Neuropilin-1 mediates neutrophil elastase uptake and cross-presentation in breast cancer cells

Neutrophil elastase (NE) can be rapidly taken up by tumor cells that lack endogenous NE expression, including breast cancer, which results in cross-presentation of PR1, an NE-derived HLA-A2-restricted peptide that is an immunotherapy target in hematological and solid tumor malignancies. The mechanism of NE uptake, however, remains unknown. Using the mass spectrometry-based approach, we identify neuropilin-1 (NRP1) as a NE receptor that mediates uptake and PR1 cross-presentation in breast cancer cells. We demonstrated that soluble NE is a specific, high-affinity ligand for NRP1 with a calculated Kd of 38.7 nm. Furthermore, we showed that NRP1 binds to the RRXR motif in NE. Notably, NRP1 knockdown with interfering RNA or CRISPR-cas9 system and blocking using anti-NRP1 antibody decreased NE uptake and, subsequently, susceptibility to lysis by PR1-specific cytotoxic T cells. Expression of NRP1 in NRP1-deficient cells was sufficient to induce NE uptake. Altogether, because NRP1 is broadly expressed in tumors, our findings suggest a role for this receptor in immunotherapy strategies that target cross-presented antigens.

Neutrophil elastase (NE) 2 is a serine protease stored in the azurophil granules of neutrophils. NE plays a major role in host defense mechanisms and coordinates innate immune responses (1). NE also plays a key role in physiologic inflammation, as well as cancer metastasis and cell proliferation (2)(3)(4)(5)(6)(7).
Moreover, NE expression can predict the responses of trastuzumab, a monoclonal antibody that targets HER2, in metastatic breast cancer (8) and could be a prognostic marker (9). A number of studies have shown that NE is taken up by a variety of malignancies that lack endogenous NE expression (2, 10 -12). Houghton and co-workers (2,13) demonstrated that NE uptake promotes lung cancer cell proliferation by degrading insulin receptor substrate-1 and activating PI3K. We demonstrated that solid tumors, including melanoma and breast cancer, take up exogenous soluble and neutrophil-derived NE (10,12). In addition, we have shown that PR1, an HLA-A2-restricted peptide derived from NE and proteinase 3, was cross-presented on HLA-A2 breast cancer (BrCa) cells after co-culture with NE or proteinase 3 (10). PR1 is a well defined peptide that has been targeted clinically with a PR1-peptide vaccine (14 -16) and preclinically with an anti-PR1/HLA-A2 monoclonal antibody (8F4) and cellular therapies (17)(18)(19). Importantly, we have shown that PR1 cross-presentation on BrCa increased the susceptibility of the cells to lysis by PR1-specific cytotoxic T cells (CTLs) and 8F4 mAb (18).
Critically, we showed that NE uptake was dose-and time-dependent and was saturable in BrCa, which strongly suggests a receptor-mediated mechanism of uptake (12). NE was partially localized to an early endosome compartment 10 min after uptake, which further supports the presence of a receptor. Furthermore, Gregory et al. (2) provided evidence that NE binds to the surface of tumor cells and is internalized via clathrin-pit-mediated endocytosis. In that study, radioligand-binding experiments showed that NE binds to the surface of lung cancer cells and displayed a dissociation constant (K d ) of 284 nM, implicating the presence of a low affinity surface receptor for NE. Other investigators have also suggested the presence of an NE receptor, but to date no receptor has been identified (20,21). Based on this knowledge, and on our previous work demonstrating that NE is taken up by a variety of tumor types (10), we hypothesized that an NE receptor is expressed in different solid tumors and mediates uptake. Identification and characterization of the receptor could lead to a better understanding of tumor mechanisms of antigen cross-presentation and proliferation and, broadly, to the interactions of the innate and adaptive immune responses.
In the current study, we confirmed that NE is taken up via a receptor-mediated mechanism in solid tumor cells. Herein, using a combination of immunoprecipitation and MS techniques, we found neuropilin-1 (NRP1), a protein overexpressed by solid tumors (22), as the receptor for NE on the surface of MDA-MB-231 BrCa cells. We also showed that NE binds to NRP1 via its RRAR motif. NRP1 mediates NE uptake and subsequent cross-presentation of PR1, which is required for the specific lysis of MDA-MB-231 cells by PR1-CTL. Altogether, our data demonstrate a novel mechanism for NE uptake, which could lead to a better understanding of antigen cross-presentation by solid tumors and potential implications in immunotherapy.

Identification of NRP1 as a cell surface receptor for NE
Using the triple-negative (ER Ϫ /PR Ϫ /HER2 Ϫ ) breast cancer cell line MDA-MB-231, which does not express NE (8), we first assessed the ability of NE to bind to the cell surface. After incubating cells with soluble NE at 10 g/ml for 10 min at 4°C, we detected NE surface staining on MDA-MB-231 cells by FACS (Fig. 1A) and confocal microscopy (Fig. 1B). Furthermore, we found both intracellular and surface staining of NE when soluble NE was incubated with MDA-MB-231 cells at 37°C (Fig. 1,  C and D). To determine whether the catalytic activity of NE is required for its short-term uptake, we preincubated NE with elafin (6 kDa), a small endogenous reversible inhibitor of NE enzymatic activity, and PMSF, a nonspecific and irreversible serine protease inhibitor. The efficiency of the protease inhibitors was assessed using EnzChek enzyme assays and presented as supplemental Fig. S1 and supplemental Methods. As shown . Cell surface and intercellular NE were determined by flow cytometry (A and C) and confocal microscopy (B and D) using the anti-NE Ab conjugated to Alexa 647. Representative histogram and image from three independent experiments were shown. E-G, NE uptake is enzyme-independent and clathrin-and PI3K-dependent. MDA-MB-231 cells were incubated with soluble NE (10 g/ml) Ϯ elafin (10 g/ml) or PMSF (0.1 mM) for 10 min at 4°C (E) and 37°C (F). Intercellular (E) and cell surface (F) were determined by flow cytometry. G, MDA-MB-231 cells were pulsed for 10 min with NE (10 g/ml) Ϯ elafin (10 g/ml) after a preincubation with wortmannin or CPZ, presented as a percentage of inhibition ϭ [1 Ϫ (MFI of cells treated with inhibitors and ligand/MFI cells treated with ligands)] ϫ 100. **, p Ͻ 0.01; *, p Ͻ 0.05 (F). Representative histograms and images from at least three independent experiments are shown.
in Fig. 1, elafin and PMSF does not abolish the NE uptake ( Fig.  1E) or NE binding on cell surface (Fig. 1F). The data suggest the existence of a cell surface receptor that binds to NE and mediates internalization independently of its enzyme activity.
To determine whether NE is internalized by a clathrin-or PI3K-dependent pathway, MDA-MB-231 cells were pretreated with chlorpromazine (CPZ), known to inhibit clathrin-coated pit-mediated endocytosis, or wortmannin, an inhibitor of PI3K-dependent internalization (23). The cells were pulsed for 10 min with 10 g/ml NE. As observed with transferrin, NE internalization was significantly inhibited in the presence of CPZ and wortmannin (Fig. 1G). The addition of elafin did not reverse this inhibition. Our results strongly indicate the existence of a rapid receptor-mediated uptake mechanism that results in NE uptake via clathrin pits and independently of its enzyme activity.
Because our results in Fig. 1 indicated the existence of a NE receptor, we decided to identify the unknown receptor without bias. We immunoprecipitated NE-binding surface proteins with an anti-NE Ab on MDA-MB-231 cells, and the captured immunoprecipitation was analyzed using nano-HPLC LC-MS/ MS (24). As shown in Fig. 2 (A and B), the tryptic peptide from NE and the binding partner NRP1 were clearly identified on the mass spectra of the NE cross-linked complex. Among the seven cell surface proteins identified by mass spectra, only NRP1 and LRP12 were transmembrane receptors (Fig. 2C). Using the protein database generated after global surface protein biotinylation followed by LC-MS to identify and quantify proteins expressed on cell surface (24), we found NRP1 to be an enriched surface protein in MDA-MB-231 cells, whereas LRP12 was less abundant (Fig. 2D). Furthermore, no difference between LRP12 expression on the surface and total cell extract was observed (Fig. 2D). Notably, NRP1 is also expressed on dendritic cells (25), which have the ability to cross-present soluble antigens including NE (26).

Specificity of NE binding to NRP1
First, to confirm the direct binding of NE to NRP1, we coated microtiter wells with human recombinant NRP (Phe22-644) and used ELISA to determine the specificity of binding of soluble NE. As shown in Fig. 3A, NE bound to NRP1 in a dose-dependent manner but not to the negative control ovalbumin (OVA). The affinity of NE binding to the NRP1 was measured by biolayer interferometry with a calculated K d of 38.7 nM (Fig. 3B). NRP1 has previously been described as an isoform-specific VEGF-165 receptor (27), and thus VEGF-165 was used as a positive control. We found that both NE and the VEGF-165 bound human NRP1, whereas no binding of ovalbumin was

Neuropilin-1 mediates NE uptake and cross-presentation
detected. As expected, we found that NRP1 binds to VEGF-165 at a K d value within a double-digit nanomolar range (63.1 nM), indicative of a strong interaction with NRP1 (Fig. 3C). Interestingly, NE with a K d of 38.7 nM demonstrated a stronger interaction with NRP1 than VEGF-165 (Fig. 3C). Taken together, our data demonstrate that NE is a specific and high-affinity ligand for NRP1.
Various peptides that expose the consensus sequence (R/K/ X)X(R/K) at the C terminus (C-end rule) were shown to bind to NRP1 and to internalize in the cells (28). More precisely, a systematic study on mutation of this sequence demonstrated that RRAR, RDAR, RPDR, RPRR, and RPPR are able to bind NRP1 (29). To validate NE binding, we first check the presence of such a motif inside the X-ray structures of NE (PDB entry 3Q76) by using Swiss PDB viewer software. X-ray crystallography indicated that a consensus sequence RRAR (in blue) is contained within an extracellular ␤-strand of NE protein (Fig. 4A). To characterize and validate the domain of NE interactions with NRP1, ligand-binding assays by ELISA were carried out using two peptides: NE (I30 -50R) RRAR motif and another sequence that corresponds to an intracellular ␤-strand, NE (F199 -220R) (in pink, Fig. 4A). As shown on Fig. 4B, the presumed peptide, NE (I30 -50R) , binds to His-NRP1 (F22-644K) , whereas NE (F99 -220R) does not. To investigate the differences of interaction between different analogs and His-NRP1 (F22-644K) , we use some alanine substitutions replacing Arg at position 34 (R34A), 35 (R35A), 37 (R37A), and the triple mutant R34A,R35A,R37A (Fig. 4C). We also test the effect of the replacement of the Ala in position 36 by another inert amino acid Gly (A36G) on NRP1 binding (Fig.  4C). Single substitutions R34A, R35A, and R37A or combined (R34A,R35A,R37A) significantly decrease the binding of the peptide to NRP1 (Fig. 4D), whereas the Ala substitution with Gly (A36G) was not sufficient to abolish the binding to NRP1 (Fig. 4D). These results indicate that NE is able to bind specifically to NRP1 via the consensus sequence RRXR.

NE binds NRP1 on the surface of the MDA-MB-231 cells, and subsequently, this binding is followed by NRP1 internalization
To investigate whether NE binds specifically to NRP1 on the cell surface, we used the CRISPR-cas9 system to knock out NRP1 in MDA-MB-231 cell line (Fig. 5A). The validation of NRP1 knockdown in MDA-MB-231 by CRISPR-Cas9 is presented as supplemental Fig. S2 and supplemental Methods. The absence of NRP1 on the surface of the cells appears to be sufficient to abolish NE surface staining after 10 min of treatment with soluble NE at 4°C (Fig. 5, B and C).
We used Amnis Image Stream to investigate whether NE induces NRP1 internalization. As shown in Fig. 5D, supplemental Fig. S4, and supplemental Methods, NRP1 was localized on the cell surface and then internalized in the cytosol after 10 min of incubation with soluble NE. To quantify the localization of NRP1, the internalization score was calculated as the ratio of the amount of intracellular fluorescence versus the total amount of fluorescence. The results clearly demonstrated that NRP1 signal was intracellular after NE treatment (Fig. 5E). We concluded that NE treatment induced efficient internalization of NRP1 in MDA-MB-231 cells.

NRP1 mediates NE uptake on various breast cancer cell lines
To determine its role in NE uptake, we transfected the MDA-MB-231 cells with siRNA or shRNA against NRP1. The loss of NRP1 expression was confirmed by Western blot (Fig. 6, A, B, D, and E). The absence of NRP1 expression resulted in a significant decrease in NE uptake in MDA-MB-231 cells (ϳ50% versus si-control or sh-control) (Fig. 6, A and B). Furthermore, we incubated MDA-MB-231 cells with anti-NRP1 antibody prior

Neuropilin-1 mediates NE uptake and cross-presentation
to incubating with NE and found significantly decreased NE uptake (ϳ60% versus isotype control) (Fig. 6C). A similar effect was also seen in Hs 578T BrCa line treated with siRNA against NRP1 (Fig. 6D). Conversely, overexpression of NRP1 in the T47D BrCa line that lacks endogenous NRP1 resulted in a significant increase in NE uptake (3.4-fold increase in MFI versus pCMV-control) (Fig. 6E). Although we showed that NRP1 was implicated in NE uptake in various BrCa cell lines, we could not compare the variation in NE uptake between the cell lines. Indeed, the variability observed between the different cell lines, especially between the T47D and MDA-MB-231 cell lines, is explained by the fact that we performed in-house fluorophore conjugations of anti-NE antibody. These reactions often give variable conjugation efficiencies that could account for vari-able MFI. In summary, our results demonstrate that NRP1 mediates NE uptake in various breast cancer cell lines.

NRP1 is necessary for breast cancer cell susceptibility to specific lysis by PR1-CTL
We previously showed that the NE-derived peptide PR1 is cross-presented on MDA-MB-231 cells after NE uptake, leading to specific cell lysis by immunotherapies that target PR1/ HLA-A2, including PR1-specific CTLs and 8F4 Ab (10,18). In light of this, we sought to determine the role NRP1-mediated uptake of NE in the cross-presentation of PR1. Cytotoxicity assays showed that NE-treated MDA-MB-231 cells were susceptible to killing by PR1-CTLs generated from various healthy In blue is shown RRAR motif contained in NE (I30 -50R) , and in pink is shown the negative peptide control NE (F199 -220R) used for this study. B, binding of the recombinant human NRP-1, His-NRP1 (F22-644K) , to immobilized NE peptide NE (Iso30 -50) by ELISA. NE (I30 -50R) peptide contains the RRAR motif. NE (F199 -220R) was used to evaluate for nonspecific binding. The results were expressed as the mean of absorbance value obtained from three experiments in triplicate. C, binding of recombinant human NRP-1 to the immobilized RRAR sequence (NE (I30 -50R) ) and its analogs. The results were expressed as the mean of absorbance value obtained from one representative experiment of three, realized in triplicate.

Neuropilin-1 mediates NE uptake and cross-presentation
donors (Fig. 7, A and B), confirming our previous observations (10). However, MDA-MB-231 cells lacking NRP1 were not lysed by PR1-CTL derived from the same donor, suggesting that NRP1 is implicated in cross-presentation in MDA-MB-231 cells (Fig. 7, A and B). Our results clearly demonstrate that NRP1 is involved in NE uptake and cross-presentation by BrCa, which is necessary for the specific lysis by PR1-CTL.

Discussion
In this report, we have shown that NRP1 mediates NE uptake by BrCa. More interestingly, we identified NRP1 as an endocytic receptor that transports soluble NE into MDA-MD-231 cells by immunoprecipitation and LC-MS. Direct binding of NE to NRP1 was confirmed by ELISA and bio-layer interferometry. More precisely, we demonstrated that NE interacts with NRP1 through the RRAR sequence contains in its N terminus. Utilizing interfering RNA and the CRISPR-cas9 system, we showed that NRP1 is a receptor for NE binding and uptake on various BrCa cell lines. Moreover, we demonstrated that PR1 crosspresentation by BrCa and cell lysis by PR1-CTL are dependent on NRP1.
Our published data showed that NE is co-localized with EEA-1 following uptake by BrCa cell lines and subsequently traffics in the endosomes (12). We showed that NE internalization is temperature-sensitive and is significantly decreased in presence of drugs known to inhibit receptors internalization such as PI3K and wortmannin (23) (Fig. 1). In this study, we identified NRP1 as a NE receptor (Fig. 2). NRP1 is known to act as a co-receptor for numerous extracellular ligands and receptors, such as SEMA3A/4A (30) and growth factor receptors VEGF-165 (31) and transforming growth factor-␤1 (32). Interestingly, we showed that NE binds to NRP1 with a similar affinity (K d ϭ 38 nM) to its natural ligand VEGF-165 (K d ϭ 63 nM) (Fig. 3, B and C). The binding observed for NE in our study accounts for the physiological binding and is supported by published reports showing NE blood concentrations up to 1 M during inflammatory states (11,33,34). It has also been proven that VEGF-165 interacts through the motif CRCDKPRR in its C terminus (27,35). Recent studies revealed that a consensus sequence (R/K)XX(R/K), found in the C-terminal region of different peptides, is required for binding to NRP1 and internalization (28). By analyzing the crystallography structure (Fig. 4A), we found that NE has a unique RRAR sequence that might explain its ability to bind to NRP1. To validate this sequence, we performed alanine substitution of each residue of the peptide (Fig. 4C). We found that the Arg in positions  34,35, and 37 are essential for NE binding to NRP1 (Fig. 4C). These data are in agreement with the results of Zanuy et al. (29), which demonstrated that a RRAR mutant presents high affinity for the b1 domain of NRP1. However, the RRXR is contained in an internal loop and not in the C-terminal region of NE, and the RRXR motif is followed by a proline that can potentially create an unfavorable conformation for the binding to NRP1. We therefore compared the binding of the C-terminal peptide from VEGF-165 (VEGF-165 (R170 -191R) ), NE peptide (Iso30 -50) , and intact NE at equimolar concentrations (supplemental Fig. S3 and supplemental Methods). We demonstrated that the intact NE and NE peptide (Iso30 -50) bind similarly to NRP1 (supplemental Fig. S3C). Surprisingly, we found that the NE peptide (I30 -50R) binds NRP1 with higher avidity than VEGF-165 (R170 -191R) (supplemental Fig.  S3, A and B).

Neuropilin-1 mediates NE uptake and cross-presentation
Our laboratory previously reported that NE-derived peptide PR1 is cross-presented on MDA-MB-231 cells after NE uptake, leading to their susceptibility to killing by PR1-CTL (10). In this study, we demonstrated that NRP1 silencing in MDA-MB-231 cells was sufficient to suppress PR1 cross-presentation and subsequent killing by PR1-CTL. Furthermore, whereas NE treatment in MDA-MB-231 WT was able to increase the PR1/ HLA-A2 presentation on the cell surface as detected by the 8F4 Ab, the knockdown of NRP1 by si-NRP1 or CRISPR KO NRP1 significantly decreased the cross-presentation of PR1/HLA-A2 to a level similar to that seen in the untreated controls (data not shown). These data suggest that NRP1 is required for PR1/ HLA-A2 cross-presentation by NE. Nevertheless, NE uptake was not completely abrogated in MDA-MB-231 cells after NRP1 knockdown, suggesting that another receptor may also be capable of internalization (Fig. 5, A-D). These data indicate that NRP1, although likely not the sole receptor for NE uptake, does play a critical role in the uptake of NE and subsequent antigen processing necessary for cross-presentation (Fig. 7, A  and B). In contrast to NRP1-mediated internalization, which we have shown to correlate with uptake and cross-presentation, this alternate uptake pathway could potentially be involved in NE-related proliferative effects, as described in numerous studies (2, 3, 11, 36).

Neuropilin-1 mediates NE uptake and cross-presentation
Recent work provides evidence that tumor-associated neutrophils are prognosticators, especially in breast and lung tumors (37). By calculating the Pearson's correlation coefficients from individual patient-derived samples using the TCGA database, we demonstrated a moderate to strong positive relationship between CD16, a cell-surface marker of neutrophils that we used as a surrogate marker for NE, and NRP1 mRNA expression in different tumor types (data not shown). Although there is a preponderance of data that link inflammation, tumor-associated neutrophils, and NRP1 with poor outcomes in cancer (9,38,39), our published data (10) along with our observations here point to NRP1 as a potential molecule that could be targeted to favorably alter the tumor microenvironment and the innate immune system to generate an effective anti-tumor immune response.
Furthermore, several studies demonstrated that NRP1 is overexpressed in solid tumors compared with normal samples (40 -44), indicating NRP1 as a potential therapeutic target. Clinical studies of patients also showed significant correlation between NRP1 expression and clinical staging.

Experimental procedures
Cell culture MDA-MB-231, Hs 578T, and T47D cells were obtained from American Type Culture Collection (Manassas, VA). All cell lines were cultured in Dulbecco's modified Eagle's medium/ F-12 or in RPMI 1640, supplemented with 10% fetal bovine serum (v/v) (Sigma-Aldrich), 100 units/ml penicillin, and 100 g/mg streptomycin (Sigma-Aldrich) and maintained in 5% CO 2 at 37°C. Cell lines were authenticated by short tandem repeat methodology at the Characterized Cell Line Core Facility at MDACC. Healthy donor apheresis samples were obtained from Gulf Coast Regional Blood Center and MDACC Blood Bank. Peripheral blood mononuclear cells were isolated using standard Histopaque 1077 gradient centrifugation (Sigma-Aldrich).

Immunoprecipitation of cell surface receptor for NE
MDA-MB-231 cells (1 ϫ 10 8 ) were treated with 80 g of NE (Athens Research and Technology) for 30 min at 4°C and then incubated with a cross-linker according to the manufacturer's instructions (Pierce Protein Biology by Thermo Scientific). The cells were washed and lysed in nondenaturing lysis buffer (20 mM Tris base, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% CHAPS, 20 mM iodoacetamide, and 0.2 mM PMSF) for 1 h at 4°C. Insoluble material was removed by centrifugation at 13,200 rpm for 10 min. Lysates (1.5 mg) were incubated overnight at 4°C with protein A/G-agarose that had been preconjugated with anti-NE antibody (clone NP-57, sc-53388; Santa Cruz Biotechnology) or mouse IgG1 isotype. The immune complexes were washed four times with ice-cold PBS. Immunoprecipitates were eluted with 2ϫ Laemmli buffer (120 mM Tris-Cl, pH 6.8, and 4% SDS) and boiled at 100°C for 10 min. Eluted fractions were desalted using GL-Tip SDB (GL Sciences, CA). Samples were reduced in 15 mM Tris (2-carboxyethyl) phosphine hydrochloride, alkylated with acrylamide (7:1 ratio), and treated with trypsin before being subjected to mass spectrometry analysis.

Mass spectrometry analysis
Tryptic peptides were analyzed by an Orbitrap ELITE MS (Thermo Scientific) coupled to an EASY-nLC 1000 nanoflow chromatography system (Thermo Scientific). The LC separation was performed in a 25-cm, 75-m-inner-diameter column packed with 3 m C18 resin (Column Technology Inc.) using a 100-min B pump linear gradient from 5-50% in 300 nl/min, where mobile phase A (0.1% formic acid in 2% acetonitrile) and mobile phase B (0.1% formic acid in 98% acetonitrile) were used. The mass spectrometry was used with MS1 spectrum resolution of 60,000, automatic gain control target of 1 ϫ 10 6 , and maximum injection time of 150 min. The top 20 precursors were selected for MS2 analysis. MS2 was done by collisioninduced dissociation fragmentation with automatic gain control of 3 ϫ 10 4 , maximum injection time of 10 min, normalized collision energy of 35, isolation width of 2.0 m/z, activation q value of 0.25, and activation time of 10 min. The LC-MS/MS data were processed through Proteome Discoverer version 1.4 (Thermo Scientific) using Sequest HT search engine and Uniprot 1308 database with a false discovery rate of less than 5%, estimated by Target Decoy PSM Validator. The search parameters included Cys alkylated with propionamide (71.03711@C) as a fixed modification and Met oxidation (15.99491@M) as a variable modification. Semi-tryptic search with a maximum of two missed cleavage sites was allowed, and precursor mass tolerance of 10 ppm and fragment mass tolerance of 0.5 Da were applied.

ELISA
For the evaluation of the binding of NE to NRP1, 96-well Maxisorb Immunoplates (Nunc) were coated with human recombinant NRP1 (His-NRP1 (Phe22-644) ) (0.5 g/ml; R&D Systems), blocked with buffer containing 1% w/v BSA in PBS, and incubated for 1 h with a serial dilution (2-fold) of NE (1 g/ml) or OVA at 37°C. The plates were then washed and incubated with anti-NE antibody (NP-57) or ovalbumin antibody for 30 min at room temperature. Bound antibodies were detected by incubation with horseradish peroxidase-conjugated goat anti-mouse antibodies (1:2000; Jackson Immuno-Research Laboratories) followed by 3,3Ј,5,5Ј-tetramethylbenzidine substrate (Becton Dickinson). Parallel ELISA with EndoGrade OVA (Hyglos) was used as the negative control to demonstrate the specificity of NE binding to recombinant NRP1. For characterizing the motif contained in NE that binds NRP1, 1 g/ml of NE (Iso30 -50) (WT) and its analogs were coated and blocked with 1% w/v BSA in PBS. His-NRP1 (Phe22-644) (0.5 g/ml; R&D Systems) was added to each well and incubated for 1 h. The same experiment was realized on a plate coated with BSA to subtract the unspecific binding. After washing the plate, the bound NRP1 was detected with an antibody HRP-conjugated His antibody (Life Technologies). Signal was developed with 3,3Ј,5,5Јtetramethylbenzidine (TMB), and the optical density determined at 450 nm.

Flow cytometry and confocal microscopy
For surface staining, the cells were incubated with NE in 1% BSA in PBS at the indicated times, temperature, and concentrations of NE (Athens Research and Technology). Internalization studies were performed in serum-free RPMI 1640. Inhibition studies were performed with cells that had been preincubated for 1 h at 37°C with different concentrations (from 0 to 5 g/ml) of anti-NRP1 antibody (AF 3870; R&D Systems) or isotype control (sc-2717; Santa Cruz Biotechnology). Surface staining was performed for 30 min on ice with anti-NE antibody (clone NP-57, sc-53388; Santa Cruz Biotechnology) directly conjugated to AlexaFluor-647 or anti-NRP1 antibody followed by phosphatidylethanolamine-conjugated secondary antibody (sc-3751; Santa Cruz Biotechnology). For intracellular staining, the cells were permeabilized in perm/wash buffer (BD Biosciences) prior to antibody staining. The data were acquired using a BD LSR Fortessa flow cytometer and were analyzed using FlowJo software (Tree Star, Ashland, OR).
For NRP1 internalization, the cells were analyzed using Amnis ImageStream Mark II Imaging Flow Cytometer (EMD Millipore Corporation) with 40ϫ objective. The results were analyzed by IDEAS image analysis software. A mask representing the whole cell was defined by the bright-field image, and an internal mask was defined by eroding the whole cell mask by 9 pixels, which were eroded to measure only the internalized part. The values of the internalization score were calculated for at least 500 cells/experimental condition. Confocal imaging was performed using a Leica Microsystems SP2 SE confocal microscope (Buffalo Grove) with 63ϫ/1.4 oil objective. Leica LCS software (version 2.61) was used for image analysis.

Neuropilin-1 mediates NE uptake and cross-presentation
accordance with the manufacturer's protocol. The transfection efficiency was confirmed by Western blot analysis of NRP1 48 h after transfection. The cells transduced with lentiviruses (neuropilin shRNA lentiviral particles, sc-36038-V; control shRNA lentiviral particles-A, sc-108080-V; Santa Cruz Biotechnology) were selected in 2 g/ml puromycin (Sigma-Aldrich) for 3 weeks. The MDA-MB-231 KO CRISPR NRP1 was generated using the CRISPR-cas9 system based somatic knock-out technology at the Gene Editing/Cellular Model Core Facility from MDACC.

Biolayer interferometry by ForteBio Octet RED384
Biolayer interferometry experiments for binding kinetics were performed on a ForteBio Octet RED384 instrument using anti-human Fc capture (18-5060) biosensor (Pall Fortebio). NRP1-Fc (NR1-H5252; ACRO Biosystems) as a ligand was loaded to the anti-human Fc capture biosensor. Following a baseline incubation in 10ϫ kinetics buffer (10 mM phosphate, 150 mM NaCl, 0.02% Tween 20, 0.05% sodium azide, 1 mg/ml BSA, pH 7.4), the loaded biosensors were exposed to a series of analyte concentrations for VEGF165 (VE5-H5248; ACROBiosystems), NE (Athens Research and Technology) and OVA (VWR International) for an association step. Dissociation was monitored in 10ϫ kinetics buffer, and background subtraction was used to correct for sensor drifting. All experiments were performed with shaking at 1,000 rpm at 30°C. ForteBio data analysis software was used to fit the data to a 1:1 binding model to extract association and dissociation rates. The K d was calculated using the ratio k off /k on .

CTL assay
Peripheral blood mononuclear cells from healthy HLA-A2 ϩ donors were isolated and stimulated with PR1 peptide (Biosynthesis Inc.) in vitro, and calcein-AM cytotoxicity assays were performed as previously described (5,9). Briefly, 1000 target cells were labeled with 10 g/ml calcein-AM (Invitrogen) for 15 min at 37°C. The cells were washed and co-cultured with PR1-CTL (effector cells) at indicated effector:target ratios in a 60-well Terasaki plate. After 4 h of incubation, trypan blue was added, and fluorescence was measured using a Cytation 3 imaging reader (Biotek, Winooski, VT).

Statistical analysis
Comparisons between groups were carried out using Student's t test and one-way analysis of variance with the Tukey's multiple comparison test. The p value of 0.05 was set as the criterion for significance. The analyses were performed using Prism 6.0 software.
Author contributions-C. K. was responsible for conception and design, acquisition and analysis of data, and writing the article; S. C. T., H. K., and S. M. H. were responsible for acquisition of data and analysis and interpretation of data content; P. S., K. E. R., K. C.-D., A. V. P., A. A. P., N. Q., D. Z., J. M. M., and C. T. were responsible for development of methodology; A. S., S. L., K. C.-D., and H. L. P. were responsible for conception and design; G. A., L. S. S. J., S. L., E. A. M., A. C. H., and Q. M. were responsible for analysis and interpretation of data and revising the paper; J. J. M. was responsible for study supervision, conception and design, analysis and interpretation of data, and final approval of the paper.