T lymphocyte–specific deletion of SHP1 and SHP2 promotes activation-induced cell death of CD4⁺ T cells and impairs antitumor response

Significance

SHP1 and SHP2 are related protein-tyrosine phosphatases (PTPs) that associate with several of the same immunoreceptor tyrosine-based inhibitory/switch motif (ITIM/ITSM)-containing receptors or T cell receptor (TCR) signaling molecules. The individual roles of SHP1 and SHP2 in T cells have been reported previously, but potentially redundant functions are less well understood. Here, we uncover an essential function in CD4⁺ T cells that is manifest only in the absence of both enzymes and is critical for the control of tumors.

Abstract

SHP1 (PTPN6) and SHP2 (PTPN11) are closely related protein-tyrosine phosphatases (PTPs), which are autoinhibited until their SH2 domains bind paired tyrosine-phosphorylated immunoreceptor tyrosine-based inhibitory/switch motifs (ITIMs/ITSMs). These PTPs bind overlapping sets of ITIM/ITSM-bearing proteins, suggesting that they might have some redundant functions. By studying T cell–specific single and double knockout mice, we found that SHP1 and SHP2 redundantly restrain naïve T cell differentiation to effector and central memory phenotypes, with SHP1 playing the dominant role. Surprisingly, loss of SHP2 alone in T cells enhanced the antitumor effects of anti-PD-1 antibodies, whereas there was no effect of SHP1 deletion. Also unexpectedly, the absence of both PTPs resulted in poorer tumor control and failure to respond to Programmed Cell Death Protein 1 (PD-1) blockade, associated with reduced frequency and activation of T cells and dendritic cells. Mechanistic studies revealed that CD4⁺, but not CD8⁺, T cells lacking SHP1 and SHP2 show increased activation-induced cell death upon anti-CD3/CD28 stimulation. Adoptive transfer of antigen-specific CD4⁺ T cells restored normal levels of tumor control in mice lacking both PTPs. Together, our results demonstrate that SHP1 or SHP2 is required to prevent activation-induced cell death of CD4⁺ T cells and is critical for tumor immunity, raising the possibility that inhibition of SHP2 might augment the therapeutic efficacy of PD-1-based immune therapy.

The protein-tyrosine phosphatases SHP1 (encoded by PTPN6) and SHP2 (encoded by PTPN11) have high structural similarity (~53% identity), comprising two SH2 domains at the N terminus, a central protein-tyrosine phosphatase (PTP) domain, and a C-terminal tail (1). SHP1 has an additional sequence in its C-terminus that promotes recruitment to lipid rafts and is important for its inhibitory activity in T cells (2). Under basal conditions, SHP1 and SHP2 are autoinhibited (closed conformation) owing to binding of the non-tyrosine-binding portion of the N-SH2 domain to the PTP domain (3, 4). Both PTPs “open” and become active when their SH2 domains bind to tandem immunoreceptor tyrosine-based inhibitory motifs (ITIMs; consensus sequence S/I/V/LxYxxI/V/L) or immunoreceptor tyrosine-based switch motifs (ITSMs; consensus sequence TxYxxV/I) (1, 5). SHP1 and SHP2 bind a variety of proteins, including receptor tyrosine kinases, cytokine receptors, scaffolding adaptors, and inhibitory receptors. Some of these proteins bind only one of the SHPs (SHP1 or SHP2), while others bind both, raising the possibility that the SHPs might have some redundant functions.

Several early studies used motheaten or motheaten viable mice to characterize the role of SHP1 in T lymphocytes. However, due to the effects of SHP1 deficiency on other lymphohematopoietic cells, it is unclear which of these effects seen in these mice are T cell autonomous. Our group and others have used conditional deletion to study SHP1 in thymocytes and peripheral T cells. SHP1 inhibits T cell receptor (TCR) signaling during thymocyte development; consequently, Ptpn6 deletion leads to fewer thymocytes surviving negative selection (6). SHP1 also inhibits TCR signaling in peripheral T cells, lowering the threshold for proliferation and limiting total expansion (7–9). Owing to its role as a negative regulator of IL-4 signaling, SHP1 restricts differentiation of naïve T cells to effector and central memory phenotypes and limits Th2 polarization of CD4⁺ T cells (10).

SHP2 is expressed ubiquitously, and global deletion in mice is embryonic lethal (11, 12). SHP2 has a well-documented positive role in RAS-RAF-MEK-ERK signaling, acting upstream of the guanine nucleotide exchange factors SOS1/SOS2 (13). Two previous reports indicate that SHP2 has no impact on thymocyte development (14, 15). However, SHP2 reportedly limits expansion of and cytokine secretion by mature antigen-specific CD8⁺ T cells in vivo (14). SHP2 is also reported to associate with the CD3ζ chains and the adaptor, SLP76, upon stimulation of Jurkat T cells (16, 17), although it appears to be dispensable for T cell intracellular Ca²⁺ release and CD3ζ chain phosphorylation (18). In primary human T cells, SHP2 inhibition modestly decreases IL-2 production in response to CD3/CD28-mediated stimulation, consistent with a positive role for the phosphatase in the RAS-RAF-MEK-ERK pathway (19, 20). SHP2 also binds directly to some cytokine receptors and to the scaffolding adaptors GAB1/GAB2 and therefore could impact multiple pathways relevant in T cells (13, 21).

PD-1 is a critical inhibitory receptor that limits antitumor immune responses. Consequently, anti-PD-1 antibodies have had a major therapeutic impact on multiple malignancies (22). Early studies suggested a role for SHP2 in PD-1 signaling (23, 24). In primary human CD4⁺ T cells, PD-1 binds SHP1 and SHP2 (24), although quantitative mass spectrometry of PD-1 immunoprecipitates from primary murine T cells revealed preferential binding to the latter SHP (25). PD-1-bound SHP2 dephosphorylates CD28 and, to a lesser extent, LCK in vitro (26). However, functional studies interrogating the role of SHP2 in PD-1 signaling are, at best, conflicting. For example, cocultures of Raji cells overexpressing PD-L1 (a PD-1 ligand) and PD-1-overexpressing Jurkat cells have yielded mixed results. One study identified both SHP1 and SHP2 as mediators of PD-1 action, while another concluded that SHP2 is the primary mediator, with a minor contribution from an unidentified signaling component (25, 27). A prior report found that mice with a T cell-specific deletion of Ptpn11 responded to systemic PD-1 blockade in an MC38 tumor model, apparently suggesting that SHP2 was not required for PD-1 signaling in T cells (14). However, a more recent report suggests that in mice, PD-1 action in the myeloid cell compartment is important for its anti-tumor effect against MC38 tumors (28, 29). While all published in vitro studies suggest that SHP2 mediates PD-1 signaling, it remains unclear whether SHP1 does so as well (25–27). Besides PD-1, multiple inhibitory receptors expressed on T cells bind SHP1 and SHP2 to varying degrees, including 2B4, BTLA, CEACAM1, LAIR1, and, in humans, LILRB1 (25, 30–33). Consequently, the SHPs might redundantly restrict T cell activation downstream of these molecules. Here, we investigate the effects of combined Ptpn6/Ptpn11 deletion on T cell development and the anti-tumor response in the presence and absence of PD-1 blockade.

Results

Effect of SHP1 and SHP2 on Thymocyte Development.

To dissect the roles of SHP1 and SHP2 in T cell biology, we generated mice that were hemizygous for Cd4-Cre and homozygous for floxed Ptpn6 (SHP1 knockout/KO) and/or Ptpn11 (SHP2 knockout/KO) alleles. We validated that splenic T cells from Cd4-Cre Ptpn6^fl/fl Ptpn11^fl/fl (double knockout/DKO) mice lacked expression of both SHPs (SI Appendix, Fig. S1A). CD4-Cre expression commences at the double positive (DP) stage of thymocyte development (34), so we assessed the effects of SHP1 and/or SHP2 deficiency in DP cells and their progeny, single positive (SP) cells. DKO mice had fewer total thymic cells compared with their CD4-Cre Ptpn6^+/+ Ptpn11^+/+ (control) counterparts, mainly due to SHP1 deficiency (SI Appendix, Fig. S1B). Positive and negative selection occur during the DP stage, and TCR signal strength is a critical determinant of survival during selection. Signal strength that is too weak or too strong results in thymocyte death (35). Notably, SHP1 KO mice had significantly fewer DP thymocytes than controls (SI Appendix, Fig. S1C), possibly resulting from increased DP thymocyte deletion and consistent with a report that SHP1 deficiency results in fewer DP thymocytes surviving negative selection (6). Deficiency of SHP2 alone had no effect on the total number of DP thymocytes, but deficiency of both SHPs resulted in a further decrease compared with SHP1 deficiency.

CD5 and CD69 expression on DP thymocytes correlate with TCR signal strength (36, 37). Consistent with previous reports of increased TCR signal strength in SHP1-deficient thymocytes, our SHP1 KO mice showed increased percentages of CD5^hi and CD69⁺ DP thymocytes compared with control mice (SI Appendix, Fig. S1 D and E) (6). DKO mice also had an increased CD5^hi fraction, although they did not have an increased CD69⁺ fraction of DP thymocytes compared with control mice. CD4⁺ and CD8⁺ SP thymocyte numbers were similarly decreased in both SHP1 KO and DKO mice compared with controls (SI Appendix, Fig. S1 F and G), consistent with fewer thymocytes surviving selection. These data suggest that SHP1 and SHP2 play a modest role in thymic T cell development linked to TCR signal strength, with SHP1 having the more prominent contribution.

SHP1 and SHP2 Redundantly Regulate T Cell Phenotypic Differentiation.

Spleens of control, SHP1 KO, SHP2 KO, and DKO mice had similar cell counts and fractions of leukocytes among live cells (Fig. 1 A and B). Frequencies of splenic CD4⁺ T cells in SHP1 KO, SHP2 KO, and DKO mice did not significantly differ from that of controls (Fig. 1C). However, SHP1 KO and DKO mice exhibited reduced numbers of CD4⁺ T cells in inguinal lymph nodes (LNs) compared with control and SHP2 KO mice, and while the frequency in DKO mouse LNs trended lower than in SHP1 KO mouse LNs, the difference was not statistically significant. We showed previously that SHP1 restrains differentiation of naïve T cells to effector/central memory cells via negative regulation of the IL-4 receptor (10). As expected, SHP1 KO mice had fewer naïve (CD62L^hi CD44^lo) CD4⁺ T cells in the spleen with a concomitant increase in effector (CD62L^lo CD44^hi) CD4⁺ T cells (Fig. 1 D–F). LN populations were not significantly affected. SHP2 KO alone did not affect naïve, effector, and central memory (CD62L^hi CD44^hi) CD4⁺ populations in either spleen or LNs. However, combined deficiency of both SHPs accentuated the effects of SHP1 deficiency: DKO mice had significantly fewer naïve CD4⁺ T cells than SHP1 KO mice and significantly more effector CD4⁺ T cells in LNs and more central memory CD4⁺ T cells in both spleen and LNs. DKO mice did not have a significantly greater percentage of effector CD4⁺ T cells in the spleen than SHP1 KO mice.

Fig. 1.

SHP1 and SHP2 redundantly restrict naïve T cell differentiation. (A) Total splenocytes. (B) %leukocytes in the spleen. (C–L) Splenic or LN T cell populations from the indicated mice, (C–F) CD4⁺ cells, (G–L) CD8⁺ cells: naïve (CD44^loCD62L^hi), effector (CD44^hiCD62L^lo), central memory (CM; CD44^hiCD62L^hi), true memory (TM; CD44^hiCD49d^hi), virtual memory (VM; CD44^hiCD49d^lo). Data are combined from at least two experiments. Each point represents a value from one mouse. Means ± SE are indicated. Significance was evaluated by one-way ANOVA with multiple comparisons, not significant (ns), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

SHP1 deficiency led to decreased frequencies of CD8⁺ T cells in spleen and LNs, and while dual SHP deficiency did not enhance this decrease in the spleen, it did in LNs (Fig. 1G). SHP1 KO mice had significantly fewer naïve CD8⁺ T cells relative to control mice with increases in both the effector and central memory populations in spleen and/or LNs (Fig. 1 H–J). Again, SHP2 deficiency had no independent effect on differentiation, but combined SHP deficiency accentuated the effects of SHP1 deficiency with DKO mice having even more effector and central memory CD8⁺ T cells. Collectively, these results indicate that SHP1 and SHP2 redundantly restrict T cell differentiation with SHP1 playing the dominant role. The reason why SHP1 KO and DKO mice had fewer T cells in spleen and LNs is unclear but could relate to decreased output from the thymus.

Within the memory compartment, CD44^hi CD49d^hi CD8⁺ T cells are reported to be antigen-experienced (true memory; TM), whereas CD44^hi CD49d^lo CD8⁺ T cells are antigen-inexperienced (virtual memory; VM) (38). SHP1 KO and DKO mice had small increases in the splenic CD8⁺ TM compartment, but populations of these cells in all genotypes, including controls, were low (Fig. 1K). Nevertheless, SHP1 deficiency significantly elevated the percentage of CD8⁺ VM T cells, which increased even further in mice with combined SHP deficiency in both spleen and LNs (Fig. 1L). Together, these data suggest that, at least in CD8⁺ T cells, SHP1/SHP2 may also redundantly control the percentage of cells with a memory phenotype that is independent of antigen exposure. Results from Fig. 1 are summarized in SI Appendix, Table S1.

Either SHP1 or SHP2 Is Required for the Antitumor T Cell Response.

PD-1, along with other T cell inhibitory receptors, binds both SHP1 and SHP2. Therefore, we hypothesized that DKO mice would exhibit an enhanced antitumor T cell response against PD-1-responsive tumors compared with control, SHP1 KO, or SHP2 KO mice. We expected that deletion of essential PD-1 signaling mediators would phenocopy PD-1 blockade or deletion, resulting in smaller tumors, regardless of PD-1 antibody treatment. To this end, we injected control, SHP1 KO, SHP2 KO, and DKO mice subcutaneously with MC38 cancer cells and treated them with anti-PD-1 or isotype control antibody. Surprisingly, isotype-treated DKO mouse tumors grew significantly larger and faster than those from control, SHP1 KO, and SHP2 KO mice, which all had similar tumor growth (Fig. 2A). As expected, tumors grew more slowly in control mice treated with anti-PD-1 than in those treated with isotype antibody (Fig. 2B and SI Appendix, Fig. S3A). Moreover, these mice were more likely to be responders (6/23 responders; see Materials and Methods) than isotype-treated mice (0/20 responders; SI Appendix, Fig. S3B). We also found that SHP1 KO mice treated with anti-PD-1 had reduced tumor growth and were more likely to be responders (3/9 responders) compared with isotype-treated mice (0/11 responders; Fig. 2C and SI Appendix, Fig. S3 C and D). In contrast to a previous report, we found that SHP1 KO mice were equally (i.e., not less) likely to respond to treatment than control mice (Fig. 2D). SHP2 KO mice treated with anti-PD-1 also had reduced tumor volume over time and were more likely to be responders (16/29 responders) than those treated with isotype antibody (2/23 responders; Fig. 2E and SI Appendix, Fig. S3 E and F). To our surprise, PD-1 blockade was actually more effective in SHP2 KO mice than in control mice (Fig. 2F). Although only a minority of control mice responded to anti-PD-1 treatment (26%), the majority of SHP2 KO mice were responders (55%). These data indicate that SHP2 restricts the T cell–mediated antitumor response in a T cell–autonomous fashion via a pathway independent of PD-1.

Fig. 2.

Either SHP1 or SHP2 is required for antitumor T cell responses. (A–H) MC38 tumor growth in mice with isotype or anti-PD-1 antibody (αPD-1) treatment. (A) Tumor growth in only isotype-treated mice analyzed by two-way ANOVA, control (n = 12), SHP1 KO (n = 14), SHP2 KO (n = 15), DKO (n = 13). (B, C, E, and G) Tumor growth, analyzed by two-way ANOVA, (B) control + isotype (n = 24, 4 euthanized) or + αPD-1 (n = 25, 2 euthanized), (C) SHP1 KO + isotype (n = 11, 0 euthanized) or + αPD-1 (n = 10, 1 euthanized), (E) SHP2 KO + isotype (n = 26, 3 euthanized) or + αPD-1 (n = 29, 0 euthanized), (G) DKO + isotype (n = 13, 2 euthanized) or + αPD-1 (n = 13, 1 euthanized). (D, F, and H) Histograms of responders and nonresponders in anti-PD-1-treated mice comparing (D) control and SHP1 KO, (F) control and SHP2 KO, (H) control and DKO, analyzed by the chi-squared test. All comparisons were combined from 2 to 5 experiments. Mice humanely euthanized before day 23 were necessarily excluded from chi-squared comparisons. Means ± SE are indicated. Not significant (ns), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Remarkably, and in contrast to our expectation, isotype- and anti-PD-1-treated DKO mice had similar tumor growth rates, and no DKO mice qualified as anti-PD-1 responders (Fig. 2 G and H and SI Appendix, Fig. S3 G and H). While our experiments were in progress, Ventura et al. also reported that DKO mouse tumors were larger than control tumors and did not respond to PD-1 blockade (39).

Our findings collectively suggest DKO T cells are dysfunctional, leading to an impaired antitumor T cell response. Thus, either SHP1 or SHP2 is required for T cell–mediated control of tumor growth and response to PD-1 checkpoint blockade. However, because of the incompetent antitumor response in DKO mice, we cannot determine from the data whether both SHP1 and SHP2 can mediate PD-1 signaling in T cells.

Combined SHP Deficiency in T Cells Leads to a Dysregulated Tumor Immune Microenvironment.

We next examined the immune microenvironment in tumors from DKO and control mice at 24 d after injection of MC38 cells. Consistent with the growth measurements in Fig. 2, tumors from DKO mice had greater mass than those from control mice (Fig. 3A). Single-cell suspensions from tumors were analyzed by flow cytometry. We found that the proportions of total leukocytes in tumors were similar in both genotypes (Fig. 3A). However, the frequency of CD4⁺ T cells in DKO mouse tumors was nearly half that of controls (Fig. 3B). CD4⁺ T cells in tumors from DKO mice and controls showed similar levels of phenotypic exhaustion, as measured by PD-1/TIM3 double positivity (Fig. 3B) or by expression of LAG3 or CTLA4 (SI Appendix, Fig. S4 A and B). Consistent with increased IL-4 signaling, we observed polarization of DKO CD4⁺ T cells away from Th1 and toward Th2 (SI Appendix, Fig. S4 C–E).

Fig. 3.

DKO mouse tumors have a dysregulated immune microenvironment. (A–E) Flow cytometric analysis of cells from isotype-treated control and DKO mouse tumors. (A) Tumor mass, %leukocytes. (B) CD4⁺ T cells, PD1/TIM3 DP CD4⁺ T cells, (C) CD8⁺ T cells, CD8⁺ T cell to T_reg ratio, CD69⁺ CD8⁺ T cells, PD1⁺ CD8⁺ T cells, PD1/TIM3 DP CD8⁺ T cells. (D) Cell suspensions were treated with PMA + ionomycin + Brefeldin A. IFNγ⁺ CD8⁺ T cells, TNFα⁺ CD8⁺ T cells, IL-2 + CD8⁺ T cells. (E) Dendritic cells (DCs), MFI of MHC class II on DCs, tumor-associated macrophages (TAMs). Data were combined from two experiments. Each point represents a value from one mouse. Means ± SE are indicated. Significance was evaluated by two-tailed unpaired t tests, not significant (ns), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

CD8⁺ T cell numbers were also decreased in DKO mouse tumors (Fig. 3C). The proportion of T_regs in the CD4⁺ T population was similar in control and DKO mouse tumors (SI Appendix, Fig. S4F), while the ratio of CD8⁺ T cells to T_regs was actually greater in the latter (Fig. 3C). These results indicate that T_regs are unlikely to account for the impaired antitumor response in DKO mice. The percentages of intratumoral CD8⁺ T cells expressing CD69 or PD-1 were also significantly decreased (Fig. 3C), indicating decreased CD8⁺ T cell activation. However, the percentage of phenotypically exhausted (PD-1/TIM3 DP) CD8⁺ T cells was lower in the tumors from DKO mice (Fig. 3C), arguing against exhaustion as the explanation for their dysfunctional anti-tumor response. CD8⁺ T cells taken from DKO mouse tumors and stimulated ex vivo also had significantly lower production of IFNγ, TNFα, and IL-2 compared with control CD8⁺ T cells (Fig. 3D), highlighting their decreased effector capacity.

We also found that tumors from DKO mice had significantly fewer dendritic cells (DCs) than those from control mice (Fig. 3E). Intratumoral DCs from DKO mice had significantly lower MHC class II expression than those from control mice, indicating deficient activation (Fig. 3E). While DKO mouse tumors had fewer T cells than control mouse tumors, they also had significantly increased proportions of tumor-associated macrophages (TAMs; Fig. 3E), which had decreased proportions of PD-L1 expression (SI Appendix, Fig. S4H). Populations of granulocytic myeloid-derived suppressor cells and monocytic myeloid-derived suppressor cells did not differ between genotypes (SI Appendix, Fig. S4 I and J), and while DKO mouse tumors had a statistically significant decrease in NK cells relative to control mouse tumors, the frequencies in both groups and the difference between them were negligible (SI Appendix, Fig. S4K). These data highlight major differences in the effects of combined SHP1/SHP2 deficiency on CD4⁺ versus CD8⁺ T cells, and the consequences of SHP1/SHP2 loss in T cells on tumor-infiltrating DCs and TAMs (Discussion).

DKO T Cells Maintain Cytotoxic Capacity.

A possible explanation for the impaired DKO antitumor T cell response is that SHP1 or SHP2 is required for T cell cytotoxicity. To generate target cells for testing cytotoxic capacity, we transduced MC38 cancer cells with a construct encoding mouse anti-CD3 Fab fused to the GPI-anchored protein human CD14 (SI Appendix, Fig. S6A). This approach yielded MC38 cells with an outwardly facing, membrane-bound anti-CD3 (MC38-αCD3) to activate murine T cells ex vivo in an antigen-independent manner. To test whether DKO T cells could kill target cells, we isolated splenic T cells from control or DKO mice and stimulated them with plate-bound anti-CD3 and anti-CD28 for 48 h in the presence of 20 ng/mL recombinant IL-2 (SI Appendix, Fig. S6B). We cultured these cells for 7 d in IL-2 (in the absence of TCR stimulation), then added them to MC38 or MC38-αCD3 cells and monitored killing. Control and DKO T cells both killed MC38-αCD3 at multiple effector-to-target (E:T) ratios with more efficient killing of MC38-αCD3 cells than MC38 cells (SI Appendix, Fig. S6 C and D). Hence, neither SHP1 nor SHP2 is essential for T cell cytotoxicity.

SHP1 or SHP2 Is Required to Prevent Activation-Induced Cell Death in CD4⁺ T Cells.

Another potential explanation for the dysfunctional antitumor T cell response in DKO mice is defective T cell expansion upon TCR stimulation. To explore this possibility, we stimulated control, SHP1 KO, SHP2 KO, and DKO T cells ex vivo with anti-CD3 and anti-CD28 in the presence of IL-2 and measured T cell numbers and fractions of CD4⁺ and CD8⁺ cells over several days. Control CD8⁺ T cells expanded more rapidly than CD4⁺ T cells (Fig. 4 A–D), consistent with existing literature (40). CD8⁺ T cells from all genotypes showed similar net expansion at day 5 poststimulation (Fig. 4 A and B). By contrast, while control, SHP1 KO, and SHP2 KO CD4⁺ T cells expanded similarly over the course of 5 d, DKO CD4⁺ T cell numbers dropped to near zero (Fig. 4 C and D).

Fig. 4.

SHP1 or SHP2 is required to prevent activation induced cell death (AICD) in CD4⁺ T cells. (A–D) T cell proliferation following stimulation with αCD3 and αCD28, (A) Absolute number of CD8⁺ T cells, (B) Fold change in CD8⁺ T cell number from baseline, (C) Absolute number of CD4⁺ T cells, (D) Fold change in CD4⁺ T cells from baseline. (E) CTV dye dilution assays showing CD4⁺ T cell divisions after 4 d after stimulation, (F and G) Annexin V staining at 3 h poststimulation of (F) CD4⁺ T cells and (G) CD8⁺ T cells. (H) Percentage of CD4⁺ T cells at day 4 poststimulation with/without FASL blockade. Data are representative of (E and H), or combined from (A–D, F, and G), at least two experiments. Each point represents a value from one mouse. Means ± SE are indicated. Significance was evaluated by one-way ANOVA with multiple comparisons, not significant (ns), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Defective CD4⁺ T cell expansion could result from impaired cell proliferation and/or increased cell death. To assess the proliferation of DKO CD4⁺ T cells, we stimulated CTV-stained T cells and measured dye dilution. At day 4 poststimulation, similar fractions of control and DKO CD4⁺ cells had undergone between 1 and 7+ divisions (Fig. 4E), indicating that combined SHP1/SHP2 deficiency did not interfere with proliferation. By contrast, at 3 h post-TCR stimulation, DKO CD4⁺ T cells exhibited threefold greater Annexin V positivity than controls (27.5% and 8.9%, respectively; Fig. 4F). No other genotype showed significant elevations in Annexin V binding on CD4⁺ T cells at this time point. The percentage of apoptotic CD8⁺ T cells was lower than in CD4⁺ T cells and was similar in all genotypes (Fig. 4G).

Excessive or chronic TCR stimulation can result in activation-induced cell death, an apoptotic process that contributes to peripheral tolerance (41, 42). SHP1 KO T cells treated with SHP099 (SHP2 inhibitor) at the time of stimulation did not have increased Annexin V staining at 3 h poststimulation (SI Appendix, Fig. S7 A and B), suggesting that the increased sensitivity of DKO CD4⁺ T cells to apoptosis results from chronic, as opposed to acute, signaling changes. Activation-induced cell death is mediated by FAS ligand (FASL) binding to FAS on T cells (43, 44). To test whether DKO CD4⁺ T cells were dying by FASL/FAS-mediated activation-induced cell death, we stimulated T cells with or without FASL-blocking antibody. Neither CD44^lo nor CD44^hi CD4⁺ T cells from DKO mice had significantly different levels of CD3, CD28, or FAS expression compared with controls at baseline (SI Appendix, Fig. S7 C–H). Four days after stimulation, FASL blockade increased control CD4⁺ T cell percentages from 24.6 to 33.8% (1.4-fold), reflecting a rescue of normal activation-induced cell death in these cells (Fig. 4H). By contrast, FASL blockade increased DKO CD4⁺ T cell percentages from 5.6 to 21.9% (3.9-fold increase), accounting for 86% of the excess DKO CD4⁺ T cell death. Collectively, these results indicate that lack of both SHPs impairs CD4⁺ T cell expansion mainly by increasing FAS-mediated activation-induced cell death.

Adoptive Transfer of Antigen-Specific CD4⁺ T Cells Restores the Antitumor Response in DKO Mice.

To test whether the defective antitumor response in DKO mice resulted from loss of functional CD4⁺ T cells, we generated OVA-expressing MC38 cells (MC38-OVA) and injected mice subcutaneously on their right flanks with MC38-OVA cells and on their left flanks with parental MC38 cells. On the same day, we adoptively transferred control 2D2 CD4⁺ T cells (expressing a transgenic TCR specific for the unrelated MOG peptide) or wild type OT-II CD4⁺ T cells (expressing a transgenic TCR specific for OVA_323-339 peptide) into control and DKO mice (SI Appendix, Fig. S8). As expected, parental MC38 tumors on the left flank grew significantly larger in DKO mice than in control mice, regardless of 2D2 or OT-II CD4⁺ T cell treatment (Fig. 5A). Adoptive transfer of OT-II CD4⁺ T cells did not enhance the antitumor response against MC38-OVA tumors on the right flank in control mice (Fig. 5B), arguing that tumor-infiltrating CD4+ cells are not limiting in these animals. By contrast, transfer of OT-II CD4⁺ T cells, but not 2D2 CD4⁺ T cells, restored the dysfunctional antitumor response to MC38-OVA tumors on the right flank in DKO mice (Fig. 5C). Thus, lack of competent CD4⁺ T cells can explain the dysfunctional antitumor response in DKO mice.

Fig. 5.

CD4⁺ T cell deficiency is responsible for dysfunctional DKO antitumor response. (A–C) MC38 and MC38-OVA tumor growth in mice following adoptive transfer of 2D2 or OT-II CD4⁺ T cells. (A) MC38 tumors in control (black, n = 21) and DKO (purple, n = 32) mice, (B) MC38-OVA tumors in control mice following adoptive transfer of 2D2 (black, n = 10) or OT-II (teal, n = 11) CD4⁺ T cells, (C) MC38-OVA tumors in DKO mice following adoptive transfer of 2D2 (purple; n = 16) or OT-II CD4⁺ T cells (pink; n = 16); control + OT-II treatment group seen in panel B is also shown. All comparisons are combined from two experiments and analyzed by two-way ANOVA. Mean ± SE are indicated. Not significant (ns), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

DKO Mice Also Have Abnormal Responses to Acute Viral Infection.

To determine whether DKO mice had impaired immune responses in a context other than tumors, we infected mice with Vaccinia-OVA (VACV-OVA) through scarification of their ears. Multiple reports indicate that CD4⁺ T cells are important for expansion of VACV-specific CD8⁺ T cells and/or elimination of virus, and previous work found that CD4⁺ T cell depletion impairs viral clearance at day 15 (45–47). We found that DKO mice had significantly increased ear thickness over the duration of the experiment (SI Appendix, Fig. S9A), reflecting impaired resolution of the immune response. At day 15, 3/4 control mice had cleared infection, while the fourth only yielded 30 plaque forming units (PFU; SI Appendix, Fig. S9B). By contrast, 2/4 DKO mice failed to clear the infection with 2,000 and 20,000 PFU. Moreover, total T cells were decreased in the ears of DKO mice relative to control mice, with CD4⁺ T cells being significantly lower and CD8⁺ T cells trending downward (SI Appendix, Fig. S9 C–E). Unfortunately, I-Ab OVA_232-339 tetramers, used to identify OVA-specific CD4+ T cells, showed poor staining. However, there were fewer antigen-specific CD8⁺ T cells in the ear parenchyma of DKO mice compared with control mice (as measured by SIINFEKL H2-Kb tetramers; SI Appendix, Fig. S9F). Collectively, these data are consistent with a defective CD4⁺ T cell response, impairing antigen-specific CD8⁺ T cell proliferation and clearance of virus.

Discussion

SHP1 and SHP2 have substantial structural and sequence similarity and share several binding proteins, suggesting that they could have some redundant functions. In T lymphocytes, however, the effects of deletion of each SHP are distinct: SHP1 KO mice have increased effector/central memory CD4⁺ and CD8⁺ populations and Th2 skewing, whereas SHP2 KO mice reportedly have expanded CD8⁺ T cells. To identify potentially redundant functions, we generated mice with combined SHP1/SHP2 deficiency. Our results reveal that the absence of both SHPs in T cells leads to significantly more peripheral T cell differentiation, enhanced CD4⁺ T cell susceptibility to activation-induced cell death, and an impaired antitumor immune response.

Prior work has assessed the effects of individual deficiency of SHP1 or SHP2 on thymocyte development. Martinez et al. found that TCR signal strength was increased in SHP1 KO (Cd4-Cre Ptpn6^fl/fl) thymocytes, leading to fewer CD4⁺ and CD8⁺ SP thymocytes compared with controls (6). Consistent with these findings, we observed increased markers of TCR signal strength and decreased numbers of DP and SP SHP1 KO thymocytes relative to control mice (SI Appendix, Fig. S1). Notably, in an earlier study of T cell–specific SHP1 KO (Cd4-Cre Ptpn6^fl/fl) mice performed at another institution, we failed to see any difference in the thymic compartment. The reason for this discrepancy is unclear; conceivably, differences in animal facilities play a role. One earlier report argued that SHP2 affects thymocyte development by limiting TCR signaling (48). However, these results might have been confounded by the use of a proximal Lck-Cre to drive Ptpn11 deletion; notably, this driver itself causes a similar thymic phenotype (49). In agreement with work by Rota et al., who also analyzed Cd4-Cre Ptpn11^fl/fl mice (14, 39), we observed no gross differences in thymocyte development in SHP2 KO mice. We also found that compound SHP1/SHP2 deficiency is similar to SHP1 deficiency except for a small decrease in DP thymocyte number in DKO mice. Our results contrast with those of Ventura et al., who recently reported that DKO mice (Cd4-Cre Ptpn6^fl/fl Ptpn11^fl/fl) and Ptpn6^fl/fl Ptpn11^fl/fl controls have similar numbers of DP thymocytes, although they also observed lower numbers of SP CD4⁺ and CD8⁺ thymocytes (39).

We reported previously (10) and confirm herein (Fig. 1) that SHP1 restricts differentiation of peripheral naïve CD4⁺ and CD8⁺ T cells to effector and central memory cells and inhibits polarization of CD4⁺ T cells toward the Th2 phenotype. In our earlier study, we attributed these effects to enhanced IL-4R signaling, as IL-4 deficiency or blockade reversed the phenotype. Notably, the phosphorylated ITIMs of IL-4Rα bind both SHPs, suggesting that SHP2 might be able to at least partially substitute for SHP1 in this pathway (50). Indeed, while we found that SHP2 deficiency alone does not affect naïve T cell differentiation, combined SHP1/SHP2 deficiency drives even more effector/central memory cell differentiation relative to SHP1 deficiency alone (Fig. 1). We also observed that SHP1 KO mice have an increased CD8⁺ VM compartment, which is further enlarged in DKO mice. Consistent with this finding, IL-4R-deficient mice have reduced populations of CD8⁺ VM T cells (51). Notably, transgenic expression of the catalytically inactive mutant SHP2-C459S (purported to act as a dominant negative mutant) specifically in T cells resulted in increased proportions of CD44^hi T cells and enhanced expression of IL-4, IL-5, and IL-10 (52). Conceivably, this putative “dominant negative SHP2” mutant, in addition to competing with endogenous SHP2, also blocks SHP1 binding to IL4Rα, thereby mimicking the DKO phenotype.

We also determined the effects of SHP1, SHP2, and combined SHP1/SHP2 deficiency on the antitumor response using the well-characterized subcutaneous MC38 model (Fig. 2). Because PD-1, among other inhibitory receptors that bind the SHPs, is known to impair the antitumor T cell response, we assessed the effects of PD-1 blockade in our mice. Deletion of an essential mediator of PD-1 signaling should phenocopy PD-1 deletion or blockade, yielding smaller tumors regardless of treatment. Rota et al. reported that SHP2 KO mice respond to PD-1 blockade, indicating that T cell expression of SHP2 was not required for the anti-tumor effects of PD-1 antibodies (14). Notably, Rota et al. did not compare PD-1 treatment efficacy between SHP2 KO mice and CD4-Cre-expressing controls (14). Not only did we confirm that T cell-SHP2 is dispensable for the effects of PD-1 blockade, but we observed that SHP2 deficiency in T cells actually enhances these effects via an as yet unclear mechanism (Fig. 2). Conceivably, SHP2 negatively regulates (an)other pathway(s), whose effects are masked in the presence of intact PD-1 signaling. Indeed, while tumor growth in isotype-treated SHP2 KO mice and control mice was similar, 2/23 SHP2 KO mice qualified as responders despite having received no anti-PD-1 treatment (Fig. 2). This finding suggests SHP2 may indeed mediate signaling by (a) parallel inhibitory pathway(s). SHP2 could be acting downstream of inhibitory receptors 2B4, BTLA, LAIR1, and/or CEACAM1, all of which have been reported to bind SHP2 and are expressed on murine T cells (25, 30–33). SHP2 has been implicated in positive or negative regulation of multiple cytokine signaling cascades, and conceivably, the milieu of altered cytokine signaling could enhance PD-1 blockade responses in SHP2 KO mice. Regardless of the precise underlying mechanism, our results argue that combining SHP2 inhibitors with PD-1 blockade might have therapeutic benefit and raise the question of whether SHP2 deficiency in T cells might sensitize tumors unresponsive to PD-1 blockade.

In agreement with a recent study, we found that SHP1 expression in T cells is also dispensable for the response to anti-PD-1 antibodies (39). However, while Ventura et al. found that PD-1 blockade was less effective at impeding tumor growth in SHP1 KO mice than in controls, we observed no difference in anti-PD-1 response in our SHP1-deficient animals (Fig. 2) This discrepancy likely stems from their use of floxed littermate controls instead of the CD4-Cre-expressing controls used in our study. Cre has deleterious effects in multiple cell types, and we found that CD4-Cre mice have a blunted response to PD-1 blockade (SI Appendix, Fig. S2). Like Ventura et al. we observed that DKO mice fail to respond to PD-1 blockade and have larger tumors than controls, indicating that DKO mice do not phenocopy PD-1 blockade or deficiency. These results also indicated that the antitumor T cell response was dysfunctional, making it impossible to determine whether PD-1 signaling in T cells requires SHP1 and/or SHP2. While our study was in progress, Strauss et al. reported that PD-1 inhibits antitumor immune responses in their MC38 model primarily by acting in myeloid cells, with only a smaller contribution by T cells (29). Thus, it remains to be determined whether either SHP1 or SHP2 is required for PD-1 function in T cells.

Tumors in DKO mice had decreased frequencies of CD4⁺ and CD8⁺ T cells as well as impaired activation of DCs compared with controls (Fig. 3). These findings comport with a model in which loss of CD4⁺ T cells leads to decreased licensing of DCs and results in suboptimal CD8⁺ T cell activation. Notably, tumors from DKO mice contained less than a third of the number of DCs observed in controls (Fig. 3). They also had increased percentages of TAMs, and these TAMs expressed lower levels of PD-L1 than TAMs in control mice. Prior studies have found that GM-CSF produced by T cells promotes upregulation of PD-L1 on TAMs and recruitment of DCs (53–57). CD4⁺ T cells also recruit CD8⁺ T cells (58), which could help explain the lower frequencies of CD8⁺ T cells in tumors from DKO mice.

Combined deficiency of SHP1 and SHP2 caused a failure of CD4⁺ T cell expansion in vitro (Fig. 4), potentially explaining the decrease in CD4⁺ T cells in tumors from DKO mice. Importantly, the expansion failure of CD4⁺ T cells from DKO mice cannot be explained by the increased percentage of effector T cells in these mice: SHP1 KO mice had similar proportions of splenic effector CD4⁺ T cells but did not show increased rates of apoptosis upon stimulation and had no defect in expansion.

Fas^lpr/lpr (Lpr) mice, which express extremely low levels of FAS, have T cells incapable of activation-induced cell death, leading to autoimmunity (43, 44, 59). FASL blockade largely rescued DKO CD4⁺ T cell death in vitro, indicating these cells were undergoing activation-induced cell death (Fig. 4). Notably, SHP1 KO CD4⁺ T cells treated with SHP2 inhibitor just before stimulation did not phenocopy DKO CD4⁺ T cells by showing increased apoptosis. This suggests that DKO CD4⁺ T cells are primed to undergo apoptosis from chronic signaling changes, which likely lead to an as yet undefined change in cell state. High and repeated TCR stimulation in vivo results in deletion of reactive clones (44, 60). Future studies looking at basal TCR signaling and single-cell resolution transcriptional changes could clarify how SHP1/SHP2 dual deficiency increases DKO CD4⁺ T cells’ susceptibility to activation-induced cell death (40, 41, 56). Unlike SHP1, SHP2 has not been reported to regulate TCR signal strength. However, it has been reported to associate with CD3ζ and SLP76 (16, 17). Notably, THEMIS forms a complex with both SHPs, and THEMIS knockdown in primary human peripheral CD4⁺ T cells results in significantly enhanced activation-induced cell death upon TCR stimulation (61, 62). Conceivably, SHP1 and SHP2 both act via THEMIS to inhibit TCR signaling and prevent activation-induced cell death. It will be interesting to determine whether THEMIS knockout mice also have enlarged tumors and impaired anti-PD-1 response compared with controls. Also of note, CD4⁺ and CD8⁺ T cells deficient in ERK2 have impaired survival but no defect in CD3/CD28-mediated proliferation (63, 64). ERK2-deficient CD4⁺ T cells also show reduced Th1 differentiation, and mice have impaired response to LCMV Armstrong infection (16, 17, 64). These data suggest that the increased activation-induced cell death seen in DKO CD4⁺ T cells could, in part, be explained by chronically reduced ERK2 signaling. Why CD8⁺ T cells are not affected is unclear. Most importantly, we found that adoptively transferring normal CD4⁺ T cells to DKO mice restored their antitumor response (Fig. 5), consistent with the idea that CD4⁺ T cell deficiency is a major reason for impaired tumor control.

In contrast to the effects of combined SHP1/SHP2 deficiency on CD4⁺ T cells, CD8⁺ T cells from DKO mice showed no increase in activation-induced cell death. Control CD8⁺ T cells also showed lower levels of activation-induced cell death than control CD4⁺ T cells, consistent with previous results (65). By contrast, Ventura et al. reported decreased CD8⁺ T cell expansion in their DKO mice. Notably, our two studies monitored T cell expansion over different durations, used different Cre driver strains (CD4-Cre versus GZMB-Cre), and employed different controls (Cre-expressing controls here versus floxed controls in their report) (39), any or all of which could explain these discrepancies. Although our data suggest a more dominant role of SHP1/SHP2 double deficiency on CD4⁺ T cells, it is possible that it may also have direct effects on CD8⁺ T cells and play an important function in tumor immunity, especially in different tumor models beyond MC38.

In summary, we conclude that SHP1 and SHP2 serve an indispensable role in CD4⁺ T cell function. T cell deficiency of both PTPs in mice results in a severely impaired T cell–mediated antitumor response, including a failure to respond to PD-1 blockade. Tumors in DKO mice have a dysregulated immune microenvironment, characterized by changes consistent with loss of CD4⁺ T cell help. CD4⁺ T cells, but not CD8⁺ T cells, from DKO mice undergo increased FASL/FAS-dependent apoptosis in response to TCR stimulation, and loss of normal CD4⁺ T cell function can account for the impaired antitumor T cell response in DKO mice. DKO mice similarly show dysregulated immune responses when challenged with VACV-OVA. (SI Appendix, Fig. S9) Collectively, our findings provide insight into how SHP1/SHP2 regulate CD4⁺ T cell survival upon activation and suggest additional therapeutic opportunities for patients with cancer.

Materials and Methods

Cell Lines.

MC38 cells were maintained at 37 °C in 5% CO₂ in Dulbecco's modified eagle medium (DMEM) containing 10% fetal bovine serum (FBS) and penicillin plus streptomycin (Thermo Fisher Scientific; 15140163). To generate MC38-αCD3 cells, we cloned the scFv sequence of murine-reactive αCD3 clone 145-2C11 into the CD5L-OKT3-scFv-CD14 lentiviral plasmid generated by Leitner et al., replacing the OKT3 region (66). Lentiviral particles were generated by transfecting HEK 293 T cells with lentiplasmid, psPAX2, and pMD2.G plasmids. MC38 cells were then transduced with CD5L-145-2C11-scFv-CD14 lentivirus and purified by fluorescence-activated cell sorting (FACS) for human CD14. MC38-OVA cells were generated by transducing MC38 cells with pLVX-puro-cOVA (Addgene plasmid #135073) followed by selection in puromycin (4 µg/mL).

Mice.

All animal studies were approved by the Institutional Animal Care and Use Committee at New York University Grossman School of Medicine. Ptpn6^fl/fl and Ptpn11^fl/fl mice were described previously and were maintained on C57Bl/6 background (67). CD4-Cre-expressing mice were provided by Christopher B. Wilson (34).

Immunoblotting.

Splenic T cells isolated by negative selection were stained with anti-CD45 FITC and anti-CD3 BV421 (BioLegend; 100336) and CD45⁺ CD3⁺ cells were recovered by FACS and centrifugation. Pellets were lysed in 1% NP-40 buffer with 0.1% SDS supplemented with phosphatase inhibitors (10 nM NaF, 1 mM Na₃VO₄, 10 mM bβ-glycerophosphate, 10 mM sodium pyrophosphate) and proteases (40 µg/mL PMSF, 2 µg/mL antipain, 2 µg/mL pepstatin A, 20 µg/mL leupeptin, and 20 µg/mL aprotinin), and clarified by centrifugation at 20,000 rcf for 10 min at 4 °C in a microfuge. Protein was quantified by Coomassie Blue assay (Pierce™ Bradford Protein Assay Kit; 23200). Lysates were boiled in sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer, resolved on 10% SDS-PAGE gels, and transferred onto polyvinylidene fluoride membranes in transfer buffer (glycine 14.4 g/L, Tris 3.03 g/L, 15% methanol; pH 8.6). Membranes were blocked with 2% milk, incubated with primary antibodies [anti-SHP1 (Santa Cruz; sc-287; 1:1,000), anti-PTPN11 (Santa Cruz; sc-7384; 1:1,000), and anti-ERK2 (Santa Cruz; sc-1647; 1:1,000)] at 4 °C overnight in 5% bovine serum albumin in TBST, washed in TBST, and incubated with secondary antibodies [goat anti-rabbit 680LT (LI-COR; 926-68021; 1:20,000) and goat anti-mouse 800CW (LI-COR; 926-32210; 1:20,000)] for 1 h at room temperature. Membranes were washed in tris-buffered saline with tween 20 (TBST) and dried before imaging on the LI-COR Odyssey CLx. Signals were quantified using Image Studio from LI-COR.

Ex Vivo T Cell Stimulations and Killing Assays.

For cell expansion assays, splenic T cells were isolated by negative selection (STEMCELL Technologies; 19851) and added to 96-, 24-, or 6-well plates [precoated overnight at 4 °C with goat anti-hamster IgG (MP Biomedicals; 856984)] in T cell stimulation medium containing anti-CD3 (1 µg/mL; BioLegend; 100238) and anti-CD28 (1 µg/mL; BioLegend; 102116) at 10⁶ cells/mL (1 mL for 24-well plates and 3 mL for 6-well plates) or 2 x 10⁶ cells/mL (200 µL for 96-well plates). For SHP2 inhibition, the indicated samples were mixed with SHP099 at a final concentration of 10 µM as they were added to the plate for stimulation. After incubation for 48 h, cells were resuspended in fresh T cell medium [Roswell Park Memorial Institute with 10% FBS, penicillin plus streptomycin (Thermo Fisher Scientific; 15140163; 100 U/mL and 100 mg/mL, respectively), 20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, GlutaMAX (Thermo Fisher Scientific 1:100), fresh 2-mercaptoethanol (55 µM), and recombinant human IL-2 (20 ng/mL; STEMCELL Technologies; 78036)] and transferred to a new plate to discontinue stimulation. T cells were counted and resuspended in T cell medium daily at 1 to 3 × 10⁶ cells/mL. T cells were stained at 3 h poststimulation with Annexin V (BD; 559763) according to the manufacturer’s instructions and anti-CD4 BUV737 and anti-CD8 BV785 (BioLegend; 100750). T cells were also stained with anti-CD4 BUV737 and anti-CD8 BV785 on days 4 and 5. For killing assays, MC38 or MC38-αCD3 cells were added at 5 × 10⁴ cells/well to 96-well RTCA E-plates (Agilent; 300600910). On day 7 post-T cell stimulation, T cells were added at various E:T ratios to the wells. Killing was measured by adherent cell impedance every 15 min. Impedance at each time point was normalized to the impedance value at time of T cell addition.

Lymphoid Tissue Characterization.

Thymuses/spleens/bilateral inguinal LNs were collected from 4-wk-old (thymus) or 9- to 11-wk-old (spleen/LNs) age- and sex-matched mice, weighed (spleen/thymus), mechanically dissociated, and washed by passage through a 70 µm strainer, treated with 180 µL DNase I (STEMCELL Technologies; 07900; 1 mg/mL stock) in 3 mL FACS buffer [2% fetal bovine serum in phosphate-buffered saline (PBS)] with 5 mM MgCl₂ for 15 min (thymus only), washed, and treated with ACK lysis buffer for 3 min. After washing again, viable cells were counted using trypan blue exclusion. Cells (10⁶) were treated with mouse FcR blocking reagent (Miltenyi; 130-092-575) and stained with the appropriate panel (SI Appendix, Tables S2 and S3) for 30 min in the dark at 4 °C. Samples were washed in PBS and stained with LIVE/DEAD Fixable Blue (Thermo Fisher Scientific; L23105) for 15 min in the dark at 4 °C, washed, fixed in 2% paraformaldehyde for 20 min in the dark at 4 °C, and washed again before acquiring data on an LSR II UV flow cytometer. CD5 and CD69 expression were assessed on samples gated on cells, single cells, live cells, B220^- cells, and CD4⁺CD8^- or CD4^-CD8⁺ cells. CD5 and CD69 gates were set based on fluorescence minus one controls (FMOs). CD44, CD62L, and CD49d expression was assessed on samples gated on cells, single cells, live cells, CD45⁺ cells, and CD4⁺CD8^- or CD4^-CD8⁺ cells. The CD49d gate was set based on FMOs.

Tumor Assays.

Mice (8 to 12-wk-old, both male and female, sex-matched) were injected subcutaneously with 5 × 10⁵ MC38 or MC38-OVA cells in 100 µL PBS on day 0. Where indicated, mice were treated with 200 µg anti-PD-1 (Bio X Cell; BE0146; clone RMP1-14) or isotype control antibody (Bio X Cell; BE0089; clone 2A3) on days 5, 8, 11, and 14. Tumors were measured by using calipers, and volume was calculated as V = 0.5(L × W²) where V = volume, L = length of longest dimension, and W = length of perpendicular dimension. Similar to prior studies (68–70), we sorted mice into responders and nonresponders. We defined responders as mice whose tumors had a less than 30% increase in volume between day 8 (the first day of measurement) and day 23 (the last day of measurement). Nonresponders were defined as mice whose tumors exhibited a ≥30% increase in volume. This cutoff represents a stringent threshold, as even a 1 mm fluctuation in each tumor dimension due to caliper measurement error could result in apparent tumor volume changes exceeding 50% at the average control mouse tumor size on day 8 (116 mm³). Mice humanely killed for tumor ulceration or poor health before day 23 were excluded from statistical analysis (by the chi-square test). Groups of mice were only compared if from the same experiment (e.g., isotype versus αPD-1, isotype versus isotype, αPD-1 versus αPD-1). For adoptive transfer of wild type 2D2 or OT-II CD4⁺ T cells, donor splenic T cells were isolated by negative selection (STEMCELL Technologies; 19852), and 10⁵ cells were injected retro-orbitally in PBS into sex-matched recipients after tumor injection on day 0.

For characterization of the immune microenvironment, mice were anesthetized with ketamine and perfused with 30 mL ice-cold PBS containing 2 to 5 mM ethylenediaminetetraacetic acid to flush out leukocytes remaining in the vasculature. Tumors were weighed, and a portion of tumor was placed in 2 mL digestion mix [9 parts DMEM, 1 part 10X collagenase/hyaluronidase (STEMCELL Technologies; 07912), 0.1 parts DNase I], cut into 1 to 2 mm pieces, and incubated for 30 min at 37 °C and 5% CO₂ with intermittent gentle swirling. Digested tumors were pressed through 70 µm strainers, washed, and treated with ACK lysis buffer for 1 min before washing again. After counting, 1.5 × 10⁶ cells were treated with mouse FcR blocking reagent for 10 min at 4 °C and stained in the dark at room temperature for 20 min with the appropriate panels (SI Appendix, Tables S2 and S3). Aliquots were also treated with 81 nM PMA, 1.39 µM ionomycin, and 5 µg/mL Brefeldin A and incubated at 37 °C, 5% CO₂ for 4 h before staining with P4 (SI Appendix, Tables S2 and S3). Samples were washed in PBS, stained with LIVE/DEAD Fixable Blue (Thermo Fisher Scientific; L23105) for 15 min in the dark at 4 °C, washed again, fixed overnight (eBioscience FoxP3 transcription factor staining buffer set; Thermo Fisher Scientific; 00-5523-00), and permeabilized. Samples were then incubated with FcR blocking reagent, rat serum, and mouse serum for 10 min before staining with panel antibodies targeting intracellular antigens for 30 min at 4 °C. See SI Appendix, Fig. S5 for examples of gating strategies for P1, P2, and P3. For P1, gates for CD69, LAG3, OX40, and TIM3 were set based on FMOs. For P2, FMO was used for PD-L1 gating. For P3, FMOs were used for gating CTLA4, RORγt, T-BET, and GATA3. For P4, IFNγ, TNFα, and IL-2 expression were assessed by gating on cells, single cells, live cells, CD45⁺ cells, and CD3⁺CD8⁺ cells. FMOs were used to gate on IFNγ, TNFα, and IL-2.

Vaccinia Virus Infection.

VACV-OVA was propagated in BSC-40 cells. On day 0, mice were infected by adding 10 µL of 5 × 10⁶ PFU of VACV-OVA to the ventral side of each pinna and subsequently scarifying with a 29-gauge needle. Ear thickness was measured by calipers on days 0, 7, 9, 11, and 15. Five minutes before being killed, mice were retro-orbitally injected with anti-CD45 BUV395 (2 µg) to label leukocytes still in the vasculature (for analytical exclusion). On day 15, mice were killed, and one ear was taken for plaque assay. The other ear was collected for flow cytometry. Dorsal and ventral sides of the ear were separated and incubated at 37 °C for 45 min in HBSS (Hyclone) containing CaCl₂, MgCl₂, 125 U/mL collagenase D (Invitrogen), and 60 U/mL Dnase-I (Sigma-Aldrich). Ears were ground on a scored plastic dish and poured through a 70 µm strainer. Cells (1.5 × 10⁶) were stained with H-2 Kb SIINFEKL (OVA_257-264) tetramer BV421 (NIH tetramer core facility) for 45 min at room temperature. Samples were washed, FcR-blocked, and stained with anti-CD3 BV605, anti-CD4 BUV737, and anti-CD8a BV785. Samples were then stained with LIVE/DEAD Fixable Blue for 15 min in the dark at 4 °C, washed, fixed in 2% paraformaldehyde for 20 min in the dark at 4 °C, and washed again before acquiring data on a BD FACSymphony A5 flow cytometer.

Quantification and Statistical Analysis.

Graphs were generated and statistical analyses were performed using GraphPad Prism v.10. Groups were compared by two-way unpaired t tests, one-way ANOVA, two-way ANOVA, or chi-squared test as indicated. Area under the curve was calculated with RStudio. Means with error bars representing SE are shown. Relevant P values are indicated in the figures.

Illustrations.

Illustrations were created with BioRender.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.