Calcium store sensor stromal-interaction molecule 1-dependent signaling plays an important role in cervical cancer growth, migration, and angiogenesis

Abstract

Store-operated Ca²⁺ entry (SOCE) is the principal Ca²⁺ entry mechanism in nonexcitable cells. Stromal-interaction molecule 1 (STIM1) is an endoplasmic reticulum Ca²⁺ sensor that triggers SOCE activation. However, the role of STIM1 in regulating cancer progression remains controversial and its clinical relevance is unclear. Here we show that STIM1-dependent signaling is important for cervical cancer cell proliferation, migration, and angiogenesis. STIM1 overexpression in tumor tissue is noted in 71% cases of early-stage cervical cancer. In tumor tissues, the level of STIM1 expression is significantly associated with the risk of metastasis and survival. EGF-stimulated cancer cell migration requires STIM1 expression and EGF increases the interaction between STIM1 and Orai1 in juxta-membrane areas, and thus induces Ca²⁺ influx. STIM1 involves the activation of Ca²⁺-regulated protease calpain, as well as Ca²⁺-regulated cytoplasmic kinase Pyk2, which regulate the focal-adhesion dynamics of migratory cervical cancer cells. Because of an increase of p21 protein levels and a decrease of Cdc25C protein levels, STIM1-silencing in cervical cancer cells significantly inhibits cell proliferation by arresting the cell cycle at the S and G2/M phases. STIM1 also regulates the production of VEGF in cervical cancer cells. Interference with STIM1 expression or blockade of SOCE activity inhibits tumor angiogenesis and growth in animal models, confirming the crucial role of STIM1-mediated Ca²⁺ influx in aggravating tumor development in vivo. These results make STIM1-dependent signaling an attractive target for therapeutic intervention.

In most types of cells, modulation of intracellular Ca²⁺ levels provides versatile and dynamic signaling that mediate various cellular processes, such as proliferation, migration, and gene expression (1). Dysregulation of Ca²⁺ signaling has been identified in tumor progression (2). The store-operated calcium entry (SOCE) is a major Ca²⁺ entry in nonexcitable cells (3, 4). SOCE, by definition, is activated by Ca²⁺ efflux from the internal store. Two genes, STIM1 (stromal-interaction molecule 1) and Orai1, are responsible for SOCE activation (3). Once endoplasmic reticulum Ca²⁺ depleted, STIM1 proteins aggregate into multiple puncta and translocate to the close proximity of plasma membranes (5). Orai1 molecule, an essential pore-forming component of the SOCE channel, translocates to the same STIM1-containing structures during store depletion and opens to mediate Ca²⁺ entry (6). The physiological function of STIM1 has mostly been studied in immune systems (7). STIM proteins are required for the development and function of regulatory T cells (8) and STIM1-deficiency causes several autoimmune diseases and myopathy in human subjects and mouse models (9). The chromosomal location of human STIM1 is in 11p15.5, a region related to pediatric malignancies, which makes STIM1 a considerable factor in mediating cell growth and transformation (10). STIM1 was first suggested as a tumor-suppressor gene to induce cell death (11). STIM1 mRNA is widely expressed in different human tissues, and strongly presented in lymphoid and myeloid cells (9). STIM1 knockdown accelerates the cell motility of melanoma cells and is defined as an antimetastasis gene (12). However, Yang et al. (13) demonstrated that STIM1 or Orai1 silencing inhibits the migration and metastasis of breast cancer cells by suppressing focal adhesion turnover. In contrast, another study showed that Orai1 could regulate mammary tumorigenesis by STIM1-independent pathways (14). Several works demonstrate that STIM1 could mediate cell proliferation. STIM1 knockdown decreases the protein levels of cyclin D1 in a hepatoma cell line (15) and STIM1-silencing inhibits cell proliferation in endothelial cells by arresting the cell cycle at the G2/M phase (16). However, the cell cycle stopped at the G1/S checkpoint after STIM1 depletion in rat vascular smooth-muscle cells (17).

Cervical cancer is strongly associated with infection by oncogenic types of human papillomavirus, but only a small fraction of those infected develop cancer, indicating that other factors contribute to the progression of the disease (18). Although intensive studies have been carried out, the tumor biology of this disease is largely unknown and little is known about the role of Ca²⁺ dysregulation involved in cervical malignancy. Using the model of cervical cancer, we studied the role of STIM1 in promoting cancer-malignant behaviors and how these processes are remodeled in cancer cells. The results indicate that STIM1 is important for Ca²⁺ signaling, which is necessary for cervical cancer cell growth, migration, and angiogenesis. We thus suggest STIM1-mediated Ca²⁺ signaling as a potential therapeutic target.

Results

Tumor Expression Level of STIM1 Is Associated with Clinical Outcome.

To clarify whether STIM1 expression has clinical significance in tumor progression, 24 cases of early-stage cervical cancer with pair tissues of carcinoma and adjacent nonneoplastic epithelia were analyzed by immunoblotting (Fig. 1A). Compared with that of nonneoplastic tissue, the level of STIM1 expression in tumor tissues was elevated in 71% cases (Fig. 1B). The immunofluorescent stainings confirmed that cervical cancer tissues clearly expressed STIM1 protein, whereas STIM1 protein was rarely detected in the adjacent normal or nonneoplastic cervical epithelia (Fig. 1C). More importantly, the level of STIM1 expression in tumor tissues was closely correlated with tumor size (Fig. 1D) (linear fit, R = 0.76, P < 0.001), an important indicator of human cervical carcinoma progression in vivo (19). In addition, the level of STIM1 expression in tumor tissues was significantly higher in the group of pelvic lymph-node metastasis (Fig. 1E), the primary cause of treatment failure and subsequent death in cervical cancer patients. We also recruited cases to check the potential redundancy between STIM proteins in tumor samples, but no trend of redundancy was found between STIM1 and STIM2 in primary tumor (Fig. S1 A and B). The 5-y overall survival rate of the patients recruited in this study (n = 35) was 72%. We found a poorer clinical outcome that was statistically significant in primary tumor with STIM1 up-regulation (P < 0.05) (Fig. S1C).

Fig. 1.

STIM1 expression is associated with cancer metastasis and clinical outcome. (A) Expression pattern of STIM1 in early-stage cervical cancer. Cervical cancer (n = 24) with the pair tissues of carcinoma and adjacent nonneoplastic epithelia were analyzed by immunoblotting. N, nonneoplastic epithelia. T, tumor tissues. (B) Quantitative analyses of STIM1 immunoblotting. STIM1 expression level in normal squamous epithelia was used as control and those in tumor tissues were expressed as the relative of control. (C) Immunofluorescent stainings of STIM1 and E-cadherin in normal cervical epithelia and adjacent cervical cancer tissues. E-cadherin, normal epithelial marker. Nuclei were stained with Hoechst 33258 (blue). Representative images of six different cases. (Scale bars, 20 μm.) (D) The association between STIM1 expression level and tumor size in the same surgical specimen of cervical cancer tissues (n = 24). (E) The tumor expression level of STIM1 was significantly higher in the groups of local pelvic lymph node metastasis. Dashed lines, mean ± SEM.

STIM1 Regulates Tumor Growth in Vivo.

To study the role of STIM1 in cancer behavior, we established the stable pools of cervical cancer cells with different levels of STIM1 expression (Fig. 2 A and B, and Figs. S2 A and B and S3A). Cervical cancer SiHa or CaSki cells with different levels of STIM1 expression were subcutaneously injected into the bilateral dorsal sites of SCID mice (Fig. 2 C and E). STIM1 overexpression remarkably enhanced tumor growth, local spread (Fig. 2C, arrow) and angiogenesis (Fig. 2 C and D), whereas shRNA-mediated knockdown of STIM1 significantly decreased tumor growth and tumor vessel numbers (Fig. 2E and Fig. S3 D–F). During tumor angiogenesis, tumor cells secrete VEGF, which is critical for the formation of new blood vessels (20). The quantitative analyses of VEGF-A by ELISA showed that VEGF-A production is correlated with the expression level of STIM1 (Fig. 2F). The results indicate that STIM1 contributes to tumor growth and angiogenesis in vivo.

Fig. 2.

STIM1 is involved in tumor growth and angiogenesis. (A) STIM1 knockdown by shRNA in cervical cancer SiHa and CaSki cell lines. (B) Establishment of STIM1-overexpressed cervical cancer cell lines. Endogenous STIM1 and exogenous STIM1 were differentiated by immunoblotting with STIM1 (Left) and EGFP (Right), respectively. (C and D) STIM1 overexpression enhances tumor angiogenesis and growth. Bilateral dorsal sites of SCID mice were subcutaneously inoculated with mock-transfected (Control) or STIM1-overexpressed cervical cancer cells. Representative tumor xenografts (C), the mean tumor vessel numbers and mean tumor weight (D) 21 d after inoculation. Arrowhead, local spread of tumor mass. (Scale bars, 0.5 cm.) Columns, mean ± SEM (n = 6), *P < 0.01. (E) STIM1 knockdown attenuates tumor growth and angiogenesis. Representative tumor xenografts (Left), mean tumor vessel numbers (Top Right), and mean tumor weight (Bottom Right) 15 d after inoculation of control shRNA- or shSTIM1-transfected cervical cancer cells. Columns, mean ± SEM (n = 6), *P < 0.01. (Scale bar, 1 cm.) (F) STIM1 regulates VEGF-A production. The VEGF-A secretion in various stable pools of cervical cancer cells were measured by ELISA. *P < 0.01, compared with wild type. Columns, mean ± SEM (n = 5).

STIM1 Is Involved in Cancer Cell Migration and Invasion.

We have previously shown that EGF is a potent stimulator for cervical cancer migration and invasion (21, 22). Accordingly, we studied whether STIM1 abundance is important for EGF-stimulated cancer cell migration. In the presence of a STIM1-specific siRNA, endogenous migrations of cervical cancer SiHa and CaSki cells were attenuated by 40% to 50% and the residual migrations were much less sensitive to EGF stimulation (Fig. 3A). In contrast, STIM1 overexpression enhanced cancer cell migration and invasion (Fig. 3B and Fig. S4A).

Fig. 3.

STIM1 modulates the focal-adhesion complex. (A and B) STIM1 is involved in cervical cancer cell migration. Columns, mean ± SEM from at least four different experiments. EGF, 100 ng/mL epidermal growth factor. (C) EGF-induced Ca²⁺ influx is a STIM1-dependent process. Gray lines, [Ca²⁺]_i oscillations of individual cervical cancer SiHa cells. Black lines, the mean trace of [Ca²⁺]_i oscillations. (D and E) STIM1 is necessary for EGF-stimulated calpain activation. (D) Quantitative fluorescent analyses of intracellular calpain activities, measured with a fluorogenic membrane-permeable calpain substrate t-Boc-LM-CMAC. PD151746 (50 μM), a specific inhibitor targeting Ca²⁺-binding site of calpain; PD145305 (50 μM), a negative control for PD151746. Columns, mean ± SEM from at least 100 cells. (E) STIM1 siRNA abolishes EGF-induced cleavage of α-spectrin. Arrow and arrowhead, the full-length (280 kDa) and calpain-digested cleaved form (150/145 kDa) of α-spectrin, respectively. (F) STIM1 affects EGF-stimulated Pyk2 activation. (Left) Representative immunoblots from at least three different experiments. (Right) Densitometric quantification of Pyk2 phosphorylation (Tyr402) levels. Points, mean ± SEM; *P < 0.01, compared with control group. (G) STIM1 modulates the focal adhesion turnover. (Left) Representative images showing focal adhesions. FAK (green), focal-adhesion marker. (Right) Quantitative analyses of focal adhesion size. Focal-adhesion sizes were quantified by measuring the area of FAK staining. Columns, mean ± SEM from at least 20 individual cells of three different experiments. N.S., nonsignificant. (Scale bars, 10 μm.)

We then studied whether STIM1 affects Ca²⁺ waves in response to EGF stimulation. When stimulated by EGF in physiological conditions, individual cervical cancer SiHa cells displayed a spectrum of intracellular Ca²⁺ ([Ca²⁺]_i) responses that included variable oscillations and failures to respond (Fig. 3C, Left). On average, 70% to ∼80% of cells responded to EGF, and ∼70% of these responses involved initial oscillations followed by the sustained elevations in baseline [Ca²⁺]_i. In nominally extracellular Ca²⁺ ([Ca²⁺]_o)-free conditions, the initial [Ca²⁺]_i oscillations did not sustain beyond 5 min after EGF stimulation (Fig. 3C, Center), suggesting that Ca²⁺ entry across plasma membranes is required to maintain oscillations. We studied whether STIM1 contributes to EGF-induced Ca²⁺ oscillations. EGF-induced sustained elevations of [Ca²⁺]_i were specifically reduced in STIM1 depletion (Fig. 3C, Right). These results indicate that EGF-induced Ca²⁺ influx is a STIM1-dependent process.

STIM1 Modulates Focal Adhesion Complex.

We examined the signal pathways in which STIM1 is involved in EGF-induced cancer cell migration. Cervical cancer cell migration is critically regulated by Ca²⁺-regulated protease calpain (Fig. S4B). EGF stimulated the calpain activity by 10-fold. PD151746, a specific inhibitor targeting the Ca²⁺-binding site of calpain, abolished EGF-induced calpain activation, whereas PD145305, a negative control for PD151746, showed no effect (Fig. 3D). STIM1 knockdown inhibited 70% to 80% of EGF-induced calpain activation. We also analyzed the calpain activation by monitoring the appearance of α-spectrin, a downstream target of calpain, which is a cytoskeletal scaffold protein that plays an important role in maintaining actin architecture (23). Cleavage of α-spectrin into 150/145-kDa breakdown products is a characteristic of calpain activation. By stimulating cervical cancer cells with EGF for 20 min, the calpain-specific 150/145-kDa breakdown products of α-spectrin appeared (Fig. 3E). Concomitantly, the full-length α-spectrin significantly decreased. The breakdown products of α-spectrin disappeared upon STIM1 depletion. These results indicate that STIM1 affects cell motility through altering calpain activity and spectrin processing.

Focal-adhesion kinase (FAK) and proline-rich tyrosine kinase 2 (Pyk2) are uniquely located in focal adhesions and regulate cell migration (24). Pyk2 was initially identified as a Ca²⁺-dependent tyrosine kinase but it is still unclear how intracellular Ca²⁺ levels would regulate Pyk2 phosphorylation. Ca²⁺ could lead to Pyk2 activation indirectly through increased levels of reactive oxygen species in T lymphocytes (25). We showed that Pyk2 is a Ca²⁺-regulated kinase in cervical cancer cells (Fig. S5A). The Tyr402 autophosphorylation and focal adhesion targeting of Pyk2 are required for Pyk2-mediated cytoskeletal reorganization and the subsequent determination of focal adhesion turnover and cell migration (24). We studied the correlation between EGF stimulation, Pyk2 phosphorylation, and STIM1 abundance. Incubation of cervical cancer SiHa cells with EGF induced Pyk2 phosphorylation at Tyr402 that was inhibited by STIM1 knockdown (Fig. 3F). The immunofluorescent images showed that pTyr402-Pyk2 was recruited to focal adhesions upon EGF stimulation (Fig. S6). In contrast, STIM1 knockdown abolished focal adhesion targeting of pTyr402-Pyk2, regardless of EGF stimulation.

We further studied the focal-adhesion dynamics by immunostaining with FAK, a major component of the focal-adhesion complex (Fig. 3G). In control cells, FAK staining showed a punctate pattern of focal adhesion and EGF induced small peripheral adhesions. In contrast, STIM1 knockdown cells displayed the quiescent cell morphology, with large focal adhesions around the cell periphery regardless of EGF stimulation. The effect of STIM1 siRNA on focal adhesion turnover was rescued by the subsequent transient expression of STIM1 cDNA in siRNA-treated cells (Fig. S7), suggesting that STIM1 is important for focal adhesion turnover. The immunoblotting also showed that EGF increased the levels of pTyr397-FAK, which was significantly inhibited by STIM1 knockdown (Fig. S5B).

EGF Enhances the Interaction Between STIM1 and Orai1.

It has been shown that STIM1 is essential for the regulation of Orai1, a pore component of the SOC channel (6). The immunoprecipitation and living cell images showed that the interaction between STIM1 and Orai1 was enhanced by EGF stimulation (Fig. 4 A–C). Although cervical cancer SiHa cells express Orai2, Orai3, TRPC1, and TRPC6 (Fig. 4A and Fig. S2C), STIM1 interacts with none of them with or without EGF stimulation (Fig. 4 A, D, and E).

Fig. 4.

EGF enhances the interaction between STIM1 and Orai1. (A) STIM1 specifically interacts with Orai1, but not with Orai2, Orai3, TRPC1, or TRPC6 in cervical cancer SiHa cells upon EGF stimulation. Representative immunoblots from at least three different experiments. IP, immunoprecipitation; WCL, whole-cell lysates. (B and C) Time-lapse confocal images of SiHa cells co-overexpressed with EGFP-STIM1 (green) and mOrange-Orai1 (red) in response to EGF stimulation. Arrows, the colocalization between EGFP-STIM1 and mOrange-Orai1 at the cell periphery. The quantitative results with pixel-by-pixel analyses were shown in C. Columns, mean ± SEM from at least 30 cells of three different experiments. (Scale bars, 10 μm.) (D and E) EGF does not enhance the interaction between STIM1 and TRPC1 or TRPC6. (Right) Representative images of STIM1 and TRPC family, such as TRPC6 (D) and TRPC1 (E) in the absence or presence of EGF stimulation. (Left) Colocalization ratio between STIM1 and TRPC6 (D) or TRPC1 (E) at cell periphery by pixel-by-pixel analyses. Columns, mean ± SEM from at least 30 cells of three different experiments. (Scale bar, 10 μm.)

STIM1 Influences Cell Cycle Progression.

Our clinical studies indicate that the tumor sizes of surgical specimens are significantly associated with STIM1 abundance (Fig. 1D). In the cervical cancer SiHa cell line, STIM1 knockdown significantly inhibited cell proliferation by 40 ± 3% at the third day posttransfection (Fig. 5A, and Fig. S3 A and B). The simultaneous FACS measurements revealed that STIM1 knockdown significantly increased the proportion of cells at S and G2/M phases of cell cycle (P < 0.01) (Fig. 5B and Fig. S8A). More importantly, an increase of p21 protein level and a decrease of Cdc25C protein level were noted in STIM1-silencing cells compared with control cells (Fig. 5C, and Figs. S3C, and S8 B and C). We also found that p21 and Cdc25C turnover depended on Ca²⁺ homeostasis in cervical cancer SiHa cells (Fig. S9 A and B). Several cell-cycle regulators were not affected by STIM1 knockdown, such as cyclin A, cyclin B1, and Cdk1 (Fig. S8B).

Fig. 5.

STIM1 influences cell-cycle progression. (A) STIM1 knockdown significantly inhibits the proliferation of cervical cancer SiHa cells. Points, mean ± SEM (n = 5); ^#P < 0.05, *P < 0.01, compared with control groups. (B) STIM1-silencing in cervical cancer cells inhibits cell proliferation by arresting cell cycle at S and G2/M phases. Representative FACS measurements to determine cell-cycle stages at the third day posttransfection. (C) STIM1 knockdown induces p21 up-regulation and Cdc25C down-regulation. Representative immunoblots from five different experiments. (D) STIM1 knockdown modestly increases mRNA levels of p21, demonstrated by semiquantitative RT-PCR with β-actin as the internal control. Columns, mean ± SEM (n = 4). (E) STIM1 knockdown slows down p21 protein degradation. The protein stability of p21 was examined in the presence of translational inhibitor cycloheximide (CHX, 20 μg/mL). (Upper) Representative immunoblots from four different experiments. (Lower) Densitometric quantification of p21 protein levels. Points, mean ± SEM (n = 4). (F) STIM1 knockdown inhibits the proteasome-mediated degradation of p21. Cells were preincubated with the protein translation inhibitor cycloheximide (20 μg/mL) for 30 min before the treatment of the proteasome inhibitor MG132 (40 μM), lysosome inhibitor NH₄Cl (40 μM), or protease inhibitor mixture (PIC, 1 unit) for 8 h. (Left) Representative immunoblots from three different experiments. (Right) Densitometric quantification of p21 protein levels. Columns, mean ± SEM (n = 3).

We studied how STIM1-silencing affects the p21 protein turnover. The mRNA levels of p21 was modestly increased by STIM1-silencing (P < 0.05) (Fig. 5D). In contrast, STIM1-silencing remarkably slowed down p21 protein degradation in the presence of translational inhibitor cycloheximide (P < 0.001) (Fig. 5E). We further studied the mechanism by which STIM1 affects p21 protein degradation. MG132, NH₄Cl and protease inhibitor mixture were used to specifically inhibit the proteasome-, lysosome-, and protease-mediated protein degradation, respectively. In control cells, the p21 protein abundance was increased by 110 ± 6% after the treatment of 26S-proteasome-specific inhibitor MG132 for 8 h (Fig. 5F). However, MG132 only increased p21 protein levels by 70 ± 6% (P < 0.01, compared with control cells) in STIM1 knockdown cervical cancer SiHa cells. Neither NH₄Cl nor a protease inhibitor mixture changed the p21 protein levels in control or STIM1 knockdown cells. We thus propose that the posttranslational regulation via the 26S-proteasome-dependent pathway contributes to the mechanisms controlling p21 up-regulation in STIM1 knockdown cancer cells. In the case of Cdc25C turnover, STIM1-silencing remarkably decreased the mRNA levels of Cdc25C by 60–70% (Fig. S9C). However, STIM1-silencing did not change the degradation rate of Cdc25C protein in the presence of translational inhibitor cycloheximide (Fig. S9D). This finding implies that the transcriptional regulation is the main mechanism controlling Cdc25C down-regulation in STIM1 knockdown cervical cancer cells.

SOC Channel Activity Affects Tumor Growth in Mouse Model.

Because SOCE is the immediate downstream target of STIM1 activation, we studied whether the manipulation of SOCE activity alters tumor growth in vivo. As shown in Fig. 6A, the SOCE activation in cervical cancer SiHa cells was specifically reduced in the presence of a STIM1-specific siRNA, 2-APB and SKF96365, indicating that 2-APB and SKF96365 are two potent inhibitors for SOCE. Rapid tumor growth with obvious angiogenesis was noted in SCID mice subcutaneously inoculated with cervical cancer SiHa cells (Fig. 6 B and C). In contrast, intraperitoneal injection of 2-APB or SKF96365 caused the obliteration of blood supply (Fig. 6C, black arrowhead), significantly diminished numbers of tumor vessels (Fig. 6D), and inhibited tumor growth (Fig. 6E). Interestingly, 2-APB and SKF96365 induced the extravasation of tumor feeding vessels (Fig. 6C, blue arrowhead), implying the damage of blood vessels. These results indicate that blockade of SOCE activity causes tumor growth regression.

Fig. 6.

Store-operated Ca²⁺ entry affects tumor growth in vivo. (A) STIM1 siRNA, 2-APB (20 μM), or SKF96365 (50 μM) inhibits SOCE of cervical cancer SiHa cells. To measure Ca²⁺ entry, SiHa cells loaded with Fura-2/AM (2 μM) were preincubated in Ca²⁺-free media plus 2 μM thapsigargin for 30 min to deplete the internal Ca²⁺ store. Each trace was averaged from at least 30 single cells. (B–E) Blockade of SOCE retards tumor growth and angiogenesis. (B) Female SCID mice bearing tumor xenograft of SiHa cells were intraperitoneally injected every 3 d (arrows) with control vehicle (n = 6), SKF96365 (2.5 mg/kg; n = 6), or 2-APB (50 μg/kg; n = 6) from the sixth day postinoculation. Representative tumor xenografts (C), mean tumor vessel numbers (D), and mean tumor weight (E) at the 15th day postinoculation. Arrows, rapid tumor growth with obvious angiogenesis. Black arrowheads, the obliteration of blood supply. Blue arrowheads, the extravasation of tumor feeding vessels. Columns, mean ± SEM (n = 6); *P < 0.01, compared with control group. (Scale bars, 1 cm.)

Discussion

This study highlights the novel role of Ca²⁺ store-sensor STIM1 in tumor malignant behavior. We show that STIM1 is important for cervical cancer cell proliferation, migration, and angiogenesis. This conclusion is supported by the following evidence. (i) STIM1-silencing in cervical cancer cells significantly inhibits cell proliferation by arresting the cell cycle at the S and G2/M phases. (ii) STIM1 overexpression enhances the invasive migration of cervical cancer cells, whereas STIM1 knockdown attenuates it. (iii) STIM1 expression regulates VEGF-A secretion from cancer cells. (iv) The animal model confirms that tumor abundance of STIM1 is linked with tumor growth, angiogenesis and local invasion. (v) The tumor expression level of STIM1 is closely associated with the clinical outcome of early-stage cervical cancer. To the best of our knowledge, this study is unique in showing the multiple functions of STIM1 in tumor biology, as evidenced by cell-line studies, animal models, and clinical sample analyses.

Our clinical studies indicate that more than 70% cases of early-stage cervical cancer display STIM1 overexpression. High abundance of tumor STIM1 indicates the high risk of metastasis. We studied how STIM1 overexpression leads to the enhancement of cancer cell migration and proposed that STIM1 affects cell motility through altering calpain activity and spectrin processing based on the following evidence. (i) Calpain activity is involved in cervical cancer cell migration. (ii) STIM1 affects calpain activation and spectrin processing. (iii) STIM1 is important for focal adhesion turnover. We also highlight the molecular identity of EGF-induced Ca²⁺ influx in cervical cancer cells. STIM1 is the molecular linker from endoplasmic reticulum Ca²⁺ store depletion to the plasma membrane SOCE (5, 26). STIM1 can interact with various plasma membrane Ca²⁺ channels, such as Orai proteins and the transient-receptor potential channel family, to form the functional pore-subunit of the SOC channel (27). The living cell imaging and immunoprecipitation demonstrate that EGF stimulates the aggregation and translocation of STIM1 toward the proximity of the plasma membrane, where STIM1 specifically interacts with Orai1 to mediate SOCE.

The findings of Yang et al. (13) show the emerging importance of STIM1 in tumor biology. Yang et al. proposed that blocking STIM1-mediating Ca²⁺ influx impairs focal adhesion turnover, which can be rescued by the small GTPases Ras and Rac. Our findings suggest that STIM1 is the key regulator of EGF-induced Ca²⁺ influx that is necessary for the activation of Ca²⁺-regulated protease calpain and tyrosine kinase Pyk2. The activation of calpain and Pyk2 regulates multiple signaling events crucial for the focal adhesion turnover and the locomotion of cervical cancer cells. The pathological significance of tumor STIM1 overexpression can be proposed to benefit the locomotion of cancer cells. In animal studies, we carefully dissected the tumor implants and unexpectedly found that, in addition to blocking tumor growth, the SOCE inhibitors, such as 2-APB and SKF96365, obliterated tumor feeding vessels. We thus suggest that both tumor cells and tumor vessels are possible targets of these drugs. In the study of Yang et al. (13), a single inhibitor (SKF96365) at a higher concentration was shown to inhibit lung metastasis. However, the information on the antiangiogenic effect of SKF96365 is not available in their work.

It has been reported that STIM1 mediates SOCE in endothelial and vascular smooth-muscle cells, and STIM1 knockdown elicits cell-cycle arrest (16, 17). Our results show that STIM1-silencing in cervical cancer cells leads to a decrease in proliferation by arresting the cell cycle at the S and G2/M phases. Several proteins alter G2/M checkpoint progression, such as p21, Cdc25C, and the Cdk1 (Cdc2)-cyclin B complex (28). Cdk1 can be dephosphorylated and activated by Cdc25C, leading to cell mitosis (29). In contrast, p21 inhibits the activity of Cdk1 and causes cell-cycle arrest (30). STIM1-silencing cervical cancer cells exhibited an increase of p21 protein levels and a decrease of Cdc25C protein. Compared with modest changes in p21 mRNA levels, the protein stability of p21 was remarkably increased in STIM1 knockdown cells, suggesting that STIM1 regulates p21 expression mainly through the posttranslational modification. STIM1 is also involved in angiogenesis through its effect on VEGF production from cancer cells. Further study needs to explore how STIM1 regulates VEGF production in detail. Taken together, the molecular mechanism by which STIM1 affects cancer malignant behaviors involves the regulation of Ca²⁺ signaling essential for cancer cell proliferation, migration, and tumor angiogenesis. These results make the STIM1-mediated signaling an attractive target for therapeutic intervention.

Materials and Methods

Cell Cultures, Transfection, RNA Interference.

Cultures of human cervical cancer cell lines (SiHa and CaSki) were prepared as described previously (21). Stable pools of cervical cancer cells overexpressing human STIM1 were used in this study. The full-length cDNA of EGFP-STIM1, mOrange-STIM1, and mOrange-Orai1 were subcloned into the eukaryotic expression vector pCMV6-XL5 (31). EGFP- or mOrange-STIM1 cDNA plasmids were transfected into cervical cancer SiHa cells by Lipofectamine 2000 (Invitrogen) and cells with STIM1 overexpression were selected by G418 (Sigma-Aldrich). FACSAria cell sorter (BD Biosciences) was used to isolate the stable pools of cells overexpressing human STIM1. For RNAi-mediated STIM1 knockdown, siRNAs or shRNAs were used (Table S1). Cells were introduced with siRNA or shRNA with Lipofetamine 2000 (Invitrogen) or by electroporation. The details of RNAi targeting human STIM1 and RNAi rescue are listed in SI Materials and Methods.

Surgical Specimens.

We collected the frozen tissues from the patients with early stage (International Federation of Gynecology and Obstetrics staging Ib) cervical cancer who were scheduled for radical hysterectomy and pelvic lymphadenectomy at National Cheng Kung University Hospital, Taiwan. The patients with local pelvic lymph node metastases did not exhibit distant metastases, such as paraaortic lymph nodes or distant organs. The collection of surgical specimens was approved by the institutional review board of National Cheng Kung University Hospital.

Migration and Invasion Assay.

In brief, cells were allowed to migrate across a membrane (8-μm pore) toward the medium containing 10 μg/mL fibronectin at 37 °C (21, 22). The invasive migration was done in the BD Matrigel invasion chamber (BD Biosciences) as an index of invasive activity of tumor cells (32).

VEGF-A Secretion Assay.

Briefly, 2 × 10⁶ cells were plated for 72 h and treated with 0.3 mM suramin for 3 h to release surface matrix-bound VEGF into culture medium. The VEGF secretion in culture supernatants were quantified using ELISA (R&D Systems) according to the manufacturer's instructions.

Immunoblotting and Immunoprecipitation.

The preparation of cell lysate and procedures of SDS/PAGE are described in detail in SI Materials and Methods. Bands in immunoblots were quantified using Vision WorksLS software (UVP).

Immunofluorescence, Confocal Microscopy, and Image Analyses.

The fluorophores were excited by laser at 405, 488, or 543 nm and detected by a scanning confocal microscope (FV-1000, Olympus). Cells with EGFP-STIM1 or mOrange-Orai1 overexpression were directly activated by laser at 488 or 543 nm, respectively (33). Cells were maintained in phenol red-free medium at 37 °C throughout the recording period. For the measurement of μ-calpain activity, an artificial μ-calpain fluorescent substrate t-butoxycarbonyl-Leu-Met-chloromethylaminocoumarin (t-Boc-LM-CMAC; Invitrogen) was used (33). The fluorescent intensity of t-Boc-LM-CMAC hydrolytic product was quantitatively analyzed as an index of μ-calpain activity of individual cell. A pixel-by-pixel analysis by FV-1000 software was used to assess the colocalization of STIM1with Oria1, TRPC1, or TRPC6 in confocal images. Focal-adhesion sizes were analyzed by dividing FAK staining pixels by the number of focal-adhesion patches.

Single Cell [Ca²⁺]_i Measurement.

[Ca²⁺]_i was measured at 37 °C with the Fura-2 fluorescence ratio method on a single-cell fluorimeter, as previously described (33). In brief, cells loaded with 2 μM Fura-2/acetoxymethyl ester (Fura-2/AM) were excited alternatively between 340 nm (I₃₄₀) and 380 nm (I₃₈₀) using the Polychrome IV monochromator (Till Photonics). The fluorescence intensity of excitation at 510 nm was monitored to calculate [Ca²⁺]_i using the TILLvisION 4.0 program (Till Photonics).

Proliferation Assay and Cell Cycle Analysis.

Cell proliferation was assessed by cell counting with Trypan blue exclusion (0.08%) to monitor viability. Cell-cycle stage was determined by FACS. Cellular DNA content was determined after propidium iodide staining. Cells were classified as in G0/G1, G2/M, and S phase based on the fluorescence intensity and the cell cycle distribution was analyzed by the Cell Fit software (BD Biosciences).

RT-PCR.

Total RNAs were isolated by RNeasy Mini Kit (Qiagen) and cDNA was prepared with SuperScript II (Invitrogen). The primers targeting human β-actin, GAPDH, p21, Cdc25C, and STIM1 are described in SI Materials and Methods.

Animal Models.

For the inoculation of xenograft, the bilateral dorsal sites of female BALB/c SCID mice aged 6 to 8 wk were subcutaneously injected with 5 × 10⁶ cervical cancer SiHa cells with different STIM1 expression level. The animal experiments were performed according to the ethical guidelines and approved by the institutional ethical committee of National Cheng Kung University.

Statistics.

All values were reported as mean ± SEM. Student's paired t test, unpaired t test, or linear regression was used for statistical analyses. Survival data were assessed by Kaplan–Meier methods and differences were compared by the log-rank statistic. Differences between values were considered significant when P < 0.05.