Immune-Checkpoint Protein VISTA Regulates Antitumor Immunity by Controlling Myeloid Cell–Mediated Inflammation and Immunosuppression

The B7 family of immune-checkpoint receptors are valuable targets for cancer immunotherapy. Antibodies targeting two B7 family coinhibitory receptors, CTLA-4 and PD-1, have elicited durable clinical outcomes in previously refractory cancer types, and are considered a breakthrough therapy for cancer (1, 2). Despite this success, the response rate following CTLA-4, PD-L1, or PD-1–blocking monoclonal antibody (mAb) therapies is generally less than 30%, indicating that additional nonredundant immune-checkpoint pathways hamper antitumor immunity (2).

VISTA (gene Vsir, RIKEN cDNA 4632428N05, also known as Gi24, Dies-1, PD-1H, and DD1α) stands for V-domain immunoglobulin suppressor of T-cell activation and is an inhibitory B7 family immune-checkpoint molecule (3, 4). VISTA is constitutively expressed on multiple immune cell types such as CD11b⁺ myeloid cells, naïve CD4⁺ and CD8⁺ T cells, Foxp3⁺CD4⁺ regulatory T cells, and TCRγδ T cells. Studies from our group and others have shown that VISTA controls T cell–mediated autoimmunity and antitumor immunity (3, 5). Although other immune-checkpoint regulators such as CTLA-4 and PD-L1/PD-1 control T-cell activation by recruiting SHP1/2, which antagonizes proximal T-cell receptor (TCR) signaling (6, 7), VISTA plays a broader role in regulating both myeloid cell–mediated and T cell–mediated immune responses (4). Our previous studies in a murine model of psoriasiform inflammation show that VISTA inhibits TLR7-induced IL23 production in myeloid dendritic cells (DC) and dampens the IL17 production in TCRγδ T cells (8). VISTA expressed on myeloid antigen-presenting cells (APC) engages an unidentified inhibitory receptor on CD4⁺ and CD8⁺ T cells and suppresses their proliferation and cytokine production (3). VISTA expressed on CD4⁺ T cells also limits T-cell activation in a T cell–intrinsic manner (9).

VISTA knockout (KO, Vsir^−/−) mice demonstrate the loss of T cell–mediated peripheral tolerance and develop inflammatory phenotypes in multiple organs (10). When bred onto an autoimmune-prone background, VISTA deficiency accelerates disease development in the experimental autoimmune encephalomyelitis model of multiple sclerosis and the sle1.sle3 model of lupus (10, 11).

The critical role of VISTA in regulating antitumor immunity has been demonstrated by genetic deletion of VISTA gene (Vsir) or treatment with a VISTA-blocking mAb (4, 10, 12, 13). Blocking VISTA deters tumor growth in multiple preclinical models (10, 12), and use of a VISTA-blocking mAb further synergizes with either the PD-L1–blocking mAb or a Toll-like receptor (TLR) agonistic peptide vaccine and CD40-specific agonistic mAb to optimally control tumor growth and achieve long-term survival (12, 13). The mechanism of synergy has been attributed to the role of VISTA in directly inhibiting TCR signaling and T-cell activation (13). Whether or not VISTA regulates the activation of myeloid cells that indirectly control tumor-specific T-cell responses has not been elucidated.

This study uncovered a function of VISTA in regulating the TLR/TRAF6-mediated signaling axis and the effector function in myeloid cells. Blocking VISTA together with a TLR-agonistic vaccine abolished the suppressive functions of myeloid-derived suppressor cells (MDSC) and tumorigenic DCs and augmented the production of proinflammatory cytokines. These effects collectively led to a T cell–permissive tumor microenvironment (TME) that facilitated tumor rejection. The reprogramming of tumor-associated myeloid cells contributes to the protective antitumor immunity following VISTA inhibition.

C57BL/6 (wild-type, WT) mice were purchased from The Charles River Laboratories. Vsir^−/− mice on a fully backcrossed C57BL/6 background were obtained from Mutant Mouse Regional Resource Centers (www.mmrrc.org; stock # 031656-UCD; refs. 10, 13). Myd88^−/− mice were purchased from The Jackson Laboratory. OT-I CD8⁺ TCR transgenic mice were purchased from Jax (stock # 003831). Animals were maintained in a specific pathogen-free facility at the Medical College of Wisconsin, Milwaukee, WI. All animal protocols were approved by the institutional animal care and use committee of the Medical College of Wisconsin. All methods were performed in accordance with the relevant guidelines and regulations.

B16-BL6 murine melanoma cells were cultured in RPMI-1640 supplemented with 10% FBS, 0.1% β-mercaptoethanol, and penicillin (100 U/mL)/streptomycin (100 μg/mL). B16-BL6 (female origin) was originally obtained from I. Fidler (The University of Texas MD Anderson Cancer Center, Houston, TX) and were verified by in vivo growth in syngeneic mice and their expression of melanoma antigens TRP1, TRP2, and gp100 but have not been authenticated further by other methods. Murine thymoma EG7 (female origin), human THP-1 (male origin) monocytes, and human HEK293T cells were authenticated and obtained from ATCC and were cultured in the same RPMI-1640 media as above. To generate a VISTA-overexpressing THP-1 cell line, THP-1 cells were transduced with lentivirus carrying human VISTA gene that was cloned into a lentiviral vector pFLRcmv-FH-puro. Transduced cells were stained with VISTA-specific mAb and FACS sorted to homogeneity. Lentivirus carrying the empty vector was used to generate the control THP-1 cell line. All cell lines were used at fewer than five passages and after short-term (less than 1 week) culture time. Cell lines were tested for Mycoplasma contamination every 5 to 6 months, using MycoAlert Mycoplasma Detection Kit (Lonza).

WT or Vsir^−/− mice (two to three of each genotype) were treated with 3 mL 3% thioglycollate broth (cat. # DF0430-17-1, Thermo Fisher Scientific). Peritoneal macrophages were harvested from peritoneal lavage on day +4 following thioglycollate injection. Cells were rested in RPMI media supplemented with 1% FBS, 2 mmol/L L-glutamine, and 50 μmol/L β-mercaptoethanol for 1 hour before stimulation with TLR agonists including LPS (1 μg/mL), CpG (1μg/mL; OND1826), Poly (I:C)(25 μg/mL), Pam3cys4 (10 μg/mL), and R848 (5 μg/mL). All TLR agonists were purchased from InvivoGen. mRNA samples were collected after 3 hours using the RNeasy Mini Kit (QIAGEN) following the manufacturer's instructions. Expression of genes Il12p35, Il12p40, and Il6 was examined by real-time qRT-PCR as described below. Culture supernatants were harvested after 5 hours and examined by ELISA to assess the level of cytokines IL6, IL12p40, GM-CSF, IFNγ, MIP-1α, and MIP-1β as described below. To determine the effect of TNFα, macrophages were stimulated with CpG (1 μg/mL) in the presence of a TNF receptor–specific blocking mAb (Bio X Cell) or isotype control IgG (25 μg/mL) for 6 hours before culture supernatants were harvested and examined by ELISA.

For measuring TLR-mediated signaling, macrophages or THP-1 cells were stimulated with Pam3 or CpG, respectively, for 0, 15, and 30 minutes. When indicated, cells were pretreated with MG132 (10 μg/mL; Sigma-Aldrich) or DMSO vehicle control for 30 minutes before TLR stimulation. Cells were lysed in RIPA buffer (50 mmol/L Tris pH 7.4, 150 mmol/L NaCl, 1.0% NP-40, 0.5% deoxycholic acid, 0.1% SDS, 1 mmol/L EDTA, 1 mmol/L EGTA, 5 mmol/L sodium pyrophosphate, 50 mmol/L NaF, 10 mmol/L β-glycerophosphate) supplemented with 1× Halt Protease and Phosphatase Inhibitor Cocktail (cat. # 78442, Thermo Fisher Scientific) at indicated time points. Phosphorylated and total levels of proteins including Jnk1/2, Erk1/2, p38, IΚB, and IKKα/β were examined by Western blotting as described below.

For denaturing immunoprecipitation (IP) of TRAF6, WT or Vsir^−/− macrophages were lysed in lysis buffer (1% NP-40, 50 mmol/L Tris-HCl, pH 7.5, 150 mmol/L NaCl, 5 mmol/L EDTA) containing 1% SDS and denatured by heating for 5 minutes. Lysates (200 μg) were diluted with SDS-free lysis buffer until the concentration of SDS reached 0.1%. TRAF6 was immunoprecipitated using a mouse mAb (4 μg for 200 μg lysates; sc-8409) from Santa Cruz Biotechnology and 20 μL protein G agarose beads (MilliporeSigma). Bound proteins were eluted using 1× SDS sample buffer and analyzed by Western blotting as described below. For HEK293T cells that were transfected with HA-tagged mutant ubiquitin and VISTA-expressing or control vectors, cells were stimulated with R848 (10 μg/mL) for 15 minutes before lysed in the denaturing IP buffer, and TRAF6 was immunoprecipitated as described above.

Electrophoretic mobility shift assays (EMSA) were used to measure the DNA binding activity of the transcription factors AP-1 and NF-κB, using radiolabeled DNA probes as previously described (14). Briefly, WT or Vsir^−/− macrophages (2 × 10⁶ to 3 × 10⁶ cells) were lysed in 50 μL buffer containing 20 mmol/L HEPES pH 7.9, 350 mmol/L NaCl, 1 mmol/L MgCl₂, 0.5 mmol/L EDTA, 0.5 mmol/L DTT, 20% glycerol, and 1% NP-40. Cell lysates (3 μL) were incubated with [32P]-labeled NF-κB probe (cat. # SC-2505) or AP-1 probe (cat. # SC-2501; 25,000 cpm per reaction; Santa Cruz Biotechnology) for 15 minutes at room temperature and then resolved on a 4% polyacrylamide gel.

Single-cell suspensions of tumor-draining lymph nodes (LN) were obtained by passing cells through 70-μm cell strainers. For peritoneal macrophages, cells (1–2 million) were stimulated with CpG (1 μg/mL) or R848 (5 μg/mL) for 3 hours before RNA isolation. Total RNA of LN cells or macrophages was prepared using the RNeasy Kit (QIAGEN). One microgram of total RNA was reverse transcribed in 50 μΛ of TaqMan reverse transcription reagents (Applied Biosystems) using random hexamer primers and according to the manufacturer's instructions. Ten nanogram cDNA and 1 μmol/L primers were used in a 20 μL PCR reaction with SYBR Green Supermix (Bio-Rad). PCR reaction was performed in the Bio-Rad iQ5 Real-Time PCR System (Bio-Rad). Each sample has triplicate replication, and the 2^–ΔΔCT method was used to determine gene expression. Expression of the target genes was normalized to 18s and displayed as fold change relative to the WT sample. Primers were purchased from Integrated DNA Technologies.

Primer sequences are described as follows: Il23-p19 (forward: CCAGCAGCTCTCTCGGAATC; reverse: TCATATGTCCCGCTGGTGC); IL12 p40 (forward: ACA GCA CCA GCT TCT TCA TCA; reverse TCT TCA AAG GCT TCA TCT GCA A); Ifnb (forward: CAGCCCTCTCCATCAACTATAAG; reverse: TCTCCGTCATCTCCATAGGG); Il6 (forward: GCAGAAAAAGGCAAAGAATC; reverse: CTACATTTGCCGAAGAGC); Tnfa (forward: AGGCAGTCAGATCATCTTC; reverse: TTATCTCTCAGCTCCACG); and 18s RNA (forward: AACCCGTTGAACCCCATT; reverse: CCATCCAATCGGTAGTAGCG).

For Western blotting, antibodies specific for K63-linked and K48-linked polyubiquitin chains, p-Erk1/2 (Thr202/Tyr204), Erk1/2, p-Jnk1/2 (Thr183/Tyr185), Jnk1/2, p-p38, p38, p-IKKα/β, IKKα/β, p-NF-κBp65, NF-κBp65, IκB, TRAF6, pTAK1 (Thr184/187), and TAK1 were purchased from Cell Signaling Technology. Antibodies specific for actin were purchased from Santa Cruz Biotechnology. VISTA-blocking mAb was as described previously (12). HA-tagged mutant ubiquitin pRK5-HA-Ubiquitin-K63 and pRK5-HA-Ubiquitin-K48 were a gift from Ted Dawson (Addgene plasmid #17606 and #17065; ref. 15). All primary antibodies were used at 1:1,000 dilution. HRP-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories) were used at 1:5,000 dilution.

THP-1 cells or peritoneal macrophages (5 × 10⁶ to 10 × 10⁶) stimulated with relevant TLR agonists were lysed in RIPA buffer (50 mmol/L Tris pH 7.4, 150 mmol/L NaCl, 1.0% NP-40, 0.5% deoxycholic acid, 0.1% SDS, 1 mmol/L EDTA, 1 mmol/L EGTA, 5 mmol/L sodium pyrophosphate, 50 mmol/L NaF, and 10 mmol/L β-glycerophosphate) supplemented with Halt Protease and Phosphatase Inhibitor Cocktail (100×; Thermo Fisher Scientific) at indicated time points. Total lysates were cleared by spinning at 12,000 rpm for 10 minutes. Lysates (20 μg) were subjected to SDS-PAGE and Western blotting. For detecting TRAF6 in WT or Vsir^−/− macrophages and HEK293T cells, 50% of the eluates from the denaturing IP (described above) were used for SDS-PAGE and Western blotting.

All primary antibodies were used at 1:1,000 dilution. Appropriate secondary antibodies were used at 1:5,000 dilution. Proteins were detected using Clarity Western ECL Substrate (Bio-Rad) and X-ray films. Results were scanned using a digital scanner at 600 dpi resolution and quantified using ImageJ software.

For ELISA analyses, culture supernatants from peritoneal macrophages were diluted 5-fold, homogenized tumor tissues were diluted 5-fold, plasma samples were diluted 5-fold, and culture supernatants from MDSC suppression assays were diluted 50-fold. All dilutions were using PBS with 2% BSA as diluent. ELISA kits for detecting murine IL12p40, IL12p70, IL23p19, IL6, TNFα, IFNβ, IL27, GM-CSF, MIP-1α, MIP-1β, CCL4, CCL5, CXCL9, and CXCL10 were purchased from BioLegend. OD₄₅₀ values were read using Typhoon 9400.

EG7 (150,000) or B16-BL6 (30,000) tumor cells were inoculated intradermally on the right flanks of female and male C57BL6 mice of 7 to 9 weeks of age. A similar number of female and male mice were pooled and allocated for each experimental group (n = 10). The peptide vaccine mixture consists of CpG (ODN1826, 50 μg), R848 (100 μg), melanocyte antigen peptide TRP1 (106–130; 50 μg), and a mutated TRP2 peptide DeltaV-TRP2 (180–188; 50 μg) dissolved in PBS (16). All peptides were synthesized by Atlantic Peptides. The vaccine mixture was injected subcutaneously on day +3 following tumor inoculation. Tumor-bearing mice were treated subcutaneously with VISTA-specific mAbs or hamster control IgG (200 μg per injection for both) every 2 to 3 days until the analytical endpoint or for 5 weeks if mice remained tumor free. Tumor size was measured with a caliper every 2 to 3 days.

For measuring cytokines within tumor tissues, mice bearing tumors 5 to 6 mm in diameter were treated with CpG (20 μg), R848 (50 μg), and VISTA-specific mAb or control IgG as described above. Tumor tissues were harvested after 3 hours and homogenized using the QIAGEN tissue homogenizer in buffer containing 1× PBS, 0.1% Tween 20, and 1× Halt protease inhibitor cocktail (Thermo Fisher Scientific). The homogenates were cleared by centrifugation at 10,000 × g for 20 minutes. Cytokines and chemokines in the supernatant were detected by ELISA as described above. For detecting cytokines in blood, heparinized blood samples were harvested via cardiac puncture and centrifuged at 2,500 rpm for 10 minutes. Plasma was collected and also examined by ELISA.

Thirty to 40 B16-BL6 tumors (7–8 mm in diameter) were harvested from untreated tumor-bearing C57BL6 WT mice. Tumors were pooled, cut into smaller pieces, and digested for 20 minutes at 37°C in RPMI buffer containing Liberase TL (150 μg/mL) and DNase I (120 μg/mL; Sigma-Aldrich). Single-cell suspensions were generated by passing cells through 70-μm cell strainers. Myeloid cells, including polymorphonuclear MDSCs (PMN-MDSC; CD11b⁺CD11C⁻LY6C^int LY6G⁺), monocytic MDSCs (M-MDSC; CD11b⁺CD11C⁻LY6C^hiLY6G⁻), myeloid conventional DCs (cDC; CD11b⁺CD11C⁺MHCII⁺LY6C⁻F4/80⁻CD103⁻), inflammatory DCs (Inf-DC; CD11b⁺CD11C⁺MHCII⁺LY6C⁺ LY6G⁻FcϵRI⁺), and CD103⁺ DCs (CD103⁺CD11C⁺MHCII⁺ F4/80⁻), were purified from dissociated tumor tissues by FACS using BD FACSAria II, with 99% purity.

To test TLR-mediated cytokine production, purified myeloid cell subsets (10,000 cells) were stimulated with TLR9 agonist CpG (1 μg/mL) in the presence of VISTA-blocking mAb (20 μg/mL) or isotype control IgG (20 μg/mL) as indicated. Culture supernatants were harvested after 24 hours. Secreted cytokines (IL12p40, IL6, and TNFα) were quantified by ELISA as described above.

For testing the T cell–suppressive function of MDSCs, sorted myeloid cDCs and Inf-DCs (10,000) were cocultured with 30,000 whole splenocytes from OT-I CD8⁺ TCR transgenic mice together with ovalbumin peptide 257–264 (1 μg/mL; GenScript), CpG (1 μg/mL), and VISTA-blocking mAb or control IgG (20 μg/mL) as indicated. Culture supernatants were harvested after 48 hours, and secreted IFNγ was quantified by ELISA. For testing the T cell–stimulatory function of CD103⁺ DCs, sorted CD103⁺ DCs (10,000) were cocultured with 30,000 purified OT-I CD8⁺ T cells, ovalbumin peptide 257–264 (1 μg/mL), CpG (1 μg/mL), and VISTA-blocking mAb or control IgG (20 μg/mL) as indicated. Culture supernatants were harvested after 48 hours, and secreted IFNγ was quantified by ELISA.

For flow cytometry, antibodies specific for CD4 (GK1.5), CD8 (53-6.7), CD16/CD32, and IFNγ (XMG1.2) were purchased from BioLegend. Single-cell suspension tumor tissues and spleens were stained with lineage-specific antibodies (CD4, CD8, CD11b, CD11c, Ly6C, Ly6G, MHCII, F4/80, and FcϵRI) to identify CD4⁺ and CD8⁺ T cells and myeloid cell populations. To detect intracellular cytokines in T cells, single-cell suspensions of tumor tissues (2–3 million) were stimulated for 5 hours in RPMI medium containing phorbol 12-myristate 13-acetate (PMA; 50 μg/mL), ionomycin (1 μg/mL), 10% FBS, 2 mmol/L L-glutamine, 50 μmol/L 2-mercaptoethanol, 1% penicillin–streptavidin, 1× monensin, and 1× Brefeldin A (BioLegend). Cells were stained with the LIVE/DEAD Fixable Violet Dead Cell Stain Kit first (Thermo Fisher Scientific), then stained with T-cell–lineage markers in the presence of pan-CD16/CD32 blocking antibody before fixation with 1% paraformaldehyde, and then permeabilized with 0.5% saponin and stained for intracellular IFNγ. Cells were analyzed on an LSR II flow cytometer (BD Biosciences). Data were analyzed with FlowJo version 9.9.6 analysis software (Tree Star).

All graphs and statistical analysis were generated using Prism 7 (GraphPad Software). Unpaired two-tailed t tests were used for comparing WT and Vsir^−/−cells or mice, or comparing treatments using VISTA-blocking mAb or control IgG. Survival differences of tumor-bearing mice were assessed using Kaplan–Meier curves and analyzed by log-rank testing. A P value less than 0.05 was considered as statistically significant. *, P < 0.05; **, P < 0.025; ***, P < 0.005; ****, P < 0.0001.

TLR stimulation initiates MyD88- or TRIF-dependent signaling pathways to induce the expression of proinflammatory cytokines, chemokines, and type I interferons (17). Owing to the high expression of multiple TLRs on peritoneal macrophages, we chose this cellular system as a model to dissect the role of VISTA in regulating TLR-mediated signaling pathways. Peritoneal macrophages were isolated from WT and Vsir^−/− mice following thioglycollate treatment and stimulated ex vivo with TLR agonists CpG (TLR9) and R848 (TLR7; Fig. 1A–C). The expression of cytokines IL12 and IL6 was examined by real-time quantitative RT-PCR (qRT-PCR) and ELISA. CpG and R848 significantly upregulated gene expression and protein production of IL12 and IL6 in Vsir^−/− macrophages (Fig. 1A and B). The expression of multiple cytokines and chemokines, including GM-CSF, IFNγ, MIP-1α, and MIP-1β, was also significantly increased in Vsir^−/− macrophages (Fig. 1C). In addition to CpG and R848, Vsir^−/− macrophages were hyperresponsive toward additional TLR agonists, including Pam3Cys4 (TLR2), LPS (TLR4), and poly (I:C) (TLR3; Fig. 1D).

Although the results above indicated that VISTA directly regulated TLR-mediated signaling, we wanted to assess if cell-extrinsic factors such as TNFα or additional soluble factors may have contributed to the hyperresponses of Vsir^−/− cells in an autocrine manner. We first tested a TNFR-specific neutralizing mAb and found minimal effects (Supplementary Fig. S1A). Next, we harvested supernatants from CpG-treated WT and Vsir^−/− cell cultures (designated as WT and KO sups, respectively) and tested their effects on freshly isolated peritoneal macrophages (Supplementary Fig. S1B). The residual CpG in the supernatants directly stimulated cells, and adding CpG to the culture further boosted cytokine production. Under all tested conditions, WT and KO supernatants stimulated cells to similar levels. These data support the conclusion that the hyperresponsiveness in Vsir^−/− macrophages was due to cell-intrinsic TLR signaling rather than secondary responses to soluble factors.

Transcription factors AP-1 and NF-κB are critical for the gene expression of many inflammatory cytokines and chemokines (18). Previous studies have established the role of MAP kinases (MAPK Erk1/2, Jnk1/2, and p38) in the activation of AP-1 (19, 20). The activation of NF-κB requires the phosphorylation of IκB kinase IKKα/β and degradation of the inhibitor IκB (21). We, therefore, examined the activation of MAPKs/AP-1 and IKK/NF-κB signaling cascades following TLR stimulation. Cell lysates from WT and Vsir^−/− macrophages were stimulated with CpG or R848. Phosphorylation of MAPKs and IKKα/β, as well as total IκB, was examined by Western blotting. Both CpG (Fig. 2A) and R848 (Fig. 2B) induced greater phosphorylation of Erk1/2, Jnk1/2, and p38 in Vsir^−/− macrophages. Phosphorylation of IKKα/β and degradation of IκB were enhanced in Vsir^−/− macrophages following CpG stimulation but were not significantly increased following R848 stimulation.

Next, we carried out EMSAs to measure the DNA binding activity of AP-1 and NF-κB. As shown in Fig. 2C, CpG stimulation induced higher activation of AP-1 and NF-κB in Vsir^−/− macrophages than in WT cells. These results indicate that VISTA inhibited TLR-mediated activation of MAPKs/AP-1 and IKK/NF-κB signaling cascades.

Previous studies show that Vsir^−/− mice develop systemic tissue inflammation (10). It is possible that Vsir⁻/⁻ macrophages were partially activated due to exposure to the existing tissue inflammation. To rule out this possibility and validate the cell-intrinsic role of VISTA, we examined the effects of VISTA overexpression in THP-1 human monocyte cells. VISTA⁺ and control THP-1 cells were stimulated with TLR2 agonist Pam3Cys4 (10 μg/mL) for the indicated time, and total cell lysates were harvested and analyzed by Western blotting (Fig. 3). VISTA expression in THP-1 cells suppressed the phosphorylation of MAPKs (Jnk1/2, Erk1/2, and p38; Fig. 3A and C), the phosphorylation of IKKα/β and NF-κB p65, and the degradation of IκB (Fig. 3B and C) following TLR2 stimulation.

TLRs utilize either MyD88-dependent (TLR2/4/5/7/8/9) or TRIF-dependent (TLR3/4) signaling pathways (22). Following TLR2/4/5/7/8/9 engagement, the MyD88/IRAK1/4 complex recruits and activates the RING-domain E3 ubiquitin ligase TRAF6, which is required for the activation of MAPKs and NF-κB (22). TRAF6 also mediates the activation of NF-κB downstream of TLR3 by binding to TRIF (23). Because Vsir^−/− macrophages are hyperresponsive to all TLR agonists (Figs. 1–3), we hypothesized that VISTA targeted a shared signaling adaptor such as TRAF6 (24). We first examined the total protein level of TRAF6 and observed increased TRAF6 expression in freshly isolated Vsir^−/− macrophages compared with WT cells (Fig. 4A). Because K48-linked polyubiquitination could target proteins to proteasomes for degradation and K63-linked polyubiquitination activates TRAF6 (25), we analyzed the ubiquitination status of TRAF6 after immunoprecipitating TRAF6 under denaturing conditions (Fig. 4B). Western blots showed that VISTA deletion resulted in lower K48-linked polyubiquitination and higher K63-linked polyubiquitination of TRAF6 following CpG stimulation (Fig. 4B).

Next, we utilized HA-tagged mutant ubiquitin containing only K48 (HA-Ub-K48) or K63 (HA-Ub-K63) to determine the effect of transient VISTA expression on TRAF6 ubiquitination. HEK293T cells were cotransfected with mutant ubiquitin together with a VISTA-expressing vector or empty control vector and analyzed at 24 hours after transfection (Fig. 4C). Transient VISTA expression significantly augmented K48-linked polyubiquitination of TRAF6 and simultaneously decreased K63-linked polyubiquitination levels (Fig. 4C).

The TAK1/TAB1/TAB2 complex is recruited to the MyD88/IRAK/TRAF6 complex by binding to the K63-polyubiquitin chains of TRAF6 and mediates the activation of the Jnk1/2/AP-1 and IKKα/β/NF-κB signaling pathways (25). Consistent with the higher protein level and K63 polyubiquitination of TRAF6, phosphorylation of TAK1 at Thr^184/187 in Vsir^−/− macrophages was significantly enhanced compared with WT cells following CpG stimulation (Fig. 4D). Reduced protein expression of TRAF6 and diminished TAK1 activation were also seen in VISTA-expressing THP-1 cells compared with control THP-1 cells following Pam3Cys4 treatment (Fig. 4E and F).

These results prompted us to test whether proteasomal inhibition reversed the inhibitory effects of VISTA. As shown in Fig. 5A, pretreatment with proteasome inhibitor MG132 effectively reversed the inhibitory effects of VISTA in THP-1 cells by rescuing the phosphorylation of Jnk1/2, Erk1/2, p38, and IKKα/β. MG132 pretreatment also enhanced TLR signaling in WT but not VISTA-deficient macrophages (Fig. 5B). These results, therefore, support the conclusion that VISTA downregulates TLR-mediated signaling by promoting TRAF6 protein degradation and restricting the activation of MAPKs/AP-1 and IKKα/β/NF-κB signaling pathways.

We have previously shown that treatment with a VISTA-blocking mAb deters tumor growth in preclinical models (12). Our current study indicated that the proinflammatory responses induced by VISTA blockade promote the development of antitumor immunity. Supporting this hypothesis, treatment with the VISTA-blocking mAb led to a significant accumulation of IL12p40, IL12p70, TNFα, and IL27 in the serum of mice (Fig. 6A). This cytokine response was absent in tumor-bearing MyD88 KO hosts (Fig. 6B). Consistent with the lack of cytokine induction, the therapeutic efficacy of the VISTA-blocking mAb was diminished in MyD88 KO hosts (Fig. 6C). These results indicated that blocking VISTA elicited an endogenous TLR/MyD88-mediated proinflammatory response that contributed to the development of antitumor immune responses.

TLR agonists are promising vaccine adjuvants for cancer therapy (26, 27). We hypothesized that VISTA inhibitors would synergize with TLR-agonistic vaccines owing to an augmented proinflammatory response. We tested this hypothesis in the B16-BL6 melanoma model (Fig. 6C). Mice bearing 3-day established B16-BL6 tumors were treated with either VISTA-blocking mAb, or a vaccine containing TLR7/8/9 agonists (CpG and R848) and peptides derived from melanoma antigens TRP1 and TRP2, or both. Combining VISTA-blocking mAb with the TLR vaccine resulted in tumor-free long-term survival in ∼50% mice, whereas either monotherapy transiently delayed tumor growth without long-lasting effects (Fig. 6C). This stronger therapeutic effect correlated with significantly increased numbers of IFNγ-expressing tumor-infiltrating CD8⁺ and CD4⁺ T cells (Fig. 6D and E).

To dissect the inflammatory TME, we examined the expression of inflammatory mediators within tumor tissues following CpG/R848 treatment (Fig. 6F and H). The VISTA-blocking mAb significantly enhanced the expression of cytokines (IL12p40, IL12p70, IL23p19, IL6, IFNβ, and TNFα) and chemokines (CCL4, CCL5, CXCL9, and CXCL10). To exclude the possibility that this inflammatory response may be due to additional effects of the antibody (i.e., Fc receptor–mediated activation of myeloid cells), we examined WT and Vsir^−/− tumor-bearing mice treated with CpG/R848 (Fig. 6G and I). Consistent with the antibody treatment, Vsir^−/− mice produced higher amounts of inflammatory mediators. In addition to tumor tissues, enhanced cytokine production was detected in tumor-draining LNs (Supplementary Fig. S2). These results indicated that VISTA blockade augments the proinflammatory responses in the periphery and within the TME, which are critical for the development of antitumor T-cell responses.

Tumor tissues are enriched with immunosuppressive MDSCs and dysfunctional DCs (28, 29). Importantly, VISTA is expressed on tumor-associated myeloid cDCs (CD11b⁺CD11C⁺ MHCII⁺LY6C⁻), Inf-DCs (CD11b⁺CD11C⁺MHCII⁺Ly6C⁺Ly6G⁻FcϵRI⁺), PMN-MDSCs (CD11b⁺CD11C⁻Ly6C^intLy6G⁺), and M-MDSCs (CD11b⁺CD11C⁻Ly6C⁺Ly6G⁻), as well as on CD103⁺ DCs (Fig. 7A). This result led us to examine how VISTA directly regulated the effector function of these myeloid cell types in tumor-bearing hosts.

We first analyzed the accumulation of MDSCs and DC subsets in tumor-bearing mice treated with CpG/R848 and VISTA-blocking mAb (Fig. 7B). VISTA inhibition reduced the accumulation of M-MDSCs and PMN-MDSCs and concurrently expanded Inf-DCs within tumor tissues (Fig. 7B). A similar reduction of PMN-MDSCs and expansion of Inf-DCs was observed in the spleen, whereas the number of M-MDSCs in the spleen remained unaltered (Supplementary Fig. S3). The number of CD103⁺ DCs and myeloid cDCs within tumor tissues and spleen was not significantly altered.

Next, we assessed whether VISTA blockade augmented the ability of tumor-associated myeloid cells to produce immune-stimulatory cytokines, such as IL12 (30). We purified MDSCs and DC subsets from tumor tissues and examined their cytokine expression following CpG/TLR9 stimulation. Our results showed that, with the exception of PMN-MDSCs, VISTA blockade significantly augmented the expression of IL12p40 in M-MDSCs, Inf-DCs, myeloid cDCs, and CD103⁺ DCs (Fig. 7C).

In addition to cytokine production, we sought to determine whether VISTA regulated the immunosuppressive functions of tumor-associated myeloid cells. Tumor-associated MDSCs (Fig. 7D and E), Inf-DCs (Fig. 7F), and myeloid cDCs (Fig. 7G) inhibited IFNγ production by OT-I CD8⁺ T cells stimulated with splenic APCs. The VISTA-blocking mAb or TLR stimulation partially reversed the suppressive functions of M-MDSCs, myeloid cDCs, and Inf-DCs, whereas combined treatments maximally abolished suppression (Fig. 7D–F). MyD88 KO M-MDSCs failed to respond to the VISTA-blocking mAb, thus validating the role of MyD88 in mediating the myeloid-intrinsic function of VISTA (Fig. 7D). On the other hand, PMN-MDSCs were not responsive to VISTA inhibition or TLR stimulation (Fig. 7E).

In contrast to the tolerogenic tumor-infiltrating myeloid cDCs and Inf-DCs, Batf3-dependent CD103⁺ DCs are the most potent cross-presenting DCs that prime CTLs at the tumor site (29, 31, 32). We showed that VISTA was expressed on CD103⁺ DCs (Fig. 7A), and treatment with VISTA-blocking mAb significantly augmented IFNγ production from purified OT-I CD8⁺ T cells cocultured with CD103⁺ DCs, whereas CpG treatment had only minimal impact (Fig. 7H). Combining VISTA blockade with CpG treatment led to optimal T-cell activation and IFNγ production (Fig. 7H). Together, these results indicated that blocking VISTA promotes the development of antitumor immunity by reprogramming tumor-associated tolerogenic MDSCs and DCs.

Previous studies have shown that VISTA is expressed on myeloid cells at steady state and in tumor-bearing hosts, but the myeloid-intrinsic functions of VISTA have not been elucidated (3, 12, 33, 34). This study demonstrated that VISTA downregulated the TLR/TRAF6/TAK1 signaling pathway by modulating the polyubiquitination and protein expression of TRAF6. At cellular levels, VISTA regulated the effector functions of MDSCs and DC subsets. Targeting VISTA in tumor-associated myeloid cells induced inflammatory cytokines and promoted antitumor T-cell responses.

TLRs expressed on DCs and other APCs maintain antitumor immunosurveillance by recognizing endogenous danger-associated molecular patterns derived from commensal/invading microbes or necrotic tumors (35, 36). Our study showed that in the absence of a TLR-agonistic vaccine, the efficacy of the VISTA-blocking mAb relied upon endogenous TLR/MyD88-mediated signaling and correlated with the production of immune-stimulatory cytokines such as IL12, IL27, and TNFα. Thus, VISTA inhibited antitumor immunosurveillance at the steady state.

Synthetic TLR agonists such as imiquimod (a TLR7 agonist) and PF-3512676 (CpG ODN for TLR9) are promising vaccine adjuvants (27, 37). We showed that in conjunction with the TLR7/8/9 agonistic vaccine, VISTA blockade augmented the expression of chemokines (CXCL9/10 and CCL4/5) and cytokines (IFNβ, IL12, IL27, IL23, IL6, and TNFα) within tumor tissues. Intratumor CXCL9/10 and CCL4/5 are critical for recruiting Th1 CD4⁺ T cells and CD8⁺ cytotoxic T cells and are associated with better clinical outcomes in various cancer types (38–40). Although IL12, IL27, TNFα, and IFNβ play well-established roles in promoting antitumor T-cell responses (30, 41, 42), IL6 impairs antitumor immunity by inducing STAT3 activation in tumor cells and MDSCs (43). IL23 can both promote and inhibit tumor growth, depending upon the stage of tumor development (44). It remains to be determined whether neutralizing IL6 and IL23 will improve the antitumor efficacy of VISTA inhibitors and reduce immune-related adverse events (irAE).

Tumor tissues are enriched with dysfunctional and immunosuppressive MDSCs and myeloid DCs (29, 45). MDSCs utilize diverse mechanisms such as the production of reactive oxygen species, nitric oxide (NO), and the secretion of arginase I, IL10, and TGFβ to inhibit T-cell responses (45). Tumor-associated myeloid DCs impair T-cell activation by expressing PD-L1 or secreting soluble mediators such as galectin-1 (46, 47), and the frequencies of circulating MDSCs have been correlated with resistance to ipilimumab therapy in melanoma patients (45, 48). Myeloid-inflamed TME profile associates with poor clinical outcomes in pancreatic cancer and head and neck cancer following GVAX vaccine therapy (49). Therefore, effective strategies that reduce the accumulation and suppressive functions of tumorigenic myeloid cells are urgently needed.

Our study showed that VISTA was an immune-checkpoint protein that regulated the immunosuppressive functions of tumor-associated myeloid cells. Blocking VISTA together with TLR stimulation abolished the suppressive function of M-MDSCs and myeloid DCs and induced IL12 production. These myeloid cell–dependent effects converted the immunosuppressive TME into a T cell–stimulatory one that may sensitize tumors to additional immunotherapies. Supporting this notion, we have previously shown that blocking VISTA synergizes with PD-L1 inhibitors in murine tumor models (13).

In contrast to M-MDSCs, the suppressive function of PMN-MDSCs was not reversed by the VISTA-blocking mAb despite high VISTA expression on these cells. This result indicated established that tumors enriched with PMN-MDSCs may be resistant to VISTA-targeted inhibitors. Previous studies have identified approaches to reduce the accumulation and suppressive functions of PMN-MDSCs, including treatment with sildenafil (50), gemcitabine (51), inhibitors of DNA methyltransferase or histone deacetylase (52), and PI3Kδ/γ inhibitor (53). However, it remains to be determined if combining VISTA inhibitors with these reagents will maximally block both PMN-MDSCs, M-MDSCs, and other tumorigenic myeloid DCs, thereby achieving optimal therapeutic outcomes.

In conclusion, this study established VISTA as an immune-checkpoint protein that regulates the functions of myeloid cells and established that myeloid cell activation contributes to the overall antitumor mechanisms of VISTA inhibitors. Unlike VISTA, immune-checkpoint proteins CTLA-4 and PD-1 do not directly regulate the function of myeloid cells (28). Targeting VISTA may be a valuable approach for patients resistant to the existing immune-checkpoint inhibitors. Results from this study provide insights for monitoring myeloid cell–dependent inflammatory responses that precede T cell–mediated antitumor responses following VISTA-targeted therapies.

D. Wang is Adjunct Professor at Medical College of Wisconsin and Fujian Normal University. L. Wang has ownership interest (including stock, patents, etc.) in ImmuNext Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: W. Xu, M.S. Ernstoff, D. Wang, S. Malarkannan, L. Wang

Development of methodology: W. Xu, Y. Zheng, L. Wang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W. Xu, J. Dong, Y. Zheng, J. Zhou, Y. Yuan, H.E. Miller, M. Olson

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): W. Xu, M. Olson, K. Rajasekaran, D. Wang, S. Malarkannan, L. Wang

Writing, review, and/or revision of the manuscript: W. Xu, H.M. Ta, H.E. Miller, M.S. Ernstoff, S. Malarkannan, L. Wang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H.M. Ta, H.E. Miller, M. Olson, S. Malarkannan, L. Wang

Study supervision: D. Wang, S. Malarkannan, L. Wang

Other (performed experiments and analyzed data): K. Rajasekaran

This study was supported by research funding from NCI R01 CA164225 and NCI R01 1R01CA223804 (L. Wang), Advancing A Healthier Wisconsin Research and Education Program (AHW REP) fund (L. Wang), the Office of the Assistant Secretary of Defense for Health Affairs through the Peer Reviewed Cancer Research Program under Award No. W81XWH-14-1-0587 (L. Wang), Worldwide Cancer Research Foundation (UK) research grant 16-1161 (L. Wang), and a Research Scholar Grant, RSG-18-045-01 - LIB, from the American Cancer Society (L. Wang). This work is also supported by NIH R01 AI102893 (S. Malarkannan), NCI R01 CA179363 (S. Malarkannan), Nicholas Family Foundation (S. Malarkannan), Gardetto Family (S. Malarkannan and D. Wang), PO1 CA 206980 and P30 CA 016056 (M.S. Ernstoff), and NIH grants AI079087 and HL130724 (D. Wang).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.