Delayed administration of anti‐PD‐1 antibody reverses immune dysfunction and improves survival during sepsis

Introduction

Sepsis, the systemic, inflammatory response syndrome that occurs during severe infection, kills more than 210,000 people in the United States annually [1]. Sepsis initiates a complex immune response that varies during the duration of the illness [2, 3, 4, 5]. Although recent studies have shown that proinflammatory and anti‐inflammatory responses occur simultaneously during the first phase of the disease, the early net result is often a potent “cytokine storm”‐mediated hyperinflammatory response [4]. With contemporary standard‐of‐care measures, the majority of patients survive the early hyperinflammatory phase and enter a stage of protracted immunosuppression, which has been termed “immunoparalysis.” This immunosuppression in sepsis is manifest by failure to clear the primary infection, reactivation of HSV‐I and cytomegalovirus, and development of new, secondary infections, often with organisms that are not particularly virulent to the immunocompetent patient [6, 7, 8]. A recent autopsy study of 235 patients who had a diagnosis of sepsis in the surgical intensive care unit showed that ∼80% had a persistent site of infection on postmortem examination, thus highlighting the significant degree of immune compromise in these patients [9].

Although a number of mechanisms of immune suppression have been shown to play a role in sepsis, the precise basis for the immunoparalysis is not well‐delineated. It is likely that a number of interacting processes are responsible, and these mechanisms include apoptotic depletion of immune effector cells, increased T regulatory cells [10], increased myeloid‐derived suppressor cells [11], and a shift from a TH1 immune phenotype to an anergic or TH2 phenotype [3, 4, 5]. Another possibility for the immunosuppression of sepsis is an up‐regulation of inhibitory proteins or a down‐regulation of costimulatory proteins [12, 13, 14, 15]. The relatively recently identified receptor, PD‐1, has been shown to be a crucial modulator of host immune responses [16, 17, 18]. PD‐1, a member of the B7‐CD28 superfamily, is a negative, costimulatory molecule that is inducibly expressed, primarily on the cell surface of activated CD4 and CD8 T cells [16, 17, 18]. Signaling through PD‐1 inhibits the ability of T cells to proliferate and produce cytokines and attenuates cytotoxic T cell function [12, 13]. The key immunomodulatory role of PD‐1 is revealed by studies showing that PD‐1 knockout mice develop autoimmune diseases, including a lupus‐like syndrome and cardiomyopathy [14, 15]. The two known agonistic ligands to PD‐1, i.e., PD‐L1 (B7‐H1) and PD‐L2 (B7‐DC), are regulated by different mechanisms and expressed by different cell types. PD‐L1 is inducibly expressed across a broad spectrum of immune and nonimmune cells, including T cells, B cells, DCs, macrophages, endothelial cells, epithelial cells, and pancreatic islet cells [17, 18]. PD‐L2 is mainly expressed on DCs and macrophages. Ligation of PD‐1 with PD‐L1, in concordance with TCR or BCR activation, leads to recruitment of SHP‐2, dephosphorylating the essential scaffolding protein Gab2 and preventing activation of PI3K, which limits downstream activation of Akt, the essential kinase in costimulatory receptor signaling [12].

PD‐1 can inhibit a robust immune response, thereby facilitating chronic viral and fungal infections [16, 19, 20, 21, 22, 23]. In a mouse model of lymphocytic choriomeningitis, functional exhaustion of virus‐specific CD8 T cells was shown to correlate with increased PD‐1 expression, and CD8⁺ T cells lacking PD‐1 are more resistant to tolerance than wild‐type CD8⁺ T cells [19]. In patients infected with HIV‐1, PD‐1 expression renders infected T cells more susceptible to apoptosis, and it is postulated that the virus down‐regulates PD‐1 to ensure its survival [24]. In a mouse model of Histoplasma capsulatum, PD‐L1 was up‐regulated on macrophages and splenocytes during infection, and PD‐1‐deficient mice had improved survival compared with wild‐type mice [23]. Treatment with a blocking antibody to PD‐1 was shown to improve survival in wild‐type mice infected with Histoplasma [23]. Blockade of PD‐1/ligand interactions has been shown to reverse T cell tolerance and exhaustion and is associated with resolution of chronic infection models [12, 13].

Relatively little work has been performed on the role of PD‐1 in the severe bacterial infections characterizing sepsis. Huang et al. [25] showed recently that mice deficient in PD‐1 have increased bacterial clearance and improved survival in the CLP model of sepsis. Their work showed that PD‐1 contributes to macrophage dysfunction during sepsis and concluded that PD‐1 could be an important therapeutic target. Our previous studies have shown that PD‐1 mRNA is highly induced in CD4 T and B cells, beginning at 8 h and persisting at 18 h after sepsis [26].

The purpose of this study was to determine if the survival advantage observed in the PD‐1 null mice in sepsis could be recapitulated therapeutically with an antagonistic antibody to PD‐1. A potential problem with using any knockout mice to examine particular pathways and therapies is that compensatory pathways have been shown to occur with genetic deletions, thereby resulting in misleading findings about the true nature of the gene in question. Thus, the results in null mice may not reflect the effects of acute inhibition of the gene in a clinical setting. Second, the results in PD‐1 null mice may not reflect the effects of blocking PD‐1 after sepsis has already occurred. In addition, this study examined potential mechanistic insights into the putative, beneficial effect of PD‐1 antagonism in sepsis.

MATERIALS AND METHODS

RESULTS

DISCUSSION

PD‐1 is an important modulator of T cell function. It mediates peripheral tolerance, induces T cell anergy and exhaustion in chronic infections, and renders T cells more susceptible to apoptosis [12, 13]. PD‐1 deficiency or blockade has been shown to improve survival in a small number of viral, bacterial, and fungal infections [16, 23, 24, 36]. We hypothesized that blockade of PD‐1 with a mAb would improve survival in a mouse model of intra‐abdominal sepsis. Data herein show that PD‐1 blockade improves survival in the widely used and clinically relevant CLP model of sepsis. It is important to note that the administration of anti‐PD‐1 did not occur until 24 h after the onset of sepsis, during the time‐point when PD‐1 receptor expression was beginning to increase. If these beneficial effects in animal models of sepsis translate into clinical studies, it will mean that delayed administration of the therapy can still be effective in patients with sepsis, a major benefit in treatment, as patients with sepsis may not present to the hospital until many hours after sepsis onset.

The present studies complement the work of Huang and associates [25], who noted that PD‐1 null mice had improved survival in sepsis. The present study does significantly strengthen the hypothesis that inhibiting PD‐1 in sepsis is a viable therapeutic approach. Studies involving knockout mice, in which the gene has been deleted, are hostage to the fact that compensatory pathways may have occurred in response to the genetic deletion. These compensatory changes may confound the proper interpretation of the nature of the gene of interest. Thus, the findings in these gene knockout mice, with the exception of conditional knockout mice, may not reflect the results of acute inhibition of the gene. Most significantly, the results in knockout mice do not allow for investigation of the effects of blocking the gene after the particular insult in question, i.e., sepsis. Despite the methodological differences between the two studies, there is broad agreement on a number of points. The present study showing increased PD‐1 expression on CD4 and CD8 T cells is consistent with the work from Huang et al. [25], who reported a time‐dependent increase in PD‐1 expression on macrophages from septic mice. The studies by Huang et al. [25] pointed to a key role of PD‐1 expression in macrophages in modulating microbial clearance and the inflammatory response. These investigators also determined that there was decreased tissue destruction in jejunum and kidney and less evidence of compacted and fragmented nuclei (indicators of cellular apoptosis) in spleens and thymi from PD‐1 null mice. Collectively, the two studies suggest that PD‐1 is involved in the innate and adaptive immune responses to infection and emphasize the important “crosstalk” between the innate and adaptive immune system in sepsis.

An important effect of anti‐PD‐1 therapy was its action to prevent the sepsis‐induced loss in the DTH response. A key feature of patients with sepsis is their loss of the DTH response to recall antigens [35]. This finding is emblematic of the profound immunosuppression that characterizes patients with sepsis. PD‐1 is recognized to be an important factor in the “exhaustion” of T cells during chronic viral infection. Work by several laboratories has demonstrated that blockade of the PD‐1 inhibitory pathway leads to restoration of T cell function, including enhancing their ability to proliferate and produce appropriate cytokines with stimulation [12, 13, 37]. The ability of the septic mice that were treated with anti‐PD‐1 to mount an appropriate DTH response in the present study is consistent with the re‐establishment of effective immunity. In patients with sepsis, this re‐establishment of immune function may allow the host to eradicate the primary infection and resist new secondary infections.

An interesting and highly important action of blocking PD‐1 was the effect to markedly diminish sepsis‐induced apoptosis. A key determinant of life or death of immune cells is the relationship between the pro‐ and antiapoptotic Bcl‐2 family members. Few studies have examined the relationship of PD‐1 on Bcl‐2 family members [38]. Lin et al. [38] transferred splenocytes from mice infected with lymphocytic choriomeningitis virus into lymphopenic T cell‐deficient hosts and showed that a PD‐1‐positive and a PD‐1‐negative population of T cells resulted. The population of cells that were positive for PD‐1 was dysfunctional, had lower Bcl‐2 levels, and disappeared from the host by Day 70, presumably because of apoptosis. Petrovas et al. [20] examined CD8 T cells from patients with HIV‐1‐positive donors and reported that high expression of PD‐1 correlated with increased ex vivo, spontaneous apoptosis, low expression of Bcl‐2, and decreased IL‐7R expression. The most likely explanation for the inter‐relationship between PD‐1 and Bcl‐xL is a result of the inhibitory action of PD‐1 on CD28 signaling [12, 39], which plays an essential role in the translation of Bcl‐xL via a PI3K/Akt‐dependent pathway [39]. PD‐1 acts to increase SHP‐2, which inhibits PI3K signaling, thereby preventing phosphorylation of AKT [17], and inhibiting production of BcL‐xL [12, 39]. Work presented herein showing that the anti‐PD‐1 antibody markedly increased Bcl‐xL is particularly important, given previous studies from our group showing that transgenic mice that overexpress Bcl‐xL have improved survival in sepsis [32].

The present results add to the increasing body of evidence demonstrating that immune modulation represents a viable therapeutic approach to sepsis. Although there are a number of possible methods to enhancing the immune response in sepsis, one such approach involves manipulation of the costimulatory molecules that have diverse functions on cells of the innate and adaptive immune system. Animal studies have now shown that modulation of the positive and negative costimulatory receptors on immune cells using agonistic antibodies to CD40, anti‐PD‐1, or anti‐CTLA‐4 can ameliorate adverse effects of sepsis and improve survival ([25, 40] and unpublished data). Collectively, these diverse studies make a compelling argument for a new approach to sepsis that involves manipulation of the host immune response. Anti‐PD‐1‐ and anti‐PD‐L1 antibody‐based therapy is currently in several Phase 2 and 3 clinical trials in patients with cancer (http://clinicaltrials.gov). Although there are some concerns about the potential for autoimmune reactions in patients who are on long‐term anti‐PD‐1/anti‐PD‐L1 therapy, these antibodies appear to be well‐tolerated to date. Given the fact that patients with sepsis would not need prolonged therapy with anti‐PD‐1/anti‐PD‐L1 antibodies, the concerns for development of autoimmunity would be much diminished. If additional animal studies confirm its efficacy, anti‐PD‐1 should be strongly considered for clinical trials in sepsis.

AUTHORSHIP

P.M. performed many experiments, analyzed data, and helped write the manuscript. S.I. helped perform experiments. J.U. helped perform experiments and analyzed data. K.C.C. helped perform experiments. J.E.M. helped design experiments and write the manuscript. R.S.S. helped design experiments and write the manuscript.