Neuropilin-1 (NRP-1) was first described as a coreceptor implicated in neuronal guidance that bound members of the semaphorin/collapsin family. NRP-1 is also expressed in endothelial cells and is believed to promote angiogenesis by acting as a coreceptor with vascular endothelial growth factor (VEGF) receptor 2. Recent studies suggest that NRP-1 can function through both a VEGF-dependent and VEGF-independent fashion. Expression of NRP-1 has been shown in many human tumors, including pancreatic adenocarcinomas. The exact role of NRP-1 in tumor cells is unknown, particularly in cells that lack the NRP-1 coreceptors VEGF receptor 2 and Plexin-A1. To discern the regulatory role(s) of NRP-1 in pancreatic adenocarcinoma that lack these coreceptors, we overexpressed both full-length NRP-1 and a deletion form of NRP-1 that does not interact with semaphorin or VEGF. Overexpression of either isoform reduced several key tumorigenic properties, including anchorage-independent cell growth and migration in vitro, and resulted in reduced tumor incidence and tumor volume in vivo. Conversely, reduction of NRP-1 expression by small interfering RNA targeting led to enhanced tumor growth. Thus, NRP-1 may play distinct growth regulatory roles in different tumor types, and altering NRP-1 expression or function may be a means of influencing the growth of pancreatic cancers.
- vascular endothelial growth factor
- pancreatic adenocarcinoma
Pancreatic adenocarcinoma, one of the deadliest human malignancies, has a mortality-to-incidence ratio approaching 1.0. The 5-year survival rate of patients diagnosed with pancreatic cancer is <4%. The disease is typically in an advanced stage at diagnosis and surgical resection is often not a viable option. Treatment is further complicated by the predisposition of pancreatic cancers for early metastasis and resistance to conventional chemotherapeutic and radiation treatments ( 1). Thus, understanding the molecular basis for the growth and metastasis of pancreatic adenocarcinoma is critically needed to develop effective therapeutic treatments for this disease.
Neuropilin-1 (NRP-1) was initially characterized as an axonal membrane glycoprotein involved in neuronal guidance and development. Neuropilin serves as a coreceptor for semaphorin 3a (Sema3a) and vascular endothelial growth factor (VEGF). In neuronal cells, NRP-1 regulates axon guidance, in part by acting as a coreceptor with Plexin-A1 and the ligand Sema3a ( 2– 4) but may also guide the accurate migration of somata by binding VEGF164 ( 5). NRP-1 is also expressed in endothelial cells, where it is involved in the regulation of angiogenesis and endothelial cell migration ( 6– 8), possibly via its function as a coreceptor for VEGF isoforms ( 6, 9, 10). Overexpression of NRP-1 in a transgenic mouse model increases capillary and blood vessel formation and hemorrhage ( 11), whereas functional inactivation of NRP-1 in mice has been shown to lead to embryonic lethality with multiple vascular abnormalities, including avascular regions, heterogeneous blood vessel size, and abnormally formed dorsal aorta ( 12). These results are consistent with the idea that NRP-1 is a key regulator of developmental angiogenesis.
NRP-1 consists of an 860-amino-acid extracellular glycoprotein region, a 22-amino-acid transmembrane region, and a 40-amino-acid intracellular region. NRP-1 has not been shown to possess intrinsic enzymatic activity. The extracellular region consists of five domains, the COOH terminus MAM (meprin, A5, μ-phosphatase), the a1 and a2 (CUB) domains, and the b1 and b2 (CF V/VIII) domains ( 13). Mutational analysis suggests that both the a and b domains are required for Sema3a signaling, whereas interactions with VEGF require only the b domains ( 14). Although the intracellular region of NRP-1 may not be required for Sema3a signaling, this region may be necessary for VEGF-independent control of endothelial cell migration ( 14, 15). Further, NRP-1 regulates endothelial cell adhesion to extracellular matrix proteins independently of VEGF receptor 2 (VEGF-R2; ref. 16). Consequently, NRP-1 may affect endothelial cell function through several different mechanisms by interacting with multiple proteins.
NRP-1 is expressed in various human tumors, including prostate cancer, breast cancer, melanoma, and pancreatic adenocarcinoma, but not in corresponding normal epithelial tissues ( 17– 21). In some model systems, NRP-1 expression has been shown to increase tumorigenicity, possibly by promoting VEGF-mediated angiogenesis ( 22, 23). However, the role of NRP-1 in regulating the malignant characteristics of these cells is not well known. We sought to determine whether neuropilin can mediate cell signaling and affect tumor cell biology in pancreatic tumor cells that lack the NRP-1 coreceptors, VEGF-R2, VEGR-R1, and Plexin-A1.
We report here that in the pancreatic adenocarcinoma cell line PANC-1, overexpression of NRP-1 down-regulated Akt and extracellular signal-regulated kinase-1/2 (Erk-1/2) phosphorylation and inhibited in vivo tumor growth. These characteristics do not require the VEGF/Sema3a-binding domains. Reducing NRP-1 expression by small interfering RNA (siRNA) targeting resulted in increased Akt and Erk-1/2 phosphorylation, cell migration, and in vivo tumorigenicity. These results suggest that NRP-1 may perform VEGF/Sema3a-dependent and VEGF/Sema3a-independent functions and that whether these functions are protumorigenic or antitumorigenic depends on which coreceptors and ligands are expressed.
Materials and Methods
NRP-1 deletion construct. Deletion constructs were created with a PCR-based mutagenesis strategy using the Excite PCR-based site-directed mutagenesis kit (Stratagene, La Jolla,CA) according to the manufacturer's protocol ( Fig. 1A ). The primers used to exclude the b1 domain were 5′-ACATTTGAAATCTTCTGAGACACTGCTC-3′ (reverse) and 5′-TGCTCTGGAATGTTGGGTATGGTGTCTGG-3′ (forward; Invitrogen, Carlsbad CA). PCR consisted of 95°C for 1 minute followed by 20 cycles of 95°C for 30 seconds, 74°C for 1 minute, and 70°C for 14 minutes. The PCR product was then purified using Sephadex G-50 (Roche Diagnostics Corp., Indianapolis, IN) and DH-5α chemical competent Escherichia coli bacteria (Invitrogen) were transformed with these plasmids. The clones obtained (termed DelC1 and DelC2) were sequenced to confirm specific deletions.
Cell line and cell culture conditions. The human pancreatic adenocarcinoma PANC-1 cell line was obtained from American Type Culture Collection (Manassas, VA) and maintained in MEM with 10% fetal bovine serum (FBS), 2 mmol/L l-glutamine, penicillin, and streptomycin (Life Technologies, Grand Island, NY) at 37°C with 5% CO2.
Creation of small interfering RNA expression plasmids and small interfering RNA cell lines. siRNA expression plasmids were created using Ambion pSilencer 1.0-U6 (Ambion, Austin, TX) according to the manufacturer's instructions. NRP-1–specific target sequences were designed using the Ambion siRNA Web design tool (http://www.ambion.com). The two target sequences were 5′-aagctctgggcatggaatcag-3′ and 5′-aaagccccgggtaccttacat-3′. Oligonucleotides corresponding to these sequences with flanking Apa1 (5′) and R1 (3′) ends were purchased from Invitrogen/Life Technologies (Carlsbad, CA) and ligated into the expression plasmid at compatible sites. Isolated clones (termed SiC1 and SiC2) were confirmed by sequencing. PANC-1 cells were then transfected with 0.5 ng of each siRNA plasmid and 10 ng of pcDNA without the promoter for the G418 resistance gene so that transfectants could be selected on the basis of G418 resistance. Cells were then grown in selective media containing G418, as previously described ( 24). Negative controls were transfected with scrambled NRP-1 target sequences and pcDNA plasmids at identical concentrations. NRP-1 expression levels of siRNA clones were determined by immunoprecipitation and Western blot analysis.
Development of cell lines expressing full-length NRP-1; NRP-1 deletion construct. A total of 1.0 × 106 subconfluent cells were transfected with plasmids containing either full-length NRP-1 or the NRP-1 VEGF/Sema3a-binding domain deletion construct using FuGENE 6 (Roche, Mannheim, Germany) and 1 μg of DNA at a 6:1 ratio according to the manufacturer's instructions. Twenty-four hours after transfection, the cells were rinsed and selected in medium containing 600 μg/mL G418 (Invitrogen/Life Technologies). Single colonies of stable transfectants were isolated and expanded for further analysis.
Immunoprecipitation and Western blot analysis. Immunoprecipitation and Western blot analysis were done as previously described ( 25) with minor modifications. Cells were grown to a confluence of 85% to 90% in complete MEM and solubilized in 20 mmol/L Tris-Cl (pH 8.0), 137 mmol/L NaCl, 1% Triton X-100, 1 mmol/L Na3VO4, 2 mmol/L EDTA, and 1 complete Mini Protease Inhibitor Cocktail Tablet (Roche Diagnostics). NRP-1 was immunoprecipitated from 500 μg cell lysate with 1 μg polyclonal goat anti-NRP-1 antibody (C-19; Santa Cruz Biotechnology, Santa Cruz, CA) and protein A/G Sepharose beads (Roche Diagnostics). Expression levels of secreted Sem3a protein were determined by concentrating 10 mL complete MEM with a Centricon YM-50 (Millipore, Billerica, MA) centrifugation filtration columns as described by the manufacturer. Immunoprecipitated proteins, whole-cell lysates, and concentrated supernatants were separated by SDS-PAGE, transferred to polyvinylidene difluoride membranes (Amersham, Arlington Heights, IL), and probed with monoclonal mouse anti–NRP-1 antibody (A-12 and A312; Santa Cruz Biotechnology), anti-Akt antibody, anti-Erk-1/2 antibody, or anti-Sema3a antibody (Upstate Biotechnology, Lake Placid, NY). Antibodies were diluted in TBS and 0.1% Tween 20 (v/v) with 5% dried milk. Blots were incubated with the appropriate horseradish peroxidase–conjugated secondary antibodies (Bio-Rad, Hercules, CA). Labeled proteins on Western blots were visualized using the Chemiluminescence Reagent Plus detection system (New England Nuclear, Boston, MA).
Soft-agar colony-forming assay. To determine the effect of NRP-1 expression on anchorage-independent growth, 250 cells of each clone were plated per well onto six-well plates in 1 mL complete MEM containing 0.5% agarose, which were previously overlaid with 1 mL MEM with 0.8% agarose. Cells were incubated at standard conditions for 12 days (37°C, 5% CO2). Colonies >50 μm in diameter were counted under a light microscope at ×20 magnification. Five fields of each clone were counted and expressed as mean ± SD.
3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide analysis of cell proliferation. In vitro proliferation analysis was done by the reducing tetrazolium salt, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), as described previously ( 26). Briefly, 2,000 cells of each clone were plated per well onto 96-well microtiter plates filled with complete MEM. Twenty-four hours later, the medium was removed and replaced with either fresh complete MEM or with medium containing 1% FBS in the presence or absence of VEGF (20 ng/mL). One plate was developed immediately after the medium change and other plates were developed every 24 hours for 4 days. Assays were initiated by adding 20 μL of MTT substrate to each well and incubating the cells for an additional 3 hours. Finally, the medium was removed and 200 μL DMSO was added to each well. Plates were read at a wavelength of 570 nm.
Flow cytometry and cell cycle analysis. Cells of each clone were grown to 85% to 90% confluence in complete MEM. After trypsinization, cells were washed in PBS and fixed in 70% ethanol at 4°C for 2 hours. For DNA staining, cells were incubated with 10 mg propidium iodide per milliliter of PBS and 2.5 μg DNase-free RNase (Roche Diagnostics) per milliliter of PBS for at least 30 minutes. Samples were kept at 4°C, subjected to flow cytometry, and analyzed with CellFit or ModFit software.
Boyden chamber assay. Migration assays were conducted as described previously ( 27) with minor modifications. Cells (30,000) in 0.5 mL complete MEM were placed in the top compartment of a standard Boyden chamber with 8 μm membrane pores and 0.5 mL complete MEM was added to the bottom compartment (Fisher Scientific, Houston, TX). To ensure that the minor differential growth rates of the clones (as determined by MTT assay) tested did not affect the results, we did the assays 8 hours after adding the cells. Nonmigrating cells were scraped from the top compartment. Cells that had migrated to the bottom compartment were fixed and stained using the Protocol HEMA 3 stain set (Fisher Scientific), and the membranes were mounted on a standard microscope slide (Curtis Matheson Scientific, Houston, TX). Numbers represent the mean of five fields of migrated cells as visualized in ×20 magnification.
Vascular endothelial growth factor ELISA. VEGF production in culture supernatants was examined using a human VEGF-specific ELISA (Quantikine; R&D Systems, Minneapolis, MN). VEGF concentration was normalized to the total protein content of each clone grown to 85% confluence in a 100 mm cell culture dish in 1% complete MEM as measured by the Bradford assay.
In vivo tumor models. Male athymic nude mice 6 to 8 weeks old were purchased from the National Cancer Institute-Frederick Cancer Research Facility (Frederick, MD) and were maintained under specific pathogen-free conditions. All animal experiments were approved by University of Texas M. D. Anderson Cancer Center Animal Care facility, which follows NIH and M. D. Anderson Cancer Center Institutional Animal Use and Care committee guidelines.
PANC-1 cells at 80% confluence were rinsed in Ca2+- and Mg2+-free HBSS and overlaid for 1 minute with 0.05% trypsin and 0.02% EDTA solution. Cells were gently removed by pipetting in complete MEM with 10% FBS to produce a single-cell suspension. Cells were washed with HBSS, spun in a centrifuge, and resuspended in HBSS at a concentration of 2 × 107 viable cells per milliliter (as determined by trypan blue exclusion). The tumor cells (2 × 106 in 0.1 mL of PBS) were then injected s.c. into the right flank of the nude mice. A total of 10 mice per clone were used. Ten days after the injection and every third day thereafter, the length and width of the tumors that developed were measured. Mice were sacrificed at 90 days and tumor mass and volume were recorded. Volume was calculated as (length / 2) × (width2).
Quantitation of autoradiographs. The linear range of autoradiographic films was scanned using a Hewlett-Packard Scanjet scanner and quantitated with Scion Image software program. Immunoprecipitated samples were evaluated as the ratio of the average area of the protein of to that of baseline or control conditions. Each Western blot sample measured was calculated as the ratio of the average area of the protein of interest to that of control (parental). Actin or vinculin was used to ensure equal loading of all samples.
Statistical analyses. Statistical analyses were done using In Stat 2.01 statistical software (GraphPad Software, San Diego, CA) with Student's t test or Fisher's exact test where appropriate. Significance (P < 0.05) was determined with 95% confidence.
PANC-1 cells express NRP-1 but not NRP-1 coreceptors. In determining the VEGF- and Sema3a-independent functions of NRP-1, we chose the pancreatic adenocarcinoma cell line PANC-1. These cells have been shown to express high levels of NRP-1, but very low levels of NRP-2, and do not express detectable levels of VEGR1, VEGFR2, or VEGFR3 ( 28). To examine the expression of Plexin-1A, we did Western blot analysis. Expression of Plexin-1A was not observed in PANC-1 cells, but was abundantly expressed in U251 glioblastoma cells used as a positive control ( Fig. 1C; ref. 29). Thus, these cells represented an excellent model with which to examine NRP-1 signaling independently of the coreceptors VEGF-R1, VEGF-R2, and Plexin-A1.
Expression of full-length and deletion isoforms of NRP-1 in PANC-1 cells. G418-resistant clones ectopically expressing full-length NRP-1, NRP-1 deletion construct, or pcDNA were examined for NRP-1 expression. Mock-transfected pcDNA clones expressed a level of NRP-1 similar to that of parental cells (wild type; Fig. 1B). NRP-1 clones C21 and C26 showed similar levels (a 3.5- to 4-fold increase with respect to vector controls) of NRP-1 overexpression. NRP-1 clones DelC1 and DelC2, which lack the b1 domain, expressed similar levels (a 4- to 5-fold induction) of the expected ∼85 kDa protein ( Fig. 1B) that was recognized by anti-NRP-1 antibodies to both the COOH terminus and the NH2 terminus (data not shown).
NRP-1 overexpression reduces soft-agar colony-forming ability. We assessed the effect of NRP-1 overexpression on anchorage-independent growth by observing the ability of PANC-1 cells to form colonies >50 μm in diameter in soft agar. Wild-type (parental) and pcDNA-transfected PANC-1 cells formed abundant large colonies (mean ± SD, 176 ± 12 and 181 ± 1 colonies, respectively; Fig. 2A and B ). In contrast, NRP-1 overexpression significantly reduced the ability of PANC-1 cells to form colonies in soft agar. NRP-1 clones C21 and C26 formed very few colonies (15.5 ± 5 and 13 ± 3 colonies, respectively), which were also smaller overall than the colonies formed by control cells. NRP-1 clones DelC1 and DelC2 also formed few colonies (15.5 ± 4 and 18.5 ± 9 colonies, respectively) and were also smaller overall than control colonies. Increasing the incubation time from 12 to 24 days did not increase the number of colonies formed by any of these clones (data not shown). These results show that overexpression of NRP-1 suppresses the ability of PANC-1 cells to grow under anchorage-independent conditions and that this suppression occurs independently of the VEGF/Sema3a-binding domain.
NRP-1 overexpression does not significantly affect cell proliferation in vitro. To determine the effect of NRP-1 expression on in vitro growth rates, MTT analysis was done. Control and pcDNA cells grew at similar rates ( Fig. 3 ). MTT analysis of siRNA clones showed no significant change in proliferation rate relative to that of control cells. The doubling time was 25.6 hours for parental cells, 24.6 hours for pcDNA-transfected cells, 28.3 hours for full-length NRP-1–transfected cells, 26.5 hours for NRP-1 deletion construct-transfected cells, and 25.4 hours for siRNA-transfected cells.
To ascertain whether the reduced proliferation in the NRP-1–overexpressing clones was due to increased apoptosis, we did flow cytometry analysis. No significant changes were observed in the sub-G0/G1 cycle, apoptosis, or in any other aspect of cell cycle profiles of each clone ( Table 1 ). These results suggest that the modest NRP-1–induced antiproliferative effect is only partially dependent on the VEGF/Sema3a domain and is not a result of increased apoptotic cell death.
NRP-1 overexpression down-regulates Akt and Erk-1/2 phosphorylation. Having shown that overexpression of NRP-1 (full-length and deletion constructs) modestly decreased cell proliferation and greatly decreased anchorage-independent cell growth, we determined the effects of NRP-1 on the signaling pathways that in part mediate these processes. Western blot analysis of whole cell extracts showed that all clones overexpressing the full-length NRP-1 or the NRP-1 deletion construct constitutively down-regulated both Akt phosphorylation (by >80%) and Erk-1/2 phosphorylation (by >50%) compared with controls ( Fig. 4 ). Clones also expressed increased total levels of Akt relative to parental and pcDNA-transfected clones. Because the mitogen-activated protein kinase pathway (as measured by Erk-1/2 phosphorylation) and the phosphotidylinositol-3′ kinase pathway (as measured by Akt phosphorylation) are associated with cell proliferation and survival, the down-regulation of these pathways is consistent with our phenotypic data. Moreover, these data confirm that these signaling changes do not require the NRP-1 VEGF/Sema3a domain.
Reduction of endogenous NRP-1 expression by small interfering RNA up-regulates Akt and Erk-1/2 phosphorylation. Because ectopic overexpression of molecules regulating signal transduction can lead to nonphysiologic results, we examined the effect of reducing endogenous NRP-1 expression by using siRNA technology. Immunoprecipitation and Western blot analysis of the siRNA clones SiC1 and SiC2 showed a >80% reduction in NRP-1 expression compared with controls ( Fig. 5 ). To ascertain the effect of reduced endogenous NRP-1 levels on signaling, we did Western blot analysis of signaling intermediates. Down-regulation of NRP-1 expression by siRNA led to large increases in phosphorylation of both Akt and Erk-1/2 (by >60%) compared with controls. Total levels of Akt protein were reduced in clones expressing both full-length and deletion isoforms of NRP-1. These results confirm that levels of NRP-1 in PANC-1 cells are inversely correlated with Akt and Erk-1/2 activities.
NRP-1 expression does not change expression levels of Sema3a or vascular endothelial growth factor. Because PANC-1 cells do not make the NRP-1 coreceptors Plexin-A1 or VEGF-R2, we sought to determine whether the NRP-1 ligands Sema3a and VEGF are produced and, if so, whether their expression changes in response to altered NRP-1 levels. VEGF expression (as determined by ELISA) was approximately the same in all clones ( Fig. 6A ). Western blot analysis of Sema3a expression in supernatants from controls (pcDNA), clones overexpressing full-length NRP-1 and the NRP-1 deletion construct, and clones with reduced NRP-1 expression via siRNA all expressed approximately the same level of Sema3a ( Fig. 6B).
NRP-1 represses migration by a vascular endothelial growth factor/Sema3a-independent mechanism. Results from a recent study suggested that in addition its role in angiogenesis, NRP-1 regulates cellular migration ( 15). To assess the effects of NRP-1 on the migration of PANC-1 cells, we used a standard Boyden chamber assay. Overexpression of full-length NRP-1 significantly reduced migration in all clones tested ( Fig. 7A and B ) compared with controls (0.4 ± 0.25, 0.8 ± 0.4, 20.6 ± 2.1, and 20.8 ± 2.9 migrated cells in C21, C26, parental, and pcDNA-transfected clones, respectively; P < 0.05). Clones expressing the NRP-1 deletion construct showed similar decreases in migration (0.2 ± 0.4 and 0.6 ± 0.8 migrated cells in DelC1 and DelC2 clones, respectively; P < 0.05). Conversely, reduction of NRP-1 expression by siRNA resulted in a significant increase in migration in both SiC1 and SiC2 clones relative to controls (69.4 ± 7.6 and 80.6 ± 7.7 migrated cells in SiC1 and SiC2 clones, respectively; P < 0.05). These data show that NRP-1 represses migration and that this inhibition does not require the VEGF/Sema3a domain of NRP-1.
NRP-1 expression reduces tumor growth in vivo. To discern the effects of NRP-1 expression in vivo, we s.c. injected control cells and PANC-1 transfectants expressing full-length NRP-1, the NRP-1 deletion construct, and NRP-1 siRNA clones into nude mice. The mice were monitored for 90 days, during which tumor growth was assessed periodically. Mice were sacrificed and tumor incidence and tumor volume were determined ( Fig. 8 ). All mice were of approximately the same weight when sacrificed (parental, 29.5 ± 1.4 g; pcDNA, 30.6 ± 1.8 g; NRP-1 C21, 30.1 ± 1.7 g; NRP-1 C26, 32.0 ± 1.5 g; SiC1, 29.3 ± 2.0 g; SiC2, 31.1 ± 1.8 g) Tumor incidence in controls (parental and pcDNA) was 50% and 60%, respectively ( Table 2 ). Tumor incidence was reduced in mice inoculated with full-length NRP-1 clones C21 and C26 and in mice inoculated with the deletion constructs DelC1 and DelC2. These results show that overexpression of NRP-1 significantly reduces tumor incidence compared with controls and that the VEGF/Sema3a domains are not responsible for this reduction. In marked contrast, tumor incidence was highest in the mice inoculated with the siRNA clones SiC1 and SiC2. These results show an inverse correlation between NRP-1 expression and tumor incidence.
In tumor-bearing mice, clones expressing full-length NRP-1 had an average tumor volume that was 50% to 90% smaller than controls, and clones expressing the NRP-1 deletion construct had an average tumor volume that was 70% to 80% smaller than controls ( Table 2). The average tumor volume of siRNA clones was 3- to 4-fold greater than controls (P < 0.05). These observations suggest that NRP-1 expression in pancreatic adenocarcinoma represses in vivo tumor growth independently of the VEGF/Sema3a domain and that down-regulation of endogenous NRP-1 expression reverses these effects.
In this study, we show that NRP-1 expression in PANC-1 cells suppressed cell motility and tumor growth and that these functions were independent of VEGF and Sema3a. The role of NRP-1 was assessed by overexpression of full-length NRP-1, expression of an NRP-1 mutant with the VEGF/Sema3a-binding domain deleted, and down-regulation of endogenous NRP-1 by siRNA technology. Overexpression of either isoform of NRP-1 significantly reduced soft-agar colony formation, moderately decreased proliferation, and reduced phosphorylation of Akt and Erk-1/2 in PANC-1 cells without affecting the level of VEGF expression. In marked contrast, reduction of endogenous NRP-1 by siRNA increased phosphorylation of these critical signaling intermediates. Because Akt and Erk-1/2 play central roles in cell survival, proliferation, and motility, their concomitant down-regulation by NRP-1 overexpression provides a biological mechanism that may account for the observed reductions of anchorage-independent growth, proliferation, and motility in PANC-1 cells ( 30– 32). To our surprise, expression of Akt increased with down-regulation of NRP-1. Interactions with proteins, such as Hsp90 and cdc37, are known to affect Akt half-life ( 33), and whether NRP-1 forms complexes with Akt that affects its stability remains to be determined. The effects of NRP-1 overexpression and of reduced endogenous NRP-1 on tumor cell motility were also examined. In our study, overexpression of the full-length NRP-1 and the NRP-1 deletion construct led to an almost complete loss of motility, whereas reduced NRP-1 expression by siRNA significantly increased motility. This repression cannot be attributed exclusively to an increased availability of Sema3a-binding sites because overexpression of NRP-1 constructs lacking the Sema3a-binding domain also significantly reduced migration.
Sema3a is produced by PANC-1 cells; however, its expression level does not change with NRP-1 expression. The NRP-1/Sema3a complex coreceptor Plexin-A1 is not expressed in PANC-1 cells. Consequently, NRP-1–dependent repression of migration seems to be independent of Sema3a. NRP-1 also inhibits migration in endothelial cells and breast carcinoma cells; however, in these cell types, inhibition depends on the concentration of Sema3a, which competes with VEGF for available NRP-1 ( 34, 35). One explanation for the different results from those studies and ours may be the presence or lack of VEGF-R2. Activated VEGF-R2 increases cell motility by destabilizing components of the cell adhesion complex and is present in endothelial cells and most breast carcinomas but not in most pancreatic adenocarcinomas, including PANC-1 ( 28, 36– 38). Excessive Sem3a may bind available NRP-1 and reduce activation of pathways promoting motility. Because PANC-1 cells lack both VEGF receptors, the repression mediated by NRP-1 may result through other ligands yet to be identified.
In our study, the repression of in vitro properties of tumorigenicity corresponded with the in vivo growth of these cells. Overexpression of either NRP-1 isoform decreased tumor incidence and volume, whereas the opposite effect was observed for siRNA clones in which endogenous NRP-1 was reduced. Thus, NRP-1 expression in PANC-1 cells in vitro and in vivo strongly inhibited tumorigenic properties independently of the VEGF/Sema3a-binding domains of NRP-1. We could not examine the affects of NRP-1 expression on blood vessel formation in vivo because of the drastic reduction in final tumor volume that accompanied NRP-1 overexpression.
Our results are distinct from those obtained from studies in prostate and breast carcinoma in which NRP-1 expression promotes tumor progression and survival primarily although VEGF-mediated pathways ( 17, 23). Those studies focused on the ability of NRP-1 to act as a coreceptor that increases the affinity of a ligand (e.g., VEGF) for a tyrosine kinase receptor (e.g., VEGFR-2). In the absence of VEGF receptors, NRP-1 may store or sequester VEGF165 and activate VEGF-dependent mechanisms on adjacent cells ( 6, 23). In our study, treatment of cells expressing either isoform of NRP-1 with antibodies that block VEGF-R1 or soluble VEGF-R1 showed no changes in Akt and Erk-1/2 phosphorylation (data not shown), which further supports a VEGF-independent function for NRP-1. Results from a recent study have shown that a chimeric NRP-1 protein can itself induce signals in human umbilical endothelial cells ( 15). Alternatively, an unidentified coreceptor that does not require the VEGF/Sem3a-binding domain may bind NRP-1. As a consequence, our data suggest that NRP-1 signaling is complex, and targeting this coreceptor for potential cancer therapy needs to consider both the tumor type and which coreceptors are expressed.
Grant support: NIH 2R01 CA65527 and DOD PC020017 (G.E. Gallick), NIH U54 CA 090810-01 (G.E. Gallick and L.M. Ellis), Lockton Foundation (M.J. Gray, G.E. Gallick, D.B. Evans, and L.M. Ellis), Lustgarten Foundation (L.M. Ellis), NIH T-32 09599 (J.S. Wey, J.T. Trevino).
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 Elizabeth L. Hess (Department of Scientific Publications, M.D. Anderson Cancer Center, Houston, Texas) for editorial assistance and Dr. Juinn-Lin Liu (M.D. Anderson Cancer Center, Houston, Texas) for the kind gift of the U251 glioblastoma cells.
Note: M.J. Gray is the Lockton Fellow for Pancreatic Cancer Research.
- Received July 10, 2004.
- Revision received January 13, 2005.
- Accepted February 24, 2005.
- ©2005 American Association for Cancer Research.