Potential conflict of interest: Nothing to report.
Both hepatitis B and C viruses frequently establish chronic infection, raising the question whether T cells are poorly primed in the liver. To determine the role of different cell types in the activation of CD8+ T cells against hepatocellular antigens, we used an Adeno-associated virus to deliver ovalbumin to hepatocytes. In contrast to CD8+ T cells, CD4+ T cells were not activated. The CD8+ T cells were activated even in the absence of endogenous CD4+ T cells; however, in the liver, these cells were high in the programmed death-1 protein and low in CD127. Chimera experiments revealed that these CD8+ T cells were activated on a solid tissue cell. Conclusion: Priming of CD8+ T cells directly on nonhematopoietic cells, in the absence of CD4+ T cell help, results in suboptimal T cell activation. This could explain the impaired function of CD8+ T cells seen in chronic liver infection. (HEPATOLOGY 2010)
Most people infected with hepatitis C virus (HCV) progress to chronic infection. This is partly due to an inadequate CD8+ T cell response that lacks breadth, intensity, and CD4+ T cell help.1-3 The CD8+ T cells generated in response to HCV often display an “exhausted” phenotype expressing high levels of programmed death-1 (PD-1) and low levels of CD127.4 Inadequate immunity is also seen in hepatitis B virus, and against the liver stage of the malaria parasite. The common factor in these diseases is infection of hepatocytes, bringing up the idea that the liver environment is contributing to the development of a defective immune response. This may be due to the liver's constant exposure to endotoxin, raising the threshold for immune activation.5, 6
Multiple liver cell types may present antigens. In the mouse, the liver contains plasmacytoid and myeloid dendritic cells (DCs), as well as more unusual DC subsets7 and Kupffer cells. In addition, the liver sinusoidal endothelial cells (LSECs) and the hepatic stellate cells both have credentials as antigen-presenting cells (APCs).8-10 Hepatocytes also present antigens.11-13 This profusion of potential APCs raises the issue of which are actually important in priming immune responses against hepatocellular antigens. To clarify these issues, we used an adeno-associated virus 2 (AAV2)-based gene therapy vector (AAV2-ova) delivered by direct injection into the liver. This vector was expressed exclusively in the liver, based on reverse transcription polymerase chain reaction analysis of multiple tissues, and exclusively in hepatocytes, based on immunohistochemistry.14
Here, we examine the priming of CD8+ T cells against this AAV vector. Previous work suggested that AAV vectors did not generate cross-primed immunity that could engage transduced hepatocytes15 and that AAV could induce tolerance in CD4+ T cells.16 However, we found that CD8+ T cells were effectively primed against the same antigen. Activation of these OT-1 cells was CD4+ T cell help–independent, and was independent of bone marrow–derived APCs. The implication is that hepatocellular antigens are excluded from bone marrow–derived “professional” APCs, including DCs and macrophages, but can nevertheless engage CD8+ T cells.
AAV, adeno-associated virus; APC, antigen-presenting cell; CD, clusters of differentiation; CFSE, carboxyfluorescein succinimidyl ester; DC, dendritic cell; GFP, green fluorescent protein; HCV, hepatitis C virus; IFN, interferon; MFI, mean fluorescence intensity; MHC, major histocompatibility complex; PBS, phosphate-buffered saline; PD-1, programmed death-1; PLN, peripheral lymph node.
Materials and Methods
Male C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Gene-targeted male B6.129-H2-Ab1tm1GruN12 (major histocompatibility complex [MHC] class II−/−) mice were purchased from Taconic Farms (Germantown, NY). The OT-1 mice were on either a CD45.1/CD90.2 or CD45.2/CD90.1 background. The OT-II transgenic mice were on the CD45.2 background. B6.C-H-2bm8 (bm8) mice were a gift from L. R. Pease (Mayo Clinic, Rochester, MN). Mice were raised in a specific pathogen-free environment, were used between 8 and 12 weeks of age, and experiments were approved by our Institutional Animal Care and Use Committee.
Bone marrow chimeras were made in the strain combinations: B6→B6, B6→bm8, and bm8→B6. Donor and host differed in CD45 allotype. Eight-week-old recipients were irradiated (10 Gy) using an RS2000 x-ray irradiator (Rad Source Technologies, Coral Springs, FL). T cell–depleted bone marrow (12 × 106 cells) was injected intravenously within 6 hours of irradiation. Thirty days later, chimeras were intravenously injected with 200 μL of clodronate liposomes from Encapsula NanoSciences (Nashville, TN) to deplete radio-resistant Kupffer cells.17 Vectors were injected 2 weeks later, after the repopulation of the liver with Kupffer cells derived exclusively from the donor bone marrow.
Serotype 2 AAV vectors encoding either ova or enhanced green fluorescent protein (GFP) under the control of the cytomegalovirus promoter were obtained from the Columbus Children's Research Institute Viral Vector Core Facility (Columbus, OH).18
Intrahepatic Vector Delivery.
Mice aged 8-12 weeks were anesthetized using Avertin, and the central lobe of the liver was exposed through a 2-cm ventral midline incision. Using a 29-gauge insulin syringe, 60 μL (7.2 × 1010 deoxyribonuclease-resistant particles diluted in phosphate-buffered saline [PBS]) was slowly injected directly into the liver. The peritoneal cavity was sutured with 4-0 Vicryl (Ethicon) and the skin was closed with wound clips.
Spleen and peripheral lymph node (PLN) cells from either OT-1, OT-II, or D0.11.10 transgenic mice were depleted of red blood cells by using Lympholyte-M (Cedarlane, Ontario, Canada). Miltenyi MACS (magnetic cell sorting) kits were used to isolate CD8+ or CD4+ T cells. For in vivo proliferation, cells in PBS were stained with 4 μM carboxyfluorescein succinimidyl ester (CFSE) for 10 minutes at 37°C and washed in PBS. At 3-4 weeks after vector injection, either 1 × 106 (Fig. 1C; Fig. 5) or 5 × 106 (Fig. 1A–B,D; Fig. 2; Fig. 3; Fig. 4; Fig. 6) T cells (>90% pure) were injected intravenously in the tail vein.
Isolation of DCs.
DCs were enriched from the spleen using the technique of Livingstone,19 with modifications.20 Spleens from C57BL6 mice were digested in HBSS containing 2.4 mg/mL collagenase IV (Sigma Aldrich), and 1 mg/mL deoxyribonuclease (Sigma Aldrich) at 37°C for 30 minutes. Cells were resuspended in 60% Percoll and overlayed with 2 mL HBSS +5% fetal bovine serum. This gradient was spun at 650g for 20 minutes. DCs from the interface were allowed to attach for 90 minutes and nonadherent cells washed away. Adherent cells were incubated overnight with 1 ng/mL granulocyte-macrophage colony-stimulating factor and 1 μM SIINFEKL peptide, and harvested the next day by gentle washing. DCs (1 × 106) were given intravenously with the OT-1 cells.
Intrahepatic lymphocytes were isolated as described.14
Flow Cytometric Analysis.
Cells in staining buffer (1% fetal bovine serum in PBS) were first incubated with Fc-block (Pharmingen) for 5 minutes. Antibodies used were anti-CD62L (phycoerythrin [PE]), anti-CD44 (PE and PE-cyanin5 [Cy5]), anti-CD8 (peridinin chlorophyll protein [PCP], allophycocyanin [APC], and PE-Cy7), and anti-CD4 (PCP and Pacific Blue) all from Pharmingen. Pacific Blue–conjugated anti-CD127, anti-PD-1 (PE), anti-CD45.1 (APC and PE-Cy7), anti-CD45.2 (AlexaFluor 700), anti-CD62L (APC-AlexaFluor 750) were from eBioscience. Data were acquired using FACSCalibur or LSRII flow cytometers,21 and analyzed using FlowJo (TreeStar) on an iMac computer. Live lymphocytes were gated based on forward scatter and side scatter (FSC/SSC).
Data in the figures represent the mean ± standard error of the mean (SEM). A Student t test was used to analyze the results where applicable, and probability values of P < 0.05 were considered significant.
Activation of CD8+ but not CD4+ T Cells.
To test the capacity of the AAV2-ova vector to activate CD4+ T cells in vivo, mice received an intrahepatic injection of either AAV2-ova, or a control vector AAV2-gfp. After 3 weeks, mice were given CFSE-labeled OT-II transgenic CD4+ T cells, specific for the ISQAVHAAHAEINEAG peptide (ova323-339). These T cells did not respond, similar to OT-II T cells infused into mice that had been given the antigen-negative AAV2-gfp control vector (Fig. 1A, two upper left panels). However, the OT-II cells were competent to proliferate in vivo, revealed by their response to peptide-pulsed splenocytes (marked “pep” in Fig. 1); after this treatment, we observed divided OT-II T cells in the liver,19 spleen (“SPL” in Fig. 1) and PLN.6 To detect T cell activation, we also measured the expression of the lymph node homing receptor CD62L (Fig. 1B). Nondividing OT-II T cells maintained high expression of CD62L, whereas responding T cells expressed less. These results were confirmed using D0.11.10 transgenic CD4+ T cells, which recognize the ISQAVHAAHAEINEAG peptide in the context of the I-Ad molecule in BALB/c mice (Fig. 1C,D).
To confirm the expression of the antigen, mice that were given the AAV2-ova vector were infused with OT-1 T cells, a CD8+ T cell population specific for the SIINFEKL peptide (ova257-264). These T cells divided and down-regulated CD62L (Fig. 1B, right panels), verifying that the antigen was expressed. We conclude that AAV-2-ova vector stimulated CD8+ but not CD4+ T cell responses.
The CD8+ T Cell Response Is CD4+-Independent.
Two different T cell receptor transgenic CD4+ T cells failed to respond to AAV-OVA (Fig. 1). To test whether endogenous CD4+ T cells were helping the CD8+ cells, MHC class II–deficient mice, which lack CD4+ T cells, were given AAV2-ova vector and then OT-1 T cells. Figure 2A shows the response of OT-1 T cells, measured using CFSE, at day 3 (D3), day 5 (D5), and week 8 (W8) after adoptive transfer (shaded profiles). In the liver, OT-1 cells divided as early as day 3, and almost all of the cells were CFSE-low by day 5; these cells were also present at week 8. In control mice given the AAV2-gfp vector (nonshaded profiles), there was very little OT-1 T cell division at days 3 and 5, although the cells were subject to some loss of CFSE staining by week 8. Strikingly, there was no difference in the division of OT-1 T cells between normal B6 mice and MHC class II–deficient mice. In the spleen and the PLN, there was essentially no cell division in any of the mice at day 3, but divided cells appeared in these tissues on day 5, as previously reported14; again, there was no difference between normal and MHC class II–deficient mice.
Figure 2B shows the outcome of these experiments in terms of the numbers of OT-1 T cells in the liver (left panel of Fig. 2B), spleen (center), and PLN of normal B6 versus MHC class II–deficient mice at day 3, day 5, and week 8 after adoptive transfer. In liver, there was a statistically significant expansion of OT-1 T cells at all three time points, but no significant difference between B6 and MHC class II–deficient mice. In spleen, we observed significant clonal expansion only on day 5, but not later. Again, there was no difference between normal B6 and MHC class II–deficient mice. In PLN, we observed no clear effects on overall OT-1 T cell numbers on days 3 and 5, although there was a significant increase in the numbers of OT-1 T cells in MHC class II–deficient mice on day 5, followed by a significant loss of OT-1 T cells in the AAV2-ova–transduced mice at week 8. The overall conclusion is that MHC class II–restricted helper T cells did not influence the response of OT-1 CD8+ T cells to the AAV2-ova vector.
To determine whether the absence of MHC class II–restricted CD4+ T cell help was modifying the immune response of CD8+ T cells in more subtle ways, we evaluated the expression of the lymph node homing receptors CD62L, the global activation marker CD44, the interleukin-7 receptor alpha (IL-7Rα) (CD127), and the coinhibitory molecule PD-1. In every case, OT-1 T cells “parked” in mice transduced with the AAV2-gfp control vector served as the control. The results are represented as the mean fluorescence intensity (MFI) in groups of at least five mice (Fig. 3A). In the liver, the presence of AAV-OVA caused the down-regulation of CD62L and up-regulation of CD44 at all time points. The CD127 marker was down-regulated at day 3 and 5, but was restored by 8 weeks. The PD-1 marker was powerfully induced in the AAV-OVA mice from day 3 to week 8. None of these effects was modified in the absence of MHC class II.
To determine if this PD-1 high phenotype correlated with impaired function, we tested the ability of these cells to produce interferon-gamma (IFN-γ). Graphs in Fig. 3B,C show OT-1 cells in wild-type versus MHC II–deficient mice on day 5 (B) and week 8 (C). On day 5, OT-1 cells in both wild-type and MHC II–deficient hosts were capable of making IFN-γ in the presence of antigen. However, by week 8, these cells made less IFN-γ than those without antigen. Thus, the high expression of PD-1 correlated with loss of function.
In the spleen, the down-regulation of CD62L was clear-cut only at week 8, whereas increased CD44 was seen at day 5 and week 8. These data are consistent with our previous demonstration that the anti-AAV immune response starts in the liver, rather than in lymph nodes.14 The down-regulation of CD127 expression on OT-1 T cells in the spleen was not seen on day 3, but was present at day 5 and week 8. PD-1 was up-regulated in OT-1 T cells on day 5 and week 8, but the level of PD-1 expression was at least 10-fold less than with the OT-1 T cells in the liver; the PD-1 MFI data are shown on the same scale to emphasize this difference. None of these effects were different between normal B6 mice and MHC class II–deficient mice. Effects on OT-1 T cells in the PLN were smaller, but there was up-regulation of CD44 and PD-1 expression on day 5 and at week 8. Again, there was no effect of MHC class II–restricted help on any of these phenotypic changes. These effects on CD8+ T cell surface phenotype in B6 versus MHC class II–deficient mice agree with Fig. 2, and support the conclusion that CD4+ T helper cells are not involved in the CD8+ T cell response to AAV2-ova–transduced liver cells.
These effects of the OT-1 T cell phenotype could be summarized as follows: whereas other markers fluctuated in a similar way in both help-intact and help-deficient mice in all of the organs sampled, the expression of PD-1 was dramatically different. Its expression was very high on OT-1 T cells in the liver; however, this expression was not influenced by the presence or absence of CD4+ T cell help.
High PD-1 Due to Priming in the Liver.
Figure 3 shows that high PD-1 expression is unique to OT-1 cells in the liver. This could be due to the liver environment causing all liver CD8+ T cells to become PD-1 high, or alternatively by intrahepatic priming. We investigated this by comparing host CD8+ T cells in the liver to OT-1 cells. The majority of host CD8+ T cells in the liver are PD-1 negative (Fig. 4); however, there is a population of PD-1–positive cells. These PD-1–positive cells are also CD62L-low, indicating that they are recently activated CD8+ T cells. We have previously shown that activated CD8+ T cells are trapped in the liver in a TLR4-dependent manner.20 However, we cannot assume that the host liver PD-1 high cells were trapped in this manner.
To test the hypothesis that activation in the liver up-regulates PD-1 on CD8+ T cells, we compared OT-1 cells activated by AAV-OVA in the liver and OT-1 cells activated in primary lymphoid tissues by SIINFEKL-pulsed DCs. Figure 5 shows that PD-1 expression is an indication of activation, because this molecule is expressed on OT-1 cells both in the liver of AAV-OVA–transduced mice, and in liver and lymphoid organs of DC-SIINFEKL–stimulated mice. Furthermore, OT-1 cells activated by DC-SIINFEKL in all organs, and OT-1 cells activated by AAV-OVA in the liver, expressed a significantly higher level of PD-1 than did OT-1 cells taken from untreated mice. However, expression of PD-1 was significantly higher in OT-1 cells in the liver of mice stimulated with AAV-OVA, compared both with OT-1 cells from unstimulated mice and with those in any organ of mice stimulated with DC-SIINFEKL. This shows that whereas PD-1 expression follows activation, the generation of PD-1hi cells is unique to cells primed in the liver. We can conclude that high PD-1 expression is not simply due to activated CD8+ T cells migrating to liver.
Antigen Presentation by Non–Bone Marrow–Derived Cells.
Cross-presentation depends on the transfer of antigen from an antigen-expressing cell to a distinct APC, and can lead either to cross-priming or to cross-tolerance.23-25 The APCs are generally MHC class I+ II+ bone marrow–derived cells such as macrophages or DCs. To test the participation of such cells in the OT-1 T cell response to AAV2-ova, we used bm8 mice. These mice harbor several mutations in the Kb MHC class I molecule, which prevent the presentation of the SIINFEKL peptide.26 We created radiation bone marrow chimeras in which bm8 bone marrow was used to reconstitute lethally irradiated B6 mice or vice versa. Because a subset of bone marrow–derived Kupffer cells is resistant to depletion by radiation alone, mice were additionally treated with clodronate liposomes after the bone marrow transplant. This treatment effectively depletes both subsets of Kupffer cells.17
The response of OT-1 T cells to AAV2-ova in the liver in such chimeras is shown in Fig. 6. The negative controls were B6→B6 chimeras transduced with antigen-negative AAV2-gfp vector, and the positive control was B6→B6 mice given AAV2-ova. All mice received an intravenous adoptive transfer of CFSE-labeled OT-1 T cells. Flow cytometric measures of the T cell response are shown in Fig. 6A. In the negative controls, there was no dilution of CFSE, no down-regulation of CD62L, and no up-regulation of CD44. In the positive controls, the T cells proliferated, CFSE staining was lost, and CD62L was down-regulated. The effectiveness of direct presentation was revealed by the dilution of CFSE and down-regulation of CD62L expression in bm8→B6 chimeras; conversely, the lack of cross-presentation was shown by the lack of CFSE dilution and lack of CD62L down-regulation in B6→bm8 chimeras. The cell surface expression of CD44 also indicated cell activation in bm8→B6 chimeras, but not in B6→bm8 chimeras. These data also are consistent with a complete lack of cross-presentation by bone marrow–derived cells.
Figure 6B shows the estimated numbers of OT-1 T cells in the liver, spleen, and PLN in each experimental group. In the liver and spleen, there was a significant increase in the number of OT-1 T cells in B6→B6 chimeras given AAV2-ova, compared to similar chimeras given AAV2-gfp. Critically, there was no increase in cell numbers in B6→bm8 chimeras, but there was in bm8→B6 chimeras. Statistical tests confirm the significance of the difference between B6→B6 positive controls and the B6→bm8 chimeras. Although the differences in the PLN did not give a significant result, the data from livers and spleens confirm that direct priming, and not cross-priming, leads to the clonal expansion of OT-1 T cells in response to AAV2-ova. In Fig. 6C, we show the percentage of lymphocytes that were OT-1 T cells in each tissue. These results support the same conclusion, but also reveal that OT-1 T cells were most frequent in the liver, as we have documented in previous studies.14 Taken together, this data set argues that AAV2-ova antigens are presented directly to OT-1 CD8+ T cells, without the involvement of bone marrow–derived APCs. The liver contains multiple populations of potential APCs, including hepatic stellate cells and LSECs. Neither of these cell types is significantly replaced in bone marrow chimeras, so their contribution to CD8+ T cell activation could not be excluded by this approach. However, we previously published that antigen was expressed exclusively in hepatocytes and not in other cell types,14 suggesting that direct priming was probably on hepatocytes.
In chronic HCV and HBV infections, virus-specific CD8+ T cells are impaired and ineffective at eliminating the virus. This correlated with an “exhausted” phenotype displaying high PD-1 and low CD127 expression.4, 21, 27-29 Impaired virus-specific CD8+ T cells may explain why these infections become chronic. Here, we documented liver-resident CD8+ T cells with this “exhausted” phenotype in the response to AAV-transduced hepatocytes in mice.
The initial CD8+ T cell response to AAV-encoded hepatocellular antigen is at first sight surprising, because the immune response to liver antigens is associated with many tolerance phenomena. Most dramatically, orthotopic transplantation of the liver between inbred strains of mice frequently results in tolerance without the need for immunosuppression.30 This effect could result from early T cell apoptosis,31, 32 from the immunosuppressive effects of antigen presentation by LSECs,10 the milieu created by the synthesis of both cytokines and PGE2 prostaglandins by Kupffer cells,33 or the action of regulatory T cells.30
The data reported here support a different model in which the CD8+ T cells respond to an AAV2 vector-encoded transgene expressed in hepatocytes and are primed without CD4+ T cell help by direct presentation of antigen. The absence of cross-priming is consistent with several other features of the AAV vectors. First, cell death and the subsequent phagocytosis of apoptotic bodies is a route by which cellular antigen enters the cross-presentation pathway,34 but these vectors are noncytopathic so this pathway of antigen presentation would not be favored. Second, whereas transduction of liver cells with adenovirus resulted in robust synthesis of type 1 IFN, this type of response was not elicited by AAV vectors.35 The type 1 IFNs play diverse roles in induction of specific immunity, and one of these roles is the promotion of cross-presentation.36
The evidence in favor of a direct-primed response raises the question of how the T cells could actually be engaged. Although hepatocytes are separated from circulating T cells by an endothelial barrier, this endothelium is fenestrated and direct contact between T cells and underlying hepatocytes can occur.12 In our model, the hepatocytes are the only cells transduced with the AAV vector. Thus, both the ultrastructural evidence and the expression pattern of AAV favor the model that hepatocytes are directly priming the CD8+ T cells.
Our data also support the interpretation that AAV2-ova did not induce a CD4+ T cell response, and that, furthermore, the CD8+ T cell response was not modified by the absence of CD4+ T cell help. In similar studies using D0.11.10 T cells, the lack of a readily detectable immune response was attributed either to the induction of T cell anergy16 or to the differentiation of the T cells into classic Tr3 cells expressing CD25 and the transcription factor forkhead box P3 (FoxP3).37 However, even if such cells were generated, a local immune response occurred in their presence.
The CD8+ T cell response to AAV was associated with elevated expression of the coinhibitory molecule PD-1. The PD-1 molecule is associated with a senescent phenotype, characteristic of inactive T cells found in the liver during chronic infections, such as with lymphocytic choriomeningitis virus and HCV.38, 39 The regulation of PD-1 expression on CD8+ T cells is not fully understood, but our data contribute to this question by showing that the antigen-specific induction of PD-1 on CD8+ T cells in the liver was not different in mice lacking MHC class II; therefore, we do not see an essential role either for CD4+ T helper cells, or for CD4+ T regulatory cells, in the induction of PD-1. Synthesizing the results from studies of lymphocytic choriomeningitis virus and HCV with our own, we would suggest the hypothesis that it is the continued contact with antigen in the liver environment and importantly, direct priming on hepatocytes, that renders CD8+ T cells both high in PD-1 expression and functionally incompetent. Gaining a better understanding of how CD8+ T cells are activated in the liver, and in particular in chronic viral hepatitis, will give us more insight into how to better generate effective therapies.
We thank Dr. K. Reed Clarke for the preparation of the AAV vectors.