Liver Biology and Pathobiology
Dendritic cells are required for effective cross-presentation in the murine liver†
Article first published online: 19 DEC 2007
Copyright © 2008 American Association for the Study of Liver Diseases
Volume 47, Issue 4, pages 1343–1351, April 2008
How to Cite
Plitas, G., Burt, B. M., Stableford, J. A., Nguyen, H. M., Welles, A. P. and DeMatteo, R. P. (2008), Dendritic cells are required for effective cross-presentation in the murine liver. Hepatology, 47: 1343–1351. doi: 10.1002/hep.22167
Potential conflict of interest: Nothing to report.
- Issue published online: 25 MAR 2008
- Article first published online: 19 DEC 2007
- Accepted manuscript online: 19 DEC 2007 12:00AM EST
- Manuscript Accepted: 30 NOV 2007
- Manuscript Received: 9 AUG 2007
- National Institutes of Health. Grant Numbers: AI70658, DK068346
The liver harbors a diversity of cell types that have been reported to stimulate T cells. Although most hepatic dendritic cells are immature, a small population of CD11chigh conventional dendritic cells (cDCs) exists that expresses high levels of costimulatory molecules. We sought to determine the relative contribution of cDCs to cross-presentation by the liver. In vitro, liver nonparenchymal cells (NPCs) depleted of cDCs induced only minimal proliferation and activation of antigen-specific CD8+ T cells when loaded with soluble protein antigen. Using a transgenic mouse with the CD11c promoter driving expression of the human diphtheria toxin receptor, we found that selective depletion of cDCs in vivo reduced the number and activation of antigen-specific CD8+ T cells in the liver after intravenous administration of soluble protein antigen. Adoptive transfer of DCs, but not CD40 stimulation, restored the hepatic T-cell response. Conclusion: Our findings indicate that the ability of the liver to effectively cross-present soluble protein to antigen-specific CD8+ T cells depends primarily on cDCs. Despite costimulation, other resident liver antigen-presenting cells cannot compensate for the absence of cDCs. (HEPATOLOGY 2008.)
Antigen-presenting cells (APCs) play a critical role in modulating the hepatic immune response to infection, autoimmunity, and malignancy. Cross-presentation, the process by which APCs internalize antigens from the extracellular environment and present them as MHC class I–bound peptides is essential for generating CD8+ T-cell immunity against antigens synthesized in cells.1 Unlike spleen CD11c+ cells, most CD11c+ cells in the liver express low levels of MHC class II and costimulatory molecules.2 Although liver DCs have been shown to prime CD8+ T cells by cross-presentation, several other hepatic cell types have been shown to possess this function as well.3, 4 Limmer et al. reported that liver sinusoidal endothelial cells (LSECs) loaded with soluble protein stimulated T cell receptor (TCR) transgenic CD8+ T cells.5 Hepatocytes infected with Plasmodium presented antigen to CD8+ T cells, which then acquired antigen-specific lytic activity and secreted interferon-gamma (IFN-γ).6 Induction of adaptive immunity against the intracellular pathogen Listeria monocytogenes (LM) is mediated primarily by cross-presentation.7 Hepatic stellate cells infected with LM elicited antigen-specific CD8+ T cell responses in vitro and when loaded with ovalbumin, mediated protection in vivo against an ovalbumin-expressing strain of LM.8 Kupffer cells, the resident macrophages of the liver, have been reported to cross-present antigen and also secrete interleukin 10, which has potent immunoregulatory effects on antigen presentation.9 NPCs also contain neutrophils and B cells, which are known to cross-present antigen in other anatomic locations.10, 11
In the mouse, CD11chigh cells have been termed cDCs to distinguish them from plasmacytoid DCs (pDCs), which are CD11cint.12, 13 To determine the relative contribution of hepatic cDCs to cross-presentation of soluble protein in the liver, where there are other cells with APC function, we performed in vitro studies with bulk NPCs devoid of cDCs. Furthermore, we utilized CD11c-diphtheria toxin receptor (CD11c-DTR) transgenic mice, in which cDCs are selectively depleted, to monitor the accumulation of antigen-specific T cells in the liver after antigen exposure. Our results indicated that liver cDCs play a dominant role in effective cross-presentation of soluble antigen and that other hepatic APCs cannot substitute for cDCs in the induction of a maximal effector T-cell response by the liver.
Materials and Methods
Animals and Infections.
Six-week-old to 10-week-old male wild-type, β2-microglobulin−/− (β2m−/−), and CD11c-DTR mice, each on a C57BL/6 (B6) background, were purchased from the Jackson Laboratory (Bar Harbor, ME). The CD11c-DTR mice were bred in house and genotyped from tail DNA using published primer sequences.7 TCR transgenic OT-I Rag1−/− mice were purchased from Taconic Farms (Germantown, NY). Mice were inoculated intravenously with 5 × 103 colony-forming units (CFUs) of the wild-type LM strain 10403S or the recombinant 10403S strain expressing the SIINFEKL epitope (LMova, kindly provided by Dr. Eric Pamer, Sloan-Kettering Institute, New York, NY). The animals were maintained in a pathogen-free animal housing facility at Memorial Sloan-Kettering Cancer Center. All procedures were approved by the Institutional Animal Care and Use Committee.
Liver NPCs were isolated as previously described with minor modifications.2 After harvesting and processing the liver, the supernatant was centrifuged (300g for 7 minutes) to isolate NPCs. The NPCs were further enriched by a 40% (wt/vol) Optiprep (Sigma-Aldrich, St. Louis, MO) density gradient per the manufacturer's protocol. To obtain NPCs devoid of cDCs (cDCneg NPCs), the enriched NPCs were labeled with a CD11c antibody, and then CD11chigh cells were removed by fluorescence-activated cell sorting (FACS) with a FACSAria cell sorter (BD Biosciences, San Jose, CA). The bulk NPC group was treated in an identical manner except that the CD11chigh cells were included in the sorted population. Post-sort purity was routinely greater than 99%. Splenic DCs were purified with anti-CD11c immunomagnetic beads (Miltenyi Biotec, Auburn, CA). In vivo expansion of DCs was accomplished with daily intraperitoneal (IP) injections of 10 μg of recombinant human fms-like tyrosine kinase 3 ligand (Flt3L; Amgen, Seattle, WA) for 10 days. Adoptively transferred DCs were uniformly CD11chigh with approximately 1% pDC (B220+mPDCA+) contamination (data not shown).
Flow cytometry was performed on a FACSAria. Prior to antibody staining or immunomagnetic bead selection, all cells were incubated with 1 μg of anti-FcγRIII/II antibody (2.4G2, Fc block; Monoclonal Core Facility, Sloan-Kettering Institute, New York, NY) per 106 cells. Cells were labeled with 0.06 μg of antibody/106 cells. Stains included fluorescein isothiocyanate, phycoerythrin (PE), allophycocyanin, peridinin chlorphyll protein, PE-Cy7, allophycocyanin-Cy7, and biotin-conjugated antibody (BD Bioscience). Biotinylated antibodies (Abs) were secondarily stained with streptavidin-PE. NPCs were stained for CD11c (HL-3), B220 (RA3-6B2), CD80 (16-10A1), CD86 (GL-1), MHC II (AF6-120.1, I-Ab), CD62L (MEL-14), CD44 (IM7), and CD8α (53-6.7), all from BD Biosciences, and mPDCA-1 (JF05-1C2.4.1), from Miltenyi. Dead cells were excluded with 7-amino-actinomycin D (BD Biosciences). Identification of OT-I T cells in vivo and in vitro was facilitated by the use of a fluorochrome-labeled MHC class I H-2Kb-SIINFEKL tetramer (OVAtet; Tetramer Core Facility, Sloan-Kettering Institute, New York, NY). To characterize the hepatic DC populations, we enriched NPCs with anti-CD45 immunomagnetic beads (Miltenyi). Intracellular cytokine analysis (ICC) of in vitro– and in vivo–stimulated T cells was performed following in vitro restimulation with SIINFEKL or irrelevant TRP peptide (Peptide Synthesis Facility, Sloan-Kettering Institute, New York, NY)–pulsed splenic DCs in the presence of Brefeldin A for 5 hours. The cells were subsequently stained with surface antibodies, fixed, permeabilized, and stained for intracellular IFN-γ (XMG1.2, BD Biosciences), perforin (eBioOMAK-D eBioscience; San Diego, CA), or the appropriate isotype controls. Flow cytometry data was analyzed using FlowJo software (Tree Star, Ashland, OR).
T Cell Assays.
In vitro antigen-specific T cell activation was assayed with OT-I TCR transgenic T cells specific for the SIINFEKL peptide. All T cells were purified from spleens and lymph nodes using a CD8+ T cell isolation kit per the manufacturer's protocol (Miltenyi). Stimulator cells were loaded with SIINFEKL peptide (1 μg/mL), or ovalbumin protein (2 mg/mL; Sigma Aldrich) and cultured with an agonistic CD40 antibody (10 μg/mL anti-CD40, FGK45; Monoclonal Core Facility, Sloan-Kettering Institute, New York, NY) or an isotype control for 2 hours, washed 3 times with 1% FBS in PBS, and then added at varying concentrations to 5 × 104 OT-I T cells in a 96-well U-bottom plate for 3 days. [3H]Thymidine (PerkinElmer; Waltham, MA) was added to the wells after 3 days, and 20 hours later radioactive uptake was measured as counts per minute.
In Vivo T Cell Assay.
OT-I T cells were labeled with 50 μM 5,6-carboxy-fluorescein diacetate succinimidyl ester (CFSE; Molecular Probes). The cells were then washed 3 times in 5% FBS in PBS and resuspended in normal saline. Then 2 × 106 CFSE-labeled OT-I T cells were injected in the lateral tail vein. The mice then received an intraperitoneal injection of 100 ng diphtheria toxin (DT; Sigma Aldrich) in 200 μL of PBS or vehicle alone. Eighteen hours later, the animals received 500 μg ovalbumin, 10 ng SIINFEKL peptide, or 5,000 CFU LMova. In addition, some of the animals received an intraperitoneal injection of 100 μg anti-CD40 or isotype control. The T cell controls received equivalent amounts of bovine serum ovalbumin (BSA), irrelevant TRP peptide (Peptide Synthesis Facility, Sloan-Kettering Institute, NY), or non-OVA-expressing LM. Three days later, liver T cells were isolated from NPCs with immunomagnetic microbeads (Miltenyi) specific for CD90, and flow cytometry was used to measure the dissolution of CFSE fluorescence or activation marker expression among CD3+CD8+OVAtet+ cells.
Statistical significance was determined by the Student t test using Prism 4.0 statistical software. P values < 0.05 were deemed significant.
Liver Contains a Small Population of Mature DCs.
Analysis of freshly isolated CD11c+ NPCs revealed a discrete population of liver cDCs (that is, CD11chigh cells) that coexpressed high levels of MHC class II, CD80, and CD86 (Fig. 1A,B), consistent with previous findings.14 We found that liver cDCs accounted for 3% of NPCs and approximately 10% of bulk liver CD11c+ cells, but there were only approximately 47,000 cDCs per liver (Fig. 1A,C). Further analysis revealed that CD11cint cells contained approximately 40% pDCs (B220+mPDCA+) and almost all pDCs were CD11cint(Fig. 1D), consistent with the findings of Jomantaite et al.14
Hepatic cDCs Are Required for Presentation of Soluble Antigen In Vitro.
To determine the relative requirement of liver NPCs to contain cDCs in order to process and cross-present soluble protein antigen, we loaded NPCs or cDCnegNPCs (NPCs depleted of CD11chigh cells by FACS) with ovalbumin and cultured them with OT-I T cells. Ovalbumin-loaded NPCs induced CD8+ T cell proliferation in a dose-dependent fashion, which was further augmented by CD40 stimulation (Fig. 2A). In contrast, ovalbumin-loaded cDCneg NPCs induced minimal proliferation, even in the presence of anti-CD40. To determine the direct antigen-presenting ability of cDCnegNPCs, we repeated the experiments with the ovalbumin-derived MHC class I peptide SIINFEKL. Substituting protein for the peptide antigen bypassed the need for internalization and processing of protein antigen and permitted analysis of just antigen presentation. We found that cDCnegNPCs loaded with peptide induced T cell proliferation, although slightly less than bulk NPCs (Fig. 2B). CD40 stimulation eliminated the disparity.
cDCs Dictate Phenotype of OT-I T Cells Stimulated by Bulk NPCs In Vitro.
OT-I T cells cultured with ovalbumin-loaded NPCs displayed an activated phenotype, as indicated by increased expression of CD44 and a concomitant decrease in CD62L expression (Fig. 3A). In contrast, cDCnegNPCs induced less T cell activation. Similarly, intracellular perforin staining was slightly lower for OT-I T cells cultured with cDCnegNPCs (Fig. 3B). T cell phenotype was not altered by CD40 stimulation (Fig. 3A,B). To establish that the phenotype of the T cells correlated with effector activity, we tested the ability of the OT-I T cells to differentiate into IFN-γ-producing effector cells. OT-I T cells cross-primed by ovalbumin-loaded bulk NPCs produced IFN-γ (Fig. 3C). This response was antigen specific, as evidenced by the lack of IFN-γ staining when an irrelevant peptide was used during restimulation. Cross-presentation by cDCs played an important role in the generation of antigen-specific effector T cells, as indicated by the lower percentage of IFN-γ+ OT-I T cells when cDCnegNPCs were used, regardless of CD40 mediated costimulation.
In Vivo Depletion of cDCs Decreases the Number of Liver OT-I T Cells Following Ovalbumin Administration.
In vitro, it was apparent that cDCs were required for efficient cross-priming of antigen-specific CD8+ T cells by NPCs. To test the relevance of our findings in vivo, we utilized CD11c-DTR mice, in which low doses of DT lead to the transient ablation of cDCs (that is, CD11chigh cells).13 In CD11c-DTR mice, we found that DT effectively depleted liver cDCs (Fig. 4A). In contrast, CD11cint cells, including pDCs, were not significantly depleted (Fig. 4B).
To determine the contribution of cDCs to the accumulation of antigen-specific T cells in the liver, we adoptively transferred CFSE-labeled OT-I T cells into CD11c-DTR mice and then administered DT. Eighteen hours later, ovalbumin was injected intravenously, and 3 days later liver T cells were isolated and analyzed by flow cytometry. Thus, cross-presentation was necessary in order for the T cells to be primed. Depletion of cDCs decreased the percentage and number of hepatic OT-I T cells, and systemic CD40 stimulation did not affect this finding (Fig. 5A). Nevertheless, based on CFSE dissolution, OT-I T cells did undergo some division despite cDC depletion. Administration of DT to wild-type control mice did not affect the expansion of OT-I T cells on delivery of antigen (data not shown). To determine if liver cDCs were required in vivo for direct antigen presentation, we repeated our in vivo experiments using SIINFEKL peptide. We found that cDCs were not necessary for OT-I T cells to accumulate in the liver in response to peptide (Fig. 5B). Curiously, in the setting of CD40 stimulation and peptide injection, there was an approximately 6-fold decrease in the percentage of liver OT-I T cells following depletion of cDCs.
Although liver cDCs were required for the maximal T cell response to soluble protein antigen, we investigated whether liver cDCs were necessary for a T cell response to LMova intracellular bacterial infection, which is mediated primarily by cross-presentation.15 We found a 3-fold reduction in the percentage of OT-I T cells in the liver following cDC depletion (Fig. 5C). The number of OT-I T cells in the DT-treated group and the CFSE profile indicated that the T cells did not proliferate. Our findings demonstrated that cDCs are required for the proliferation of intrahepatic OT-I T cells following infection with LMova.
To determine if the T cell proliferation seen in cDC-depleted animals that received ovalbumin was specific to the liver, we also analyzed splenocytes. Consistent with a prior report, cDC depletion abrogated OT-I proliferation in the spleen (data not shown).16 With the addition of systemic CD40 stimulation, spleen OT-I T cells remained CFSEhigh, indicating that they had not divided (Fig. 5D). In contrast, liver OT-I T cells of DT-treated mice became mostly CFSElow/neg (Fig. 5A). Spleen OT-I T cells were also found to be CFSEhigh 48 hours following ovalbumin injection; likewise, OT-I T cells in the lymph nodes were CFSEhigh 48 and 72 hours after antigen delivery, whereas in the liver OT-I T cell CFSE dissolution was observed (data not shown).
Accumulation of OT-I T Cells in the Liver Depends on Antigen Presentation Capacity of cDCs.
To assess if the decreased frequency of proliferating OT-I T cells observed in the liver on depletion of cDCs actually depended on cross-presentation and not another DC function such as cytokine production, we performed a rescue experiment. Flt3L-expanded wild-type DCs or MHC class I–deficient (β2m−/−) DCs were adoptively transferred into DT-treated or saline-treated CD11c-DTR transgenic mice, and the frequency of liver OT-I T cells was determined. Adoptive transfer of wild-type DCs partially reversed the effect of cDC depletion by increasing the percentage of liver OT-I T cells almost 4-fold over the DT-only group (Fig. 6A). Adoptive transfer of β2m−/− DCs had no effect, indicating that the APC activity of cDCs is the major factor involved in the accumulation of OT-I T cells in the liver following in vivo cross-presentation of ovalbumin protein. This finding was not a result of the differential ability of the 2 DC groups to traffic to the liver because adoptively transferred CFSE-labeled Flt3L-expanded wild-type or β2m−/− cDCs constituted equal percentages of recipient hepatic lymphocytes (Fig. 6B).
cDCs Determine Activation State and Effector Function of Liver OT-I T Cells.
To establish the effect of cDC depletion on the functional phenotype of antigen-specific T cells, we analyzed OT-I T cell expression of activation markers and production of intracellular IFN-γ. OT-I T cells from cDC-depleted animals treated with ovalbumin were less activated, as indicated by greater CD62L staining, which was not altered by CD40 stimulation (Fig. 7A). CD44 expression was not changed by cDC depletion. In nondepleted animals, intracellular cytokine analysis revealed that 48% of the OT-I T cells stained positive for IFN-γ, which increased to 87% with CD40 stimulation (Fig. 7B). Strikingly, depletion of cDCs in vivo decreased the percentage of IFN-γ+ OT-I T cells to 12%, and this was not rescued by CD40 stimulation. There was no difference in intracellular staining for perforin following cDC depletion in vivo (data not shown).
Recent investigations have demonstrated that the liver is able to regulate antigen processing and presentation.17 This regulation depends on anatomical and physiological factors such as sluggish blood flow in the hepatic sinusoids and the continuous exposure to large amounts of foreign antigen from the portal circulation. A number of resident liver cells have been shown to possess APC function. In general, liver APCs have been grouped as “professional” APCs, which include DCs and macrophages, and “nonprofessional” APCs, which include hepatocytes, LSECs, and stellate cells. The relevant contribution of DCs and other hepatic APCs to cross-presentation has been unclear.18 In an elegant orthotopic liver transplant model where microchimerism was eliminated by bone marrow transplantation, the liver was shown to be an excellent priming site for CD8+ T cells by direct presentation of peptide antigen by liver parenchymal cells.19 However, antigen presentation was restricted to non–bone-marrow-derived cells of the liver, thereby eliminating the potential role of DCs. Notably, cross-presentation was not addressed in the liver transplant model. T cell activation, albeit abortive, has been described in transgenic mice where MHC alloantigen expression is restricted to the liver.20 In that model, the antigen may have been confined to hepatocytes, as it could not be detected on DCs by flow cytometry or immunohistochemistry. Thus, the relative role of cross-presentation by liver DCs in generating T cell immune responses in the liver has been uncertain.
Our in vitro findings demonstrated the critical role of cDCs in cross-presentation by NPCs and their redundant role in direct antigen presentation. We then extended our findings by using an in vivo model of cDC depletion to account for potential APC function of hepatocytes and other components of NPCs that may have been excluded in our cell preparations. Furthermore, the in vivo model accounted for the unique architecture of the liver, which may provide selective access of APCs to T cells or antigen.21 We primarily chose to use soluble protein antigen to avoid restricting antigen availability, and to focus on cross-presentation, a pathway that is critical in generating an adaptive immune response to both pathogens and malignancy.22–26 We also provided a potent maturation stimulus with CD40 stimulation, which up-regulates costimulatory molecules and enhances IL-12 production to promote CD8+ T cell expansion and differentiation.27–29In vivo cDC depletion decreased the intrahepatic proliferation and accumulation of adoptively transferred OT-I T cells following ovalbumin administration (Fig. 5A). CD40 stimulation did not rescue the ability of the liver to present ovalbumin administered intravenously (Fig. 5A) nor did nonspecific systemic inflammation by concomitant LM infection (data not shown). The requirement of cDCs for maximal T cell proliferation in vivo depended on their cross-presenting capability because only adoptively transferred wild-type DCs and not β2m−/− DCs restored T cell proliferation (Fig. 6). It was surprising that cDC depletion in vitro and in vivo markedly impaired T cell proliferation because LSECs, stellate cells, and Kupffer cells, which are components of NPCs, have been reported to cross-prime T cells.3, 5, 8, 9 There was no inherent defect in the direct antigen-presenting ability of these accessory APCs in vitro because they induced T cells to proliferate when cDCneg NPCs were loaded with peptide instead of protein (Fig. 2B).
Recently, it has been reported that CD8α+ DCs are required for efficient entry of LM into the spleen and that the previously reported notion that CD11c+ cells are absolutely required for CD8+ T cell priming may have been confounded by the dynamics of LM infection in the spleen in the absence of cDCs.7, 30 In addition, it was demonstrated that adoptive transfer of LM-infected macrophages could induce the proliferation of OT-I T cells in the spleens of cDC-depleted mice. In contrast, LM entry into the liver was not affected by cDC depletion, but the investigators did not analyze the liver for the presence of proliferating T cells. We found that the presence of cDCs in the liver was necessary for the proliferation of antigen-specific T cells during infection (Fig. 5C).
In the absence of cDCs, liver OT-I T cells still proliferated somewhat as indicated by the dissolution of CFSE and their increased frequency over T cell controls (Fig. 5A). This has not been observed in the spleen in previous studies using the CD11c-DTR mouse.7, 16, 31, 32 Instead, cross-presentation of antigen in multiple different forms was completely abrogated by the toxin-mediated depletion of cDCs. These studies, however, focused primarily on the spleen for the detection of antigen-specific T cell proliferation. Because the liver is a unique immunological organ that harbors other APCs such as LSECs, hepatocytes, and stellate cells that do not exist in lymphoid organs, it might be that a low level of T cell proliferation can be induced in the liver in the absence of cDCs. This seemed to be the case in our model, as we did not detect any T cell proliferation in the spleen (Fig. 5D) or lymph node, even though the magnitude of cDC depletion was similar in these compartments (data not shown). In addition to their reduced proliferation, the liver T cells of cDC-depleted animals were functionally tolerogenic, as they were less activated and less likely to produce IFN-γ (Fig. 7). To exclude the possibility that T cell priming occurred in the spleen and the cells merely trafficked to the liver, we repeated the experiments in mice subjected to splenectomy and found no change in the absolute number, CFSE dissolution, or activation state of liver OT-I T cells 3 days following systemic ovalbumin administration (data not shown). Furthermore, with cDC depletion, OT-I T cells were dividing only in the liver, and this was not secondary to earlier activation in the spleen or lymph node as even 48 hours following ovalbumin administration, OT-I T cells were dividing in the liver but not in the spleen or lymph node (data not shown). Taken together, our findings demonstrate that although liver cDCs may not be completely necessary to stimulate the proliferation of CD8+ T cells by cross-presentation, T cell activation state and effector function critically depend on cDCs. The presence of cDCs is also necessary for the maximal accumulation of OT-I T cells in the liver, and it is possible that the lack of cDCs leads to apoptosis of the proliferating T cells. Thus, altered cDC function is a potential mechanism of T cell tolerance in the liver. Such organ-specific dependence of cDCs for antigen presentation has also been shown for natural killer T (NKT) cells, where cDC presentation of the glycolipid α-galactosylceramide in vivo is critical for splenic but not hepatic NKT cells.33
Several liver diseases have been shown to subvert DC function.4, 34 Defective DC function has been proposed as an explanation for the inefficient immune response against hepatitis C virus infection.35, 36 Patients with hepatocellular carcinoma have circulating DCs that are relatively immature and demonstrate reduced allostimulatory capacity.37, 38 Our findings indicate that hepatic cDCs play a critical role in generating a maximal effector antigen-specific CD8+ T cell response by cross-presentation. In the absence of effective cDCs, the balance between tolerance and immunity in the liver may be biased toward tolerance, thereby allowing malignancy and certain pathogens to evade immune detection. An understanding of this balance is necessary to develop therapeutic interventions based on liver DC immunomodulation.
We gratefully acknowledge Jan Hendrikx, Vinod Sahi, Patrick Anderson, and the staff of the Memorial Sloan-Kettering Flow Cytometry Core for their advice and assistance.