CACNA2D2 Promotes Tumorigenesis by Stimulating Cell Proliferation and Angiogenesis

ORIGINAL ARTICLE CACNA2D2 promotes tumorigenesis by stimulating cell proliferation and angiogenesis

M Warnier, M Roudbaraki, S Derouiche, P Delcourt, A Bokhobza, N Prevarskaya and P Mariot

In the present study, we have assessed whether a putative calcium channel α2δ2 auxiliary subunit (CACNA2D2 gene) could be involved in prostate cancer (PCA) progression. We therefore carried out experiments to determine whether this protein is expressed in PCA LNCaP cells and in PCA tissues, and whether its expression may be altered during cancer development. In addition, we evaluated the inﬂuence on cell proliferation of overexpressing or downregulating this subunit. In vitro experiments show that α2δ2 subunit overexpression is associated with increased cell proliferation, alterations of calcium homeostasis and the recruitment of a nuclear factor of activated T-cells pathway. Furthermore, we carried out in vivo experiments on immuno-deﬁcient nude mice in order to evaluate the tumorigenic potency of the α2δ2 subunit. We show that α2δ2-overexpressing PCA LNCaP cells are more tumorigenic than control LNCaP cells when injected into nude mice. In addition, gabapentin, a ligand of α2δ2, reduces tumor development in LNCaP xenografts. Finally, we show that the action of α2δ2 on tumor development occurs not only through a stimulation of proliferation, but also through a stimulation of angiogenesis, via an increased secretion of vascular endothelial growth factor in cells overexpressing α2δ2.

Oncogene (2015) 34, 5383–5394; doi:10.1038/onc.2014.467; published online 26 January 2015

INTRODUCTION coding for a putative auxiliary subunit of voltage-dependent 13 Although the implications of calcium signaling in genetic calcium channels (α2δ2 subunit). This subunit was identified as 18 cardiovascular, neurological or metabolic diseases have been promoting apoptosis, and was therefore thought to be a tumor 13 known for years,1 it has only recently been admitted to also be suppressor gene. We have, in the present study, addressed the involved in processes leading to cancer development,2 such as question of whether the α2δ2 subunit could be involved in PCA unregulated cell growth, resistance to apoptosis, enhanced progression. We show that α2δ2 is expressed in prostate epithelial angiogenesis and invasion. Indeed, it has been shown that cells and tissues and that its expression is enhanced during cancer α δ calcium transport mechanisms (through pumps, exchangers or development. We demonstrate that 2 2 regulates calcium channels) or calcium targets (calcium-binding proteins, calcium- homeostasis, nuclear factor of activated T-cells (NFAT) activity dependent kinases) may be subject to remodeling or altered and cell proliferation. Furthermore, we carried out xenograft fi expression in cancer cells.3 Voltage-gated calcium channels have experiments on immuno-de cient nude mice and we showed that α δ been shown either to be overexpressed, such as in colon cancer 2 2-overexpressing LNCaP PCA cells are more tumorigenic than α δ (Cav1.2 channels4) or downregulated (for instance, in lung cancer control LNCaP cells. Finally, we propose that the action of 2 2on (Cav3.1 channel5)) during cancer progression. tumor development probably occurs through a stimulation of Calcium channels are implicated in cancer development in both proliferation and angiogenesis. different tissues or organs, including prostate, breast, brain.6,7 Indeed, they have been demonstrated to participate in cell invasion, migration, differentiation or proliferation.8–11 We have RESULTS previously shown that voltage-dependent Cav3.2 T-type calcium α2δ2 expression in prostate tissues and cell lines channels are overexpressed in prostate cancer (PCA) cell lines, We have investigated whether α2δ (CACNA2Dx) proteins could be progressing toward a more aggressive phenotype.9 These expressed in human PCA cell lines and tissues. As shown in channels are expressed in human PCA acini and allow the Figure 1A, PCA LNCaP cells express the α2δ2 transcript. The other secretion of paracrine factors that may participate in cancer α2δ subunits investigated here (α2δ1 and α2δ3) are not expressed progression.12 Moreover, auxiliary voltage-dependent calcium in LNCaP cells. Other prostate cell lines express the α2δ2 transcript channels subunits are potentially involved in tumor growth. (DU145, Figure 1 and PC3; supplementary Figure S1A). Similarly, Indeed, several putative calcium channel subunit genes have been α2δ2 transcripts are present in both normal and cancerous human identified as being either down- or upregulated in some cancer prostate tissues (Figures 1B and C). Immuno-fluorescence (IF) tissues such as CACNA2D2 and CACNA2D3 in lung and gastric experiments showed that LNCaP cells express the α2δ2 subunit cancers,13–15 CACNB3 in recurrent non-small cell lung cancer16 or (Figure 1D). Immuno-staining for α2δ2 is particularly evident on involved in resistance of gliomas to chemotherapy (CACNG4).17 the cell periphery, close to plasma membrane area. Similar results A cluster of genes in locus 3p21 has been identified as a tumor were obtained with DU145 and PC3 cells (supplementary Figure S1B). suppressor gene cluster. In this cluster lies a CACNA2D2 gene In western blot experiments, we observed the presence of a band

Laboratoire de Physiologie Cellulaire, INSERM U1003, Villeneuve d'Ascq Cédex, France. Correspondence: Dr P Mariot, Laboratoire de Physiologie Cellulaire, INSERM U1003, Bâtiment SN3, Université Lille1, Villeneuve d'Ascq Cédex 59655, France. E-mail: [email protected] Received 9 July 2014; revised 5 November 2014; accepted 19 December 2014; published online 26 January 2015 CACNA2D2 stimulates tumor cell growth M Warnier et al 5384 at the expected size of 160 kD for α2δ2 (supplementary Figure S2). Immuno-histochemical (IH) staining and IF were carried out on The expression of this band was signiﬁcantly reduced when LNCaP parafﬁn-embedded and frozen (IF) sections of human prostate cells were treated with siRNA-targeting α2δ2 (si-α2δ2a and tissues, respectively. As illustrated in Figures 1E and F, there was si-α2δ2b, supplementary Figure S2), showing that the antibody a consistent staining of α2δ2 in epithelial glandular acini in both used here successfully detects the α2δ2 protein. non-cancerous and cancerous human prostate tissues. In addition,

Oncogene (2015) 5383 – 5394 © 2015 Macmillan Publishers Limited CACNA2D2 stimulates tumor cell growth M Warnier et al 5385 Figures 1E and F show that there was a stronger staining on the and α2δ2 subunits,19 on LNCaP cell proliferation. Figure 3C shows apical membranes of the epithelium than on basolateral that gabapentin (100 μM) reduced cell proliferation by about 35% membranes. after 48 h incubation. Gabapentin-induced inhibition then slightly In order to evaluate potential disorders in the expression of decreased, falling to 20% after 4 days, which may be due to α2δ2 between normal and tumoral tissues, we carried out a gabapentin desensitization. Another α2δ2 ligand (pregabalin; 20 comparison of the α2δ2 transcript expression using an RT–PCR 100 μM ) also slightly but significantly reduced LNCaP cell approach in 18 PCA tissues and 20 non-cancerous prostate tissues. proliferation (10.4 ± 1.07% inhibition at 48 h, supplementary As illustrated on Figures 1B and C, α2δ2 is more frequently Figure S3C). In addition, two different α2δ2-overexpressing LNCaP expressed (significantly different, Po0.001, Χ2) in cancer tissues clones expressing similar levels of α2δ2 subunits (clones C9 and (95% of cancer tissues expressed α2δ2) than in non-cancerous C11, supplementary Figure S2) displayed faster rates of prolifera- tissues (only 40% of non-cancerous prostate tissues expressed tion than did control LNCaP cells (Figure 3D), with a doubling time α2δ2). of 47.5 h for LNCaP against 40.5 h for clone 9 and 36 h for clone 11. Similarly, we compared the expression of α2δ2 protein on We then carried out IF experiments using Ki-67 staining (Figure 3E) prostate tissue microarray (TMA) in 24 different tissues (16 cancers on LNCaP and LNCaP-α2δ2 cells (clone C11). Ki-67 is a marker of and 8 non-cancerous tissues). As shown in Figure 2, epithelial cells proliferating cells. It is only expressed in phases G1, S, G2 and M, lining the acini were stained by the α2δ2 antibody in normal, whereas cells in the G0 phase do not express significant levels of benign prostatic hyperplasia and cancer samples. In advanced Ki-67.21 It can either exhibit punctuated nuclear localization in the cancers, glandular structure was, as expected, progressively lost G1 and S phases or a more homogeneous localization in the and stained cells were disseminated throughout the cancer tissue nucleus during the G2 and M phases. We demonstrate here that (see Figure 2B, Gleason Scores 4+5 and metastasis). Both stained α2δ2 is associated with a decrease in the number of cells in the surface and 3,3'-diaminobenzidine density increased in advanced G0 phase of the cell cycle. In our experiments, non-proliferating grades of PCA as compared with normal or hyperplasic prostate. cells represent 35.6 ± 1.6% of the whole-LNCaP cell population, Cellular localization differs between normal, benign prostatic whereas they represent only 21.6 ± 3.5% of the whole-LNCaP-α2δ2 hyperplasia, early stages of cancers and advanced cancers. In cell population. This correlates with an increase in the number normal, benign prostatic hyperplasia tissues and early stages of of cells in the G1/S and G2/M phases of the cell cycle in LNCaP- cancer (well-differentiated cancers: Gleason Scores o6), staining α2δ2 cells. Altogether, these results show that α2δ2 enhances is mostly apical and intracellular, whereas in advanced stages of prostate cell proliferation. cancer (from intermediately differentiated cancers—Gleason – — Scores 6 7(3+4) to poorly differentiated cancers, Gleason Scores Role of α2δ2 in calcium homeostasis from 7(4+3) and higher), staining increases in every cellular As α2δ2 has been shown to modulate voltage-dependent calcium compartment and remarkably on cell nuclei (Figure 2C). We also channel activity and plasma membrane expression,22 we studied analyzed staining density using color deconvolution to extract whether α2δ2 could regulate calcium homeostasis in LNCaP cells. 3,3'-diaminobenzidine staining from counterstaining (Azure blue). Though not decreased by si-α2δ2 (Figure 4a), the cytosolic-free We show that 3,3'-diaminobenzidine density, and thus α2δ2 calcium concentration was increased by α2δ2 overexpression, in expression, was significantly enhanced in PCA tissues as compared either 2 or 10 mM extracellular calcium concentrations (Figure 4b). with non-cancerous tissues (Figure 2C). Furthermore, cytosolic Furthermore, to assay whether calcium homeostasis, and internal staining was significantly increased between grades 2 and 3, calcium store capacity, was altered in LNCaP-α2δ2 cells, we showing a correlation between α2δ2 expression and cancer 2+ challenged the cells with ionomycine (1 μM), a Ca ionophore able progression (Figure 2Cb). to induce calcium mobilization from endoplasmic reticulum 23 2+ stores and thapsigargin (0.5 μM), an endoplasmic reticulum Ca α2δ2 regulates cell proliferation ATPase inhibitor.24 Our results show that both ionomycine and Different approaches were used to evaluate the role of α2δ2in thapsigargin led to slightly stronger calcium releases in LNCaP- prostate cell proliferation (Figure 3). First, LNCaP cell proliferation α2δ2 cells than in LNCaP-ctl cells (Figures 4c and d). On the was reduced by siRNA-targeting α2δ2 (50 nM si-α2δ2a or si-α2δ2b) contrary, si-α2δ2 lessened the calcium release induced by either as measured either by MTS assay (Figure 3A) or manual counting thapsigargin or ionomycine (Figure 4d). This demonstrates that (supplementary Figure S3A). Kinetic experiments (Figure 3A) α2δ2 regulates calcium homeostasis in LNCaP cells. indeed illustrate that both si-α2δ2 had a significant inhibitory We thus investigated whether the NFAT-calcineurin pathway effect on cell proliferation only after 3–4 days’ incubation and could be involved in α2δ2 signaling. First, we show that NFAT reached their maximal action after 5–7 days (inducing a 40–80% activity was greater in LNCaP-α2δ2 than in LNCaP-ctl cells reduction in cell proliferation). Treatment of DU145 and PC3 cells with (Figure 4e). To evaluate the role of the calcineurin/NFAT pathway si-α2δ2 similarly reduced proliferation (supplementary Figure S3B). in cell proliferation, we used two different calcineurin inhibitors Fluorescence-activated cell sorting analysis of cell cycle shows that (cyclosporin A and FK506), which both reduced LNCaP cell si-α2δ2 increased the number of LNCaP cells in G1 phase and proliferation. Reduction of cell proliferation by these inhibitors of decreased the proportion of cells in G2/M and S phases calcineurin was more pronounced in LNCaP-α2δ2 cells than in (Figure 3B). We therefore tested gabapentin, a ligand of α2δ1 LNCaP-ctl cells (Figure 4f).

Figure 1. Expression of α2δ2 in prostate cells and tissues. (A) Prostate cancer LNCaP and DU145 cell lines were screened by RT–PCR for the different members of the α2δ family. Only the α2δ2 transcript is detected in LNCaP and DU145 cells at the expected size of 505 bp. Human brain is used as a positive control. (B) Human prostate tissues were screened for the expression of α2δ2. Among the prostate samples tested (n = 20 for non-cancer tissues, n = 18 for cancer tissues), α2δ2 is more frequently expressed in cancer tissues (***Po0.001, Χ2-test). (C) Examples of RT–PCR performed on non-cancer and cancer tissues (numbered from 1 to 10: N1 to N10 for normal and C1 to C10 for cancer). (D) Confocal immuno-fluorescence experiments showing the expression of α2δ2 in LNCaP cells (×20 objective). Staining is particularly pronounced at the cell periphery. (E) Confocal immuno-fluorescence experiment showing the expression of α2δ2 in prostate hyperplasic tissues (a and b: different magnifications of the same field). (F) Confocal immuno-fluorescence experiments showing the expression of α2δ2in prostate cancer tissues. Panel a shows a cancer (grade 3) with small acini still observable. Panel b is an enlargment of the acinus labeled in panel a. In panel c is shown a cancer (grade 4) where acini are fused and lumen is not observable. Bar size: 40 μm. Lum: lumen of the acini. Arrow: apical side of the epithelium. Diamond headed arrow: basal side.

Figure 2. Immuno-histochemical study of α2δ2 expression in normal and diseased prostate. (A and B) Overview of prostate TMA showing different normal, hyperplasic and cancer tissues. (A) × 5 objective, (B) × 40 objective). Gleason Scores are indicated on each picture. M: rectal metastasis, BPH: hyperplasia, N: normal prostate tissue. Immuno-histochemical staining against α2δ2 was revealed using peroxidase-DAB reaction. (C) Color images were deconvoluted, and intensity of cell staining was analyzed and converted into optical density (OD). (a) As illustrated, OD is enhanced in cancer tissues in acini surrounding cells. (b) Histogram showing the average OD for different Gleason grades in cytosol and in nucleus. The number of cells analyzed for each bar was comprised between 143 and 882. OD was compared between normal and other conditions (*Po0.05, **Po0.01, ***Po0.001) and between grade 2 and other conditions (#Po0.05, ##Po0.01, ###Po0.001). Comparison was assessed with Student–Newman–Keuls tests.

Figure 3. α2δ2 promotes cell proliferation in prostate cancer LNCaP cell lines. (A-a) A treatment with siRNA (50 nM)-targeting α2δ2 (si-α2δ2a) inhibits LNCaP cell growth as measured with MTS assay. Kinetic experiments show that inhibition by si-α2δ2a reaches its peak (40%) 6 days after the onset of treatment. (b) LNCaP cell proliferation was signiﬁcantly reduced by two different siRNA-targeting α2δ2(si-α2δ2a and si-α2δ2b, 50 nM) 6 days after treatment. (B) FACS analysis shows that treatments with si-α2δ2a increased the proportion of cells in G0/G1 phase, whereas decreased the proportion of cells in the S and G2/M phases. (C) Gabapentin (100 μM), a ligand of α2δ2 reduces cell proliferation as measured by MTS assay. (D) Kinetics of cell proliferation in two different clones of LNCaP cells overexpressing α2δ2 (LNCaP-α2δ2-C9 and LNCaP-α2δ2-C11). (E-a) Bar representation of the proportion of LNCaP cells in G0, G1/S or G2/M phase as measured by Ki-67 staining in LNCaP- ctl and LNCaP-α2δ2-C11 cells. (b) Example of Ki-67 immuno-staining. The overexpression of α2δ2 reduces the number of cells in the G0 phase, whereas increases the number of cells in the G1/S phases. (*Po0.05, **Po0.01, ***Po0.001).

Overexpression of α2δ2 promotes in vivo tumor development, increase after 4–7 weeks (Figure 5b). The latent period was a blood vessel formation and VEGF secretion little bit longer in LNCaP-α2δ2 tumors (49.5 ± 3, 34.5 ± 3.5, Immunodeficient nude mice were injected on both flanks with 37.8 ± 2.8 days for LNCaP-α2δ2, LNCaP-ctl and LNCaP-Neo, either LNCaP-ctl cells, LNCaP-Neo cells or with LNCaP-α2δ2 cells respectively, Figure 5c). Once the tumors were detectable, they (Figure 5). Though we did not compare the level of expression of grew with a doubling time of 12.2 ± 1.9 days in LNCaP tumors and α2δ2 between human prostate tumors and LNCaP-α2δ2 clones, 10.7 ± 1.2 days for LNCaP-Neo tumors to reach an average size of this last model was particularly useful to determine the role of 940 ± 146 mm3 (1.26 ± 0.22 g). For their part, LNCaP-α2δ2 tumors α2δ2 in proliferation and tumor growth. Indeed, the increased grew with a doubling time of 8.3 ± 0.8 days to reach a size expression of α2δ2 in LNCaP-α2δ2 clones may qualitatively reflect of 1760 ± 154 mm3 (2.3 ± 0.21 g) 11 weeks after cell injection the increased expression of α2δ2 occurring in advanced PCA. (Figure 5d). In different tumors, maximum tumor size was Tumor size, which was measured twice a week, usually began to usually reached 4 weeks after the onset of tumor growth.

Figure 4. Calcium imaging studies show that α2δ2 is involved in calcium homeostasis. (a) Downregulating α2δ2 expression in LNCaP cells (using siRNA-targeting α2δ2 (si-α2δ2a and si-α2δ2b, 50 nM)) did not change the basal cytosolic calcium concentration in both 2 and 10 mM external calcium concentration conditions. (b) Overexpressing α2δ2 in LNCaP cells (LNCaP-α2δ2) slightly increased the basal cytosolic calcium concentration in both 2 and 10 mM external calcium concentration conditions. (c) Application of 0.5 μM Thaspigargin (TG) induced a calcium release, as shown by the increase in the F340/F380 ratio, of lower amplitude in LNCaP-ctl cells as compared with LNCaP-α2δ2 cells. (d) Average peak amplitude of calcium release induced by thapsigargin (TG, 0.5 μM) and ionomycine (iono 1 μM) in LNCaP-ctl, LNCaP-α2δ2 cells and in LNCaP cells treated with si-α2δ2a, si-α2δ2b or si-ctl. (e) Measurement of NFAT activity in LNCaP cells shows that the overexpression of α2δ2is associated with a signiﬁcant stimulation of NFAT activity. (f) Inhibitors of calcineurin (cyclosporin A (CsA 1 and 5 μM) and FK506 (10 μM)) reduced cell growth stimulated by the overexpression of α2δ2. (*Po0.05, **Po0.01, ***Po0.001).

Altogether, LNCaP-α2δ2 tumors grew faster and reached a larger In addition, histological slides show that there are more blood size than LNCaP-ctl and LNCaP-Neo tumors. We also assessed the vessels in LNCaP-α2δ2 tumors than in LNCaP-ctl tumors (Figures effect of gabapentin on nude mice bearing LNCaP-ctl tumors. As 6Aa and Ab). We thus stained blood vessels with CD31 (Figures 7A seen in Figure 5e, gabapentin consumption (via drinking water) and B), which is a common marker for endothelial cells and used from the onset of the experiment considerably slowed tumoral to detect angiogenesis in tumors. We show that there is a twofold development. increase in the surface stained by CD31 in LNCaP-α2δ2 tumors as Both LNCaP-ctl and LNCaP-α2δ2 tumors (Figures 5a and 6) were compared with LNCaP-ctl tumors (Figure 7C). This was confirmed abundantly irrigated and characterized by signs of hemorrhages using western blots experiments, which demonstrate that there is with leakage of red blood cells in surrounding tissues and a 50% increase in CD31 expression in LNCaP-α2δ2 tumors numerous foci of inflammation and necrosis (Figures 6A and B). (Figure 7D). We did not detect any significant differences in necrosis between As α2δ2 is associated with blood vessel formation, we measured LNCaP-ctl and LNCaP-α2δ2 tumors. On the contrary, proliferating vascular endothelial growth factor (VEGF) expression and secre- cell nuclear antigen IF studies, which allow the identification of tion in prostate cells and tumors. As shown in Figure 7D, western cells in the S phase of the cell cycle, show that there are more blot experiments demonstrate that VEGF expression was slightly proliferating cells in LNCaP-α2δ2 tumors than in LNCaP-ctl tumors but significantly increased in tumors overexpressing α2δ2. In (Figure 6C). addition, as shown in Figure 7E, enzyme-linked immunosorbant

Figure 5. α2δ2 potentiates in vivo tumor growth. (a) Mice were injected in both flanks with LNCaP, LNCaP-Neo or LNCaP-α2δ2 cells. After killing of the animals, tumors were extracted from the animals, weighed and prepared for RNA or protein extraction or immuno-histochemistry. (b) Kinetics of tumor growth in both control and α2δ2 tumors. (c) Bar plot of the latency period before the development of detectable tumors (4100 mm3) in LNCaP-ctl, LNCaP-Neo and LNCaP-α2δ2 tumors. (d) Average doubling time was lower in LNCaP-α2δ2 tumors than in LNCaP- control and LNCaP-Neo tumors. (e) Administration of Gabapentin via drinking water (400 mg/l) from the onset and throughout the experiment reduced the average kinetic of tumor development in animals injected with LNCaP-ctl cells. (*Po0.05, **Po0.01, ***Po0.001). assay tests show that overexpression of α2δ2 significantly itself is not hypermethylated. In a recent study, it was even shown enhanced in vitro VEGF secretion in two different clones of that CACNA2D2 is overexpressed in breast cancer cell lines LNCaP-α2δ2 cells. compared with non-tumorigenic cells.26 As we show that it may be overexpressed in a variety of cancer cells (prostate, breast, Implication of α2δ2 presence in other tissues pancreas, lung, colon), CACNA2D2 may possess oncogenic proper- α δ In order to assess the potential relevance of α2δ2 in tumorigenesis ties. Interestingly, another protein of the same family, the 2 1 in other tissues, we compared the expression of α2δ2 on TMA in subunit (CACNA2D1) has recently been shown to be expressed in α δ liver cancer stem cells and that RNAi knockdown of α2δ1 other cancer tissues (lung, breast and colon). We show that 2 2is 27 consistently expressed in lung, breast and colon cancer tissues expression reduces tumorigenicity in cancerous liver cells, (two cases for each cancer) and that stained surface is greater in suggesting that different proteins of the CACNA2D family may cancer vs normal tissues (Figure 8a). In addition, we carried out have oncogenic properties. RT–PCR in various pancreatic cell lines. Our results show that α2δ2 The CACNA2D2 gene may therefore, according to the tissues or is expressed in pancreatic tumorigenic carcinoma cell lines circumstances, behave as a tumor suppressor gene or an (ASPC1, BxPC3, Capan1, MiaPaca2, Panc1), but not in the human oncogene. Such a property is known for various genes, such as non-tumorigenic immortalized pancreatic ductal H6c7 epithelial REST (repressor element-1 silencing transcription factor), which cell line (Figure 8b). acts as an oncogene in neural cells or a tumor suppressor gene in breast or lung cells,28 or other transcription factors such as WT1, KLF4, SnoN and Runx (for review see reference 28). The RASFF1 DISCUSSION gene has also been reported to display both oncogenic and tumor We demonstrate for the first time in this study that α2δ2, a suppressive properties in lung neuroendocrine tumors,29 depend- putative calcium channel accessory subunit,25 is able to promote ing on the isoform expressed. Interestingly, the RASFF1 gene is cell proliferation in vitro and tumorigenesis in vivo.Wefirst show located in the same 3p21.3 locus as CACNA2D2. that α2δ2 is expressed in PCA cell lines and tissues and that α2δ2 In vitro experiments were carried out using various proliferation is more frequently expressed in cancer tissues than in non- assays, and all the methods used here demonstrate that α2δ2 cancerous prostate tissues obtained from surgical human samples. activates proliferation of PCA cells, decreasing the number of cells We demonstrate that α2δ2 is expressed mainly in the prostate in the Go and G1 phases and increasing the proportion of cells in epithelium and that its expression is higher in cancer tissues than the S and M phases. This was shown using overexpression studies, in normal tissues. This result was unexpected because other siRNA strategy and ligand-inhibition of α2δ2. Furthermore, we publications had previously shown that the CACNA2D2 gene, demonstrate that α2δ2 overexpression is associated with tumor coding for the α2δ2 protein, is located in a tumor suppressor gene growth, VEGF secretion in vitro and blood vessel formation in vivo. cluster13 (the 3p21.3 chromosomal region). This region may be We thus suggest that α2δ2, through a stimulation of VEGF deleted or hypermethylated in various cancers, such as lung or secretion, promotes angiogenesis. Therefore, both proliferative pancreatic cancers.14 In addition, CACNA2D2 promotes apoptosis and angiogenic actions of α2δ2 may be responsible for α2δ2- in non-small cell lung cancer cells18 and reduces tumor growth in induced tumor development in mice xenografts. In addition, we nude mice. However, some studies have shown that CACNA2D2 show that gabapentin, a known α2δ2 ligand, reduced tumor size

Figure 6. Histology of LNCaP tumors. (A) Low magnification views (×5 objective) of a control LNCaP tumor (a) and an LNCaP-α2δ2 tumor (b). (B) Different views of a control LNCaP tumor (a) and an LNCaP-α2δ2 tumor (c) at higher magnification (×20). Numerous foci of necrosis and inflammation or hemorrhages could be observed in tumors ((b) LNCaP and (d) LNCaP-α2δ2). (C) PCNA immuno-fluorescence (a) and quantification of the percentage of PCNA-positive nuclei normalized to the total number of nuclei (DAPI) (b). Red arrows: examples of hemorrhages and leakage of red blood cells. Blue arrows: inflammatory cells. Black arrows: necrosis sites. (*Po0.05).

in mice xenografts, conﬁrming that α2δ2 may be involved in enhances cell growth, whereas its downregulation leads to tumor development. reduced proliferation.32 We show here that the α2δ2 expression It has been proposed that α2δ2 is a voltage-dependent calcium regulates calcium homeostasis. As Cav3.2 and α2δ2 exert similar channel accessory subunit and could regulate both targeting and actions on both calcium signaling and cell growth, they may be activity of α1 pore subunits.30,31 As PCA cells express functional involved in a common pathway, leading to increases in cell Cav3.2 T-type calcium channels,9 α2δ2 involvement in tumorigen- proliferation, through a possible interaction. However, various esis could rely on its interaction with Cav3.2 channels. Further- interacting domains, including a large C-terminal sequence that more, we have shown previously that Cav3.2 overexpression protrudes into the extracellular environment, have been

Figure 7. α2δ2 promotes blood vessel formation in xenografted LNCaP tumors through VEGF secretion. (A) Immuno-ﬂuorescent identiﬁcation of CD31 (a): CD31 (green), (b): DAPI (blue), (c): overlay of DAPI and CD31. (B) Immuno-histochemical (peroxydase/DAB) staining of CD31 (top panel: LNCaP tumor, bottom panel: LNCaP-α2δ2 tumor. (C) CD31-positive area in LNCaP and LNCaP-α2δ2 tumors. (D-d) Examples of western blots of CD31 and VEGF in eight different tumors (LNCaP tumors: T-LNCaP, LNCaP-α2δ2 tumors: T-α2δ2) from four different mice (Mx). (b) Relative density of CD31 bands (normalized to β-actin) observed in western blots (mean ± s.e.m. from 10 tumors). (c) Relative density of VEGF bands (normalized to β-actin) observed in western blots (mean ± s.e.m. from 10 tumors). (e) VEGF secretion measured using ELISA assay in LNCaP cells and LNCaP cells overexpressing α2δ2 (clones C9 and C11). Overexpression of α2δ2 was associated with an increased secretion of VEGF.

© 2015 Macmillan Publishers Limited Oncogene (2015) 5383 – 5394 CACNA2D2 stimulates tumor cell growth M Warnier et al 5392 MATERIALS AND METHODS Cell culture and human tissues LNCaP, DU145 and PC3 PCA cell lines were purchased from the American Type Culture Collection and cultured in a humidified atmosphere at 37 °C 12 (95% air—5% CO2) as described by Gackière et al. in RPMI 1640 (GIBCO, Life Technologies, St Aubin, France), supplemented with 10% fetal bovine serum (Sigma, St Quentin Fallavier, France), and 2 mM L-glutamine (Sigma). LNCaP cells stably expressing CACNA2D2 (LNCaP-α2δ2) or transfected with an empty pcDNA3 plasmid (LNCaP-Neo) were generated as described in supplementary methods. Pancreatic cell lines (H6C7, ASPC1, BxPC3, Capan1, MiaPaca2, Panc1) used to detect the expression of CACNA2D2 were obtained and cultured as described by Kondratska et al.38 Paraffin-embedded TMAs (from prostate (Figure 2), lung, colon and breast (Figure 8)) were obtained from US Biomax (Rockville, MD, USA). Human brain cDNA were purchased from Amsbio (Abington, UK). Prostate tissue samples (prostate adenocarcinoma and hyperplasia) were obtained from consenting patients, following the local ethical considerations and as described in supplementary methods. Unless specified, all products were purchased from Sigma.

Cell transfection Two small RNAs (Eurogentec, Angers, France) interfering with the human coding sequence of CACNA2D2 (Genebank accession # NM_001174051) were designed and referred to as si-α2δ2a (5′-GACCAACGUUCUGAUCUGC (dTdT)-3′) and si-α2δ2b (5'-AACAAGGUCAACUAUUCAUAC(dTdT)-3'). The siRNAs used in this study included a nonspeciﬁc control siRNA against Luciferase (5′-CUUACGCUGAGUACUUCGA (dTdT)-3′accession # FN554878). Cells were transfected with 50 nM siRNA using HiPerFect Transfection Reagent (Qiagen, Courtaboeuf, France) as described previously.12 The efﬁciency of the siRNA was validated by western blot analysis (supplementary Figure S2).

Analysis of the α2δ2 subunit gene expression by RT–PCR RNA extraction was carried out using Trizol according to the method α δ 39 Figure 8. Expression of 2 2 in non-prostatic cancer tissues (breast, originally described by Chomczynski and Sacchi. RT–PCR was carried a α δ lung and colon). ( ) Immuno-histochemical staining against 2 2 out as previously described.12 The PCR primers in this study were: was revealed using peroxidase-DAB reaction in breast, lung and 5′-TGCAAGAAGACCTTGTCACA-3′ (forward) and 5′-ACACAATTGTTGAG colon cancer and normal tissues (×5 objective). Grade and stage are CCCTCA-3′ (reverse) for the 400 bp CACNA2D1 amplicon (accession #: GI b – indicated on the right corner of each tissue microarray. ( )RT PCR 179761), 5′-GATGCGGAGCTAGAGGATGA-3′ (forward) and 5′-GCTGTACGTG carried out in six pancreatic cell lines, a human immortalized, non- ′ fi TTGAGATGCT-3 (reverse) for the 505 bp CACNA2D2 amplicon (accession #: tumorigenic pancreatic ductal epithelial cell line, H6c7 and ve BC 152438), 5′-CCATGGAGGTGAAGAAGACA -3′ (forward) and 5′-TTGT tumorigenic carcinoma cell lines ASPC1, BxPC3, Capan1, MiaPaca2, ′ α δ CACCAGCTTTCAGGAA-3 (reverse) for the 500/439 bp CACNA2D3 amplicon Panc1. 2 2 transcript was present only in carcinoma cell lines. (accession #: BC NM_018398), and 5′-CAGAGCAAGAGAGGCATCCT-3′ (forward) and 5′-GTTGAAGGTCTCAAACATGAT-3′ (reverse) for the 220 bp demonstrated in α2δ2, which would suggest that α2δ2 may β-actin amplicon (accession #:: NM_001101). interact with proteins other than calcium channels. For example, 33 the Von Willebrand A domain, in the extracellular portion of Western blotting α δ 2 2, is also present in proteins such as integrins involved in the After washing in phosphate-buffered saline, cells were collected in a lysis buffer interaction with the extracellular matrix.34 The adhesion of LNCaP (Triton X-100 1%, Na deoxycholate, 1%, NaCl 150 mM,PO4NaK 10 mM,pH7.2), cells on the extracellular matrix could thus be affected by the with an anti-protease cocktail and incubated on ice for 45 min. The lysates were overexpression of α2δ2 and this could delay tumor onset as centrifuged at 12 000 g for 10 min at 4 °C. The protein concentration of the observed in our experiments. This extracellular domain may also supernatant was determined by the BCA assay (Pierce Chemical Biology, serve as a receptor to extracellular ligands such as thrombos- Courtaboeuf, France). Western-blot analysis was performed as described pondin 1.35,36 Thrombospondin have themselves been shown to elsewhere12 using the following primary antibodies: anti-CACNA2D2 (130 kDa, control angiogenesis through their interaction with their numer- Santa Cruz sc-66822, Heidelberg, Germany; 1/200 in WB, 1/100 in IF/IH), anti- ous receptors.37Although thrombospondin 1 may inhibit angio- CACNA2D2 (Abnova H00009254-M12, Walnut, CA, USA; 1/50 in IF/IH), anti- genesis through their interaction with CD36 or VLDL-R membrane proliferating cell nuclear antigen (36 kDa, Santa Cruz sc-56, 1/200 in WB, 1/100 in IF), anti-Ki-67 (Santa Cruz sc101861, 1/100 in IF/IH), anti-CD31 (130 kDa, Abcam, receptors, they can also promote angiogenesis after binding with α β 37 α δ Paris, France; ab28364, 1/200 in WB, 1/50 in IF/IH), anti-VEGF (43 kDa, Abcam 3 1 integrins. It is therefore conceivable that 2 2 promotes ab46154, 1/200 in WB, 1/50 in IF/IH), anti-β-actin (43 kDa, Sigma A5441, 1/4000 in cell proliferation or angiogenesis through a signaling pathway, WB), anti-calnexin (95 kDa, Millipore MAB3126, Guyancourt, France; 1/2000 in which does not involve voltage-dependent calcium channels. We WB). Each experiment was repeated at least three times. will thus go on to investigate whether α2δ2 subunits could regulate angiogenesis in response to thrombospondin 1 binding. Immuno-staining and confocal analysis Altogether, our results suggest that α2δ2 is a potentially relevant marker of PCA progression and could therefore be The protein expression studies of PCA cells and tissues were carried out using indirect IF analysis on acetone-fixed cells or frozen section of evaluated as relevant target in diagnostic assays. In addition, our prostate and immuno-histochemistry (IH) on formalin-fixed paraffin- experiments show that molecular or pharmacological inhibition of embedded tissues (TMA). TMA (US Biomax) and mice tumoral tissues α2δ2 leads to a reduction of tumor growth, suggesting its were studied using IH procedures as previously described.40 Frozen relevance as a basis for the development of alternative PCA sections of human prostate were prepared for IF experiments as described treatments. previously.12

Oncogene (2015) 5383 – 5394 © 2015 Macmillan Publishers Limited CACNA2D2 stimulates tumor cell growth M Warnier et al 5393 Calcium imaging multiple comparisons), and Χ2-tests to compare proportions. Differences Calcium imaging was carried out in Hank's Balanced Salt Solution were considered signiﬁcant with *Po0.05; **Po0.01; ***Po0.001. containing 142 mM NaCl, 5.6 mM KCl, 1 mM MgCl2, 2 or 10 mM CaCl2, 10 mM HEPES and 5.6 mM glucose. The osmolarity and pH of external Declaration of approval for animal studies solutions were adjusted to 310 mOsm l − 1 and 7.4, respectively. Cytosolic In vivo experiments were conducted on mice according to the agreement Ca2+ concentration was measured and analyzed using Fura2-loaded cells 41 provided by the local ethical comity (protocol CEEA 202012). (2 μM) as described elsewhere.

NFAT assay CONFLICT OF INTEREST Cells were seeded in 60-mm dishes and were transfected with a NFAT- The authors declare no conflict of interest. luciferase plasmid or with luciferase only (1 μg of plasmid per dish). Cells were lysed and proteins extracted using the Luciferase Assay System Kit (Promega, Madison, WI, USA). NFAT-luciferase activity was then measured ACKNOWLEDGEMENTS ’ according to the manufacturer s protocol. NFAT activity was normalized to We thank the members of the imaging platform BICEL, E Richard and C Slommiany the number of cells measured using a MTS assay. Each experiment was for their helpful contribution. This work was supported by INSERM, the University of repeated at least three times. Lille1 and the Region Nord-Pas de Calais. This work was supported by INSERM, the University of Lille1 and the Region Nord-Pas de Calais. Cell proliferation and cell cycle analysis Cell viability was assessed by a colorimetric MTS assay (CellTiter 96) REFERENCES according to manufacturer’s instructions. Cell proliferation was also assessed by manual counting of cells extruding Trypan Blue. Ki-67 or 1 Bidaud I, Mezghrani A, Swayne LA, Monteil A, Lory P. Voltage-gated calcium proliferating cell nuclear antigen staining were also used to discriminate channels in genetic diseases. Biochim Biophys Acta 2006; 1763: 1169–1174. cells in quiescence (G0 phase) and cycling cells. 2 Monteith GR, Davis FM, Roberts-Thomson SJ. Calcium channels and pumps in For cell cycle analysis, cells were grown in three 60-mm dishes. After cancer: changes and consequences. J Biol Chem 2012; 287: 31666–31673. treatment, cells were harvested and treated as described previously32 with 3 Lee JM, Davis FM, Roberts-Thomson SJ, Monteith GR. Ion channels and trans- propidium iodide (25 μg/ml). Propidium iodide fluorescence was measured porters in cancer. 4. Remodeling of Ca(2+) signaling in tumorigenesis: role at 520 nm (excitation at 488 nm) using a flow cytometer (Beckman Coulter of Ca(2+) transport. Am J Physiol Cell Physiol 2011; 301: C969–C976. Epics XL4-MCL, Roissy CDG, France). Data were interpreted using Multicycle 4 Wang XT, Nagaba Y, Cross HS, Wrba F, Zhang L, Guggino SE. The mRNA of L-type for Windows (Phoenix Flow system). Each experiment was repeated at least calcium channel elevated in colon cancer: protein distribution in normal and three times. cancerous colon. Am J Pathol 2000; 157: 1549–1562. 5 Toyota M, Ho C, Ohe-Toyota M, Baylin SB, Issa JP. Inactivation of CACNA1G, a T-type calcium channel gene, by aberrant methylation of its 5' CpG island in VEGF secretion assay human tumors. Cancer Res 1999; 59: 4535–4541. Secretion of VEGF was measured in vitro by enzyme-linked immunosorbant 6 Fiske JL, Fomin VP, Brown ML, Duncan RL, Sikes RA. Voltage-sensitive ion channels assay as described by the manufacturer (Abcam). Cells were cultured in and cancer. Cancer Metastasis Rev 2006; 25:493–500. 12-well plates for 3 days before the culture medium was changed. VEGF 7 Kunzelmann K. Ion channels and cancer. J Membr Biol 2005; 205:159–173. was assayed in the culture medium after 24 h’s incubation (three wells per 8 Bennett ES, Smith BA, Harper JM. Voltage-gated Na+ channels confer invasive condition). Each experiment was repeated at least three times. properties on human prostate cancer cells. Pflugers Arch 2004; 447:908–914. 9 Mariot P, Vanoverberghe K, Lalevee N, Rossier MF, Prevarskaya N. Overexpression of an alpha 1H (Cav3.2) T-type calcium channel during neuroendocrine differ- In vivo experiments entiation of human prostate cancer cells. J Biol Chem 2002; 277: 10824–10833. Six-week-old male swiss nude mice (Charles River Laboratories, St Germain 10 Skryma R, Van Coppenolle F, Dufy-Barbe L, Dufy B, Prevarskaya N. sur l'Arbresle, France) were injected with LNCaP, LNCaP-Neo or LNCaP- Characterization of Ca(2+)-inhibited potassium channels in the LNCaP human α2δ2 cells. Six million cells were injected into both flanks of each mouse. prostate cancer cell line. Receptors Channels 1999; 6: 241–253. 42,43 Such double-injection protocols are often used in xenograft assays. 11 Spitzner M, Ousingsawat J, Scheidt K, Kunzelmann K, Schreiber R. Voltage-gated Though one cannot exclude that a tumor may affect the other, LNCaP cells K+ channels support proliferation of colonic carcinoma cells. Faseb J 2007; 21: 44 injected subcutaneously do not metastasize 12 weeks after injection and 35–44. should not diffuse to the other tumor. We did not observe any sign of 12 Gackiere F, Bidaux G, Delcourt P, Van Coppenolle F, Katsogiannou M, Dewailly E metastasis. Cells were prepared in a mixture composed of 50% phosphate- et al. CaV3.2 T-type calcium channels are involved in calcium-dependent secre- buffered saline and 50% BD-Matrigel (BD Bioscience, Rungis, France). tion of neuroendocrine prostate cancer cells. J Biol Chem 2008; 283: 10162–10173. Tumors were measured twice a week using a caliper, and animals were 13 Lerman MI, Minna JD. The 630-kb lung cancer homozygous deletion region on killed 12 weeks after injection unless the mice had to be killed earlier if the human chromosome 3p21.3: identification and evaluation of the resident can- total tumor size reached 10% of the animal weight. Tumor volume was didate tumor suppressor genes. The International Lung Cancer Chromosome calculated using the following formula: 3p21.3 Tumor Suppressor Gene Consortium. Cancer Res 2000; 60:6116–6133. Volume (in mm3) = length (in mm) × depth (in mm) × width (in mm) × π/6. 14 Ghosh S, Ghosh A, Maiti GP, Alam N, Roy A, Roy B et al. Alterations of 3p21.31 On the day of killing, tumors were weighed and divided for further IH, tumor suppressor genes in head and neck squamous cell carcinoma: Correlation western blot and RT–PCR experiments. At least 10 animals per condition with progression and prognosis. Int J Cancer 2008; 123: 2594–2604. were used. In order to study gabapentin action on tumor development, 15 Wanajo A, Sasaki A, Nagasaki H, Shimada S, Otsubo T, Owaki S et al. gabapentin was administered via the drinking water, which was renewed Methylation of the calcium channel-related gene, CACNA2D3, is frequent and every other day. Water consumption was assessed to be 5 ml/day/mouse. a poor prognostic factor in gastric cancer. Gastroenterology 2008; 135:580–590. Gabapentin concentration in the drinking water was set at 400 mg/l in 16 Mitra R, Lee J, Jo J, Milani M, McClintick JN, Edenberg HJ et al. Prediction of order for each animal to consume 60 mg/day/kg, which is to the maximal postoperative recurrence-free survival in non-small cell lung cancer by using an dose used in human medical treatments. The stability of gabapentin being internationally validated gene expression model. Clin Cancer Res 2011; 17: very high in water even at room temperature, its concentration should not 2934–2946. be affected in our experiments.45 17 Halatsch ME, Low S, Mursch K, Hielscher T, Schmidt U, Unterberg A et al. Candidate genes for sensitivity and resistance of human glioblastoma multiforme cell lines to erlotinib. Laboratory investigation. J Neurosurg 2009; 111:211–218. Statistical analysis 18 Carboni GL, Gao B, Nishizaki M, Xu K, Minna JD, Roth JA et al. Plots were produced using Origin 7.0 (Microcal Software Inc., North- CACNA2D2-mediated apoptosis in NSCLC cells is associated with alterations of ampton, MA, USA). Results are expressed as mean ± s.e.m. Statistical the intracellular calcium signaling and disruption of mitochondria membrane analyses were performed using unpaired t-tests (for comparing two integrity. Oncogene 2003; 22: 615–626. groups), analysis of variance tests followed by either Dunnett (for multiple 19 Dooley DJ, Taylor CP, Donevan S, Feltner D. Ca2+ channel alpha2delta ligands: control vs test comparisons) or Student–Newman–Keuls post-tests (for novel modulators of neurotransmission. Trends Pharmacol Sci 2007; 28:75–82.

© 2015 Macmillan Publishers Limited Oncogene (2015) 5383 – 5394 CACNA2D2 stimulates tumor cell growth M Warnier et al 5394 20 Li Z, Taylor CP, Weber M, Piechan J, Prior F, Bian F et al. Pregabalin is a potent and 33 Canti C, Nieto-Rostro M, Foucault I, Heblich F, Wratten J, Richards MW et al. selective ligand for alpha(2)delta-1 and alpha(2)delta-2 calcium channel subunits. The metal-ion-dependent adhesion site in the Von Willebrand factor-A domain of Eur J Pharmacol 2011; 667:80–90. alpha2delta subunits is key to trafficking voltage-gated Ca2+ channels. Proc Natl 21 Starborg M, Gell K, Brundell E, Hoog C. The murine Ki-67 cell proliferation antigen Acad Sci USA 2005; 102: 11230–11235. accumulates in the nucleolar and heterochromatic regions of interphase cells and 34 Whittaker CA, Hynes RO. Distribution and evolution of von Willebrand/integrin A at the periphery of the mitotic chromosomes in a process essential for cell cycle domains: widely dispersed domains with roles in cell adhesion and elsewhere. progression. J Cell Sci 1996; 109: 143–153. Mol Biol Cell 2002; 13: 3369–3387. 22 Dolphin AC. Calcium channel auxiliary alpha2delta and beta subunits: trafficking 35 Eroglu C, Allen NJ, Susman MW, O'Rourke NA, Park CY, Ozkan E et al. Gabapentin and one step beyond. Nat Rev Neurosci 2012; 13:542–555. receptor alpha2delta-1 is a neuronal thrombospondin receptor responsible for 23 Liu C, Hermann TE. Characterization of ionomycin as a calcium ionophore. J Biol excitatory CNS synaptogenesis. Cell 2009; 139:380–392. Chem 1978; 253: 5892–5894. 36 Garcia K, Nabhani T, Garcia J. The calcium channel alpha2/delta1 subunit is 586 – 24 Lytton J, Westlin M, Hanley MR. Thapsigargin inhibits the sarcoplasmic or endo- involved in extracellular signalling. J Physiol 2008; : 727 738. plasmic reticulum Ca-ATPase family of calcium pumps. J Biol Chem 1991; 266: 37 Bornstein P. Thrombospondins function as regulators of angiogenesis. JCell 3 – 17067–17071. Commun Signal 2009; : 189 200. 25 Davies A, Hendrich J, Van Minh AT, Wratten J, Douglas L, Dolphin AC. Functional 38 Kondratska K, Kondratskyi A, Yassine M, Lemonnier L, Lepage G, Morabito A et al. biology of the alpha(2)delta subunits of voltage-gated calcium channels. Trends Orai1 and STIM1 mediate SOCE and contribute to apoptotic resistance 1843 Pharmacol Sci 2007; 28:220–228. of pancreatic adenocarcinoma. Biochim Biophys Acta 2014; : 2263–2269. 26 da Costa Prando E, Cavalli LR, Rainho CA. Evidence of epigenetic regulation of the 39 Chomczynski P, Sacchi N. Single-step method of RNA isolation by acid tumor suppressor gene cluster flanking RASSF1 in breast cancer cell lines. guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 1987; 162: Epigenetics 2011; 6: 1413–1424. 156–159. 27 Zhao W, Wang L, Han H, Jin K, Lin N, Guo T et al. 1B50-1, a mAb raised against 40 Katsogiannou M, El Boustany C, Gackiere F, Delcourt P, Athias A, Mariot P et al. recurrent tumor cells, targets liver tumor-initiating cells by binding to the calcium Caveolae contribute to the apoptosis resistance induced by the alpha(1A)- channel alpha2delta1 subunit. Cancer Cell 2013; 23:541–556. adrenoceptor in androgen-independent prostate cancer cells. PLoS One 2009; 28 Negrini S, Prada I, D'Alessandro R, Meldolesi J. REST: an oncogene or a tumor 4: e7068. suppressor? Trends Cell Biol 2013; 23:289–295. 41 Gackiere F, Bidaux G, Lory P, Prevarskaya N, Mariot P. A role for voltage gated 29 Pelosi G, Fumagalli C, Trubia M, Sonzogni A, Rekhtman N, Maisonneuve P et al. T-type calcium channels in mediating "capacitative" calcium entry? Cell Calcium Dual role of RASSF1 as a tumor suppressor and an oncogene in neuroendocrine 2006; 39: 357–366. 30 – tumors of the lung. Anticancer Res 2010; : 4269 4281. 42 Oh C, Park S, Lee EK, Yoo YJ. Downregulation of ubiquitin level via knockdown of 30 Hobom M, Dai S, Marais E, Lacinova L, Hofmann F, Klugbauer N. Neuronal polyubiquitin gene Ubb as potential cancer therapeutic intervention. Sci Rep distribution and functional characterization of the calcium channel alpha2delta-2 2013; 3: 2623. 12 – subunit. Eur J Neurosci 2000; :1217 1226. 43 Wilhelm MT, Rufini A, Wetzel MK, Tsuchihara K, Inoue S, Tomasini R et al. Isoform- 31 Dolphin AC, Wyatt CN, Richards J, Beattie RE, Craig P, Lee JH et al. The effect of specific p73 knockout mice reveal a novel role for delta Np73 in the DNA damage alpha2-delta and other accessory subunits on expression and properties of the response pathway. Genes Dev 2010; 24:549–560. calcium channel alpha1G. J Physiol 1999; 519:35–45. 44 Lim DJ, Liu XL, Sutkowski DM, Braun EJ, Lee C, Kozlowski JM. Growth of an 32 Gackiere F, Warnier M, Katsogiannou M, Derouiche S, Delcourt P, Dewailly E et al. androgen-sensitive human prostate cancer cell line, LNCaP, in nude mice. Prostate Functional coupling between large-conductance potassium channels and Cav3.2 1993; 22: 109–118. voltage-dependent calcium channels participates in prostate cancer cell growth. 45 Zour E, Lodhi SA, Nesbitt RU, Silbering SB, Chaturvedi PR. Stability studies of Biol Open 2013; 2: 941–951. gabapentin in aqueous solutions. Pharm Res 1992; 9:595–600.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)