|Home | About | Journals | Submit | Contact Us | Français|
CD4+CD25+Foxp3+ Tregs suppress autoimmune responses. In addition, they limit T cell responses during chronic infection, thereby minimizing T cell–dependent immunopathology. We sought to investigate how Tregs are regulated in the livers of patients chronically infected with HCV, where they control the balance between an adequate protective immune response and suppression of immunopathology. We found that, despite accumulating and proliferating at sites of infection in the livers of patients chronically infected with HCV, Tregs were relatively less expanded than CD4+CD25+Foxp3– effector T cells. The relative lower expansion of intrahepatic Tregs coincided with their upregulation of programmed death–1 (PD-1). PD-1 expression inversely correlated with both Treg proliferation and clinical markers of immune suppression in vivo. Consistent with the possibility that PD-1 controls Tregs, blockade of the interaction between PD-1 and programmed death–1 ligand 1 (PD-L1) enhanced the in vitro expansion and function of Tregs isolated from the livers of patients chronically infected with HCV. Blockade of the interaction between PD-L1 and B7.1 also improved the proliferation of these cells. Interestingly, both PD-1 and phosphorylated STAT-5 were overexpressed in intrahepatic Tregs in a parallel fashion in steady disease conditions, and in an alternate-fluctuating fashion during the course of severe hepatitis reactivation. Notably, PD-L1 blockade upregulated STAT-5 phosphorylation in Tregs ex vivo. These data suggest that PD-L1 negatively regulates Tregs at sites of chronic inflammation by controlling STAT-5 phosphorylation.
CD4+CD25+Foxp3+ Tregs are essential in maintaining peripheral immunological self tolerance (1–11). Natural Tregs develop intrathymically in response to specific combinations of TCR and costimulatory signals and represent about 5%–15% of peripheral CD4+ T cells, whereas adaptive Tregs are generated extrathymically from conventional CD4+ T cells, acquiring CD25, Foxp3, and suppression function in response to appropriate stimuli (1–11). Tregs require TCR engagement to nonspecifically suppress via various mechanisms, including cell-cell contact by only partially identified membrane molecules, suppressive cytokine production, and dominant adsorption of IL-2 by high CD25 expression (CD25hi) (1–13). Although several biomarkers have previously been determined to be useful in defining Tregs in different physiological or pathological conditions (14–17), the discovery of Foxp3 represented a milestone in the identification of Tregs, because it has been related to the development and suppression function of these cells (4–6). In addition, Foxp3 seems to determine critical molecular patterns of Tregs, including anergy and the incapacity to produce IL-2 or other survival cytokines (18–21). The overexpression of IL-2 receptor (IL-2R) may compensate for this defect by adsorbing high amounts of paracrine IL-2, essential for Treg proliferation (22–24). Absent or decreased Foxp3 expression evolves into different scenarios, ranging from X-linked massive multiorgan autoimmunity (25–27) to impaired suppression function observed in various human and experimental autoimmune disorders (28–31).
Apart from their role in controlling autoimmunity, Tregs seem to perform a critical task in curtailing harmful responses associated with chronic infections (32). Indeed, high Foxp3 expression has been related to the control of immunopathology established by a wide array of persistent pathogens (32). In particular, Tregs have been proposed to participate in the establishment of a fine equilibrium between immunopathology and immune protection, ultimately resulting in long-lasting survival of the host during chronic infections. However, it is unclear how Tregs fine-tune this balance, thus enabling the control of excessive immunopathology without completely suppressing the protective immune responses.
An additional mechanism limiting effector T cell (Teff) responses is related to the propensity of activated T cells to overexpress the programmed death 1 (PD-1) receptor (33, 34). PD-1 can bind both programmed death–1 ligand 1 (PD-L1), which is upregulated on a wide variety of hematopoietic and nonhematopoietic cells, and PD-L2, which is only upregulated on DCs, macrophages, and cultured bone marrow mast cells (33, 34). Upon the simultaneous engagement of PD-1 and B or T cell antigen-specific receptors to their respective ligands, the cytoplasmic domain–associated immunoreceptor tyrosine-based inhibitory motif of PD-1 is phosphorylated and recruits a series of Src homology 2–containing tyrosine phosphatases (SHPs), leading to the dephosphorylation of the TCR- or BCR-associated effector molecules (33, 34). The inhibitory PD-1/PD-L1 pathway has previously been demonstrated to inhibit both microbial and self-reactive T cell responses and thus to participate in the establishment of chronic infections on the one hand or self tolerance on the other (33–43). In addition, both PD-1 and PD-L1 have been demonstrated to be expressed on Tregs, but whether they influence Treg function is unknown (34, 44–46).
To investigate the roles of Tregs, PD-1 expression, and the possible interplay between their functions in the establishment of long-lasting inflammatory diseases, we studied chronic HCV infection, which is known to establish a lifelong disease in the majority of infected individuals (47, 48). Here, we provide evidence suggesting that a negative regulatory mechanism (referred to herein as contraregulatory) takes place at the level of chronic inflammatory sites via PD-1 overexpression that controls the phosphorylation of STAT-5 in Tregs.
First, Tregs were identified in PBLs isolated from 65 patients with chronic HCV infection and 30 healthy donors (HDs) as well as in intrahepatic lymphocytes (IHLs) isolated from liver biopsies of 57 of the 65 HCV patients studied. PBLs isolated from HDs are referred to herein as HD-PBLs, and cell populations isolated from HCV patients are referred to as HCV-PBLs and HCV-IHLs. CD4+CD25+ T cells appeared markedly more accumulated at the intrahepatic than at the peripheral level (Figure (Figure1,1, A and B). More importantly, both the absolute number of CD4+CD25+Foxp3+ Tregs and their level of Foxp3 expression were significantly higher in HCV-IHLs than in either HCV-PBLs or HD-PBLs (Figure (Figure1,1, A, C, and D). Likewise, the number of CD4+CD25+Foxp3– Teffs was significantly higher in IHLs than in PBLs (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI36604DS1). However, the relative percentage of Foxp3+ cells, compared with that of Foxp3– cells, was significantly lower in the intrahepatic than in the peripheral CD4+CD25+ T cells in all HCV patients studied (Figure (Figure1,1, A and E). Intrahepatic CD4+CD25+Foxp3– T cells were bona fide Teffs because they alone — none of the other intrahepatic or peripheral CD4+ T cell subsets (including Tregs) — expressed significant amounts of IL-2 ex vivo when IHLs were stimulated for 6–8 h with a combination of mAbs to CD3 and CD28 conjugated to Dynabeads (anti-CD3/CD28; Figure Figure2,2, A and B). Both the percentage of IL-2+ Teffs (Figure (Figure2,2, A and B) and their level of IL-2 mean fluorescence intensity (MFI; data not shown) were significantly higher in IHLs than in PBLs, which suggests that memory effector cells with the highest functional potentiality are recruited in the inflamed tissues. The finding that virtually all the IHLs producing IL-2 were Foxp3– (Figure (Figure2,2, A and B) argues against the possibility that the intrahepatic Foxp3 upregulation was the result of a transient, nonfunctional expression by Teffs infiltrating the inflamed liver (49). To investigate whether the intrahepatic Treg accumulation reflected recruitment or increased proliferation, we enumerated Tregs expressing the proliferation marker Ki67 in fresh HCV-IHLs and HCV-PBLs. In comparison to intrahepatic CD4+CD25–Foxp3– T cells or peripheral Tregs, a considerable proportion of intrahepatic Foxp3+ Tregs overexpressed Ki67, albeit to a significantly lesser extent than did intrahepatic CD4+CD25+Foxp3– Teffs (Figure (Figure2,2, C and D).
To examine Treg function in vitro, highly purified peripheral CD4+CD25–Foxp3– responder T cells (Tresps) (50) were labeled with CFSE and stimulated with anti-CD3/CD28 for 6 d in the presence or absence of highly purified unlabeled intrahepatic Tregs, peripheral Tregs, or CD4+CD25–Foxp3– T cells (control cells; Figure Figure3A).3A). Because of the paucity of IHLs recovered from a single biopsy (no more than 1 × 105 cells), intrahepatic Tregs used in the suppression assay were purified from pooled HCV-IHLs isolated from 3–4 liver biopsies from a total of 16 patients (51). Both intrahepatic (Figure (Figure3B)3B) and peripheral Foxp3+ Tregs (data not shown), but not control Foxp3– cells (data not shown), efficiently performed suppression function, as detected by the CFSE dilution of Tresps (Figure (Figure3B).3B). The suppression was exerted via a cell-cell contact mechanism (52), as it was inhibited by the separation of Tresps and Tregs in a transwell plate system (Figure (Figure3B).3B). Intrahepatic Foxp3+ cell frequencies were directly related to the suppression function (Figure (Figure3C),3C), but inversely correlated with the expression of Ki67 in intrahepatic Teffs and Tregs (Figure (Figure3,3, D and E). In addition, intrahepatic Foxp3+ cells correlated with the clinical consequences of the immune suppression, demonstrated by the rise of plasma viremia (Figure (Figure3F)3F) and the decline of hepatic damage (Figure (Figure3G),3G), as calculated by the histological activity index (HAI) score (53). This finding is clearly different from a previous immune histochemistry description showing that increased inflammation directly correlated with increased Tregs in the liver (54). This disparity is likely attributable to the different systems used. Immune histochemistry can detect more precisely the localization of Foxp3+ cells in the different liver areas and is a preferential system to visualize the inflammatory scenario in situ, but it is less sensitive than flow cytometry assay, particularly in multiple staining assays. Foxp3+ cell frequencies did not correlate with Ki67 expression, viral load, or HAI score at the peripheral level (data not shown). In addition, purified intrahepatic and peripheral CD4+CD25+ cells did not significantly differ in their capacity to perform suppressor function, despite the lower percentage of Foxp3+ cells in the former (Figure (Figure1,1, A and E). This finding is likely due to the higher level of Foxp3 expression in intrahepatic than in peripheral CD4+CD25+ cells (Figure (Figure1D),1D), which may compensate for the lower percentage of Foxp3+ cells in the liver (Figure (Figure1E). 1E).
To determine why intrahepatic Treg expansion and proliferation were constantly lower than intrahepatic Teff expansion in vivo (Figure (Figure1E1E and Figure Figure2,2, C and D), we first examined the expression of the inhibitory PD-1 molecule on both HCV-PBL and HCV-IHL T cell populations from 31 of the 65 HCV patients studied (33, 34). In agreement with previous studies (42, 43), PD-1 was upregulated on activated CD4+ or CD4– CD25+Foxp3– Teffs derived from HCV-infected livers compared with the expression of PD-1 by peripheral Teffs of PBLs derived from HDs or HCV patients (HD-PBLs or HCV-PBLs, respectively; Supplemental Figure 2, A–D). However, more important in the context of this study is the finding that Tregs overexpressed PD-1 to a significantly higher extent than did activated Teffs in both HCV-IHLs and HCV-PBLs in all patients studied (Figure (Figure4,4, A–C). In addition, HCV-IHL Tregs upregulated PD-1 significantly more than did HCV-PBL or HD-PBL Tregs (Figure (Figure4,4, A–C, and Supplemental Figure 3, A and B). Importantly, the ratio between the PD-1+ and Foxp3+ Treg percentages (PD-1/Foxp3 cell ratio) infiltrating the liver of HCV patients correlated inversely with the viral load (Figure (Figure4D)4D) and directly with the HAI score (Figure (Figure4E),4E), thus indicating that PD-1 on Tregs restrains the Foxp3-related suppression of both antiviral and immunopathology-related responses. No correlation among the parameters was shown at the level of the peripheral T cell populations (data not shown).
To determine the functional capacity of PD-1 expression, we first selected HCV-PBL samples with abundant expression of PD-1 (Figure (Figure4)4) and purified both Tregs and Tresps that were stained or not with CFSE. After 6 d of stimulation, CFSE-labeled Tresps proliferated in response to either antigenic stimulus, with autologous APCs pulsed with both HCV-NS3 and HCV-core antigens (HCV-Ags/APCs; Figure Figure5A),5A), or nonspecific stimulus, with anti-CD3/CD28 (data not shown), and an even higher extent of proliferation was observed in the presence of a blocking anti–PD-L1 mAb (Figure (Figure5A).5A). Even in the presence of anti–PD-L1 mAb, less than 1% of these cells became Foxp3+ (Figure (Figure5A),5A), which supports the in vivo evidence suggesting that Foxp3 expression was not the result of a transient, nonfunctional expression by Teffs in our model (Figure (Figure2,2, A and B, and ref. 49). Cumulative experiments with purified Tresps from a total of 9 HCV patients confirmed that PD-L1 blockade resulted in a significant increase of HCV-specific T cell proliferation compared with isotype control treatment (Figure (Figure5B),5B), thus highlighting the recently observed role of PD-1 in the exhaustion of Teffs during chronic infections (36–43).
Next, we assessed the effect of PD-L1 blockade on the expansion of Tregs. Consistent with the finding that Tregs are anergic in vitro (8), CFSE-labeled Tregs were unable to proliferate after 6 d of contact with antigenic (Figure (Figure6A)6A) or nonspecific stimulus (data not shown) or when cultured in a transwell plate system, in which they were seeded with the antigenic stimulus in the lower wells and unstimulated Tresps were cultured in the upper wells (Figure (Figure6A).6A). However, Tregs were rescued from anergy and efficiently proliferated, as demonstrated by the substantial expansion of Foxp3+ cells when Tregs and Tresps were independently stimulated in the separated transwell plates (i.e., Tregs stimulated with HCV-Ags/APCs and Tresps stimulated with anti-CD3/CD28; Figure Figure6A).6A). Conversely, Tregs remained anergic when an anti–IL-2 mAb was added to the wells containing the stimulated Tresps, suggesting that IL-2 produced by Tresps plays a major role in the expansion of Tregs (Figure (Figure6A).6A). The antigen specificity of Foxp3+ Tregs was highlighted by the evidence that their antigen-driven expansion was consistently inhibited by the addition of anti–MHC class II mAb, but not anti–MHC class I mAb, to the wells containing Tregs (Figure (Figure6A).6A). The addition of anti–PD-L1 mAb to cultures containing Tregs alone in the presence of HCV-Ags/APCs or anti-CD3/CD28 was unable to rescue Tregs from anergy in vitro (Figure (Figure6B),6B), even though these cells expressed both PD-1 (Figure (Figure4)4) and PD-L1 (data not shown). On the contrary, the expansion of intrahepatic CD25+Foxp3+ Tregs in response to the combination of the antigenic stimulus and either autologous IL-2–producing Tresps or soluble IL-2 substantially improved with PD-L1 blockade compared with culture conditions in the absence of anti–PD-L1 (Figure (Figure6B).6B). Similar results were obtained by stimulating Tregs with anti-CD3/CD28 and IL-2 in the presence of anti–PD-L1, which suggests that Tregs can suppress each other via the PD-1/PD-L1 interaction (data not shown). Cumulative experiments with Tregs from a total of 8 HCV patients showed that PD-L1 blockade resulted in a significant increase in IL-2–dependent, HCV-specific Treg proliferation compared with treatment with isotype control (Figure (Figure6C).6C). Similar results were obtained by using intrahepatic Tregs purified from a pool of IHLs that in turn had been isolated from 3 independent liver biopsies (data not shown). To investigate whether the inhibitory PD-1 effect on Treg proliferation accounts for the inverse correlation we observed between Foxp3+ and Ki67+ Tregs in vivo (Figure (Figure3,3, D and E), we sorted PD-1+ and PD-1– Tregs from purified CD4+CD25+Foxp3+ cell populations. Under conditions in which cells were stimulated for 6 d with anti-CD3/CD28 and IL-2 in the presence of anti–PD-L1, a considerable increase in Ki67+ cells was shown in the PD-1+ Treg population, but not in the PD-1– Treg population (Supplemental Figure 4). Control experiments revealed that the anti–PD-1 mAb used for cell sorting in this as well as the following experiments did not interfere with the PD-1/PD-L1 interaction. Indeed, proliferation of CFSE-stained Foxp3+ cells in response to anti-CD3/CD28 and IL-2 was not affected by the addition of anti–PD-1 (data not shown). These data suggest that this anti–PD-1 antibody recognizes a sequence that seems not involved in the PD-1/PD-L1 interaction. On the other hand, the evidence that B7.1 was overexpressed, although at a level consistently less than that of PD-1, on both fresh intrahepatic Tregs and in vitro–activated Tregs (Supplemental Figure 5, A and B) led us to investigate whether the inhibitory effect by PD-L1 on Treg proliferation might be mediated, at least in part, by interaction with B7.1 (55). Consistent with this possibility, blocking of PD-L1/B7.1 interaction with a soluble form of CTL-associated antigen 4 (CTLA-4) via the use of a fusion CTLA-4/Ig protein that binds B7.1 with high affinity (56) improved Treg proliferation in response to anti-CD3/CD28 and IL-2 in vitro (Supplemental Figure 6, A and B). However, anti–PD-L1 improved Treg proliferation consistently more than did CTLA-4/Ig (Supplemental Figure 6, A and B).
Next we investigated whether the enhanced expansion of Tregs by PD-L1 blockade correlated with the increased suppression function. Under conditions in which purified CFSE-labeled CD4+CD25+ Tregs were stimulated with anti-CD3/CD28 in the presence of IL-2 and the mAb to PD-L1 or isotype control for 6 d of culture, PD-L1 blockade was confirmed to upregulate the Foxp3+ cell proliferation, compared with the Tregs stimulated in the presence of the isotype control (Figure (Figure7A).7A). Importantly, a marked suppression of the CFSE-labeled Tresp proliferation in response to anti-CD3/CD28 was highlighted when the cells were cocultured with Tregs demonstrating substantial expansion of Foxp3+ cells after previous stimulation in the presence of anti–PD-L1 mAb compared with suppression by Tregs previously stimulated in the absence of anti–PD-L1 mAb (Figure (Figure7,7, A and B). On the contrary, sorted, fresh PD-1+ or PD-1– Foxp3+ Tregs without previous stimulation (Supplemental Figure 7A) efficiently suppressed proliferation in a similar fashion, and the PD-L1 blockade did not increase the intrinsic suppression capacity of PD-1+ Tregs (Supplemental Figure 7B).
After determining that the PD-L1 blockade enhanced IL-2–dependent expansion of Tregs, we analyzed the effect of PD-L1 on STAT-5 phosphorylation, known to be implicated in IL-2R signaling (57, 58) and proposed to play a key role in Treg homeostasis (59, 60). STAT-5 appeared to be spontaneously phosphorylated in a notable proportion of all cell populations expressing CD25, significantly more so in the CD25hi than in the CD25lo/– cell subset (Figure (Figure8,8, A and B). More detailed analyses of CD25+ populations showed that pSTAT-5+ cells were more highly represented in Foxp3+ Tregs than in Foxp3– Teffs, particularly in those expressing PD-1 (Figure (Figure8,8, A and B). In addition, pSTAT-5 was more highly expressed in all populations infiltrating the inflamed liver (Figure (Figure8,8, A and B) than in those derived from the peripheral blood (Figure (Figure8A8A and Supplemental Figure 8). To verify that homeostatic interplay between PD-1 and pSTAT-5 expression takes place in vivo, we performed longitudinal analyses in 2 selected patients showing a severe hepatitis reactivation that is generally considered a rare event during chronic HCV infection (61). Interestingly, we observed a peak of both Treg expansion and pSTAT-5 intensity, which followed the raised levels of serum alanine aminotransferase and frequencies of T cells promptly producing IL-2 ex vivo (i.e., disease activity) and preceded a dramatic contraction of disease activity (Figure (Figure8C).8C). Subsequently, both Treg expansion and pSTAT-5 intensity decreased in relation with the upregulation of PD-1, which anticipated the establishment of steady disease activity (Figure (Figure8C).8C). Similarly, PD-1 and pSTAT-5 were overexpressed in Teffs — albeit to a lower extent than in Tregs — by the same patients (data not shown), evidence in support of the parallelism of the Teff and Treg expansions. No correlation between the parameters above and the serum HCV-RNA levels in the kinetic studies was shown. Notably, PD-L1 inhibited STAT-5 phosphorylation in fresh intrahepatic Tregs ex vivo. Indeed, upon 6 h stimulation with anti-CD3/CD28 and IL-2 in the presence of anti–PD-L1 mAb, both pSTAT-5+ cell expansion and MFI increased to a greater extent in Tregs than in Teffs, and to a greater extent in HCV-IHLs than in HCV-PBLs (Figure (Figure9A,9A, Supplemental Figure 9A, and data not shown) or HD-PBLs (data not shown). Amazingly, pSTAT-5 upregulation was also observed in the PD-1– cell populations, albeit to a lesser degree than in the PD-1+ cell populations (Figure (Figure9A9A and Supplemental Figure 9A). Therefore, we investigated whether this phenomenon was caused by increased capacity of PD-1+ cells to provide bystander signals to the PD-1– cells after stimulation of the total IHLs or PBLs (containing both PD-1+ and PD-1– cells) in the presence of blocking anti–PD-L1 antibody. This hypothesis was ascertained by experiments in which we used fresh PD-1+ or PD-1– Foxp3+ Tregs (Figure (Figure9B9B and Supplemental Figure 9B) sorted from HD-PBLs and HCV-PBLs. Under these conditions, pSTAT-5 was upregulated in Foxp3+ Tregs to a much greater extent than in Foxp3– Teffs, and the addition of anti–PD-L1 mAb caused a consistent increase of pSTAT-5 upregulation in the PD-1+ Tregs (Figure (Figure9B),9B), but not in the PD-1– cells (Supplemental Figure 9). Control experiments revealed that the levels of nonphosphorylated STAT-5, constitutively expressed by all cells, did not change in this experimental setting (data not shown). Importantly, kinetics experiments demonstrated that after IL-2 contact, STAT-5 phosphorylation was notably more rapid, but less vigorous and sustained, in Tresp than in Tregs (Supplemental Figures 10 and 11). In particular, CD4+ Tresps remained both CD25– and pSTAT-5– after 30 minutes of stimulation with anti-CD3/CD28 (Supplemental Figure 10). However, after 1 h of IL-2 culture, the CD25+ cell population demonstrated prompt upregulation of STAT-5 that decreased with time. Conversely, the CD4+CD25hiFoxp3+ cells, but not the CD25loFoxp3– or the tiny CD25loFoxp3+ populations, displayed pSTAT-5 later, but to a higher extent and more sustained over time, compared with Tresps (Supplemental Figure 11). In addition, the PD-L1 blockade was confirmed to improve pSTAT-5 more so in PD-1+ Tregs than in PD-1+ Teffs (data not shown).
Our findings suggest that PD-1 expression correlates with the regulation of expansion and suppression capacity of Tregs during chronic HCV infection. We first observed substantial Treg accumulation in the inflamed liver of HCV patients and that some of this accumulation was caused by increased proliferation in vivo, as calculated by Ki67 expression. It will be interesting to study whether Treg proliferation can initiate directly at the site of infection or whether it occurs in the draining lymph nodes, causing divided cells to migrate into the inflamed tissue. Moreover, accumulation and proliferation of Tregs were synchronized with those of Teffs at the site of infection, but at a level significantly and consistently lower than the parallel expansion of Teff IHLs. Nonetheless, intrahepatic Tregs were intrinsically professional, because they exerted efficient suppression function that was directly related with the level of Foxp3 expression, the decline of proliferating (Ki67+) T cells, and the clinical consequences of downregulation of antiviral and immunopathological responses (i.e., increased viral load and decreased HAI score, respectively) in vivo. Because we were able to demonstrate notable production of IL-2 by Teffs derived from inflamed liver, which is known to be the most important source of HCV antigens, the capacity of Tregs to proliferate in vivo is likely caused by the sustained amounts of viral antigens and paracrine IL-2 at the site of infection. The results of our antigen-specific Treg proliferation assay in vitro supported this view because the combination of viral antigens and IL-2 efficiently rescued Tregs from anergy in the majority of HCV patients tested, indicating a key role for HCV in the generation of functional Tregs. Further analyses, such as those using class II tetramers (62), are needed to determine whether most Tregs are antigen specific or expand in response to non–antigen-specific stimuli such as inflammation-, necrotic cell–, or virus-derived products.
To understand why the expansion and proliferation of intrahepatic Tregs were, although parallel, constantly less than those of intrahepatic Teffs in all HCV patients tested, we investigated the possibility that Tregs are more susceptible than the effectors to the PD-1–dependent exhaustion mechanism in the inflamed tissue (33, 34). The lower Treg expansion correlated with significantly higher expression of PD-1 on Tregs than on Teffs infiltrating the inflamed liver. Further in vivo experiments revealed that the PD-1/Foxp3 cell ratio correlated inversely with the viral load and directly with the HAI score, which suggests that PD-1 expression on Tregs contraregulates the suppression of both the anti-viral and the immunopathological responses. The PD-L1 blockade significantly increased both the expansion and the relative suppression function of Tregs when stimulated with either specific or nonspecific TCR ligands plus paracrine IL-2 in the presence of a blocking mAb to PD-L1 in vitro, thus providing evidence for the hypothesis that the inhibition of suppression is PD-1 dependent. The potential role of PD-1–dependent control of Treg proliferation is emphasized by the observation that PD-L1 blockade (by anti–PD-L1) increased the expansion of Ki67+ cells in the Treg population upon TCR-dependent stimulation in the presence of IL-2 in vitro. This finding is consistent with the inverse correlation observed between Foxp3+ and Ki67+ Tregs in vivo. Importantly, the interaction of PD-L1 with both PD-1 and B7.1 was involved in controlling Treg proliferation in our system in vitro, supporting the recent identification of human B7.1, but not B7.2, as an additional inhibitory partner for PD-L1 (55). However, blocking with anti–PD-L1 improved Treg proliferation consistently more than did blocking with CTLA-4/Ig, likely because PD-L1 interacts with PD-1 with both higher affinity (55) and higher avidity than it does with B7.1, as indicated by our observation that both in vivo– and in vitro–activated Tregs expressed higher levels of PD-1 than B7.1. Nevertheless, the PD-L1–mediated inhibition of Treg function is dependent on its capacity to decrease expansion, and ultimately the number of Tregs, rather than the intrinsic Treg suppression capacity. This finding is consistent with both the previous finding that Foxp3 expression dictates and fixes the suppression function of the single Tregs regardless of the presence of inhibitory signals (18, 19) and the hypothesis that decreased expansion of Tregs by PD-1 can decrease suppression function in a given inflammatory context. In contrast, the lack of evident correlation when we considered the peripheral Tregs further demonstrates that peripheral immune settings do not necessarily reflect the local immune responses present at a given inflammatory site (51, 63–67).
Tregs overexpress the heterotrimeric IL-2R, consisting of the α (CD25), β, and common γ chains. These chains induce different signaling pathways, including the activation of JAK-3 and the resulting phosphorylation of STAT-5, which then dimerizes, translocates to the nucleus to activate gene transcription, and supports cellular survival and proliferation (57, 58, 68). Herein, we demonstrated that pSTAT-5 was spontaneously overexpressed in intrahepatic Tregs, to a significantly greater extent than in intrahepatic Teffs, and paralleled both CD25hi and PD-1 expression in vivo. In addition, kinetic studies in patients with severe hepatitis reactivation showed that pSTAT-5 and PD-1 expression in Tregs were upregulated in an alternate fashion in vivo: pSTAT-5 upregulation followed the peak and anticipated the dramatic suppression of disease activity, whereas PD-1 upregulation preceded the return of both pSTAT-5 and disease activity to steady levels. Finally, PD-L1 blockade increased both proliferation and pSTAT-5 expression by fresh intrahepatic Tregs significantly more than by intrahepatic Teffs ex vivo. Taken together, these data support the hypothesis that PD-1 controls Treg proliferation by limiting STAT-5 phosphorylation that is likely caused by the capture (via the highly expressed IL-2R) of high amounts of paracrine IL-2 produced by CD25lo Teffs at the site of infection (22–24). The findings that PD-1 is consistently more expressed than B7.1 and that anti–PD-L1 enhanced Treg proliferation more so than CTLA-4/Ig led us to hypothesize that the pSTAT-5 control is likely the result of the PD-L1/PD-1 more than the PD-L1/B7.1 interaction. The potential existence of a homeostatic loop, whereby the upregulation of pSTAT-5 in IL-2–consuming Tregs causes the upregulation of PD-1 that in turn controls STAT-5 phosphorylation via the activation of the SHP2 (33, 58), remains an interesting and relevant issue that requires further study. Our results are consistent with the recent finding supporting dominant IL-2 adsorption by Tregs as a critical mechanism of suppression that results in IL-2 deprivation–induced apoptosis of Teffs (69). In this context, the inhibitory PD-1 effect on STAT-5 phosphorylation of Tregs may play a key role in controlling the suppression caused by IL-2 deprivation by limiting the expansion of IL-2–capturing Tregs. As a consequence, intrahepatic Teffs can escape from excessive expansion of Tregs, which have been recently suggested to exert IL-2 deprivation via cell-cell contact (69). However, our model does not exclude additional mechanisms by which the PD-L1–dependent pathway may interfere with Tregs by contrabalancing various mediators of Treg suppression (1–11, 70–72). Kinetics experiments showed that STAT-5 phosphorylation after IL-2 contact in vitro was notably more rapid in Tresps than in Tregs, whereas it increased and was sustained over time in Tregs more than in Tresps in the later phases. The interesting issue of whether Foxp3-induced anergy can explain the requirement of more sustained IL-2 contact to initiate the STAT-5 phosphorylation cascade in Tregs may be the subject of further study. However, once appropriately rescued from anergy (i.e., via the capture of paracrine IL-2 in an inflamed tissue), CD25hi Tregs can capture higher IL-2 amounts than can CD25lo Teffs and induce more vigorous STAT-5 phosphorylation. In this scenario, the homeostasis between Treg and Teff expansion would be maintained by the higher expression of PD-1 on Tregs, which would prohibit exaggerated suppression of Teffs.
In conclusion, our results indicate that PD-L1–mediated inhibition limits the expansion of Tregs by controlling STAT-5 phosphorylation, thus providing a mechanism — which has not to our knowledge been described previously — by which Treg function can be contraregulated. Persisting pathogens are usually incompletely controlled by Teffs in primary infected individuals because of their capacities to generate escape mutants, to exert immunosuppressive effects, to induce Tregs, or to cause PD-1–dependent Teff exhaustion (32, 48, 51, 73–78). Our data suggest that under these circumstances, the balance between suppression (Foxp3) and contrasuppression (PD-1) may fine-tune Tregs in order to minimize the harmful responses without completely switching off those intended to limit excessive viral spread, and ultimately to maintain a long-lasting host/pathogen relationship. Studies are in progress to determine whether, on the other hand, predominant PD-1 contrasuppression of Tregs plays a key role in the degeneration leading to autoimmunity. Nevertheless, the inhibitory role of PD-1 in Tregs must be taken into account when targeting PD-1 or PD-L1 for therapeutic rationale, because such a strategy might result in boosting not only Teff responses, but also Treg-mediated immune suppression, which may have an advantageous role by inhibiting unwanted (e.g., autoreactive) responses.
We studied 65 HCV patients (Supplemental Table 1) and 30 HDs (age, 21–48 years; 21 males, 9 females) according to the ethical guidelines of the 1975 Declaration of Helsinki and a priori approval by the Institutional Review Boards of Sapienza Università di Roma and IRCCS Policlinico San Matteo di Pavia. In addition, 2 patients with chronic HCV infection (genotype 2c) showing a severe hepatitis reactivation were followed for a period of at least 50 d without receiving any treatment. Serum HCV-RNA levels and HCV genotype were determined by Versant HCV RNA 3.0 bDNA and INNO-LiPA, HCV II, respectively (Bayer Corp.). None of the patients were being treated with IFN-α with or without ribavirin, alternative antiviral therapies, or immunosuppressive therapies at the time of the study. Of the 65 HCV patients, 57 underwent percutaneous needle liver biopsy, part of which was processed for diagnostic and part for immunological analyses. Histology was graded as HAI, calculated as follows: periportal necrosis, intralobular necrosis, and portal inflammation were scored with values ranging 0–10, 0–4, and 0–4, respectively, and their sum provided the grading or inflammatory index; fibrosis was scored with a value ranging 0–4, providing the staging or fibrosis index; and the total score was calculated as grading plus staging (53).
PE-cyanine 7–conjugated anti-CD4, peridinin-chlorophyll-protein complex–conjugated anti-CD25, and FITC-conjugated goat anti-rabbit IgG were from Caltag Laboratories. Allophycocyanin-conjugated anti-Foxp3 and PE-conjugated anti–PD-1 were from eBioscence. PE-conjugated anti–IL-2, FITC-conjugated anti-Ki67, FITC-conjugated anti-B7.1, control isotype–matched mAbs, and BD Cytofix/Cytoperm Kit were from BD Biosciences — Pharmingen. CFSE was from Invitrogen. Purified anti–PD-L1 mAb was from R&D Systems. Purified polyclonal rabbit anti–pSTAT-5 was from Cell Signalling. Dynabeads CD3/CD28 T Cell Expander (anti-CD3/CD28) was from Dynal Biotech. CD4+CD25+ Regulatory T Cell Isolation Kit was from Miltenyi Biotec. HCV-core and HCV-NS3 antigens were from ViroGen Co. and Biodesign International Inc., respectively. Brefeldin-A was from Sigma-Aldrich. Recombinant IL-2 was from Chiron. Cells producing the recombinant fusion protein constituted by CTLA-4 and human Ig were donated by F. Sallusto (Institute for Research in Biomedicine, Bellinzona, Switzerland; ref. 56). Complete medium was RPMI 1640 medium containing 5% fetal calf serum, 2 mM glutamine, 1% nonessential amino acids, 1% sodium pyruvate, and 50 μg/ml kanamycin from Invitrogen.
PBLs and IHLs from liver biopsies were isolated as previously described (51, 73, 79). CD4+CD25+ T cells were purified from PBLs using the CD4+CD25+ Regulatory T Cell Isolation Kit (Miltenyi Biotec). Briefly, non-CD4+ cells were depleted by separation over a MACS column (Miltenyi Biotec) after labeling with a cocktail of biotin-conjugated mAbs (anti-CD8, -CD14, -CD16, -CD19, -CD36, -CD123, -TCRγ/δ, and –Glycophorin A) and anti-Biotin MicroBeads (Miltenyi Biotec). Then, the enriched CD4+ T cell fraction was conjugated with CD25 MicroBeads and separated using a MACS column. The unlabeled CD4+CD25– T cell fraction was not retained in the column and obtained by collecting the effluent. Magnetically labeled CD4+CD25+ T cells were retained on the column and eluted after removal of the column from the magnetic field. Flow cytometry analysis using a FACSCanto flow cytometer (BD) and FACSDiva analysis software (version 5.0.3; BD) demonstrated 95%–99% and less than 5% CD4+CD25+ T cells in the labeled and unlabeled fraction, respectively. In addition, purified CD4+CD25+ cells were stained with PE-conjugated anti–PD-1 and then processed with FACSAria (BD) to sort CD4+CD25+PD-1+ and CD4+CD25+PD-1– cells. Greater than 90% PD-1+ cells were found in the positively sorted PD-1+CD4+CD25+ cell population, and less than 3% were found in the PD-1–CD4+CD25+ cells.
The combination of differently labeled mAbs to surface molecules described above was used to stain PBLs or IHLs (51, 76, 77). After staining, cells were washed, fixed, and permeabilized using Cytofix/Cytoperm solution (BD Biosciences) at 4°C for 20 min, rewashed with Perm Wash Buffer (BD Biosciences), and stained at intracellular level with labeled mAbs to Foxp3 and Ki67 for 30 min at 4°C. When stained with unlabeled specific antibody to detect intracellular pSTAT-5, cells were washed and stained with the appropriate secondary labeled antibody. In selected experiments, cells that had been previously double stained with labeled mAbs to CD4 and CD25 were stimulated or not with anti-CD3/CD28 in the presence or absence of IL-2 (50 or 100 U/ml) for 6 h at 37°C. At the second hour, 5 μg/ml brefeldin-A was added. Cells were washed, fixed, permeabilized as described above, and stained at the intracellular level with labeled mAbs to Foxp3 and IL-2 for 30 min at 4°C. Negative controls were obtained by staining cells with irrelevant isotype-matched mAbs in each experiment. Stained cells were acquired with a FACSCanto flow cytometer and analyzed using FACSDiva software (BD).
Purified peripheral Tregs or Tresps, as well as sorted PD-1+ or PD-1– Treg populations, were labeled with 1 μM CFSE for 10 min at 37°C, stopped with serum for 10 min at 37°C, washed with complete medium, and cultured alone (5 × 104) or cocultured at a 1:1 cell ratio in a 96-well plate in the presence or absence of different combinations of HCV-Ags/APCs, anti-CD3/CD28, 50 U/ml IL-2, blocking anti–PD-L1, CTLA-4/Ig, or isotype control mAb. Alternatively, Tregs and Tresps were cocultured in separated 96-transwell plates (Corning Inc.) and independently stimulated as described above. In selected experiments, in which we were able to obtain about 5 × 104 intrahepatic Tregs from pools of 4 liver biopsies, these cells were cultured alone or cocultured with CFSE-stained peripheral Tresps (isolated from 6 randomly selected HCV patients) at a 1:1 cell ratio as described above. After 6 d stimulation, cells were stained with labeled mAbs to CD4, CD25, and Foxp3. In some experiments, PD-1+ and PD-1– Treg populations were stimulated or not for 6 d with anti-CD3/CD28 and IL-2, in the presence or absence of anti–PD-L1, and stained with the labeled mAbs to CD4, CD25, PD-1, Foxp3, and Ki57 described above. Finally, cells were analyzed with a FACSCanto flow cytometer (BD).
All statistical analyses were performed with both SPSS (version 11; SPSS Inc.) and Prism (version 4; GraphPad) software using Pearson’s correlation test and nonparametric Mann-Whitney U test. Differences were considered significant at P < 0.05.
This work was supported by European Union financial supports; by Fondo per gli Investimenti della Ricerca di Base and Programmi di Ricerca scientifica di rilevante Interesse Nazionale, Ministero dell’Istruzione, dell’Università e della Ricerca projects; by Ministero della Sanità-Istituto Superiore di Sanità (Progetti AIDS); by Associazione Italiana per la Ricerca sul Cancro; by Fondazione Italiana Sclerosi Multipla; and by Contributo Regione Lazio per la realizzazione di opere strutturali ed iniziative sociali, culturali e sportive di carattere locale 2006.
Conflict of interest: The authors have declared that no conflict of interest exists.
Nonstandard abbreviations used: CTLA-4, CTL-associated antigen 4; HAI, histological activity index; HD, healthy donor; IHL, intrahepatic lymphocyte; IL-2R, IL-2 receptor; PD-1, programmed death–1; PD-L1, programmed death–1 ligand 1; Teff, effector T cell; Tresp, responder T cell.
Citation for this article: J. Clin. Invest. 119:551–564 (2009). doi:10.1172/JCI36604
Vittorio Francavilla’s present address is: Centro Nazionale AIDS, Istituto Superiore di Sanità, Rome, Italy.
See the related Commentary beginning on page 450.