Recent studies suggest that SIPA1 encoding a Rap GTPase-activating protein SPA-1 is a candidate metastasis efficiency-modifying gene in human breast cancer. In this study, we investigated the expression and function of SPA-1 in human prostate cancer (CaP). Immunohistochemical studies of tumor specimens from CaP patients revealed a positive correlation of SPA-1 expression with disease progression and metastasis. The correlation was recapitulated in human CaP cell lines; LNCaP that rarely showed metastasis in SCID mice expressed an undetectable level of SPA-1, whereas highly metastatic PC3 showed abundant SPA-1 expression. Moreover, SIPA1 transduction in LNCaP caused prominent abdominal lymph node metastasis without affecting primary tumor size, whereas shRNA-mediated SIPA1 knockdown or expression of a dominant-active Rap1 mutant (Rap1V12) in PC3 suppressed metastasis. LNCaP transduced with SPA-1 (LNCaP/SPA-1) showed attenuated adhesion to the precoated extracellular matrices (ECM) including collagens and fibronectin, due to defective ECM-medicated Rap1 activation. In addition, LNCaP/SPA-1 showed a diminished level of nuclear Brd4, which is known to bind SPA-1, resulting in reduced expression of a series of ECM-related genes. These results suggest that SPA-1 plays an important role in controlling metastasis efficiency of human CaP by regulating the expression of and interaction with ECM in the primary sites. (Cancer Sci 2011; 102: 828–836)
Prostate cancer (CaP) is the most common male malignancy in the Western world.(1) Data from the Surveillance, Epidemiology, and End Results program of the National Cancer Institute between 1999 and 2006 indicated that 16% of patients presented regional lymph node involvement or distant metastasis at the time of diagnosis. Importantly, those with distant metastasis showed poor prognosis with a 5-year survival rate of 30.2% (http://www.seer.cancer.gov/). Thus, clarification of molecular mechanisms of metastasis(2) is a crucial step towards establishing a new therapeutic strategy.
Cancer metastasis consists of a series of complex processes, and multiple genetic factors may be involved in the metastatic efficiency, such as mutations accumulated during primary tumor evolution, genetic features intrinsic to the original tumor cell clones and host genetic backgrounds.(3,4) One of the initial steps of metastasis is invasion and detachment of tumor cells from the primary tissues, in which the interaction of tumor cells with ECM plays an important role.(5,6) In agreement, among numerous metastasis-predictive genes in various tumors, the common molecular signature of metastasis involves those related to ECM interactions.(7–9)
Accumulating evidence indicates an important role of Rap signaling in controlling cell–ECM and cell–cell adhesion, in part via regulation of integrins and cadherins.(10,11) Rap signaling is controlled by multiple regulatory factors, including Rap GDP/GTP exchange factors and Rap GTPase-activating proteins (GAP),(12) and the deregulation may profoundly affect the adhesive behavior of tumor cells. For instance, loss-of-function mutations of DOCK4 were reported to result in the defective Rap activation, leading to impaired intercellular adherence junctions and aggressive invasion of certain human tumor cells.(13) Recently, it was reported that SIPA1, encoding a specific RapGAP SPA-1,(14) was a candidate dominant locus of the host genetic background controlling the metastatic efficiency of mammary tumors.(15,16) More recently, two other genes, Rrp1b and Brd4, whose products interacted physically with SPA-1, were also reported to be predictive of breast cancer survival in humans.(17,18) Although SPA-1 plays an important role in normal lymphohematopoietic development and leukemia genesis,(19,20) its possible roles in other cancer cells remain to be elucidated.
In the current study, we investigated the roles of SPA-1 in human CaP. We show that the expression levels of SPA-1 in primary CaP positively correlate with disease progression and metastasis, and demonstrate that SPA-1 plays a crucial role in the metastasis efficiency of human CaP cell lines in SCID mice. We provide evidence that SPA-1 promotes metastasis of CaP by inhibiting the adhesive anchoring to ECM as well as the expression of a series of ECM-related genes.
Materials and Methods
Clinical tissue specimens. Biopsy samples from 94 patients with CaP were obtained from a tissue bank in the Department of Urology at Kyoto University Hospital under the protocols approved by the institutional review board.
Mice. Five-week-old male C.B-17/IcrCrj SCID mice were obtained from Charles River Japan (Yokohama, Japan) and were maintained in a specific pathogen-free condition. Animal experiments were performed according to the guidelines of the Kyoto University Experimental Animal Center.
Cell lines. Human CaP cell lines, LNCaP and PC3, were obtained from the American Type Culture Collection (Rockville, MD, USA), and MCF10A, a mammary epithelial cell line, was kindly provided by Dr Masahiko Itoh (Dokkyo University, Tochigi, Japan). Cap cells were maintained in RPMI supplemented with 10% FBS. MCF10A cells were cultured in DMEM supplemented with 10% FCS, 20 ng/mL epidemal growth factor (EGF), 10 μg/mL insulin and 0.5 mg/mL hydrocortisone.
Gene transfection. Cells were transfected with a pcDNA3.1 vector (Invitrogen, Carlsbad, CA, USA) containing SIPA1 or Frag-tagged Rap1V12 cDNA using Lipofectamine 2000 (Life Technologies, Inc., Gaithersburg, MD, USA) and selected with 1 mg/mL G418. LNCaP cells were also infected with the retrovirus containing SIPA1, which was obtained by transfecting PMX-SIPA1-IGFP plasmid in G3T-hi cells using a Retrovirus Packaging kit Ampho (Takara Bio, Shiga, Japan), and GFP+ cells were sorted with FACS Vantage (BD Biosciences, Franklin Lakes, NJ, USA). SIPA1 knockdown was performed by infecting pSINsi-hU6 (Takara Bio) retrovirus containing short hairpin RNA for SIPA1 (shSPA-1) or scramble RNA as follows: SPA-1 RNAi (1); 5′-GATCCGCAACGACATTGTGACCATTAGTGCTCCTGGTTGATGGTCACAATGTCGTTGCTTTTTTAT-3′ and 5′-CGATAAAAAAGCAACGACATTGTGACCATCAACCAGGAGCACTAATGGTCACAATGTCGTTGCG-3′, SPA-1 RNAi (2); 5′-GATCCGCTACTTGCAACACCATTCTTAGTGCTCCTGGTTGAGAATGGTGTTGCAAGTAGCTTTTTTAT-3′ and 5′-CGATAAAAAAGCTACTTGCAACACCATTCTCAACCAGGAGCACTAAGAATGGTGTTGCAAGTAGCG-3′, scramble RNAi; 5′-GATCCGTACAGCGGTCCAATCATAGTAGTGCTCCTGGTTGCTATGATTGGACCGCTGTACTTTTTTAT-3′ and 5′-CGATAAAAAAGTACAGCGGTCCAATCATAGCAACCAGGAGCACTACTATGATTGGACCGCTGTACG-3′.
Antibodies. Rabbit anti-SPA-1 antibody has been described previously.(14) Other antibodies were as follows: anti-Rap1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-β1 integrin, anti-active form of β1 integrin, anti-β-actin (Abcam, Cambridge, UK), anti-Brd4 (Abcam), anti-FLAG (Sigma, St Louis, MO, USA) and biotinylated anti-rabbit IgG (Vector Laboratories, Burlingame, CA, USA).
Immunohistochemistry and immunostaining. Paraffin-embedded sections of the specimen were subjected to immunohistochemical staining as previously described.(21) The samples were examined blindly by two independent researchers and graded from 0 to 2 to represent from no staining to strong staining, respectively. For immunostaining, cells cultured on coverslips were fixed with 1% formaldehyde, blocked with 1% BSA, incubated with primary antibodies followed by secondary antibodies, embedded in Mowiol (Calbiochem, San Diego, CA, USA) and observed under a fluorescence microscopy (Carl Zeiss, Oberkochen, Germany) as before.(22)
Immunoblotting and flow cytometry. Cells were lysed in lysis buffer (50 mM Tris–HCl [pH 7.4], 1% Triton X-100, 150 mM NaCl, 2 mM EDTA, 1 mM EGTA, phosphatase and protease inhibitor cocktail) and subjected to immunoblotting. Rap1 GTP was detected by a pull-down assay. Subcellular components of cells were separated with the use of ProteoExtract Subcellular Proteome Extraction kit (Calbiochem). For detecting activated β1-integrin, cells were incubated with an antibody specific for activated β1-integrin (HUTS-4, Chemicon, Temecula, CA, USA) at 37°C for 30 min in the presence of 1 mM MnCl2 followed by secondary antibody on ice and analyzed with FACSCanto (BD Biosciences).
Cell invasion and attachment assay. A cell invasion assay was done with the use of BD Biocoat Matrigel Invasion Chambers (BD Biosciences) as before.(23) For the ECM attachment assay, cells were seeded on ECM-coated six-well plates (BD BioCoat variety pack II) at 2 × 105 cells/well for 24 h.
In vivo metastasis assay. CaP cells (5 × 106 cells in 20 μL medium) were inoculated into the left testes of SCID mice.
Statistical analysis. The Mann–Whitney U test was used, and statistical analyses were performed using StatView ver. 5.0 (SAS Institute, Cary, NC, USA).
Positive correlation of SPA-1 expression levels with disease progression and metastasis in human primary CaP. We first investigated the expression of SPA-1 in CaP tissues from 94 patients with the use of immunohistochemistry. SPA-1 expression levels and Gleason scores were scored independently for each sample. Cancer tissues with Gleason scores 7 and above exhibited abundant SPA-1 expression, whereas those with scores less than seven hardly expressed SPA-1 (Fig. 1). Also, SPA-1 expression levels were increased significantly according to the progression of clinical stages from organ-confined, extracapsular to metastatic diseases (Fig. 1B).
SPA-1 regulates the metastasis efficiency of human CaP cell lines. We then investigated the involvement of SPA-1 in the metastasis of CaP with the use of two representative human CaP cell lines, LNCaP and PC3. PC3 showed much greater “invasion” activity through the Matrigel-coated filters in vitro than LNCaP (Fig. S1A). When these cells were inoculated into the testis of immunodeficient SCID mice, PC3 developed prominent metastasis in the abdominal lymph nodes, whereas LNCaP hardly showed any nodal involvement (Fig. S1B). Metastatic PC3 exhibited abundant SPA-1 expression, whereas nonmetastatic LNCaP showed undetectable SPA-1 expression; the difference was attributable to that of SPIA1 transcripts (Fig. S1C,D). As anticipated, the basal level of Rap1GTP in PC3 was much less than that in LNCaP (Fig. S1C). To directly examine the role of SPA-1 in metastasis, two independent LNCaP lines stably expressing SPA-1 (LNCaP/SPA-1) were established, both of which showed an undetectable level of basal Rap1GTP (Fig. 2A). Both LNCaP/SPA-1 lines formed primary tumors in the Scid testis of comparable sizes to control LNCaP/Vect; however, LNCaP/SPA-1 lines showed significantly enhanced invasion activity in vitro compared with LNCaP/Vect and developed massive metastasis in the abdominal lymph nodes in SCID mice, whereas LNCaP/Vect rarely metastasized (Fig. 2B–D). As a reverse experiment, we stably knocked down SPIA1 in PC3 by using two independent shRNA, which caused markedly reduced endogenous SPA-1 and a concomitant increase in basal Rap1GTP (Fig. 3A). Both PC3/shSPA-1 lines showed markedly reduced invasion activity in vitro as well as diminished lymph node metastasis in vivo again without affecting the primary tumor sizes (Fig. 3B–D). Altogether, these results demonstrate that endogenous expression of SPA-1 is required and sufficient for the lymph node metastasis of human CaP cell lines.
Expression of SPA-1 results in attenuated attachment of CaP cells to ECM. Given that Rap signaling regulates cell adhesion, we next examined the effects of SPA-1 expression in LNCaP on the attachment to precoated ECM. MCF10A with features close to normal epithelial cells was used as a control. None of these lines were attached to uncoated bacterial grade dishes (Fig. 4A). MCF10A cells were firmly attached to all the major components of ECM, including collagen type I (Col-I), type IV (Col-IV), fibronectin (FN) and laminin (LM). LNCaP/Vect cells were also attached to the ECM, except for LM, albeit they were spread with a spindle-like form and less adherence junctions (Fig. 4A). The ECM attachment of LNCaP/SPA-1 was significantly reduced compared with that of LNCaP/Vect, whereas the attachment to poly-D-lysine (pDl)-coated dishes via the charge effect was comparable (Fig. 4A,B). Both LNCaP/SPA-1 and LNCaP/Vect expressed the same levels of β1-integrin, which could be activated comparably in the presence of Mn2+ (Fig. 4C). However, although LNCaP/Vect showed strong Rap activation on contact with ECM (Col-IV) much more than on poly-D-lysine (pDI), which was sustained during the adhesion, minimal Rap activation was detectable in LNCaP/SPA-1 upon contact with the ECM (Fig. 4D). The results suggest that CaP cells with abundant SPA-1 expression exhibit the attenuated cell adhesion to ECM, at least in part due to the defective ECM-induced Rap1 activation.
Expression of SPA-1 causes diminished nuclear Brd4 and reduced ECM-related gene expression. We previously reported that SPA-1 was capable of interacting directly with a chromatin adaptor, Brd4.(24) Immunostaining analysis indicated that Brd4 was mostly, if not exclusively, in the nuclei of LNCaP/Vect cells (Fig. 5A). In LNCaP/SPA-1 cells, which strongly expressed SPA-1 largely in the cytosol, nuclear Brd4 expression appeared to be reduced compared with LNCaP/Vect cells (Fig. 5A). It was confirmed by immunoblotting analysis that Brd4 in the nuclear fraction of LNCaP/SPA-1 was significantly diminished compared with LNCaP/Vect cells, whereas cytosolic Brd4 was largely comparable (Fig. 5B). Although some SPA-1 was detected in the “nuclear” fraction, we previously indicated that it was attributable in part to the SPA-1 in the insoluble cytoskeletal fraction.(25) It was reported that Brd4 regulated a series of ECM-related gene expression in mammary tumor cells,(18) and therefore we investigated it with the use of PCR-array analysis. LNCaP/SPA-1 showed significantly diminished expression of a number of ECM-related genes, including COLs, FN1 and ITGA1 (integrin α1; Fig. 5C, Table S1). We also examined the expression of Brd4 in primary human CaP specimens. Whereas Brd4 was detected almost exclusively in the nuclei of low-grade CaP cells (Gleason score 6) with little SPA-1 expression, high-grade CaP cells (Gleason score 8) strongly expressing SPA-1 showed abundant expression of Brd4 in the cytosol (Fig. 5D). An inverse correlation was observed between SPA-1 expression and Brd4 staining in the nuclei (Table S2). These results suggest that SPA-1 expression also attenuates the expression of a series of ECM-related genes per se in CaP cells.
Constitutive Rap1 activation suppresses metastasis of PC3. We finally addressed whether the metastasis-enhancing effect of endogenous SPA-1 in CaP cells was mediated via deregulated Rap signaling. Toward this end, we introduced a dominant active mutant of Rap1 (Rap1V12) resistant to the GAP activity of SPA-1 in PC3. Despite abundant expression of endogenous SPA-1, two independent PC3/Rap1V12 lines expressed high basal levels of Rap GTP and showed reduced invasion activity in vitro compared with PC3/Vect (Fig. 6A,B). Moreover, both of them showed markedly diminished lymph node metastases in Scid mice (Fig. 6C,D). Additionally, PC3/Rap1V12 lines also developed smaller primary tumors than PC3/Vect (Fig. 6C,D). The results suggest that the deregulated Rap activation plays an important role in the metastasis-enhancing effect of SPA-1 in CaP cells.
In the current study, we demonstrated that the endogenous expression levels of SPA-1 in primary human CaP cells showed a significantly positive correlation with disease progression and metastasis. The correlation was recapitulated in the representative human CaP cell lines; thus, a highly metastatic PC3 line abundantly expressed SPA-1, whereas a non-metastatic LNCaP line showed negligible SPA-1 expression. Transduction of SPA-1 in LNCaP induced a potent metastatic activity to abdominal lymph nodes in vivo in SCID mice. Because primary tumor size in the testis was unaffected, the effect was specific for metastasis efficiency rather than secondary to the tumor growth rate. Conversely, shRNA-mediated knockdown of endogenous SPA-1 in PC3 strongly suppressed the metastatic activity, again without affecting the primary tumor growth. Furthermore, transduction of dominant active Rap1V12, a GAP–resistant Rap1 mutant, in PC3 also inhibited the metastatic activity. However, it was noted that PC3/Rap1V12 showed reduced growth in the primary site as well. The effect may be due to the “unphysiological” level of exogenous Rap1V12 activation, which may inhibit the Ras-ERK pathway,(26) and thus the results may need further careful verification. Nonetheless, these results suggest that SPA-1 plays a crucial role in promoting the metastasis efficiency of CaP cells in vivo, in part via regulation of Rap signaling.
LNCaP/SPA-1 showed significantly attenuated adhesion to pre-established ECM, such as Col-I, Col-IV and FN, compared with LNCaP/Vect, whereas charge-mediated attachment via pDl was unaffected. Although SPA-1 expression did not affect the expression and Mn2+-mediated activation of integrin β1, Rap1 activation hardly occurred in LNCaP/SPA-1 on contact with ECM, whereas LNCaP/Vect showed strong and sustained Rap activation during the ECM attachment. It is well established that Rap signaling plays a crucial role in “inside out” activation of integrins,(10,27) and thus it is strongly suggested that the attenuated adhesion of LNCaP/SPA-1 is attributable in part to the impaired Rap activation via ECM. It remains to be seen how such an effect is related to the increased transmigration through Matrigels in vitro. It may be conceivable that impaired anchoring to ECM promotes the transmigration via chemoattractive factors such as TGF-β in serum and Matrigels.
SPA-1 directly interacts with a number of proteins in various subcellular compartments, such as AF-6 (affadin), aquaporin-2 and Brd4.(24,28,29) It was reported that ectopic expression of Brd4 in a mammary tumor cell line caused marked alteration of a series of ECM-related gene expression and suppressed the metastasis efficiency.(18) We found that expression of SPA-1 in LNCaP resulted in the diminution of endogenous Brd4 preferentially in the nuclei, probably due to cytoplasm sequestration.(24) Concomitantly, LNCaP/SPA-1 revealed reduced expression of a series of ECM-related genes, including Col’s and FN. Thus, SPA-1 may additionally contribute to the attenuated ECM-anchoring of CaP cells via reduced ECM production per se by cancer cells. The association of cytosolic SPA-1 with Brd4 may additionally enhance the Rap GAP activity of SPA-1,(24) further promoting Rap1 inactivation.
Cancer cells may invade tissues as individual cells, or in sheets or clusters, which is called collective migration.(5) Collective migration may occur as a continuous protrusion from primary tumors.(30) It may also result from the detachment of cell clusters and their dissemination away from primary tumors into interstitial tissue gaps or lymphatics,(6,31) where integrin-mediated ECM-anchoring of cancer cells may function as a barrier. In a clinical setting, therefore, it may be conceivable that CaP cells with abundant SPA-1 expression show a propensity to detach from the primary tissue, leading to the collective migration to lymph nodes via lymphatic flow. Recently, on the other hand, it was reported that the expression of Rap1E63, an active yet GAP-susceptible Rap1 mutant, in PC3 cells rather enhanced the experimental blood-born metastasis in nude mice.(32) Thus, roles of Rap signaling in distinct aspects of metastasis remain to be carefully investigated.
A meta-analysis of gene expression data (Oncomine website: https://www.oncomine.org/resource/login.html) indicates that human primary prostate cancer tissues express higher levels of SIPA1 transcripts than normal prostate specimens;(33) moreover, SIPA1 expression was significantly higher in metastatic than primary lesions.(34) Intriguingly, expression of RAP1A transcripts shows reverse profiles to those of SIPA1.(34,35) These data are in good agreement with the present results at protein levels. The mechanisms for SIPA1 activation during human CaP progression remain to be elucidated. Nonetheless, the current results strongly suggest that SPA-1 plays a crucial role in controlling metastasis efficiency of CaP cells.
This work was supported by Grant-in-Aid from Ministry of Education, Culture, Sports, Science and Technology, Japan, Yamaguchi Endocrine Research Association, Organon Urology Academia, Takeda Science Foundation and the Japanese Foundation for Prostate Research.
The authors declare no competing financial interests.