PlexinB1 Inactivation Reprograms Immune Cells in the Tumor Microenvironment, Inhibiting Breast Cancer Growth and Metastatic Dissemination

Breast cancer (BC) is one of the most frequent tumor types and the most common malignancy in women (1). The tumor microenvironment (TME) is essential to dynamically regulate cancer progression, and due to its influence on the therapeutic outcome, multiple therapies directed to TME components have been developed in the last few years. In particular, several new anticancer approaches target immune cells, e.g., cytotoxic T lymphocytes (to unleash their antitumor properties) or myeloid cells (to block their immune-suppressive activities; ref. 2). A crucial step in the antitumor immune response is the recruitment and activation of adaptive immune cells. In fact, tumors characterized by higher levels of pro-inflammatory cytokines and T-cell infiltrate display a better response to immunotherapy, e.g., immune checkpoint inhibitors (ICI; ref. 3). Hence, turning a “cold” tumor into a “hot” one (in terms of inflammation) has been the focus of many recent efforts in the field. Even though BC is not generally considered a highly infiltrated tumor, the triple-negative BC (TNBC) subtype is characterized by a consistent inflammatory infiltrate and could potentially benefit from immune checkpoint blockade. In line with that, immunotherapy is currently approved as the first line of treatment for a subset of patients diagnosed with advanced TNBC (4).

Semaphorins comprise a large family of evolutionarily conserved proteins. Originally identified as signaling cues for axon navigation, semaphorins are involved in the regulation of diverse biological and pathological processes, from developmental angiogenesis to bone homeostasis and immune responses (5). Plexins comprise a large family of transmembrane receptors for the semaphorins. Accumulating evidence indicates that a number of so-called “immune” semaphorins play a major role in the pathogenesis of immunological diseases and control cancer-associated inflammation, thus representing potential therapeutic targets of interest (6). SEMA4D is a transmembrane semaphorin initially described for its role in the immune system, but it has also been associated with neural development, angiogenesis, and cancer growth (7). Moreover, preclinical data suggest that SEMA4D is a key regulator of immune cells in the TME (8, 9). PLXNB1, the main canonical receptor for SEMA4D, is expressed by cancer cells and by multiple players in the TME. For instance, it has been found to control the migration of immature dendritic cells (10) and to sustain B-cell proliferation and survival (11). PLXNB1 expression is also found in endothelial cells (EC), whereby SEMA4D promotes the angiogenic process (12). However, the functional relevance of PLXNB1 in the TME has been so far poorly characterized. Moreover, the intracellular tail of SEMA4D seems to have a signaling role via poorly studied mechanisms (13).

The SEMA4D-targeting monoclonal antibody pepinemab has been tested for clinical use in patients with cancer (14). However, recent findings raised concerns about the safety of targeting SEMA4D in tumors as they revealed increased metastatic spreading in a pancreatic cancer model, possibly due to antibody-mediated retrograde SEMA4D signaling in macrophages (15). In this context, targeting PLXNB1 could represent an alternative way to tackle this signaling axis and to overcome resistance to therapy. Thus, in this study, we evaluated the relevance of PLXNB1 in regulating the TME and tumor progression in mouse models of TNBC. Herein, we unveiled a role for PLXNB1 in the regulation of immune cells in the TME, identifying a potential target to enhance immunotherapy in TNBC.

4T1 cells (ATCC Cat# CRL-2539, RRID:CVCL_0125) were purchased from the ATCC in 2015 and cultured in RPMI medium (EuroClone, #ECB9006L) supplemented with 1% L-glutamine (EuroClone, #ECB3000D), 10% FBS (Gibco, #10270-098), and 1% penicillin and streptomycin (EuroClone, #ECB3001D). Py230 cells (ATCC Cat# CRL-3279, RRID:CVCL_AQ08) were purchased from ATCC in 2020 and cultured in F-12K Medium (Kaighn’s Modification of Ham’s F-12 Medium, ATCC, #30-2004) supplemented with 0.1% MITO+ serum extender (Corning, #355006), 5% FBS (Gibco, #10270-098), and 1% penicillin and streptomycin (EuroClone, #ECB3001D). EMT6 cells (ATCC Cat# CRL-2755, RRID:CVCL_1923) were purchased from ATCC in 2023 and cultured in Waymouth’s MB 752/1 Medium (Gibco, #11220-035) supplemented with 15% FBS (Gibco, #10270-098) and 1% penicillin and streptomycin (EuroClone, #ECB3001D). SVEC4-10 cells (RRID:CVCL_4393) were purchased from ATCC in 2005 and cultured in DMEM (EuroClone, #ECB7501L) supplemented with 1% L-glutamine (EuroClone, #ECB3000D), 10% FBS (Gibco, #10270-098), and 1% penicillin and streptomycin (EuroClone, #ECB3001D). HEK-293T cells (RRID: CVCL_0063) were purchased from ATCC in 2019 and cultured in IMDM (Gibco, #21980) supplemented with 1% L-glutamine (EuroClone, #ECB3000D), 10% FBS (Gibco, #10270-098), and 1% penicillin and streptomycin (EuroClone, #ECB3001D). All cell lines were thawed from the initial expansion of the stock, without reauthentication, and used within 10 passages. All cell lines were cultured in a humidified incubator with 5% CO₂ at 37°C. Raji cells and Jurkat cells expressing mFcgRIV were included in the Promega kit (Promega mFcγRIV ADCC Bioassay Kit, #M1201), purchased in 2023; they were used without expansion according to the vendor’s request and cultured in RPMI1640 containing 4% low IgG serum (provided with the kit). PlxnB1_4C6 cells derive from a cell clone of Expi293F cells (Thermo Fisher #A14527 RRID:CVCL_D615) stably expressing human PLXNB1 that was established in 2015 as described in ref. 16. The cells, after reviving from the frozen stock, were passaged 20 to 30 times over 3 months, cultured in DMEM (Wako Pure Chemical Co. #041-30081), 10% FCS (Gibco, #10270-098), 1 mg/mL G418 (Nacalai Tesque #16512-52), and 5 μg/mL puromycin (Sigma #P7255), and validated for the high expression levels of human PLXNB1 by flow cytometry. All cell lines were regularly tested for Mycoplasma contamination.

Six- to eight-week-old female WT BALB/c and C57BL/6 mice were purchased from Charles River. BALB/c Plxnb1^−/− (C.Cg-Plxnb1^tm1.1Ltam/Cnrm RRID:IMSR_EM:12796), C57BL/6 Plxnb1^−/− (B6.129-Plxnb1^tm1.1Ltam/Cnrm RRID:IMSR_EM:12800), and BALB/c Sema4d^−/− (C.cg-Sema4d^tm1Kik/Cnrm RRID:IMSR_EM:12799) mice were previously generated and described (17, 18). 1 × 10⁶ 4T1, 4 × 10⁶ Py230, or 2 × 10⁵ EMT6 cells were orthotopically surgically implanted in the fourth mammary fat pad under anesthesia with 2.5% isoflurane. Tumor growth was monitored by caliper measurement twice a week, and tumor volume was calculated with the formula (a² × b) × 0.52, where a and b stand for the minor and major tumor diameters, respectively. Lung macro-metastasis was counted after counterstaining the airways with Indian ink. For pharmacological treatments, the mice were randomly assigned to different cohorts and treated as described below. Anti-CD8 (Bio X Cell Cat# BE0117, RRID:AB_10950145) was administered intraperitoneally at a dosage of 200 µg at days −1, 0, +1, and +8 (with respect to cancer cell transplantation). Mice were sacrificed at day 15, following the approved animal experimentation protocols, in order to avoid severe adverse effects and the acceleration of tumor progression due to CD8⁺ cell depletion. Anti PD-1 (Bio X Cell Cat# BE0146, RRID:AB_10949053) was administered intraperitoneally at a dosage of 250 µg every 4 days starting from day 7. For anti-PLXNB1 in vivo treatment, an antagonistic PLXNB1-dimerizing protein reported previously [Fc(m6A9)B3; ref. 19] was used. This protein is a human IgG1 Fc-based homodimeric protein grafted with a PLXNB1-binding peptide m6A9 (WRPYIERWTGRLIV) at the tip of E–F loop of the CH3 domain (20). Purified Fc(m6A9)B3 or Fc control was administered intravenously once a week at a dosage of 400 µg/mice, starting 4 weeks post Py230 tumor inoculation or 7 days after EMT6 tumor inoculation. Py230-bearing mice were sacrificed after 3 weeks of treatment with Fc(m6A9)B3, whereas EMT6-bearing mice were sacrificed after 2 weeks. For survival experiments, mice bearing 4T1 and Py230 tumors were euthanized when tumors reached the volume of 700 mm³ and 500 mm³, respectively. Biochemical analysis of renal and hepatic function parameters and the hematological profiling of mice was outsourced to the analytical laboratory at the “Clinica Veterinaria Città di Torino”. Whole blood was withdrawn from the submandibular vein at sacrifice. All animal procedures were approved by the Ethical Committee of the University of Turin (Candiolo, Turin, Italy) and by the Italian Ministry of Health and were conducted in compliance with European laws and policies.

Tumors from 4T1 or Py230 cells were dissected and collected. Tissues were fresh frozen in OCT (Killik, Bio-Optica, #05-9801) and 10-μm-thick frozen sections were serially cut using a Leica 2135 cryostat. Dried tumor slices were fixed with Zn fixative (0.5 g/L Ca acetate; 5 g/L Zn acetate; 5 g/L Zn chloride; 0.1 mol/L Tris buffer) for 10 minutes and permeabilized with Triton X-100 (Sigma-Aldrich) 0.1% PBS solution. After saturation with 2% goat serum (Vector-D.B.A. #S1000) at room temperature for 1 hour, tumor slices were incubated overnight with optimized primary antibody dilution, washed, and incubated with appropriate Alexa Fluor-conjugated secondary antibody (Thermo Fisher Scientific Cat# A-21206, RRID:AB_2535792; Cat# A-21434, RRID:AB_2535855; Cat# A-31572, RRID:AB_162543; Cat# A-21208, RRID:AB_2535794; Cat# A-21110, RRID:AB_2535759; Cat# A-21247, RRID:AB_141778; Cat# A-31573, RRID:AB_2536183) for 45 minutes. After staining with DAPI (Roche #10236276001), tumor slices were mounted, and images were acquired with Leica TCS SPE II-DM5500 CSQ confocal microscope equipped with a 20× oil immersion objective lens (ACS APO 20×/0,60 IMM CORR) using Leica LAS AF software. Images (at least four fields/tumor) were acquired with a 20× objective and analyzed using (NIH) ImageJ (RRID:SCR_003070) software. For IF staining, the following primary antibodies were used: anti-CD68 1:100 (Bio-Rad Cat# MCA1957, RRID:AB_322219), anti-CD206 1:100 (R&D Systems Cat# FAB2535G, RRID:AB_10971285), anti-iNOS 1:50 (Abcam Cat# ab15323, RRID:AB_301857), anti-Meca32 1:100 (BD Biosciences Cat# 550563, RRID:AB_393754), anti-Ng2 (Millipore Cat# AB5320, RRID:AB_11213678), anti-CD8 1:100 (Thermo Fisher Scientific Cat# 14-0081-82, RRID:AB_467087), anti-CA9 1:100 (Novus Cat# NB 100-417, RRID:AB_350323), anti-ki-67 1:100 (Thermo Fisher Scientific Cat# MA5-14520, RRID:AB_10979488), anti-CD11c 1:100 (Bio-Rad Cat# MCA1369, RRID:AB_324490), anti-CD8 1:100 (Thermo Fisher Scientific Cat# 14-0081-82, RRID:AB_467087), and anti-GRZB 1:50 (Abcam Cat# ab255598, RRID:AB_2860567).

Mouse tumors were cut into small pieces, disaggregated with 1.5  mg/mL type IV collagenase (Worthington, #LS004186) and 100 μg/mL DNase (Sigma-Aldrich, #D5025), and filtered through strainers. 1 × 10⁶ cells were stained with specific antibodies. Tumor-draining lymph nodes were harvested and filtered to obtain a single-cell suspension through a strainer before staining. Cell permeabilization and fixation were performed with the FOXP3/Transcription Factor Staining Buffer Set (eBioscience #00-5523-00). Flow cytometry was performed using the BD LSR Fortessa 20× and analyzed with FlowJo V10 (RRID:SCR_008520) software (absolute cell counts are displayed in Supplementary Table S1). Phenotype analysis was performed with the following antibodies: Fixable Viability Stain 780 (BD Horizon, #565388, RRID:AB_2869673), BV786 Rat Anti-Mouse CD45 (BD Biosciences Cat# 564225, RRID:AB_2716861), BB700 Rat Anti-Mouse Ly-6G (BD Biosciences Cat# 566435, RRID:AB_2739730), BV421 Rat Anti-Mouse Ly-6C (BD Biosciences Cat# 562727, RRID:AB_2737748), BV480 Rat Anti-CD11b (BD Biosciences Cat# 566117, RRID:AB_2739519), PE-CF594 Rat Anti-Mouse F4/80 (BD Biosciences Cat# 565613, RRID:AB_2734770), BB700 Armenian Hamster Anti-Mouse CD3e (BD Biosciences Cat# 566494, RRID:AB_2744393), BB515 Rat Anti-Mouse CD8a (BD Biosciences Cat# 564422, RRID:AB_2738801), BV480 Rat Anti-Mouse CD4 (BD Biosciences Cat# 565634, RRID:AB_2739312), and APC Rat Anti-Mouse IFNγ (BD Biosciences Cat# 554413, RRID:AB_398551). Additionally, the following antibodies were used in some analyses: BUV395 Rat Anti-Mouse CD45 (BD Biosciences Cat# 564279, RRID:AB_2651134), BB700 Armenian Hamster Anti-Mouse CD3e (BD Biosciences Cat# 566494, RRID:AB_2744393), BV605 Rat Anti-Mouse CD8a (BD Biosciences Cat# 563152, RRID:AB_2738030), BUV737 Rat Anti-Mouse CD4 (BD Biosciences Cat# 612844, RRID:AB_2870166), BV480 Rat Anti-Mouse CD25 (BD Biosciences Cat# 566202, RRID:AB_2739593), PE Mouse Anti-Mouse Foxp3 (BD Biosciences Cat# 566881, RRID:AB_2869932), APC Rat Anti-Mouse Granzyme B (Thermo Fisher Scientific Cat# 17-8898-82, RRID:AB_2688068), BV711 Mouse Anti-T-bet (BD Biosciences Cat# 563320, RRID:AB_2738136), BV421 Mouse Anti-GATA3 (BD Biosciences Cat# 563349, RRID:AB_2738152), BUV563 Rat Anti-CD11b (BD Biosciences Cat# 741242, RRID:AB_2870793), PE-CF594 Rat Anti-Mouse F4/80 (BD Biosciences Cat# 565613, RRID:AB_2734770), PE Rat Anti-Mouse CD206 (Thermo Fisher Scientific Cat# MA5-16871, RRID:AB_2538349), FITC Mouse Anti-iNOS/NOS Type II (BD Biosciences Cat# 610331, RRID:AB_397721), and BUV737 Hamster Anti-Mouse CD11c (BD Biosciences Cat# 612797, RRID:AB_2870124).

IHC for CD8 was performed on formalin-fixed, paraffin-embedded lung sections (10-μm thick) from 4T1 WT and Plxnb1^−/− mice. Paraffin sections were deparaffinized with xylene and rehydrated with decreasing concentrations of ethanol in water, and the sections were then pretreated with 1% H₂O₂ in methanol for 30 minutes. Antigen retrieval was achieved by heating sections in sodium citrate buffer (pH 6.0) for 6 minutes in a 750‐W microwave oven. Tissues were incubated with rabbit anti-CD8 1:100 (Novus Cat# NBP2-29475, RRID:AB_2904552) overnight at 4°C. Samples were then washed and incubated for 1 hour with an anti-rabbit HRP-conjugated secondary antibody (Dako #K4003), and antigens were revealed with 3,3-diaminobenzidine (Dako #K3468) according to the manufacturer’s instructions. Sections were counterstained with hematoxylin (Bio-Optica #05-06002L) and visualized with MorphoLens 6 (Morphle) slide scanner at 40×. The expression levels of CD8 were quantified by ImageJ software (RRID:SCR_008520) by analyzing at least four fields/lungs.

Tumor-infiltrating cell populations were isolated by magnetic separation. Briefly, murine explanted tumors were cut into small pieces, disaggregated with 1.5 mg/mL type IV collagenase (Worthington, #LS004186) and 100 μg/mL DNase (Sigma-Aldrich, #D5025), and filtered through strainers. CD4⁺ and CD8⁺ lymphocytes were isolated together after the single-cell suspension was incubated with CD8/CD4 (TIL) Microbeads (Miltenyi, #130-116-480) and magnetically separated using LS Columns (Miltenyi, #130-042-401) accordingly to the manufacturer’s instructions. CD8⁺, CD4⁺, CD31⁺, CD11b⁺, and F4/80⁺ cells were separated using Miltenyi’s #130-116-478, #130-116-475, #130-097-418, #130-126-725, and #130-110-443 microbeads, respectively.

T cells were isolated from WT or Plxnb1^−/− spleens using the Pan T Cell Isolation Kit II (Miltenyi, #130-095-130) accordingly to the manufacturer’s instructions. T-cell activation was performed with the mouse T Cell Activation/Expansion Kit (Miltenyi, #130-093-627) in TexMACS Medium (Miltenyi #130-097-196) supplemented with 10% FBS and Mouse IL2 IS 50 U/mL (Miltenyi #130-120-332) for 10 minutes for subsequent Western blot analysis or for 48 hours for subsequent gene expression analysis.

RNA was extracted with Maxwell RSC miRNA Tissue Kit (Promega, #AS1460). cDNA preparation was performed according to standard procedures with High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, #4368814). Gene expression was evaluated by Real-Time PCR using the Taqman Gene Expression Master Mix (Applied Biosystems, #4369016) and the following Taqman probes: PLXNB1 (Mm00555359_m1), SEMA4D (Mm00443147_m1), actin (Mm01205647_g1), PDCD1 (Mm01285676_m1), IL1a (Mm00439620_m1), IL6 (Mm00446190_m1), NOS2 (Mm00440502_m1), TNF (Mm00443258_m1), Arg1 (Mm00475988_m1), CD86: (Mm0044540_m1), IL10 (Mm01288386_m1), MRC1 (Mm01329359_m1), CXCR3 (Mm99999054_s1), GRZB (Mm00442837_m1), IFNg (Mm01168134_m1), Perf1 (Mm01168134_m1), IL2 (Mm00434256_m1), IL12 (Mm0043169_m1), IL4 (Mm00445259_m1), IL5 (Mm00439646_m1), and IL13 (Mm00434204_m1). An average of 30 to 90 ng of template/reaction was used, and the samples were run in triplicate on QuantStudio 7 Pro (Applied Biosystems). Gene expression fold change variations, normalized to β-actin levels and then to a reference control condition, were calculated by the 2^−ΔΔCt method. The expression of Plxnb1 and Sema4d mRNAs in different cell types was normalized to the amount of β-actin housekeeping transcripts in each individual sample, by the ΔCt method. The relative counts of specific gene transcripts/10⁴ β-actin mRNA molecules (per each sample and cell type) were then calculated by the formula: 2^−ΔCt × 10⁴. Gene expression in TILs (isolated as described in “Isolation of tumor-infiltrating cell populations”) of targets known to have implications in immune responses was assessed using TaqMan Mouse Immune Array (Applied Biosystems, #4367786). Differential gene expression was considered significant with fold change >2 and P < 0.05.

4T1, Py230, and EMT6 cells were seeded into 96-well cluster plates (1.500 cell/well) in 5% FBS. Cell viability was evaluated at 24, 48, and 72 hours using the CellTiter-Glo Luminescent Cell Viability Assay (Promega, #G7573) according to the manufacturer’s recommendations and using the VICTOR X Multilabel Plate Readers (Perkin Elmer).

Femurs and tibiae of 6-week-old wild-type or Plxnb1^−/− female BALB/c mice were used to generate bone marrow-derived macrophages (BMDM). Cells were flushed by centrifuging the bones cut at the knee joint. The isolate was filtered through a 40 μm cell strainer, and red blood cells were lysed with RBC Lysis Buffer (BioLegend, #420301). The harvested cells were plated in RPMI medium (EuroClone, #ECB9006L) supplemented with 1% L-glutamine (EuroClone, #ECB3000D), 10% FBS (Gibco, #10270-098), 1% penicillin and streptomycin (EuroClone, #ECB3001D), and 50 ng/mL recombinant mouse CSF (Bio-Techne #416-ML). BMDMs were polarized at day 7 for M1-like polarization using 100 ng/mL LPS (Sigma-Aldrich Cat# L5668) and 200 U/mL mouse IFNγ (Miltenyi, #130-105-785). To obtain the M2-like phenotype, we used 20 ng/mL mouse IL-4 (Sigma-Aldrich Cat# I1020). BMDMs were polarized for 24 hours.

Nonreplicating viral particles containing the targeted shRNA for Plxnb1 TRCN0000078913 from Sigma Mission Library (or empty vector pLKO.1 noncoding plasmids) were produced in HEK-293T (RRID: CVCL_0063) packaging cells by co-transfection of the transfer vector, of the packaging plasmids pCMVΔR8.74, of the pRSV.REV plasmid, and of the vesicular stomatitis virus (VSV) envelope plasmid pMD2.VSV-G using the calcium phosphate precipitation method as previously described (21). Target cells were then incubated with the conditioned media derived from transfected HEK-293T (RRID: CVCL_0063) cells, in the presence of 8 μg/mL polybrene (Merk, #H9268-5G), for 8 to 12 hours. Red fluorescent protein (RFP)-expressing 4T1 cells were generated by infection with lentiviral particles encoding Turbo-RFP (BPS Bioscience # 78347-P).

For in vivo extravasation assays, 1 × 10⁶ RFP-expressing 4T1 cells were injected into the lateral mouse tail vein of 6- to 8-week-old WT and Plxnb1^−/− BALB/c mice. Mice were sacrificed 48 hours after injection, and their lungs were perfused with PAF 4% (ChemCruz; #sc-281692). RFP detection was performed with LEICA DMI3000 B fluorescence microscope equipped with a 20× objective lens (Leica, HI PLAN 20×/0,40 PH1) and acquired with Photometrics Cool SNAP HQ camera. Quantification of metastatic cells in the lungs was done by analyzing at least four microscopic fields per lung, using (NIH) ImageJ (RRID:SCR_003070) software.

SVEC4-10 (RRID:CVCL_4393) ECs were plated at a density of 2.5 × 10⁵ in 9.6 cm² wells, the day before the experiment. 4T1 cancer cells were labeled with Vybrant DID cell-labeling solution (Life Technologies, #V22887) according to the manufacturer’s instructions. Labeled cells were gently detached with 1 mmol/L EDTA and plated on sub-confluent EC layers. Cancer cells were let to adhere for 30 minutes at 37°C and 5% CO₂ and then washed 3 times with PBS. Images were acquired immediately with a DMI4000 B Leica optical microscope equipped with a 10× objective lens. (Leica, HCX FL PLAN 10X/0.25) at 10× magnification and analyzed using (NIH) ImageJ (RRID:SCR_003070) software.

Cells were lysed in RIPA buffer supplemented with 1 mmol/L Na₃VO₄ and protease inhibitor cocktail (Sigma-Aldrich, #P8340). Protein concentration of cell extracts was determined using Pierce bicinchoninic acid (BCA) reagent (Thermo Fisher Scientific, #23227) according to the manufacturer’s instructions. Protein samples were denaturated by adding a 4× loading buffer (β-mercaptoethanol 0.6 mol/L; SDS 8%; Tris-HCl 0.25 mol/L pH 6,8; glycerol 40%; bromophenol blue 0.2%) and incubated at 95°C for 5 minutes. Samples containing equivalent amounts of protein were subjected to 7.5% SDS-PAGE. Proteins were transferred onto a nitrocellulose membrane using the Trans-Blot Turbo Transfer System (Bio-Rad) according to manufacturer’s instructions, probed with antibodies targeting proteins of interest, and antibody binding was revealed by enhanced chemiluminescence technique using a ChemiDoc Image Lab analyzer and software (Bio-Rad), according to the manufacturer’s protocols. Approximately 10% BSA was used for filter blocking in all conditions. The following antibodies were used: totAKT (Cell Signaling Technology Cat# 9272, RRID:AB_329827), pAKT (Cell Signaling Technology Cat# 9271, RRID:AB_329825), totSTAT3 (Santa Cruz Biotechnology Cat# sc-8019, RRID:AB_628293), p-Stat3 (Santa Cruz Biotechnology Cat# sc-8059, RRID:AB_628292), and vinculin (Sigma-Aldrich Cat# V9131, RRID:AB_477629).

ADCC activity was evaluated using the Promega mFcγRIV ADCC Bioassay Kit (Cat#M1201) according to the manufacturer’s instruction. The anti-CD20 (mouse IgG2a class antihuman CD20 antibody) was provided with the kit. Fc(m6A9)B3-elicited ADCC was tested using human PlexinB1-expressing stable cells [clone 4C6, established previously (16)] as the target cells. Briefly, 1 × 10⁴ target cells (Raji or PlxB1_4C6) were plated in a 96-well plate together with increasing concentrations of test compounds and 7.5 × 10⁴ cells/well effector cells (Jurkat cells expressing mouse FcγRIV), followed by incubation for 6 hours at 37°C and 5% CO₂. After the addition of the substrate (Bio-Glo Reagent), luminescence was measured using a GloMax Navigator.

The statistical significance of quantitative data was analyzed by GraphPad Prism (RRID:SCR_002798) 10.0.0 software, applying the most appropriate methods and correction tests, specified in individual figure legends. Statistical tests used in this study were as follows: unpaired t test, log-rank (Mantel–Cox) test, and two-way ANOVA test. P < 0.05 was considered significant. ^∗, P < 0.05; ^∗∗, P < 0.01; ^∗∗∗, P < 0.001; ^∗∗∗∗; P < 0.0001.

The data generated in this study are available in the manuscript and its supplementary files or upon request from the corresponding author.

In order to assess the role of PLXNB1 in the TME, we performed in vivo orthotopic tumor transplants of the murine TNBC model 4T1 in syngeneic WT and PLXNB1-deficient animals (Plxnb1^−/− mice). The ability of 4T1 cells to develop tumors was significantly reduced when they were transplanted in the mammary fat pad of Plxnb1^−/− mice compared with WT controls (Fig. 1A and B). Next, we evaluated the role of PLXNB1 expression in the TME for the metastatic process and found that 4T1 tumors growing in Plxnb1^−/− mice gave rise to fewer macro-metastasis in the lungs compared with those growing in a WT environment (Fig. 1C). Tumors grown in PLXNB1-deficient animals also displayed a significantly lower metastatic index, defined as the number of lung metastasis normalized to the primary tumor weight (Fig. 1D). This was confirmed also when analyzing tumors of comparable size, indicating that the reduced metastatic burden was not simply a result of a lesser primary tumor mass (Supplementary Fig. S1A). In addition, to assess whether PLXNB1 inhibition would also enhance mice survival, we set up an experiment in which we predefined an end point for sacrifice. In this context, we observed enhanced survival in tumor-bearing Plxnb1^−/− mice compared with WT controls (Fig. 1E). The same effect on tumor progression of PLXNB1 deficiency was observed in another in vivo orthotopic TNBC model, by transplanting Py230 cancer cells in syngeneic C57/BL6 mice. In fact, although showing different and slower growth kinetics compared with 4T1 (22), Py230 BC cells injected in the mammary fat pad of female Plxnb1^−/− mice displayed a significant delay in tumor growth, compared with those grown in WT mice (Fig. 1F), and this was then associated with increased survival (Fig. 1G).

A short-term metastatic dissemination assay revealed no significant differences between WT and PLXNB1-deficient mice (Supplementary Fig. S1B). SEMA4D is known to participate in monocyte–EC adhesion via coupling with PLXNB receptors (23). Although we observed that SEMA4D was expressed by 4T1 cancer cells (Supplementary Fig. S1E) and PLXNB1 is known to be highly expressed by EC (12), no differences in 4T1 cancer cell adhesion were observed in Plxnb1-silenced EC (Supplementary Fig. S1C). These data indicate that the reduced ability of 4T1 cells to metastasize from primary tumors is not accounted for by differences in their ability to extravasate in the lungs or to adhere to the EC, thus suggesting that other mechanisms account for the effect of Plxnb1 silencing on TNBC progression and metastatic dissemination.

To understand the mechanisms of PLXNB1-dependent regulation of tumor progression, we first assessed the distribution of PLXNB1 and its main ligand SEMA4D in the TME of 4T1 tumors. Thus, we purified endothelial, myeloid, and lymphoid cells from tumor samples, and, by means of real-time PCR, we analyzed the expression levels of these transcripts. In line with previous literature (12), we found that PLXNB1, besides its expression in cancer cells, is highly expressed in EC of 4T1 tumors (Fig. 2A; Supplementary Fig. S1D). Furthermore, PLXNB1 was detectably expressed both in CD4⁺ and CD8⁺ T cells and myeloid cells in the TME (Fig. 2A). Similarly, SEMA4D was expressed both in T lymphocytes and myeloid cells, as previously shown (7–9) at a comparable level as in EC (Fig. 2B). Because SEMA4D inhibition has been shown to have an impact on tumor growth in diverse mouse models (7, 8), and it is currently assessed for therapeutic purposes in patients with cancer (8, 9, 24), we wished to compare the impact of SEMA4D or PLXNB1 loss on tumor progression in our mouse models. To this end, we analyzed the growth of 4T1 tumor cells orthotopically implanted in Sema4d^−/−, Plxnb1^−/−, and WT mice. Although we detected a significant reduction of the tumor burden in both knockout mouse strains, compared with WT mice, tumor growth suppression was significantly greater in Plxnb1^−/− mice (Fig. 2C).

The PLXNB1-ligand SEMA4D is a well-recognized pro-angiogenic factor (12), and tumor-induced angiogenesis is impaired in SEMA4D-deficient mice (7). We hence hypothesized that the reduced primary tumor growth and metastatic dissemination observed in Plxnb1^−/− mice could be explained in part by impaired tumor angiogenesis. Confocal microscope analysis of EC of tumor sections did not indicate a significant reduction in tumor vessel area of Plxnb1^−/− compared with WT mice, indicating that tumor-induced angiogenesis was not reduced (Supplementary Fig. S2A and S2B). We did observe a higher degree of pericyte coverage of tumor vessels in Plxnb1^−/− mice, compared with the WT, as assessed by co-immunostaining of tumor sections for EC and pericyte markers (Meca32 and NG2, respectively; Supplementary Fig. S2A and S2C). Increased pericyte recruitment to tumor vessels indicates that Plxnb1^−/− mice tumor vasculature is more mature, displaying a phenotype previously described as “normalized” (25). We then wondered if the increase in vessel coverage observed in Plxnb1^−/− mice was functionally relevant in terms of vascular normalization. In tumors grown in PLXNB1-deficient mice, we observed a significant reduction of tumor hypoxia markers, e.g., lower levels of carbonic anhydrase IX (CAIX; Supplementary Fig. S2D), a well-known hypoxia-inducible enzyme that has been previously associated with acidosis, invasiveness, and drug resistance (26). These data show an unexpected impact of PLXNB1 inactivation, different from that of its best-known ligand SEMA4D, to induce tumor vessel normalization and efficiently reduce tumor hypoxia, which is widely considered a tumor-inhibiting mechanism (25).

Antibody-mediated targeting of SEMA4D enhances the recruitment of activated monocytes and lymphocytes to the tumor site, thus skewing the balance of stromal cells and cytokines in the TME toward an antitumor environment (8, 9). Hence, we investigated immune cell recruitment in the TME of Plxnb1^−/− and WT mice. We observed an overall increase of infiltrating CD68⁺ macrophages in tumors grown in PLXNB1-deficient mice compared with those grown in WT mice (Supplementary Fig. S3A). We furthermore investigated the polarization state of infiltrating macrophages and found that PLXNB1 deficiency caused a shift from an M2-like (protumoral) to an M1-like (antitumoral) prevalent phenotype, compared with controls (Fig. 3A–C). To further investigate the major myeloid cell subtype recruited to tumors, we assessed infiltrating dendritic cells (CD11c⁺ cells). We observed increased numbers of CD11c⁺ cells in the tumor stroma of PLXNB1-deficient mice compared with controls (Fig. 3D). To further assess the effect of PLXNB1 depletion on macrophage polarization, we set up a series of experiments by purifying myeloid cells from the bone marrow of Plxnb1^−/− and WT mice, followed by treatments inducing macrophage polarization. PLXNB1-deficient BMDMs showed downregulated expression of Arg-1 (M2 marker) and a trend toward an upregulation of iNOS (M1 marker) compared with WT counterparts (Supplementary Fig. S3C and S3D). These data are in line with our observations in tumor tissues demonstrating a switch from an M2-like toward an M1-like macrophage phenotype in 4T1 Plxnb1^−/− versus WT tumors.

Based on the observed shift in macrophage polarization, we next analyzed the adaptive immune response, by assessing the presence, activation, and co-stimulation of lymphoid cells, which are required to develop a successful antitumor immune response (27). In particular, we investigated the differential recruitment of TILs in the TME of WT and Plxnb1^−/− mice. Although we did not observe significant changes in T-cell or dendritic cell populations in draining lymph nodes (Supplementary Fig. S4), we detected an increased number of CD3⁺ T lymphocytes in the primary tumors grown in Plxnb1^−/− mice (Fig. 4A). We also observed enhanced cytotoxic CD8⁺ T-cell recruitment in Plxnb1^−/− TME compared with WT controls (Fig. 4B and C), further highlighted by an increase of infiltrating CD8⁺ along with a reduction of CD4⁺ lymphocytes in Plxnb1^−/− mice (Fig. 4C). Consistent with what was seen in the primary tumors, we found enhanced recruitment of CD8⁺ T cells in lung metastases of tumors grown in Plxnb1^−/− mice compared with WT controls (Supplementary Fig. S5D and S5E).

We also analyzed the CD4⁺ T-cell populations in the primary tumors, observing that the immunosuppressor regulatory T cell (Treg) subset of CD4⁺ T cells was unchanged in Plxnb1^−/− mice compared with WT controls (Fig. 4D), thereby implying an increased ratio of CD8⁺/Treg cells (Fig. 4D). Further assessment of the phenotype of CD4⁺ T cells showed a significant reprogramming toward a Th1 polarization state (Fig. 4E).

It has been shown previously that 4T1 cells show greater capacity to grow tumors and metastasize when transplanted into IFNγ^−/− mice (28). We found increased levels of IFNγ in tumors grown in Plxnb1^−/− mice (Fig. 5A and B). Because cytotoxic CD8⁺ T cells, which we found to be increased in the TME of Plxnb1^−/− mice, are a major source of IFNγ, we aimed at identifying IFNγ-producing cells in our tumors. Indeed, infiltrating Plxnb1^−/− CD8⁺ T cells produced higher levels of IFNγ compared with their WT counterparts (Fig. 5C and D). Indicative of their functional activation, we found increased intratumoral proliferation of CD8⁺ cytotoxic T cells in 4T1 tumors grown in Plxnb1^−/− mice compared with WT controls (Fig. 5G); moreover, PLXNB1 deficiency was associated with an enhanced production of the protein GrzB (Fig. 5E and F). Furthermore, phenotypic analysis revealed that Plxnb1^−/− and WT cytotoxic T cells maintained the expression of the immune checkpoint inhibitor PD-1 (Supplementary Fig. S5A). Thus, to get further insights into cancer immunity associated with PLXNB1 deficiency, we measured the expression of immune response genes in TILs purified from 4T1 tumors growing in Plxnb1^−/− and WT mice (Fig. 5H). An array analysis identified 13 (out of 96) genes that were significantly differentially expressed. Among the most upregulated genes in Plxnb1^−/− TILs were well-known positive regulators and effectors of T-cell responses, such as the inducible T-cell co-stimulatory receptor (Icos; ref. 29), the effector protein Perforin1 (Prf1; ref. 30), transcription factors Stat3 and Stat4, chemokines such as Ccl5 (31), and E-selectin (Sele). Among the top down-modulated genes, we found Vcam1, the extracellular matrix component Fn1, and Cd34. Consistent with this, we found increased STAT3 and AKT signaling in ex vivo activated T cells derived from Plxnb1^−/− mice compared with T cells derived from WT mice (Supplementary Fig. S5B). To further assess the effect of PLXNB1 deficiency on T-cell phenotype, we performed gene expression analysis of CD4⁺/CD8⁺ T cells purified from the spleens of Plxnb1^−/− or WT mice, and these activated and expanded in vitro. We observed an upregulation of IFNγ and IL10 mRNA levels in Plxnb1^−/− T cells compared with WT counterparts (Supplementary Fig. S5C), suggesting that PLXNB1 targeting renders T cells more inclined to immune activation. Together, these data suggest that PLXNB1 deficiency in the 4T1 TME leads to a significant recruitment of CD8⁺ T cells programmed toward an antitumor phenotype.

Given the strong impact of environmental PLXNB1 deficiency in increasing recruitment and activation of CD8⁺ cytotoxic T cells, we assessed the functional importance of their antitumor activity in Plxnb1^−/− mice by inactivating these cells with depleting CD8-specific antibodies. In this experiment, the mice were sacrificed after 2 weeks of treatment to avoid adverse effects due to acceleration of tumor progression upon CD8⁺ cell depletion. The depletion of CD8⁺ T cells nullified the tumor- and metastasis-suppressive effect due to PLNXB1 deficiency, strongly suggesting that the activity of these cells is enhanced in Plxnb1^−/− mice (Fig. 6A–C). It is known that increased levels of infiltrating TILs correlate with good prognosis in patients with TNBC, even in the absence of adjuvant therapy (32). Given the strong increase in the number and activation of infiltrating TILs observed in the Plxnb1^−/− TME and their critical functional role, we sought to investigate whether the combination of PLXNB1 deficiency in the TME with treatment with an ICI, such as an anti-PD-1, could achieve a greater effect in inhibiting tumor progression in mice. Indeed, whereas PD-1 blockade in 4T1 tumors engrafted in WT mice barely reduced primary tumor size, in Plxnb1^−/− mice, anti-PD-1 treatment almost achieved complete tumor rejection (Fig. 6D–F). As shown above, the number of lung metastases was lower in Plxnb1^−/− mice compared with controls, but anti-PD-1 treatment further reduced the number of metastasis in Plxnb1^−/− mice only. These results show that CD8⁺ T-cell activation induced by PLXNB1 inactivation in the TME enhances the efficacy of immunotherapy, providing the rationale for combining PLXNB1-blocking drugs with ICIs, in TNBC.

To better assess the potential clinical application of PLXNB1 inhibition in the treatment of TNBC, we assessed the antitumor efficacy of systemic delivery of a recently developed protein inhibitor of PLXNB1 signaling, Fc(m6A9)B3. This is an Fc-based engineered protein with a PLXNB1-binding peptide moiety, which can strongly antagonize ligand-induced PLXNB1 activation by dimerizing cell surface PLXNB1 in a signaling incompetent conformation (19). The same peptide moiety, when synthesized as cyclic peptide and dimerized via a PEG linker, was successfully used to block PLXNB1 signaling in vivo (33).

We found that the administration of Fc(m6A9)B3 efficiently reduced the growth of Py230 orthotopic tumors compared with Fc-treated controls (Fig. 7A–C). Analysis of tumor tissues of Fc(m6A9)B3-treated Py230 mice showed significant changes in the TME similar to those found in 4T1 tumor-bearing Plxnb1^−/− mice. In particular, fewer MRC1⁺ M2-like macrophages (Supplementary Fig. S6A) and enhanced recruitment of CD8⁺ T cells and increased expression of GrzB (Supplementary Fig. S6B) were found in Fc(m6A9)B3-treated tumors compared with Fc control-treated tumors. This pharmacological targeting of PLXNB1 was well tolerated and impacted neither the hematological profile nor hepatic or renal functions (Supplementary Fig. S7A and S7B), when compared with untreated tumor-bearing mice.

To further assess whether genetic and pharmacological targeting of PLXNB1 would have similar antitumor effect in another independent TNBC mouse model, we injected EMT6 TNBC cells orthotopically in the mammary fat pad of syngeneic Plxnb1^−/− Balb/c mice. We observed a significant inhibition of EMT6 tumor growth in Plxnb1^−/− mice compared with WT controls (Supplementary Fig. S8A). In addition, we treated the EMT6 mouse model with Fc(m6A9)B3, and, similarly to the Py230 model, we observed a significant inhibition of tumor growth (Supplementary Fig. S8A).

Because a pharmacological systemic inhibition of PLXNB1 could also directly impact cancer cells in addition to reprogramming the TME, we incubated Fc(m6A9)B3 with 4T1, Py230, and EMT6 TNBC cells in culture and observed no significant inhibition of cell viability (Supplementary Fig. S8B–D). Treatment of EMT6 tumor-bearing Plxnb1^−/− mice with Fc(m6A9)B3 did not achieve any additional inhibition of tumor growth compared with Fc control-treated knockout mice (Supplementary Fig. S8A), suggesting that targeting PLXNB1 in the TME but not on the tumor cells is sufficient to achieve maximal tumor suppression. Finally, to assess whether Fc(m6A9)B3 could potentially induce ADCC, we performed an ADCC bioassay. Although the positive control Raji cells incubated with anti-CD20 yielded a strong ADCC response (estimated EC50 value = 43 ng/mL), no ADCC activity toward PLXNB1-expressing cells was detected upon incubation with Fc(m6A9)B3 (Supplementary Fig. S7C).

Based on the changes in the TME observed upon treatment with Fc(m6A9)B3 in TNBC models, we sought to investigate whether treatment with Fc(m6A9)B3 would enhance the antitumor efficacy of ICIs, as observed in the context of genetic depletion in 4T1 tumor-bearing Plxnb1^−/− mice. Although PD-1 blockade in Py230 tumor-bearing mice could barely reduce the tumor burden, the combined treatment with Fc(m6A9)B3 and anti-PD-1 induced a greater antitumor effect compared with single treatments (Fig. 7D and E), suggesting that the two drugs unleash complementary mechanisms suppressing tumor growth. Altogether, these data strongly suggest that both genetic and pharmacological targeting of PLXNB1 in diverse TNBC preclinical models efficiently suppresses tumor growth by reprogramming the immune response in the TME and thereby enhancing the antitumor efficacy of ICIs.

The importance of semaphorin signaling in cancer onset and progression is raising increasing interest. The immune semaphorin SEMA4D, in particular, has been intensely studied in the tumor context, and its antibody-mediated blockade is currently being translated into the clinic (8, 9, 24). Moreover, it was shown that anti-SEMA4D can enhance the recruitment of activated monocytes and lymphocytes into the tumor, modifying the balance of stromal cells and cytokines in the TME toward a pro-inflammatory setting (8, 9). Intriguingly, the main receptor for SEMA4D, PLXNB1, is known to be expressed at varying levels in cancer cells; however, its role in the regulation of the TME has been poorly characterized so far.

Here, we show that the growth of three different TNBC models is strongly reduced in Plxnb1^−/− mice compared with controls and that systemic treatment with a PLXNB1 inhibitor efficiently hampered tumor progression. PLXNB1 is expressed in all TNBC cells used in the in vivo experiments but was also found in different cell populations in the TME, such as EC, myeloid, and T cells. Herein, we demonstrated, in vivo and in vitro, that PLXNB1 inhibition did not directly affect tumor cells but instead profoundly reprogrammed the TME of TNBC preclinical models. Thus, because SEMA4D inhibition has been shown to impinge on tumor growth and the TME in diverse mouse models (8, 9), we compared the growth of syngeneic 4T1 tumor cells orthotopically implanted in Sema4d^−/− and in Plxnb1^−/− mice. Although we detected a strong reduction of the tumor burden in both knockout mice, cancer growth suppression appeared particularly profound in Plxnb1^−/− mice, suggesting that the efficacy of PLXNB1 targeting for cancer therapy is surely not inferior to SEMA4D inhibition or could even ensure a wider blockade of tumor-promoting ligands found in the TME. Even if the underlying mechanisms of this enhanced antitumor effect of PLXNB1 inhibition remain to be clarified, it may be conceivable that other Sema4s (e.g., Sema4A and/or Sema4C; ref. 34) can contribute to the observed effects.

Indeed, in vivo studies by Sierra and colleagues (7) demonstrated that tumor angiogenesis is reduced in tumors growing in a SEMA4D-deficient environment. However, in the PLXNB1 knockout model, we observed that tumor-induced angiogenesis was not affected. This is consistent with previous data and may be accounted for by the presence of alternative SEMA4D receptors (e.g., PLXNB2; ref. 17). We did observe an increase in vessel pericyte coverage upon PLXNB1 loss and a reduction in intratumoral hypoxia, suggestive of normalization of the tumor vasculature. Indeed, SEMA4D–PLXNB1 signaling has already been linked to vascular permeability alterations and in the regulation of pericyte function in rodent models of stroke and in a diabetic retinopathy model (35, 36). Moreover, a growing body of evidence demonstrated that the immune-vascular crosstalk and a mutual regulation between normalized vessels and T cells (37) can improve the efficacy of immunotherapy in cancer (38). Future experiments will aim at deciphering the precise molecular mechanisms responsible for this tumor vessel normalization and its potential involvement in the immune cell activation that we observed upon PLXNB1 depletion. At this stage, we sought to better characterize the differential immune infiltrate within tumors growing in Plxnb1^−/− mice, finding an enhanced fraction of M1-polarized macrophages, compared with the TME of WT mice. Macrophages are one of the most represented immune cell populations in the TME and strongly impact on tumor growth and metastatic process. High density of TAM was significantly associated with late clinical staging in patients with BC, and TAMs are increasingly considered as a potential therapeutic target in BC (39). In addition, the inhibition of Arg-1 (M2 marker) and the trend toward enhanced expression of iNos (M1 marker) in BMDMs purified from Plxnb1^−/− mice suggest that PLXNB1 may directly affect macrophage activity and polarization. In addition, we observed greater recruitment of CD11c⁺ cells in the TME of PLXNB1-deficient animals, suggestive of increased stromal infiltration of professional antigen-presenting cells compared with WT mice. Even though CD11c⁺ dendritic cells were more abundant in the TME of Plxnb1^−/− mice, we cannot formally exclude that other subtypes could be present and functionally relevant in an environment lacking PLXNB1. Further experiments are needed to better characterize the diverse myeloid cell subtypes recruited in the absence of PLXNB1.

PLXNB1 has been found in activated human T lymphocytes (11) and dendritic cells (40). Our characterization of the lymphoid immune infiltrate showed it to be one of the most profound PLXNB1-dependent changes in the TME. In fact, we observed a strong increase in the amount of infiltrating CD8⁺ cytotoxic T cells in the TME of Plxnb1^−/− mice. Although CD8⁺ cytotoxic T cells mediate tumor-specific adaptive immunity, CD4⁺ T helper cells exist in two main subtypes: Th1 and Th2. Th1 cells have a prominent antitumor activity due to their production of TNFα, IFNγ, IL2, and IL1. On the other hand, Th2 cells rather sustain tumor growth (41). Naive CD4⁺ T-cell differentiation into Th1 cells can be induced by STAT4, a transcription factor that we found induced in Plxnb1^−/− TILs. Moreover, Th1 CD4⁺ T cells were increased in the PLXNB1-deficient TME compared with WT, and at the same time, tumor-infiltrating Plxnb1^−/− CD8⁺ T cells showed an increased production of IFNɣ and GrzB and increased intratumoral proliferation, suggesting that PLXNB1 signaling may be involved in the inhibition of lymphocyte activation. The precise molecular mechanism for increased T-cell recruitment awaits elucidation, but a number of mechanisms can be proposed. On the one hand, increased TILs recruitment could be an indirect effect of increased pro-inflammatory macrophage or dendritic cell recruitment. On the other hand, PLXNB1 deficiency could be directly involved in regulating T-cell homing, as suggested by the altered expression of genes mediating T-cell activation and migration. We can speculate that PLXNB1 deficiency results in increased R-RAS signaling due to PLXNB1 intrinsic GAP activity (42). Previous work showed that R-RAS signaling plays a critical role in T-cell proliferation, migration, and activation (43). In line with what we observed in primary tumors, we noticed a significant recruitment of CD8⁺ T cells within the fewer lung metastases identified in 4T1 tumor-bearing Plxnb1^−/− mice, which was barely detectable in WT controls. These findings suggest that PLXNB1 targeting induced profound changes in CD8⁺ T-cell activation and recruitment in the TME, which was appreciable not only in the primary tumor but also in distant sites, and may be partly accountable for the strong reduction of lung metastasis observed in tumor-bearing Plxnb1^−/− mice. Although further experimental evidence in other preclinical models is warranted, these findings bear a remarkable potential relevance in clinical perspective. From a functional perspective, we wondered if the increase in CD8⁺ T cells could be sufficient to explain the PLXNB1-deficient mice phenotype. Upon anti-CD8 treatment, we found that the difference in growth and metastatic dissemination between tumors grown in Plxnb1^−/− and WT mice was lost, rescuing the phenotype seen in WT animals. CD8⁺ T-cell targeting also impaired the metastatic ability of 4T1 tumors grown in WT mice, suggesting that cytotoxic T cells are fundamental in maintaining disease control in this model. Taken together, these data suggest that the enhanced CD8⁺ T-cell recruitment into tumors grown in Plxnb1^−/− mice is mainly accountable for the suppression of tumor progression.

In line with the observed antitumor cytotoxic activity of CD8⁺ T cells, TIL gene expression analysis revealed that PLXNB1 deficiency was associated with the upregulation of T cell response genes such as Icos and Perforin1. Ctla4 levels also were increased, possibly reflecting enhanced TCR signaling (44). Furthermore, Stat3 and Stat4 transcription factors were strongly induced, together with the chemokine Ccl5, a potent promoter of monocyte and T-cell migration. It should be noted that STAT3-deficient human CD8⁺ T cells fail to upregulate perforin and GrzB expression in response to IL21 and IL15 (45), whereas STAT4 is required for CD4⁺ T-cell differentiation into Th1 cells (46). In addition, we found that Sele (E-selectin) expression was strongly induced in Plxnb1-deficient T cells; this protein is usually expressed by ECs and participates in leukocyte rolling and adhesion to the endothelium (47). Interestingly, it has been previously shown that E-selectin can be induced by membrane TNFα in CD4⁺ T cells upon activation (48). Among the top down-modulated genes, we found Vcam1, an adhesion molecule expressed by thymocytes and T cells undergoing apoptosis (49); Fn1, a major constituent of the extracellular matrix, synthesized by T cells upon activation (50); and Cd34, possibly suggesting a reduction in the recruitment of HSCs. T cells purified ex vivo from Plxnb1^−/− mice carried an enhanced immune activation gene expression profile, compared with WT counterparts, a further indication that PLXNB1 inhibition directly affects T-cell function. In addition, in line with the differential gene expression observed in TILs, we found that Plxnb1 silencing in stimulated T cells in vitro increased AKT signaling and STAT3 phosphorylation, indicative of enhanced activation of major pathways promoting T-cell expansion, survival, differentiation, and functional activity (51). We can speculate that PLXNB1 deficiency results in increased R-RAS or RAC1 signaling, which play a critical role in T-cell proliferation, migration, and activation (43, 52), potentially explaining the increased T cell-mediated immune response in PLXNB1-deficient tumors. For example, PLXNB1 loss might increase STAT3 activity by de-sequestering GTP-bound RAC1 in the cytosol or interfering with its activation (53, 54). Together, these data suggest that PLXNB1 deficiency appears to intrinsically stimulate T cells.

As a potential clinical application, we demonstrated for the first time that systemic pharmacological targeting of PLXNB1, using the targeted engineered inhibitor protein Fc(m6A9)B3, efficiently and safely inhibited cancer growth in two diverse TNBC models, faithfully recapitulating the results obtained in the gene-deficient mouse models, without inducing any toxic effect.

The TME of solid tumors is frequently defined as “cold” or “hot,” based on the amount of T-lymphocyte infiltration and pro-inflammatory cytokine production. Hot tumors, characterized by higher levels of pro-inflammatory cytokines and T-cell infiltrate, display a better response to immunotherapies, such as ICIs targeting the PD-L1/PD-1 axis (3, 55). Hence, turning a cold tumor into a hot one has been in the focus of many recent therapeutic efforts, including for patients with BC. In our experiments, we found that PLXNB1 targeting enhances CD8⁺ T-cell infiltration in both primary and metastatic sites, featuring their conversion into hot tumor tissues. Indeed, several trials employing anti-PD-1 as a new therapeutic strategy are ongoing for patients with TNBC (56). In this respect, although ineffective in 4T1 or Py230 tumor-bearing mice proficient for PLXNB1, a PD-1 blocking antibody treatment prompted a strong antitumor response in Plxnb1^−/− mice and in mice systemically treated with the selective PLXNB1 inhibitor.

In conclusion, herein we unveil a specific role for PLXNB1 in regulating the TME of TNBC. Our results demonstrated that genetic or pharmacological targeting of PLXNB1 in TNBC models hampered cancer progression through reprogramming the TME toward a tumor-suppressor response, by recruiting and enhancing the activity of CD8⁺ T cells, which in turn increased the efficacy of anti-PD-1 immunotherapy. These findings are promising from a translational perspective, providing preclinical evidence for new combination strategies with immunomodulatory agents and standard-of-care therapies for BC.

J. Takagi reports grants from Japan Agency for Medical Research and Development (AMED) during the conduct of the study and has a patent for WO2019026920A1 issued, licensed, and with royalties paid from MiraBiologics Inc. Dr. J. Takagi is a cofounder and shareholder of MiraBiologics Inc. No disclosures were reported by the other authors.

G. Franzolin: Data curation, formal analysis, validation, investigation, methodology, writing–original draft. S. Brundu: Data curation, formal analysis, validation, investigation, methodology, writing–original draft. C.F. Cojocaru: Data curation, formal analysis, validation, writing–review and editing. A. Curatolo: Formal analysis, validation, investigation. M. Ponzo: Data curation, formal analysis, validation, investigation, methodology. R. Mastrantonio: Validation, investigation. E. Mihara: Data curation, formal analysis. A. Kumanogoh: Resources, writing–review and editing. H. Suga: Resources, methodology, writing–review and editing. J. Takagi: Resources, funding acquisition, methodology, writing–review and editing. L. Tamagnone: Conceptualization, resources, supervision, funding acquisition, methodology, writing–original draft, writing–review and editing. E. Giraudo: Conceptualization, data curation, supervision, funding acquisition, validation, methodology, writing–original draft, project administration, writing–review and editing.

The authors would like to thank Gabriella Cagnoni and Massimo Accardo for discussions and technical support. This work was supported by FPRC 5xmille Ministero Salute 2017 PTCRC-INTRA 2020, progetto SEE-HER to E. Giraudo; by Fondazione AIRC per la Ricerca sul Cancro (AIRC), IG #19957 and IG 2022 #27516 to E. Giraudo and IG #19923 and # 29255 to L. Tamagnone; by AIRC 5 per Mille, Special Program on Metastatic Disease (# 21052) to E. Giraudo and L. Tamagnone; by the Italian Ministry of Health, Ricerca Corrente 2022-23 to E. Giraudo and L. Tamagnone; and by the Platform Project for Supporting Drug Discovery and Life Science Research (22ama121011j0001) from the Japan Agency for Medical Research and Development (AMED) to J. Takagi. Università Cattolica del Sacro Cuore [Intramural Grant (Linea D1.1)] also contributed to the funding of this research project and to its publication.