The immune response to foreign antigens in the liver is often suboptimal and this is clinically relevant in chronic persistence of hepatotropic viruses. In chronic infection with the hepatitis C virus, activated CD8+ T cells specific for viral epitopes are present in the peripheral blood and the liver, yet viral clearance is unusual. To define the fate of activated CD8+ entering the liver, we developed a mouse model of portal vein injection of activated CD8+ T cells in vivo. Activated CD8+ T cells are retained very efficiently by the liver and undergo an approximately 8-fold expansion in the first 48 hours. This expansion is followed by apoptosis and a decline in numbers of the retained cells over the next 4 days. The presence of high affinity (HA) antigen does not affect the initial retention by the liver but greatly limits the expansion in the first 48 hours by increasing apoptosis of the retained cells. In the absence of Kupffer cells, the initial retention and expansion are unchanged, but HA antigen does not limit the expansion of the liver CD8+ T cell pool. In conclusion, these data identify a previously unknown phase of CD8+ T cell expansion after entering the liver, demonstrate that HA antigen limits the hepatic CD8+ T cell pool by inducing apoptosis, and that this effect requires Kupffer cells. Interfering with antigen presentation by Kupffer cells may be a strategy to limit HA antigen-induced deletion of activated CD8+ T cells entering the liver. (HEPATOLOGY 2004;39;1017–1027.)
The healthy liver is at the interface between the external and internal environments and has a number of unique immunological features. It contains a large and diverse immune cell population, yet immune responses to foreign antigens in the liver are frequently suboptimal.1 Due to its large blood flow and unique anatomy, a large proportion of T cells activated in the lymph nodes (LN), and all T cells activated in the spleen, flow into the liver. Such activated T cells often mount an inadequate effector response to foreign antigens in the liver.2 The reasons for the poor effector response by activated T cells entering the liver are of great clinical relevance in understanding why infection with hepatitis B and C frequently progress to chronicity despite the presence of virus-specific T cells in the peripheral blood and liver.3, 4
The healthy liver is unusual in retaining activated CD8+ T cells entering it. Such retention is due to a combination of unique anatomy, with low flow rates, a very mobile system of tissue macrophages, and the presence of adhesion molecules ICAM-1 and VAP-1.5 Liver retention has a high degree of specificity for activated T cells, and particularly CD8+ rather than CD4+ T cells. On entering the liver, CD8+ T cells are in physical contact with Kupffer cells, and a large proportion undergo apoptosis.5–7 In between the initial retention and final apoptosis, very little is known about activated CD8+ T cells that enter the liver.
In this study, we quantify the expansion of a well-defined effector population of CD8+ T cells on entering the liver. Such a population of activated CD8+ T cells will expand approximately 8-fold in 48 hours, with a gradual reduction over the next 4 days. If the retained CD8+ T cells interact with high affinity (HA) peptide and major histocompatibility complex (MHC) complexes on liver cell populations, the expansion is severely limited, and the removal of the retained CD8+ T cells is accelerated. Previously, we have shown that activated T cells retained in the liver are in physical contact with Kupffer cells. By using mice deficient in mature tissue macrophages, we demonstrate here that the ability of HA peptide to limit the expansion of the effector pool of CD8+ T cells is dependent on mature Kupffer cells. Inhibition of antigen presentation by Kupffer cells may be a therapeutic strategy for minimizing HA antigen-induced apoptosis of activated CD8+ T cells in the liver.
HA, high affinity; MHC, major histocompatibility complex; CSF-1, colony stimulating factor-1; TCR, T cell receptor; LN, lymph node; PBS, phosphate buffered saline; FBS, fetal bovine serum; CFSE, carboxyfluorescein diacetate; RBC, red blood cells; TUNEL, terminal deoxynucleotidyl transferase mediated dUTP biotin nick end labeling; CTL, cytotoxic T lymphocyte; ICAM, intercellular adhesion molecule; LFA, lymphocyte function associated antigen; TNF, tumor necrosis factor.
Materials and Methods
C57BL/6J mice and CSF-1 deficient mice (op/op mice)8 were purchased from The Jackson Laboratory (Bar Harbor, ME). A colony of OT-1 T cell receptor (TCR) transgenic mice was maintained on the CD45.1 allotype background. All animals were on the C57BL/6J background and housed in a specific pathogen-free environment, in accordance with institutional guidelines for animal care.
CD8+ OT-1 Lymphocyte Activation and Adoptive Transfer.
Axillary and inguinal LN and spleen were dissected from 6- to 8-week-old OT-1 mice. A cell suspension was obtained by mechanical homogenization under sterile condition. The cells were washed several times with 5% fetal bovine serum (FBS) in phosphate buffered saline (PBS) and resuspended in RPMI (Roswell Park Memorial Institute Media) 1640 (GIBCO, Grand Island, NY) supplemented with 10% FBS, 100 U/ml of penicillin G (SIGMA, St. Louis, MO), 100 μg/ml of streptomycin sulfate (SIGMA), and 50 μM of 2-ME (2-Mercaptoethanol, SIGMA) at the concentration of 2 × 106 cells/ml. HA peptide SIINFEKL (20 nM) was added and incubated under 37°C and 5% CO2. After 3 days, the cells were washed three times with PBS 5% FBS. In some experiments, the cells were incubated at 37°C with 2 μM carboxyfluorescein diacetate (CFSE) (Molecular Probes, Eugene, OR) for 20 minutes and washed for another three times. By day 3 of activation in vitro, more than 95% of these cells were CD8+, and more than 99% CD8+ cells were CD25 high. This confirmed that we had a very pure population of highly activated CD8+ T cells.
Age- and gender-matched C57BL/6J mice and CSF-1 deficient mice were used as recipients. Under ketamine/xylazine anesthesia, the abdominal cavity of the mice was exposed by a midline incision. Activated cells (5 × 106) prepared as described above were injected directly into the portal vein. After the injection, the abdominal cavity of mice was closed using proline sutures and Autoclips (Becton Dickinson, Parsippany, NJ). Twenty minutes before portal vein injection, each recipient mice received an intraperitoneal (ip) injection of either 25 nmol of the HA SIINFEKL in PBS, 25 nmol of the control peptide (ovalbumin 323-339), or PBS alone. Peptide or PBS were injected ip every 24 hours subsequently.
At various time points between 10 minutes and 7 days after injection of activated CD8+ OT-1 cells, the mice were sacrificed and the lymphocytes from the liver, blood, spleen, and lungs were isolated. In brief, mice were anesthetized, heparinized, and exsanguinated by cutting the abdominal aorta and vena cava. The liver was perfused immediately with digestion buffer consisting of Bruff's medium containing 0.05% collagenase IV (SIGMA), 0.002% DNase I (SIGMA), and 5% FBS. After perfusion, the liver was dissected out of the abdominal cavity and homogenized. The homogenized liver was incubated in 10 ml digestion buffer at 37°C for 40 minutes in a shaking water bath. After removing hepatocytes and cell clumps by centrifugation at 20g, the supernatant was then centrifuged at 500g and a pellet was collected. The pellet was suspended in 22% Opti-prep (Axis-Shield, Oslo, Norway) and centrifuged at 1,500g. The cells at the interface were collected, washed, and analyzed. Lymphocytes from spleen, LN, and blood were obtained by mechanical homogenization. Red blood cells (RBC) were lysed using Ack Lysing Buffer (SIGMA) for spleen and blood.
Staining Reagents and Flow Cytometric Analysis.
The antibodies used for staining were anti-TCRαβ (clone H57-597), anti-CD8 (clone 53-6.7), anti-CD4 (clone H129.19), anti-CD45.1 (clone A20), and anti-CD25 (clone PC61). Avidin-fluorescein-5-isothiocyanate, phycoerythrin, or allophycocyanin was used when biotinylated antibody was used at primary staining. All these reagents were purchased from BD Biosciences (San Jose, CA). For terminal deoxynucleotidyl transferase mediated dUTP biotin nick end labeling (TUNEL) staining, Apop Tag Fluorescein Direct In Situ Apoptosis Detection Kit (Intergen, NY) was used according to the manufacturer's instructions. Cells (1–3 ×106) obtained from each organ were stained using reagents listed previously. After staining, the samples were fixed with 2% paraformaldehyde.
The samples were analyzed using FACS Calibur flow cytometer (BD Bioscience) and CellQuest software (BD Bioscience). Each sample data was first gated based on forward scatter and side scatter on lymphocytes and then further analyzed for fluorescence.
In Vivo Cytotoxic T Lymphocyte (CTL) Assay.
Target cells for in vivo evaluation of cytotoxic activity were prepared as described in detail elsewhere.9 Briefly, spleen and LN cells were isolated from age- and gender-matched C57BL/6 mice and RBC were lysed. The cells were washed and divided into two populations. One population was pulsed with 1 μM of OT-1 specific peptide SIINFEKL, whereas the other was left unpulsed at 37°C for 1 hour. The peptide-pulsed population was labeled with high (5 μM) CFSE concentration and the unpulsed population was labeled with low (0.5 μM) concentrations. Both population were then washed five times to remove the residual peptide and CFSE. The cells from each population were counted, mixed in a 1:1 ratio, and 2 × 107 of the cell mixture (in 0.5 ml PBS) was injected into the tail vein of mice that had portal vein injections of PBS (control group) or portal vein injections of activated CD8+ OT-1 cells (test group) 36 hours before. Specific in vivo cytotoxicity was determined by analyzing isolated lymphocytes from liver, spleen, and LN of the recipient mice 12 hours after injection of the target cells using FACS. The target and control cell population can be detected separately at different CFSE fluorescence intensities. Specific target cell lysis was described using a value estimated as (1 – % SIINFEKL pulsed cells / % non-pulsed cells) × 100. To minimize the subset difference among target cells in each organ and/or each mice, the isolated cells were further stained with anti-TCR antibody and gated on TCR positives at FACS analysis.
Mean values ± SE for various groups were calculated and statistical significance determined using the Student's t test.
Efficient Activation of OT-1 In Vitro.
Culture of naive OT-1 cells from the spleen and LN with the HA peptide (SIINFEKL) for 3 days results in efficient activation of the OT-1 CD8+ T cells (Fig. 1A and B). In addition, by day 3, the cultured cell population had a purity of greater than 96% activated OT-1 CD8+ T cells.
Activated CD8+ T Cells Are Efficiently Retained by the Healthy Liver.
We previously showed in a single pass perfusion model that the normal liver selectively retains activated CD8+ T cells in an antigen non-specific manner. To test the efficiency of activated CD8+ T cell retention in vivo, activated CD8+ T cells were directly injected into the portal vein of normal B6 mice. Ten minutes after injection, the percentages of donor CD8+ T cells in various organs were determined by cell isolation and FACS analysis.
In a direct confirmation of the single pass perfusion model, 10 minutes after injection, donor cells were found in large numbers in the liver and virtually none were present in other organs including the blood, lung, or spleen (Fig. 2A–D) and also draining or distal LN. This distribution of portal vein injected activated CD8+ T cells staying in the liver was essentially unchanged for 24 hours after cell injection (Fig. 2E–H).
Increase in the OT-1 CD8+ T Cell Population Retained by the Liver.
By 48 hours, there is a substantial increase in the percentage of OT-1 CD8+ T cells in the liver compared to the initial retention at 10 minutes (Fig. 3). On average, the number of OT-1 CD8+ T cells increased 8-fold in 48 hours. This increase between the 10 minute and 48 hour time points could have occurred due to intra-hepatic division of OT-1 CD8+ T cells retained in the liver, or due to their exiting from the liver, dividing extrahepatically, and then trafficking back to the liver. But because only a very few donor cells were seen outside the liver during the early time course after portal vein injection (Fig. 2), the latter possibility is unlikely. Between days 2 and 7 after injection, the percentage of OT-1 CD8+ T cells in the liver decreased substantially.
Presence of HA Antigen in the Liver Severely Limits the Degree of Expansion of OT-1 CD8+ T Cells.
To study whether the presence of HA antigen in liver affects the fate of retained cells, recipient mice were injected with the HA peptide for the OT-1 T cell receptor (SIINFEKL) 20 minutes before injection of activated OT-1 CD8+ T cells into the portal vein. At 10 minutes after injection of OT-1 CD8+ T cells, there was no difference between the HA peptide-injected mice and the PBS-injected controls (Fig. 3A and D). This demonstrates that the presence of HA peptide does not affect the initial retention of activated OT-1 CD8+ T cells by the liver. By 48 hours after transfer of OT-1 CD8+ T cells there is a significant difference in the percentage of OT-1 CD8+ T cells in the livers with and without HA peptide (Fig. 3B and E). The increase in OT-1 CD8+ T cells in the first 48 hours after they enter a liver without HA peptide is severely limited in the presence of HA peptide. From 14 experiments and 5–15 animals for each time point, this phenomenon was confirmed and statistically evaluated (Fig. 4). There was no difference between the control ovalbumin peptide (ova 323–339) and the PBS-injected mice. Each value (mean ± SE) was 10 minutes; PBS: 4.214 ± 1.429 versus HA peptide: 4.970 ± 1.512 (NS), 48 hours; PBS: 35.797 ± 4.801 4 versus HA peptide: 15.242 ± 3.711 (P < .005), 96 hours; PBS: 16.772 ± 2.837 versus HA peptide: 2.200 ± 1.004 (P < .001), day 7; PBS: 10.220 ± 4.220 versus HA peptide: 1.737 ± 0.756 (NS). In summary, this shows that the approximately 8-fold increase that occurs in the pool of activated OT-1 CD8+ T cells in the first 48 hours is severely limited to an approximately 3-fold increase if HA peptide is present but is unaltered in the presence of a control ovalbumin peptide.
Transferred OT-1 CD8+ T Cells Are Functional CTLs In Vivo.
To establish that the portal vein injected CD25+ CD8+ OT-1 cells had cytotoxic function, an in vivo CTL assay was performed. Figure 5(A–C) shows data from control mice that did not receive any CD8+ OT-1 cells, but did receive CFSE labeled T cell populations (CFSE high loaded with HA peptide and CFSE low not loaded with peptide). There is no loss of peptide loaded CFSE high target cells. Figure 5(D–F) shows data from mice that received CD8+ OT-1 cells and CFSE labeled T cell populations (CFSE high loaded with HA peptide and CFSE low not loaded with peptide). There is loss of the peptide loaded CFSE high cell population, demonstrating cytotoxic function of the injected CD8+ OT-1 cells. Liver, spleen, and LN were analyzed separately and the degree of loss of target cells was similar and there was no difference in cytotoxic activity among these organs.
OT-1 CD8+ T Cells Undergo Proliferation and Apoptosis, With Enhancement of the Apoptosis by HA Peptide.
To test that the increase in the number of OT-1 CD8+ T cells occurs via cell division, the cell membrane dye CFSE was used. CFSE binds to cell surface protein and decreases in intensity as the cells divide. Thus, proliferation in vivo can be demonstrated. Activated OT-1 CD8+ T cells were labeled with CFSE and injected into the portal vein of mice previously injected ip with PBS or HA peptide. After 48 hours, liver lymphocytes were isolated and the CFSE intensity of donor cells was analyzed by FACS analysis by gating on CD45.1+ and CD8+ cells. This revealed an approximately 30-fold decrease in CFSE intensity in the donor cells in the HA peptide- and PBS-injected groups during the first 48 hours (Fig. 6). This demonstrates that retained activated CD8+ T cells do proliferate in the liver and this is most likely the cause of the dramatic increase of donor cells seen in the liver after 48 hours. The decline of CFSE intensity was identical in PBS- and HA-peptide injected mice, suggesting that the degree of cell division is not significantly affected by the interaction of the retained cells with HA peptide.
Death of donor cells was quantitatively analyzed looking at apoptotic cells detected by TUNEL method. Figure 7 shows TUNEL staining for the CD45.1 donor T cells isolated from livers of mice injected with PBS or HA peptide. The presence of HA peptide in the liver resulted in significantly greater percentage of the donor cell being TUNEL positive 48 hours after injection (Fig. 7). The possibility that donor T cells escape more out of the liver when the HA peptide is present is unlikely because significantly fewer donor T cells were detected in the HA peptide given group in every extra-hepatic site at any time point (Fig. 8).
HA Peptide-Induced Apoptosis of Intrahepatic CD8+ T Cells Is Dependent on Kupffer Cells.
We previously showed that the majority of activated CD8+ T cells retained in the liver are in contact with Kupffer cells. In addition, we also demonstrated that ICAM-1, which is present on Kupffer cells and sinusoidal endothelium, is required for efficient retention of activated CD8+ T cells by the liver. To test the hypothesis that Kupffer cells are required for the retention and the antigen-induced apoptosis of activated CD8+ T cells, we studied the retention and antigen-induced deletion of activated OT-1 CD8+ T cells in the livers of wild-type and colony stimulating factor 1 (CSF-1) deficient mice. CSF-1 deficient mice (op/op mice) lack CSF-1 due to an inactivating frameshift mutation in the coding region of the CSF-1 gene.8 CSF-1 is essential for development of tissue macrophages, and CSF-1 deficient mice have no mature functional tissue macrophages including Kupffer cells. Total Kupffer cell numbers are less than 30% compared to hemizygous littermates, and the Kupffer cells present are small and unable to phagocytose.8 Using the CSF-1 deficient mice as recipients, an experiment was designed in order to reveal the role of Kupffer cells in liver retention and deletion of activated CD8+ T cells.
Activated OT-1 CD8+ T cells were transferred via the portal vein into CSF-1 deficient mice or wild type littermates as control that had received ip injections of PBS or HA peptide. At 10 minutes, there was no difference between CSF-1 deficient mice and wild type littermates in the PBS- and peptide-injected groups (Fig. 9A), indicating no difference in the initial retention by liver. After 48 hours, there was no difference between CSF-1 deficient mice and wild type mice when the HA antigen was not present. Both showed a similar increase in the percentages of donor cells in the liver (Fig. 9A). This demonstrates that the initial retention and expansion is not dependent on Kupffer cells. In the HA peptide-injected mice there was significant difference between the wild type and CSF-1 deficient mice. In wild type mice, the presence of HA peptide severely limited the increase of the OT-1 CD8+ T cells at 48 hours (as demonstrated in Fig. 4A), but in the CSF-1 deficient mice, HA peptide was unable to limit the increase in OT-1 CD8+ T cells (Fig. 9A) (% of donor CD8+ T cells in the liver lymphocyte pool. HA peptide-injected wild type: mean ± SE, 5.4 ± 2.244, P < .02, HA peptide-injected CSF-1 deficient 28.259 ± 6.443). This difference strongly suggests that the mechanism of HA peptide-induced CD8+ T cell death is lacking in CSF-1 deficient mice, and we propose it is due to the absence of Kupffer cells in these mice. TUNEL staining of donor cells in the wild type and CSF-1 deficient mice receiving HA peptide did not reveal any significant difference between the two groups (data not shown). We cannot, therefore, demonstrate that the lack of peptide-induced reduction in the CSF-1 deficient mice injected with peptide was due to less apoptosis, but because the mechanism for phagocytosis of apoptotic cells may be also reduced in CSF-1 deficient mice, it is difficult to interpret TUNEL data.
Almost no donor cells are detectable in the blood, lungs, or spleen at 10 minutes after injection (Fig. 2A–D). The distribution of donor cells in the blood, lungs, and spleen in CSF-1 deficient mice were identical to wild type, suggesting that the absence of functional Kupffer cells has minimal effect on the ability of the liver to retain activated CD8+ T cells entering it. At 48 hours, when the HA antigen is not present, some donor cells start to appear in other organs, including spleen (Fig. 8A–C), but there were no differences between CSF-1 deficient mice and wild type mice (Fig. 9B). The presence of HA peptide results in very few donor cells being detected in the blood, lungs, and spleen of wild type mice (Fig. 8). This may be because the hepatic retention and apoptosis in the presence of peptide is so efficient that they do not exit the liver, or because they undergo apoptosis outside of the liver due to interaction with HA peptide. In contrast to the liver, where there was a discordance in the percentage of donor cells in the wild type peptide-injected and CSF-1 deficient peptide-injected groups, in the blood, lungs, and spleen almost no donor cells were present in the wild type and CSF-1 deficient peptide-injected groups (Fig. 9B).
The liver contains a large and varied population of immune cells and has unique functional features.10, 11 During in vivo T cell activation, there is massive accumulation of the apoptosing T cells in the liver.12 This has been confirmed in a wide variety of models ranging from transgenic animals, to adoptive transfer studies, to viral infection of unmanipulated wild type mice.13–15 The demonstration in a single pass perfusion model, of the ability of the healthy liver to retain activated CD8+ T cells further suggested that the liver functions very differently from other non-lymphoid organs.5 Collectively, it has become clear that the healthy liver retains activated CD8+ T cells, and many or most of these cells undergo apoptosis. These two facts are consistent with a number of hypothesis, with two contrasting views being 1) the liver is trapping T cells that are programmed for apoptosis (elephants graveyard model) or 2) the liver is retaining T cells and inducing apoptosis of the T cells that interact with cognate antigen (responder trap).16
In this study, we provide detailed information on the fate of T cells retained by the liver. From the previous adoptive transfer of activated T cells into the systemic circulation, it has not been possible to address in detail issues of liver T cell retention, function, proliferation, and apoptosis because it is not known when the adoptive transferred population actually enters the liver.12, 17 To clarify this we used a strategy of portal vein injection, giving us an exact time of T cell entry into the liver. Our data confirm the previous observations that activated CD8+ T cells entering the liver are retained and undergo substantial apoptosis.5 In between these two events there is not uniform apoptosis and decline in the retained population. Rather, there is significant expansion that peaks at approximately 8-fold by day 2 and then a gradual decline (Fig. 4). The decline is associated with apoptosis of the retained cells, but the gradual increase in the percentage of donor cells in the LN and spleen as they are decreasing in the liver suggests that at least some of the retained cells escape from the liver. So it is a graveyard, but death is preceded by division and some cells escape the graveyard altogether.
The consequences for an activated CD8+ T cell entering a liver containing HA peptide are quite clear. There is no difference in retention, with virtually complete initial retention by the liver regardless of whether the cells entering the liver do or do not interact with HA peptide (Fig. 3). However the expansion of the retained population is significantly curtailed if the activated T cells interacts with HA peptide on a liver cell population. The presence of HA peptide diminishes the peak size of the retained cell population by approximately half, and we believe this is mostly due to increased apoptosis of the retained cells (Fig. 7). These data are consistent with the responder trap hypothesis. This mechanism will apply to all T cells activated in the spleen because they exit into the portal circulation. These findings were obtained using HA peptide, and the response to low affinity agonists, or partial antagonists, is yet to be determined.
By day 7, the presence of HA peptide results in approximately 6-fold fewer donor cells in the liver (Fig. 4). Similarly, in the blood, lung, and spleen, there are many fewer donor cells in the presence of HA peptide (Fig. 8). We infer that in the presence of HA peptide, activated CD8+ T cells are retained by the liver and undergo massive apoptosis, leaving very few cells to exit into the circulation. The proposed mechanism for the enhanced retention is the upregulation of LFA-1 after engagement of the T cell receptor.18 In fact, interaction of LFA-1 with ICAM-1, in the absence of adequate costimulation, may also result in apoptosis, particularly of CD8+ T cells.19–21 We, however, cannot exclude that the decrease of cells in the spleen, LN, and lungs of the peptide-injected animals occurs due to apoptosis at these sites rather than the liver. The numbers of donor cells outside the liver were too few to generate consistent TUNEL data. Our peptide-injected model does not have the tissue specificity of antigen presentation, but is a relevant model for many conditions, such as viral hepatitis, in which viral antigen is distributed widely throughout the body, and virus-specific T cells enter the liver.
It is of great interest to identify the cellular and molecular mechanism required for peptide-induced deletion of activated CD8+ T cells retained by the liver. From bone marrow chimera studies we have shown that antigen presentation on bone marrow derived cells enhances peptide-induced deletion of activated CD8+ T cells.14 To test if Kupffer cells are the bone marrow cell population enhancing apoptosis, it was necessary to use a mouse model lacking Kupffer Cells. A number of maneuvers have been used to impair Kupffer cell function in vivo, including silica and gadolinium chloride injections.22–24 These strategies have significant drawbacks. First, they are reasonably good at impairing Kupffer cell phagocytosis but do not physically remove mature Kupffer cells.25 Second, they are associated with Kupffer cell activation with release of cytokines, and this has unpredictable effects. To avoid these issues, we made use of the natural mutant mice lacking CSF-1.8 In these mice, the maturation of all tissue macrophages is grossly impaired and mature Kupffer cells are not present in the liver.26
In wild type mice, the presence of HA peptide dramatically reduces the number of OT-1 CD8+ T cells at day 2 by enhancing intrahepatic apoptosis of these cells (Figs. 3 and 4). In contrast to wild type mice, the presence of peptide in the liver of CSF-1 deficient mice does not result in any decrease in the number of OT-1 CD8+ T cells. This strongly supports the hypothesis that antigen presentation by Kupffer cells is required for peptide-induced deletion of the retained activated OT-1 CD8+ T cells. It may therefore be possible to selectively interfere with antigen presentation by Kupffer cells, thus blocking the peptide-induced CD8+ T cell deletion without interfering with the CD8+ T cell effector response against hepatocytes. This is therapeutically desirable in situations of suboptimal intrahepatic immune responses, such as chronic infection with hepatitis B and C viruses. The ability of Kupffer cells to induce apoptosis of T cells in vitro by a CD95 dependent mechanism, and the inability to induce oral tolerance in gadolinium chloride treated animals, is consistent with our hypothesis.27–29
These data support a model of hepatic antigen presentation in which antigen presentation by hepatocytes and endothelium leads to conventional CD8+ T cell effector function with release of cytokines and cytotoxic function. Antigen presentation by Kupffer cells, however, enhances CD8+ T cell apoptosis, thus limiting the overall hepatic CD8+ T cell response. Kupffer cells have a number of unique features that makes them good candidates for inducing antigen-specific deletion in activated CD8+ T cells. They are highly mobile and phagocytic.30 In addition, Kupffer cells possess a variety of molecules that may induce CD8+ T cell apoptosis. These include the production of TNF-α, CD95-ligand, galectin-1, and indolomine deoxygenase (IDO).28, 31–33 Of these, at least TNF-α and CD95-ligand are upregulated by the interferon-gamma that is produced in abundance by activated CD8+ T cells. The role of antigen recognition by activated CD8+ T cell in inducing apoptosis may be to increase the affinity of LFA-1 and induce stronger binding to Kupffer cell ICAM-1, or it may be to increase production of interferon gamma by the CD8+ T cell, resulting in upregulation of the Kupffer cell apoptotic mechanisms mentioned previously. After induction of apoptosis, Kupffer cells are able to phagocytose apoptotic cells, thus ensuring efficient removal of the CD8+ T cell apoptotic bodies.34, 35 All these features ensure that Kupffer cells are uniquely poised to sample the entire immunogenic environment of the liver, present it to activated CD8+ T cells, induce their apoptosis, and also phagocytose the apoptotic debris.
The presence of multiple mechanisms on Kupffer cells for the induction of apoptosis complicates the prospect of inhibiting CD8+ T cell apoptosis. Because antigen presentation is required for Kupffer cell induced apoptosis, it may be possible to block CD8+ T cell apoptosis by interfering with class I MHC processing in Kupffer cells rather than trying to block all the pro-apoptotic signals. A number of stages of the class I processing pathway are potentially amenable to blockage. One approach is the delivery of inhibitors of proteasomes to Kupffer cells to block class I processing.
In summary, we have shown that between hepatic retention and apoptosis, the pool of retained activated CD8+ T cells undergoes significant expansion when they enter a liver that does not contain HA peptide. Such expansion does not occur in the presence of HA peptide due to a high degree of apoptosis. In the absence of Kupffer cells, the retention of activated CD8+ T cells by the liver is not altered, but HA peptide does not limit the size of the retained CD8+ T cell pool.
The authors are grateful to Richard Flavell for invaluable mentorship, and Nicholas Crispe for helpful discussion.