CIB2 Negatively Regulates Oncogenic Signaling in Ovarian Cancer via Sphingosine Kinase 1

The family of calcium and integrin binding proteins (CIB) is composed of four EF-hand Ca²⁺-binding proteins termed CIB1–4 (1). Of these, CIB1 has been most widely studied, with numerous binding partners identified (2), including the αIIbβ₃ integrin (3) and sphingosine kinase 1 (SK1; ref. 4). A number of cellular functions have been proposed for CIB1, including roles in platelet activation, spermatogenesis, cell survival, and as a Ca²⁺-myristoyl switch protein mediating the cell signaling of SK1 (2, 4, 5). While considerable work has examined the roles of CIB1, comparatively little is known regarding the function of the other CIB family members. Despite various cellular functions of CIB1, the Cib1^−/− mice display no overt phenotype other than male infertility (5, 6), leading to suggestions that functional redundancies may exist between some of the CIB proteins (6).

CIB2 shows approximately 60% amino acid sequence similarity to CIB1, and both proteins appear to share a similar hydrophobic target binding pocket (7, 8). Indeed, both proteins bind Ca²⁺ and both interact with the cytoplasmic domain of αIIb integrin, at least in vitro (1). Despite these similarities, and the observations that both proteins are ubiquitously expressed in human tissues (8, 9), the only reported shared function between CIB1 and CIB2 involves mediating infection of HIV in human T cells (10), while the other reported role of CIB2 involves the regulation of Ca²⁺ homeostasis in the sensory neurons of the ear and eye (8, 9).

Recently, we described that CIB1 is overexpressed in a variety of cancers and can exert oncogenic effects by enhancing the plasma membrane localization of SK1 (11). Here, based on the sequence and presumed structural similarity between CIB1 and CIB2, we examined the potential role for CIB2 in regulation of SK1. We found that, indeed, CIB2 interacts with SK1 in the same binding site as CIB1. Unlike CIB1, however, we found that CIB2 lacks a Ca²⁺-myristoyl switch function, and as a result has a surprising opposite function to CIB1 by blocking the plasma membrane localization of SK1 and inhibiting its subsequent prosurvival and pro-oncogenic signaling. Furthermore, we found that CIB2 is significantly downregulated in ovarian cancer, and acts as a potential tumor suppressor in the disease via inhibiting tumor growth and cell migration/invasion, and enhancing chemosensitivity of ovarian cancer cells.

Human embryonic kidney (HEK293 and HEK293T), human cervical carcinoma (HeLa), and murine fibroblast (NIH3T3) cells were obtained from the ATCC over the period 2007–2014, and cultured as described previously (4, 12). SKOV3 and OV90 human ovarian cancer cell lines were obtained from ATCC prior to 2010 and cultured in RPMI medium supplemented with 10% FCS, 100 U/mL penicillin, 100 μg/mL streptomycin, and 250 ng/mL Fungizone (Gibco) in a humidified atmosphere with 5% CO₂. All human cell lines were authenticated prior to publication by short tandem repeat DNA fingerprinting, tested regularly for mycoplasma by PCR analysis (13), and used after less than 25 passages from resuscitation. Immortalized murine embryonic fibroblasts (MEF) generated from 14.5 days postcoitum embryos of Sphk1^−/− mice (14) were cultured in the same manner, but maintained in a humidified atmosphere with 10% CO₂. Details of plasmids and cell transfection are provided in Supplementary Methods. Primary ovarian cancer cells were derived from ascites of serous ovarian cancer patients and cultured as described previously (15) with informed patient consent and approval from the Royal Adelaide Hospital Research Ethics Committee in accordance with the ethical guidelines of the Declaration of Helsinki. Focus formation and colony formation assays were performed as described previously (16). Myristoylation of HA-tagged CIB2 was assessed in HEK293T cells labeled with [³H]myristic acid as described previously (4).

Sources of all antibodies employed are provided in Supplementary Methods. Lysates for coimmunoprecipitation were prepared in 50 mmol/L Tris/HCl buffer (pH 7.4) containing 150 mmol/L NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 10 mmol/L NaF, 10 mmol/L Na₃VO₄, and 10% glycerol and incubated with Protein G MACS MicroBeads (Miltenyi Biotec) and anti-FLAG antibodies for 1 hour on ice. The immune complexes were captured using μMACS separation columns (Miltenyi Biotec), washed, and eluted according to the manufacturer's instructions. Samples were subjected to SDS-PAGE followed by Western blotting. To examine interaction between endogenous SK1 and CIB2, clarified brain lysates from wild-type C57Bl/6 mice were prepared in 50 mmol/L Tris/HCl buffer (pH 7.4) containing 150 mmol/L NaCl, 2 mmol/L Na₃VO₄, 10 mmol/L NaF, 1 mmol/L EDTA, 10% glycerol, 0.05% Triton X-100, and 10 mmol/L β-glycerophosphate and incubated with anti-SK1 antibodies or control rabbit IgG overnight at 4°C. The immune complexes were then incubated with Protein A MACS MicroBeads for 30 minutes on ice, applied to μMACS separation columns, and analyzed by Western blotting as described above.

Interaction between endogenous SK1 and CIB2 was assessed using the Duolink In Situ Proximity Ligation Assay Kit (Sigma-Aldrich), according to the manufacturer's instructions, employing anti-SK1 and anti-CIB2 antibodies.

Recombinant GST-CIB2 protein was generated from E. coli as described previously for GST-CIB1 (4). Pull-down analyses were performed by incubating with glutathione-sepharose–bound GST-CIB2 or GST as described previously (4).

Cells were plated onto poly-l-lysine–coated glass slides, cultured for 24 hours, and then immunofluorescently stained for HA and endogenous SK1 as described previously (4). Overexpressed SK1 visualized by its EGFP tag. To assess agonist-induced translocation of SK1, CIB1, and CIB2, the transfected HeLa cells were stimulated with 1 μg/mL phorbol 12-myristate 13-acetate (Sigma-Aldrich) for 30 minutes or 2 μmol/L ionomycin (Calbiochem) for 2 minutes prior to immunofluorescence staining.

SK1-selective activity assays were performed as described previously (17). S1P generation in intact cells was assessed by labeling cells with ³H-sphingosine (Perkin Elmer) as described previously (18). S1P concentration in cells was assessed by S1P ELISA (Echelon Biosciences), according to the manufacturer's instructions.

RNA extracted from ovarian cancer cells was used to generate cDNA using the QuantiTect Reverse Transcription Kit (Qiagen) according to the manufacturer's protocol. CIB2 expression was assessed by quantitative RT-PCR (qRT-PCR) using a Rotor-Gene 6000 (Qiagen). CIB2 and SK1 expression in TissueScan starter kit and TissueScan ovarian cancer tissue panel II cDNA panels (Origene) was assessed by quantitative RT-PCR on a 7500 ABI system (Thermo Fisher Scientific) and normalized to the level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). PCR conditions and primer sequences are described in Supplementary Methods.

SKOV3 or OV90 cells (3 × 10⁴ cells) were seeded in serum-free medium in the top chamber of a CIM-plate 16 (AECA Biosciences). Cell migration into the bottom chambers containing RPMI medium supplemented with 10% FCS was measured using the xCELLigence RTCA DP system (AECA Biosciences) according to the manufacturer's instructions. Invasion of SKOV3 cells was assessed by chick chorioallantoic membrane (CAM) assay as described previously (19), with 2.5 × 10⁵ cells per Matrigel graft. At the end of the experiments, Matrigel grafts with surrounding CAM were harvested, fixed with 4% paraformaldehyde, paraffin embedded, and serially sectioned (6 μm). Invasion of SKOV3 cells into the mesoderm layer of the CAM was assessed by hematoxylin and eosin staining as well as IHC staining with anti-human anti-cytokeratin antibodies as described previously (19).

TNFα-induced apoptosis of transfected HeLa cells was assessed as described previously (4). SKOV3 and OV90 cells were cultured for 24 hours to assess chemosensitivity, and treated with carboplatin (CBP, Hospira) or paclitaxel (Sigma-Aldrich), either alone or in combination with SK1-I (Sapphire Bioscience). Viable cells were assessed after 72 hours of treatment using the CellTiter 96 AQ_ueous MTS assay (Promega) according to the manufacturer's protocol.

SKOV3 cells (5 × 10⁶) were subcutaneously injected into the flanks of 7-week-old female NOD-SCID mice. Tumor growth was monitored daily for 4 weeks and tumor volumes were calculated by the formula (width² × length)/2. Animal studies were performed with approval from the SA Pathology/Central Adelaide Local Health Network Animal Ethics Committee.

Our previous studies demonstrated an important interaction of CIB1 with SK1 (4, 11). Because of the high sequence and presumed structural similarity between CIB1 and CIB2, we examined whether CIB2 also interacted with SK1. To do this, we initially employed coimmunoprecipitation from lysates of HEK293T cells coexpressing both SK1 and CIB2. This demonstrated a clear interaction between CIB2 and SK1, with each protein present in the isolated immunocomplexes of the other protein (Fig. 1A and B). Interaction between endogenous CIB2 and SK1 was also observed by coimmunoprecipitation using lysates from mouse brains, where high levels of both CIB2 (20, 21) and SK1 (22) occur (Fig. 1C). This physiologic interaction was further validated by in situ proximity ligation assays (PLA) in HEK293 cells that demonstrated positive cytoplasmic PLA signals only in the presence of both anti-CIB2 and anti-SK1 antibodies (Fig. 1D). Similar results were found with two other different sets of anti-CIB2 and anti-SK1 antibodies (Supplementary Fig. S1A and S1B). This is indicative of interactions between endogenous SK1 and CIB2, and also suggests that the interaction mainly occurs in the cytoplasm. We also tested whether this interaction was affected by Ser225 phosphorylation of SK1, an important regulator of both SK1 activity and plasma membrane localization (23). Pull-down experiments using purified GST-CIB2 protein and lysates from HEK293T cells expressing SK1 demonstrated that CIB2 was able to bind the Ser225→Ala mutant of SK1 in a comparable manner to wild-type SK1 (Fig. 1E), suggesting that like CIB1 (4), the interaction between CIB2 and SK1 occurs independently of SK1 phosphorylation.

To further define the interaction of CIB2 and SK1, we then characterized the CIB2-binding region of SK1. Our previous studies showed that the CIB1-binding site resides on the α8 helix of SK1 and includes Phe197 and Leu198 as critical residues for the interaction (4). Pull-down analysis with CIB2 revealed that the Phe197 and Leu198 in SK1 were also important for the association with CIB2 as mutation of these residues attenuated the interaction (Fig. 1F).

As the Ca²⁺-myristoyl switch function of CIB1 is critical for the plasma membrane localization and signaling of SK1 (4), we investigated whether CIB2 also acts in the same manner. Ca²⁺-myristoyl switch proteins like CIB1 are dependent on myristoylation, and as CIB2 also contains the N-terminal myristoylation consensus sequence of MGxxxS/T (where x represents any amino acid; ref. 7), we initially examined whether CIB1 was myristoylated. HeLa cells expressing CIB2 (with a C-terminal HA-tag) were labeled with ³H-myristic acid. Fluorography was then performed on CIB2 immunoprecipitated from the cell lysates via its HA-tag to determine incorporation of the radiolabeled myristic acid. Consistent with previous reports (20), we found that CIB2 was, indeed, myristoylated (Fig. 2A).

We next investigated whether the interaction of CIB2 with SK1 was dependent on Ca²⁺, another common determinant of Ca²⁺-myristoyl switches (24). Notably, in contrast to the situation for CIB1 (4), this showed that the interaction of CIB2 with SK1 was not dependent of the presence of Ca²⁺ or Mg²⁺ (Fig. 2B). In fact, the presence of Ca²⁺ consistently resulted in a slight reduction in detection of the SK1–CIB2 interaction. As Ca²⁺-myristoyl switch function is also characterized by the Ca²⁺-induced relocation of the protein to cell membranes, we next examined whether this occurred following treatment of HeLa cells with the Ca²⁺-ionophore ionomycin or phorbol ester (PMA). Consistent with previous findings (4), stimulation of cells with ionomycin resulted in clear translocation of CIB1 to the plasma membrane (Fig. 2C). In stark contrast, however, no such translocation of CIB2 was observed in response to either ionomycin or PMA (Fig. 2C and D). Together, these findings indicate that despite being myristoylated, CIB2 lacks Ca²⁺-myristoyl switch behavior.

To assess whether the myristoylation of CIB2 altered its interaction with SK1, we employed coimmunoprecipitation of SK1 with wild-type or the non-myristoylatable Gly2→Ala mutant of CIB2 (20). Both versions of CIB2 interacted with SK1 in a similar manner (Supplementary Fig. S2), suggesting that myristoylation has no effect on the interaction between CIB2 and SK1.

The previous results suggested that CIB2 was unlikely to function in the same manner as CIB1 in mediating the translocation of SK1 to the plasma membrane. However, as CIB2 associates with SK1 at the same site as CIB1, we examined the possibility that CIB2 may act as an endogenous inhibitor of SK1 translocation. Thus, we examined the localization of endogenous SK1 in cells overexpressing CIB2, with and without exposure to PMA, a known inducer of SK1 translocation (23, 25). Consistent with previous observations by us and others (4, 25), in addition to cytoplasmic SK1, strong nuclear staining of endogenous SK1 was also visible in all cells examined. In control cells, however, endogenous SK1 showed clear translocation to the plasma membrane following stimulation with PMA (Fig. 2D). On the contrary, no such translocation of SK1 was observed in cells overexpressing CIB2 (Fig. 2D). As agonist-induced translocation of SK1 to the plasma membrane is dependent on ERK1/2–mediated phosphorylation (23), we next examined whether CIB2 had any unexpected effect on ERK1/2 activation directly upstream of SK1 phosphorylation. We found that the PMA-induced ERK1/2 (Fig. 2E) and SK1 activation (Fig. 2F) were both unaffected by the expression of CIB2, suggesting that the effects of CIB2 were likely to be mediated via direct binding of SK1 rather than via indirect modulation of upstream pathways.

As CIB2 expression blocked the agonist-induced translocation of endogenous SK1 to the plasma membrane, we next investigated whether CIB2 expression also affected S1P generation in cells. Indeed, despite not altering the total SK1 activity in the cells (Fig. 2F), overexpression of CIB2 significantly inhibited S1P generation (Fig. 3A) and reduced cellular S1P levels (Fig. 3B). This suggests that by blocking the plasma membrane association of SK1, CIB2 prevents SK1 from accessing pools of its substrate, sphingosine, and hence synthesis of S1P.

As the subcellular localization of SK1 is crucial for its role in cellular signaling (26), we then examined the effects of CIB2 on the prosurvival and pro-oncogenic signaling of SK1. TNFα engagement of its receptors can elicit both prosurvival and proapoptotic signaling (27). Previous studies have shown that TNFα-induced activation of SK1 is important for prosurvival signaling by TNFα, with targeting SK1 activity or blocking translocation of SK1 to the plasma membrane enhancing TNFα-induced apoptosis (4, 28, 29). Thus, we next examined whether CIB2 expression could affect the TNFα-associated prosurvival signaling, and we found that, indeed, CIB2 significantly sensitized HeLa cells to apoptosis following treatment with TNFα (Fig. 3C; Supplementary Fig. S3).

Targeting SK1 with chemical inhibitors or a dominant-negative SK1 has been previously shown to attenuate neoplastic cell transformation induced by oncogenic Ras (16). As we recently also showed that plasma membrane localization of SK1 is crucial for oncogenic signaling by Ras (11), we next investigated whether CIB2 could affect this process. Consistent with previous observations (11, 30), oncogenic Ras resulted in potent relocalization of SK1 from the cytoplasm to the plasma membrane (Fig. 3D). This relocalization of SK1, however, was blocked by the overexpression of CIB2 (Fig. 3D), as was Ras-induced neoplastic transformation, assessed by the ability of NIH3T3 cells to overcome contact inhibition in the focus formation assays (Fig. 3E). No effect was observed by CIB2 on ERK1/2 activation by oncogenic Ras (Fig. 3F), further suggesting the inhibitory effect of CIB2 directly on SK1.

As our findings suggested that CIB2 has a negative effect on Ras and SK1-mediated oncogenic signaling this raised the possibility that CIB2 acts as a tumor suppressor protein. Thus, we next examined by qRT-PCR the expression of CIB2 in a panel of various human tumors, compared with corresponding normal tissues. While CIB2 expression was highly variable in lung tumor tissues and showed no difference in breast or kidney tumors, it was markedly (8- to 25-fold) reduced, compared with normal ovary, in all ovarian tumor tissues examined in this small survey panel (Fig. 4A). Further qRT-PCR examination of these findings in a more extensive panel of human ovarian cDNA showed that CIB2 expression in cancerous ovarian tissues was significantly lower than that detected in normal ovarian tissues (Fig. 4B). Some variability in CIB2 expression was observed in these cancer tissues; however, this was not related to tumor stage or tumor grade (Supplementary Fig. S4), potentially suggesting loss of CIB2 expression is an early event contributing to ovarian cancer development. Examination of CIB2 expression by qRT-PCR in the human ovarian cancer cell lines SKOV3 and OV90 also showed significantly reduced CIB2 expression compared with normal ovarian tissues (Fig. 4C). Together, these findings are consistent with a potential tumor suppressor role for CIB2.

Using the same panel of ovarian cDNA, we also assessed the gene expression of SK1. This demonstrated that, consistent with a previous report (31), SK1 was significantly upregulated in ovarian cancer, compared with normal ovary (Supplementary Fig. S5A). However, no correlation was observed between SK1 expression and tumor stage nor tumor grade (Supplementary Fig. S5B).

To investigate the potential tumor suppressor role of CIB2 in ovarian cancer, we first generated SKOV3 and OV90 cells stably reexpressing CIB2 (Fig. 5A; Supplementary Fig. S6A). Immunofluorescence analysis showed that in both control SKOV3 and OV90 cells endogenous SK1 showed clear localization at the plasma membrane (Fig. 5B; Supplementary Fig. S6B), which is consistent with the low level of CIB2 in these cells (Fig. 4C). This localization of SK1 was completely blocked by reexpression of CIB2, with no SK1 found at the plasma membrane of these cells (Fig. 5B; Supplementary Fig. S6B), suggesting a negative regulatory role for CIB2 in the oncogenic signaling by SK1. Consistent with this, we observed that neoplastic growth of SKOV3 cells, as measured by colony formation in soft agar, was substantially hindered by reexpression of CIB2, which resulted in significantly reduced number of colonies (Fig. 5C). A similar reduction in neoplastic growth was also observed in OV90 cells reexpressing CIB2 (Supplementary Fig. S6C). Subsequent xenografting of the CIB2 reexpressing SKOV3 cells into NOD/SCID mice demonstrated that CIB2 reexpression also resulted in reduced tumor growth (Fig. 5D), further supporting CIB2 as a tumor suppressor in ovarian cancer cells.

The mortality of ovarian cancer is largely attributed to the high motility and invasiveness of ovarian cancer cells (32). As SK1 has been implicated in regulating cell migration and invasion of multiple cancer cell types, including ovarian cancer (33), we interrogated the role of CIB2 in this process. Initially, we employed transwell migration assays, where we observed that reexpression of CIB2 in SKOV3 cells resulted in an approximate 50% reduction in cell migration compared with control cells (Fig. 6A). Similarly, a significant reduction in cell migration was also observed in OV90 cells stably reexpressing CIB2 (Supplementary Fig. S6D). Next, we examined the effect of CIB2 using chicken CAM assays, which have been widely utilized to assess tumor cell invasion in vivo. Matrigel grafts containing SKOV3 cells reexpressing CIB2 or control SKOV3 cells were placed on top of the CAM of chick embryos 11 days following fertilization. Consistent with previous observations (19), control SKOV3 cells readily invaded through the ectoderm into the mesoderm of CAM over a 3-day period. In stark contrast, however, invasion of cells reexpressing CIB2 was significantly hindered, with the majority of the cells proliferating against the ectoderm without invading (Fig. 6B and C).

The elevated expression of SK1 is associated with the development of chemoresistance in multiple cancers (34). Consistently, Kaplan–Meier analysis of publicly available ovarian cancer datasets (GSE14764) showed a significant association between high SK1 expression and poor progression-free survival in patients that had undergone standard chemotherapy (platin and taxol; Fig. 7A). In contrast, and consistent with a negative regulatory role on SK1, higher CIB2 expression was associated with a significantly improved prognosis in these patients (Fig. 7B). To examine whether the level of CIB2 is associated with chemoresistance in ovarian cancer patients, we then examined CIB2 gene expression in a panel of primary ovarian cancer cells from a cohort of patients (Supplementary Table S1) that had undergone standard chemotherapy. Consistent with a tumor suppressor role of CIB2, we observed significantly reduced CIB2 expression in chemoresistant tumor cells, compared with those that were sensitive to chemotherapy (Fig. 7C). Consistent with the chemosensitizing effects of SK1 inhibition previously observed in various other cancer cell types (35), we found the SK1-selective inhibitor, SK1-I (36), acted synergistically with both carboplatin and paclitaxel to induce cell death in SKOV3 (Fig. 7D and E) and OV90 (Supplementary Fig. S6E and S6F) ovarian cancer cells. Similarly, reexpression of CIB2 resulted in elevated sensitivity to carboplatin in SKOV3 cells (Fig. 7F), and also the more CBP-sensitive OV90 cells (Supplementary Fig. S6G). However, CIB2 reexpression did not improve sensitivity to paclitaxel in SKOV3 cells (Fig. 7G), although some effects were observed in OV90 cells (Supplementary Fig. S6H). Together, these results suggest CIB2 is involved in regulating chemosensitivity of ovarian cancer cells.

Ovarian cancer is the most lethal gynecologic malignancy. Because of lack of early detection, over 70% of ovarian cancer cases are diagnosed at advanced stage, where the cancer cells have metastasized through the abdominal cavity (32). The current standard therapy for ovarian cancer consists of debulking surgery followed by adjuvant chemotherapy with carboplatin and paclitaxel (37). Initial response to treatment is high, but the majority of patients eventually relapse and acquire chemotherapeutic resistance, which is the primary cause of ovarian cancer–related death. Prognosis with five-year survival rates of about 40% is therefore very poor (37). Hence, more efficient novel treatment strategies are warranted, as is a better understanding of the molecular mechanisms underlying the pathogenesis and development of the disease. In the current study, we demonstrated that CIB2 associates with and inhibits oncogenic signaling by SK1, a lipid kinase previously implicated in ovarian cancer progression (33). We found loss of CIB2 expression is common in ovarian cancer and that its reexpression inhibits ovarian tumor growth, migration, and invasion; and enhances the chemosensitivity of ovarian cancer cells. Thus, this work has identified CIB2 as a novel tumor suppressor and potential prognostic marker for ovarian cancer. It has also established the therapeutic potential, and mechanistic basis, of targeting SK1 in ovarian cancer.

CIB2 is a ubiquitously expressed Ca²⁺-binding protein with largely unknown molecular functions. Early studies proposed roles for CIB2 in signaling from N-methylo-d-aspartate (NMDA) receptors in hippocampal neurons (20) and integrin α7Bβ1D in skeletal muscle (21). More recently, CIB2 has been described to have a role in the development, maintenance, and function of mechanosensory stereocilia of inner ear hair cells, as well as in phototransduction in the retina and protection from light-dependent retinal degeneration through the regulation of Ca²⁺ homeostasis in these cells (8, 9). A further recent study also reported a shared role between CIB2 and CIB1 in assisting receptor-mediated viral entry of human immunodeficiency virus into CD4⁺ T lymphocytes. Notably, however, despite the identification of several binding partners for CIB2, the mechanisms and downstream mediators underlying these highly tissue-specific roles of CIB2 remain unknown. Furthermore, the ubiquitous expression of CIB2 suggests much broader roles for this protein. As described in the current study, this includes the inhibition of signaling by the ubiquitously expressed SK1, which has broad implications across various tissues.

SK1 has a demonstrated role in cancer development, tumor growth, angiogenesis, and resistance to chemotherapies in many neoplasms (38), including ovarian cancer (33). Our previous studies showed that the oncogenic signaling of SK1 depends on its localization to the plasma membrane (12), a process mediated by CIB1 via its Ca²⁺-myristoyl switch function (4). Consistent with this notion, blocking the plasma membrane localization of SK1 through ablating CIB1 expression or via expression of an engineered myristoylation-deficient version of CIB1 was found to inhibit both TNFα-induced cell survival (4) and Ras-induced neoplastic transformation (11). Notably, in the current study, CIB2 expression elicited a similar inhibitory effect on plasma membrane localization and oncogenic signaling by SK1. This, together with our findings that CIB2 and CIB1 interact with the same region on SK1, suggest that CIB2 is an endogenous negative regulator of the SK1/CIB1 system and likely competes with CIB1 for SK1 binding.

Although CIB1 and CIB2 share sequence and presumed structural similarity, they are reported to undergo different conformational changes upon engagement with Ca²⁺(1), which may differentially alter their association with SK1. Indeed, CIB1 has a high dependency on Ca²⁺ for most of its protein interactions, including with SK1 (2, 4). In contrast, CIB2 has been previously shown to interact with other proteins in a manner not dependent on Ca²⁺ (1), like we observed for SK1 where Ca²⁺ appears to even moderately weaken the CIB2–SK1 interaction (Fig. 2B). These different modes of SK1 binding, together with their observed differential subcellular trafficking, provide a possible mechanism for the coordinated regulation of SK1 by the CIB proteins. Thus, it is likely that under basal conditions, with low cytoplasmic Ca²⁺ levels, SK1 occurs preferentially in a complex with CIB2 in the cytoplasm, where it is sequestered away from the plasma membrane. Upon cell stimulation, however, associated increases in cytoplasmic Ca²⁺ may allow CIB1 to temporarily appropriate SK1 from CIB2 and transport it to the plasma membrane where it has known cell signaling (12). Thus, via this mechanism, both CIB1 and CIB2 play key roles in regulating the signaling activity of SK1. Disruption of this system, either by upregulation of CIB1, which we recently reported in multiple cancers, including ovarian cancer (11), or by downregulation of CIB2, such as we have shown in the current study that occurs in ovarian cancer, will lead to elevated levels of SK1 at the plasma membrane, and resultant aberrant oncogenic signaling.

Despite the well-established role of SK1 in other cancers, its therapeutic value in ovarian cancer is only beginning to emerge. SK1 expression is elevated in ovarian cancer (31, 39, 40) and a higher level of SK1 is found in metastatic tumors and has been associated with poor patient survival (31, 41). Furthermore, studies have shown that targeting SK1 by either RNAi knockdown or chemical inhibition can impede proliferation and induce apoptosis of ovarian cancer cells through p38 activation and Akt inhibition (39, 40, 42). Interestingly, the S1P receptor agonist, FTY720, which also weakly inhibits SK1 (43) attenuates the growth of xenograft tumors arising from both ovarian cancer cell lines and patient-derived cells (40). Notably, our studies suggest targeting SK1 signaling via reexpression of CIB2 also attenuates the neoplastic growth of ovarian cancer cells in vitro and in vivo (Fig. 5C and D; Supplementary Fig. S6C).

Upregulation of SK1 is associated with chemoresistance in many solid cancers, including colon, breast, prostate, and pancreatic cancers (38). Our analysis suggests this is also the case in ovarian cancer, where higher SK1 expression correlated with poorer survival of ovarian cancer patients treated with platinum and taxane-based chemotherapies (Fig. 7A); findings that are consistent with recent studies showing higher SK1 expression correlated with poorer overall survival in ovarian cancer patients (31). Notably, however, we also found a clear correlation between low levels of CIB2 and poor survival of ovarian cancer patients treated with platinum and taxane-based chemotherapy (Fig. 7B). This, together with our analysis of primary ovarian cancer patient cells indicating a link between low CIB2 levels and the development of resistance to chemotherapy (Fig. 7C), suggests CIB2 as a promising prognostic marker for this disease. This, however, should be analyzed further in prospective studies with tumors from larger ovarian patient cohorts.

Few studies have examined the chemosensitizing effects of targeting SK1 in ovarian cancer. SKI-II, a poorly selective sphingosine kinase inhibitor, does sensitize A2780 ovarian cancer cells to the N-(4-hydroxyphenyl)retinamide, a synthetic retinoid that has cytotoxic effects on various cancer cell lines through its ability to induce elevated cellular ceramide levels (42). Our studies, using the SK1-selective inhibitor, SK1-I demonstrate that inhibition of SK1 is effective in sensitizing ovarian cancer cells to the first-line chemotherapeutic agents carboplatin and paclitaxel (Fig. 7D and E; Supplementary Fig. S6E and S6F). Furthermore, we observed similar chemosensitizing effects with targeting SK1 through reexpression of CIB2 in ovarian cancer cells also enhancing cell killing by these chemotherapeutics (Fig. 7F and G; Supplementary Fig. S6G and S6H).

S1P levels are elevated in the ascites of metastatic ovarian cancer patients (44) and several studies have demonstrated its ability to stimulate migration and invasion of ovarian cancer cells (44–46). Similarly, elevated SK1 expression is also associated with ovarian cancer metastasis (41). Targeting of SK1 by siRNA knockdown or chemical inhibition has been reported to suppress both ovarian cancer cell migration and invasion in vitro (40, 41). Notably, our findings also demonstrated that targeting SK1 through reexpression of CIB2 suppressed the migratory and invasive properties of ovarian cancer cells in vitro and in vivo (Fig. 6; Supplementary Fig. S6D). The migration and invasion of ovarian cancer cells relies on mediators such as metalloproteinases (MMP), p38 kinase, and PI3K signaling, which are also known to be induced by S1P (33). Whether these pathways are also altered by the reexpression of CIB2, however, requires further elucidation.

In conclusion, the current study has identified CIB2 as a potential tumor suppressor with prognostic impact and promising therapeutic target in ovarian cancer. The strength of this study is that it has also defined the mechanistic basis for the tumor suppressor action of CIB2 via the novel discovery of its role in attenuation of oncogenic signaling by SK1. Indeed, we found that low CIB2 or high SK1 expression correlated with poorer survival of ovarian cancer patients treated with platinum- and taxane-based chemotherapies, highlighting the prognostic significance of this pathway. Our findings also indicate that targeting the CIB2/SK1 axis has positive therapeutic impact on multiple aspects of ovarian cancer cell biology, including cell survival, migration, invasion, chemotherapeutic sensitivity, and tumor growth. This, combined with the knowledge that SK1 is a promising target in multiple human cancers (33, 38, 47), highlights the potential utility of SK1 inhibitors in the treatment of ovarian cancer. Thus, CIB2 and SK1 are novel prognostic biomarkers that can be used to aid ovarian cancer patient management but also offer opportunities for new molecular targeted therapies to improve clinical outcome in this disease. Whether CIB2 also plays a similar role in some other cancers remains to be determined.

No potential conflicts of interest were disclosed.

Conception and design: W. Zhu, K.E. Jarman, M.R. Pitman, S.M. Pitson

Development of methodology: W. Zhu, K.E. Jarman, N.A. Lokman, P.A.B. Moretti, M.K. Oehler, M.R. Pitman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W. Zhu, K.E. Jarman, H.A. Neubauer, L.T. Davies, B.L. Gliddon, H. Taing, M.K. Oehler, M.R. Pitman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Zhu, K.E. Jarman, H.A. Neubauer, B.L. Gliddon, M.R. Pitman, S.M. Pitson

Writing, review, and/or revision of the manuscript: W. Zhu, P.A.B. Moretti, M.K. Oehler, M.R. Pitman, S.M. Pitson

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W. Zhu, M.K. Oehler

Study supervision: M.R. Pitman, S.M. Pitson

We thank J. Downward (Imperial Cancer Research Fund, London) for providing the HRas plasmid and M. Okabe (Osaka University) for providing the pCX-EGFP plasmid.

This work was supported by the Fay Fuller Foundation, Project Grant (1004695) and Senior Research Fellowship (1042589) from the National Health and Medical Research Council of Australia (to S.M. Pitson), a Royal Adelaide Hospital (RAH) Clinical Research Grant, University of Adelaide Graduate Research and Training Scheme Scholarships (to W. Zhu), an Australian Postgraduate Award and RAH Dawes Scholarship (to K.E. Jarman), RAH Early Career Fellowship (to M.R. Pitman) and a Project Grant from the Ovarian Cancer Research Foundation (to M.K. Oehler).

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.