Inhibition of prostate cancer osteoblastic progression with VEGF₁₂₁/rGel, a single agent targeting osteoblasts, osteoclasts, and tumor neovasculature

Abstract

Purpose

A hallmark of prostate cancer (PCa) progression is the development of osteoblastic bone metastases, which respond poorly to available therapies. We previously reported that VEGF₁₂₁/rGel targets osteoclast precursors and tumor neovasculature. Here we tested the hypothesis that targeting non-tumor cells expressing these receptors can inhibit tumor progression in a clinically relevant model of osteoblastic PCa.

Experimental Design

Cells from MDA PCa 118b, a PCa xenograft obtained from a bone metastasis in a patient with castrate-resistant PCa, were injected into the femurs of mice. Osteoblastic progression was monitored following systemic administration of VEGF₁₂₁/rGel.

Results

VEGF₁₂₁/rGel was cytotoxic in vitro to osteoblast precursor cells. This cytotoxicity was specific as VEGF₁₂₁/rGel internalization into osteoblasts was VEGF₁₂₁ receptor driven. Furthermore, VEGF₁₂₁/rGel significantly inhibited PCa-induced bone formation in a mouse calvaria culture assay. In vivo, VEGF₁₂₁/rGel significantly inhibited the osteoblastic progression of PCa cells in the femurs of nude mice. Microcomputed tomography analysis revealed that VEGF₁₂₁/rGel restored the bone volume fraction of tumor-bearing femurs to values similar to those of the contralateral (non–tumor bearing) femurs. VEGF₁₂₁/rGel significantly reduced the number of tumor-associated osteoclasts but did not change the numbers of peritumoral osteoblasts. Importantly, VEGF₁₂₁/rGel-treated mice had significantly less tumor burden than control mice. Our results thus indicate that VEGF₁₂₁/rGel inhibits osteoblastic tumor progression by targeting angiogenesis, osteoclastogenesis, and bone formation.

Conclusions

Targeting VEGFR-1 – or VEGFR-2–expressing cells is effective in controlling the osteoblastic progression of PCa in bone. These findings provide the basis for an effective multitargeted approach for metastatic PCa.

Introduction

Bone metastases are prevalent (90%) in patients with advanced prostate cancer (PCa) (1) and are a major cause of mortality and morbidity (2). Indeed, currently no curative therapy is available for men with PCa bone metastases, and only a modest survival advantage is achieved with chemotherapy. Skeletal metastases of PCa are unique in that they consistently produce bone-forming lesions, although an osteolytic component is also present (3–7). This high tropism for bone and the consistent osteoblastic phenotype suggests that PCa cells interact with bone and that this interaction influences the progression of the disease. Thus, bone-targeted therapies have been explored by many investigators as a new avenue for treating the disease.

We previously reported that VEGF₁₂₁/rGel, a vascular endothelial growth factor (VEGF) fusion construct composed of human VEGF₁₂₁ and the highly cytotoxic plant toxin gelonin (rGel), had efficacy against a model of prostate tumor growth in bone that had predominantly osteolytic features. We also found that VEGF₁₂₁/rGel targets mVEGFR-1⁺ osteoclast precursor cells in vitro and reduces the number of mature osteoclasts at the tumor–bone interface in vivo (8). However, although osteoclast-targeted therapies are effective in controlling bone-related complications, clinical trials have failed to show a survival benefit in men with PCa bone metastases (9).

Numerous investigators have suggested that VEGF-A has a direct role in bone vascularization and formation during normal bone development (10–14), bone repair (12;13;15), and the bone-remodeling process that takes place in PCa skeletal metastasis (16;17). Indeed, the metastases of cancer cells to bone are known to alter bone architecture and mineral homeostasis, and tumor cells in the bone marrow generate a number of cytokines, including VEGF, that can directly affect the proliferation and maturation of osteoclasts, osteoblasts, and their precursors, markedly affecting bone remodeling (1;18;19). VEGF₁₂₁/rGel binds selectively to the VEGF receptors VEGFR-1 (Flt-1/FLT-1) and VEGFR-2 (Flk-1/KDR), which are normally overexpressed on the vasculature of most solid tumors (20–22), resulting in a reduction of angiogenesis (23–25). But again, clinical trials with antiangiogenic therapies failed to show a survival benefit in men with PCa bone metastases.

Thus, this work was undertaken to gain insight into the potential clinical relevance of the use of VEGF₁₂₁/rGel in men with PCa bone metastases. To that end, we evaluated the antitumor effect of VEGF₁₂₁/rGel in an osteoblastic model of castrate-resistant PCa growing in bone (MDA PCa 118 xenograft) (26). We also assessed whether VEGF₁₂₁/rGel had a direct effect on PCa-induced bone formation. Our findings suggest that osteoblast precursors are also a target of VEGF₁₂₁/rGel and that simultaneously targeting several cellular components of the bone microenvironment (including osteoblasts) will be an important therapeutic modality for controlling PCa growth in bone.

Materials and Methods

Cell Culture

Porcine aortic endothelial (PAE) cells transfected with the human VEGFR-2 receptor (PAE/KDR) and PAEs transfected with the human VEGFR-1 receptor (PAE/FLT-1) were a generous gift from Dr. Johannes Waltenberger (University Hospital, Maastricht, The Netherlands) and were propagated as previously described (27). The number of VEGFR-2 and VEGFR-1 receptor sites on these cells have been determined to be 150,000 and 50,000 per cell, respectively (28). Murine brain endothelial cells (bEnd.3) and human umbilical vein endothelial cells (HUVECs) were kind gifts from Dr. Sophia Ran (Southern Illinois University, Springfield, IL). Human PCa cell lines PC-3, LNCaP and C4-2B, and the mouse preosteoblast cell line MC3T3-E1 were purchased from the American Type Culture Collection (Manassas, VA). Differentiation of confluent MC3T3 cells was performed as previously described (29) and confirmed by Alizarin Red S staining (30). Primary mouse osteoblasts (PMOs) were obtained from CD1 mice as reported previously (29). MDA PCa 2b (31) and MDA PCa 118b (26) cell lines are a bone-derived PCa cell line and xenograft, respectively, established in Dr. Nora Navone’s laboratory. MDA PCa 118b cells were maintained in vivo by subcutaneous passage in immunodeficient mice (26).

Animals

Male athymic Balb/c nude mice (National Cancer Institute, Frederick, MD) were maintained under specific pathogen–free conditions according to the American Association for Accreditation of LaboratoryAnimal Care (AAALAC) standards.

Purification of VEGF₁₂₁/rGel

VEGF₁₂₁/rGel construction and purification were performed essentially as previously described (25), followed by SP Sepharose chromatography (pH 6.0) with a NaCl gradient to separate the biologically active dimeric form from other species. VEGF₁₂₁/rGel was concentrated and stored in sterile PBS at −20 °C.

Cytotoxicity and Internalization of VEGF₁₂₁/rGel and rGel

Cytotoxicity of VEGF₁₂₁/rGel, rGel, and VEGF₁₂₁ against log-phase MC3T3, PMO, MDA PCa 2b, and C4-2B cells was evaluated over 72 h as previously described (25). Cytotoxicity against 50,000 or 100,000 MDA PCa 118b cells was evaluated in short-term cultures obtained from subcutaneous tumors and grown over 72 h in CnT52 medium. Cytotoxicity against differentiated MC3T3 cells was evaluated after the cells were grown in differentiation conditions for 1, 2, or 3 weeks. For internalization, PMOs were treated with 4 μg/ml (48 nM) VEGF₁₂₁/rGel for 24 h and then washed with glycine buffer (500 mM NaCl, 0.1 M glycine, pH 2.5) to remove cell surface–bound VEGF₁₂₁/rGel. Cells were incubated with a rabbit anti-gelonin polyclonal antibody (1:200) followed by a FITC-conjugated anti-rabbit secondary antibody (1:80). Nuclei were stained with propidium iodide (1 μg/ml) in PBS. The cells were mounted on slides with 1,4-diazabicyclo[2.2.2]octane and visualized under a fluorescence microscope (Nikon Eclipse TS1000).

RNA Extraction

Total RNA was extracted using an RNeasy mini-kit (Qiagen, Valencia, CA), and its integrity was verified by electrophoresis on a denaturing formaldehyde-agarose gel and on a 2100 Bioanalyzer (Agilent, Foster City, CA).

Polymerase Chain Reaction (PCR) and Reverse Transcription (RT)-PCR Analysis

Relative levels of VEGFR-1, VEGFR-2, and VEGF-A transcript were assessed by RT-PCR analysis. GAPDH primers were used as controls (24). mVEGF transcript was detected using primers previously described (8). We utilized primers for VEGFR-1 and VEGFR-2 that recognize conserved sequences in both mouse and human cDNA (24), and also generated the following primers unique to mouse VEGFR: M R2 forward − 5′ TCTGTGGTTCTGCGTGGAGA; M R2 reverse − 5′ GTATCATTTCCAACCACCCT; M R1 forward − 5′ CTCTGATGGTGATCGTGG; M R1 reverse − 5′ CATGCCTCTGGCCACTTG. Amplified RT-PCR products were subjected to densitometric analysis using FluorChem8900 (Alpha Innotech, San Leandro, CA).

Organ Culture Bone-Formation Assay

Calvariae from 4-day-old CD1 mice (Charles River Laboratories International, Inc., Wilmington, MA) were excised, cut in half, and cultured for 7 d in six-well plates as previously described (26). Briefly, calvariae were placed on a metal grid bathed in BGJ medium (Sigma-Aldrich, St. Louis, MO) with 0.1% bovine serum albumin in the presence and absence of 100 nM VEGF₁₂₁/rGel or MDA PCa 2b cells cultured in the wells. The medium was changed on day 3, and VEGF₁₂₁/rGel was supplemented as appropriate. At the end of culturing, the calvariae halves were processed as previously described (26). Sections were also stained with hematoxylin and eosin (H&E). Histomorphometric analysis was performed by the Bone Histomorphometry Core facility at The University of Texas MD Anderson Cancer Center (M.W.S). Osteo II software version 8.40.20 (Bioquant, Nashville, TN) was used to measure the ratio of osteoid volume to bone volume, the ratio of osteoid surface to bone surface, and the osteoid area. Measurements of all samples were obtained approximately 160 μm from the frontal suture of the calvaria for a distance of 2 mm. The osteoid covering the bone was measured first, followed by the total bone surface; the degree of bone surface covered by osteoid was expressed as the ratio of osteoid surface area to bone surface area. All experiments were performed in triplicate.

Intrabone Injections and Bone Tissue Sample Processing

MDA PCa 118b tumors growing subcutaneously in mice were harvested as previously described (26), washed, and resuspended in PBS in preparation for implantation into the mice. Nu/nu male mice (5–6 weeks old, 10 mice per group) were anesthetized with intramuscular injections of ketamine (100 mg/kg) plus acepromazine (2.5 mg/kg). Aliquots of 1 × 10⁶ of 118b cells were diluted in 5 μl of growth medium and then injected into the distal epiphysis of the right femur of each mouse using a 28-gauge Hamilton needle as previously described (26). The contralateral femur was used as an internal control. Twenty mice were randomized to receive intravenous injections of either saline or VEGF₁₂₁/rGel (14 mg/kg) every other day for 9 d. Treatment began 1 week after tumor placement. Treatment was stopped after five cycles, and the mice were then monitored radiographically once a week for tumor bulk and new bone formation, with no further treatment. Mice were euthanized 8 weeks after tumor placement. Femurs bearing MDA PCa 118b tumors and the contralateral non–tumor-bearing femurs were resected, fixed in ethanol, and subjected to micro computed tomography (μCT) analysis to assess bone mass, as previously described (26). MicroCT was performed in the Small Animal Imaging Facility at MD Anderson Cancer Center. The ratio of bone volume to total volume was calculated as previously described (26). Fixed specimens were subsequently decalcified, embedded in paraffin, and sectioned as previously described (29). Histopathologic analysis included H&E, toluidine blue (osteoblasts) and tartrate-resistant acid phosphatase (TRAP) staining (osteoclasts) (26;32). The total tumor content in bone samples was determined as previously described (26). Overall tumor burden was defined as the sum of tumor soft tissue and mineralized tissue.

Statistical Analysis

All statistical analyses were done with Microsoft Excel software (Microsoft, Redmond, WA). Data are presented as means ± SEM. P values were obtained using the 2-tailed t test with 95% confidence intervals to evaluate statistical significance; P < 0.05 was considered statistically significant.

Results

PMOs and the Mouse Preosteoblastic Cell Line MC3T3 Express VEGFR-1, whereas PCa Cells Express Little or No VEGFR

RT-PCR analysis revealed that MC3T3 cells expressed high levels of VEGFR-1 mRNA but did not express the VEGFR-2 transcript (Fig. 1A), suggesting that osteoblasts are candidates for targeting by VEGF₁₂₁/rGel.

Figure 1.

(A–C) RT-PCR analysis. VEGFR primers that recognize conserved sequences in both mouse and human cDNA that result in an identical length of transcript were used. (A) LNCaP and MDA PCa 2b human prostate cancer cells showed no PCR product for VEGFR-1 or VEGFR-2. PC-3 cells expressed low levels of transcript for both VEGFR-1 and VEGFR-2. Primary mouse osteoblasts (PMOs) and MC3T3 cells expressed VEGFR-1 but not VEGFR-2. mRNA transcript from cells expressing human VEGFR-1 (PAE/FLT-1) and VEGFR-2 (PAE/KDR) were used as controls. (B) MDA PCa 118b tumor tissue did not express VEGFR-1 or VEGFR-2. Murine bEnd.3 cells and human umbilical vein endothelial cells (HUVECs) were used as controls. GAPDH was utilized as a loading control. (C) VEGFR-1 mRNA was down-regulated during MC3T3 cell differentiation. VEGFR-1, VEGFR-2, and VEGF-A levels are shown. GAPDH primers were used as controls. Low levels of VEGF₁₆₄ and VEGF₁₂₀ transcripts were also detected. (D–E) VEGF₁₂₁/rGel is specifically targeted to primary mouse osteoblasts (PMOs). (D) VEGF₁₂₁/rGel, rGel, and VEGF₁₂₁ cytotoxicity on PMOs over 72 h. Each experiment was performed in triplicate. (E) VEGF₁₂₁/rGel internalization into PMOs was driven by VEGF₁₂₁. PMOs were treated with either 10 nM VEGF₁₂₁/rGel or the untargeted toxin rGel for 24 h. Only nuclei were visible in rGel-treated PMOs whereas fluorescent rGel staining was observed in the cytoplasm of VEGF₁₂₁/rGel-treated PMOs (arrows).

We also assessed the expression of VEGFRs in human PCa cells known to induce osteolytic or osteoblastic lesions in vivo and found that the osteolytic PC-3 PCa cell line expressed low levels of both VEGFR-1 and VEGFR-2. However, no VEGFR-1 or VEGFR-2 transcripts were detected in two other PCa cell lines, LNCaP and MDA PCa 2b (Fig. 1A). VEGFR was also not detected in mRNA harvested from MDA PCa 118b tumor tissue (Fig. 1B).

VEGFR-1 Transcript Levels Decrease during Osteoblast Differentiation

We previously showed that VEGFR-1 levels decrease in osteoclast precursor cells at onset of differentiation (8). We assessed whether VEGF receptors on osteoblast precursors undergo a similar fate. RT-PCR analysis revealed that MC3T3 cells exhibited a gradual down-regulation of the VEGFR-1 transcript during differentiation (Fig. 1C). RT-PCR analysis also revealed low levels of VEGF₁₆₄ and VEGF₁₂₀ murine isoforms but no VEGF₁₈₈ isoform, suggesting that these precursors promote angiogenesis or mitogenesis in the tumor microenvironment before differentiation.

VEGF₁₂₁/rGel Has Cytotoxic Effects on Osteoblasts but Not on PCa Cells

To determine whether VEGF₁₂₁/rGel could directly target osteoblasts, we evaluated the effect of VEGF₁₂₁/rGel on PMOs and MC3T3 cells grown in differentiation medium for 0–3 weeks. The 50% inhibitory concentration (IC₅₀) of VEGF₁₂₁/rGel on the PMOs was 15 nM, whereas the IC₅₀ of rGel alone was 200 nM, indicating that the cytotoxicity of VEGF₁₂₁/rGel was mediated through VEGF₁₂₁ (Fig. 1D). This IC₅₀ is similar to that previously reported for mouse osteoclast precursor cells and bone marrow monocytes (8). As expected, we found that MC3T3 sensitivity to VEGF₁₂₁/rGel was significantly reduced when the cells were grown in differentiation medium (Table 1), a finding that matched the PCR data showing down-regulation of VEGFR-1 during differentiation. Confirming the absence of VEGF receptors that bind to VEGF₁₂₁/rGel, we observed no specific cytotoxicity of this construct on the MDA PCa 2b, C4-2B, or MDA PCa 118b PCa cells (Table 1).

Table 1.

VEGF₁₂₁/rGel and rGel cytotoxicity on MC3T3 cells and prostate cancer cell lines

Cell Line	Cell Type	Status	IC50 (nM), VEGF121/rGel	IC50 (nM), rGel
MC3T3	Mouse preosteoblast	Log phase	30	100
MC3T3	Mouse preosteoblast	Differentiated, 1 week	500	500
MC3T3	Mouse preosteoblast	Differentiated, 2 weeks	> 1000	> 1000
MC3T3	Mouse preosteoblast	Differentiated, 3 weeks	> 1000	> 1000
MDA PCa 2b	Human Prostate	Log phase	290	1150
C4-2B	Human Prostate	Log phase	> 1000	> 1000
MDA PCa 118b	Human Prostate	Log phase	> 1000	> 1000
PC-3	Human Prostate	Log phase	225*	100*

^*Veenendaal et al.(25)

VEGF₁₂₁/rGel Internalization into PMOs Is Driven by VEGF₁₂₁

Immunostaining revealed that VEGF₁₂₁/rGel, but not rGel alone, localized in the cytoplasm of PMOs, which suggests that VEGF₁₂₁ mediates rGel internalization (Fig. 1E).

VEGF₁₂₁/rGel Inhibits PCa-Mediated Bone Formation in Neonatal Mouse Calvariae

Organ-culture assays revealed that calvariae treated with medium alone did not show new bone formation, whereas coculturing them with MDA PCa 2b cells stimulated new bone formation (Fig. 2A, upper panel). The addition of 100 nM VEGF₁₂₁/rGel inhibited MDA PCa 2b–cell induced new bone formation (Fig. 2A, lower panel), whereas treatment with VEGF₁₂₁/rGel alone did not have a significant effect on new bone formation (Supplemental Figure 1). Histomorphometric analysis of undecalcified mouse calvariae revealed that VEGF₁₂₁/rGel inhibited PCa-induced new bone formation, as assessed by osteoid-related parameters. The ratio of osteoid volume to bone volume normalized to control decreased by 63% in the presence of VEGF₁₂₁/rGel (P < 0.05; double-sided t test; Fig. 2B). Similarly, the ratio of osteoid surface to bone surface decreased by 54% (Fig. 2B), and the total osteoid area decreased by 55% (P < 0.075; double-sided t test; Fig. 2B). These results suggest that VEGF₁₂₁/rGel can block PCa-induced new bone formation.

Figure 2.

(A) VEGF₁₂₁/rGel inhibited prostate cancer–induced bone formation. Upper panel: representative neonatal mouse calvariae cultured in vitro in the absence (left) or presence (right) of MDA PCa 2b cells. Lower panel: representative neonatal mouse calvariae cocultured with MDA PCa 2b cells in the absence (left) and presence (right) of 100 nM VEGF₁₂₁/rGel. Calcified bone matrix and newly formed osteoid are indicated. (B) Quantification of VEGF₁₂₁/rGel’s effect on new bone formation. VEGF₁₂₁/rGel significantly decreased the ratio of osteoid volume to bone volume (left panel) primarily by reducing osteoid surface (middle panel) and area (right panel).

VEGF₁₂₁/rGel Inhibits the Osteoblastic Growth of MDA PCa 118b Cells in Bone

Eight weeks after MDA PCa 118b tumor-bearing mice were injected with either saline or VEGF₁₂₁/rGel, 90% (9 of 10) of the saline-treated mice developed robust osteoblastic lesions in their right femurs, an indication of tumor growth (Fig. 3A). In contrast, only 25% (2 of 8) of the VEGF₁₂₁/rGel–treated mice developed areas of osteoblastic reaction in their right femurs. The areas of increased bone density in the treated group were not associated with major alterations in the architecture of the epiphysis (Fig. 3A), indicating a significant reduction of tumor burden in the VEGF₁₂₁/rGel–treated mice. Accordingly, μCT analysis revealed that the total bone volume in the femurs injected with MDA PCa 118b cells was significantly higher in the saline-treated mice than in the VEGF₁₂₁/rGel–treated mice (P = 0.011; two-tailed t test) (Fig. 3B and C) and that the total bone volume of the femurs injected with MDA PCa 118b cells was essentially the same as the total bone volume of the contralateral femurs in the VEGF₁₂₁/rGel–treated mice (Fig. 3C). MicroCT analysis of a 1-mm cortical segment of each bone’s midshaft revealed that the femurs in the VEGF₁₂₁/rGel–treated mice had higher bone mineral density than did those of the saline-treated mice (P < 0.005; two-tailed t test) (Fig. 3D). We found no significant difference in the overall bone mineral content between the saline- and VEGF₁₂₁/rGel–treated mice (P < 0.25; two-tailed t test; Fig. 3E), suggesting that the lower bone mineral density in the saline-treated mice was due to the distribution of the bone mineral content over a larger volume.

Figure 3.

VEGF₁₂₁/rGel inhibited osteoblastic growth of MDA PCa 118b cells in bone. (A) Radiographic analysis of saline- and VEGF₁₂₁/rGel–treated mice. Radiographs show right femurs 8 weeks after implantation of MDA PCa 118b cells. Arrows indicate the increased bone density in the saline-treated mice (left panels) and significantly lower bone density in the VEGF₁₂₁/rGel–treated mice (right panels). (B) Representative microcomputed tomography (μCT) images reveal higher bone mass in the tumor-bearing femurs of saline-treated mice (left panels) than in the tumor-bearing femurs of VEGF₁₂₁/rGel–treated mice (right panels). (C) Quantification of μCT data revealed that VEGF₁₂₁/rGel normalized the bone volume fraction of 118b tumor-containing bone, as compared with the contralateral bone in the VEGF₁₂₁/rGel-treated mice. (D) VEGF₁₂₁/rGel treatment increased bone mineral density and reduced overall bone volume. (E) However, VEGF₁₂₁/rGel treatment did not significantly affect overall bone mineral content.

VEGF₁₂₁/rGel Reduces Osteoblastic PCa Tumor Burden

Histopathologic analysis of the femurs injected with MDA PCa 118b cells in the saline- and VEGF₁₂₁/rGel–treated mice confirmed the presence of osteoblastic lesions in the tumor-bearing legs of all saline-treated mice (Fig. 4A). In contrast, osteoblastic growth of MDA PCa 118b cells was severely impaired in the majority of femurs of the VEGF₁₂₁/rGel–treated mice (Figs. 4B–E). Only isolated pockets of MDA PCa 118b cells were visible in some areas of the right femurs in the VEGF₁₂₁/rGel–treated mice (Fig. 4C). It was notable that no tumor cells were visible in the bone shafts of two VEGF₁₂₁/rGel–treated mice (Figs. 4D and E). None of the mice’s contralateral legs showed evidence of osteoblastic lesions (Fig. 4F).

Figure 4.

VEGF₁₂₁/rGel inhibited osteoblastic growth of MDA PCa 118b cells in bone. (A–F) H&E staining of MDA PCa 118b tumors in the femurs of nude mice. T, prostate cancer cells; M, bone matrix. Insets represent Faxitron images of bone specimens at the time of harvest. (A) Representative femur from a 118b tumor–bearing leg of a saline-treated mouse. (B–E) Tumor-bearing femurs from VEGF₁₂₁/rGel–treated mice with varying levels of tumor growth. (B) Visible, but reduced, osteoblastic growth of tumor cells. (C) Isolated pocket of tumor cells in the diaphyseal region of a femur, exhibiting limited osteoblastic growth. (D, E) Tumor-free femurs from two VEGF₁₂₁/rGel–treated mice. (F) Contralateral leg. No tumors were observed in the contralateral legs of any mice in the study. (G) Quantitative analysis of tumor soft tissue as a percentage of total bone volume showed a significant reduction of prostate cancer cells in the femurs of VEGF₁₂₁/rGel–treated mice relative to those in the saline-treated mice. (H) Overall tumor burden (tumor soft tissue + new bone matrix) was also significantly reduced in VEGF₁₂₁/rGel–treated mice relative to that in saline-treated mice.

To further understand the effect of VEGF₁₂₁/rGel treatment on tumor growth, we quantified the total tumor burden in bone. The percentage of tumor soft tissue (i.e., tumor cells and stroma) in bone as a function of overall bone volume was significantly reduced in VEGF₁₂₁/rGel–treated mice relative to that in saline-treated mice (mean values, 18.8% vs 33.1%; P < 0.05, two-tailed t test; overall reduction, 43.2%) (Fig. 4G). Similar results were observed for the overall tumor burden (i.e., tumor soft tissue and new bone matrix), suggesting that reduction in overall tumor burden may be directly correlated to the reduction in tumor cells (mean values, 36.7% vs 52.9%; P < 0.05, two-tailed t test; overall reduction, 30.6%) (Fig. 4H). Thus, VEGF₁₂₁/rGel statistically significantly prevented tumor growth in bone.

VEGF₁₂₁/rGel Reduces the Number of Osteoclasts at the Tumor–Bone Interface but Not the Number of Peritumoral Osteoblasts

As an indirect measure of the effect of VEGF₁₂₁/rGel treatment on osteoblast proliferation, we assessed osteoblast numbers at the tumor–bone interface and in the peritumoral space on histologic samples stained with toluidine blue. Osteoblasts were easily identified on the bone surface of morphologically normal bone, whether contralateral or VEGF₁₂₁/rGel-treated bones, in which no tumor was visible (Fig. 5A). As previously reported, MDA PCa 118b cells growing in bone demonstrated an increased number of osteoblasts in the tumor-bearing legs relative to that in the contralateral normal femur (Figs. 5B). Osteoblasts in the vicinity of surviving tumor cells in VEGF₁₂₁/rGel-treated mice were also observed (Fig. 5C). It was interesting that we counted similar numbers of peritumoral osteoblasts in both saline- and VEGF₁₂₁/rGel-treated specimens (Fig. 5D). Because most PCa osteoblastic lesions have an osteolytic component, we investigated the incidence of multinucleated, TRAP-positive osteoclasts. Almost all bone specimens from the saline-treated mice had TRAP-positive osteoclasts prominently lining the tumor–bone interface (Fig. 5E). Both osteoblasts and osteoclasts were identified near tumor cells that were surrounded by bone matrix (Fig. 5F). However, treatment of MDA PCa 118b tumor–bearing mice with VEGF₁₂₁/rGel dramatically reduced the number of osteoclasts lining the tumor–bone interface (Fig. 5G). Fewer osteoclasts were observed near tumor cells surrounded by bone matrix in VEGF₁₂₁/rGel-treated mice, whereas osteoblast numbers were unchanged (Fig. 5G and 5H). Quantification of the overall number of osteoclasts revealed a significant reduction (54%; P < 0.05, two-tailed t test) in VEGF₁₂₁/rGel–treated specimens relative to that in the saline-treated mice (Fig. 5I).

Figure 5.

VEGF₁₂₁/rGel reduced the number of osteoclasts at the tumor–bone interface but did not reduce peritumoral osteoblasts. T, prostate cancer cells; M, bone matrix; black arrows, osteoblasts; white arrows, osteoclasts. (A) Toluidine blue staining of contralateral (top) and tumor-bearing (bottom) femurs shows morphologically normal bone tissue with visible osteoblasts. Original magnification, ×200. (B, C) Toluidine blue staining of tumor cells that are surrounded by bone matrix: (B) control, (C) VEGF₁₂₁/rGel. (D) Quantitation of peritumoral osteoblasts per field. (E) TRAP-positive osteoclasts are present throughout the tumor–bone interface in control mice. Original magnification, ×100; inset original magnification, ×200. (F) TRAP (left) and toluidine blue (right) staining of identical specimens from control mice indicate the presence of both osteoclasts and osteoblasts in the vicinity of tumor cells that are surrounded by bone matrix. Original magnification, ×200. (G) TRAP and (H) toluidine blue staining of identical specimens from VEGF₁₂₁/rGel-treated mice. Fewer osteoclasts line the tumor–bone interface or are present in the vicinity of tumor cells surrounded by bone matrix, while osteoblast numbers are unchanged. Original magnification, ×100. (I) Quantification of osteoclasts at the tumor–bone interface revealed that VEGF₁₂₁/rGel significantly suppressed osteoclast numbers compared to controls. Osteoclast numbers in the contralateral leg are provided as a reference.

Discussion

In this study, we obtained evidence that targeting neovascularization, osteoblasts, and osteoclasts effectively controls the osteoblastic progression of PCa cells in bone. Briefly, VEGF₁₂₁/rGel a) specifically targeted osteoblast precursors in vitro, b) had no cytotoxic effect against PCa cells in vitro, c) inhibited PCa-induced new bone formation in an organ culture assay, d) blocked osteoblastic PCa growth in vivo, and e) significantly reduced the number of tumor-associated osteoclasts.

These findings implicate VEGF and its receptors in the progression of PCa metastasis (16;33–36) and suggest that tumor-induced bone remodeling is a central step in skeletal growth. In addition to its angiogenesis-regulating function in bone formation, VEGFR-1 is involved in recruiting osteoclast precursors to the site of bone resorption and osteoclastogenesis (37–39). VEGFR-1 has also been implicated in the maintenance of bone marrow functions in op/op mice, with tyrosine kinase–deficient VEGFR-1 and reduced numbers of osteoclasts and osteoblasts (40). Kitagawa et al. (17) reported that mouse osteoblasts express VEGFR-1 and neuropilin-1 and demonstrate activity of PTK787, a tyrosine kinase inhibitor that binds to the ATP binding sites of VEGFRs, against PCa-induced osteoblastic lesions in bone. Otsuka et al. (41) demonstrated that administering bevacizumab to mice bearing experimental bone metastases reduced the tumor-induced formation of osteoblastic lesions. No TRAP-positive osteoclasts were observed in these lesions, even though the osteoblastic bone metastases were smaller in the bevacizumab-treated mice than they were in control mice. These studies’ findings suggest that VEGF receptors not targeted by VEGF₁₂₁ are also involved in the development and progression of osteoblastic and osteolytic lesions and are consistent with our finding that VEGF plays a role in this process.

We previously reported that VEGF₁₂₁/rGel readily targets CD11b⁺VEGFR-1⁺ osteoclast precursor cells, which are derived from bone marrow, in vitro. VEGF₁₂₁/rGel may target other VEGFR-1⁺ or VEGFR-2⁺ cells that may play a role in tumor growth in bone. A subset of VEGFR-expressing, bone marrow–derived cells—primarily VEGFR-1⁺ hematopoietic progenitor cells and VEGFR-2⁺ circulating endothelial progenitor cells—have been shown to migrate from the bone marrow to metastasis sites in organs, where they create microenvironments conducive to the efficient development of secondary tumors (42). Indeed, Erler et al. (43) have shown that CD11b⁺ cell recruitment is necessary for the formation of a premetastatic niche and subsequent metastasis of MDA-MB-231 and 4T1/luc breast cancer cells. Other investigators (44;45) have shown that the recruitment of CD11b⁺VEGFR-1⁺ cells of various lineages plays a role in the establishment and growth of metastases. Studies to identify other VEGFR-1⁺/VEGFR-2⁺ cell populations that may play a role in tumor growth in bone are currently under way in our laboratory.

The process of metastasis involves a sequential series of events, including tumor cell entry into the circulation, arrest in the capillary beds of distant organs, extravasation, and proliferation within the organ parenchyma (46). One caveat of the MDA PCa 118b intrabone-injection system that we used in this study is that it does not recapitulate the entire process of bone metastasis. However, it does adequately recapitulate the process of PCa progression in bone, and thus, the results from our studies are relevant to the effect of VEGF₁₂₁/rGel in established bone metastases of PCa.

We previously reported that VEGF₁₂₁/rGel inhibits angiogenesis (23–25) and reduces the number of osteoclasts at the tumor–bone interface in a PCa model of osteolytic bone growth (8). In this study, we found that VEGF₁₂₁/rGel also targets osteoblast precursors. The reduction in overall tumor burden in VEGF₁₂₁/rGel-treated mice corresponded to a reduction in the overall number of osteoclasts. However, we did not find a reduction in the numbers of peritumoral osteoblasts in treated bone specimens relative to those in controls. These findings suggest that VEGF₁₂₁/rGel affects primarily osteoblast function (including osteoclast activation) rather than osteoblast numbers in vivo. These results may also indicate that although targeting osteoblast precursors may affect the initial stages of tumor development, surviving tumor cells retain the ability to recruit osteoblasts effectively on cessation of treatment, which occurred 6 weeks prior to tissue harvest. In summary, our study indicates that the survival advantage of PCa cells was marginalized by the cytotoxic effect of the treatment on osteoblast and osteoclast precursor cells and on the ability of the latter to differentiate to mature osteoclasts.

Two interesting VEGF₁₂₁ fusion toxins have recently been reported in the literature. Smagur et al. (47) recently reported fusing Abrin, a potent plant toxin, to VEGF₁₂₁. Unlike VEGF₁₂₁/rGel, the VEGF₁₂₁ moiety is at the C-terminus. Also unlike VEGF₁₂₁/rGel, the fusion construct did not express as a soluble protein in E. coli and was refolded. The cytotoxicity profile against endothelial cells, which are targeted via VEGFR-2, appears to be similar to that of VEGF₁₂₁/rGel. Because access to receptor-binding determinants can depend on molecular orientation, it will be interesting to see if the construction design of Abrin-VEGF₁₂₁ results in a different activity profile than VEGF₁₂₁/rGel against osteoclast precursor cells, which are effectively targeted via VEGFR-1. A more recent VEGF₁₂₁ fusion construct is SLT-VEGF, a fusion protein comprising VEGF₁₂₁ and catalytically active A subunit of Shiga-like toxin 1 (SLT-1) (48). SLT-VEGF is internalized through VEGFR-2 mediated endocytosis, and its cytotoxicity correlates with VEGFR-2 expression. In addition, this protein exhibits efficacy in a clinically relevant orthotopic nude mouse model of pancreatic cancer. The authors reported that SLT itself can bind to the cellular receptor globotriaosylceramide known as Gb3/CD77 and enter cells through CD77-mediated endocytosis (48). This increases the potential of non-specific toxicity should the fusion protein be cleaved during circulation. In contrast, Gelonin cannot traverse the mammalian cell membrane without a carrier. The authors also reported that modification of SLT-VEGF with no more than one molecule of Cy5 dye resulted in a greater than 50-fold reduction in the IC₅₀, which may be the result of modification of VEGF lysine residues that are critically involved in the interaction of VEGF with VEGFR-2. In contrast, IC₅₀ of VEGF₁₂₁/rGel labeled with ⁶⁴Cu-DOTA (average of 3.3 DOTAmolecules per VEGF₁₂₁/rGel molecule) did not significantly change compared to VEGF₁₂₁/rGel alone, allowing multimodality imaging of tumor growth (49).

Our results indicate that VEGF₁₂₁/rGel has a significant therapeutic effect against the osteoblastic progression of metastatic PCa cells and that the antitumor effect is mediated by targeting PCa stroma. Because PCa development in bone appears to be dependent on tumor-induced bone remodeling, trials of agents targeting multiple bone-cell components, such as VEGF₁₂₁/rGel alone or in combination with conventional chemotherapeutic agents, may enable a better understanding of the relative effect of each of the processes involved in new bone formation and metastasis, such as tumor homing, recruitment of osteoclast and osteoblast precursors, and the role of the stroma. VEGF₁₂₁/rGel treatment is a novel concept of a single agent targeting several bone-cell components and may form the basis for combination therapies with cytotoxic agents for skeletal tumors in their most lethal phase.