N2E4, a Monoclonal Antibody Targeting Neuropilin-2, Inhibits Tumor Growth and Metastasis in Pancreatic Ductal Adenocarcinoma via Suppressing FAK/Erk/HIF-1α Signaling

Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive malignancy with extremely limited treatment; the effective targeting strategy stays an urgent unmet need. Neuropilin-2 (NRP2), a multifunctional transmembrane non-tyrosine-kinase glycoprotein, enhances various signal transduction pathways to modulate cancer progression. However, the application value of NRP2 as a therapeutic target in pancreatic cancer is still unclear. Here, we detected the elevated NRP2 was associated with the poor prognosis of pancreas carcinoma. The mouse monoclonal antibody targeting NRP2 (N2E4) that could specifically bind to PDAC cells was developed. Moreover, N2E4 inhibits PDAC proliferation, migration, and invasion in vitro, and repressed growth and metastasis in vivo. Mechanistically, the effect of N2E4 was mainly related to the blocking of interaction between NRP2 with integrinβ1 to inhibit FAK/Erk/HIF-1a/VEGF signaling. Therefore, N2E4 has the potential for targeting therapy of PDAC. This study lays a foundation for the future development of NRP2-based targeted therapy for PDAC.

However, little progress arises in the targeting treatment of PDAC compared to other cancers. Therefore, scientific trials to evaluate new targets for PDAC remain unmet medical needs.
Neuropilins (NRPs), transmembrane non-tyrosine-kinase glycoproteins, are the receptors of class-3 semaphorins, including two highly conserved homologous members, neuropilin-1 (NRP1) and neuropilin-2 (NRP2) (5). A great deal of researches shows that NRP2 is upregulated and associated with unfavorable prognosis in a variety of tumors (6,7). As multifunctional membrane proteins, NRP2 enhances or modifies signal transduction pathways through interactions with various receptors or ligands to modulate cancer progression (8). For example, NRP2 interacts with and functions as a coreceptor for the a6b1 integrin in breast cancer cells, which facilitates a6b1 signaling and activates focal adhesion kinase (FAK) (9,10). In particular, these studies demonstrate that NRPs can promote vascular endothelial growth factor (VEGF) signaling without VEGF receptor (VEGFR) involvement by coopting integrin signaling (11). Furthermore, JJ Ou et al. found that the integrin/FAK/Erk pathway induces NRP2 activation, promoting tumor progression (12). The Erk cascades are not only mainly considered to be correlated with cancer cell proliferation, survival, and actin remodeling, but also upregulate the expression of hypoxia-inducible factor-1 (HIF-1) followed by VEGF (13). The upregulated VEGF then acts on the same cell via an autocrine manner to further enhance VEGF action (14).
According to Caunt, et al., blocking NRP2 function inhibited metastasis of breast cancer and glioma cells (15). Although not as intensely explored for functions of NRP-2 in PDAC, it is reliable that NRP-2 is involved in biological processes of pancreatic cancer such as cell survival and invasion (16). These findings suggest that NRP2 is a potential target for cancer therapy, but it is unclear whether PDAC therapy targeting NRP2 can improve tumor progression or cancer metastasis.
Given these findings, we have successfully obtained a monoclonal antibody targeting NRP2 (N2E4) that stably binds to the NRP2 b1b2 fragment by genetic engineering and hybridoma technology in the previous study (17). Large-scale production and purification (>95%) were established in our laboratory to guarantee to be sufficient for the study. In this work, we comprehensively assessed the expression of NRP2 in pancreatic cancer, identified the N2E4 target protein (NRP2) in PDAC cells, and further elucidated the positive therapeutic effect of N2E4 in vitro and in vivo and the underlying mechanisms. We provided evidence that N2E4 had the potential application value in the treatment of PDAC, which offered a theoretical basis for targeting therapy of pancreatic cancer.

GEPIA Database
Gene Expression Profiling Interactive Analysis (GEPIA) database is a visualized big data cancer analytics platform based on the cancer genome atlas (TCGA) and the genotype-tissue expression (GTEx). In this study, the difference of neuropilin-2 mRNA (NRP2) level between tumor with normal tissues in PDAC, and the relationship between NRP2 levels with the survival prognosis of pancreatic cancer was analyzed by "Expression Plots" and "Survival Plots" modules in the GEPIA database respectively.

Immunohistochemistry
Tissue arrays (HPan-Ade120Sur-01) were purchased from Shanghai Outdo Biotech Co. Ltd (Shanghai, China), including 63 pairs of pancreatic carcinoma tissues and adjacent paracarcinoma tissues. The antigen retrieval was conducted in a microwave oven with the citrate buffer. Arrays were incubated in endogenous peroxidase blocking solution (reagent A, P003IH, Auragene) for 10 min at room temperature and blocked by normal non-immunized goat serum (reagent B) for 10 min at room temperature. Tissues were incubated with the NRP2 antibody (Affinity, DF7604) at 4°C overnight. Biotinconjugated (reagent C) secondary antibody and HRP-labeled streptavidin (reagent D) were implemented in turn. After diaminobenzidine (DAB) signals generated, the slides were counterstained with hematoxylin and observed by the micro digital sectioning scanning system (Motic VM1, Hong Kong).
The relative expression levels of mRNAs were calculated using the 2 -(DCt sample-DCt control) method. Experiments were performed in triplicate.

Immunofluorescence
Cells were cultured on specialized culture dishes (Sorfa, 201100) overnight at 37°C and fixed with 4% paraformaldehyde. After blocked by PBST with 2% BSA, slides were incubated using the primary antibody at 4°C overnight, then the fluorochromeconjugated secondary antibody at 37°C for 1 h in the dark and Hoechst 33258 (B1155, Sigma) at room temperature for 5 min in the dark. The fluorescence images were acquired using a confocal microscope (Olympus FV1000MPE-B, Japan).

CCK8 Assays
To assess cell proliferation capability, the Cell Counting Kit-8 (CCK8) was obtained from Yeasen Biotech Co., Ltd. (Shanghai, China). Cells were seeded in 96-well plates (2~5 × 10 3 cells/well) and maintained for 24 h, 48 h, and 72 h with different doses of mAb. CCK8 solution was then added into each well followed by incubation for 2 h. Finally, absorbance was measured at 450 nm using a SPECTRAmax Microplate Reader (model 680, Bio-Rad, Tokyo, Japan). Each experiment was independently repeated three times.

Scratch Assays
Scratch assay was performed on 90% confluent cells and inoculated in 24-well plates. Scratches were established using sterile 200-ml pipette tips gently. Then cells were cultured with fresh medium containing 400 mg/ml N2E4 or corresponding PBS. When cells migrated into the scratches at 24 h, images of the scratch areas were captured by phase-contrast microscope (Olympus FV1000, Japan) and analyzed by ImageJ software. Each scratch coverage was estimated in quadruplicate at four different locations.

Matrigel-Coated Invasion Assays
The cell suspension containing 5.0 x 10 4 cells/ml was prepared in the serum-free medium. Subsequently, 600 ml of complete medium containing 10% fetal bovine serum was added to the lower chamber. 500 ml of cell suspension described above was added to each upper chamber (Corning 3422, USA) to the culture at 37°C for 48 h. Cells below the membranes were stained by 0.5% crystal violet for 20 min at room temperature, washed with water, and air-dried. The numbers of invaded cells were observed and counted under the microscope (Olympus BX53, Japan).

Immunoprecipitation (IP)/Co-IP
The cells were lysed on ice with 1×lysis buffer containing 1 mM PMSF and centrifuged at 13,000 g for 30 min. The supernatant was then collected. The total protein was incubated with 1 mg NRP2 antibody and 10 mL protein A/G beads overnight at 4°C, C-9 antibody and protein A/G beads as the positive control, IgG and protein A/G beads as the negative control. Wash the protein A/G-beads 3 times. After the last washing, 1× Lysis Buffer was mixed with the same volume of 2×SDS sample buffer and then boiled at 100°C for 10 minutes. The mixture was analyzed by SDS-PAGE electrophoresis. Silver staining or western blotting was performed.

ELISA
The supernatant culture medium from the co-culture was collected and cytokine levels were measured separately. The supernatant was centrifugated at 12,000 × g for 10 min at 4°C following being collected at 12 h, 24 h, and 48 h. This was then used to measure the VEGF level using a Human VEGF-A Elisa kit (EY-00H100681, EYBIO, Shanghai, China), according to the manufacturer's protocol.

Tumor Xenograft Model
All animals were applied under guidelines established by the Animal Care and Use Committee of Xiamen University. Female BALB/c nude mice were gained from the Animal Center of Xiamen University (Fujian, China) and kept in a specific pathogen-free barrier facility indoor. 5 × 10 6 cells were injected subcutaneously into the back of the right forelimb of 4~6-weekold female BALB/c nude mice. When tumor volume reached 40 70 mm 3 , estimated as V = L × W 2 /2 (V: Tumor volume; L: Length; W: Width), mice were randomly divided into three groups with 5 animals in each group. Then mice were treated with PBS (vehicle control), low dose (20 mg/kg), and high dose (40 mg/kg) of N2E4 once daily via tail vein injection. Tumor volume and body weight were measured three times per week for 2∼3 weeks. After mice were euthanized using CO 2 , tumor tissues were isolated and measured.

Luciferase Reporter Assay
The BALB/c nude mice were injected about 1 × 10 6 luciferaselabeled BxPC-3 cells into the lateral tail vein. Mice were anesthetized and injected with D-luciferin potassium salt (E011306, ENERGY, Shanghai, China) for bioluminescent imaging weekly. Images were captured, and the bioluminescent signal was quantified using IVIS Lumina II Imaging System (Caliper, USA). The mice were sacrificed after 8 weeks. The lungs were isolated for hematoxylin and eosin (H&E) staining.

Statistical Analysis
Statistical analysis was implemented via SPSS 17.0 (SPSS Inc, Chicago, IL, USA) and GraphPad Prism8 (GraphPad Software Inc., San Diego, CA, USA) statistical software. Statistical data were expressed as mean ± SD, and the data represent at least three separate experiments. For comparisons of two treatment groups, Student's t-test was used as appropriate. P < 0.05 was seen as statistically significant.

Elevated Expression of NRP2 Is Associated With Unfavorable Prognosis in Pancreas Carcinoma
We assessed the role of NRP2 in pancreatic carcinoma via the GEPIA database. The level of NRP2 was significantly upregulated in pancreatic cancer tissues compared with normal pancreatic tissues (P< 0.05) ( Figure 1A). The survival analysis showed that Overall Survival and Disease-Free Survival ( Figure 1B) of pancreatic cancer patients with high NRP2 levels were significantly lower than patients with low NRP2 levels (p < 0.05).
Given the above findings, we further analyzed the level of NRP2 in 60 pairs of pancreatic carcinoma tissues and adjacent normal tissues via immunohistochemistry. The data confirmed that NRP2 expression was significantly higher in PDAC ( Figure 1C). The staining intensity score was higher in PDAC tissues than para-carcinoma tissue (P < 0. 0001, Figure 1D). Strongly positive (+++) NRP2 expression was confirmed in 55.36% of PDAC tissues and 17.31% of para-carcinoma tissues, weekly positive (+) NRP2 expression was evaluated in only 14.29% of PDAC tissues and 51.92% of para-carcinoma tissues ( Figure 1E). The ROC curve was created by sensitivity and specificity of NRP2 scores and provided the AUC (0.7887) to distinguish pancreatic cancer from normal tissues (95% CI: 0.7028 to 0.8745; P < 0.0001). The point close to (0.0, 1.0) on the curve maximized sensitivity (84.2%) and specificity (65.5%) ( Figure 1F). Collectively, the level of NRP2 was significantly raised in pancreatic carcinoma tissues and had higher sensitivity and specificity to differentiate pancreatic cancer from paracarcinoma tissues. Therefore, NRP2 could represent an attractive therapeutic target for PDAC.

The N2E4 Antibody Specifically Binds to NRP2 in PDAC Cells
To detect NRP2 levels in three pancreatic ductal adenocarcinoma cell lines, RT-PCR was conducted. The level of NRP2 mRNA in BxPC-3 cells was the highest, followed by PANC-1 then MIA PaCa-2 ( Figure 2A). Flow cytometry analysis indicated that the binding capability of N2E4 to NRP2 proteins of cell surface was consistent with NRP2 mRNA levels of three cell lines. The increase of affinity was reflected in a conspicuous fluorescent profile offset in the case of N2E4 ( Figure 2B). Therefore, BxPC-3 with high NRP2 level and MIA PaCa-2 with low NRP2 level were selected and verified using western blotting. The results showed the levels of NRP2 proteins were much higher in BxPC-3 than MIA PaCa-2, and N2E4 exhibited more significant abundance differences suggesting the affinity of N2E4 could reach the level of the commercial polyclonal antibody H-300 (sc-5542, Santa) ( Figures 2C, D). After immunoprecipitation, the N2E4-protein complex was precipitated by protein A/G beads and then separated on an SDS-PAGE gel. A specific band of around 110 kDa was detected in PDAC cells compared with the IgG control and commercial NRP2 monoclonal antibody C-9 (sc-13117, Santa), which proved that the N2E4 target protein in PDAC cells was NRP2 ( Figure 2E). Immunofluorescence assay displayed the binding of N2E4 with BxPC-3 cells with strong FITC-fluorescence intensity ( Figure 2F). Thus, the N2E4 antibody was specifically combined with NRP2 in PDAC cells.

N2E4 Inhibits Proliferation, Migration, and Invasion in BxPC-3 Cells
To determine the effects of the N2E4 antibody on PDAC cells (MIA PaCa-2 and BxPC-3), cell viability, migration, and invasion were measured using cell biology assays. CCK-8 assay displayed that the cell viability of MIA PaCa-2 was hardly changed, but there was slightly decreased after 48 and 72 h of N2E4 treatment (400 mg/ml) in BxPC-3 cell lines ( Figure 3A). The IC50 values of BxPC-3 cells for N2E4 were calculated using GraphPad and were 698.1 mg/ml, 426.9 mg/ml, and 399.9 mg/ml at 24h, 48h, and 72h, respectively. The results of scratch assays presented that the migration capacity of BxPC-3 cells was significantly inhibited after 24 h of N2E4 treatment compared with the PBS group ( Figures 3B, C). Similarly, the invasion capacity of BxPC-3 cells was visibly repressed after 48 h of N2E4 treatment using Matrigelcoated assays (Figures 3D, E). These data demonstrated that N2E4 contributed to the inhibitions of proliferation, migration, and invasion in BxPC-3 cells.

N2E4 Affects the Formation of Pseudopods and the E-Cadherin/ N-Cadherin Switch in EMT
PDAC is characterized by aggressive migration and invasion (18,19). The increased pseudopods caused the aggressive movement of cancer cells. Lee et al. found formations of pseudopods and stress fibers contributed to cancer cell migratory and invasive ability and tumor metastasis in orthotopic lung cancer animal models (20). In this work, we observed strong skeletal structures, abundant stress fibers, and filamentous pseudopodia (marked with green arrows) in the PBS group via rhodamine-labeled phalloidin. In contrast, the N2E4 treatment group showed minimal pseudofoot protrusions, blurred skeleton structures, and reduced stress fibers ( Figure 4A). Besides, EMT is the initial cause of tumor invasion and metastasis, one of which is characterized by the loss of epithelial markers (e.g. Ecadherin) and the gain of mesenchymal markers (e.g. N-cadherin) (21,22). To demonstrate whether N2E4 referred to the regulation of EMT, western blotting and cellular immunofluorescence assays were performed. Compared with the PBS group, the N-cadherin level decreased gradually over time, while the level of E-cadherin increased slightly under the N2E4 treatment ( Figures 4B-D). And the results of the immunofluorescence (Figures 4E, F) were consistent with those of Western blotting analysis. These data supported that N2E4 inhibited the formation of pseudopods and EMT processes in PDAC cells.

N2E4 Attenuates the Activation of FAK/ Erk/HIF-1a by Blocking the Interaction Between NRP2 With Integrinb1
To detect whether N2E4 inhibits the interaction between NRP2 and integrinb1, we verified the co-localization of NRP2 and integrinb1 using immunofluorescence and detected the interaction intensity between them using Co-IP. As was depicted in Figure 5A, significant co-localization of NRP2 with integrinb1 was exhibited in BxPC-3 cells. The reduced interaction between NRP2 and integrinb1 under the treatment of N2E4 was further confirmed by Co-IP ( Figure 5B). Moreover, western blotting was proceeded to examine the changes in protein levels after N2E4 treatment in BxPC-3 cells. We found that phosphorylated FAK and Erk1/2 remarkably decreased with the prolonged action of N2E4, while total FAK, Erk1/2, or Akt and phosphorylated Akt level remained stable (Figures 5C-F). It prompted that the activation of FAK and Erk1/2 was suppressed with the treatment of N2E4. As Erk is involved in the regulation of synthesis and transcriptional activation of HIF-1a (23), we hypothesized that N2E4 might have a role in the transcription and expression of VEGF. So western blotting was firstly performed to analyze whether or not N2E4 was able to affect the expression of HIF-1a. Treatment with N2E4 could effectively reduce the level of HIF-1a protein ( Figure 5G). Secondly, we analyzed the VEGF mRNA expression level by RT-PCR and the autocrine VEGF protein level using ELISA. The data indicated the levels of both VEGF mRNA ( Figure 5H) and autocrine protein ( Figure 5I) were inhibited after N2E4 treatment. The above results indicated that N2E4 might attenuate the activation of FAK/Erk/HIF-1a by blocking the interaction between NRP2 with integrinb1, then inhibit VEGF signal transduction. vein injection ( Figure 6A). And the results showed that N2E4 prompted a more significant growth inhibition effect on BxPC-3 xenograft. The inhibition effect for tumor volume was more evident with a high dose (40 mg/kg) of N2E4, P < 0.05 ( Figure 6B). In terms of tumor weight, two treatment groups showed significant inhibitory effects (P < 0.01) with a nearly 40% inhibitory rate, however, the two doses did not have different effects ( Figure 6C). According to Burvenich et al., the receptor occupancy to the antibody-drug in vivo correlates with antitumor efficacy, and the saturation of receptors corresponds to maximal antitumor efficacy in vivo (24), Therefore, we speculated that the antibody dose of 20 mg/kg might approach the saturation state of NRP2 receptors. By contrast, it was not statistically significant for inhibition effect in tumor growthmediated by MIA PaCa-2 ( Figures 6D, E). These results indicated that N2E4 suppressed the NRP2-associated growth of pancreatic cancer cells in vivo, consistent with the results in vitro.

N2E4 Inhibits the Growth and Metastasis of Pancreatic Cancer In Vivo
All the mice were well tolerated with stable body weight and without obvious toxicity signs during the study period. Furthermore, a xenograft nude mouse model was established to further explore whether N2E4 could restrain metastasis of PDAC in vivo. The BxPC-3 cells labeled luciferase were injected into the tail vein of nude mice to simulate the spread of cancer cells throughout the blood. All mice survived until the end of the experiment. The heightened metastasis was observed in mice without N2E4 treatment; however, metastasis was markedly inhibited in mice from the N2E4 treatment group ( Figure 6F). The average radiance of lungs in the N2E4 treatment group was lower compared with the control (P < 0.05), indicating decreased metastasis ( Figure 6G). Also, the number of lung metastasis nodules was significantly reduced in the N2E4 treatment group using H&E ( Figure 6H). The data indicated that N2E4 attenuated the metastasis of pancreatic cancer cells.

DISCUSSION
In this study, the expression of NRP2 is elevated and associated with unfavorable prognosis in pancreas carcinoma (Figure 1). Based on our previous report, we further confirm that the N2E4 monoclonal antibody specifically combines with the NRP2 in PDAC cells (Figure 2). Our findings suggest that N2E4 inhibits proliferation, migration, invasion (Figure 3), and represses growth and metastasis of pancreatic cancer cells in vivo ( Figure 6).
In the migration of pancreatic cancer, cells are often powered by filament pseudopodia involved in the extension in the direction of cell movement (25)(26)(27). Ouyang  cancer cells compared with doxorubicin, which was associated with suppression of filopodia formation by these agents (28). Therefore, through the specific fluorescence staining of the cytoskeleton, it is clear that with N2E4 treatment, skeletal structures under the cell membrane become blurred, filamentous pseudopods and stress fibers absent gradually ( Figure 4A), which could mean the aggressive movement and EMT weakens (22). Besides, the increased epithelial phenotype markers (E-cadherin) and the decreased mesenchymal phenotype marker (N-cadherin) under N2E4 treatment further confirmed our hypothesis ( Figures 4B-F).
The change in cell stiffness determined by the cytoskeleton is a crucial feature of cancer cells that affects the metastatic ability. Zou et al. demonstrated that actin cytoskeleton remodeling led to decreased cell stiffness via integrin/FAK/Erk Pathways, enhancing the migration of mesenchymal stem cells (29). Also, NRP2 is necessary for the association of a6b1-integrin with the cytoskeleton and also facilitates integrina6b1-mediated activation of FAK (10). We demonstrated that N2E4 inhibited the activation of FAK/Erk by blocking the interaction between NRP2 with integrinb1 in BxPC-3 cells using Co-IP and western blotting, but Akt signaling inhibition is not involved in the mechanism ( Figures 5A-F). Erk is not only involved in the regulation of HIF-1a synthesis but also its transcriptional activation. The phosphorylated Erk could increase the transcriptional activation function of HIF-1a, and thus stimulates the biological function of VEGF (23), for instance, survival and metastasis of pancreatic cancer (16). Interestingly, our data revealed that the expression of HIF-1a was attenuated with the N2E4 treatment ( Figure 5G), and VEGF mediated signal transduction might be further inhibited ( Figures 5H, I).
One limitation of our research is that pancreatic cancer characteristically contains dense stroma in the tumor tissue, which might affect the drug response and resistance. However, mimicking the dense stroma by xenograft tumor is relatively complicated, and the genetically-engineered mouse model might have an advantage on this issue. Besides, for the treatment of PDAC with highly therapy-resistant (30), it is crucial to Bars are the means ± SD, n = 3. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. determine which patients will benefit most from an approach and tailor precise drugs and treatments to individual patients (31). For example, the PD-L1 expression level has been accepted as the companion diagnostic (for lung cancer) or the complementary detection (for melanoma and bladder cancer) on anti-PD-1 treatments (32). As the treatment requires a direct interaction between NRP2 and N2E4, N2E4 shows more practical effects on pancreatic cancer cell lines and xenografts with high NRP2 expression. Therefore, the expression of the receptor and its assessment would constitute vital steps for clinical application, which is likely significant for the development of precision medicine.
Although this is merely the tip of the icebergs in assessing the application potential of NRP2 pathway blockade by N2E4, these findings suggested that N2E4 could restrain interaction between NRP2 with integrinb1 to inhibit FAK/Erk/HIF-1a/VEGF signaling, limiting growth and metastasis of pancreatic cancer ( Figure 7). Actually, antibody drugs have two main physiological activities in vivo: one is to recognize and specifically bind antigen substances through the Fab segment, causing the neutralization or apoptosis of the antigen. The second is the immunological effect of antibodies, for instance, antibody-dependent cellmediated cytotoxicity (ADCC). The Fc segment of the antibody has effector cell ligand binding site and can bind to FcgRIIIA on the surface of NK cells (effector cells) to mediate the ADCC action (33). It is worth mentioning that Liu et al. found that NRP1 determines how T cells develop and establish immune memory, and knocking out NRP1 can enhance the immune memory ability of T cells, and the immune response is more robust when the tumor cells are "seen" again (34). Therefore, the role of NRP2, which is highly homologous to NRP1, in memory immunity may be worth looking forward to. Or maybe the inhibitory effect of N2E4 on pancreatic cancer is also associated with enhanced immune response.
In conclusion, as a unique therapeutic monoclonal antibody inhibiting the crucial signaling pathway, N2E4 suppresses PDAC cell proliferation, migration, invasion in vitro, and significantly inhibited tumor growth and metastasis in vivo. Consequently, this work suggests that N2E4 has the potential for targeting therapy of PDAC, contributes to understanding the NRP2

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by Animal Care and Use Committee of Xiamen University.

AUTHOR CONTRIBUTIONS
LW, JY, FL, and TW conceived the idea. LW designed research. LW and LLW performed experiments. SW and ZZ purified antibodies. LW, PX, and XL collected data and interpreted the results. LW and ZL wrote the paper. All authors contributed to the article and approved the submitted version.