Inhibition of Prostate Tumor Growth and Bone Remodeling by the Vascular Targeting Agent VEGF₁₂₁/rGel

Prostate cancer is highly metastatic to bone. The survival rate of patients with prostate cancer metastases is about 31%, compared with a ∼100% 5-year survival rate if the cancer is locally confined. Skeletal metastases of prostate tumors are, therefore, a major source of mortality and morbidity. One of the properties characteristic of prostate cancer is that metastases to bone produce pure osteoblastic lesions (1, 2). Osteoblastic/osteolytic lesions late in the disease process may also form. However, despite the osteoblastic nature of most prostate cancer skeletal metastasis, several studies suggest that simultaneous up-regulation of osteolytic bone resorption also occurs (3). Osteoclast precursor cells may be stimulated to differentiate or may be activated directly by tumor-secreted soluble factors, such as macrophage colony-stimulating factor (M-CSF), interleukins (IL), transforming growth factor-β, and vascular endothelial growth factor (VEGF) among others (4–6). The secretion of some of these factors by cancer cells regulates expression on the surface of stromal osteoblasts of receptor activator of nuclear factor-κB ligand (RANKL), a maturation factor and dominant mediator of osteoclastogenesis, thereby increasing osteoclast-mediated bone resorption (7).

Molecular engineering has enabled the synthesis of novel chimeric molecules having therapeutic potential. Chimeric fusion constructs targeting the receptors for IL-2, EGF, VEGF, and other growth factor have been described (8–11). Receptor kinase inhibitors (12), antibodies targeting the receptors for VEGF (13), and other approaches to reduce VEGF expression or its interaction to its receptors (14–17) are under development to disrupt tumor angiogenesis.

We previously developed and characterized VEGF₁₂₁/recombinant gelonin (rGel), a novel growth factor fusion construct composed of VEGF₁₂₁ and the highly cytotoxic plant toxin gelonin (rGel; ref. 18). VEGF₁₂₁ binds only to Flt-1/FLT-1 [VEGF receptor (VEGFR)-1] and Flk-1/KDR (VEGFR-2). Both receptors are overexpressed on the endothelium of tumor vasculature (19–21), including lung, brain, breast, colon, prostate, skin, and ovarian cancers. In contrast, these receptors are almost undetectable by immunohistochemistry in normal tissues other than pancreas and kidney (22). These receptors have also been identified on osteoclasts (23–25), and Flt-1/FLT-1 may play a role in recruitment of osteoclast precursors to the site of bone resorption (24). These receptors, therefore, seem to be excellent targets for the development of therapeutic agents that inhibit tumor growth and metastatic spread through inhibition of tumor neovascularization or through inhibition of osteoclastogenesis.

Several reports (26–28) have implicated VEGF in prostate carcinogenesis and in metastatic spread in addition to its well-established role in angiogenesis. VEGF is also produced by tumor cells to facilitate nesting of metastatic cells in bone and to promote neovascularization (23), which are two critical events necessary for the successful formation of skeletal metastases. Previous studies (29, 30) have suggested a link between VEGF expression and metastatic spread (including skeletal metastases) in prostate tumors. However, the relative contribution of neovasculature versus osteoclast activity in the development and maturation of skeletal metastasis is not well understood. Kitagawa et al. (31) showed recently the activity of PTK787, a tyrosine kinase inhibitor designed to bind to the ATP-binding sites of VEGFRs, against prostate cancer–induced osteoblastic lesions in bone. In the current study, we used the vascular ablative fusion toxin VEGF₁₂₁/rGel to inhibit prostate cancer tumor growth in an osteolytic model and show that targeting VEGFR-1 on osteoclast precursor cells is important in the resulting suppression of skeletal osteolytic lesions.

Bacterial strains, pET bacterial expression plasmids, and recombinant enterokinase were obtained from Novagen (Madison, WI). Rabbit anti-gelonin antisera were obtained from the Veterinary Medicine Core Facility at M.D. Anderson Cancer Center (Houston, TX). Anti-Flt-1 and anti-Flk-1 polyclonal antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). R-phycoerythrin (R-PE)–conjugated Flk-1 monoclonal antibody and Alexa Fluor 488–conjugated CD11b (Mac-1) monoclonal antibody were purchased from BD PharMingen (San Diego, CA). R-PE-conjugated secondary antibodies were purchased from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA). Neutralizing antibodies to Flt-1 (AF471), Flk-1 (AF644), recombinant M-CSF, and recombinant placenta growth factor (PlGF) were purchased from R&D Systems (Minneapolis, MN).

Cell lines. Porcine aortic endothelial cells transfected with the human KDR (PAE/KDR) or FLT-1 (PAE/FLT-1) receptors were a generous gift from Dr. J. Waltenberger (University Hospital, Maastricht, Netherlands). Mouse brain endothelial (bEnd3) cells, mouse osteoclast precursor cells (RAW264.7), and the prostate cancer cell line PC-3 were obtained from American Type Culture Collection (Manassas, VA). Bone marrow–derived monocytes (BMM) were harvested as described below.

Primary bone marrow cell culture. BMM cells were harvested from the tibiae and femora of 8- to 12-week-old mice. Bone marrow cells were washed and plated in complete α-MEM overnight. Nonadherent cells were then washed, resuspended in α-MEM, and plated in 96-well or 24-well plates in the presence of M-CSF (10 ng/mL) as described below. After 3 days, cells were washed and treated as described below.

Animals. Male athymic BALB/c nude mice (National Cancer Institute, Frederick, MD) and black 6 mice (C57BL/6J, Charles River, Wilmington, MA) were maintained in a laminar airflow cabinet under specific pathogen-free conditions according to American Association for Accreditation of Laboratory Animal Care (AAALAC) standards.

Expression and purification of VEGF₁₂₁/rGel. The construction, expression, and purification of VEGF₁₂₁/rGel has been described previously (18). The fusion toxin was stored in sterile PBS at −20°C.

Cytotoxicity of VEGF₁₂₁/rGel and rGel. Cytotoxicity of VEGF₁₂₁/rGel and rGel against log-phase PC-3, RAW264.7, and BMM cells was done as described (18). Log-phase cells (3 × 10³ PC-3, 5 × 10³ RAW264.7, or 2.5 × 10⁴ BMM) were plated in 96-well flat-bottomed tissue culture plates and allowed to attach overnight. Purified VEGF₁₂₁/rGel and rGel were diluted in culture medium and added to the wells in 5-fold serial dilutions. Cells were incubated for 72 hours. The remaining adherent cells were stained with crystal violet (0.5% in 20% methanol) and solubilized with Sorenson's buffer [0.1 mol/L sodium citrate (pH 4.2) in 50% ethanol]. Absorbance was measured at 630 nm. Competition assays were done by plating RAW264.7 (5 × 10³) and BMM (2.5 × 10⁴) cells in 96-well plates and by preincubating the cells for 1 hour with 25 nmol/L PlGF before addition of 40 nmol/L (RAW264.7) or 8 nmol/L (BMM) VEGF₁₂₁/rGel. Results were analyzed by staining the remaining adherent cells with crystal violet as described above.

In vitro osteoclast differentiation. BMM and RAW264.7 cells were cultured in 96-well plates at a density of 5 × 10³ per well and 3 × 10³ per well, respectively. Cell cultures were initially treated with 100 ng/mL RANKL and 10 ng/mL M-CSF (for BMM) and also subject to a medium change on day 3. Osteoclast differentiation was assessed by counting the total number of multinucleated (more than three nuclei), tartrate-resistant acid phosphatase (TRAP)–positive cells per well 96 hours after treatment using the Leukocyte Acid phosphatase kit (Sigma-Aldrich, St. Louis, MO).

Time-dependent cytotoxicity and effect on osteoclastogenesis. Nonadherent, primary monocytes (50 × 10³ per well) were cultured overnight in 24-well plates. Cells were then treated with RANKL (100 ng/mL) in the absence or presence of various concentrations of VEGF₁₂₁/rGel or rGel added either simultaneously with RANKL or 60 hours after RANKL stimulation. TRAP staining was done at 60 or 96 hours after stimulation with RANKL. Cells were fixed and stained for TRAP and TRAP-positive multinucleated cells were counted in each condition.

RNA extraction. BMM and RAW264.7 cells were treated with their respective IC₅₀ VEGF₁₂₁/rGel doses in the absence and presence of RANKL (and presence of M-CSF for BMM cells) for 24 to 96 hours. Control cells, including bEnd3, PAE/KDR, and PAE/FLT-1 cells, were treated with PBS. Total RNA was extracted using the 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).

Reverse transcription-PCR analysis. Levels of Flt-1/FLT-1, Flk-1/KDR, and VEGF-A were assessed by reverse transcription-PCR (RT-PCR) analysis. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) primers were used as controls. In addition to primers for Flk-1/KDR, Flt-1/FLT-1, and GAPDH described previously (22), the following primers were also used: Flk-1, 5′-CATGCACAGTCTACGCCAACC (forward) and 5′CGCAACATGTTTACACTTCGGT (reverse); Flt-1, 5′-ACTGAAACTAGGCAAATCGCTCG (forward) and 3′-GCGGATCTGTCACAGTCGTCG (reverse); and VEGF-A, 5′-TGAAGTGATCAAGTTCATGGACGT (forward) and 5′-TCACCGCCTTGGCTTGTC (reverse). Amplified RT-PCR products were subjected to densitometric analysis using Alpha Innotech FluorChem 8900 (San Leandro, CA).

Western blot analysis. Total cell extracts of PAE/KDR and PAE/FLT-1 cells were obtained as described previously (22). Blots were incubated overnight with appropriate antibodies (1:200 in 5% milk) followed by incubation with goat anti-rabbit IgG horseradish peroxidase, developed using the Amersham (Piscataway, NJ) enhanced chemiluminescence detection system and exposed to X-ray film.

Fluorescence-activated cell sorting analysis. RAW246.7 and BMM cells were harvested and washed with PBS and resuspended in staining buffer (PBS + 2% fetal bovine serum) at a concentration of 10 × 10⁶ cells/mL. One hundred microliter aliquots of RAW264.7 and BMM cells were incubated with 2 and 0.5 μg anti-Flt-1 or R-PE-conjugated anti-Flk-1 antibody, respectively. Cells were incubated with primary antibodies for 1 hour at 4°C. In addition, BMM cells were incubated for 1 hour (4°C) with 0.2 μg Alexa Fluor 488–conjugated CD11b antibody for detection of the monocyte/macrophage population. R-PE-conjugated secondary antibody (0.5 μg) was used to detect anti Flt-1. Samples were analyzed on a BD FACSCalibur flow cytometer (BD Biosciences, San Jose, CA) using CellQuest Pro acquisition software (BD Biosciences). Instrumentation setup and electronic compensation for spectral overlap was done using cell samples single-stained with FITC or R-PE. At least 10,000 events were collected for each sample.

Internalization of VEGF₁₂₁/rGel into RAW264.7 and BMM cells. Cells were incubated with various concentrations of VEGF₁₂₁/rGel or rGel at the time points indicated. To show receptor specificity, Cells were pretreated with Flt-1 or Flk-1 neutralizing antibodies for 1 hour before treatment with VEGF₁₂₁/rGel or rGel. Glycine buffer [500 mmol/L NaCl, 0.1mol/L glycine (pH 2.5)] was used to strip the cell surface of noninternalized VEGF₁₂₁/rGel. Cells were fixed with 3.7% formaldehyde and permeabilized with 0.2% Triton X-100. Nonspecific binding sites were blocked with 5% bovine serum albumin in PBS. Cells were then 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 slides were fixed with DABCO medium, mounted, and visualized under fluorescence (Nikon Eclipse TS1000, Melville, NY) and confocal (Zeiss LSM 510, Thornwood, NY) microscopes.

Intrabone injections. The PC-3 cells were harvested by a 1-minute treatment with trypsin/EDTA. The culture flask was tapped to detach the cells. The cells were washed in PBS and resuspended in PBS in preparation for implantation into the mice. Animals (10 mice per group) were anesthetized with i.m. injections of ketamine (100 mg/kg) plus acepromazine (2.5 mg/kg). Aliquots of 5 × 10⁴ of PC-3 cells were diluted in 5 μL growth medium and then injected into the distal epiphysis of the right femur of each mouse using a 28-gauge Hamilton needle. The contralateral femur was used as an internal control. Twenty mice were randomized into two treatment groups. Treatment began 1 week after tumor placement. A maximum tolerated dose of 45 mg/kg for VEGF₁₂₁/rGel given i.v. under the conditions described below has been established previously (22). Accordingly, the animals were treated (i.v.) with the following protocol: group 1, 200 μL saline every other day for nine days (5 treatments) and group 2, 180 μg VEGF₁₂₁/rGel in 200 μL saline every other day for nine days (5 treatments). Mice were monitored weekly for tumor bulk and bone loss. Mice were killed in case of excessive bone loss as per AAALAC guidelines and pathologic examination of the subject bones was done.

Processing of bone tissue samples. Formalin-fixed, paraffin-embedded tissue samples from the tumors were prepared as described (32) and stained with H&E. TRAP staining was done as described by the manufacturer of the kit (Sigma-Aldrich) and the total number of osteoclasts was determined by counting the number of multinucleated (more than three nuclei), TRAP-positive cells.

VEGF₁₂₁/rGel inhibits growth of intrafemoral PC-3 tumors and reduces the number of tumor-induced osteoclasts. The antitumor effect of the fusion protein VEGF₁₂₁/rGel was evaluated in a prostate cancer bone model by injecting PC-3 tumor cells into the distal epiphysis of the right femur of athymic nude mice. Tumor growth was monitored by X-ray analysis and animals with large osteolytic lesions or bone lysis were sacrificed. One hundred percent of tumor-inoculated mice treated with saline developed osteolytic lesions (Fig. 1A,, left) and 50% survival occurred 40 days after tumor placement (P < 0.05; Fig. 1B). In contrast, treatment with VEGF₁₂₁/rGel resulted in suppression of intrafemoral growth of tumor osteolytic lesions as assessed radiologically (Fig. 1A,, right); 50% of the VEGF₁₂₁/rGel-treated mice survived past 140 days without sign of osteolysis (Fig. 1B). H&E staining showed PC-3 cells throughout the bone marrow of mice treated with saline (Fig. 1C,, left). In contrast, isolated pockets of PC-3 cells were visible in some bone marrow sections from VEGF₁₂₁/rGel-treated mice (Fig. 1C,, middle, arrows). Tumor cells were not visible in some bone sections from mice treated with VEGF₁₂₁/rGel (Fig. 1C , right). Thus, VEGF₁₂₁/rGel significantly prevented osteolysis and tumor growth in bone compared with saline-treated animals.

We next examined PC-3 tumor cells to identify whether the observed VEGF₁₂₁/rGel-mediated inhibition of PC-3-induced osteolysis was due to a direct effect on the tumor cells. We have shown previously that the VEGF₁₂₁/rGel fusion construct is not specifically cytotoxic to PC-3 cells in vitro compared with the rGel toxin alone, suggesting that these cells express an insufficient number of VEGFRs to mediate specific VEGF₁₂₁/rGel cytotoxicity (18). RT-PCR analysis indicated that PC-3 cells express relatively low levels of FLT-1 and KDR transcript (data not shown).

Because PC-3 tumors result in osteolytic lesions, we analyzed the number of osteoclasts in tissue sections. As expected, TRAP staining revealed a dramatic increase in the number of osteoclasts in the tumor-bearing leg of mice treated with saline (P < 0.05; Fig. 1D). However, bone sections of VEGF₁₂₁/rGel-treated mice surprisingly showed the same number of osteoclasts as those present in the contralateral (control) leg, suggesting that VEGF₁₂₁/rGel inhibits tumor-mediated osteoclast proliferation and/or differentiation.

VEGF₁₂₁/rGel kills osteoclast precursor cells but not terminally differentiated osteoclasts. To understand the effect of VEGF₁₂₁/rGel in the bone microenvironment and test if VEGF₁₂₁/rGel may be directly targeting osteoclast precursor cells in vivo, we next evaluated the effect of VEGF₁₂₁/rGel on RANKL-induced osteoclast differentiation of RAW264.7 and BMM cells in vitro. Treatment with increasing concentrations of VEGF₁₂₁/rGel, but not rGel, showed a dramatic decrease of TRAP-positive multinucleated osteoclasts in both RAW264.7 (Fig. 2A and B) and BMM (Fig. 2C and D) cells. The observed effect was not mediated by either VEGF₁₂₁ or gelonin alone but is a characteristic unique to the combined fusion protein (data not shown). The IC₅₀ of VEGF₁₂₁/rGel on undifferentiated RAW264.7 cells was 40 nmol/L compared with 900 nmol/L for rGel, indicating that the cytotoxicity of VEGF₁₂₁/rGel was mediated through VEGF₁₂₁ and suggested the presence of a receptor recognizing VEGF₁₂₁ (Supplementary Fig. S1A). Similar to the RAW264.7 cells, the IC₅₀ of VEGF₁₂₁/rGel (8 nmol/L) on undifferentiated BMM cells was substantially lower than that of rGel (Supplementary Fig. S1B). VEGF₁₂₁/rGel showed a greater cytotoxic effect on BMM compared with undifferentiated RAW264.7 cells. In addition, we observed that VEGF₁₂₁/rGel, but not rGel, inhibited the M-CSF-dependent survival of monocytes (data not shown). Thus, VEGF₁₂₁/rGel not only inhibited RANKL-mediated differentiation of osteoclast precursors but also exhibited cytotoxicity toward undifferentiated cells in a targeted manner.

The observed inhibitory effect of VEGF₁₂₁/rGel on osteoclastogenesis could be due to a cytotoxic effect of VEGF₁₂₁/rGel on osteoclast progenitor cells. Insufficient number of osteoclast progenitor cells may lead to impaired contact between committed progenitors, leading to inability to form multinucleated osteoclasts. Therefore, we next investigated the susceptibility of primary mouse monocytes and their terminally differentiated counterparts to VEGF₁₂₁/rGel cytotoxicity. BMM cells were treated with VEGF₁₂₁/rGel or rGel at various times after RANKL and M-CSF stimulation. Addition of VEGF₁₂₁/rGel with RANKL and M-CSF resulted in inhibition of osteoclastogenesis. The degree of inhibition was dependent on the concentration of VEGF₁₂₁/rGel (Fig. 3A and B). However, osteoclastogenesis proceeded normally if BMM cells were first allowed to differentiate for 60 hours before the addition of VEGF₁₂₁/rGel (Fig. 3C). The apparent cytotoxicity of VEGF₁₂₁/rGel to osteoclast precursor cells, but not to mature osteoclasts, was further investigated by adding VEGF₁₂₁/rGel at different time points after RANKL stimulation of both RAW264.7 and BMM cells (Table 1). For both cells, the increase in the IC₅₀ of VEGF₁₂₁/rGel corresponded to the length of time the cells were allowed to differentiate before the addition of VEGF₁₂₁/rGel. For RAW264.7 cells, the IC₅₀ of VEGF₁₂₁/rGel increased from 30 nmol/L when added simultaneously with RANKL to 300 nmol/L when added 96 hours after RANKL stimulation, whereas the IC₅₀ of rGel did not change significantly. The change in IC₅₀ of VEGF₁₂₁/rGel on BMM cells was more dramatic, increasing from 8 to 100 nmol/L if VEGF₁₂₁/rGel was added 24 hours after RANKL stimulation. Addition of VEGF₁₂₁/rGel 48 hours after RANKL stimulation increased the IC₅₀ to >400 nmol/L.

VEGF₁₂₁/rGel, but not rGel, is internalized into RAW264.7 and BMM cells through a specific mechanism. We next examined by immunostaining whether VEGF₁₂₁/rGel was delivered into the cytoplasm of the osteoclast precursor cells. VEGF₁₂₁/rGel, but not rGel, localized in the cytoplasm of RAW264.7 cells, suggesting that internalization is mediated by VEGF₁₂₁ (Fig. 4). We examined further whether internalization of VEGF₁₂₁/rGel was mediated by VEGF₁₂₁ by examining if VEGF₁₂₁/rGel activated pp44/42, a known downstream target of VEGFR activation (33). Western blot analysis revealed that treatment of RAW264.7 cells with VEGF₁₂₁/rGel resulted in activation of pp44/42 at similar levels as treatment with an equimolar amount of VEGF₁₂₁ (Supplementary Fig. S2). rGel did not induce stimulation of pp44/42, indicating that the effect of VEGF₁₂₁/rGel is mediated by VEGF₁₂₁ rather than a nonspecific mechanism. Treatment of RAW264.7 cells with PlGF, an Flt-1-specific ligand, also resulted in pp44/42 activation, underscoring the presence of this receptor on the surface of osteoclast precursor cells and a possible role in mediating VEGF₁₂₁/rGel internalization.

Localization of VEGF₁₂₁/rGel into RAW264.7 and BMM cells is mediated by Flt-1. To determine the role of VEGF₁₂₁ receptors in VEGF₁₂₁/rGel-mediated cytotoxicity of osteoclast precursor cells, we preincubated RAW264.7 and BMM cells with neutralizing antibodies to Flt-1 and Flk-1 for 1 hour before the addition of VEGF₁₂₁/rGel and monitored internalization of VEGF₁₂₁/rGel. Pretreatment of RAW264.7 and BMM cells with neutralizing antibodies to Flt-1, but not Flk-1, inhibited the localization of VEGF₁₂₁/rGel into these cells (Fig. 5A). We assessed the role of each receptor in VEGF₁₂₁/rGel-mediated cytotoxicity by preincubating RAW264.7 and BMM cells with Flt-1 or Flk-1 neutralizing antibodies or with PlGF for 1 hour before the addition of VEGF₁₂₁/rGel. PlGF was able to inhibit the VEGF₁₂₁/rGel-mediated cytotoxicity in both RAW264.7 (Fig. 5B) and BMM cells (Fig. 5C). Taken together, this indicates that the Flt-1 receptor, but not the Flk-1 receptor, is responsible for mediating VEGF₁₂₁/rGel-induced cytotoxicity in osteoclast progenitor cells.

RAW264.7 and BMM cells express Flt-1. We determined the levels of Flk-1 and Flt-1 in RAW264.7 and BMM cells. RT-PCR analysis indicated low levels of Flt-1, but no Flk-1 transcript, in RAW264.7 cells. Western blot analysis of RAW264.7 cells confirmed this observation (Supplementary Fig. S3A). Fluorescence-activated cell sorting (FACS) analysis indicated that 99% of the RAW264.7 cells expressed Flt-1 (Fig. 6A) and 8% expressed Flk-1 (Fig. 6B). RT-PCR analysis of BMM cells showed an amplification of both Flt-1 and Flk-1 (data not shown) and FACS analysis showed that 41.9% of the CD11b-positive BMM cells expressed Flt-1 and 5.4% expressed Flk-1 (Fig. 6C-E). RT-PCR analysis of BMM cells following stimulation of RANKL-mediated osteoclastogenesis showed no change in the levels of Flk-1, but treatment seemed to down-regulate the Flt-1 transcript (Supplementary Fig. S3B). RT-PCR analysis of mVEGF isoforms detected low levels of VEGF₁₆₄ and VEGF₁₂₀ transcript but no VEGF₁₈₈ (data not shown). The down-regulation of Flt-1 mRNA in BMM cells following stimulation of osteoclastogenesis by RANKL was confirmed by testing samples at different cycles of the RT-PCR analysis (Supplementary Fig. S3C) and further validated by densitometric analysis as described in Materials and Methods. All bands were individually compared with its internal GAPDH standard loaded in the same fashion to normalize the data. Flk-1 did not show a significant change in expression 96 hours after RANKL stimulation, whereas Flt-1 exhibited a down-regulation of 3.4-fold 96 hours after RANKL stimulation compared with untreated BMM (Supplementary Fig. S3D).

Osteoclastogenesis plays a central role in bone development and maintenance. Normal bone requires a close coordination of osteoblastic matrix deposition and osteoclastic resorption (for review, see ref. 34). Interference with the process of osteoclastogenesis alters the kinetics of bone remodeling resulting in abnormal bone development (35, 36). There is general consensus that the hematopoietically derived osteoclast is the pivotal cell in the degradation of the bone matrix (37) and stimulation of osteoclastic bone resorption is the primary mechanism responsible for bone destruction in metastatic cancer (38). We tested the efficacy of VEGF₁₂₁/rGel, a known vascular ablative agent, in a model that mimics tumor growth in bone following skeletal metastases. Our results show that (a) VEGF₁₂₁/rGel dramatically inhibits prostate tumor growth in bone, (b) VEGF₁₂₁/rGel treatment suppressed tumor-mediated osteoclastogenesis in vivo, (c) VEGF₁₂₁/rGel is selectively cytotoxic to osteoclast precursor cells rather than to mature osteoclasts, and (d) VEGF₁₂₁/rGel cytotoxicity is mediated by the Flt-1 receptor, which is down-regulated as osteoclast precursor cells differentiate to mature osteoclasts.

Although the VEGF-A cytokine family plays an essential role in the regulation of embryonic and postnatal physiologic angiogenic processes, its role in skeletal growth, endochondral bone formation, and differentiation of osteoclast precursor cells is substantially less well understood. Osteoclast precursor cells are recruited to the future site of resorption by VEGF-A and RANKL, two cytokines that are expressed in the immediate vicinity of the bone surface (39, 40). VEGF mRNA is expressed by hypertrophic chondrocytes in the epiphyseal growth plate, suggesting that a VEGF gradient is needed for directional growth and cartilage invasion by metaphyseal blood vessels (23). Treatment with a soluble FLT-1 antibody to block the VEGFR results in almost complete suppression of blood vessel invasion and impaired trabecular bone formation (23), a development that is reversed by cessation of the anti-VEGF treatment. A similar phenotype is observed when Vegf is deleted in the cartilage of developing mice by means of Cre-loxP-mediated, tissue-specific gene ablation (41). In addition, examination of VEGF^120/120 mice shows delayed recruitment of blood vessels into the perichondrium as well as delayed invasion of vessels into the primary ossification center, indicating a significant role of VEGF at both early and late stages of cartilage vascularization (42). VEGF plays an important role in the vascularization of bone tissues (43); increases osteoclast-mediated bone resorption (25); induces osteoclast chemotaxis and recruits osteoclasts to the site of bone remodeling (40); and can partially rescue M-CSF deficiency in op/op mice (25).

Both of the major receptors of VEGF-A have been observed in osteoclasts (23–25), although some reports cite only the presence of Flt-1 (24, 44). The VEGF-Flt-1 interaction has been implicated in the recruitment process of osteoclast precursor cells from hematopoietic tissue to the site of bone resorption (24, 45). Conflicting data at present suggest that the function and signaling of Flt-1 is complex but dependent on developmental stage and on cell type (46, 47). Flt-1/FLT-1, but not Flk-1/KDR, is associated with inhibition of hematopoietic stem cell cycling, differentiation, and hematopoietic recovery in adults and hematopoietic cell motility (45, 48). Recently, Niida et al. (49) introduced a VEGFR-1 tyrosine kinase domain-deficient mutation into op/op mice and showed that signaling between Flt-1 and the CSF-1 receptor plays a key role in osteoclastogenesis and maintenance of bone marrow functions, including vascularization.

Previous studies in our laboratory have shown that the VEGF₁₂₁/rGel fusion construct can bind to both Flt-1/FLT-1 and Flk-1/KDR but cytotoxicity and cellular internalization of VEGF₁₂₁/rGel on vascular endothelial cells is mediated through internalization of Flk-1/KDR and not Flt-1/FLT-1. In addition, endothelial cells that express the KDR receptor below a threshold number (∼2 × 10³ receptors per cell) are several hundred-fold more resistant to VEGF₁₂₁/rGel than are cells that overexpress KDR (1× 10⁵ to 3 × 10⁵ receptors per cell; ref. 18). Therefore, our finding that VEGF₁₂₁/rGel can disrupt osteoclast activity was unanticipated. Our observation that the antiangiogenic VEGF₁₂₁/rGel construct can significantly affect prostate tumor growth in a skeletal setting is a novel finding for agents in this class. The current study shows that Flt-1 and not Flk-1 on osteoclast precursor cells seems to be primarily responsible for mediating the cytotoxic effects of VEGF₁₂₁/rGel and that the biological role and character of the two receptors for VEGF is different on osteoclasts compared with vascular endothelial cells (46, 47).

VEGFRs have been the focus of attention as therapeutic agents, both by us (18, 22) and others (8, 11), and the context has usually been in terms of vascular targeting and antiangiogenic therapy. The ability of VEGF₁₂₁/rGel to target osteoclast precursor cells represents a novel application of this fusion protein with significant clinical potential. Because VEGF is involved in both angiogenesis and osteolysis, VEGF₁₂₁/rGel may disrupt tumor growth both by preventing angiogenesis and by inhibiting the process of bone remodeling. This in turn prevents further tumor invasion and osteolytic penetration of tumor into bone. We plan to examine the effects of VEGF₁₂₁/rGel on cancer cell–mediated neoangiogenesis in the bone marrow and delineate the relative contribution of both processes to the development of osteolytic lesions. Because osteolysis is an integral component of osteoblastic lesions as well, it remains to be seen if VEGF₁₂₁/rGel is as efficacious in these models and xenograft model studies are currently under way to address this issue.

Kaplan et al. (50) showed recently that VEGFR1-positive hematopoietic cells play a key role in the development of metastasis by forming clusters in tumor-specific sites in advance of tumor metastases. Apparently, VEGF₁₂₁/rGel may be used to target hematopoietic cells that play a role in preparation of this “premetastatic niche,” thereby impeding, or preventing, metastases from occurring.

Based on our data, the “Trojan horse” approach used by VEGF₁₂₁/rGel might be useful for the treatment of bone-related malignancies, such as bone metastases and the osteoclast component of Paget's disease as well as hematopoietic diseases, such as multiple myeloma. Our study also shows that Flt-1/FLT-1 may play an important role in bone biology and warrants further study about its importance in the normal bone remodeling process and in various pathologic states.

Grant support: Clayton Foundation for Research (M.G. Rosenblum), Gillson Longenbaugh Foundation (Houston, TX) and Peregrine Pharmaceuticals, Inc. (P.E. Thorpe), Specialized Program of Research Excellence Program Career Development Award of the Prostate Cancer (B.G. Darnay), NIH grant CA96797 and Prostate Cancer Foundation awards (N.M. Navone), and P30 CA 16672 (Cancer Center Core grant).

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.

We thank W. Hittelman for his assistance in confocal imaging.