Selective depletion of virus-specific CD8 T cells from the liver after PD-1 therapy with Fc-intact antibody during chronic infection

Abstract

Anti-PD-1 antibody therapy is now widely used in various cancers. However, the role of the antibody Fc region in PD-1 directed immunotherapy is not well understood. Preclinical studies commonly use species-mismatched rat anti-mouse antibodies, which may not accurately reflect antibody-Fcγ receptor (FcγR) interactions. Here, we used mouse anti-mouse PD-1 antibodies to investigate how the Fc region influences therapeutic efficacy for enhancing CD8 T cell responses using mouse models of chronic lymphocytic choriomeningitis virus infection and CT26 tumors. Treatment with these mouse anti-mouse PD-1 antibodies caused preferential depletion of PD-1+ virus-specific CD8 T cells in the liver, resulting in increased viral titers. These effects of mouse anti-PD-1 antibodies were Fc dependent since mutating the Fc region to block FcγR interaction prevented PD-1+ CD8 T cell depletion and resulted in effective immunotherapy. Using mice lacking activating FcγR III or inhibitory FcγR IIb, we found that depletion of PD-1+ CD8 T cells was mediated via activating FcγR III. Furthermore, we determined that phagocytic cells, not NK cells, were the in vivo effectors that mediated depletion of PD-1+ CD8 T cells. Similar depletion of tumor-specific CD8 T cells and reduced tumor control were observed in the CT26 model with Fc-intact mouse anti-mouse PD-1 treatment. These findings highlight potential negative effects of Fc-functional anti-PD-1 antibodies in therapies for liver cancer, liver metastases, and chronic hepatotropic viral infections. Conversely, FcγR-mediated depletion could benefit “agonistic” anti-PD-1 antibodies for treatment of autoimmunity. Our research emphasizes the importance of Fc region in tailoring PD-1 therapies for diverse clinical applications.

Introduction

Immune checkpoint inhibitors targeting the PD-1 (programmed cell death 1)/PD-L1 pathway have transformed cancer treatment and show promise for treating persistent viral infections, with over 10 drugs approved worldwide for more than 20 different cancer types (1–3). A key mechanism of action of PD-1 targeted therapy is the release of inhibitory signals mediated by PD-1/PD-L1 interaction, which enhances PD-1-expressing antigen-specific CD8 T cell responses in the presence of the persistent antigens. However, despite these advances, the majority of patients do not respond to PD-1 therapy, highlighting the need for a deeper understanding of the mechanisms underlying its variable efficacy (4, 5).

While the therapeutic efficacy of monoclonal antibodies is often attributed to their ability to block or stimulate specific pathways, their activity is also influenced by interactions between their Fc regions and the host’s Fc gamma receptor (FcγR) system (6–12). The Fc region of antibodies, which binds to FcγRs on immune cells, plays a critical role in modulating immune responses by recruiting effector functions such as antibody-dependent cellular cytotoxicity (ADCC) and antibody-dependent cellular phagocytosis (ADCP). However, the role of the Fc region in PD-1 directed immunotherapy remains incompletely understood.

Currently, anti-PD-1 antibodies with human IgG4 (hIgG4) backbones are most widely used, because hIgG4 is considered neutral for FcγR-mediated antibody effector function (7, 12–14). However, hIgG4-based anti-PD-1 antibodies may still induce FcγR-mediated antibody effector functions, as hIgG4 can bind to all activating human FcγRs, including FcγRI (13–15). Indeed, hIgG4 antibodies can exert FcγR-mediated effector function, including depletion of target cells and cytokine storm (16, 17). Furthermore, hIgG4-based anti-PD-1 antibodies bind human FcγRI with high affinity, mediating crosslinking between PD-1 and FcγRI that induces FcγRI+ macrophages to phagocytose PD-1+ T cells in vitro (18). Additionally, this hIgG4-based anti-PD-1 antibody therapy reduces tumor-infiltrating PD-1+ CD8 T cells and fails to control tumor growth in a xenograft allogeneic cancer model. Similar observations have been reported in various tumor models with diverse immune environments (19).

Additionally, “agonistic” antibodies targeting PD-1 have been recently developed to suppress unwanted immune responses, such as in autoimmune diseases (20–28). For instance, peresolimab showed promising efficacy in a phase 2a rheumatoid arthritis trial (26), with Fc engagement with FcγR suggested to play a critical role (21). Similarly, positive data for rosnilimab, a depleting agonist targeting PD-1+ T cells, have emerged from an ongoing phase 2b rheumatoid arthritis trial (20, 28). These findings highlight the importance of elucidating Fc-dependent mechanisms to optimize anti-PD-1 antibodies and expand their therapeutic applications to new disease contexts.

Preclinical studies should use the species-matched therapeutic antibodies to better recapitulate antibody-FcγR interactions found in human patients (13). Here, we generated panels of mouse anti-mouse PD-1 antibodies and investigated how Fc regions influence therapeutic efficacy and antigen-specific CD8 T cell responses. We used mouse models of chronic lymphocytic choriomeningitis virus (LCMV) infection and CT26 tumors. This work offers new insights into the mechanisms of PD-1-directed immunotherapy and aims to inform the design of next-generation therapeutic antibodies, including “agonistic” anti-PD-1 antibodies.

Results

Treatment with mouse anti-mouse PD-1 antibody depletes LCMV-specific CD8 T cells and increases viral titer selectively in the liver during chronic infection

We generated panels of anti-mouse PD-1 antibodies in PD-1 knockout mice (SI Appendix, Table S1). To evaluate the therapeutic efficacy of mouse anti-mouse PD-1 antibodies in enhancing LCMV-specific CD8 T cell responses and reducing viral load during chronic infection, we used the PD-1 antibody clone 332.8H3 (mouse IgG1 (mIgG1)) (29–32). LCMV chronically infected mice were generated by transient CD4 T cell depletion, followed by LCMV clone 13 infection. While total CD4 T cell numbers including Foxp3+ regulatory CD4 T cells come back to near-normal levels, these mice remain highly deficient in LCMV-specific CD4 T cell numbers as well as LCMV-specific antibody responses (33, 34). This approach establishes lifelong viremia with high levels of virus in almost every tissue in the mouse and serves as a stringent model for studying CD8 T cell exhaustion as well as PD-1 blockade effects (33–36). This also prevents the generation of excessive amounts of immune complexes that mitigate Fc-mediated antibody effector function (37–39). PD-1 is highly expressed by virus-specific CD8 T cells in various tissues during chronic LCMV infection (SI Appendix, Fig. S1A), as opposed to virus-specific memory CD8 T cells in mice that have cleared an acute LCMV Armstrong infection (35) (SI Appendix, Fig. S1B). LCMV chronically infected mice (> day 40 post-infection) were either left untreated or treated with the mouse PD-1 antibody for two weeks, and LCMV-specific CD8 T cell responses and viral titer were examined in spleen, lungs, and liver (Fig. 1A). While the number of LCMV-specific D^bGP276+ CD8 T cells increased in the spleen and lungs, we observed a reduction in the liver among mice treated with the mouse anti-PD-1 antibody (Fig. 1 B and C). A similar trend was noted for LCMV-specific D^bGP33+ CD8 T cells (Fig. 1 D and E). Consequently, treatment with the mouse anti-mouse PD-1 antibody improved viral control in the spleen and lungs but led to increased viral titer in the liver (Fig. 1F). Additionally, viral titer and the numbers of LCMV-specific CD8 T cells were inversely correlated in the liver and other tissues (Fig. 2), suggesting that the decrease in LCMV-specific CD8 T cells directly contributed to impaired viral control in the liver following treatment with a mouse anti-mouse PD-1 antibody. It is interesting to note how tissue-specific viral control can be seen within an individual mouse with viral titer decreasing in the spleen and lung due to PD-1 therapy but actually increasing in the liver due to depletion of LCMV-specific CD8 T cells from the liver.

Figure 1. Treatment with mouse anti-mouse PD-1 antibody depletes LCMV-specific CD8 T cells and increases viral titer selectively in the liver during chronic infection.

(A) Experimental design. LCMV chronically infected mice (> day 40 post-infection) were left untreated or treated with a mouse anti-mouse PD-1 antibody (clone 332.8H3, mIgG1) every 3 days for 2 weeks, followed by analysis of LCMV-specific CD8 T cells and viral titer in the tissues. (B) Representative FACS plots of D^bGP276-specific CD8 T cells by tetramer staining in the indicated tissues. (C) Summary of D^bGP276-specific CD8 T cell numbers. (D) Representative FACS plots of D^bGP33-specific CD8 T cells. (E) Summary of D^bGP33-specific CD8 T cell numbers. (F) Summary of viral titer in the indicated tissues. Results were pooled from 3–5 experiments with n=4–5 mice per group in each experiment. Statistical comparisons were performed using an unpaired Mann-Whitney test. Results for the liver are highlighted with a red square (B, C). Bars and error bars represent the geometric mean and 95% confidence interval (B, C). Untx, untreated.

Figure 2. Inverse correlation between the number of LCMV-specific CD8 T cells and viral titer during chronic infection.

Correlation between viral titer in the indicated tissues and the number of LCMV-specific (D^bGP33+ and D^bGP276+) CD8 T cells in each tissue. Data are derived from Fig. 1. Statistical significance of the Pearson correlation coefficient was assessed using a two-tailed test. Results for the liver are highlighted with a red square. Untx, untreated.

To determine whether the selective reduction in LCMV-specific CD8 T cells and impaired viral control in the liver were exclusive to mice treated with a mouse anti-mouse PD-1 antibody, we administered a rat anti-mouse PD-1 antibody (clone 29F.1A12, rat IgG2a (rIgG2a)) to LCMV chronically infected mice (SI Appendix, Fig. S2A). Interestingly, this treatment, as shown before (35), resulted in an increase in the number of LCMV-specific D^bGP276+ CD8 T cells in the spleen, lungs, and liver (SI Appendix, Fig. S2 B and C). Similar trend was seen for LCMV-specific D^bGP33+ CD8 T cells (SI Appendix, Fig. S2 D and E). As a result, impaired viral control in the liver was not observed in mice treated with the rat anti-mouse PD-1 antibody, and viral titer decreased in all examined organs (SI Appendix, Fig. S2F). These results indicated species-matched mouse Fc region may play a role in depleting LCMV-specific CD8 T cells from the liver following anti-PD-1 antibody treatment during chronic infection.

Treatment with an Fc-mutated mouse anti-mouse PD-1 antibody does not deplete LCMV-specific CD8 T cells and improves viral control in the liver during chronic infection

We hypothesized that the depletion of PD-1+ LCMV-specific CD8 T cells selectively from the liver following treatment with a mouse anti-mouse PD-1 antibody during chronic infection is mediated by the Fc region of the antibody. To explore this, we engineered the Fc region of the anti-PD-1 antibody clone 332.8H3 (mIgG1) to create a variant antibody (clone 2203, mIgG1-D265A) that lacks detectable binding to FcγR (40). We assessed the therapeutic effectiveness of this Fc mutant mouse anti-mouse PD-1 antibody in a two-week treatment regimen (Fig. 3A). The numbers of LCMV-specific CD8 T cells in the liver were slightly increased, when the Fc mutant anti-PD-1 antibody was used (Fig. 3B). Notably, the Fc mutant antibody significantly enhanced viral control in the liver, whereas viral titer was elevated with the Fc wild-type (wt) mouse anti-mouse PD-1 antibody (Fig. 3C). Furthermore, treatment with the Fc mutant anti-PD-1 antibody resulted in a more pronounced increase in LCMV-specific CD8 T cells and a significant reduction in viral titer in the spleen and lungs compared to the wild-type antibody treatment (Fig. 3 B and C).

Figure 3. Treatment with an Fc-mutated mouse anti-mouse PD-1 antibody does not deplete LCMV-specific CD8 T cells and improves viral control in the liver during chronic infection.

(A) Experimental design. LCMV chronically infected mice (> day 40 post-infection) were left untreated or treated with Fc wild type (wt) (332.8H3, mIgG1) or Fc mutant anti-PD-1 antibody (2203, mIgG1-D265A) every 3 days for 2 weeks, followed by analysis of LCMV-specific CD8 T cells and viral titer in the tissues. (B) Summary data for the numbers of LCMV-specific (D^bGP33+ and D^bGP276+) CD8 T cells. (C) Viral titer in the indicated tissues. Results were pooled from 5 experiments with n=2–3 mice per group in each experiment. Statistical comparisons were performed using (B) the Kruskal-Wallis test with Dunn’s correction or (C) one-way ANOVA with Dunnett’s correction for multiple comparisons. Bars and error bars represent the geometric mean and 95% confidence interval (B) or the mean and standard deviation (C). Untx, untreated; wt, wild type.

Fc-intact mouse anti-mouse PD-1 antibody rapidly depletes LCMV-specific CD8 T cells from the liver within 24 hours after treatment of chronically infected mice

We further examined the dynamics of LCMV-specific CD8 T cell depletion in the liver. A single dose of the Fc wt or Fc mutant mouse anti-mouse PD-1 antibody was administered to LCMV chronically infected mice, and the liver and other tissues were analyzed 24 hours post-treatment (Fig. 4A). Notably, we observed a significant and rapid decline in the number of LCMV-specific CD8 T cells in the liver, with reductions of 97-fold for D^bGP33+ and 257-fold for D^bGP276+ CD8 T cells just 24 hours after injection of the Fc wt anti-PD-1 antibody (Fig. 4 B and C). In contrast, the populations of LCMV-specific CD8 T cells remained stable in the liver of mice treated with the Fc mutant anti-PD-1 antibody, clearly distinguishing it from the effects of the wild-type antibody (Fig. 4 B and C). These findings collectively underscore the critical role of the Fc region of the mouse anti-mouse PD-1 antibody in modulating LCMV-specific CD8 T cell responses, particularly in the liver during PD-1 therapy in the context of chronic viral infection.

Figure 4. Fc-intact mouse anti-mouse PD-1 antibody rapidly depletes LCMV-specific CD8 T cells from the liver within 24 hours after treatment of chronically infected mice.

(A) Experimental design. LCMV chronically infected mice (> day 40 post-infection) were left untreated or treated with a single injection of Fc wild type (wt) (332.8H3, mIgG1) or Fc mutant anti-PD-1 antibody (2203, mIgG1-D265A). LCMV-specific CD8 T cells in the liver were analyzed 24 hours later. (B) Representative FACS plots of D^bGP276-specific CD8 T cells by tetramer staining. (C) Summary data for the numbers of D^bGP33- and D^bGP276-specific CD8 T cells in the liver. Results were pooled from 5 experiments with n=2–3 mice per group in each experiment. Statistical comparisons were performed using the Kruskal-Wallis test with Dunn’s correction for multiple comparisons. Contrasting FACS plots in mice treated with Fc wt and Fc mutant anti-PD-1 antibody are highlighted with a red square (B). Error bars represent the geometric mean and 95% confidence interval (C). Untx, untreated; wt, wild type.

Depletion of LCMV-specific CD8 T cells caused by treatment with mouse anti-mouse PD-1 antibodies is not limited to a specific antibody clone or IgG subclass

It is unexpected that an anti-PD-1 antibody of the mIgG1 subclass can deplete PD-1+ antigen-specific CD8 T cells, as mIgG1 has traditionally been considered to exert minimal, if any, antibody effector functions in mice due to its preferential binding to the inhibitory FcγR IIB rather than the activating FcγR III (7, 12–14, 41). To determine whether the capacity of the anti-PD-1 antibody to deplete PD-1+ CD8 T cells in the liver was a specific phenomenon of clone 332.8H3 (mIgG1), we evaluated a panel of mouse anti-mouse PD-1 antibodies of various IgG subclasses (2 mIgG1, 6 mIgG2a, 1 mIgG2b), along with two rat anti-mouse PD-1 antibodies (rIgG2a) in LCMV chronically infected mice (SI Appendix, Table S1). Remarkably, a single dose of any mouse anti-mouse PD-1 antibody, but not the rat anti-mouse PD-1 antibodies (29F.1A12 and RMP1–14), resulted in significant reductions (> 10-fold) in LCMV-specific CD8 T cells in the liver within 24 hours (Fig. 5). These findings indicate that mouse anti-mouse PD-1 antibodies, regardless of clone or IgG subclass, effectively target PD-1+ CD8 T cells in the liver during chronic LCMV infection.

Figure 5. Depletion of LCMV-specific CD8 T cells caused by treatment with mouse anti-mouse PD-1 antibodies is not limited to a specific antibody clone or IgG subclass.

LCMV chronically infected mice (> day 40 post-infection) were treated with a single injection of each of 9 mouse or 2 rat anti-mouse PD-1 antibodies. LCMV-specific CD8 T cells in the liver were analyzed 24 hours later. Summary data for the numbers of LCMV-specific (D^bGP33+ and D^bGP276+) CD8 T cells in the liver are shown. Results were pooled from >10 experiments with n=1–4 mice per group in each experiment. Statistical comparisons were performed using the Kruskal-Wallis test with Dunn’s correction for multiple comparisons, where **P < 0.01, ***P < 0.001, ****P < 0.0001. Bars and error bars indicate geometric mean and 95% confidence interval. ns, not significant; Untx, untreated; mIgG, mouse immunoglobulin G; rIgG, rat immunoglobulin G.

Rat anti-mouse PD-1 antibody becomes a “depleting” antibody when its Fc region is replaced with a species-matched mouse Fc region

We then examined the impact of Fc regions from mouse and rat antibodies by comparing the rat anti-mouse PD-1 antibody RMP1–14 (rIgG2a) with a modified version in which its Fc region was replaced with mIgG1. These anti-PD-1 antibodies were administered to mice chronically infected with LCMV, and the quantity of LCMV-specific CD8 T cells in the liver was assessed 24 hours later. The number of LCMV-specific CD8 T cells rapidly decreased when the rat anti-mouse PD-1 antibody was modified to contain the mouse Fc region (Fig. 6 A and B), emphasizing the importance of the physiological interactions between the Fc portions of therapeutic antibodies and mouse FcγR.

Figure 6. Rat anti-mouse PD-1 antibody becomes a “depleting” antibody when its Fc region is replaced with a species-matched mouse Fc region.

LCMV chronically infected mice (> day 40 post-infection) were treated with a single injection of rat-mouse chimeric (RMP1–14-CP162, mouse IgG1 Fc) or rat (RMP1–14, rat IgG2a Fc) anti-PD-1 antibody. LCMV-specific CD8 T cells in the liver were analyzed 24 hours later. (A) Representative FACS plots of D^bGP33- and D^bGP276-specific CD8 T cells by tetramer staining in the liver. (B) Summary data for the numbers of LCMV-specific (D^bGP33+ and D^bGP276+) CD8 T cells in the liver. Results were pooled from 2–4 experiments with n=2–3 mice per group in each experiment. Statistical comparisons were performed using the Kruskal-Wallis test with Dunn’s correction for multiple comparisons. Untx, untreated.

Depletion of LCMV-specific CD8 T cells in the liver by anti-PD-1 antibody is mediated via activating FcγR III, not inhibitory FcγR IIb

To elucidate the mechanism underlying anti-PD-1 antibody-mediated depletion of LCMV-specific CD8 T cells in the liver during chronic viral infection, we investigated the specific FcγRs involved in the antibody’s effector functions. The anti-PD-1 antibody, belonging to the mIgG1 subclass, is capable of binding to both the inhibitory FcγR IIB and the activating FcγR III (7, 11, 13, 14). To assess the contributions of specific FcγRs to the rapid depletion of LCMV-specific CD8 T cells in the liver following anti-PD-1 antibody therapy during chronic infection, we conducted experiments using FcγR IIB- or FcγR III-knockout mice. These LCMV chronically infected mice received a single injection of the Fc-intact mouse anti-mouse PD-1 antibody, and liver tissues were analyzed 24 hours post-treatment (Fig. 7A). Interestingly, depletion of LCMV-specific CD8 T cells in the liver following anti-PD-1 antibody therapy occurred in mice lacking inhibitory FcγR IIB (Fig. 7 B and C). In contrast, depletion of LCMV-specific CD8 T cells in the liver following anti-PD-1 antibody therapy did not occur in mice lacking FcγR III (Fig. 7 B and C). These results show the activating FcγR III was involved in PD-1+ CD8 T cell depletion in the liver by the anti-PD-1 antibody.

Figure 7. Depletion of LCMV-specific CD8 T cells in the liver is mediated via activating FcγR III whereas the inhibitory FcγR IIb is dispensable.

(A) Experimental design. Groups of LCMV chronically infected (> day 40 post-infection) WT mice, or FcγR IIB knockout (KO) mice, or FcγR III KO mice were left untreated or treated with a single dose of anti-PD-1 antibody (332.8H3, mIgG1). LCMV-specific CD8 T cells in the liver were analyzed 24 hours later. (B) Representative FACS plots for D^bGP33-specific CD8 T cells in the liver by tetramer staining. (C) Summary data for the numbers of D^bGP33- and D^bGP276-specific CD8 T cells in the liver. Results were pooled from 2 (B) or 3 (C) experiments with n=1–3 mice per group. Statistical comparisons were performed using an unpaired Mann-Whitney test. Results from FcγR III KO mice are highlighted with a red square (B). Bars and error bars represent the geometric mean and 95% confidence interval (C). Untx, untreated; wt, wild type.

Phagocytic cells and not NK cells play a major role in anti-PD-1 antibody-mediated depletion of LCMV-specific CD8 T cells during chronic infection

We then explored the roles of natural killer (NK) cells and phagocytic cells, both of which are known to mediate FcγR-dependent antibody effector functions (12–14). In LCMV chronically infected mice, NK cells were depleted using anti-NK1.1 antibody, while phagocytic cells were depleted with clodronate-filled liposomes. Following these depletions, Fc wt mouse anti-mouse PD-1 antibody was administered, and liver tissues were analyzed 24 hours later (Fig. 8A). Flow cytometry confirmed the effective depletion of NK cells and phagocytic cells in the liver of these mice (SI Appendix, Fig. S3 A and B). Notably, LCMV-specific CD8 T cells were still effectively depleted by the Fc wt anti-PD-1 antibody in NK cell-depleted mice. In contrast, depleting phagocytic cells prior to anti-PD-1 antibody therapy completely prevented the depletion of LCMV-specific CD8 T cells in the liver (Fig. 8 B and C). Together, these results indicated that the engagement of activating FcγR III by the Fc region of the anti-PD-1 antibody led to antibody-dependent phagocytosis of LCMV-specific CD8 T cells in the liver during chronic infection.

Figure 8. Phagocytic cells and not NK cells play a major role in anti-PD-1 antibody-mediated depletion of LCMV-specific CD8 T cells during chronic infection.

(A) Experimental design. LCMV chronically infected mice (> day 40 post-infection) were left untreated or treated with a single injection of 250 μg of anti-NK1.1 antibody (PK136) or 200 μl of clodronate-filled liposome to deplete NK cells or phagocytic cells, respectively. The following day for anti-NK1.1-treated mice or two days later for clodronate-treated mice, these groups of mice were either left untreated or given a single dose of 200 μg of anti-PD-1 antibody (332.8H3, mIgG1). LCMV-specific CD8 T cells in the liver were then analyzed 24 hours later. (B) Representative FACS plots for D^bGP33-specific CD8 T cells by tetramer staining are shown. (C) Summary data for the numbers of D^bGP33- and D^bGP276-specific CD8 T cells in the liver in the indicated groups. Results were pooled from 3 independent experiments with n=2–3 mice per group. Statistical comparisons were performed using the Kruskal-Wallis test with Dunn’s correction. Results from clodronate-treated mice are highlighted with a red square (A). Error bars represent the geometric mean and 95% confidence interval (B). Untx, untreated; wt, wild type.

FcγR-mediated anti-PD-1 antibody effector function is compromised during chronic LCMV infection with CD4 T cell help due to the presence of excessive amounts of antigen-antibody immune complexes

We so far examined FcγR-mediated anti-PD-1 antibody effector function in chronic LCMV model with life-long viremia, where transient CD4 T cell depletion compromised generation of LCMV-specific CD4 T cell responses and antibody responses. Accordingly, this model prevents the generation of excessive amounts of antigen-antibody immune complexes. High levels of immune complexes, which are composed of antibodies bound to specific antigens, are known to be associated with compromised FcγR-mediated antibody effector functions and have been shown to suppress FcγR-mediated antibody effector functions during chronic viral infection (37–39) and systemic lupus erythematosus (SLE) (42–44). Therefore, it remains unclear whether preferential depletion of LCMV-specific CD8 T cells in the liver by anti-PD-1 antibody could occur in the presence of excessive antigen-antibody immune complexes.

To address this question, we used another chronic LCMV model containing virus-specific CD4 T cells and LCMV-specific antibodies, leading to the generation of excessive amounts of immune complexes formed by LCMV-specific antibodies and viral antigens (37–39). Mice were infected with LCMV clone 13, and Fc wt or Fc mutant mouse anti-mouse PD-1 antibody was administered at day 21 post-infection. The number of LCMV-specific CD8 T cells was evaluated in the liver 24 hours later (SI Appendix, Fig. S4A). As anticipated, we found no decrease in LCMV-specific D^bGP33+ CD8 T cells and a marginal decrease of LCMV-specific D^bGP276+ CD8 T cells (2.2-fold) in the liver at 24 hours after injecting Fc wt anti-PD-1 antibody (SI Appendix, Fig. S4 B and C). Thus, anti-PD-1 antibody mediated depletion of LCMV-specific CD8 T cells in the liver was mitigated in the presence of high amounts of immune complexes during chronic infection. We then examined therapeutic efficacy of Fc wt and Fc mutant mouse anti-mouse PD-1 antibodies in a 2-week treatment regimen during chronic LCMV infection in the presence of high amounts of immune complexes. Mice were infected with LCMV clone 13 and received either Fc wt or Fc mutant anti-PD-1 antibody, administered every 3 days from day 23 to day 35 post-infection (SI Appendix, Fig. S5A). Both antibodies were similarly effective in promoting viral clearance from serum and multiple tissues including the liver (SI Appendix, Fig. S5B). Together, these results indicated that anti-PD-1 antibody-mediated effector function could be mitigated in some circumstances by excessive immune complex formation.

Our previous study showed that afucosylated antibodies with increased FcγR affinity can enhance effector function and overcome excessive immune complexes (39). Therefore, we generated an afucosylated version of mouse anti-mouse PD-1 antibody (332.8H3, mIgG2c-afuc) and administered to mice chronically infected with LCMV at day 21–22 post-infection. The number of LCMV-specific CD8 T cells was evaluated in the liver 24–36 hours later (SI Appendix, Fig. S6A). Consistent with the data from chronic LCMV model with life-long viremia and low immune complexes (Fig. 1 A–C), treatment with the afucosylated anti-PD-1 antibody resulted in a >10-fold decrease in LCMV-specific CD8 T cells in the liver at 24–36 hours, while no or minimal decrease in the spleen and lung (SI Appendix, Fig. S6 B–E). These results indicate that the liver remains the primary site for depletion of LCMV-specific CD8 T cells by anti-PD-1 antibody with enhanced FcγR affinity, even in the presence of excessive immune complexes, depending on the biological context and the antibody’s effector function.

Fc-intact mouse anti-mouse PD-1 antibody therapy has negative impact on tumor control

Finally, we investigated the impact of FcγR-mediated anti-PD-1 antibody effector functions on therapeutic efficacy in cancer treatment. To assess tumor-specific CD8 T cell responses and control of tumor growth, we utilized the mouse model of CT26 colon carcinoma. Tumor-bearing mice received either Fc wt or Fc mutant mouse anti-mouse PD-1 antibody, administered every three days from day 9 to day 27 following tumor inoculation (Fig. 9A). We found that only a minority of untreated mice (7 out of 31, 22.6%) and those treated with the Fc wt anti-PD-1 antibody (7 out of 40, 17.5%) had controlled tumor growth, as indicated by tumor volumes less than 400 mm3 by 27 days post-tumor inoculation (Fig. 9B). In contrast, mice treated with the Fc mutant anti-PD-1 antibody exhibited significantly better tumor control, with 26 out of 35 mice (74.3%) demonstrating suppressed tumor growth (Fig. 9B). Of note, tumor growth was accelerated in mice treated with the Fc wt anti-PD-1 antibody compared to untreated mice, while those receiving the Fc mutant anti-PD-1 antibody showed the most effective tumor control among all treatment groups (Fig. 9C). Additionally, tumor-specific CD8 T cell responses, assessed using L^dGP70-specific dextramer (45), were enhanced by treatment with the Fc mutant anti-PD-1 antibody but were significantly diminished in those treated with the Fc wt anti-PD-1 antibody (Fig. 9 D and E). These findings indicated that FcγR-mediated effector functions of the Fc-intact mouse anti-PD-1 antibody play a critical role in determining its therapeutic efficacy against cancer.

Figure 9. Fc-intact mouse anti-mouse PD-1 antibody therapy has negative impact on tumor control.

(A) Experimental design. CT26 tumor-bearing mice were left untreated or treated with anti-PD-1 antibody with wild type (wt) Fc (332.8H3, mIgG1) or mutated Fc (2203, mIgG1-D265A) every 3 days starting from day 9 post-tumor implantation. (B and C) Summary data for the tumor growth curve represented by tumor volume in (B) individual and (C) pooled mice (mean and standard deviation) at indicated time points (days 8–27). (D) Representative FACS plots of L^dGP70 tumor-specific CD8 T cells by dextramer staining in the tumor at day 27 post-tumor implantation. (E) Summary data for the numbers of L^dGP70 tumor-specific CD8 T cells per gram of tumor. Data represent results from 3 (B and C) or 2 (D and E) independent experiments with n=3–10 mice per group in each experiment. Dotted line represents cut-off value for tumor volume at 400 mm3 (B). Statistical comparisons were performed using one-way ANOVA with Dunnett’s correction for multiple comparisons. Error bars represent the mean and standard deviation (C) or the geometric mean and 95% confidence interval (E). Untx, untreated.

Discussion

In this study, we showed that treatment of mice with a mouse anti-mouse PD-1 antibody (mIgG1) led to depletion of antigen-specific PD-1+ CD8 T cells during chronic LCMV infection and cancer. T cell depletion by PD-1 therapy was observed preferentially in the liver during chronic LCMV infection, resulting in impaired viral control in this anatomical compartment. Similar to virus-specific CD8 T cells in the liver during chronic LCMV infection, intratumoral PD-1+ tumor-specific CD8 T cells were also depleted in CT26 tumor-bearing mice that were treated with the Fc wt anti-PD-1 antibody, and tumor growth was accelerated in those mice. In contrast, mice treated with the Fc mutant anti-PD-1 antibody did not deplete PD-1+ tumor-specific CD8 T cells and had better tumor control. The adverse therapeutic outcomes resulting from T cell depletion through anti-PD-1 antibody effector function are in stark contrast to previous studies on the mechanism of action of therapeutic antibodies against CTLA-4 (cytotoxic T-lymphocyte-associated antigen-4). These studies have demonstrated enhanced tumor control by anti-CTLA-4 therapy through the FcγR-dependent depletion of regulatory T cells (Tregs) within the tumor microenvironment (6, 8, 10, 41).

Depletion of PD-1+ CD8 T cells was facilitated by the interaction between the Fc region of the anti-PD-1 antibody and FcγRs. Therapeutic use of the Fc mutant anti-PD-1 antibody led to enhanced responses of LCMV-specific and tumor-specific CD8 T cells. As a result, treatment with the Fc mutant anti-PD-1 antibody resulted in reduced viral titer in the liver during chronic LCMV infection and improved tumor control in the CT26 tumor model. Additional mechanistic investigations uncovered that engagement of the activating FcγR III by the Fc region of the anti-PD-1 antibody (mIgG1) triggered the phagocytosis of LCMV-specific CD8 T cells in the context of chronic infection. Neither inhibitory FcγR IIb nor NK cells were found to be responsible for T cell depletion through the effector function of the anti-PD-1 antibody, aligning with prior studies that highlight the significant role of phagocytosis by macrophages expressing activating FcγRs in antibody-mediated cell depletion during chronic LCMV infection (37, 38). Since macrophages are major components of the leukocyte infiltrates in tumors (46, 47), PD-1+ T cell depletion in the CT26 cancer model in our study is likely due to anti-PD-1 antibody-dependent cellular phagocytosis of PD-1+ T cells. These findings are in line with earlier studies, demonstrating that anti-CTLA-4 antibody-mediated Treg depletion is also coordinated by tumor infiltrating macrophages that express activating FcγRs (6, 10).

One key observation of the present study is that PD-1+ T cell depletion occurred preferentially in the liver. The liver contains phagocytic cells like Kupffer cells that express activating FcγRs (14, 48–50). In addition, a unique anatomical feature of the liver, where CD8 T cells share localization with Kupffer cells in the sinusoid, enables circulating anti-PD-1 antibody to easily access PD-1+ CD8 T cells in this organ (51, 52). These factors collectively contribute to the preferential depletion of LCMV-specific CD8 T cells in the liver following anti-PD-1 antibody treatment during chronic infection. To test whether other commonly used depleting antibodies also preferentially target cells in the liver compared to the other tissues, we injected anti-CD8α depleting antibody (clone 2.43, rIgG2b) (39) into LCMV chronically infected mice (SI Appendix, Fig. S7A). We found that there was depletion of CD8 T cells from all tissues after a 2-week regimen of anti-CD8α antibody treatment but with the greatest reduction in the liver (~19-fold decrease) (SI Appendix, Fig. S7 B and C). We further assessed CD8 T cells in the spleen, lung, and liver 24 hours after a single injection of anti-CD8α antibodies (SI Appendix, Fig. S7D). It is worth noting that after a short-term antibody treatment (24 hours), there was again more effective depletion of CD8 T cells in the liver compared to other tissues (SI Appendix, Fig. S7 E and F). Thus, preferential depletion in the liver is seen after both anti-CD8α and anti-PD-1 Fc-intact antibodies.

In this context, it is worth pointing out an important difference between the anti-CD8α and anti-PD-1 antibodies. The anti-CD8α antibody is purely a depleting antibody that very rapidly depletes CD8 T cells from the liver within a day and CD8 T cells from other tissues are also depleted after more sustained antibody treatment. In contrast, the anti-PD-1 Fc-intact antibodies are not only depleting antibodies but also remove the PD-1 inhibitory signals, resulting in a proliferative burst of new effector CD8 T cells from the stem-like CD8 T cells residing in the spleen (36). Thus, after treatment of chronically infected mice with the PD-1 Fc-intact antibodies, both expansion and depletion of the LCMV-specific CD8 T cells are going on at the same time. In the liver, the depletion effect is much stronger than expansion, hence the great reduction of LCMV-specific CD8 T cells in the liver accompanied by an increase in the viral titer, whereas in the spleen, LCMV-specific CD8 T cell expansion is favored over depletion and there is an increase in the number of LCMV-specific CD8 T cells and a reduction in viral titer (Fig. 1C). These are novel findings showing the organ-specific effects of PD-1 blockade using PD-1 depleting antibodies.

Analogous findings have shown that the liver is a major site for the depletion of circulating B cells by anti-CD20 therapy, where intravascular access plays a critical role (48, 50, 53). Furthermore, during chronic infection, LCMV-specific CD8 T cells are predominantly tissue-resident and minimally circulating (54). As a result, the therapeutic effects of anti-PD-1 antibody therapy can be compartmentalized to individual organs, depending on the balance between the extent of T cell depletion and the enhanced T cell responses, including the recruitment of newly generated effector T cells from secondary lymphoid organs into circulation and peripheral tissues, through the blockade of the PD-1/PD-L1 interaction. These responses are likely to be differentially regulated across various anatomical compartments. Importantly, the organ-specific therapeutic effects observed with anti-PD-1 therapy in our preclinical model could extend to patients with cancer, as metastatic lesions in patients with non-small-cell lung cancer or mismatch repair deficiency carcinoma have exhibited varying responses to PD-1 blockade across different organ sites, with liver metastases exhibiting notably lower responsiveness (55). Interestingly, some cancer patients develop a more aggressive disease progression, known as hyperprogressive disease, following treatment with immune checkpoint inhibitors (56–58), and the presence of liver metastasis is proposed as one of the risk factors (59–61).

It should be noted that several variables influence FcγR-mediated antibody effector functions. For instance, excessive immune complexes can impair these functions (37–39). In patients with systemic lupus erythematosus (SLE) or a mouse model of lupus, the IgG fraction of lupus serum, containing autoantibodies and potentially immune complexes, resulted in resistance to B cell depletion by anti-CD20 antibody (42–44). Similarly, the effectiveness of depleting anti-PD-1 antibodies may be reduced under conditions of high immune complex levels, as shown in SI Appendix, Fig. S4 and Fig. S5. Nonetheless, our previous study demonstrated that CD8 T cell depletion using an anti-CD8α antibody achieved ~70% efficiency in the liver compared to < 20 % efficacy in the spleen after 2 days of anti-CD8α antibody injection during chronic LCMV infection, even in the presence of excessive immune complexes (39), highlighting the liver as an efficient site for antibody-mediated depletion of target cells, even in the context of elevated immune complex levels. While the mouse anti-PD-1 antibody (332.8H3, mIgG1) showed marginal depleting activity in the liver (~2.2-fold decrease = approximately 55% reduction) during chronic LCMV infection with abundant immune complexes (SI Appendix, Fig. S4), this depletion efficiency was not competitive with enhancing LCMV-specific CD8 T cell responses via blocking the PD-1/PD-L1 pathway, leading to improved viral control achieved in the chronic LCMV model with excessive immune complexes (SI Appendix, Fig. S5). Of note, when Fc affinity to FcγR was enhanced using an afucosylated version of anti-PD-1 antibody, selective depletion of LCMV-specific CD8 T cells in the liver could occur even in the presence of excessive immune complexes (SI Appendix, Fig. S6). These findings underscore the importance, in a given anti-PD-1 antibody, of the competition between anti-PD-1 mediated PD-1+ T cell depletion and the blocking activity of PD-1/PD-L1 inhibitory pathway in determining its biological activities.

Another critical factor in antibody-mediated target cell depletion is the level of target antigen expression on cell surfaces (39). For example, PD-1 is selectively expressed on LCMV-specific CD8 T cells during chronic infection (SI Appendix, Fig. S1 A) and tumor-specific CD8 T cells in tumors (Fig. 9D), enabling Fc-intact anti-PD-1 antibodies to effectively deplete PD-1+ T cell. When the target antigen is highly abundant, such as CD90.2, depleting antibodies can effectively deplete CD90.2+ T cells even in the presence of high immune complex levels during chronic LCMV infection (39). In contrast, when a target antigen is either expressed at low levels or widely distributed, antibody-mediated target cell depletion may become less effective. For instance, anti-IFNAR-1 (MAR1–5A3, mIgG1), despite its intact Fc region and frequent use for blocking type I IFN activity in vivo (62, 63), shows limited depletion activity, likely due to the ubiquitous distribution of IFNAR-1-expressing cells (64). Thus, the overall effectiveness of FcγR-mediated antibody effector functions depends on multiple factors, including the abundance and distribution of the target antigen, the presence of immune complexes, and the Fc region’s binding affinity to FcγRs. These factors should be carefully considered when using the Fc wt anti-PD-1 antibodies that may deplete PD-1+ CD8 T cells in diseases like chronic infections or cancer, where they are crucial to eliminate infected cells or cancer cells.

Conversely, Fc-intact anti-PD-1 antibodies with robust FcγR-mediated antibody effector functions hold potential for treating autoimmunity by depleting PD-1+ autoreactive T cells to dampen immune responses. Several so-called “agonistic” anti-PD-1 antibodies aimed at suppressing immune responses are currently in development (20–27). A recent phase 2a trial of the PD-1 agonist monoclonal antibody peresolimab showed promising results in patients with rheumatoid arthritis, with FcγR engagement proposed to contribute to its “agonistic” activity (21, 26). Another depleting and agonist anti-PD-1 antibody rosnilimab is also showing positive data in ongoing phase 2b trial in rheumatoid arthritis (20, 28). Our findings are highly relevant for understanding the alternative mechanism of these “agonistic” antibodies. These antibodies are Fc-intact and may well be working by depleting autoreactive T cells as opposed to “agonistic” PD-1 signalling. Future studies should clarify the relative contributions of agonism vs. depletion in their immune suppressive effects.

In conclusion, our study reveals tissue-specific PD-1 therapy effects influenced by Fc-FcγR mediated antibody effector function, particularly in the liver during chronic viral infection and in the tumor microenvironment, which have important implications for optimizing PD-1 therapy towards hepatotropic infections, liver cancer, and liver metastases. Additionally, our findings offer opportunities to explore novel applications of anti-PD-1 antibodies to dampen immune responses in situations such as autoimmunity and transplantation.

Materials and Methods

A detailed description of materials and methods is provided in SI Appendix, Materials and Methods.

Mice.

Six- to 8-week-old female C57BL/6J and BALB/cJ mice were purchased from the Jackson Laboratory (Bar Harbor, ME). FcγR IIB and FcγR III knockout mice were kindly gifted by Dr. Jeffrey V. Ravetch (The Rockefeller University).

Chronic LCMV infection model.

PD-1 blockade studies were performed using mice with life-long LCMV viremia, a stringent model of CD8 T cell exhaustion (33, 35). LCMV chronically infected mice were generated as follows. C57BL/6J mice were transiently depleted of CD4 T cells by injecting 300–500 μg of anti-CD4 antibody (GK1.5, BioXCell) intraperitoneally (i.p.) twice on days −2 and 0, followed by intravenous (i.v.) infection with 2 × 106 PFU of LCMV clone 13. For assessing the impact of excessive amounts of antigen-antibody immune complexes on PD-1 blockade, mice were infected with LCMV clone 13 without transient CD4 T cell depletion. Viral titers were determined by plaque assay on Vero E6 cells. All experiments were conducted in accordance with National Institutes of Health and the Emory University Institutional Animal Care and Use Committee guidelines.

Tumor model.

CT26 cells were purchased from the American Type Culture Collection (ATCC). BALB/cJ mice, without transient CD4 T cell depletion, were implanted with 1 × 10⁵ CT26 colon carcinoma cells suspended in 25% Matrigel (Corning) subcutaneously in the right and/or left flank. Treatment was started on day 9 post-inoculation. Mice were monitored daily and removed from the study group when the tumors exceeded 20 mm in their largest diameter. Tumor volume was measured using a caliper as the volume of an ellipsoid (length × width × height × 0.52).

Mouse anti-mouse PD-1 and rat anti-mouse PD-1 antibodies.

Nine mouse anti-PD-1 antibodies (SI Appendix, Table S1) were generated in PD-1 knockout mice (31) and prepared in-house, with endotoxin levels verified to be less than 2 EU/mg. Rat anti-mouse PD-1 antibody 29F.1A12 was prepared in-house. Rat anti-mouse PD-1 antibody RMP1–14 and rat-mouse chimeric anti-mouse PD-1 antibody RMP1–14-CP162 were obtained from BioXCell.

Recombinant anti-PD-1 antibody.

The clone 2203 anti–mouse PD-1 antibody is derived from the 332.8H3 mouse anti–mouse PD-1 antibody clone (VK and VH sequences in SI Appendix, Table S2, mIgG1, kappa) (31) but has a D265A mutation in the mouse IgG1 heavy chain. Clone 2203 was produced in the pEFGF expression vector in CHO cells using DHFR amplification (65), purified from culture supernatants by Protein G affinity chromatography, and verified to have an endotoxin level of less than 2 EU/mg.

Afucosylated anti-mouse PD-1 antibody.

Afucosylated 332.8H3 mouse anti-mouse PD-1 antibody was produced by transiently transfecting Expi293F cells with plasmids encoding the 332.8H3 light chain and the 332.8H3 heavy chain as mIgG2c and culturing cells in the presence of 50 micromolar 2F-Peracetyl-Fucose (Cayman Chemical) as previously described (39), followed by antibody purification from culture supernatants using CaptureSelect^™ IgG-Fc (multispecies) affinity matrix (Thermo Fisher Scientific).

Significance Statement.

The Fc region’s role in PD-1-directed immunotherapy is pivotal yet underexplored. This study revealed that species-matched mouse anti-mouse Fc-intact PD-1 antibodies caused activating FcγR-mediated depletion of PD-1+ CD8 T cells via phagocytic cells, particularly in the liver during chronic LCMV infection, impairing viral control there. Comparable depletion of intratumoral PD-1+ tumor-specific CD8 T cells occurred post-treatment in CT26 tumor-bearing mice, accelerating tumor growth. These findings highlight negative effects on treating hepatotropic infections and liver cancer with Fc-intact anti-PD-1 antibodies. Conversely, FcγR-mediated PD-1+ CD8 T cell depletion may enhance “agonistic” anti-PD-1 antibodies’ efficacy in autoimmunity. Our research underscores the importance of strategically customizing the Fc region to optimize PD-1 therapies for diverse clinical applications across cancer, chronic infections, and autoimmunity.

Data, Materials, and Software Availability.

All data are included in the manuscript and/or supporting information.