Immune-checkpoint proteins VISTA and PD-1 nonredundantly regulate murine T-cell responses

To determine whether VISTA and PD-1 regulate immune responses in a redundant or independent/synergistic manner, Vista^−/−Pdcd1^−/− mice (herein referred to as VISTA/PD-1 double KO mice) were generated on the C57BL/6 background and characterized. The double KO mice were born fertile and produced normal litter sizes. Normal thymic development and lymphocyte populations (T and B lymphocytes, natural killer cells, and natural killer T cells) in the bone marrow, spleen, and lymph nodes were observed in 6–8-wk-old KO mice.

Comprehensive multiorgan histological analyses were performed in 12-mo-old WT, VISTA KO, PD-1 KO, and VISTA/PD-1 double KO mice (Fig. 1). Hematoxylin and eosin (H&E) stained sections from heart, lung, liver, kidney, pancreas, salivary gland, small and large intestines, and brain were examined. Several organs, including lung, liver, and pancreas in the double KO mice, were heavily infiltrated with leukocytes (Fig. 1A, Top) and showed significant tissue necrosis, presumably as a result of immune cell-mediated destruction (Fig. 1A, Bottom). The levels of leukocyte infiltration and tissue necrosis in the KO mice were blindly quantified on the basis of a semiquantitative scoring method, and the VISTA/PD-1 double KO mice showed the highest scores compared with WT and single KO mice (Fig. 1B). Despite the significant accumulation of activated T cells, the double KO mice did not develop overt autoimmune disease. Serum levels of IgM and IgA were moderately elevated in aged double KO mice (Fig. S1). Our cohoused PD-1 KO mice also developed accumulation of spontaneously activated T cells and chronic inflammation in multiple organs but failed to develop arthritis or other previously reported autoimmune phenotypes (10). This discrepancy might be a result of the different housing conditions or the age of the mice analyzed.

Compared with VISTA KO and PD-1 KO mice, VISTA/PD-1 double KO mice at the age of 6–7 mo showed significantly increased frequencies of CD44^hiCD62L^lo CD8⁺ and CD4⁺ T cells, which is indicative of an activated or memory phenotype (Fig. 2 A and C). On in vitro restimulation, the double KO T cells produced significantly higher levels of cytokines, such as IFNγ, TNFα, and IL-17A, than WT and single KO cells (Fig. 2 B, D–F).

PD-1 binds to ligands PD-L1 and PD-L2 (20). To corroborate the results seen in the VISTA/PD-1 double KO mice, we bred VISTA KO onto the previously described PD-L1 KO (15) and generated the VISTA/PD-L1 double KO mice. Our data demonstrated spontaneous activation of peripheral CD4⁺ and CD8⁺ T cells in the VISTA/PD-L1 double KO mice, which was comparable to that of VISTA/PD-1 double KO mice (Fig. S2 A and B). Together, these data support the conclusion that VISTA and the PD-1/PD-L1 pathway nonredundantly control the peripheral tolerance of T cells.

The phenotype of spontaneous T-cell activation in the double KO mice (Fig. 2 and Fig. S2) indicates that both VISTA and PD-1 regulate the threshold of TCR activation to autoantigens. We hypothesize that disruption of both pathways will increase predisposition to autoimmune disease on a susceptible background. To test this hypothesis, the VISTA/PD-1 double KO mice were bred with the 2D2 mice, which express a TCR transgene specific for the self-antigen, myelin oligodendrocyte glycoprotein (MOG_35–55) (21). Previous studies reported that 4% of 2D2 mice developed spontaneous EAE between the ages of 3 and 5 mo (21). A similar incidence of spontaneous EAE (1/30, ∼3%) was observed in our colony of 2D2-WT mice younger than 6 mo (Fig. 3A and Table S1). 2D2-PD-1 KO mice showed similar incidence of spontaneous disease as WT mice (2/40, 5%). In contrast, genetic deficiency of VISTA accelerated disease onset, such that ∼60% (25/42) of the 2D2-VISTA KO mice rapidly succumbed to complete hind limb paralysis within 3 mo. Combined deficiency of VISTA and PD-1 further increased disease incidence to ∼90% (35/37) (Fig. 3 A and B). Analysis of CNS tissue from paralyzed 2D2-VISTA/PD-1 double KO mice confirmed the accumulation of inflammatory cells and demyelination (Fig. 3 C and D). The disease onset age ranged between 4 and 16 wk in the 2D2-VISTA/PD-1 double KO mice, which was similar to the ages observed in the 2D2-VISTA KO mice (5–16 wk) (Fig. S3). Only one WT mouse developed disease around 12 wk of age. A small percentage of diseased mice from the 2D2-VISTA KO (2/40) and 2D2-VISTA/PD-1 double KO (3/37) mice developed atypical EAE, manifested as unilateral paralysis rather than bilateral paralysis. Detailed information regarding disease onset, severity, penetrance, and mortality is presented in Table S1.

Spontaneous T-cell activation and enhanced autoimmunity in aged VISTA/PD-1 double KO mice indicate that VISTA and PD-1 control T-cell tolerance toward autoantigens. We hypothesize that these pathways also critically regulate T-cell responses toward foreign antigens. To address this question, mice were immunized with soluble antigenic peptides together with poly (I:C) (TLR3 agonist) as adjuvant. 2W1S, an MHC class II-restricted peptide, and OVA^257-264_, an MHC class I-restricted peptide, were used (22). On day +7 postimmunization, splenic T cells were isolated and restimulated ex vivo with the respective peptides. Significantly higher numbers of IFNγ-producing T cells were present in the VISTA/PD-1 double KO mice compared with in WT or single KO mice, indicating that VISTA and PD-1 nonredundantly control T-cell responses (Fig. 4 A and B).

We have previously reported that VISTA functions as a ligand that engages an unknown receptor on T cells and suppresses T-cell activation (3). Because both VISTA and PD-L1 are highly expressed on CD11b⁺ myeloid APCs (2, 3), we hypothesize that a combined deficiency of VISTA and PD-L1 on APCs maximally enhances T-cell activation. To test this hypothesis, CD11b⁺ myeloid DCs were isolated from WT, VISTA KO, PD-L1 KO, and the VISTA/PD-L1 double KO mice (15, 23) and were used to stimulate naive CD4⁺ OTII TCR transgenic T cells in the presence of cognate peptide OVA_323–339. Our data show that the combined deficiency of VISTA and PD-L1 on myeloid APCs synergistically enhanced T-cell proliferation and IFNγ production (Fig. 4 C and D).

Although the receptor for VISTA (VISTA-R) is unknown, we speculate that the engagement of VISTA-R on T cells suppresses TCR signaling independent of PD-1. In addition, we hypothesize that the coengagement of VISTA-R and PD-1 on T cells synergistically impairs TCR signaling. To test these hypotheses, proximal TCR signaling events were examined using immobilized fusion proteins VISTA-Ig and PD-L1-Ig, both of which suppressed T-cell proliferation and cytokine production in vitro (3, 12). LAT is a proximal signaling adaptor that is phosphorylated on TCR stimulation and forms a complex with multiple signaling molecules including SH2 domain containing leukocyte protein of 76kDa (SLP76) and phospholipase C (PLC)-γ1 (24). To determine whether VISTA functions by interfering with the phosphorylation of LAT, a solid phase immunoprecipitation assay was performed, in which the proximal signaling complexes could be recovered as the bound fraction to the plastic surface (25, 26). Our data show that in the presence of coimmobilized control-Ig or VISTA-Ig proteins, plate-bound anti-CD3ε mAb pulled-down comparable amounts of CD3ζ. This result excluded the possibility that the immobilized VISTA-Ig displaced anti-CD3ε mAb or impaired the binding of anti-CD3ε mAb to the TCR/CD3 complex (Fig. 5A). Despite the similar engagement of CD3ζ, immobilized VISTA-Ig significantly reduced the amount of LAT recruited to the CD3 complex, as well as its phosphorylation (Fig. 5A). When total cell lysates were examined, the phosphorylation of several proximal and downstream signaling molecules such as SLP76, PLC-γ1, Akt, and Erk1/2 were also impaired (Fig. 5B).

Next, the effect of coengaging VISTA-Ig and PD-L1-Ig was examined (Fig. 5 C and D). Preactivated T cells were analyzed, as they expressed a high level of PD-1 (27). Consistent with our hypothesis, the combined engagement of VISTA-Ig and PD-L1-Ig maximally reduced the phosphorylation of LAT and its recruitment to the CD3 complex (Fig. 5C). Furthermore, VISTA-Ig and PD-L1-Ig coengagement maximally reduced the phosphorylation of SLP76, PLC-γ1, Akt, and Erk1/2 in total cell lysates (Fig. 5D). Together, these results support the conclusion that VISTA-R and PD-1 each impairs early TCR signaling and results in the most robust suppression when combined.

Enhanced T-cell responses in the VISTA/PD-1 double KO mice suggest that these two immune checkpoints could be targeted concurrently to optimize antitumor immunity. This hypothesis was tested in transplantable murine tumor models (Fig. 6). Our data show that in the CT26 colon cancer model, the combinatorial treatment of VISTA and PD-L1 mAbs on day +2 after tumor inoculation led to tumor regression and long-term survival (8/8), whereas either mAb alone was less effective (1/8 in the VISTA mAb-treated group and 3/8 in the PD-L1 mAb-treated group rejected tumor) (Fig. 6A). Analysis of tumor-specific T-cell activation showed synergistically enhanced cytokine production (IFNγ and TNFα) and granzyme B production by tumor-specific CD8⁺ T cells from tumor-draining LN (Fig. 6B). Next, combinatorial treatment was tested on larger established tumors. When treatment was initiated on day +5 after CT26 inoculation, the average tumor size reached ∼4–5 mm in diameter (Fig. 6C). To facilitate tumor-specific T-cell activation in this therapeutic setting, mice were treated with anti-CD25 mAb on day +5 and day + 20 to transiently deplete CD25⁺Foxp3⁺CD4⁺ regulatory T cells. Synergistic efficacy was achieved with the combined treatment of VISTA and PD-L1 mAbs, which led to ∼80% tumor regression (8/10), whereas single mAb treatment failed to show any significant effect (Fig. 6C).

We next evaluated the therapeutic efficacy of the combinatorial treatment in a less immunogenic tumor model, such as the B16BL6 melanoma model, which is responsive to therapeutic blockade of CTLA-4 and PD-1 in previous studies (28–31). A GM-CSF secreting cellular vaccine (GVAX) (28) was applied on day +3, +6, +9, and +12 after tumor inoculation to boost tumor-specific T-cell responses. A sublethal whole-body irradiation (250 rads) was applied on day +3, which was shown to facilitate T-cell-mediated antitumor immune responses (32). Consistent with our hypothesis, treatment with both VISTA and PD-L1 mAbs significantly suppressed tumor growth and conferred survival advantage, whereas single mAb treatment was largely ineffective (Fig. 6D). Together, these data indicate that VISTA and PD-L1/PD-1 pathways independently control tumor-specific T-cell responses, and combined therapeutic blockade synergistically enhances antitumor immunity.

C57BL/6 mice were purchased from Charles River Laboratories. VISTA KO mice on a fully backcrossed C57BL/6 background were obtained from the Mutant Mouse Regional Resource Centers (University of California–Davis) (36). 2D2 TCR transgenic mice were purchased from the Jackson Laboratory. PD-1 KO mice were provided by Dr. Honjo (10). PD-L1 KO mice were as described (15). All animals were maintained in a pathogen-free facility at the Medical College of Wisconsin. All animal protocols were approved by the Institutional Animal Care and Use Committee of the Medical College of Wisconsin.

Age- and sex-matched WT and VISTA KO mice were killed by CO₂ asphyxiation. Organs were harvested and fixed in 10% (vol/vol) buffered formalin. H&E stain was performed on tissue sections. Tissue inflammatory status was scored in a blind manner by a pathologist using the following semiquantitative scoring criteria: 0 = normal; 1 = mild/small foci of dense lymphocytic infiltrate; 2 = moderate/multiple foci of dense/large activated lymphocytic infiltrate with/without germinal center; and 3 = marked reactive/activated or atypical lymphocytic infiltrate.

Flow cytometry analysis was performed either on FACScalibur or LSRII, using CellQuest software (BD Bioscience). Data analyses were performed using FlowJo software (Treestar).

All graphs and statistical analysis were generated using Prism 4 (GraphPad Software, Inc.). Student’s t test (two tailed) or two-way ANOVA were used for data analyses. ***P < 0.005, **P < 0.025, *P < 0.05.

Additional materials and methods are provided in the SI Materials and Methods.