Hepatic dendritic cell subsets in the mouse

Authors


Abstract

The CD11c+ cell population in the non-parenchymal cell population of the mouse liver contains dendritic cells (DC), NK cells, B cells and T cells. In the hepatic CD11c+ DC population from immunocompetent or immunodeficient [recombinase-activating gene-1 (RAG1)–/–] C57BL/6 mice (rigorously depleted of T cells, B cells and NK cells), we identified a B220+ CD11cint subset of ‘plasmacytoid’ DC, and a B220 CD11c+ DC subset. The latter DC population could be subdivided into a major, immature (CD40lo CD80lo CD86lo MHC class IIlo) CD11cint subset, and a minor, mature (CD40hi CD80hi CD86hi MHC class IIhi) CD11chi subset. Stimulated B220+ but not B220 DC produced type I interferon. NKT cell activation in vivo increased the number of liver B220 DC three- to fourfold within 18 h post-injection, and up-regulated their surface expression of activation marker, while it contracted the B220+ DC population. Early in virus infection, the hepatic B220+ DC subset expanded, and both, the B220+ as well as B220 DC populations in the liver matured. In vitro, B220 but not B220+ DC primed CD4+ or CD8+T cells. Expression of distinct marker profiles and functions, and distinct early reaction to activation signals hence identify two distinct B220+ and B220 subsets in CD11c+ DC populations freshly isolated from the mouse liver.

Abbreviations:
B6:

C57BL/6

NPC:

Non-parenchymal cells

MACS:

Magnetic cell sorting

GF:

Germ-free

MCMV:

Murine cytomegalovirus

RAG:

Recombinase-activating gene

αGalCer:

α-Galactosylceramide

IFN-I:

Interferon type I

1 Introduction

Hepatic dendritic cells (DC) are supposed to mediate tolerance induction, as well as the intrahepatic induction and restimulation of specific T cell immunity. The key role of the liver in tolerance induction has been reviewed 13. Intrahepatic DC activation and expansion can override hepatic tolerance induction, as liver allograft acceptance is abrogated by pretreating donors with the DC-expanding growth factor Flt3 ligand. It is of interest to define hepatic DC subsets and characterize their surface phenotype and function to test whether these contrasting effects of the liver on specific T cell immunity can be assigned to distinct liver DC subsets.

In this paper we identify B220+ and B220 DC subsets in the mouse liver. Particular care is taken to exclude from the analysis CD11c+ cells in hepatic non-parenchymal cell (NPC) populations that do not belong to DC lineages. We determine the surface phenotype of mouse hepatic B220+ and B220 DC subsets in the liver, and extend the analysis to subsets within the B220 DC subset. Interferon type I (IFN-I) release in response to virus infection is analyzed in vitro using cell sorter-purified hepatic DC subsets. The early changes in hepatic DC subsets and their surface phenotype after in vivo NKT cell activation or virus infection are followed. The data indicate that B220+ and B220 CD11c+ DC are distinct hepatic DC subsets in the mouse while the subset status of hepatic CD11b+ and CD8α+ DC remains uncertain.

2 Results

2.1 CD11c+ non-parenchymal cells in the liver

We isolated 6×106–7×106 NPC from the liver of a normal, young adult C57BL/6 (B6) mouse (Table 1). Similar numbers of hepatic NPC were recovered from standard pathogen-free (SPF) and germ-free (GF) B6 mice, and from immunocompetent and immunodeficient [recombinase-activating gene-1 (RAG1)–/–] B6 mice. Approximately 20% of the cells within the hepatic NPC population were CD11c+ (Table 1). A two- to threefold increase in CD11c+ cells was found in hepatic NPC populations from immunodeficient RAG1–/– B6 mice (Fig. 1A). Surface expression of CD11c by liver NPC showed a broad spectrum including subsets with intermediate (CD11cint) and high (CD11chi) CD11c surface expression (Fig. 1A). Less than 10% of hepatic CD3+ T cells were CD11c+. In contrast, a large fraction (50–80%) of hepatic DX5+ NK cells were CD11cint; this subset was strikingly expanded in RAG1–/– B6 mice (Fig. 1A). Few CD11c+ CD19+ B cells were found in the mouse liver (data not shown).

Hence, in addition to DC, hepatic CD11c+ NPC populations contain NK cells, B cells and T cells. We therefore rigorously depleted (by negative selection in vitro, antibody treatment in vivo, and/or the use of RAG1–/– B6 mice) T cells, B cells and NK cells from hepatic CD11c+ NPC populations before analyzing DC populations of the liver. In addition, NK and T/B cells were gated out electronically in four-color flow cytometry (FCM) analyses.

Figure 1.

(A) CD11c+ hepatic NPC. NPC were prepared by liver perfusion in situ and digestion ex vivo followed by centrifugation through a Percoll gradient. Hepatic NPC were isolated from normal B6 or immunodeficient RAG1–/– B6 mice. The hepatic NPC were surface-labeled by mAb binding CD11c, DX5 or CD3ϵ. Isotype-matched control antibodies were used to set markers and quadrants. (B) DC subsets in liver and spleen. Hepatic NPC and spleen cells were depleted from T, B and NK cells using anti-CD5, anti-CD19 and anti-DX5 mAb-conjugated microbeads and the MACS system. The negative fractions were surface-stained with PE-conjugated mAb binding CD11c and FITC-conjugated mAb binding B220. For the analyses, CD11c+ cells were gated on the CD11c/SSC dot plot. Isotype-matched control antibodies were used to set the gates. (C) Hepatic CD11c+ DC subsets. Hepatic NPC from B6 mice were surface-stained with FITC-conjugated mAb binding B220, allophycocyanin-conjugated mAb binding CD11c, PE-conjugated mAb binding CD3ϵ, CD19 or DX5, and biotinylated mAb binding CD8α or CD11b. Cells were analyzed by four-color FCM with gates set on CD11c+ CD3ϵ CD19 DX5 cells. CD8α or CD11b surface expression of the B220+/ B220 DC subsets is shown. Isotype-matched control antibodies were used to set the gates and quadrants. Data from a representative experiment (out of four individual experiments) are shown.

Table 1. Mouse hepatic NPC
Cellsa)×10–5 (± SEM)/mouseb)%
  1. a) NPC isolated from the perfused, collagenase-digested liver by Percoll density separation.

  2. b) Mean ± SEM of pooled data from four independent experiments (with three mice per group analyzed) are shown.

  3. c) Mean % (± SEM) of DC within the hepatic DC population is shown.

NPC recovered65.5±5.2 
CD11c+ NPC12.9±0.818.5
CD11c+ DX5+ NK cells 2.2±0.33.4
CD11c+ DX5 CD3 DC 9.9±1.214.8
Subset A: B220 CD11chi DC 1.1±0.111.1±0.9c)
Subset B: B220 CD11cint DC 5.2±0.753.0±8.3c)
Subset C: B220+ CD11cint DC 3.5±0.635.6±6.9c)

2.2 Hepatic CD11c+ DC subsets

We searched for markers that define subsets within hepatic CD11c+ (CD3 CD19 DX5) DC populations from normal or RAG1–/– B6 mice. The CD45 isoform B220 defined a B220 subset (65–70%) that contained CD11chi cells (subset A, 10–15% of hepatic DC) and CD11cint cells (subset B, 45–50% of hepatic DC); and B220+ CD11cint cells (subset C, 30–35% of hepatic DC) (Fig. 1B). Similar DC subsets were found in the spleen of immunocompetent and immunodeficient B6 mice, but the relative and absolute numbers of DC within the subsets A–C differed between liver and spleen (Fig. 1B and data not shown). The B220 CD11chi DC subset A predominated in the spleen but not in the liver (55–60% versus 15–20%); in contrast, the B220+ or B220 CD11cint DC subsets B or C were strikingly smaller in the spleen than in the liver (15–20% versus 30–35%, and 10–15% versus 45–50%, respectively). Surface expression of the CD11c and B220 molecules thus identifies distinct DC subsets in spleen and liver.

Surface expression of the integrin CD11b and the CD8α coreceptor have been used to define ‘myeloid’ (CD11b+ CD8) and ‘lymphoid’ (CD11b CD8+) DC subsets. We found non-overlapping CD8α+ and CD11b+ subsets in hepatic CD11c+ DC populations (Fig. 1C). CD8α+ CD11b DC were present in hepatic B220+ and B220 DC (Fig. 1C). CD11b+ CD8α DC formed a major subset in B220 but not B220+ hepatic DC (Fig. 1C). Many splenic and hepatic DC were CD8α CD11b (Fig. 1C, data not shown). These marker profiles of splenic and hepatic DC were found in normal and immunodeficient RAG1–/– B6 mice, and in B6 mice raised under SPF or GF conditions (data not shown). Immunohistology studies of the liver showed that CD11c+ B220 DC were predominantly found in periportal fields and around central veins, while CD11c+ B220+ DC were present mainly in the lobular parenchyma but only rarely detected in periportal fields (data not shown). B220+ and B220 DC are thus present in spleen and liver of mice, but the population size of DC within each subset strikingly differs between the two organs. As the B220 marker most clearly identified hepatic DC subsets, we asked whether these subsets express a distinct functional marker.

2.3 Selective induction of IFN-I release in hepatic B220+ DC by virus infection

We tested whether the splenic and hepatic B220+ DC have the unique ability to produce large amounts of IFN-I in response to virus infection, as reported in mice 4, 5 and humans 6, 7. Freshly isolated, cell sorter-purified B220+ or B220 CD11c+ DC subpopulations from spleen or liver were infected with the murine cytomegalovirus (MCMV) in vitro. IFN-I release during a 24-h post-infection incubation period was determined using a sensitive bioassay. B220+ but not B220 DC infected with MCMV produced IFN (Fig. 2). An IFN-I-neutralizing antibody blocked 80–90% of the antiviral activity, indicating that the released IFN-I mediates the antiviral activity (data not shown). The preparative isolation of B220+ DC did not stimulate their ‘spontaneous’ IFN-I release. Virus infection or stimulation with CpG-containing oligonucleotides, but not stimulation with poly(I/C) induced IFN-I release (data not shown). Splenic B220+ DC reproducibly produced three- to sixfold more IFN-I per DC than hepatic DC (Fig. 2). These data support the notion that the B220 marker identifies a functionally distinct mouse DC subset.

Figure 2.

B220+ but not B220 DC produce IFN-I. CD11chi/int B220 (subset A/B) and CD11cint B220+ (subset C) populations from hepatic NPC or spleen cells of B6 mice were isolated by cell sorting to >98% purity. Cells (105) of these subsets were infected with MCMV (3×105 PFU) for 1 h at 37°C, washed and cultured in flat-bottom 200-μl 96-well plates in triplicates. Non-infected cells were used as control. Supernatants were collected after 24 h for IFN-I detection by a sensitive bioassay (described in Sect. 4.3). Mean data from a representative experiment (out of two independent experiments, 14 mice per experiment) ± SEM are shown.

2.4 The surface phenotype of hepatic DC within the B220+ and B220 subset

The minor CD11chi subset A in the hepatic B220 DC population was more mature than the major CD11cint subset B, because a major fraction of DC in subset A expressed the CD40, CD80 and CD86 costimulator molecules on the surface together with high levels of MHC class II molecules (Fig. 3). In contrast, only few DC from subset B expressed low levels of CD40, CD80 and CD86 costimulator and intermediate levels of MHC class II molecules on the surface (Fig. 3). A similar surface phenotype prevailed in hepatic B220+ CD11cint DC of subset C (Fig. 3). CD8+ cells were found in all three subsets of hepatic DC; B220+ CD11cint DC (subset C) contained a clearly defined CD8+ subset, while a broad range of CD8 surface expression made it difficult to delineate CD8+ subsets in hepatic B220 DC populations (Fig. 3). A ‘mature’ CD11chi B220 DC subset A predominates the splenic DC population while an ‘immature’ CD11cint B220 DC subset B predominates the hepatic DC population (Fig. 1B). Surface expression of CD11b, CD8α, CD40, CD80, CD86 and MHC class II by the splenic DC subsets A–C showed a profile similar to that of the respective hepatic DC subsets (data not shown).

DX5+ cells as well as all NK cell-mediated functions tested (cytolytic activity against YAC-1 targets; poly(I/C)- or IL-12-induced IFN-γ release) were eliminated when NK cells were depleted from hepatic NPC invivo (by repeated injections of anti-Asialo-GM1 antibody) or in vitro [by treatment with anti-Asialo-GM1 antibody plus C', or depletion of DX5+ cells by magnetic cell sorting (MACS)] (data not shown). NK1.1+ DC were detected within these NK cell-depleted CD11c+ liver NPC populations. These DC were exclusively found in hepatic B220 CD11cint DC populations (subset B) and comprised 60–70% of the cells in this subset; no NK1.1+ DC were found in the B220+ CD11cint subset C or B220 CD11chi subset A (Fig. 4A). NK1+ DC were not cytolytic (Fig. 4B), but primed CD4+ T cell responses in vitro (Fig. 4C). Hepatic NK1.1+ CD11c+ DC showed low CD40 and intermediate CD86 and MHC class II surface expression (data not shown).

Figure 3.

Surface phenotype of hepatic DC subsets. Hepatic NPC were depleted of T/B cells and NK cells using anti-CD5, anti-CD19 and anti-DX5 mAb-conjugated microbeads and MACS. The negative fraction was stained with PE-conjugated anti-CD11c mAb, FITC-conjugated anti-B220 mAb, and a third biotinylated anti-CD8α, anti-CD40, anti-CD80, anti-CD86 or anti-MHC class II (I-Ab) mAb. Biotinylated antibody staining was followed by SA-Red 670 staining. Gated CD11chi B220 (subset A), CD11cint B220 (subset B), and CD11cint B220+ (subset C) cell populations were analyzed for expression of the third marker. Isotype-matched control antibodies and a second layer (SA-Red) were used to set gates and markers. Data from a representative experiment (out of four individual experiments, using three mice per experiment) are shown.

Figure 4.

Hepatic NK1.1+ DC. (A) Identification of hepatic NK1+ B220 CD11cint DC. The T/B cell- and NK cell-depleted liver NPC from B6 or RAG1–/– mice were surface-stained with PE-labeled anti-CD11c mAb, FITC-labeled anti-B220 mAb, and biotinylated anti-NK1.1 mAb. Biotinylated antibody staining was followed by SA-Red 670 staining. Gates were set on CD11chi B220 (subset A), CD11cint B220 cells (subset B), and CD11cint B220+ (subset C) cell populations which were analyzed for surface expression of the NK1 marker. Data from a representative experiment (out of four individual experiments, using four mice per experiment) are shown. (B) CD11c+ NK1.1+ DC show no cytotoxicity against YAC-1 targets. The cytolytic activity against YAC-1 targets of FACS-sorted NK1+ B220 DX5 CD11cint hepatic DC versus NK1+ DX5+ hepatic NK cells was tested in a 4-h cytotoxicity assay. (C) CD11c+ NK1.1+ DC can prime CD4+ T cells in vitro. Co-cultures were set up of FACS-sorted, peptide-pulsed NK1+ B220 DX5 CD11cint hepatic DC or NK1+ DX5+ hepatic NK cells with naive OT-II CD4+ T cells. After a 48-h incubation, the ability of this hepatic DC subset to prime T cells was tested (by measuring IFN-γ release).

2.5 Changes in hepatic DC subpopulations in response to NKT cell activation

Injection of the CD1d-binding glycolipid α-galactosylceramide (αGalCer) into mice activates hepatic NKT cells and DC 8, 9. All hepatic DC express CD1d on the surface (data not shown). Within 18 h post-injection, the number of hepatic DC increased moderately (Table 2). The CD11cint B220+ DC subset C strikingly decreased in absolute and relative numbers, while the CD11chi B220 DC subset A increased two- to threefold in number (Table 2, Fig. 5A). These changes in DC numbers and subset distribution in response to i.v. or i.p. αGalCer injection were more prominent in hepatic than splenic DC populations, presumably because the livers harbors a larger αGalCer-reactive NKT cell population than the spleen (Fig. 5A). Hence, within hours after αGalCer injection, the hepatic B220 CD11c+ DC subsets A and B are activated and expand while the B220+ CD11cint DC subset C (though also activated) contracts. Following αGalCer injection, the hepatic DC up-regulated CD40, CD86, CD86 and MHC class II surface expression (Fig. 5B). Similar changes were observed in splenic DC subsets from αGalCer-injected mice (data not shown). Control RAG1–/– or MHC class I-deficient (β2m–/–) B6 mice injected with αGalCer showed no changes in the hepatic DC subsets (data not shown).

Table 2. Hepatic CD11c+ DC subsets in αGalCer-injected B6 mice
CellsNon-injected (×10–5/mouse)%αGalCer-injected (×10–5/mouse)%
  1. a) NPC isolated from the perfused, collagenase-digested liver by Percoll density separation.

  2. b) Mean ± SEM of pooled data from two independent experiments (with three mice per group analyzed) are shown.

  3. c) NPC depleted of T, B and NK cells by MACS.

NPC recovereda)56.5±11.5b) 127±24 
CD11c+ DCc) 9.0±1.4  10.9±3.5 
Subset A: B220CD11chi DC 1.0±0.1 11.1±0.93.2±1.027.0±5.0
Subset B: B220CD11cint DC 4.9±0.7 53.0±4.36.8±2.254.0±9.0
Subset C: B220+CD11cint DC 3.1±0.5 35.6±2.80.9±0.38.5±1.5
Figure 5.

(A) DC subsets in liver and spleen from αGalCer-treated mice. B6 mice injected i.p. with 1 μg αGalCer were killed 18 h post-injection. Vehicle-injected mice were used as control. Hepatic NPC and spleen cells were depleted of T/B cells and NK cells. Negative fractions were surface-stained for FCM analysis with FITC-labeled anti B220 mAb and PE-labeled anti-CD11c mAb. For analysis, CD11c+ cells were gated on CD11c/SSC dot plot. Isotype-matched control antibodies were used to set the gates. Data from a representative experiment (out of two independent experiments with three mice per group) are shown. (B) Hepatic DC activation after αGalCer injection into B6 mice. B6 mice injected i.p. with 1 μg αGalCer were killed 18 h post-injection. Vehicle-injected mice were used as control. The negative fractions from the liver were surface-stained with PE-conjugated anti-CD11c mAb, FITC-conjugated anti-B220 mAb, and biotinylated anti-CD40, anti-CD80, anti-CD86 or anti-MHC class II (I-Ab) mAb. Biotinylated antibody staining was followed by SA-Red 670 staining. CD11c+ cell populations gated on CD11c/SSC dot plot, and CD11chi B220 (subset A), CD11cint B220 (subset B) and CD11cint B220+ (subset C) marker expression were analyzed for the third marker expression. Isotype-matched control antibodies were used to set the gates and markers. Data from a representative experiment (out of two independent experiments, three mice per group) are shown.

2.6 Changes in the hepatic DC population after virus infection

MCMV infection in vitro induces IFN-I release by B220+ CD11cint subset C DC from spleen and liver (Fig. 2). We tested whether MCMV infection in vivo induces changes in hepatic DC (sub)populations similar to NKT cell activation. MCMV infection of immunocompetent or immunodeficient (RAG1–/–) B6 mice leads to a moderate increase in the number of hepatic DC 36 h post-infection (Table 3). This resulted mainly from an increase in B220+ CD11cint subset C DC in the liver (Table 3). In infected mice, the hepatic subset C DC showed evidence of activation, i.e. up-regulation of CD40 and CD86, and to a lesser extent CD80 (Fig. 6). Hepatic DC of subset A up-regulated surface expression of CD40 and CD86, but not CD80 (Fig. 6). Unexpectedly, MHC class II surface expression by liver DC in subset C was down-regulated early after infection (Fig. 6).

Similar changes were seen in immunocompetent B6 mice and congenic, immunodeficient RAG1–/– B6 mice, supporting the notion that the innate but not the specific immune system induces these changes in DC in the early stage of virus infection (Fig. 6). Similar changes were evident early after virus infection in splenic DC populations of subset A and C (data not shown). Hence, within 36 h after virus infection, the IFN-I-producing subset C (B220+ CD11cint) expands or is recruited to the liver, and is activated. This response differs from the early response to NKT cell activation, in which a depletion of subset C is observed in the liver DC population.

Table 3. NPC of the liver in MCMV-infected B6 mice
Cell subsetNon-infected (×10–5/mouse)%MCMV-infected (×10–5/mouse)%
  1. a) NPC isolated from perfused, collagenase-digested liver by Percoll density separation.

  2. b) Mean ± SEM of pooled data from two independent experiments (with three mice per group analyzed) are shown.

  3. c) NPC depleted of T, B and NK cells by MACS.

Total NPC recovereda)65±2.5b) 115±2.0 
CD11c+ DCc)10.7±2.8 16.2±1.2 
Subset A: B220CD11chi DC1.2±0.211.0±0.51.4±0.19.0±1.0
Subset B: B220CD11cint DC6.6±1.462.0±3.58.2±0.650±5.5
Subset C: B220+CD11cint DC2.9±0.627.5±1.56.6±0.541±4.0
Figure 6.

Hepatic DC activation in MCMV-infected mice. Immunocompetent B6 mice or immunodeficient RAG–/– mice were infected i.p. with 3×104 PFU MCMV and killed 36 h later. Hepatic NPC were depleted of T/B cells and NK cells. The negative fractions from liver NPC were surface-stained as described in the legend of Fig. 5. The gates were set on CD11chi B220 and CD11cint B220+ subsets. Data from a representative experiment (out of two independent experiments with three mice per group) are shown.

2.7 In vitro priming of CD4+ and CD8+ T cells by hepatic B220 but not B220+ DC

Purified splenic CD4+ T cells (from OT-II mice) and CD8+ T cells (from OT-I mice) were co-cultured in vitro for 2–5 days with peptide-pulsed, splenic or hepatic B220 or B220+ CD11c+ DC. B220 but not B220+ DC from spleen or liver supported the specific in vitro priming of TCR-transgenic CD4+ or CD8+ T cells (Fig. 7). Proliferation as well as cytokine (IFN-γ, IL-2) release were induced by B220 DC, with the splenic DC being more potent APC than hepatic DC.

Figure 7.

In vitro priming of OT-I and OT-II T cells by DC subsets. Hepatic and splenic NPC were depleted of T/B cells and NK cells by MACS using anti-CD5-, anti-CD19- and anti-DX5-conjugated microbeads. CD11chi B220 and CD11c B220+ cell populations were isolated by cell sorting. DC (1×104) were loaded with increasing doses of either the Kb-binding OVA257–264 peptide SIINFEKL or the Ab-binding OVA323–339 peptide ISQAVHAAHAEINEAGR for 3 h, washed twice and co-cultured with 1×105 OT-I or OT-II cells in 96-well plates for 4 days. Release of IFN-γ and IL-2 into the supernatants after 4 days of co-culture was measured by ELISA (A). The expansion of OT-I and OT-II T cells co-cultured with the different DC subsets loaded with increasing doses of OVA peptides was measured by counting cells at day 4 (B).

3 Discussion

3.1 Identifying DC in mouse hepatic CD11c+ NPC populations

The αx-chain of the gp150/95 integrin (CD11c/CD18) is widely used to identify and preparatively isolate murine DC. Because CD11c is also expressed on the surface of NK cells, B cells and T cells, these CD11c+ cell subsets have to be distinguished from DC. CD3 or CD19 expression readily excludes T and B cells from the analysis of CD11c+ cells. The mAb DX5, which binds the α2-chain of the α2β1 integrin very-late antigen (VLA)-2 (CD49b/CD29), is expressed by most murine NK cells but not by DC. Only by depleting NK cells in vivo or in vitro, using T/B cell-deficient RAG1–/– B6 mice and/or depleting T and B cells in vitro, we could define and preparatively isolate hepatic CD11c+ DC.

In the first part of the study, we analyzed exclusively freshly isolated liver DC from normal mice. Animals were not pretreated with Flt3 ligand to expand hepatic DC, and the isolated DC were not exposed to cytokines in vitro or cultured before the analyses. Collagenase treatment, a well-established technique to isolate DC from tissues, was used to obtain NPC from the liver. The yield of NPC obtained from the liver without collagenase treatment was low, and the DC subset distribution was different. Control experiments indicated that collagenase treatment did not change the DC phenotype.

3.2 Plasmacytoid B220+ DC

Natural IFN-I-producing cells have been first discovered in human blood 7. Shortly thereafter, the mouse homologue of the IFN-I-producing cell was identified as the ‘plasmacytoid’ DC (PDC), also termed preDC or preDC2. PDC generated from precursors in mouse bone marrow or blood by Flt3 ligand 1014 develop independently of the T cell lineage 15 in an IFN regulatory factor (IRF)-8-dependent way 16. PDC from the thymus 17, spleen and LN 18, 19 or the liver 20 produce IFN-I after stimulation 4, 5, 11, 21, 22. These DC are long-lived with a slow turnover 19 and express a heterogeneous surface phenotype profile 19. B220+ DC seem to be immature APC 12 or tolerogenic 5, 23. We could not prime CD4+ or CD8+ T cells in vitro with B220+ PDC. Following maturation, PDC may be able to prime Th1 and Th2 T cell responses 24.

3.3 Hepatic DC subsets

DC subset classifications are controversial. CD11b and CD8 are frequently used to identify ‘myeloid’ and ‘lymphoid’ DC characterized by their exclusive CD8 and CD11b surface expression 2530. CD8+ CD11b ‘lymphoid’ and CD8 CD11b+ ‘myeloid’ DC are found in the mouse liver (Fig. 1, 31). The origin of these two DC subsets in lymphoid organs and in the liver is unresolved. Some data indicate that the in vivo differentiation of these DC subpopulations may be unrelated to their experimental generation invitro32. In our attempt to define subsets in freshly isolated hepatic DC populations using these markers, we encountered two problems: (1) a substantial fraction of hepatic DC are CD8 CD11b (Fig. 3); and (2) CD8 and CD11b surface expression is variable in hepatic CD11c+ DC, which makes it difficult to define a subset. These difficulties led to the search for an alternative subset definition for hepatic DC.

In liver NPC populations rigorously depleted of T/B cells and NK cells, we defined B220+ and a B220 DC subsets (Fig. 1, 3, 5). IFN-I production correlated with B220 surface expression, but not with CD8+ or CD4+ surface expression in B220+ DC of subset C, confirming recent reports 5, 20. The CD11cint B220+ DC subset C is prominent in the liver where it represents about a third of the total hepatic DC pool. Its identification is unambiguous. It contains a well-defined CD8α+ subset (20–30% of the DC in the B220+ subset C) but no CD11b+ DC. The hepatic B220 DC subsets A and B also contain CD8α+ cells. In vivo stimulation of hepatic DC, particularly by NKT cell activation, induced a depletion of CD8α+ DC (either by down-modulating marker expression, or by preferential loss of CD8α+ DC). Virus infection activates and expands hepatic B220+ DC in vivo but does not lead to up-regulation of CD8 surface expression.

About half of the B220 DC of subset A and B in the liver were CD11b+ and would be classified as ‘myeloid’ DC, while only a minority (10%) of CD11b B220 DC were CD8+, confirming the heterogeneity of this subset 20. Tissue-specific variability is a problem in the definition of DC subsets because CD11b+ and CD8+ B220 DC subsets are readily identified in splenic but not hepatic DC populations. The surface phenotype and IFN-I release thus define B220+/B220 subsets more easily than CD11b+/CD8+ subsets in hepatic DC.

NK1 (CD161), a member of the NKR-P1 gene family, is a type II transmembrane C-type lectin expressed on the surface of T cells and NK cells as disulfide-linked homodimers. The NK1.1 allelic form to which the mAb PK136 binds is expressed in selected strains of mice, including B6 mice. We found NK1 expression by subset B (CD11cint B220) DC with low/intermediate MHC class II and costimulator expression. More hepatic than splenic DC were NK1+. NK1 expression by hepatic CD11c+ NPC has been described 20, but the NK1+ CD11c+ NPC were considered non-DC and excluded from the study. This selectively excludes 70% of liver DC in subset B from the analysis and makes the DC population structure in the liver seeming more similar to that in the spleen. We demonstrate that the NK1+ CD11c+ DC mediate no NK cell functions but are APC. It remains to be elucidated whether NK1 molecules expressed by DC have a functional role.

3.4 Response of hepatic DC subsets to NKT cell activation and virus infection

The notion that B220+ and B220 DC populations in the liver are functionally distinct subsets is supported by their different early responses in vivo to stimuli that activate hepatic DC, i.e. NKT cell activation or MCMV infection. The expansion/contraction of the absolute and relative sizeof the two subsets and their changes in surface phenotype differed. In particular, the selective increase of IFN-I-producing hepatic B220+ DC shortly after virus infection suggests a role of these immature DC in the early response of this vulnerable organ to virus infection. Factors that regulate these responses in the liver are under investigation. Furthermore, the effects on specific T cell priming by the changes induced in the phenotype of hepatic DC subsets are the focus of ongoing studies.

4 Materials and methods

4.1 Mice

H-2b B6 mice, RAG1–/– B6 mice, and TCR transgenic OT-I and OT-II B6 mice were bred and kept under SPF conditions in the animal colony of the University of Ulm (Ulm, Germany).B6 mice raised under GF conditions were also used. Mice were used at 10–16 weeks of age.

4.2 Isolation of hepatic and splenic mononuclear cells

Hepatic NPC 8 and splenic DC 9 were isolated as described. Hepatic and splenic CD11chi/int B220 and CD11cint B220+ DCused for IFN-I release experiments were isolated by electronic cell sorting to >98% purity.

4.3 Bioassay to measure IFN-I release

The IFN concentration in the cell culture supernatants was determined by an antiviral assay using mouse L929 cells infected with vesicular stomatitis virus as described 33. To confirm the specificity of the antiviral activity, a neutralizing mAb directed against mouse IFN-I was added to the supernatant before addition to the test cells.

4.4 Flow cytometry analyses

Hepatic NPC or spleen cells were depleted of T, B and NK cells by MACS (Miltenyi Biotec) using anti-CD5, anti-CD19 and anti-DX5 MicroBeads (Cat. No. 130–049–301, 130–052–201 and 130–052–501). Cells were washed twice in PBS/0.3% (w/v) BSA supplemented with 0.1% (w/v) sodium azide. Nonspecific binding of antibodies to Fc receptor was blocked by pre-incubating cells with mAb 2.4G2 (Cat. No. 01241D; BD Biosciences) directed against the FcγRIII/II CD16/CD32 (0.5 μg mAb/106 cells/100 μl). After washing, the cells were incubated with 0.5 μg/106 cells of the relevant mAb for 30 min at 4°C, and washed again twice. In most experiments, cells were subsequently incubated with a second-step reagent for 10 min at 4°C. Three- or four-color FCM analyses wereperformed by FACS (Becton & Dickinson, Mountain View, CA). The forward narrow angle light scatter was used as an additional parameter to facilitate exclusion of dead cells and aggregated cell clumps.

The following reagents and mAb were obtained from BD Biosciences: FITC-conjugated anti-CD3ϵ mAb 145-2C11 (Cat. No. 553062), PE-conjugated anti-β-TCR H57, (Cat. No. 03105A), FITC- and PE-conjugated CD49b/pan-NK-cells mAb DX5 (Cat. No. 09944D and 553858), PE- and allophycocyanin-conjugated anti-CD11c mAb HL3 (Cat. No. 553802 and 550261), FITC-conjugated anti-CD4 mAb GK1.5 (Cat. No. 553729), anti-CD8α mAb 53-6.7 (Cat. No. 553031), anti-CD11b mAb M1/70 (Cat. No. 557396), anti-CD80 mAb 16-10A1 (Cat. No. 553768), and anti-CD86 mAb GL1 (Cat. No. 553691), biotinylated anti-CD8α 53–6.7 (Cat. No. 553029), anti-CD11b mAb M1/70 (Cat. No. 01712D), anti-CD40 mAb 3/23 (Cat. No. 09662D) and anti-I-Ab mAb AF6–120.1 (Cat. No. 553550). Second-step reagents used were SA-PerCP (Cat. No. 554064; BD Biosciences) and SA-Red 670 (Cat. No. 19543–024; Gibco-BRL, Berlin, Germany).

4.5 Injection of α-galactosylceramide into mice

The glycolipid αGalCer, a kind gift of Dr. Y. Koezuka (Kirin Brewery Co Ltd, Pharmaceutical Research Laboratory, Gunma, Japan), was dissolved in 0.5 ml PBS and injected i.v. into mice (1 μg/mouse).

4.6 MCMV infection of mice

B6 or RAG1–/– mice were infected by i.p. injection of 3×104 PFU MCMV (a kind gift of Dr. D. Michel, Dept of Virology, University of Ulm). Mice were killed 36 h post-infection. DC (105/well) were infected in vitro by 3×105 PFU MCMV in medium for 1 h at 37°C, 5% CO2, washed and cultured in 200-μl wells in flat-bottom 96-wel plates in triplicates. Non-infected cells were used as a control. After 24 h the supernatants were collected for IFN-I detection.

4.7 T cell priming in vitro

Cells were cultured in 200 μl round-bottom wells in RPMI 1640 medium supplemented with 5% FCS (Cat. No. A15–649; PAA Laboratories, Linz, Austria), 2 mM L-glutamine and antibiotics. DC were pulsed with the indicated doses of either the Kb-binding OVA257–264 peptide SIINFEKL (recognized by the transgene-encoded TCR expressed by OT-I mice), or the Ab-binding OVA323–339 peptide ISQAVHAAHAEINEAGR (recognized by the transgene-encoded TCR expressed by OT-II mice). After 3 h of incubation DC were washed twice, and 1×104 DC/wellwere co-cultured with 1×105 purified responder CD8+ (OT-I) or CD4+ (OT-II) T cells. After a 48-h co-culture, IFN-γ and IL-2 were detected in supernatants by double-sandwich ELISA. All purified capture and biotinylated antibodies and all recombinant mouse cytokines standards were from BD Biosciences. Extinction was analyzed at 405/490 nm on a TECAN micro plate-ELISA reader (TECAN, Crailsheim, Germany) using the EasyWin software (TECAN).

Acknowledgements

This work was supported by grants from the Deutsche Forschungsgemeinschaft (DFG Re549/10–1) and the University of Ulm (IZKF-E3) to R.S. and J.R. We greatly appreciate the expert technical assistance of Tanja Güntert and Katja Ölberger, the help of Dr. P. Conradt with cell sorting, the MCMV from Dr. D. Michel (Dept. of Virology, Ulm), and Dr. Y. Koezuka(Kirin Brewery Co Ltd, Pharmaceutical Research Laboratory, Gunma, Japan) for the αGalCer. We gratefully acknowledge the helpful comments of Dr. M. J. Wick (Gothenberg, Sweden).

Footnotes

  1. 1

    WILEY-VCH

  2. 2

    WILEY-VCH

  3. 3

    WILEY-VCH

  4. 4

    WILEY-VCH

  5. 5

    WILEY-VCH

  6. 6

    WILEY-VCH

  7. 7

    WILEY-VCH

Ancillary