Naftopidil, a Selective α₁-Adrenoceptor Antagonist, Suppresses Human Prostate Tumor Growth by Altering Interactions between Tumor Cells and Stroma

Recently, several studies demonstrated that alpha₁-adrenoceptor (α₁-AR) antagonists, which are generally prescribed for benign prostatic hyperplasia (BPH) patients, have antiproliferative effects on prostate cancer (PCa). Epidemiologically, 2 observational cohort studies have shown an apparent low prevalence of PCa in BPH patients receiving α₁-AR antagonists. Harris et al. reported that the administration of quinazoline-based α₁-AR antagonists (subtype nonselective: doxazosin or terazosin) significantly decreased the incidence of PCa (1). Murtola et al. reported that the administration of α₁-AR antagonists (subtype nonselective: alfuzosin or subtype-selective: tamsulosin) decreased the incidence of high-grade PCa; the decreasing trend correlated with the cumulative duration of α₁-AR antagonists use (2).

BPH is a common prostatic disease in elderly men. In regard to treatment for BPH, α₁-AR antagonists are the most widely used therapeutic agents. Although the incidence of BPH increases with age, this increase occurs faster than that of PCa (3). In addition, most PCa arises in the prostate concomitant with BPH elsewhere (4). This means that, in general, the administration of α₁-AR antagonists for BPH treatment often comes before diagnosis of PCa. Accordingly, it is very interesting to examine the possibility of α₁-AR antagonists suppress prostate tumor growth. Previous studies have documented the induction of apoptosis in PCa cells by subtype nonselective α₁-AR antagonists (doxazosin, terazosin, and prazosin; refs. 5, 6). In addition, the G₁ cell-cycle arrest was induced in PCa cells by subtype-selective α₁-AR antagonist naftopidil, which is only marketed in Japan (7). The underlying molecular mechanism by which α₁-AR antagonists inhibit PCa cells is gradually being elucidated.

In general, cancer cells exist in a microenvironment composed of extracellular matrix (ECM), tumorigenic soluble factors, and stromal cells including fibroblasts, smooth muscle cells, endothelial cells, macrophages, and other cells (8). In this tumor microenvironment, various cytokines or growth factors act as messengers in reciprocal tumor–stroma interactions (9). Fibroblasts are likely sources of tumorigenic soluble factors that affect the growth of cancer cells. Thus, they have a crucial influence on the growth and progression of tumors (10). The normal prostate has an abundance of stromal components, mainly well-differentiated smooth muscle cells. In prostate tumors, however, fibroblasts surrounding cancer cells are associated not only with the initiation of cancer cells but also with tumor growth and progression to androgen independence (11–13). Olumi et al. demonstrated stroma-induced malignant transformation, indicating that fibroblasts surrounding epithelia play an important role in PCa development (14). Other groups reported that coinoculation of PCa cells with fibroblasts increased tumorigenicity and implied the promotion of angiogenesis (11, 15, 16). Therefore, the inhibition of tumor–stroma interactions with addition of the suppression of cancer cell growth might lead to synergistic effects on tumor control.

In the present study, we investigated the effects of subtype-selective α₁-AR antagonists (naftopidil, tamsulosin, and silodosin) on prostate tumor growth with a focus on the role of stroma, especially commercially available fibroblast cells (PrSC). Although we have already reported that the α_1D-selective antagonist naftopidil suppresses the growth of androgen receptor-positive human PCa line LNCaP and the androgen receptor-negative human PCa line PC-3 by induction of G₁ cell-cycle arrest (7), little is known about the relationship between the growth inhibitory effects and the androgen receptor signal or the α₁-AR signal in PCa cells. Thus, we also investigated the biochemical mechanisms of subtype-selective α₁-AR antagonists in PCa cells with a focus on the androgen sensitivity of the androgen receptor-positive PCa cells and the α₁-androgen receptor selectivity of each antagonist.

Normal human prostate total Poly A+ RNA (636124) was purchased from Clontech Laboratories, Inc. The 3 clinically used subtype-selective α₁-AR antagonists were kindly provided by the following pharmaceutical companies: naftopidil from Schering-Plough Co; tamsulosin hydrochloride from Astellas Japan, Inc; and silodosin from Daiichi Sankyo Co., Ltd. Selective α_1A antagonist RS100329 and selective α_1D antagonist BMY7378 were purchased from Tocris Bioscience and Sigma-Aldrich, Inc, respectively. Mouse monoclonal anti-human Ki-67 antigen (clone MIB-1) antibody was purchased from Dako Cytomation, Inc. Rat monoclonal anti-mouse CD31/PECAM-1 (MEC13.3) and mouse monoclonal anti-human p27kip1 (57) antibodies were purchased from BD Bioscience. Mouse monoclonal anti-p21waf1 (HZ52) antibody was purchased from Millipore Co. Mouse monoclonal anti-actin (AC-15) antibody was purchased from Sigma-Aldrich, Inc.

The androgen-sensitive, androgen receptor-positive human PCa cell line LNCaP was obtained from the American Type Culture Collection. Androgen low-sensitive, androgen receptor-positive E9 cells were derived from a parental LNCaP cell population by a limiting dilution method in regular culture conditions (17). Androgen-insensitive, androgen receptor-positive AIDL cells were established from parental LNCaP cells by continuous passaging in hormone-depleted conditions (18). LNCaP and E9 cells were cultured in phenol red (+) RPMI-1640 supplemented with 10% fetal bovine serum (FBS) and 1% antibiotic/antimycotic solution. AIDL cells were cultured in phenol red–free RPMI-1640 supplemented with 10% charcoal-stripped (CS) FBS and 1% antibiotic/antimycotic solution. The parental LNCaP, E9, and AIDL lines expressed similar levels of androgen receptor in culture, but prostate-specific antigen (PSA) protein was detected only in parental LNCaP cells (19). The androgen sensitivity of parental LNCaP, E9, and AIDL was confirmed by the change of PSA mRNA expression in cell culture with synthetic androgen R1881 (19). Human prostate stromal cell PrSC was purchased from Lonza Group Ltd. PrSC was maintained using SCGM Bullet Kit (Lonza Group Ltd) as recommended.

Subconfluent cultures of PCa cells (parental LNCaP, E9, and AIDL lines) and PrSC were trypsinized and counted. Xenografts without PrSC contained 4 × 10⁵ PCa cells. Mixtures with PrSC were prepared for each PCa cell type by mixing 1 × 10⁵ cancer cells and 3 × 10⁵ PrSC in suspension. Pelleted cells were resuspended in 50 μL of neutralized type I rat tail collagen gels, and then grafted beneath the renal capsule of male athymic mice (7 to 8 weeks old).

Subsequently, we evaluated the effects of administration of subtype-selective α₁-AR antagonists on growth of tumors from an E9 + PrSC combination. Cellular combinations were prepared by mixing 1 × 10⁵ E9 cells and 3 × 10⁵ PrSC in suspension. Xenografts were transplanted into male athymic mice as described above. Treatments were started 7 days after transplantation. Each subtype-selective antagonist was administered orally (10 mg/kg/d) as a suspension in 0.5% of carboxymethylcellulose through a 22-gauge gavage needle for 28 days. The control group received only the diluent. All animals were maintained in a specific pathogen-free environment. The Mie University's Committee on Animal Investigation approved the experimental protocol.

Mice were sacrificed 35 days after transplantation. Grafts were harvested and their weights were measured. Tissues were fixed in 10% neutral-buffered formalin for regular H&E (hematoxylin and eosin) staining and human-specific Ki-67 staining or in IHC zinc fixative (BD Biosciences Pharmingen) for mouse-specific CD31 staining. Then, they were dehydrated and embedded in paraffin.

Every harvested tumor was analyzed for cell proliferation (Ki-67 labeling index) and microvessel density (MVD). Immunohistochemical (IHC) staining was performed with a Vectastain ABC elite kit (Vector Laboratories Inc) following our previously reported protocols (19). The extent of Ki-67 labeling was expressed as the percentage of positive nuclear cells. A “microvessel” was defined as mouse-specific CD31-positive endothelial cells that formed a vascular lumen. The numbers were determined by 2 investigators independently in a blinded fashion.

Total RNA was extracted from each cell line using the Qiagen mini RNA Easy kit (Qiagen Inc). cDNA was reverse transcribed from 2 μg of total RNA using High Capacity RNA-to-cDNA Kit (Life Technologies Co) in a total volume of 20 μL. RT-PCR were performed using 2 μL of cDNA with Takara Ex Taq (Takara Bio Inc) with specific primers (Table 1). After PCR, the amplified RT-PCR products were electrophoretically analyzed through 1.6% of ethidium bromide–stained agarose gel and photographed under UV light.

LNCaP (and derived) cell lines were plated onto 12-well plates (Nunc; 1.0 × 10⁴ parental LNCaP cells per well; 0.75 × 10⁴ E9 cells per well; 2.0 × 10⁴ AIDL cells per well). Two days after seeding, cells were treated with each α₁-AR antagonist (10 μmol/L) dissolved in dimethylsulfoxide (DMSO) or vehicle (0.1% DMSO) and incubated for 4 days. Then, cells were detached by trypsinization and counted using a hemocytometer. Cell viability was measured by trypan blue exclusion. The PrSC line was seeded into 96-well culture plates at a density of 5 × 10² cells per well. After 24 hours, cells were treated with various concentrations (0, 0.1, 1, and 10 μmol/L) of each α₁-AR antagonist and incubated for 3 days. Viable cells were measured using a spectrophotometric cell counting kit (Dojindo Lab).

E9 (1 × 10⁵ cells) and PrSC (2 × 10⁵ cells) were seeded onto 6-well plates (Nunc) and 90-mm culture dishes (Sumitomo Bakelit Co. Ltd). Two days after seeding, cells were treated with either 50 μmol/L (E9) or 25 μmol/L (PrSC) of each α₁-AR antagonist or vehicle (0.1% DMSO) for 48 hours. After treatment, cells were isolated and the nuclei stained using the Cycle TEST PLUS DNA Reagent kit (BD Immunocytometry Systems). To determine cell-cycle distribution, the DNA content of stained cells was analyzed on a BD FACS Cant II flow cytometer (Becton Dickinson).

E9 cells were seeded onto 6-well plates at a density of 1 × 10⁵ cells per well. Two days after seeding, E9 cells were treated with each α₁-AR antagonist (50 μmol/L) for 48 hours. Cells were trypsinized and collected, and then cells were stained with annexin V-fluorescein isothiocyanate (FITC) and propidium iodide (PI) simultaneously using the Annexin V-FITC Apoptosis Detection kit (BioVison). The cell suspensions were analyzed with a BD FACS Canto II flow cytometer (Becton Dickinson) to determine the percentage of apoptotic (FITC-stained cells) and necrotic cells (PI-stained cells). A minimum of 20,000 cells were collected for all the samples.

Extracted proteins were separated by gel electrophoresis and transferred to Immobilon polyvinylidene difluoride (PVDF) membranes following our previously reported protocol (7). The specific protein bands were detected with the LAS-4000 mini (Fuji Photo Film) using a Super Signal West Pico reagent kit (Pierce Biotechnology). Anti-p21waf1, p27kip1, and actin were used at dilutions of 1:2,500, 1:2,500, and 1:5,000, respectively.

PrSC was seeded onto 6-well plates at a density of 1 × 10⁴ cells per well. Twenty-four hours after seeding, cells were treated with each α₁-AR antagonist (10 μmol/L) for 3 days. Cells were harvested and counted, and culture supernatants were collected on day 4. For the quantitative determination of interleukin (IL)-6, the collected aliquots of medium were subjected to ELISA assays using a Quantikine human IL-6 immunoassay kit (R&D Systems).

In vivo data obtained from vehicle and α₁-AR antagonist administration groups were analyzed using Student's t test with Dunnet multiple comparison. Other data from in vitro experiments were analyzed using Student's t test. Values were expressed as the means ± SD. Values of P < 0.05 were considered statistically significant.

We used the commercially available human PrSC cell line as a source of stromal cells to simulate the tumor microenvironment. We first confirmed the characteristics of PrSC using RT-PCR analysis. PrSC expressed mRNAs of fibroblastic/myofibroblastic differentiation markers vimentin, tenascin (TN)-C, and α smooth muscle actin (SMA), but not smooth muscle differentiation markers myosin or desmin (Fig. 1A). PrSC expressed mRNA of tumorigenic growth factors, fibroblast growth factor (FGF)-2 and FGF-7, whereas the mRNA expression of insulin-like growth factor (IGF)-I was quite low (data not shown). In PrSC, the expression of androgen receptor protein was not detected, and no changes in cell proliferation were observed in culture with additional dihydrotestosterone (DHT; data not shown).

Parental LNCaP, E9, and AIDL lines were grafted alone or in combination with PrSC into athymic mice. As shown in Figure 1B, no significant difference of tumor weights was observed when LNCaP or its sublines were grafted without stroma. On the other hand, PCa tumors grown by combining E9 with PrSC resulted in significantly larger tumors (87.8 ± 21.9 mg) compared with tumors from LNCaP combined with PrSC (60.4 ± 35.0 mg; P < 0.05). There was no significant difference between LNCaP + PrSC tumors and AIDL + PrSC tumors. IHC analysis showed that cell proliferation (Ki-67 labeling index) in PCa + PrSC tumors was significantly higher than those of tumors without PrSC for all PCa cells (P < 0.01; Fig. 1C). As for the MVD in tumors, there was a statistically significant increase between grafting E9 or AIDL alone versus adding PrSC (P < 0.01; Fig. 1D).

Of the 3 cell lines, we found that E9 cells showed the highest responsiveness and most stable tumorigenesis when coinoculated with PrSC. Therefore, we chose the E9 + PrSC model to evaluate the effects of subtype-selective α₁-AR antagonists (naftopidil, tamsulosin, and silodosin) on PCa tumorigenesis in vivo.

We grafted E9 + PrSC cell mixtures into athymic mice and gave the subtype-selective α₁-AR antagonist (10 mg/kg/d) to the grafted mice daily by oral administration for 28 days. Figure 2A reveals the characteristic gross appearance and Ki-67 staining of E9 + PrSC tumors treated with vehicle or subtype-selective antagonists (naftopidil, tamsulosin, or silodosin). E9 + PrSC xenografts formed spheroidal reddish tumors filled with blood. The average relative rate of E9 + PrSC tumor weights divided by body weight in control animals, naftopidil-, tamsulosin-, and silodosin-treated animal were 0.369%, 0.192%, 0.267%, and 0.281%, respectively (Fig. 2B). A statistically significant reduction of relative rate was observed only in naftopidil-treated mice (P < 0.01). The Ki-67 index in E9 + PrSC tumors was decreased by all antagonists (P < 0.01; Fig. 2C). MVD in E9 + PrSC tumors was decreased by treatment of naftopidil and silodosin (P < 0.01; Fig. 2D).

There were differences in the mRNA expression level of the α₁-AR subtypes among the 3 PCa cell lines. α_1a-AR subtype mRNA was detected in LNCaP and AIDL, whereas its mRNA expression in E9 was quite low (Fig. 3A). Similar levels of α_1d-AR subtype mRNA were detected in LNCaP and E9 but not in AIDL. α_1b-AR subtype mRNA was not detected in any of the PCa cells. Although PrSC expressed all α₁-AR subtype mRNAs, expression of α_1a-subtype mRNA was comparatively low.

Naftopidil (α_1D-selective), silodosin (α_1A-selective), and RS100329 (α_1A-selective) showed approximately equivalent growth inhibitory effects on LNCaP, E9, and AIDL (Fig. 3B). Tamsulosin (α_1A- and α_1D-selective) and BMY 7378 (α_1D-selective) showed comparatively weak inhibitory effects on AIDL. With regard to PrSC, similar to PCa cells, growth inhibitory effects were observed by treatment with naftopidil, silodosin, and RS100329 in dose-dependent manners (Fig. 3C). The growth inhibitory effect of naftopidil was greater than that of silodosin.

To investigate whether subtype-selective α₁-AR antagonists affect the cell cycle, we carried out flow cytometric analysis and Western blot analysis of cell-cycle regulatory proteins. Compared with vehicle, naftopidil but neither tamsulosin nor silodosin increased the population of cells in the G₀/G₁ phase, which was accompanied by a decrease in the S/G₂ phase in E9 cells (Fig. 4A). Similar results were observed in PrSC with lower concentrations than that required to induce G₁ cell-cycle arrest in E9 cells (Fig. 4B). As shown in Figure 4C, naftopidil-treated E9 cells increased the level of cell-cycle regulatory protein p27, whereas no increase of p27 protein was observed in naftopidil-treated PrSC.

To confirm the apoptotic effect of subtype-selective α₁-AR antagonists, we used flow cytometry to characterize early apoptosis in annexin V–stained cells. Silodosin-treated E9 cells had a greater proportion of cells in the early phase of apoptosis (Table 2). Although statistical analysis indicated a significant difference (P < 0.05), the increased ratio was modest. Naftopidil and tamsulosin were not involved in apoptosis induction. Increased caspase-3 activity was observed only in silodosin-treated E9 cells.

To evaluate the effect of subtype-selective α₁-AR antagonists on secretion of IL-6 derived from PrSC, we performed ELISA. The total amount of IL-6 protein secreted from PrSC was decreased by naftopidil treatment (Fig. 5B). Real-time PCR analysis showed that the expression of FGF-2 and FGF-7 mRNAs by PrSC was not changed by treatment with any subtype-selective α₁-AR antagonist (data not shown).

We have demonstrated that subtype-selective α₁-AR antagonists have antiproliferative effects on PCa growth in both in vitro and in vivo systems. The major finding in this study was that oral administration of subtype-selective α₁-AR antagonist, naftopidil, suppressed E9 + PrSC tumor growth in vivo. The antiproliferative effect of naftopidil was due to inhibition of cell-cycle progression in both PCa cells and PrSC. Naftopidil was the only α₁-AR antagonist that showed strong antiproliferative effects on stromal cells among the 5 subtype-selective α₁-AR antagonists which included tamsulosin, silodosin, RS100329, and BMY7378. In addition, we showed that the antiproliferative effect on stromal cells resulted in the suppression of tumorigenic soluble factor derived from stromal cells. These results suggest that the marked decline of tumor weight caused by naftopidil treatment might be due to its antiproliferative effects on both PCa cells and stromal cells. It is very attractive that naftopidil exhibits this unique antiproliferative action on stromal cells.

Previous studies have reported that tumor–stroma interactions could play a role in the promotion of prostate carcinogenesis (16). Kawada et al. reported that PrSC-conditioned medium increased PCa cell proliferation in vitro, suggesting the significance of prostate stroma secreted factors in tumor–stroma interactions (20). FGF-2 and FGF-7 play an important role as mediators involved in tumor–stroma interactions (10). In PrSC, mRNA expressions of FGF-2 and FGF-7 were detected (data not shown). Previous report showed that responsiveness of E9 cells to FGF-2 and FGF-7 stimulation were higher than that of parental LNCaP cells (19), suggesting that E9 cells made larger tumor in combination with PrSC compared with LNCaP cells in vivo. Uemura et al. showed that the administration of angiotensin II receptor blockers (ARB), that are widely used as antihypertensive drugs, suppressed PC-3 + PrSC tumor growth in vivo by inhibition of PrSC cell proliferation and IL-6 secretion from PrSC (15). It is highly important that naftopidil shows antiproliferative effects not only on cancer cells but also on stromal cells, and that stromal cell growth was inhibited by lower concentrations than that required to inhibit cancer cells. In our in vivo tumorigenic study, although difference of a given concentration from clinically relevant therapeutic dose range was the smallest in naftopidil, the highest decline of tumor weight was observed with naftopidil treatment. From these findings, we suggest that subtype-selective α₁-AR antagonists such as naftopidil, which inhibit stromal cell growth and suppress the secretion of tumorigenic soluble factor, might prevent tumor–stroma interactions in PCa progression. For instance, naftopidil showed antitumor effects through suppression of IL-6 protein secreted from stromal PrSC. IL-6 plays an important role in tumorigenesis and is considered a positive growth factor for most PCa cells and an important messenger linking stromal and epithelial prostate cells (21). IL-6 is also involved in the regulation of PCa cell growth (22). The inhibition of IL-6 secreted from stromal cells likely contributes to the suppression of prostate tumor.

We found strong evidence for the antiproliferative effects of highly α_1A-selective RS100329 on both PCa cells and PrSC (Fig. 3). Naftopidil and RS100329 both possess a phenylpiperazine-based structure and arrest cells in G₁. Interestingly, the Ca²⁺/calmodulin-dependent protein kinase (CaM kinase) inhibitor KN-62 has a phenylpiperazine structure and induces G₁ cell-cycle arrest in small cell lung carcinoma (SCLC) cells (23, 24). In this study, the phenylpiperazine-derived, α_1D-selective antagonist BMY7378 did not induce G₁ cell-cycle arrest at low concentration (10 μmol/L; Fig. 3), whereas 5 times concentration of BMY7378 induced very weak G₁ cell-cycle arrest in E9 cells (data not shown). Taken together, it is possible that the α₁-AR antagonists with a phenylpiperazine-based structure might suppress PCa cell proliferation through G₁ cell-cycle arrest, whereas modification of the substituent group could change the properties of the compound, resulting in other effects on cells. The quinazoline-derived, subtype nonselective α₁-AR antagonist doxazosin is a well-investigated α₁-AR antagonist with antitumor effects (25–29). Kyprianou et al. focused attention on the quinazoline-based structure of doxazosin and indicated doxazosin-induced apoptosis might be independent of the α₁-AR signal and the biological characteristics of the cells (5, 30). Garrison and his colleagues demonstrated that novel lead quinazoline-derived compound (DZ-50) has demonstrated to reduce endothelial cell viability via a nonapoptotic mechanism, even though it has quinazoline structure (31). The characteristic of the compound depends strongly on substituent group. Thus, it is possible that BMY7378, which has phenylpiperazine structure, did not show any significant cell-cycle arrest at low concentration. Our studies suggest strongly that differences in antiproliferative mechanisms among α₁-AR antagonists are mainly attributable to their chemical structures.

Tamsulosin and silodosin showed the similar tendency in vivo, that is, a small decline of E9 + PrSC tumor growth. Suppression of E9 + PrSC tumor growth during silodosin treatment might have induced some early apoptosis in PCa cells. Silodosin showed antiproliferative effects on E9 cells through apoptosis induction, whereas the apoptosis-inducing action was modest in comparison with that of quinazoline-derived α₁-AR antagonists (5). On the other hand, tamsulosin did not show antiproliferative effect on either PCa cells or PrSC in vitro, similar to previous reports (5, 6). Tamsulosin induced neither induction of G₁ cell-cycle arrest, nor apoptosis, nor suppression of tumorigenic soluble factors derived from PrSC. It is interesting that these antagonists have shown different antiproliferative mechanisms for androgen low-sensitive E9 cells expressing a_1d-AR subtype but not α_1a-AR subtype.

In regard to treatment for BPH, α₁-AR antagonists are the most widely used therapeutic agents. α₁-AR antagonists are divided into 2 groups on the basis of differential selectivity to α₁-AR; subtype nonselective agents and subtype-selective agents. In Japan, using subtype-selective antagonist is mainstream treatment for BPH because such agents have fewer side effects on the cardiovascular system than subtype nonselective antagonists (32). In this study, we used 3 typical, clinically used, subtype-selective α₁-AR antagonists which were developed in Japan. Tamsulosin, an α_1A-AR and α_1D-AR selective antagonist, is commonly used around the world. Silodosin, a highly selective α_1A-AR antagonist, has been recently approved for clinical use in Japan, Europe, and the United States. Naftopidil, which is only marketed in Japan, has a comparatively higher selectivity for α_1D-AR than the other 2 antagonists. α_1D-selective naftopidil and highly α_1A-selective silodosin showed antiproliferative effects on PCa cells regardless of the androgen sensitivity and the α₁-AR subtype expression in PCa cells in vitro cell culture. These results suggest that neither the androgen receptor nor the α₁-AR signal in PCa cells may be involved in their antiproliferative effects. In clinical aspects, naftopidil has high tolerability with few adverse effects (33, 34). Thus, certain clinical benefit obtained from long-term internal use of naftopidil might be prospective. Naftopidil may be considered for long-term prevention, that is, it is possible that naftopidil could suppress progression to clinical PCa in patients with latent PCa which is concomitant with BPH. In future, further investigation for a prospective study would be necessary to confirm the antitumor effects from oral administration of naftopidil on patient with PCa.

In conclusion, our findings advanced understanding of the effects of subtype-selective α₁-AR antagonists on both PCa and stromal cells. This study provides strong evidence that oral administration of α_1D-AR antagonist naftopidil decreases PCa tumor growth by altering tumor–stroma interactions. Furthermore, the antiproliferative effect of naftopidil is not related to androgen sensitivity of the cells or the α₁-AR subtype expression in PCa cells, suggesting that naftopidil might induce G₁ cell-cycle arrest in various types of cancer cells.

No potential conflicts of interest were disclosed.

We thank Mrs. Hiroko Nishii and Mrs. Yumi Yoshikawa for technical support.

Grants-in-Aid from the Ministry of Education for Science and Culture of Japan, no 22591764 (Y. Sugimura); no 21592039 (K. Arima); no 21791499 (K. Ishii).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.