genomic instability and enhancing tumor cell growth.

Variable levels of γ-H2A.X and pATM have been reported in AML, myelodysplastic syndromes[2] and in multiple myeloma (MM)[3]. However, in blood tumors, the presence and the functional role of ongoing DNA damage in carcinogenesis have not been explored in detail. Additionally, inactivation of p53 does not appear to represent a pivotal event in the evolution from premalignancy toward cancer in these diseases, since p53 mutations are relatively rare and appear late during the course of these tumors. For example, p53 mutations and deletions are rare in newly diagnosed MM patients (5–10%), and present in only 10–20% of patients with relapsed and refractory MM[4], while losses of ATM and ATR are even rarer[5].

An alternative pathway to p53, downstream of ATM/ATR has been described, which is activated after DSB and induces apoptosis[6-10]. It is centered on the proto-oncogene ABL1, commonly translocated in chronic myeloid leukemia. Treatment of tumor cell lines with DNA damaging agents results in ABL1 relocalization from the cytoplasm to the nucleus, where it elicits apoptosis. Importantly, the nuclear shuttling of ABL1 has been demonstrated only in vitro in tumor cell lines after drug-induced DNA damage; therefore, its functional and in vivo clinical relevance remains unknown.

Here we elucidate a novel synthetic lethal approach where genetic inhibition of serine– threonine kinase 4 (STK4) reactivates the Hippo mediator YAP1, thereby triggering apoptosis in hematologic malignancies with intrinsic DNA damage. Our data provide the rationale for the development and clinical evaluation of novel STK4 inhibitors.

Results

MM cells evade apoptosis despite pervasive DNA damage

We first explored a panel of MM cell lines and MM cells to confirm the presence of widespread DNA damage[3]. Eleven of 13 MM cell lines and cells derived from subjects with MM demonstrated increased γ-H2A.X staining ( and ) and an activated DNA damage response (DDR) ( and data not shown). This pattern was not present in normal plasma cells or in peripheral blood mononuclear cells (PBMCs) derived from healthy individuals[3] (), mirroring reports in other cellular contexts where the presence of DNA damage discriminates normal tissues from pre–neoplastic and cancerous lesions[11,12]. Notably, U266 and KMS–34 MM cell lines, which did not show γ-H2A.X foci, were also negative for all markers of DDR activation ().

Surprisingly, despite this pervasive DNA damage and DDR activation, we did not detect any significant cell death under basal conditions ( and ), implying that MM cells have mechanisms to escape the apoptotic response triggered in normal cells.

ABL1 relocalization in the nucleus of MM cells

As mentioned above, p53 genetic inactivation appears to be less relevant in hematopoietic neoplasms including MM than in epithelial cancers[4]. Indeed, modifications associated with DNA damage were present at a comparable level in both p53–wild type and – mutated MM cell lines ( and data not shown). A second pathway involved in the apoptotic response after DSBs involves nuclear re–localization of ABL1 upon DNA damage[7-10]. We therefore asked whether ABL1 is localized in the nucleus in MM. Strikingly, ABL1 demonstrated a prominent and preferential localization inside the nucleus in most MM cells, regardless of their p53 mutational status (), in contrast to HeLa cells, in which ABL1 shuttles inside the nucleus only after doxorubicin treatment (). Immunohistochemical staining also confirmed prominent nuclear ABL1 localization in cells derived from individuals with MM ( and ).

Following DNA damage, ATM is phosphorylated[13]. As a result, activated JNK phosphorylates 14–3–3 proteins leading to ABL1 release, which in turn can shuttle inside the nucleus[14]. Indeed treatment of MM cell lines with an ATM inhibitor, Ku55933[15] or JNK1 inhibitor, SP600125[16] reduced nuclear and increased cytoplasmic ABL1 ( and ) in both p53–WT and p53–mutant MM cell lines. Taken together, these results suggest that ongoing DNA damage in MM activates ATM and JNK, leading to nuclear relocalization of ABL1.

To determine whether nuclear ABL1 is able to induce apoptosis in MM cells, we used the U266 MM cell line, which does not show γ-H2A.X foci, pATM or nuclear ABL1 under basal conditions. Treatment with DNA–damaging agent doxorubicin induced multiple γ-H2A.X foci and strong ATM and JNK phosphorylation (). Importantly, ABL1 moved into the nuclear compartment (), and a marked increase in apoptotic cells was evident (). Treatment with imatinib significantly increased viability, suggesting a prominent role for nuclear ABL1 in inducing cell death in MM. An increase in γ-H2AX, pATM, and pJNK after exposure to doxorubicin was also evident in MM.1S cell line that presents endogenous DNA damage and DSBs, associated with a major increase in nuclear ABL1 and apoptosis (). However, in the presence of imatinib, cell viability significantly increased. In contrast, PBMCs lacked ongoing DNA damage () and did not demonstrate nuclear ABL1 relocalization or apoptosis after doxorubicin–induced DNA damage ().

We hence propose a model whereby MM cells live in a delicate equilibrium, withstanding high levels of persistent DNA damage that, through pATM and pJNK, leads to ABL1 nuclear relocalization, that should lead to cell death. Notwithstanding, no significant apoptosis was evident in MM cells, suggesting that additional mechanisms are engaged in these cells to prevent their own demise.

YAP1 is deleted or consistently down–regulated in MM cells

ABL1 forms a complex with the tumor suppressor p73, a p53 homologue[17], and the Hippo pathway effector YAP1 (Yes-associated protein)[18,19]. YAP1 and TAZ are the main transcriptional co-activators downstream of the Hippo pathway, controlling organ size and regulating stem and progenitor cell proliferation. In response to DNA damage, ABL1 induces apoptosis through YAP1phosphorylation, which in turn stabilizes p73 and co–activates p73 pro– apoptotic target genes[19,20]. Therefore we sought to determine whether endogenous nuclear relocalization of ABL1 in MM is unable to induce apoptosis due to disruption of the ABL1/YAP1/p73 axis.

YAP1 behaves as an oncogene in several epithelial cancers (see discussion). However, a detailed analysis of published gene expression profiling datasets revealed a noteworthy pattern: YAP1 was consistently up–regulated in tumor cell lines of epithelial origin, but profoundly downregulated in hematologic malignancies including lymphomas, leukemias, and MM (). Human YAP1 maps to chromosome 11 at the 11q22.1 locus, which is a site of focal homozygous deletions in 5 to 13% MM samples[21-24]. The genes implicated as targets of this deletion are BIRC2 and BIRC3, which control the pro-oncogenic NF-κB pathway[21,22]. Reassessing previously published data by others and us[21,23,24], we noticed that the deletion in this locus consistently involves YAP1 in addition to BIRC2 and BIRC3 in all MM cell lines and most MM samples (). At the gene expression level, probe sets reporting for YAP1 reflected low values overall, including in normal hematopoietic tissues. Importantly, however, when MM samples were subdivided in two groups based on YAP1 expression, low–expressors had a significantly shorter survival than high–expressors (). Moreover, in various datasets there was a consistent, significant reduction in YAP1 expression levels, progressing from normal plasma cells to monoclonal gammopathy of undetermined significance (MGUS) to MM ( and ). Among MM cell lines, there are subsets presenting YAP1 homozygous deletions (KMS–18, KMS–20, and KMS–28PE); others with no detectable YAP1 at the mRNA and protein level despite no genomic losses at chromosome 11; and finally, cell lines with robust expression of the gene ( and ).

YAP1 expression in MM cells induces ABL1-mediated apoptosis

We next explored the functional role of YAP1 in MM. In gain–of–function experiments, reintroduction of YAP1–EGFP[25] in deleted cell lines (KMS–18 and KMS–20) significantly reduced cell number and increased apoptosis ( and ). Conversely, downregulation of YAP1 with specific shRNAs in MM cell lines expressing YAP1 induced a significant increase in proliferation and survival proportional to the reduction in YAP1 levels, while YAP1 overexpression did not affect cell count or apoptosis ( and ).

As mentioned above, YAP1 is not expressed in a consistent number of MM cell lines, in the absence of deletions at chromosome 11. We therefore assessed whether reintroduction of YAP1 was able to affect cell proliferation and apoptosis in this specific MM subset as well. YAP1 over-expression in MM.1S cell line dramatically reduced proliferation and increased apoptosis to levels comparable to YAP1–deleted cells ( and ). These results suggest that re–expression of YAP1 might induce apoptosis and reduce proliferation not only in MM cells where YAP1 is deleted, but also in the larger population of subjects with MM where YAP1 is not expressed despite normal copy number.

YAP1–induced apoptosis was mediated by the aberrant presence of ABL1 in the nucleus, since treatment with imatinib significantly reduced the apoptotic response suggesting that YAP1 phosphorylation by ABL1 is required for the apoptotic response, as previously described[19]

( and ). Imatinib treatment also specifically reduced phospho–Y357 YAP1, a crucial step for activation of proapoptotic genes mediated by YAP1[19] (). Similar effects were obtained in the subset of MM cell lines with low YAP1 levels ( and data not shown). These results indicate that apoptosis induced by the nuclear relocalization of ABL1 in MM cells is prevented, at least in part, by low YAP1 levels.

Due to the functional interaction between YAP1 and p73[19,20,26,27], we next explored the relationship between YAP1 and p73 upon DNA damage in MM. Re–expression of YAP1 in the deleted MM cell lines remarkably increased p73 protein levels with moderate effects on p73 mRNA levels, while p53 and TP63 (p63) protein levels were not altered ( and ). Accordingly, levels of transcriptional p73 targets such as BAX, PUMA, and CDKN1A (p21), significantly increased at both the mRNA and protein levels (), whereas p53/p73 target NOXA did not vary.

ABL1–mediated phosphorylation of YAP1 at Y357 enhances its affinity toward p73 binding[28]. Indeed, imatinib treatment reduced the interaction of p73 with YAP1 (). To confirm the role of p73 in driving YAP1–mediated apoptosis, we transfected KMS– 20 with a YAP1 mutant construct that lacks the WW domain necessary to interact with p73[28]. This mutant, unlike wild type YAP1, was unable to trigger apoptosis and inhibit proliferation ( Taken together, these results suggest that apoptosis in MM induced by DNA damage and YAP1 restoration is mediated by stabilization of p73 and increased expression of its downstream pro–apoptotic targets.

Inactivation of kinase STK4 enhances YAP1 and apoptosis

A cytoplasmic serine–threonine kinase, STK4, interacts with LATS1 and significantly reduces YAP1 levels[29,30]. STK4 downregulation with specific shRNAs leads to a robust increase of YAP1 protein levels, compared to scrambled shRNA (). Notably, YAP1 appeared both in the nucleus and in cytoplasm upon STK4 downregulation (). We further explored whether STK4 downregulation impacted on YAP1 mRNA levels. A moderate increase in YAP1 mRNA levels was evident after STK4 inhibition (). Of note, gene expression profiling data revealed a significant, inverse correlation between STK4 and YAP1 expression levels in MM samples (P < 0.0001, ). Additionally, treatment of MM.1S cells with the proteasome inhibitor bortezomib robustly increased YAP1 protein levels (). Taken together, these results indicate that STK4 controls YAP1 both at the mRNA and protein levels.

We then assessed whether up–regulation of YAP1 induced by STK4 knockdown was associated with reduced proliferation. Indeed, all shRNAs which effectively downregulated STK4 expression and increased YAP1 levels also significantly inhibited MM cell proliferation () and induced a robust apoptotic response ( and ). We further confirmed this phenotype using an independent set of inducible shRNA sequences inserted into another vector or in different MM cell lines (–right panel and ). Importantly, treatment with bortezomib or doxorubicin enhanced this effect (). Additionally, inhibition of STK4 failed to reduce proliferation and increase apoptosis in the YAP1–deleted cell lines KMS–18 and KMS–20 ( and ). To further confirm that YAP1 mediates the phenotypes induced by STK4 inhibition, the expression of STK4 and YAP1 was concomitantly reduced in MM.1S cells with the respective shRNAs, rescuing the phenotype (). These data demonstrate that the effects of STK4 inhibition in MM cells are mediated by restoration of YAP1.

Re–expression of a STK4 mutant devoid of kinase activity, K59R[31], in MM.1S and H929 MM cells down–regulated for STK4, failed to repress YAP1 levels, rescue proliferation, or prevent apoptosis, suggesting that STK4 kinase activity is required to suppress YAP1 thereby preventing apoptosis ( and ).

These results indicate that YAP1 downregulation, seen in MM cells and cell lines in the absence of chromosome 11 deletion, can, at least in part, be due to the inhibitory effects of STK4 upon YAP1 levels.

To demonstrate in vivo the relevance of this synthetically lethal interaction, a genetic approach conditionally knocking down STK4 in MM.1S cells injected subcutaneously in mice was performed. Tumors developed exclusively from MM.1S cells infected with a scrambled shRNA, while no growth was evident in STK4 silenced cells (P < 0.0001; ). Taken together, our results demonstrate that STK4 inhibition upregulates YAP1 levels in MM cells, thereby triggering apoptosis both in vitro and in vivo ().

DNA damage, ABL1, STK4 and YAP1 in lymphoma and leukemia

We next assessed DNA damage in a panel of lymphoma, lymphoblastic and myeloid leukemias, and Waldenström macroglobulinemia cell lines. Staining with γ-H2A.X revealed robust, ongoing DNA damage in the majority of the cell lines assessed (). Moreover, consistent nuclear localization of ABL1 was evident ( and ). YAP1 mRNA and protein levels were low, as in MM ( and ). Remarkably as in MM, cells derived from individuals with leukemia showing low YAP1 expression had a significantly worse prognosis (). The reintroduction of YAP1 in ALL (Jurkat) or AML (OCI/AML3)() cell lines decreased cell number and was associated with apoptosis and induction of p73–target genes ( and ). As in MM, STK4 reduction through STK4 shRNAs increased YAP1 levels, reduced cell number, and enhanced apoptosis ( and ).

Discussion

In this study, we have demonstrated that MM, lymphomas, and leukemias present pervasive DNA damage. As a result, the pro–apoptotic tyrosine kinase ABL1 relocalizes into the nucleus, uncovering for the first time an unexpected broad role for this protein in inducing apoptosis during the DDR response. Tumor cells nevertheless escape apoptosis due to genetic inactivation or reduced expression of the Hippo co–transcription factor YAP1. Importantly, we elucidate a novel synthetic lethal approach[32] in which inhibition of the kinase STK4 reactivates YAP1 and triggers apoptosis, providing the rationale for developing novel STK4 inhibitors, for clinical evaluation in hematological malignancies.

As in neoplasms of epithelial origin[1], activation of DNA damage checkpoint might also represent a barrier against the evolution towards cancer in hematological tissues; however, unlike epithelial cancers, hematological malignancies do not seem to require p53 inactivation. Instead, early inactivation of the ABL1/YAP1/p73 axis may substitute for p53 mutations and/or inactivation in hematological tumors.

YAP1 is focally amplified in a vast array of solid tumors including brain, colon and hepatocellular carcinomas, and has been consistently reported as an oncogene in epithelial cancers[33]. Our data support a role for YAP1 as a tumor suppressor gene in hematological cancers. A possible explanation for this differential function may relate to YAP1 forming complexes with various partners with distinct functional sequelae, depending on the cellular context[34]. For example, in the absence of DNA damage YAP1 preferentially interacts with oncogenic transcription modulator RUNX, leading to increased degradation of p73[20]. Thus, transcription modulators can shift YAP1 away from p73 towards other partners, endowing YAP1 with proliferative and anti-apoptotic features.

The heart of the Hippo pathway includes a core of two serine–threonine kinase pairs, STK4 (MST1) and STK3 (MST2), which activate two other kinases, LATS1 and LATS2. In turn, LATS1 and LATS2 regulate YAP1 and TAZ[35]. The concomitant loss of both STK4 and STK3 is required to foster the development of hepatocellular carcinoma in vivo[29]. In our study, the used STK4–specific shRNAs did not affect STK3 mRNA levels (), suggesting that STK4 exerts a prominent, specific role in controlling YAP1 levels in MM. As for other Hippo pathway members, recent intriguing findings have implicated LATS2 in DNA damage and cell cycle control[36,37], suggesting that the Hippo pathway exerts a tight control on DNA damage and reinforcing the notion of a tumor suppressive role for the this pathway in specific contexts such as hematological cancers. Ultimately, YAP1 presents a high degree of homology with TAZ. Although they often present overlapping functions, the two proteins seem to exert independent roles, exemplified by the phenotypes of the mouse knockouts and the specific functions of TAZ in stem cell maintenance and WNT regulation[38]. Unlike YAP1, we did not detect focal deletions affecting the TAZ locus in MM cell lines and cells derived from individuals with MM, suggesting that tumors of hematological lineage preferentially inactivate YAP1.

Kinases are valuable targets for cancer therapies, as already proven by imatinib, erlotinib, and gefitinib. Therefore, STK4 provides a promising therapeutic target in hematological cancers displaying low YAP1 levels and concomitant DNA damage. Importantly, STK4 knockdown induced apoptosis in both p53–WT and mutant cells, suggesting that restoring YAP1 levels could represent a novel treatment strategy even in tumors with p53 inactivation and poor prognosis.

Experimental Procedures

Materials, Reagents and cell lines

All antibodies and assay reagents were obtained from commercial sources as described in Supplemental Experimental Procedures. All MM cell lines were purchased from American Type Culture Collection (ATCC), established in our laboratory or kindly provided by our collaborators.

Primary cells

Blood samples from healthy volunteers and MM bone marrow samples were processed as described in Supplemental Experimental Procedures.

Cell Culture and Molecular Methods

Methods for cell culture, DNA and RNA extraction, mRNA quantification, subcellular fractioning, immunoblotting, immunofluorescence, as well as assays for viability, cellular growth, and apoptosis are described in Supplemental Experimental Procedures.

Lentiviral mediated gene transfer and transient transfection of MM cell lines.

MM cells (3,000,000) were transiently transfected by nucleoporation (LONZA, Cell Line Nucleofector Kit V, Amaxa Biosystems). Lentiviral infection, selection and induction were performed following standard procedures (see also Supplemental Experimental procedures).