Antitumor Effects of Thalidomide Analogs in Human Prostate Cancer Xenografts Implanted in Immunodeficient Mice

Prostate carcinoma is the most common cancer and the second leading cause of cancer death in American men. Current standard therapy for advanced prostate cancer involves androgen ablation using luteinizing hormone-releasing hormone agonists with or without an androgen antagonist such as flutamide or bicalutamide. However, the disease eventually becomes androgen independent, at which point therapeutic options are limited. Novel treatment modalities are required to improve clinical outcome.

Angiogenesis, the formation of new blood vessels from preexisting vessels, plays a significant role in solid tumor growth and metastasis (1). In particular, using the transgenic adenocarcinoma of the mouse prostate model, Huss et al. (2) have identified two distinct angiogenic switches in prostate cancer progression. It seems logical to speculate that inhibition of angiogenesis would represent an effective treatment strategy for prostate cancer. Thalidomide has been shown to inhibit basic fibroblast growth factor (bFGF)- and vascular endothelial growth factor (VEGF)-induced angiogenesis (3, 4). The antitumor activity of thalidomide has also been reported in numerous clinical trials (5, 6, 7, 8). In androgen-independent prostate cancer, thalidomide caused a decrease in serum prostate-specific antigen and an improvement of clinical symptoms in 27% of patients (9). The common side effects of thalidomide, including dose-dependent somnolence and dizziness as well as peripheral neuropathy (10), have prompted the development of thalidomide analogs with better pharmacological profiles. Our laboratory previously demonstrated the superior antiangiogenic activity of the N-substituted analog CPS11 and the tetrafluorinated analogs CPS45 and CPS49 in comparison with thalidomide in multiple in vitro assays (11). The current study investigated the in vivo therapeutic efficacy of these analogs in the treatment of prostate cancer.

Thalidomide analogs CPS11, CPS45, and CPS49 were synthesized by Dr. Kurt Eger and his group at the University of Leipzig (Leipzig, Germany). Worldwide patents of these analogs have been filed by the NIH and licensed to Celgene Corp. (Warren, NJ). Their chemical structures have been published previously (11). Thalidomide was obtained from Celgene Corp.

Human prostate cancer cell lines PC3 and 22Rv1 and Lewis lung cancer cell line LLC1 were obtained from the American Type Culture Collection (Manassas, VA). Prostate cancer cells and Lewis lung cancer cells were maintained at 37°C and 5% CO₂ in RPMI 1640 and DMEM, respectively, supplemented with 10% fetal bovine serum and antibiotics (100 units/ml penicillin, 100 μg/ml streptomycin, and 0.25 μg/ml amphotericin B).

All animal experiments were done in accordance with institutional guidelines for animal welfare. PC3 (5 × 10⁶) and 22Rv1 (3 × 10⁶) cells were injected s.c. into 5–6-week-old male severely combined immunodeficient mice. When tumor volume reached ∼150–200 mm³, animals were randomized into five groups (n = 5 each). Each group was treated with i.p. bolus injections of either the drug vehicle (0.5% carboxymethylcellulose), CPS11 (100 mg/kg), CPS45 (100 mg/kg), CPS49 (12.5 mg/kg), or thalidomide (100 mg/kg) 5 days a week for 4 weeks. These doses were the maximum tolerated doses determined in a previous study (11). Tumors were measured with a caliper once a week, and their volumes were calculated using the formula π/6 × a × b², where a is the longest dimension of the tumor, and b is the width. Before euthanasia, the mice were anesthetized with 2% isoflurane. Approximately 800-1000 μl of blood were drawn by cardiac puncture and collected in EDTA-containing microtainer tubes (Becton Dickinson, Franklin Lakes, NJ). Plasma samples were separated by centrifugation and then stored at –80°C for subsequent detection of angiogenic factors. Harvested tumors were snap frozen in OCT (Miles Inc., Elkhart, IN) in liquid nitrogen and subsequently processed for immunohistochemistry.

The Quantikine human VEGF, platelet-derived growth factor (PDGF)-AA, and bFGF ELISA kits (R&D Systems, Minneapolis, MN) were used to determine plasma levels of human VEGF, PDGF-AA, and bFGF, respectively, according to the manufacturer’s instructions.

Five-μm-thick sections obtained from each frozen tumor were stained with H&E for histological examination. For detection of microvessels, sections were stained with the polyclonal anti-CD31/PECAM-1 antibody (1:500; Santa Cruz Biotechnology, Santa Cruz, CA). Antigens were visualized using the streptavidin-biotin-peroxidase method.

Tumor sections stained with anti-CD31/PECAM-1 antibody were examined by light microscopy. Clusters of stained endothelial cells distinct from adjacent microvessels, tumor cells, or other stromal cells were counted as one microvessel. The MVD for each tumor was expressed as the average count of the three most densely stained fields identified using a ×40 objective. Five different tumors per vehicle control or treatment group were analyzed.

RNA was extracted from PC3 and 22Rv1 tumor xenografts harvested from severely combined immunodeficient mice treated with the vehicle (0.5% carboxymethylcellulose), CPS11, CPS45, CPS49, or thalidomide using the RNeasy system (Qiagen, Valencia, CA). Messenger RNAs from each of the vehicle control reference tumors (n = 5) and from each of the drug-treated tumors (n = 5 tumors/treatment group) were converted to cDNA and labeled with cyanine 3-dUTP (Cy3) and cyanine 5-dUTP (Cy5), respectively, using the LabelStar system (Qiagen). Labeled cDNAs from each n = 5 set were pooled, and aliquots were used to interrogate separate oligonucleotide array chips representing >20,000 human genes printed at the Laboratory of Molecular Technology (National Cancer Institute, Frederick, MD). Expression levels for each gene were determined after competitive binding between RNA from vehicle-treated PC3 or 22Rv1 tumors and those from CPS11-, CPS45-, CPS49- or thalidomide-treated PC3 or 22Rv1 tumors. Array slides were scanned with an Axon 4000 scanner (Axon Instruments, Foster City, CA). Data were filtered and normalized using GENEPIX software (Axon Instruments). Genes that were deregulated by ≥1.5-fold relative to the control were considered to be significant.

LLC1 (1 × 10⁶) cells were injected into 5–6-week-old male severely combined immunodeficient mice via the tail vein. On the following day, animals were randomly assigned to five groups (n = 5 each). Each group was treated with i.p. bolus injections of either the drug vehicle (0.5% carboxymethylcellulose), CPS11 (100 mg/kg), CPS45 (100 mg/kg), CPS49 (12.5 mg/kg), or thalidomide (100 mg/kg) 5 days a week for 3 weeks. All animals were euthanized at the end of the treatment period. Lungs were removed, and tumor nodules on the surface of lungs were counted.

All results are presented as mean ± SE. Comparisons were made with one-way ANOVA followed by Dunnett’s test, with P < 0.05 as the criterion for statistical significance.

CPS11, CPS45, and CPS49 significantly inhibited PC3 tumor growth by 90%, 51%, and 68%, respectively, compared with the vehicle control (Fig. 1,A). Thalidomide had no effect (Fig. 1,A). Growth of 22Rv1 tumors was slightly delayed by the analogs but not by thalidomide (Fig. 1,B). No significant reduction in body weight was observed in PC3 tumor-bearing mice treated with the analogs and thalidomide (Fig. 1,C). In 22Rv1 tumor-bearing mice, however, CPS45 significantly decreased body weight by ∼15% compared with the vehicle control (Fig. 1 D).

PC3 and 22Rv1 tumors differentially expressed VEGF, PDGF-AA, and bFGF proteins (Fig. 2, A–C). Neither the analogs nor thalidomide significantly altered VEGF levels in PC3 or 22Rv1 tumors compared with the vehicle control (Fig. 2,A). Interestingly, PDGF-AA levels were significantly reduced by 82%, 67%, and 58% in PC3 tumors treated with CPS45, CPS49, and thalidomide, respectively (Fig. 2,B). All of the analogs and thalidomide had no significant effects on 22Rv1 tumor PDGF-AA levels (Fig. 2,B). PC3 and 22Rv1 tumor bFGF levels were unchanged by the analogs and thalidomide (Fig. 2 C).

DNA microarray data for the VEGF, PDGF, and fibroblast growth factor (FGF) gene families were examined. Screening of transcriptomes sampled from PC3 tumors treated with the analogs and thalidomide revealed a 1.5-fold or greater differential down-regulation of PDGFA, FGF9, FGF14, FGF16, and FGF19 (Table 1). In 22Rv1 tumors, the analogs and thalidomide differentially caused ≥1.5-fold down-regulation of PDGFA, PDGFB, PDGF receptor α, and various isoforms of the FGF and FGF receptor family (Table 1). Transcription of VEGFA, VEGFB, and VEGFC in both PC3 and 22Rv1 tumors was not altered by the analogs or thalidomide (Table 1).

MVD was decreased by 64% in PC3 tumors treated with CPS49 (Fig. 3). In 22Rv1 tumors, CPS45 and CPS49 reduced MVD by 60% and 53%, respectively (Fig. 3). CPS11 and thalidomide did not affect MVD in PC3 and 22Rv1 tumors compared with the vehicle control (Fig. 3).

As shown in Table 2, CPS11 and CPS49 significantly reduced the number of lung metastases by 87% and 67%, respectively. CPS45 and thalidomide failed to do so.

Our results showed that the N-substituted thalidomide analog CPS11 and the tetrafluorinated analogs CPS45 and CPS49, but not thalidomide, suppress the growth of androgen-independent prostate cancer xenografts. It was demonstrated previously (11) that CPS45 and CPS49 inhibit rat aortic microvessel outgrowth as well as human umbilical vein endothelial cell proliferation and tube formation more potently than CPS11. However, CPS11 was observed to suppress prostate tumor growth to a greater extent than CPS45 and CPS49 and even to induce PC3 tumor regression in the current study. It was speculated that CPS11 may have a better pharmacokinetic profile or multiple pharmacological targets. The minimal antitumor effects of these analogs observed in the 22Rv1 xenografts remain elusive.

To understand the mechanism(s) of action underlying the in vivo antitumor effects of these analogs, tumor levels of the angiogenic factors VEGF, PDGF-AA, and bFGF were determined. It is well documented that VEGF and bFGF are potent stimulators of endothelial cell proliferation and play a significant role in tumor angiogenesis. All of the thalidomide analogs and thalidomide failed to reduce tumor VEGF and bFGF protein levels, consistent with microarray analyses that indicated their lack of effect on VEGFA and FGF2 (bFGF) gene transcription. The exception was CPS49, which caused a 1.5-fold down-regulation of the FGF2 gene but did not reduce bFGF protein levels in the 22Rv1 tumors. Interestingly, PC3 tumor PDGF-AA levels were decreased by CPS45, CPS49, and thalidomide, but not by CPS11, although microarray experiments revealed that the PDGFA gene was significantly down-regulated in response to CPS11 and CPS45, but not CPS49 and thalidomide. The analogs and thalidomide did not alter PDGF-AA levels in 22Rv1 tumors, despite reduced PDGFA gene transcription by CPS11, CPS45, and thalidomide. It should be noted that PC3 and 22Rv1 prostate cancer cell lines/xenografts have distinct genetic characteristics (e.g., androgen sensitivity, androgen receptor expression or lack thereof, and p53 status) that may differentially influence their responses to the analogs and thalidomide (12). Posttranscriptional modulation of PDGF-AA by the analogs and thalidomide may also contribute to the discrepancies between the ELISA and microarray data. For instance, the possibility that CPS11 may keep PDGF-AA protein levels high in the face of reduced PDGFA gene transcription by inhibiting enzymes involved in protein degradation or that CPS49 may reduce PDGF-AA protein levels without down-regulating the PDGFA gene by promoting protein degradation cannot be excluded. Previous studies reported that many tumors express PDGF and cognate receptors, suggesting an autocrine pathway for stimulation of tumor cell growth (13). In particular, PDGF-AA, PDGF receptor α, and PDGF receptor β proteins were detected in prostatic intraepithelial neoplasias and adenocarcinomas (14, 15). Furthermore, PDGF was shown to have proangiogenic effects (16, 17, 18). It was thus hypothesized that these thalidomide analogs inhibit prostate cancer growth in part via modulation of PDGF signaling. In accordance with this hypothesis, intratumoral MVD was significantly decreased in response to CPS45 and CPS49 treatment, suggesting that the antitumor effects of these tetrafluorinated analogs may be associated with the inhibition of PDGF-mediated angiogenesis. The observation that CPS11 exhibited a potent antitumor effect without concurrently decreasing the intratumoral levels of major angiogenesis factors and MVD is intriguing. It is possible that the N-substituted analog affects non-angiogenesis-related molecular pathways in addition to the angiogenic pathways targeted by the tetrafluorinated analogs, perhaps indicating a broader range of anticancer profile of the former. The significant reduction in the number of tumor nodules induced by CPS11 and CPS49 in the Lewis lung model indicated that these analogs also possess antimetastatic activity. Other thalidomide analogs reported in the literature have also demonstrated antiangiogenic and/or antitumor properties (19, 20). One study (20) showed that certain analogs promote cell cycle arrest and increase the expression of proapoptotic proteins and decrease that of antiapoptotic proteins. However, it should be noted that different thalidomide analogs and thalidomide itself may bind to distinct target protein(s).

In summary, the present data demonstrated the antitumor effects of thalidomide analogs in human prostate cancer xenografts. One possible mechanism of action of these analogs may involve inhibition of the PDGF signaling pathway. Detailed transcriptome-wide microarray analyses are under way with the aim of deciphering the precise mechanisms of action of these analogs and explaining the differential effects of these analogs on PC3 and 22Rv1 prostate tumors. Full toxicology and teratogenicity testings are in progress before clinical development.

We are grateful to Dr. Miriam R. Anver and Keith Rogers at the Laboratory of Histopathology and to Dr. Lisa Gangi and Sally Hobaugh at the Laboratory of Molecular Technology, National Cancer Institute-Frederick, for technical support and expertise.