Introduction

Immunotherapy has emerged as a promising treatment for patients with advanced cancer. Immune system–based cancer therapies offer a durable clinical benefit because they can potentiate a self-propagating and adaptable response once the immune system is activated (1). However, only a fraction of patients responds to such treatment (2). Most of the immunotherapy-resistant tumors are “cold,” entailing a severely impaired presence and activation status of effector T cells within the tumor microenvironment (TME; ref. 3). Thus, understanding the mechanisms underlying T-cell exclusion could translate into a broader and more durable response to this therapeutic option.

Cytotoxic CD8+ T lymphocytes (CTL) are one of the most powerful antitumor cells in the TME and their infiltration in tumors correlates with good prognosis in several tumor types (4). Immune checkpoint inhibitors (ICI) take advantage of these cells and their killing capacity, but these require both the presence and physical contact between antitumor T cells and cancer cells (1). Given the lack of preexisting CTLs in T-cell “cold” tumors, it is very unlikely that the use of currently approved ICIs will lead to robust antitumoral T-cell responses in these tumor types (5). Hence, priming of the tumor via a combination of different therapies might be used to recruit a higher number of CTLs into these tumors.

Plexins are large transmembrane glycoproteins that, in most cases, function as the receptors for semaphorins (6). In the nervous system, these proteins play a bifunctional role, having the capacity to exert both repulsive and attractive effects in neuronal wiring during development (7). Their ability to modulate the immune response in both physiological and pathological conditions (8) and their role as “cell positioning cues” within the TME have also been explored (9). Targeting plexin signals (or semaphorins) is therefore a promising therapeutic strategy to restore antitumor immunity.

Plexin-A4 (Plxna4) is a member of class A plexins (10), which interacts with Sema6A and Sema6B (11). When in association with neuropilin-1 (NRP1), it can also function as a coreceptor for Sema3A (12). In the central nervous system, Plxna4 mediates axon repulsion by the direct binding to Sema6A and Sema6B (13, 14). In the immune system, Plxna4 has been implicated in macrophage Toll-like receptor (TLR)-mediated signaling and cytokine production in sepsis (15), anti-inflammatory polarization in colitis (16), and entry into hypoxic niches in cancer (17). In T cells, Plxna4 negatively regulates T-cell–mediated immune responses, with Plxna4-deficient mice showing exacerbated disease in a mouse model of experimental autoimmune encephalomyelitis (EAE; ref. 18). On the basis of these findings, we defined the function of PlxnA4 in the control of the immune response in the context of cancer.

Materials and MethodsAnimals

Plxna4 knockout (KO) mice on a C57BL/6 background were obtained from Dr. Castellani (Institut NeuroMyoGène, Université de Lyon, Lyon, France). C57BL/6 mice were purchased from Charles River. OT-I mice were purchased from Taconic. Sema6a KO mice on a C57BL/6 background were obtained from Prof. Dr. Pasterkamp (Dept. of Translational Neuroscience, University Medical Center Utrecht, The Netherlands). All mice used were between 6 and 12 weeks old, without specific gender selection. In all experiments, littermate controls were used. Euthanasia was performed by cervical dislocation. Housing conditions and all experimental animal procedures were approved by the Animal Ethics Committee of the KU Leuven.

Bone marrow transplantation

Six-week-old C56BL/6 recipient mice were lethally irradiated with a dose of 9.5 Gy using the Small Animal Radiation Research Platform (SARRP, XSTRAHL). Femur and tibia bones were collected from donor mice of the appropriate genotype. In a sterile culture hood, bone marrow (BM) cells were obtained by flushing the bones with a syringe filled with Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco, Thermo Fisher Scientific, 21875034) supplemented with 10% heat-inactivated FBS (Biowest, S1810). The cells were subsequently filtered using a 40-μm pore–sized mesh and centrifuged for 5 minutes at 200 × g. BM cells were counted and 1 × 107 cells were injected intravenously (i.v.) via tail vein in the irradiated recipient mice. Tumor experiments were initiated 6 to 8 weeks after BM reconstitution. Red and white blood cell count was determined using a hemocytometer on peripheral blood, collected in heparin with capillary pipettes by retro-orbital bleeding.

Cell lines

Murine Lewis lung carcinoma cells (LLC) and B16F10 melanoma cells were obtained from the ATCC. E0771 medullary breast adenocarcinoma cells were obtained from CH3Biosystems. All cells were cultured in Dulbecco's Modified Eagle Medium (DMEM; Gibco, Thermo Fisher Scientific, 41965039) supplemented with 10% heat-inactivated FBS, 2 mmol/L glutamine (Gibco, Thermo Fisher Scientific, 25030024), 100 U/mL penicillin, and 100 μg/mL streptomycin (Gibco, Thermo Fisher Scientific, 15140122) at 37°C in a humidified atmosphere containing 5% CO2. For the overexpression of ovalbumin (OVA), the following plasmid was used: pCDH_CMV7-OVA-EFI-G418. LLC and B16F10 cancer cells were transduced with concentrated lentiviral vectors and further selected with G418 antibiotics (1 mg/mL, Invivogen, ant-gn) to generate a homogenous population of OVA-overexpressing cancer cells (LLC-OVA and B16F10-OVA). Overexpression of the OVA protein was confirmed by Western blot analysis. All cell lines were tested for Mycoplasma and passaged in the laboratory for no longer than 6 months after receipt.

Tumor models

Adherent growing murine cells, 1 × 106 LLC and 5 × 105 B16F10, were injected subcutaneously (s.c.) for LLC and orthotopically for B16F10 at the right side of the mouse in PBS (Gibco, Thermo Fisher Scientific, 14190094). Alternatively, 5 × 105 E0771 medullary breast adenocarcinoma cells were injected orthotopically in the mammary fat pad of the second nipple on the right side in a volume of 50 μL PBS. Tumor volumes were measured three times a week with a caliper and calculated using the formula: V = π × d2 × D/6, where d is the minor tumor axis and D is the major tumor axis. At the end stage, tumors were weighed and collected for immunofluorescence and/or flow cytometric analyses. For survival analysis, a tumor volume of 1,800 mm3 was used as the humane endpoint.

Histology and immunostaining

Tumors and lymph nodes (LN) were collected and fixed in 4% formaldehyde (37% stock, VWR, ACRO119690250) diluted in PBS overnight at 4°C, dehydrated and embedded in paraffin. Serial sections were cut at 7-μm thickness with an HM 355S automatic microtome (Thermo Fisher Scientific). Paraffin slides were first rehydrated to further proceed with antigen retrieval in Target Retrieval Solution, Citrate pH 6.1 (DAKO, Agilent, S1699). If necessary, 0.3% hydrogen peroxide (Stock 30%, Millipore, 1072090250) was added to methanol (VWR, 20848.320), to block endogenous peroxidases. The sections were blocked with the appropriate serum (DAKO, Agilent), matching the species of the secondary antibody, and incubated overnight at room temperature with the following antibodies: rat anti-CD8α (Thermo Fisher Scientific, 4SM15, 1:100), rat anti-CD4 (Thermo Fisher Scientific, 4SM95, 1:100), rat anti-F4/80 (AbD Serotec, MCA497, 1:100), rabbit anti-FITC (AbD Serotec, 4510–7604, 1:200), rat anti-CD31 (BD Pharmingen, 550274, 1:50), rat anti-CD34 (BD Pharmingen, 553731, 1:100), rabbit anti-NG2 (Millipore, AB5320, 1:200), and rat anti-PNAd (BioLegend, MECA-79, 1:100). Appropriate secondary antibodies raised against the species of the primary antibody were used: Alexa 488 (Molecular Probes, A21208, 1:200), 647 (Molecular Probes, A31573, 1:100) or 568-conjugated secondary antibodies (Molecular Probes, A11077, 1:200), biotin-labeled antibodies (Jackson Immunoresearch, 711–065–152 and 712–065–153, 1:300), and, when necessary, TSA Plus Cyanine 3 and Cyanine 5 System amplification (Perkin Elmer, Life Sciences, NEL744001KT and NEL745001KT, 1:50) were performed according to the manufacturer's instructions. Hoechst-33342 solution (Thermo Fisher Scientific, H3570, 1:1,000) was used to visualize nuclei. Mounting of slides was done with ProLong Gold mounting medium without DAPI (Invitrogen, P36930). Imaging and microscopic analysis was performed with an Olympus BX41 microscope and CellSense imaging software.

Tumor hypoxia assessment and tumor perfusion

Tumor hypoxia was detected 1 hour after intraperitoneal (i.p.) injection of 60 mg/kg pimonidazole hydrochloride (Hypoxyprobe kit, Chemicon, HP3–100Kit) in LLC tumor–bearing mice. Tumors were harvested and fixed in 4% formaldehyde overnight. To detect the formation of pimonidazole adducts, 7-μm thick sections were immunostained with rabbit anti-hypoxyprobe monoclonal (Hypoxyprobe Kit, Chemicon, HP3–100 Kit, 1:100) following the manufacturer's instructions. Perfused tumor vessels were counted on tumor sections from mice injected i.v. with 0.05 mg FITC-conjugated lectin (Lycopersicon esculentum; Vector Laboratories, B-1175–1).

Flow cytometry

Mice were sacrificed by cervical dislocation, and tumors, livers, blood, LNs (inguinal and axillary LNs) or the tumor-draining LN (tdLN; the closest LN draining the tumor bed) were collected. Tumors were minced in α-minimum essential medium (MEM) (Lonza, BE12–169F), containing 50 μmol/L β-mercaptoethanol (Gibco, Thermo Fisher Scientific, 21985023), 5 U/mL Deoxyribonuclease I 0.85 mg/mL (Roche, 10104159001), Collagenase V (Sigma-Aldrich, C9263–1G), 1.25 mg/mL Collagenase D (Roche, 11 088 882 001), and 1 mg/mL Dispase (Gibco, Thermo Fisher Scientific, 17105–041), and incubated in the same solution for 30 minutes at 37°C. Livers were processed in RPMI 1640 medium, supplemented with 10 U/mL Deoxyribonuclease I (Roche, 10104159001) and 120 U/ml Collagenase III (Worthington Biochemical, LS004182), using the gentleMACS Dissociator (Miltenyi Biotec). The digested tissues were filtered using a 70-μm pore–sized mesh and cells were centrifuged for 5 minutes at 300 × g. Blood samples were collected in heparin with capillary pipettes by retro-orbital bleeding. Red blood cell lysis was performed by using a homemade red blood cell lysis buffer (150 mmol/L NH4Cl, 0.1 mmol/L EDTA, 10 mmol/L KHCO3, pH 7.4). LNs were processed on a 40-μm pore cell strainer in sterile PBS and cells were centrifuged for 10 minutes at 300 × g. Red blood cell lysis was performed by using Hybri-Max (Sigma-Aldrich, R7757). Single cells were resuspended in FACS buffer (PBS containing 2% FBS and 2 mmol/L EDTA) and incubated for 15 minutes with Mouse BD Fc Block purified anti-mouse CD16/CD32 (BD Pharmingen, 553142). Extracellular staining was performed for 30 minutes at 4°C. When necessary, permeabilization was performed using the eBioscience Foxp3/Transcription Factor Fixation/Permeabilization Kit (Thermo Fisher Scientific, 00–5521–00) according to the manufacturer's instructions and cells were incubated overnight at 4°C with the intracellular antibodies. All antibodies used are described in Supplementary Table S1. Cells were subsequently washed and resuspended in FACS buffer before flow cytometric analysis by a FACS Canto II, Fortessa X-20, or flow sorting by a FACS Aria III, Aria Fusion (BD Biosciences). Data were analyzed by FlowJo (TreeStar, Version 10.7). Fluorescence Minus One (FMO) controls were utilized to ensure proper gating of positive populations.

Mouse T-cell isolation and activation

Naïve mouse T cells were isolated from the spleen, inguinal, and axillary LNs. In brief, tissues were processed on a 40-μm pore cell strainer in sterile PBS and cells were centrifuged for 10 minutes at 300 × g. Red blood cell lysis was performed using Hybri-Max. Total splenocytes were cultured in T-cell medium [RPMI 1640 medium supplemented with 10% heat-inactivated FBS, 100 U/mL penicillin and 100 μg/mL streptomycin, 1% MEM non-essential amino acids (NEAA, Gibco, Thermo Fisher Scientific, 11140035), 25 μmol/L β-mercaptoethanol, and 1 mmol/L sodium pyruvate (Gibco, Thermo Fisher Scientific, 11360070)] at 37°C in a humidified atmosphere containing 5% CO2.

According to the experimental requirements, T cells were activated for 3 days by adding mouse anti-CD3/CD28–coated Dynabeads (Thermo Fisher Scientific, 11453D) at a 1:1 bead-to-cell ratio. At day 3 of activation, the beads were magnetically removed and activated T cells were further expanded for a maximum of 3 additional days in the presence of 10 ng/mL recombinant murine IL2 (mIL2, PeproTech, 212–12). CD8+ T cells were isolated by using MagniSort Mouse CD8+ T Cell Negative Selection Kit (eBioscience, Thermo Fisher Scientific, 8804–6822–74) according to the manufacturer's instructions. CD4+ T cells were isolated by using MACS Mouse CD4+ T Cell Isolation Kit (Miltenyi Biotec, 130–104–454) according to the manufacturer's instructions. According to the experimental requirements, activated T cells (at day 3 of stimulation) were treated for 48 hours with 15 μg/mL anti–programmed cell death protein 1 (PD-1; RMP1–14, BioLegend) or the appropriate isotype control. FOXO inhibition in activated T cells at day 3 of stimulation was performed by 48 hours of treatment with 80 μmol/L carbenoxolone (CBX, Sigma-Aldrich, C4790) or the DMSO vehicle control (dimethyl sulfoxide, Sigma, D2438).

Human T-cell isolation and activation

Buffy coat samples from healthy donors were obtained from the Red Cross-Flanders. Human CD4+ and CD8+ T cells were directly isolated by using the StraightFrom Buffy Coat CD4 and CD8 MicroBead Kit (Miltenyi Biotec, 130–114–980 and 130–114–978, respectively) according to the manufacturer's instructions. Red blood cell lysis was performed using Hybri-Max. T cells were activated in T-cell medium for 3 days by adding human anti-CD3/CD28–coated Dynabeads (Thermo Fisher Scientific, 11132D) at a 1:1 bead-to-cell ratio. At day 3 of activation, the beads were magnetically removed and activated T cells were further expanded for a maximum of 7 additional days in the presence of 10 ng/mL recombinant human IL2 (hIL2, PeproTech, 200–02). According to the experimental requirements, activated T cells (at day 7 of stimulation) were treated for 48 hours with 15 μg/mL anti–PD-1 (J116, BioXCell) or the appropriate isotype control.

Cytospin staining

CD8+ T cells and peripheral blood leukocytes were seeded onto glass slides by cytospin centrifugation and fixed in 4% formaldehyde for 10 minutes, followed by incubation with 0.2% Triton-X (VWR, 1.086.031.000) diluted in PBS for 15 minutes. To reduce the immune background, sections were blocked with 10% donkey serum (Sigma, D9663) in PBS for 1 hour, followed by blocking with FAB fragment anti-mouse IgG (Jackson Immunoresearch, 715–007–003, 1:10) for 1 hour. Samples were then probed overnight with mouse anti-Plexin-A4 (Plxna4; R&D Systems, 707201, 1:500) and incubated with Donkey Alexa 568–conjugated secondary antibodies (Molecular Probes, A10037, 1:100) for 45 minutes. Nuclei were counterstained with Hoechst-33342 and mounting of the slides was performed with ProLong Gold mounting medium without DAPI. All steps were performed at room temperature. Microscopy was conducted with an Olympus BX41 microscope and cellSens imaging software.

Quantitative RT-PCR

RNA was extracted from T cells using the TRIzol Reagent (Life Technologies, 15596018) according to the manufacturer's instructions. Reverse transcription to cDNA was performed with the SuperScript III First Strand cDNA Synthesis Kit (Life Technologies, 18080051) according to the manufacturer's instructions. Premade assays were purchased from Integrated DNA Technologies. The cDNA, primer/probe mix, and TaqMan Fast Universal PCR Master Mix were prepared according to the manufacturer's instructions (Applied Biosystems, 4352042). Samples were loaded into an optical 96-well Fast Thermal Cycling Plate (Applied Biosystems) and qRT-PCR was performed using a QuantStudio 12K Flex Real-Time PCR System (Applied Biosystems). Samples were run in technical duplicates. Data was normalized to housekeeping gene expression (Hprt for mouse and TBP for human genes). The commercially available probes (Integrated DNA technologies) used are listed in Supplementary Table S2.

Lymphocytic choriomeningitis virus expressing OVA model

Wild-type (WT) mice were preconditioned by i.v. injection of 1 × 104 naïve OT-I T cells. Twenty-four hours later, the mice were vaccinated i.p. with 105 plaque-forming units (PFU) of a recombinant lymphocytic choriomeningitis virus expressing OVA (LCMV-OVA; a kind gift from Prof. Dr. Daniel Pinschewer, University of Basel, Basel, Switzerland), as described in Flatz and colleagues (19). After 7 days of LCMV-OVA infection, OT-I T cells were FACS sorted from the blood.

T-cell proliferation assay

To monitor cell proliferation, activated T cells were labeled with 3.5 μmol/L Violet Cell Tracer (Thermo Fisher Scientific, C34557) at 37°C for 20 minutes. The cells were subsequently washed with FACS buffer and cultured according to the experimental requirements. Absolute numbers of T cells in culture were counted by flow cytometry using Precision Count Beads (BioLegend, 424902). According to the experimental requirements, activated T cells (at day 3 of stimulation) were treated with 100 μmol/L Rac1 inhibitor NSC23766 (Selleckchem, S8031) or the DMSO vehicle control.

Annexin V/propidium iodide apoptosis assay

Activated CD8+ T cells were collected, washed, and resuspended in 100 μL Annexin V Binding Buffer (BioLegend, 422201) containing 4 μL of Annexin V (BioLegend, 640941) and 0.1 μL propidium iodide solution (1 mg/mL stock, Sigma Aldrich, P4864). After 15 minutes of incubation at room temperature, samples were analyzed by flow cytometry.

Transwell migration assay

Migration of T cells was assessed by using Transwell permeable supports with 5-μm polycarbonate membrane (Costar, 3387). To determine cell migration in response to soluble factors, the bottom chamber was loaded with 0.1% FBS, 200 ng/mL CCL21 (PeproTech, 250–13), 200 ng/mL CCL19 (PeproTech, 250–27B), 150 ng/mL CXCL9 (PeproTech 250–18), or 50 ng/mL CXCL10 (PeproTech, 250–16) in T-cell medium. T cells were incubated for 2 (naïve) or 3 hours (activated) at 37°C and migrated cells in the bottom chamber were collected and counted by flow cytometry using Precision Count Beads. According to the experimental requirements, activated T cells were pretreated for 1 hour with 10 μg/mL anti-CCR7 (R&D Systems, 4B12), 250 μg/mL anti-CXCR3 (BioLegend, CXCR3–173), 100 μmol/L Rac1 inhibitor NSC23766, or the appropriate isotype, or vehicle control.

LN homing assay

Naïve CD8+ T cells were isolated from WT and Plxna4 KO mice and labeled with either 3.5 μmol/L Violet Cell Tracer or 1 μmol/L carboxyfluorescein diacetate succinimidyl ester (CFSE) Cell Tracer (Thermo Fisher Scientific, C34554). For CFSE labeling, cells were stained in PBS for 8 minutes at room temperature with gentle agitation. To label cells with Violet Cell Tracer, the staining was conducted for 20 minutes at 37°C with gentle agitation. Afterwards, a brief wash with complete RPMI medium was performed to quench any remaining dye. Healthy WT mice were injected i.v. with a 1:1 mixture between 1–2 × 106 labeled WT and Plxna4 KO T cells. After 2 hours, LNs of the recipient mice were harvested and analyzed by IHC and/or flow cytometry. To exclude probe-specific cell toxicity, the fluorescent cell tracers were switched accordingly, showing identical experimental results.

Tumor homing assay

OT-I T cells were isolated from transgenic WT and Plxna4 KO OT-I mice, generated by the intercross of Plxna4 heterozygous mice with OT-I–positive mice. These mice have a monoclonal population of naïve T-cell receptor (TCR) transgenic CD8+ T cells (OT-I T cells) that recognize the immunodominant cytosolic chicken OVA “SIINFEKL” peptide. For activation of OT-I T cells, total splenocytes from OT-I mice were isolated and cultured for 3 days in T-cell medium with 1 μg/mL SIINFEKL peptide (IBA - LifeSciences, 6–7015–901) and 10 ng/mL mIL2. At day 3 of activation, OT-I T cells were further expanded for a maximum of 3 additional days in the presence of 10 ng/mL mIL2.

For the tumor homing assay, activated WT and Plxna4 KO OT-I T cells were labeled with either 3.5 μmol/L Violet Cell Tracer or 1 μmol/L CFSE and injected i.v. with a 1:1 mixture between 2–3 × 106 WT and Plxna4 KO OT-I T cells into WT recipient mice with established B16F10-OVA or LLC-OVA tumors. The tumors of recipient mice were harvested 24 and 48 hours after T-cell transfer and analyzed by flow cytometry.

Liver homing assay

WT mice received a plasmid DNA by hydrodynamic injection (HDI). Each mouse was injected rapidly (<8 seconds) in the tail vein with 40 μg of pcDNA3 empty vector (EV) or pcDNA3-OVA (OVA) diluted in PBS in an injection volume of 10% of the body weight. Four days after HDI, activated WT and Plxna4 KO OT-I T cells were labeled with either 3.5 μmol/L Violet Cell Tracer or 1 μmol/L CFSE and injected i.v. with a 1:1 mixture between 2–3 × 106 WT and KO OT-I T cells into mice. Twenty-four hours after T-cell transfer, blood and livers were harvested and analyzed by flow cytometry.

Plasmids and lentiviral vectors

In the overexpression experiments, the following plasmids were used: pCDH-CMV-Sema3a-DYK-EF1-Puro (Sema3a OE), pCDH-CMV-Sema6a-DYK-EF1-Puro (Sema6a OE), pCDH-CMV-Sema6b-DYK-EF1-Puro (Sema6b OE), and pCDH-CMV-MCS-EF1-Puro (EV). B16F10-OVA cancer cells were transduced with concentrated lentiviral vectors and further selected with puromycin antibiotics (1 μg/mL, Sigma, P9620) to allow the generation of a homogenous population of overexpressed (and empty vector control) cancer cells.

GTPase pull-down assay

Rac1 and Rap1 activation were measured by using a Rac1 or Rap1 Activation Assay Kit (Thermo Fisher Scientific, 16118 and 16120, respectively) according to the manufacturer's instructions. Briefly, cell lysis was performed by incubating activated T cells (at day 5 of stimulation) with the lysis buffer for 5 minutes on ice. The lysates were centrifuged for 15 minutes at 16,000 × g and subsequently incubated with the glutathione S-transferase (GST)-fused with: (i) p21-binding domain of Pak1 (GST-Pak1-PBD, 20 μg) or (ii) RalGDS-binding domain of Rap1 (GST-RalGDS-RBD, 20 μg), bound to glutathione resin at 4°C for 60 minutes with gentle rocking. After being washed three times with lysis buffer, the samples were eluted in 2× SDS reducing sample buffer and analyzed for bound Rac1 (GTP-Rac1) or Rap1 (GTP-Rap1) by Western blot analysis.

Western blotting

Protein extraction of liver samples was performed by using a homemade RIPA lysis buffer (50 mmol/L Tris HCl pH 8, 150 mmol/L NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with Complete Protease Inhibitor Cocktail (Roche, 11697498001) and PhosSTOP Phosphatase Inhibitor (Roche, 04906837001). Lysates were incubated on ice for 30 minutes before centrifuging for 15 minutes at 4°C to remove cellular debris. Protein concentration of cell extracts was determined by using Pierce bicinchoninic acid (BCA) reagent (Thermo Fisher Scientific, 23227) according to the manufacturer's instructions. Protein samples were denaturated by adding a homemade 6X 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%), incubated at 95°C for 5 minutes. Samples containing equivalent amounts of protein were subjected to 12% SDS-PAGE. Proteins were transferred onto a nitrocellulose membrane using the Trans-Blot Turbo Transfer System (Bio-Rad) according to manufacturer's instructions. The membranes were blocked for nonspecific binding in 5% nonfat dry milk (Cell Signaling Technology, 9999S) in homemade Tris-buffered saline-Tween 0.1% (50 mmol/L Tris HCl pH 7.6, 150 mmol/L NaCl, 0.1% Tween; TBS-T) for 1 hour at room temperature and incubated with primary antibody overnight at 4°C. The following antibodies were used: mouse anti-Rac1 (Thermo Fisher Scientific, 16118, 1:1,000), rabbit anti-Rap1 (Thermo Fisher Scientific, 16120, 1:1,000), mouse anti-Vinculin (Sigma-Aldrich, V9131, 1:200), and mouse anti-OVA (Abcam, ab17293, 1:500). After incubation with the primary antibodies, the membranes were washed for 15 minutes in TBS-T and incubated with the appropriate secondary antibody (1:5,000 in 5% nonfat dry milk in TBS-T) for 1 hour at room temperature. The following secondary antibodies were used: goat anti-mouse and goat anti-rabbit IgG-HRP (Santa Cruz Biotechnology, sc-2005 and sc-2004, respectively). The signal was visualized with Enhanced Chemiluminescent Reagents (ECL; Invitrogen, WP20005) or SuperSignal West Femto Chemiluminescent Substrate (Thermo Fisher Scientific, 34094) with a digital imager (ImageQuant LAS 4000, GE Health Care Life Science Technologies). The results of the GTPase pull-down assay were normalized against the corresponding band of the total proteins.

Adoptive T-cell transfer

Adoptive T-cell transfer (ACT) experiments were performed with either naïve or activated OT-I T cells. WT recipient mice carrying subcutaneous LLC-OVA or orthotopic B16F10-OVA tumors (average tumor size of 30–50 mm3) were injected i.v. with either PBS, 1–3 × 106 WT or the same number of Plxna4 KO OT-I T cells. Starting from the day of ACT, recipient mice were injected daily i.p. with 5 μg/mouse of recombinant human IL2 in a volume of 200 μL of PBS for 4 consecutive days. Recipient mice were additionally treated i.p. three times per week with 10 mg/kg anti–PD-1 (RMP1–14, BioLegend) or the appropriate isotype control, starting from an average tumor size of 200 mm3. The tumors were measured every day and were weighted and collected at the end stage for flow cytometric analysis.

Human samples

Blood samples were freshly collected from patients with metastatic melanoma before and 3 weeks after the first cycle of ICI therapy (anti–PD-1 alone or in combination with anti–CTLA-4). Response assessment of patients with melanoma with stage III and IV nonresectable disease was performed as per response evaluation criteria in solid tumors (RECIST v1.1; ref. 20). Patients with complete or partial responses were categorized as responders, while nonresponders only achieved stable or progressive disease as their best overall response. Patients with resectable stage III disease all underwent complete LN dissection, which coincided with the “on-treatment” sampling time point. Pathologic response was assessed on the resection specimen. Patients with a complete response were categorized as responders, while patients without pathologic complete response were categorized as nonresponders. All relevant clinicopathologic information of the human subjects is provided in Supplementary Table S3. Inclusion and exclusion criteria for the study can be found in Supplementary Table S4. The research using human samples was conducted according to institutional and European Union ethical standards, and all subjects ensured written informed consent to participate in this study.

In brief, peripheral blood mononuclear cells from patients and healthy volunteers were immediately isolated by density gradient centrifugation using Lymphoprep (Stemcell, 07811). CD4+ and CD8+ T cells were negatively selected using MojoSort Human CD4 and CD8 T Cell Isolation Kit (Miltenyi Biotec, 480010 and 480129, respectively) according to manufacturer's instructions. For the expression analysis of circulating monocytes, cDNA samples from patients with different tumor types and age-matched healthy controls (Supplementary Table S5) were provided by the “Monomark" clinical study (21).

Statistical analysis

Data entry and all analyses were performed in a blinded fashion. All statistical analyses were performed using GraphPad Prism Software (Version 9.2). Pairwise comparisons on two experimental conditions were performed using an unpaired Student t test or a paired t test for competition assays. Grouped data were assessed by two-way ANOVA with Bonferroni multiple comparison correction. Survival curves were compared with the log-rank (Mantel–Cox) test. Statistical details of the experiments can be found in the figure legends. Detection of mathematical outliers was performed using the Grubbs test in GraphPad. Sample sizes for all experiments were chosen based on previous experiences. All graphs show mean values ± SEM.

ResultsGenetic knockout of Plxna4 in the stroma inhibits tumor progression and increases CTL infiltration

To study the role of Plxna4 in the TME, we took advantage of Plxna4 KO mice (Supplementary Fig. S1A; ref. 22). Compared with WT controls, Plxna4 KO mice were phenotypically identical and had similar blood counts (Supplementary Table S6). By implanting LLC cancer cells s.c., we observed a significantly slower tumor growth in Plxna4 KO versus WT mice (Fig. 1A and B). Because previous experiments with human umbilical vein endothelial cells (HUVEC) have shown the involvement of Plxna4 in basic fibroblast growth factor (bFGF)-induced angiogenic signaling (23), we analyzed tumor blood vessel parameters in WT and Plxna4 KO mice. Tumor vessel density, perfusion, and pericyte coverage were comparable between WT and Plxna4 KO mice (Supplementary Fig. S1B–S1D), resulting in no differences in tumor hypoxic areas (Supplementary Fig. S1E). Plxna4 was also reported to be part of the signaling complex involved in the positioning of tumor-associated macrophages (TAM) inside hypoxic niches (17). However, we could not observe any difference in either TAM infiltration (Supplementary Fig. S1F) or localization within hypoxic regions (Supplementary Fig. S1G and S1H). In addition, gene expression markers typically used to characterize classically (M1-like) and alternatively activated (M2-like) macrophages were unaltered in sorted TAMs from WT and Plxna4 KO tumor-bearing mice (Supplementary Fig. S1I), suggesting that, at least in these conditions, Plxna4 is not required for macrophage localization or polarization within the TME.

Figure 1.

Genetic knockout of Plxna4 in the stroma or in the hematopoietic lineage abates tumor progression and increases CTL infiltration. Subcutaneous LLC tumor growth (A) and weight (B) in WT and Plxna4 KO mice. C, Flow cytometric analysis of CTLs in the tdLNs from WT and Plxna4 KO mice bearing subcutaneous LLC tumors. Histologic quantification of CTLs in the inner tumor bed (D), representative micrographs (scale bar, 50 μm; E), and paired analysis of the inner (core) and outer (border) tumor areas of LLC tumor sections obtained from WT and Plxna4 KO tumor-bearing mice (F). Flow cytometric analysis of CTLs in the tdLNs (G) and orthotopic B16F10 tumors (H) from WT and Plxna4 KO tumor-bearing mice. LLC tumor growth (I) and tumor weight (J) of lethally irradiated WT mice reconstituted with WT (WT→WT) or Plxna4 KO (KO→WT) BM cells. Tumor growth (K) and tumor weight (L) in an orthotopic E0771 breast cancer model. Flow cytometric analysis of CTLs in the tdLNs (M) and primary tumor (N) from WT→WT and Plxna4 KO→WT chimeras. For the in vivo experiments, n = 4–8 mice per group. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001 versus WT (A–D, G, and H), CTLs in the tumor border (F) and versus WT→WT (I–N). All graphs show mean ± SEM.

Plxna4 was described as a negative regulator of T-cell–mediated immune responses (18, 24), so we therefore investigated if the tumor-suppressing phenotype observed in Plxna4 KO mice was related to T-cell functions. Flow cytometric analysis showed that tumor-bearing Plxna4 KO versus WT mice had increased numbers of CTLs in the tdLNs (Fig. 1C; Supplementary Fig. S1J). Histologically, we found that mice lacking Plxna4 had increased infiltration of CTLs into the core of the tumor when compared with WT mice (Fig. 1D). Paired analysis of CTLs in the outer area versus the inner area of the same tumor suggested that, compared with their WT counterparts, Plxna4 KO CTLs had increased capacity of migrating from the outer rim into the core of the tumor (Fig. 1E and F). Therefore, we hypothesized that Plxna4 KO CTLs subvert a Plxna4-dependent T-cell exclusion mechanism seen in WT mice. Consistently, in an orthotopic B16F10 melanoma model, we observed a higher infiltration of CTLs, in both the tdLN (Fig. 1G) and primary tumor (Fig. 1H) of Plxna4 KO versus WT mice. When looking more closely into the different CD8+ T-cell subsets (naïve, central memory, effector/memory, and terminally exhausted T cells), the proportion of each CD8+ T-cell subset in both the tdLNs and tumors remained the same in Plxna4 KO versus WT mice (Supplementary Fig. S1K and S1L). Conversely, in both tumor models, total CD4+ T cells did not change, either in numbers (Supplementary Fig. S1M–S1P) or in their localization within the TME (Supplementary Fig. S1Q and S1R).

To further restrict the KO of Plxna4 to the immune system, we generated BM chimeras by transplanting BM cells from WT or Plxna4 KO mice into lethally irradiated WT recipient mice, WT→WT and Plxna4 KO→WT, respectively. Upon reconstitution, Plxna4 KO→WT chimeras displayed normal blood counts, comparable with those of WT→WT mice (Supplementary Table S7). Upon subcutaneous engraftment, LLC tumor growth in Plxna4 KO→WT chimeras was slower than in WT→WT chimeras (Fig. 1I and J), resembling our results in Plxna4 KO mice (Fig. 1A and B). In an alternative tumor model, obtained by the orthotopic injection of E0771 breast cancer cells, tumor progression was significantly decreased upon deletion of Plxna4 in the BM (Fig. 1K and L). Similar to Plxna4 KO mice, higher numbers of CTLs, but not of CD4+ T cells, were found in both the tdLN (Fig. 1M; Supplementary Fig. S1S) and primary tumor (Fig. 1N; Supplementary Fig. S1T) in Plxna4 KO→WT compared with WT→WT chimeras. Altogether, these results show that a Plxna4-deficient tumor stroma or immune system results in impaired tumor growth and in the selective increase of CTLs inside both the tdLN and the tumor core.

Plxna4 expression is dynamically regulated in CTLs

Elevated CTL infiltration in the TME correlates with a good prognosis in several tumor types (4). Because we detected higher CTL numbers in the tumor core of Plxna4 KO mice (Fig. 1D, H, and N), we investigated the direct role of Plxna4 in these cells. First, we observed that Plxna4 expression was upregulated in activated CTLs upon 3 days of anti-CD3/CD28 stimulation in vitro (Fig. 2A). Purified CD4+ T cells showed similar kinetics of Plxna4 expression, but at about one third the abundance of CTLs, both in the naïve and active state (Fig. 2A). In line with regulation at the transcript level, protein staining revealed that the number of Plxna4-expressing activated CTLs was increased compared with naïve cells (Fig. 2B). Similar to murine data, the expression of PLXNA4 in human T cells was also upregulated upon later stages of T-cell activation, and this to a greater degree in CTLs than in CD4+ T cells (Fig. 2C). On the basis of these observations, we characterized the expression of Plxna4 in CTLs sorted from the circulation of healthy or tumor-bearing mice. Compared to healthy mice, Plxna4 was upregulated in CTLs in both an orthotopic melanoma model (B16F10) and a subcutaneous LLC lung cancer model (Fig. 2D). Plxna4 was also upregulated in circulating CTLs after mouse infection with a LCMV-OVA, which has a specific tropism for dendritic cells (Supplementary Fig. S2A; ref. 19). These data suggest an involvement of Plxna4 in CTLs upon antigen recognition.

Figure 2.

Plxna4 expression is dynamically regulated in CTLs. A, Time course of Plxna4 expression in purified mouse CD8+ and CD4+ T cells activated with anti-CD3/CD28 for 3 days and further expanded in the presence of IL2. B, Representative images of Plxna4 cytospin staining on WT and Plxna4 KO CTLs before and after 3 days of anti-CD3/CD28 activation (scale bar, 50 μm). C, Time course of Plxna4 expression in purified human CD8+ and CD4+ T cells activated with anti-CD3/CD28 for 3 days and further expanded in the presence of IL2. D, Plxna4 expression in circulating CTLs sorted from healthy mice and orthotopic B16F10 or subcutaneous LLC tumor–bearing mice. E, Plxna4 expression in naïve T cells (CD8+CD44LoCD62LHi), effector/memory T cells (CD8+CD44HiCD62LLo), and central memory T cells (CD8+CD44HiCD62LHi) sorted from the circulation and LNs of healthy mice. F, Plxna4 expression in naïve T cells (CD8+CD44LoCD62LHi), effector/memory T cells (CD8+CD44HiCD62LLo), central memory T cells (CD8+CD44HiCD62LHi), and terminally exhausted T cells (CD8+PD-1HiTIM-3Hi) sorted from the circulation, tdLNs, and B16F10 orthotopic tumors of tumor-bearing mice. G, Plxna4 expression in CD8+CD4Lo and CD8+CD44Hi T cells sorted from the circulation and liver of hydrodynamically vaccinated mice. For the in vivo experiments, n = 3–5 mice per group (D–G). In vitro results (A and C) were performed in triplicates and are representative of two independent experiments. *, P < 0.05; **, P < 0.01; and ***, P < 0.001 versus naïve and activated mouse CD4+ T cells (A), naïve and activated CD4+ human T cells (C), circulating CTLs in healthy mice (D), effector/memory T cells in LNs or corresponding circulating T-cell subset (E), effector/memory T cells present in the same tissue or corresponding circulating T-cell subset (F), circulating CD8+CD44Hi T cells, and CD8+CD44Hi T cells in the liver (G). All graphs show mean ± SEM.

To assess the relation between Plxna4 expression, activation status and tissue of origin, we sorted different CD8+ T cell subsets from the blood and LNs in healthy mice (Supplementary Fig. S2B), or from the blood, tdLN, and tumor tissue in B16F10 tumor–bearing mice (Supplementary Fig. S2C). In healthy mice, Plxna4 was expressed in all the circulating T-cell subsets, but it was decreased in effector/memory T cells or undetectable in naïve and central memory T cells sorted from the LNs (Fig. 2E). In B16F10 tumor–bearing mice, Plxna4 abundance in circulating naïve and central memory T cells were comparable with those measured in healthy mice but was strongly augmented in circulating effector/memory T cells, which have encountered the antigen in the tdLNs and consequently express CD44 (Fig. 2F). However, in the tdLNs or in the tumor bed, Plxna4 expression in effector/memory T cells was reduced and was undetectable in naïve, central memory, and exhausted CTLs, the latter being the most abundant subset in the tumor bed (Fig. 2F; Supplementary Fig. S2C). Altogether, our data suggest that T-cell activation increases Plxna4 in effector/memory CTLs, but this expression is downregulated in all T-cell subsets when entering the inflammatory site. These observations held also true in a liver inflammation model, induced by HDI of a plasmid directing the expression of OVA (25). In this model, circulating antigen-primed CD8+CD44Hi T cells expressed more Plxna4 than their naïve CD44Lo counterparts (Fig. 2G). Similar to what was observed in tumor-infiltrating CTLs (Fig. 2F), Plxna4 expression was also halved in CD8+CD44Hi T cells infiltrating inflamed livers (Fig. 2G). Altogether, these data show that Plxna4 expression is induced in effector/memory CTLs upon T-cell activation and sustained in circulation, while reduced at the inflammatory sites.

Genetic knockout of Plxna4 in CTLs increases their motility and proliferation via enhanced Rac1 activity

As the expression of Plxna4 in circulating CTLs was increased in tumor-bearing mice compared with healthy mice, but decreased upon their infiltration into the tumor bed (Fig. 2D and F), we studied the functional relevance of Plxna4 in CTLs. We analyzed the proliferation of WT and Plxna4 KO CTLs in an in vitro time course experiment. From day 3 of activation, Plxna4 KO CTLs showed a higher proliferation rate than WT cells (Fig. 3A and B; Supplementary Fig. S3A). This difference in proliferation was increasing with time, reaching almost two-fold higher rate at day 5 of stimulation (Fig. 3A), consistent with the observation that Plxna4 expression increases over time following stimulation (Fig. 2A). In contrast, the apoptotic rate of WT and Plxna4 KO CTLs at different timepoints following stimulation did not change (Supplementary Fig. S3B). Downstream effector functions of in vitro activated CTLs were also not affected, because the analysis of IFNγ and granzyme B (GrzmB) expression did not show any differences between WT and Plxna4 KO CTLs (Supplementary Fig. S3C–S3F). Unlike CTL proliferation, expansion of activated CD4+ T cells did not change significantly between WT and Plxna4 KO T cells (Supplementary Fig. S3G).