Effector T cell function rather than survival determines extent and duration of hepatitis in mice
Jnl of Hepatology June 2016
Michelle Vo1,2, Lauren E. Holz1,2,3, Yik Chun Wong1,2, Kieran English1,2, Volker Benseler1,2,4, Claire McGuffog1,2, Miyuki Azuma5, Geoffrey W. McCaughan2,6, David G. Bowen1,2,⇑,y, Patrick Bertolino1,2,⇑,y
1Liver Immunology Program, Centenary Institute, Newtown, NSW, Australia; 2AW Morrow Gastroenterology and Liver Centre,
Royal Prince Alfred Hospital Newtown, NSW, and Faculty of Medicine, University of Sydney, NSW, Australia; 3Current address: Department of
Microbiology and Immunology, The Peter Doherty Institute, The University of Melbourne, Parkville, VIC, Australia; 4Current address:
Department of Surgery, University of Regensburg, Bavaria, Germany; 5Department of Molecular Immunology Graduate School, Tokyo Medical
and Dental University, Yushima, Tokyo, Japan; 6Liver Injury and Cancer Program, Centenary Institute, Newtown, NSW, Australia
See Editorial, pages 1208–1210
Background & Aims
Acute hepatitis is often mediated by cytotoxic T lymphocytes (CTLs); however, the intrinsic parameters that limit CTL-mediated liver injury are not well understood.
Methods
To investigate whether acute liver damage is limited by molecules that decrease the lifespan or effector function of CTLs, we used a well-characterized transgenic (Tg) mouse model in which acute liver damage develops upon transfer of T cell receptor (TCR) Tg CD8 T cells. Recipient Tg mice received donor TCR Tg T cells deficient for either the pro-apoptotic molecule Bim, which regulates CTL survival, or suppressor of cytokine signaling-1 (SOCS-1), which controls expression of common gamma chain cytokines; the effects of anti-PD-L1 neutralizing antibodies were also assessed.
Results
Use of Bim-deficient donor T cells and/or PD-L1 blockade increased the number of intrahepatic T cells without affecting the degree and kinetic of acute hepatitis. In contrast, SOCS-1-deficient T cells induced a heightened, prolonged acute hepatitis caused by their enhanced cytotoxic function and increased expansion. Although they inflicted more severe acute liver damage, SOCS-1-deficient T cells never precipitated chronic hepatitis and became exhausted.
Conclusions
The degree of acute hepatitis is regulated by the function of CD8 T cells, but is not affected by changes in CTL lifespan. Although manipulation of the examined parameters affected acute hepatitis, persistent hepatitis did not ensue, indicating that, in the presence of high intrahepatic antigen load, changes in these factors in isolation were not sufficient to prevent T cell exhaustion and mediate progression to chronic hepatitis.
Introduction
Acute liver injury has a variety of causes, but is often initiated by an immune response triggered by hepatotropic viruses and other pathogens targeting this organ. A robust and sustained immune response involving both innate and adaptive immunity is critical to clear infections by viruses targeting the liver, such as the hepatitis B (HBV) [1] and C viruses (HCV) [2]. Antiviral cytotoxic CD8 T lymphocytes (CTLs) play a crucial role in mediating liver injury during infection, and the induction and maintenance of broad vigorous virus-specific CD8 T cell responses is a positive predictor of spontaneous resolution of HCV infection [[3], [4], [5]]. Antibody-dependent depletion of CD8 T cells in a chimpanzee model of HCV infection led to prolonged viral infection [6]. The detection of CTLs in the periphery coincides with raised alanine transaminase (ALT) levels and manifests clinically as acute hepatitis [7]. Clearance of HCV is more common in symptomatic patients [5], and it is likely that the degree of liver damage that ensues during the acute phase of infection is critical in influencing the outcome of infection [8].
Although it is known that CTLs kill target cells via granzyme and/or Fas/FasL interactions (classical CTL killing pathway), or via the secretion of hepatotoxic cytokines (bystander killing), the molecular processes limiting the action of CTLs and responsible for terminating acute liver damage mediated by CD8 T cells are not well understood. It is important to elucidate these processes to derive strategies to increase the chances of clearing persistent hepatotropic infections, and also to prevent too sustained or vigorous CD8 T cell responses that might lead to eventual development of significant fibrosis and cirrhosis, or in the more acute setting, to the development of acute liver failure. Several regulatory processes intrinsic to T cells could be important in limiting liver damage, including: induction of inhibitory receptors such as PD-1, Tim-3, LAG3, and CTLA-4; regulation of CTL effector function (downregulation of cytokine and effector molecules); and apoptosis of CTLs. Other parameters such as the number of CD8 T cell precursors specific for the virus [[9], [10]], the affinity of the TCR recognizing viral epitopes [11], the availability of CD4 T cell help [[12], [13]], and the number of hepatocytes expressing antigen [14] have also been shown to influence the long-term CD8 T cell outcome.
In order to identify the T cell intrinsic parameters controlling the degree and kinetics of acute liver damage, we used the well-characterized Met-Kb transgenic (Tg) mouse model of immune-mediated hepatitis, in which acute liver injury is induced by TCR Tg CD8 T cells that are first activated in lymph nodes (LN) and subsequently recognize their cognate antigen in the liver, leading to hepatocellular injury [15].
By transferring donor Tg T cells deficient for genes that control T cell death (Bim) or regulate signaling of cytokines critical for effector T cell function (suppressor of cytokine signaling-1; SOCS-1), or by treating recipient mice with anti-PD-L1 blocking antibodies, we investigated the roles of regulation of T cell death, regulation of CTL function, and T cell inhibitory molecules in limiting the degree and duration of liver damage, independently of TCR affinity, T cell help and antigen dose.
Our results suggest that strategies that enhance T cell survival promoted accumulation of CD8 T cells recognizing hepatocyte-expressed antigens in the liver, without altering the severity or tempo of liver damage. In contrast, augmenting the function of effector T cells prolonged acute hepatitis and increased the severity of liver damage. While these strategies had different effects on the fate and/or function of effector CD8 T cells, none prevented the resolution of acute hepatitis associated with the eventual development of functional exhaustion of T cells within in the liver.
In the setting of high-level persistent antigen expression in the liver, the degree and duration of liver damage is thus predominantly regulated by genes controlling the function of effector cells rather than those affecting T cell lifespan. Persistence of liver damage is ultimately limited by functional T cell exhaustion. These data suggest that, to be effective, immunotherapies aimed at boosting the number of CTLs in patients with liver disease will need to be combined with strategies that enhance T cell effector function and, more importantly, interfere with T cell exhaustion.
Excerpts
Blocking PD-1/PD-L1 interactions increased T cell numbers without augmenting the severity of acute liver damage or leading to chronic hepatitis
PD-1/PD-L1 blockade is currently used in the clinic to boost the function of exhausted CD8 T cells in a range of solid tumors, has been explored in chronic HCV infection [25], and is under investigation for treatment of hepatocellular carcinoma (HCC) [26]. It was therefore important to test the role of this molecule in our model.
PD-1 was expressed at similar low levels on donor Des RAG−/− T cells following 24 h of activation in the liver and LN of Met-Kb mice, but was not expressed at any stage on T cells transferred into B10.BR mice (Fig. 2A). At 48 h post-activation, liver-activated CD8 T cells expressed more PD-1 compared to LN-activated T cells (p <0.05) (Fig. 2A, B). Similar levels of expression were detected on liver- and LN-activated CD8 T cells less than a day later, probably due to death of liver-activated T cells and recirculation of LN-activated T cells to the liver [[18], [27]]. Over time, all Des RAG−/− T cells activated in Met-Kb mice became PD-1high and expression tended to increase with time (Fig. 2B), a result consistent with the phenotype of intrahepatic CD8 T cells reported in different models when antigen persists [[28], [29]]. PD-L1, the main PD-1 ligand, was also expressed at very low levels in the hepatic parenchyma of Met-Kb mice (Fig. 2C), but not in B10.BR mice (data not shown) at day 15 post-T cell transfer. Interestingly, PD-L1 expression was positively correlated with the number of infiltrating cells, and was increased in association with the more marked infiltrates of Des Bim−/− T cells. In these livers, PD-L1 expression was found in the vicinity of infiltrating Des Bim−/− T cells in the lobules and also in portal regions (Fig. 2C).
To examine the possibility that PD-1/PD-L1 interactions contributed to the induction of tolerance in liver-activated T cells and/or to the attenuation of acute hepatitis, Met-Kb and B10.BR mice, into which wild-type Des T cells had been transferred, were treated with neutralizing anti-PD-L1 or control antibodies (rat IgG). Treatment of Met-Kb mice with anti-PD-L1 every three days, beginning from day one after adoptive transfer of T cells, led to increased numbers of donor Des T cells in the liver at day 15, in comparison to control rat IgG-treated mice (Fig. 2D), suggesting that PD-L1 blockade increased T cell proliferation and/or survival. Despite accumulation of the Des T cell progeny in the liver, treatment with anti-PD-L1 did not have a significant effect on serum ALT levels (Fig. 2E). Interestingly, commencing administration of the anti-PD-L1 antibody one day earlier (day 0) decreased the number of activated T cells and inhibited hepatitis (data not shown), suggesting anti-PD-L1 interfered with T cell priming. Consistent with the failure of anti-PD-L1 to change the outcome of acute hepatitis, the phenotype of hepatocyte-activated T cells was not affected by anti-PD-L1 when treatment started at day 1, as all cells proliferated at the same rate, maintained low CD25 expression and did not produce IL-2 or IFN-γ (Supplementary Fig. 3). PD-1 expression was slightly decreased on hepatocyte-activated T cells isolated from the anti-PD-L1-treated mice in comparison to control mice, but its expression was high in both groups (Supplementary Fig. 3), providing further evidence that PD-1 expression was not sufficient to control the lack of effector function in these cells. Anti-PD-L1 treatment from day 1 post-transfer did not alter the naïve phenotype of donor T cells transferred into B10.BR mice (data not shown).
To further test whether PD-1/PD-L1 interactions prevented Bim-deficient Des cells from mediating hepatitis, Met-Kb mice into which Bim-deficient Des T cells had been adoptively transferred were treated with anti-PD-L1 from day 1. The phenotype of the donor T cells was not influenced by the expression of Bim or antibody treatment at day 2 post-transfer, as all Bim-deficient Des T cells transferred into B10.BR mice maintained a naïve CFSEhigh CD44low PD-1low CD25low phenotype, whereas all hepatocyte-activated Des T cells in Met-Kb recipients proliferated and expressed low levels of CD25 on their surface (Supplementary Fig. 4). Anti-PD-L1 treatment beginning at day 1 post-transfer did not decrease PD-1 expression on hepatocyte-activated Bim-deficient Des T cells (Supplementary Fig. 4). Blocking PD-1/PD-L1 interactions also induced a significant increase in donor Bim-deficient Des T cell numbers at day 15 in comparison to Met-Kb mice treated with control rat IgG (Fig. 2D), but did not interfere with the severity, kinetic, or the transient nature of acute hepatitis (Fig. 2E). In a similar manner to wild-type Des cells, a large proportion of these Bim-deficient Des T cells were found in the liver (Fig. 2D). Histological analysis of the liver correlated with T cell survival data, with large numbers of T cells detected in the portal and lobular regions of Met-Kb mice receiving Des Bim−/− T cells (data not shown). To test whether the function of Bim-deficient T cells could be restored at a late time point (when T cells were silenced), we commenced treatment of recipient mice with anti-PD-L1 or control rat IgG at day 20 post-transfer of Bim-deficient T cells, administering antibody every three days till day 45. Late treatment with anti-PD-L1 altered the cell survival and/or proliferation of donor T cells, such that greater numbers of Des Bim−/− T cells could be found in Met-Kb mice receiving anti-PD-L1 antibodies compared to those receiving control IgG (data not shown). However, anti-PD-L1 treatment of recipient mice did not enhance the CTL activity of donor Des Bim−/− cells at day 38 (Supplementary Fig. 5), and hepatitis was not triggered as assessed by serum ALT (data not shown).
Collectively, these results demonstrate that inhibition of PD-1/PD-L1 interactions promoted the proliferation and/or survival of self-reactive Bim-deficient T cells, suggesting that these two molecules increased T cell survival via distinct pathways. Most importantly, although activated Bim-deficient CD8 T cells accumulated and reached impressive numbers in the liver of mice treated with anti-PD-L1, their function was not restored, suggesting that, in a persisting high antigen load setting, anti-PD-L1 blockade was unable to overcome the robust silencing imposed on T cells within this organ, even when the lifespan of liver-reactive T cells was prolonged.
Discussion
By examining the ability of Tg Des T cells deficient for Bim or SOCS-1 to induce hepatitis and the role of PD-1/PD-L1 interactions in a well-characterized mouse model of acute immune-mediated hepatitis, this study demonstrated that the regulation of effector function was more critical than the regulation of the lifespan of effector CD8 T cells in limiting the tempo and degree of acute liver damage. However none of these factors were able to interfere with the dominant exhaustion process occurring in the liver in the presence of an ongoing high antigen load, and hence chronic hepatitis did not ensue.
Hepatocytes can be injured by many agents, including alcohol, toxins, chemicals and immune cells. Immune-mediated hepatocellular injury can be mediated by soluble factors such as inflammatory cytokines, including IFN-γ and TNF-α, that kill hepatocytes in a bystander manner [41], or by direct CTL killing [42]. CTLs kill their targets in an antigen-specific manner by delivering lethal granzymes to the target cells [43] and/or by Fas/FasL interactions [44]. Although the mediators of effector T cell-dependent liver damage have been well-characterized, the intrinsic T cell parameters that regulate CTL-mediated liver damage have not been previously investigated. This is an important unanswered question, as prolonged and uncontrolled and/or excessive CTL-mediated hepatocyte killing might lead to chronic or fulminant hepatitis, respectively. By using a Tg model of acute hepatitis mediated by a monoclonal population of CTLs of known specificity activated in LN, we manipulated intrinsic CD8 T cell factors that might regulate acute hepatitis, namely their susceptibility to apoptosis (via Bim), responsiveness to activation (via PD-1/PD-L1 interactions) and cytokine regulation/effector T cell function (via SOCS-1).
The transfer of CD8 Tg T cells deficient for the pro-apoptotic molecule Bim was clearly associated with prolonged survival of these liver-reactive T cells in recipient mice, suggesting that Bim was a critical regulator of T cells survival. The five fold lower intrahepatic numbers of Bim-deficient Tg T cells observed in Alb-Kb compared to Met-Kb mice suggested that the vast majority of T cells that accumulated in the liver of Met-Kb mice at 30 days post-transfer were originally activated in the LN. One of the remarkable findings of this study was that the striking accumulation of Bim-deficient Des T cells in the livers of Met-Kb mice was not sufficient to intensify or even prolong the severity and kinetic of acute hepatitis nor induce chronic liver damage. This observation extends the conclusions of our previous studies showing that Bim-deficient Tg T cells activated solely by intrahepatic bone-marrow derived cells were silenced after inducing acute liver damage [23]. These results also confirm previous reports that Bim regulates survival without altering the effector function and autoimmune potential of T cells [18].
Interestingly, inhibiting the binding of the inhibitory receptor PD-1 with its ligand PD-L1 early post-activation had a similar effect to Bim deficiency in hepatocyte-reactive CD8 T cells, resulting in increased accumulation of antigen-specific T cells in the liver without affecting the kinetic and severity of acute liver injury (Fig. 2D, E). This suggests that PD-1/PD-L1 interactions in the earlier phase post-activation are important in inhibiting T cell expansion and/or survival, rather than in dampening CD8 T cell effector function. This effect of PD-1/PD-L1 on survival seems to be distinct from the pro-apoptotic effect of Bim, as anti-PD-L1 treatment resulted in a pronounced accumulation of Bim-deficient T cells in the liver (Fig. 2D). Consistent with this, recent data suggests that PD-1 appears to promote T cell survival via induction of the transcription factor FOXO1, which regulates PD-1 expression in a positive feedback loop [45].
The more marked hepatitis observed following adoptive transfer of hepatocyte-reactive T cells deficient for SOCS-1, a molecule that regulates signaling of several cytokines binding to the common γ chain, including IL-2 [36], suggests that cytokine expression by T cells changed the effector function of T cells with an associated increase in the severity of hepatitis. The similar T cell numbers in the Met-Kb LN during the first days after transfer regardless of SOCS-1 expression (Fig. 4D) suggests that the effect of SOCS-1 on hepatitis did not occur during primary activation in LN. Instead, SOCS-1-deficient Tg T cells acquired enhanced function at day 5 after initial LN priming. This seemed to occur not only via increased expansion of CD8 T cells in the liver, probably due to their increased sensitivity to IL-2, but also by enhanced cytotoxic function. It is not clear whether this enhanced effector function was programmed during primary activation in LN or, alternatively, acquired upon secondary restimulation in the liver. However, the high expression of CD25 by SOCS-1-deficient, but not SOCS-1-sufficient, CD8 T cells in the Met-Kb recipient liver (Fig. 5B), suggests that the absence of SOCS-1 regulation extended cytokine expression in T cells and favored increased proliferation within the liver and enhanced effector function. This data is consistent with the known regulatory role of SOCS-1 [[36], [37]]. SOCS-1 gene transcription is induced in response to initial IL-2R/IL-2 binding that occurs just after T cell activation. SOCS-1 inhibits JAK/STAT signaling that lies downstream of the IL-2R [[31], [32], [33]], acting in a negative feedback loop to regulate the IL-2 signaling pathway [36]. SOCS-1 deficiency is thus thought to augment the “responsiveness” of T cells to cytokines at a later stage of the immune response when these molecules accumulate.
Although we were able to prolong effector function and alter the outcome of hepatitis by ablating SOCS-1 in donor T cells, liver damage was still controlled and did not evolve into chronic hepatitis in this model. SOCS-1 and SOCS-3 have been reported to have redundant inhibitory activities. Thus it is possible that other cytokine inhibitory mechanisms such as SOCS-3 compensate for SOCS-1 deficiency to restrict acute hepatitis. Regardless of cytokine regulation, most Tg SOCS-1−/− T cells were also deleted following early increases in survival and expansion, suggesting that like SOCS-1-sufficient Tg T cells, SOCS-1-deficient Tg T cells also died by Bim-dependent apoptosis. It would be interesting to know whether Tg T cells deficient for both SOCS-1 and Bim, and thus displaying enhanced survival as well as dysregulated effector function, would induce fulminant hepatitis or become exhausted following transfer into Met-Kb mice. Unfortunately SOCS-1−/− Bim−/− Des TCR Tg mice could not be bred despite our several attempts to generate this line.
The exact mechanisms responsible for T cell silencing after acute hepatitis in these different models remain unclear. However, our recent study showed that CTLs become exhausted in the presence of a persisting high intrahepatic antigen load [14]. Together with the results from this study showing that Bim- and SOCS-1-deficient T cells express PD-1 and Tim-3 and become functionally exhausted, it would appear that T cell exhaustion occurs in the face of ongoing high levels of intrahepatic antigen expression, regardless of enhanced T cell function or resistance to apoptosis. Based on these findings, we predict that strategies that tend to increase the survival of antigen-specific T cells would fail to interfere with robust T cell exhaustion mechanisms that occur in the liver in this setting.
This data is highly relevant for therapies that attempt to restore T cell effector function in patients chronically infected with HBV or HCV. It is thought that in these clinical settings, persisting high antigen load might play an important role in evolution towards chronic infection by promoting T cell exhaustion; similar mechanisms may also play a role in promoting resistance of hepatocellular carcinoma to tumor-specific CTL responses [46]. The relatively limited efficacy of PD-1 inhibition in HCV infection in vivo [[25], [47]], suggests that other parameters need to be taken into consideration. The compensatory roles of other inhibitory molecules, including Tim-3, TGIT, and CTLA-4, have been proposed as key to understand these observations [48]. An alternative, non-exclusive possibility suggested by our findings is that T cell responses are difficult to restore or enhance in the presence of high antigen load (e.g. large tumor burden in HCC or high levels of intrahepatic antigen expression in HBV and HCV). This study and previous results [14] would predict that inhibition of PD-1/PD-L1 interactions would be most effective in combination with other treatments that interfere with T cell exhaustion, or in the presence of lower antigen loads, such as lowered tumor burden following resection in HCC or in combination with antiviral therapy to lower viral titers in chronic HBV and HCV.
In conclusion, this study demonstrates that a highly efficient population of CTLs that secrete hepatotoxic cytokines is critical to potentiate severe acute hepatitis. Although promoting survival of CD8 T cells enhanced their clonal expansion and accumulation in the liver, it was not sufficient to amplify liver damage or mediate chronic hepatitis. This would suggest that once activated, in the presence of high levels of hepatocyte-expressed antigen, CD8 T cells have a limited window of opportunity during which effector function is mediated, regardless of their lifespan. Past this phase, if antigen is not cleared in the liver, effector CD8 T cells become exhausted. This pathway might have been selected during evolution to avoid irreversible damage to the liver and prevent death by fulminant hepatitis. The adverse effect of this protection mechanism is, however, the persistence of viruses (e.g. HBV and HCV) that chronically infect the host by promoting silencing of the antiviral T cell response.
Collectively, these results assist in our understanding of why T cells become exhausted after acute hepatitis in the presence of ongoing high-level antigen expression, and why the extent of viral antigen-specific T cell infiltrates do not always correlate with ALT levels in patients chronically infected with HBV and HCV [49]. This current study predicts that antiviral strategies that boost the priming or effector function of CTLs would be more effective than those that increase CTL survival. The combination of treatments that increase CD8 T cell survival and function and interfere with T cell exhaustion might also improve the efficacy of these therapies. 作者: StephenW 时间: 2016-6-4 16:26