- Split View
-
Views
-
Cite
Cite
Babak Baban, Anna M. Hansen, Phillip R. Chandler, Anna Manlapat, Adam Bingaman, David J. Kahler, David H. Munn, Andrew L. Mellor, A minor population of splenic dendritic cells expressing CD19 mediates IDO-dependent T cell suppression via type I IFN signaling following B7 ligation, International Immunology, Volume 17, Issue 7, July 2005, Pages 909–919, https://doi.org/10.1093/intimm/dxh271
- Share Icon Share
Abstract
By ligating CD80/CD86 (B7) molecules, the synthetic immunomodulatory reagent CTLA4-Ig (soluble synthetic CTLA4 fusion protein) induces expression of the enzyme indoleamine 2,3-dioxygenase (IDO) in some dendritic cells (DCs), which acquire potent T cell regulatory functions as a consequence. Here we show that this response occurred exclusively in a population of splenic DCs co-expressing the marker CD19. B7 ligation induced activation of the transcription factor signal transducer and activator of transcription (STAT1) in sorted CD19+, but not CD19NEG, DCs. STAT1 activation occurred even when DCs lacked receptors for type II IFN (IFNγ); however, STAT1 activation and IDO up-regulation were not observed when DCs lacked receptors for type I IFN (IFNαβ). Thus, IFNα, but not IFNγ, signaling was essential for STAT1 activation and IDO up-regulation in CD19+ DCs following B7 ligation. Consistent with these findings, B7 ligation also induced sorted CD19+, but not CD19NEG, DCs to express IFNα. Moreover, recombinant IFNα induced CD19+, but not CD19NEG, DCs to mediate IDO-dependent T cell suppression, showing that IFNα signaling could substitute for upstream signals from B7. These data reveal that a minor population of splenic DCs expressing the CD19 marker is uniquely responsive to B7 ligation, and that IFNα-mediated STAT1 activation is an essential intermediary signaling pathway that promotes IDO induction in these DCs. Thus, CD19+ DCs may be a target for regulatory T cells expressing surface CTLA4, and may suppress T cell responses via induction of IDO.
Introduction
Induced or spontaneous indoleamine 2,3-dioxygenase (IDO) activity suppresses adaptive T cell-mediated immunity in murine models of pregnancy, allogeneic tissue transplantation and inflammatory diseases. These include tumor growth, asthma and autoimmune diseases, such as colitis, insulin-dependent diabetes mellitus, rheumatoid arthritis and multiple sclerosis (1, 2). Human and murine dendritic cells (DCs) acquire potent T cell regulatory properties following induction of IDO expression (1, 3, 4). In mice, IDO expression by murine CD8α+ DCs suppressed delayed-type hypersensitivity responses to tumor-associated peptide antigens (5). Recently, Grohmann and colleagues reported that murine DCs responded to CTLA4-Ig (soluble synthetic CTLA4 fusion protein)-mediated ligation of B7 (CD80/86) molecules by expressing functional IDO, and that the ability of CTLA4-Ig to delay tissue allograft rejection was partly IDO dependent (6). Consistent with this, we found that CTLA4-Ig-mediated blockade of destructive donor allospecific T cell responses was completely dependent on the ability of CTLA4-Ig to induce IDO in recipient mice (7). However, IDO expression was detected exclusively in specific DC subsets with distinct phenotypes (B220+, CD8α+) that acquired potent IDO-dependent T cell regulatory functions following in vivo exposure to CTLA4-Ig (7, 8). Since B7 molecules are expressed widely on a variety of antigen-presenting cells (APCs), these findings suggested that the ability to respond to CTLA4-Ig treatment by expressing functional IDO was restricted to specific populations of splenic DCs co-expressing B220 and/or CD8α.
IFNγ (IFN type II) is a potent inducer of IDO expression in multiple cell types, such as cultured cell lines, tumor cells and physiologic murine splenic CD8α+ DCs (5, 9). IFNγ induces IDO gene transcription in vitro via activation of signal transducer and activator of transcription (STAT1), a member of the signal transducer and activator family of transcription factors (6, 10). However, following in vivo treatment with CTLA4-Ig, splenic DCs from mice deficient for IFNγR expression [IFNγRα-KO (gene deficient) mice] mediated IDO-dependent T cell suppression as efficiently as DCs from wild-type mice, suggesting that IFNγ signaling was not essential for induction of functional IDO activity in DCs following B7 ligation in vivo (8).
In the current study, we show that B7 ligation mediated by CTLA4-Ig induced a highly selective pattern of IFNα (IFN type I) secretion and STAT1 activation restricted to a specific population of splenic DCs. The principal responsive subset comprised a minor DC population expressing the marker CD19. These DCs appeared thus to be similar to CD19+ plasmacytoid DCs that we recently identified as the principal IDO+ regulatory DC population present in tumor-draining lymph nodes (TDLNs) (11).
Methods
Mice
F1[CBA × B6], IDO-deficient (IDO-KO with F1[CBA × B6] backgrounds), BALB/c and BM3 (CBA) TCR transgenic mice used in these studies were bred at the Medical College of Georgia (12, 13). IFNαβR-KO, IFNγR-KO and (background matched) strain 129 wild-type mice were generous gifts from D. Moskofidis (Medical College of Georgia). IFNαβR-KO mice with BALB/c backgrounds were generous gifts from W. Portnoy (University of California, Berkeley, CA, USA). All procedures were carried out with the approval of the Institutional Animal Care and Use Committee.
CTLA4-Ig
Native CTLA4-Ig (non-mutant, catalog no. C4483) and mutant (catalog no. C4358) isotypes of CTLA4-IgG2a were purchased from Sigma (St Louis, MO, USA). Mice were injected with 100 μg CTLA4-Ig (intra-peritoneally) and DCs were incubated with 100 μg ml−1 of CTLA4-Ig. Unless otherwise stated in the text, the native CTLA4-Ig isotype was used for studies described.
Recombinant IFN
Recombinant mouse IFNα (catalog no. 12100-1) and IFNγ (catalog no. 12500-1) were purchased from PBL Biomedical Laboratories (Piscataway, NJ, USA).
Mixed lymphocyte reactions
Mixed lymphocyte reactions (MLRs) were performed essentially as described previously (14). Combinations of responders and stimulators were set up in triplicate wells in a total of 200 μl per well RPMI 1640 medium (catalog no. 15-041-CV; Cellgro, Herndon, VA, USA) supplemented with 10% fetal bovine serum (FBS, Sigma), penicillin (100 IU ml−1), 100 mg ml−1 streptomycin (Cellgro), 2 mM ml−1L-glutamine (Cellgro) and 5 × 10−5 M 2-mercaptoethanol in 96-well round-bottomed plates (Falcon, Bedford, MA, USA). Responder T cells were enriched using nylon wool (15) and used at either 1 × 105 or 5 × 104 cells per well together with equal numbers of fractionated or 5× unfractionated non-irradiated stimulators. Plates were incubated for 72 h at 37°C in a humidified 5% CO2 atmosphere. Wells were pulsed with 0.5 μCi [3H]thymidine ([3H]TdR) in 40 μl RPMI 1640 for the last 6 h of the incubation period. TdR incorporation was measured using the BetaPlate system (Wallac, Newark, NJ).
1-Methyl-tryptophan and 10× tryptophan
1-Methyl-D-tryptophan (1mT; Aldrich, Milwaukee, WI, USA) was added to give a final concentration of 100 μM. L-Tryptophan (L-α-amino-3-indole-propionic acid FW 204.2, Sigma) was used at a final concentration of 245 μM to give 10× the normal concentration used in stock RPMI (24.5 μM final).
Anti-IFN antibody
Monoclonal rat anti-murine IFNα antibody (catalog no. 22100-1) and monoclonal rat anti-murine IFNγ antibody (catalog no. 22500-1) were purchased from PBL Biomedical Laboratories.
Anti-IDO antibody
Polyclonal rabbit anti-murine IDO antibody was prepared by a commercial supplier (Biosource International, Hopkinton, MA, USA). Antisera were raised against two synthetic peptides (KPTDGDKSEEPSNVESRGC and CSAVERQDLKALEKALHD) following conjugation to ovalbumin. Antisera were affinity purified over the first peptide and screened for reactivity by ELISA.
Immunohistochemistry
Tissue sections (5 mm) were prepared from formalin-fixed paraffin-embedded tissues. Following de-paraffinization, sections were washed for 10 min in distilled water. Cytospin preparations of ∼20 000 sorted cells per sample chamber were centrifuged (700 r.p.m., 5 min), air-dried, fixed in 10% formalin and washed twice in PBS. All subsequent procedures were carried out at room temperature (RT). Endogenous peroxidase activity was blocked with hydrogen peroxide (1 : 10 w/PBS, 10 min). Tissue sections were also treated with proteinase K (catalog no. S3020; DAKO, Carpentaria, CA, USA) for 10 min. After two washes in PBS, all preparations were treated with universal blocking reagent at 1 : 10 in distilled water (catalog no. HK085-5K; BioGenex, San Ramon, CA, USA), rinsed in PBS and incubated with either anti-IDO antibody or anti-IFNα antibody (1 : 100 in PBS; 1 h for cytospins, 2 h for tissue sections). After two washes in PBS, preparations were treated with biotinylated goat anti-rabbit Ig (catalog no. HK336-9R, BioGenex). After a 5-min wash in PBS, slides were incubated for 20 min in peroxidase-conjugated streptavidin (catalog no. HK330-9k, BioGenex). IDO-expressing cells were visualized using 3-amino-9-ethylcarbazole chromogen (catalog no. HK121-5K Liquid AEC, BioGenex) for 30 s to 10 min as necessary for optimal staining. Preparations were counterstained with hematoxylin (catalog no. 7221; Richard-Allan Scientific, Kalamazoo, MI, USA) and mounted in Faramount (catalog no. S3025, DAKO). Anti-IDO antibody pre-incubated with neutralizing peptide (1.2 mg antibody : 10 mg peptide) was used as the specificity control.
Immunofluorescence (STAT1 and P-STAT1) staining
Tissue sections and cytospin preparations were prepared as above. To permeabilize, all preparations were incubated in 0.2% Triton X-100 for 5 min at RT. All slides were washed three times for 5 min at RT and then incubated in blocking buffer (20% normal donkey serum, 1% BSA, 0.02% NaN3, 1× PBS) for 45–60 min. Following treatment with the primary antibody [phospho-(Y701)-STAT1 (P-STAT1), antibody catalog no. 9171; Cell Signaling Technology, Beverly, MA, USA] overnight at 4°C, preparations were then washed three times with Tris-buffered saline (TBS) for 5 min each time. All slides were then incubated with the secondary fluorescence-labeled antibody (1 : 100, catalog no 711-166-152; Jackson Immunoresearch Laboratories, West Grove, PA, USA) for 1 h in the dark at RT, washed twice in TBS for 5 min each time in the dark and then counterstained using bis-Benzimide, Hoechst (catalog no. B-2883, Sigma).
Splenic DC isolation
Spleens were harvested into 1% FBS/HBSS. One milliliter of collagenase IV (100 CD units ml−1 in 1% FBS/HBSS-CLS-4; Worthington, Lakewood, NJ, USA) was injected into three areas of each spleen. Injected spleens were then placed in collagenase IV (1 ml per spleen of 400 CD units ml−1 in 1% FBS/HBSS). After incubation (37°C, 30 min), spleens were made into a single-cell suspension and centrifuged (1300 r.p.m., 5 min) and erythrocytes lysed (3 min) in 3 ml of ACK lysing buffer (catalog no. 10-548E; BioWhittaker, Walkersville, MD, USA). Splenocytes were washed twice (10 mM EDTA in Ca/Mg-free PBS) before fractionation (MACS) or sorting (Mo-Flo) as described below.
AutoMACS fractionation
Cell pellets were re-suspended in running buffer (1% BSA, or 2% FCS in 1 mM EDTA in Ca/Mg-free PBS), and anti-murine CD11c microbeads (catalog no. 130-052-001; Miltenyi, Auburn, CA, USA) were added (50 μl ml−1). Following incubation (30 min on ice) in the dark, cells were washed twice and CD11c+ cells were selected using the AutoMACS system. Typically, CD11c+ cells isolated by this procedure were 80–85% pure, while CD11c− cells were >99% pure.
Preparative flow cytometry
Splenocyte cell suspensions were incubated with a cocktail of APC–CD11c (catalog no. 550261; Pharmingen, San Diego, CA, USA) and PE–CD19 (catalog no. 557329, Pharmingen) for 20 min at 4°C. Preparative cell sorting was performed as described (8), using a Mo-Flo four-way flow cytometer equipped with DakoCytomation Summit™ software (DakoCytomation, Ft Collins, CO, USA) to select cells of interest. CD11c+ cell fractions were selected for high purity (>98%), which was achieved by setting sorting gates to collect cells unambiguously stained by CD11c mAb (CD11cHIGH). This procedure sacrificed some DCs with lower CD11c-staining profiles (CD11cLOW), but avoided contamination with macrophages and other cell types whose autofluorescence overlapped the CD11cLOW region. As shown in Results, essentially all IDO-dependent T cell suppressive activities segregated with the unambiguous CD11cHIGH sorted cells that co-expressed CD19, so it was not necessary to include ambiguous CD11cLOW DC populations for the purposes of this study. Sorting gates for CD19 staining were set between distinct populations of stained and unstained cells. All sorted DCs exhibited comparable light scatter properties (FSCHIGH, SSCHIGH) characteristic of large mononuclear cells.
Analytical flow cytometry
Phenotypic analyses of splenic DCs were performed using four-color flow cytometry with dye-conjugated mAbs. DC subsets were identified using a cocktail of mAb to CD11c, B220, CD19 and CD8α and cell-surface markers were identified using PE-conjugated mAb to H2Kb, H-2Ak/Ek, CD80 and CD86 (all from BD Biosciences, San Diego, CA, USA). The CD11c gate was set to match sorting parameters shown in Fig. 1(A) to permit comparisons with Mo-Flo-sorted DC populations and with our previous studies (8).
Reverse transcription–polymerase chain reaction
Analysis of IFNα gene expression in CD11c+-enriched DC population treated or untreated with 100 μg ml−1 CTLA4-Ig was performed using semi-quantitative reverse transcription–polymerase chain reaction (RT–PCR). Total RNA was isolated from cells using RNA STAT-60 (catalog no. CS-110; Tel-Test Inc., Friendswood, TX, USA). A total of 3 ng of RNA was amplified for 40 cycles (IFNα: 94°C for 30 s, 65°C for 1 min and 72°C for 2 min; β-actin: 94°C for 30 s, 52.5°C for 1 min and 68°C for 2 min), following reverse transcription in a one-step reaction (RT–PCR ‘Access’; Promega, Madison, WI, USA). A total of 10 μl of reaction was electrophoresed on a 1% agarose gel. Primers for amplification of specific IFNα subtype transcripts were IFNα1–9, forward CCTGATGGTCTTGGTGGTGATAA and reverse CAGTTCCTTCATCCCGACCAG (16) and for β-actin transcripts were forward AGCAAGAGAGGTATCCTG and reverse CTTTACGGATGTCAACGTC. As controls, 2 × 106 CD11c+ DC cells were infected with influenza A virus strain X31 (courtesy of Graeme Price) at multiplicity of infection 10 in PBS for 1 h at 37°C. Cells were pelleted (5 min, 700 × g) and re-suspended in 0.5 ml RPMI for 4 h. Total RNA was isolated from cells using RNA STAT-60.
Western blots
A total of 106 CD11c+ DCs were enriched by AutoMACS, treated or untreated with 100 μg ml−1 CTLA4-Ig in vitro for 5 h, harvested in cell lysis buffer (PBS, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS, 150 ng ml−1 phenylmethylsulfonylfluoride, 100 ng ml−1 aprotinin) and 30 μg of cell protein was electrophoresed on 10% polyacrylamide gels overlaid with a 5% stacking gel. Protein was quantitated using the bicinchonic acid assay (Pierce, Rockford, IL, USA). Antibody against activated STAT1 (P-STAT1, Tyr701, catalog no. 91H; Cell Signaling Technology) was used in combination with standard ECL techniques.
ELISA
A total of 106 CD11c+ DCs from IDO-WT mice, enriched by AutoMACS, were treated with 100 μg ml−1 non-mutant or mutant CTLA4-Ig in vitro for 5 h. Media were then harvested and measured for IFNα as per manufacturer's instructions (Mouse IFN Alpha ELISA kit; PBL Biomedical Laboratories).
Results
CD19+ DCs mediate IDO-dependent T cell suppression
In a recent study on murine DCs from TDLNs, we reported that DC populations expressing the surface marker CD19 were the only cells that mediated IDO-dependent T cell suppression (11). To examine if CD19 also identified the population of DCs from spleen that mediated IDO-dependent T cell suppression, we sorted CD19+ and CD19NEG DCs from untreated and CTLA4-Ig-treated F1[CBA × B6] mice using a rapid (Mo-Flo) flow cytometer, and then cultured sorted DCs with H-2Kb-specific CD8+ T cells from BM3 TCR transgenic mice to assess their T cell stimulatory properties (Fig. 1). DC populations were gated based on purity criteria (>98% CD11c+, see Methods). This sorting strategy selected DCs expressing relatively high levels of CD11c (CD11cHIGH) and excluded the majority of splenic plasmacytoid DCs, which express relatively low levels of CD11c (17, 18). As we have shown previously, splenic CD11cHIGH DC populations sorted in this way from untreated mice and from IDO-KO mice treated with CTLA4-Ig stimulated vigorous BM3 T cell proliferation (8). In contrast, unfractionated (total) splenocytes (Fig. 1A and B) and sorted DC populations co-expressing B220 or CD8α mediated IDO-dependent T cell suppression (8).
IDO-mediated T cell suppression segregated with sorted DC populations expressing CD19 when prepared from CTLA4-Ig-treated mice (Fig. 1C). Lack of T cell proliferation was due to IDO-mediated suppression because underlying potent stimulatory properties of DCs from CTLA4-Ig-treated mice became evident when excess tryptophan was added to cultures. Identical outcomes were obtained when IDO inhibitor 1mT was added to cultures (data not shown). Sorted CD19NEG DCs from mice exposed to CTLA4-Ig were not suppressive and stimulated robust BM3 T cell proliferation, though their T cell stimulatory properties were slightly enhanced in the presence of excess tryptophan (Fig. 1C). The suppressive effects of CD19+ DCs were potent and dominant since CD19+ DCs were a minor DC population (Fig. 1B), yet they completely suppressed T cell proliferation in the presence of CD19NEG DCs that stimulated potent T cell responses only when separated from CD19+ DCs (Fig. 1C). These data revealed that a minor DC population expressing CD19 mediated IDO-dependent T cell suppression following CTLA4-Ig treatment in vivo.
We performed multi-color flow cytometric analyses to evaluate the phenotypic characteristics of DC populations expressing the CD19 marker that mediated potent IDO-dependent T cell suppression. Approximately 50% of total CD11c+ splenocytes fell within populations gated using the criteria shown in Fig. 1(A) (CD11cHIGH). Within this gated CD11cHIGH DC population and consistent with data in Fig. 1(B), CD19 staining was heterogeneous, though highest levels of CD19 expression were detected on minor DC populations that also co-expressed B220; these B220+ cells accounted for ∼20% of DCs falling within the CD11cHIGH-gated DC population (Table 1). Much lower levels of CD19 were detected on CD8α+(B220NEG) DC populations and B220NEGCD8αNEG DCs did not express detectable CD19. CD19+ DCs expressed high levels of MHC class I and MHC class II (MHCI, MHCII) and B7 (CD80, CD86) compared with CD19NEG DCs, suggesting that CD19+ DCs were mature DCs. Identical outcomes were obtained when phenotypic analyses were performed on DCs from untreated mice and from mice exposed to CTLA4-Ig prior to flow cytometric analyses. Thus, in vivo CTLA4-Ig treatment had no detectable effect on the phenotypic characteristics of DCs or the relative proportions of DC subsets (data not shown). These data revealed that CD19 marked a minor population of splenic DCs co-expressing B220 and/or CD8α. Together with functional studies (Fig. 1C), these data suggested that DCs competent to express functional IDO activity and acquire potent IDO-dependent T cell regulatory functions following B7 ligation fell within the gated CD11cHIGHCD19+ DC population shown in Fig. 1(B).
Marker . | B220+ CD8+ (∼10%)a . | B220+ CD8NEG (∼12%) . | B220NEG CD8+NEG (∼29%) . | B220NEG CD8NEG (49%) . |
---|---|---|---|---|
CD19 | 938 ± 39b | 1169 ± 71 | 115 ± 12 | 39 ± 4 |
MHCI | 831 ± 32 | 534 ± 38 | 517 ± 28 | 277 ± 10 |
MHCII | 1371 ± 94 | 1401 ± 86 | 771 ± 65 | 844 ± 115 |
CD80 | 558 ± 2 | 348 ± 26 | 424 ± 10 | 229 ± 5 |
CD86 | 809 ± 29 | 441 ± 47 | 617 ± 41 | 233 ± 39 |
Marker . | B220+ CD8+ (∼10%)a . | B220+ CD8NEG (∼12%) . | B220NEG CD8+NEG (∼29%) . | B220NEG CD8NEG (49%) . |
---|---|---|---|---|
CD19 | 938 ± 39b | 1169 ± 71 | 115 ± 12 | 39 ± 4 |
MHCI | 831 ± 32 | 534 ± 38 | 517 ± 28 | 277 ± 10 |
MHCII | 1371 ± 94 | 1401 ± 86 | 771 ± 65 | 844 ± 115 |
CD80 | 558 ± 2 | 348 ± 26 | 424 ± 10 | 229 ± 5 |
CD86 | 809 ± 29 | 441 ± 47 | 617 ± 41 | 233 ± 39 |
Approximate percentage of CD11cHIGH DCs using gates shown in Fig. 1(A).
Mean fluorescence intensity (from more than three separate experiments).
Marker . | B220+ CD8+ (∼10%)a . | B220+ CD8NEG (∼12%) . | B220NEG CD8+NEG (∼29%) . | B220NEG CD8NEG (49%) . |
---|---|---|---|---|
CD19 | 938 ± 39b | 1169 ± 71 | 115 ± 12 | 39 ± 4 |
MHCI | 831 ± 32 | 534 ± 38 | 517 ± 28 | 277 ± 10 |
MHCII | 1371 ± 94 | 1401 ± 86 | 771 ± 65 | 844 ± 115 |
CD80 | 558 ± 2 | 348 ± 26 | 424 ± 10 | 229 ± 5 |
CD86 | 809 ± 29 | 441 ± 47 | 617 ± 41 | 233 ± 39 |
Marker . | B220+ CD8+ (∼10%)a . | B220+ CD8NEG (∼12%) . | B220NEG CD8+NEG (∼29%) . | B220NEG CD8NEG (49%) . |
---|---|---|---|---|
CD19 | 938 ± 39b | 1169 ± 71 | 115 ± 12 | 39 ± 4 |
MHCI | 831 ± 32 | 534 ± 38 | 517 ± 28 | 277 ± 10 |
MHCII | 1371 ± 94 | 1401 ± 86 | 771 ± 65 | 844 ± 115 |
CD80 | 558 ± 2 | 348 ± 26 | 424 ± 10 | 229 ± 5 |
CD86 | 809 ± 29 | 441 ± 47 | 617 ± 41 | 233 ± 39 |
Approximate percentage of CD11cHIGH DCs using gates shown in Fig. 1(A).
Mean fluorescence intensity (from more than three separate experiments).
B7 ligation selectively activates STAT1 in IDO-competent DCs
STAT1 induces transcriptional activity via interactions with specific DNA motifs, known as IFNγ-activating sequences and IFN-stimulated response elements, located in certain gene promoters, including the IDO gene promoter (19). STAT1 activation was an essential upstream event required for IDO expression in splenic DCs from B6 mice cultured with CTLA4-Ig (6). To investigate if STAT1 activation was involved in the selective response by DCs to B7 ligation, we assessed the activation status and intracellular location of STAT1 in DCs following CTLA4-Ig-mediated B7 ligation in vitro. After 5 h, 30–50% of MACS-enriched CD11c+ DCs from F1[CBA × B6] mice (data not shown) and mice genetically deficient in IFNγR expression (IFNγRα-KO mice, Fig. 2A–C) contained activated P-STAT1 protein in their cell nuclei. Nuclear P-STAT1 was not detected in DCs cultured in the absence of CTLA4-Ig, although DCs contained the inactive (un-phosphorylated) form of STAT1 in their cytoplasm, as expected (data not shown). Consistent with data from cytospin analyses, P-STAT1 protein was detected by western blot analyses of cell lysates from DCs treated with non-mutant CTLA4-Ig, but P-STAT1 was not detected in lysates from untreated DCs (data not shown). Similarly, DCs cultured with a closely related CTLA4-Ig isotype with a mutated IgG2a (Fc) domain (mutant CTLA4-Ig), which failed to induce IDO expression or IDO-dependent T cell suppression in F1[CBA × B6] mice (8), also contained no detectable P-STAT1 (data not shown). These outcomes suggested that B7 ligation induced rapid but selective activation of STAT1 in some splenic DCs.
To phenotype DCs that responded to B7 ligation by activating STAT1, we assessed the activation status and intracellular location of STAT1 in sorted splenic DC populations cultured with CTLA4-Ig in vitro (Fig. 2D–I). After culture with CTLA4-Ig for 5 h, >80% of MACS-enriched CD19+ DCs stained with anti-P-STAT1 antibody (Fig. 2G), and staining co-localized to cell nuclei. In contrast, <30% of CD19NEG DCs stained for P-STAT1, and staining was perinuclear in the majority of stained cells, rather than intra-nuclear (Fig. 2D). Analyses of Mo-Flo-sorted B220+/B220NEG and CD8α+/CD8αNEG DC populations treated with CTLA4-Ig revealed that ∼70% of B220+ (Fig. 2H) and ∼80% of CD8α+ (Fig. 2I) DCs contained activated nuclear P-STAT1 protein, while few B220NEG (Fig. 2E) and CD8αNEG (Fig. 2F) DCs (<10% in each sorted population) contained intra-nuclear P-STAT1 protein. These outcomes suggested that STAT1 activation was a selective early response to B7 ligation in some, but not all DCs, and that responsive DCs co-expressed CD19, B220 and/or CD8α. These phenotypic characteristics matched those of DCs that expressed IDO 18–24 h after CTLA4-Ig treatment, and acquired potent T cell regulatory functions as a consequence (7, 8).
To evaluate if selective DC responses to B7 ligation in vitro also occurred in vivo, we assessed the activation status of STAT1 in spleen tissues from mice exposed to CTLA4-Ig (Fig. 2K–M). Immunohistochemical analysis of spleen tissues from mice sacrificed 5 h after injection of non-mutant CTLA4-Ig (Fig. 2K and L) revealed discrete clusters of cells containing activated intra-nuclear P-STAT1 protein, which were located almost exclusively in splenic red pulp areas. P-STAT1-specific staining was not detected in tissues from mice exposed to the mutant isotype of CTLA4-Ig that failed to induce IDO (Fig. 2M). These data revealed that B7 ligation in vivo induced rapid and highly selective activation of STAT1 in minor populations of splenocytes.
Type I, but not type II, IFN signaling is essential for STAT1 activation following B7 ligation
To evaluate if IFN signaling induced STAT1 activation following B7 ligation, we isolated splenic DCs from mice genetically defective in the expression of type I (IFNαβR-KO) or type II (IFNγRα-KO) IFNRs (Fig. 3A–C). Following B7 ligation, intra-nuclear P-STAT1 was detected in 30–50% of DCs from 129/SvJ wild-type mice (Fig. 3A) and from IFNγRα-KO mice with 129/SvJ backgrounds (Fig. 3B). In contrast, anti-P-STAT1 antibody did not stain DCs from IFNαβR-KO mice exposed to CTLA4-Ig (Fig. 3C). These data suggested that signaling through IFN type I receptors was essential for STAT1 activation following B7 ligation, while signaling through type II IFN receptors was not essential for this response.
To test if IFN type I signaling was essential for IDO up-regulation following B7 ligation, we injected CTLA4-Ig into BALB/c mice with defective expression of IFNαβRs, harvested spleens 24 h later and stained tissue sections with anti-IDO antibody (Fig. 3G and H). Consistent with previous studies using F1[CBA × B6] and 129/SvJ mice (7, 8), IDO+ cells were dispersed in splenic red pulp areas of control BALB/c mice (Fig. 3G). However, no IDO+ cells were detected in spleen of IFNαβR-KO mice (Fig. 3H). Since STAT1 activation (Fig. 2A–C) and IDO expression (8) following B7 ligation occurred normally in mice deficient for IFNγR expression, these data suggested that IFNα signaling, but not IFNγ signaling, was essential for early activation of STAT1 and subsequent IDO up-regulation in minor populations of splenocytes. In addition, the pattern of IDO expression induced in spleen 24 h after in vivo CTLA4-Ig treatment (Fig. 3G) was reminiscent of the pattern of STAT1 activation observed at earlier times (Fig. 2), suggesting that selective activation of STAT1 preceded IDO expression in the same minor population of cells located in splenic red pulp.
Though cells that responded to B7 ligation were present in genetically manipulated mice lacking IFNγRs, it is possible that these cells might not develop in mice lacking IFNαβRs. To address this alternative explanation for failure to activate STAT1 and induce IDO expression in IFNαβR-KO mice, we employed a complementary approach to test the hypothesis that IFN type I, but not type II, signaling was essential for STAT1 activation following B7 ligation. DCs from F1[CBA × B6] mice were cultured with CTLA4-Ig alone (Fig. 3D) or in the presence of CTLA4-Ig and mAbs that neutralized IFNγ (Fig. 3E) and IFNα (Fig. 3F), and STAT1 activation was assessed as before. While anti-IFNγ mAb had no significant effect on the proportion of DCs containing intra-nuclear P-STAT1 (30–50% of DCs in Fig. 3D and E), anti-IFNα mAb completely blocked STAT1 activation in a dose-dependent manner, and no P-STAT1+ DCs (among ∼5000 cells inspected) were detected when >50 μg ml−1 anti-IFNα mAb was present (Fig. 3F, data not shown). These data support the hypothesis that IFNα is an essential intermediary signaling ligand that activates STAT1-mediated IDO up-regulation in DCs following B7 ligation. These findings also suggested that splenocytes were induced to express IFNα following B7 ligation in vitro. Since we used MACS-enriched CD11c+ DCs, these data suggested that DCs might be the source of IFNα, though MACS enrichment did not completely remove other (CD11cNEG) splenocytes, which might be a source of IFNα.
B7 ligation induces CD19+ DCs to express and secrete IFNα
To identify cells that produced IFNα following B7 ligation, we measured IFNα gene and protein expression by splenocytes. First, we assessed IFNα gene transcription by RT–PCR analysis and IFNα secretion by ELISA following CTLA4-Ig treatment in vitro (Fig. 4). Transcripts of the IFNα1–9 genes were detected in RNA samples prepared from CTLA4-Ig-treated and influenza virus-infected DCs (Fig. 4A). IFNα1–9 transcripts were not detected in RNA samples prepared from untreated DCs. Consistent with this, DCs secreted IFNα into culture media following treatment with non-mutant CTLA4-Ig, while DCs treated with the mutant CTLA4-Ig isotype, which did not induce IDO in DCs (8), did not secrete IFNα (Fig. 4B).
Based on the finding that B7 ligation induced DCs to express IFNα, we hypothesized that IFNα production was restricted to CD19+ DCs following B7 ligation, like STAT1 activation and functional IDO up-regulation (Figs 1 and 2). To test this hypothesis, we treated purified (Mo-Flo sorted) CD19+ and CD19NEG (CD11cHIGH) DCs with CTLA4-Ig for 5 h and stained them with anti-IFNα mAb (Fig. 4C, upper panels). Almost all CD19+ DCs contained cytoplasmic IFNα, while very few cells expressing IFNα were detected in CD19NEG DCs. To test if IDO expression was also restricted to CD19+ DCs following B7 ligation, we cultured Mo-Flo-sorted DCs for 24 h in the presence of CTLA4-Ig and stained cells for IDO expression. As expected from functional data reported in Fig. 1(C), IDO+ cells were detected exclusively in the CD19+ DC population (Fig. 4C, center panels). IDO was also selectively induced in the same CD19+ DC population when recombinant IFNα was added to culture medium (Fig. 4C, lower panels). These data showed that B7 ligation and IFNα induction were preferentially confined to DC populations expressing CD19. IDO staining after 24 h was more heterogeneous than IFNα staining after 5 h in CD19+ DCs, suggesting that the ability to express high levels of IDO may not be uniformly distributed within the CD19+ DC population. However, these data show that IFNα secretion induced following B7 ligation and IDO expression induced by STAT1-dependent IFNα signaling were both confined to the CD19+ DC population.
IFNα, but not IFNγ, promotes IDO-dependent T cell suppression
Based on data showing that IFNα signaling, but not IFNγ signaling, was essential for STAT1 activation and IDO up-regulation in splenic DCs (Figs 3 and 4), we hypothesized that IFNα produced by CD19+ DCs following B7 ligation signaled CD19+ DCs to acquire potent IDO-dependent T cell regulatory functions. To test this hypothesis, we asked if recombinant IFNα could substitute for in vivo CTLA4-Ig treatment as a stimulus to induce IDO-dependent T cell suppression. Using the experimental system described in Fig. 1, we performed MLRs using splenocytes from untreated F1[CBA × B6] mice as stimulators and assessed the effect of adding recombinant IFNα (Fig. 5A) and IFNγ (Fig. 5B) on their ability to stimulate BM3 T cell proliferation. T cell proliferation was reduced significantly in cultures containing ≥150 U ml−1 IFNα. This anti-proliferative effect of IFNα was due to induction of IDO and not an intrinsic anti-proliferative effect of IFNα because T cell proliferative responses recovered to control levels in the presence of the IDO inhibitor (1mT). Also consistent with the hypothesis that IFNα signaled IDO induction, addition of IFNα to MLRs containing splenocytes from IDO-KO mice had no effect on their ability to promote T cell proliferation.
In contrast to outcomes obtained with IFNα, addition of recombinant IFNγ did not suppress T cell proliferation (Fig. 5B), and addition of IDO inhibitor did not enhance T cell responses when recombinant IFNγ or no exogenous IFN was added to MLRs (data not shown). These data revealed that IFNα induced IDO-dependent T cell suppression, while IFNγ had no effect in this system.
To examine if IFNα acted to induce IDO-dependent T cell suppression selectively in the CD19+ DC population, we repeated the previous experiment using purified CD19+ and CD19NEG DC populations sorted by flow cytometry (Fig. 5C and D, respectively). When CD19+ DCs were used as APCs, addition of recombinant IFNα to MLRs induced potent IDO-dependent T cell suppression, which was reversed in the presence of 1mT or excess tryptophan in MLRs. In contrast, addition of recombinant IFNγ had no significant effect on T cell proliferation (Fig. 5C). Moreover, neither IFNα nor IFNγ addition had any effect on the robust T cell stimulatory activity of CD19NEG DCs (Fig. 5D). These outcomes confirmed that IFNα was the relevant upstream signaling ligand that induced functional IDO expression in CD19+ DC populations since IFNα substituted for B7 ligation in promoting IDO-dependent T cell regulatory functions of CD19+ DCs.
Discussion
In the current study, we identified a small population of splenic CD19+ DCs as the principal cell type that mediated IDO-dependent T cell suppression following CTLA4-Ig treatment in vivo. CD19+ DCs selectively responded to B7 ligation by secreting IFNα and activating STAT1. Since IFNα, but not IFNγ, could substitute for B7 ligation to promote IDO-dependent T cell suppression, these data support the hypothesis that splenic CD19+ DCs are unique in their ability to produce IFNα and respond to IFNα-mediated signaling by acquiring potent T cell regulatory functions via STAT1 activation and IDO up-regulation in response to B7 ligation.
The rationale for studying CD19+ DCs in spleen was based on our previous discovery that CD19+ DCs constituted the principal cell population that mediated IDO-dependent T cell suppression in TDLNs (11). CD19 is a component of signaling complexes expressed by B cells, and has been widely used to separate B cells from DCs; partly for this reason, splenic CD19+ DCs may not have been recognized previously. CD19+ DCs from TDLNs shared certain characteristics with the B cell lineage, including D–J region Ig gene rearrangements and expression of B220 and Pax5 (11). Similar links to the B cell lineage have been reported in plasmacytoid DC subsets from other studies (20). In spleens of F1[CBA × B6] mice, we found that CD19+ DCs constituted ∼20% of sorted splenic DCs expressing relatively high levels of CD11c and, like CD19+ DCs from TDLNs of B6 mice, these cells co-expressed B220 and many also expressed CD8α. Murine plasmacytoid DCs have been reported to express B220 and 120G8 and low/intermediate levels of CD11c, display immature phenotypes with respect to MHC and B7 expression levels and have relatively weak T cell stimulatory functions associated with T cell suppressive and tolerogenic outcomes (17, 18, 21–23). In the current study, sorted CD19+ DCs that mediated IDO-dependent T cell suppression expressed relatively high levels of CD11c, had mature phenotypes and were potent T cell stimulators. However, when IDO activity was induced following in vivo CTLA4-Ig, or in vitro IFNα, treatment these DCs became strongly suppressive. Hence, CD19 expression appears to identify the population of DCs that can be induced to acquire potent T cell regulatory functions via IDO. These DCs appear distinct from typical plasmacytoid DCs defined previously by others, although they share certain features, such as B220 expression and the ability to produce IFNα (8, 11, 18). Hence, CD19 may not be a distinct DC lineage marker but rather identifies these DC populations with particular functional characteristics, including the ability to respond to B7 ligation by up-regulating IDO.
The responsiveness of DCs to CTLA4-Ig isotypes may differ between mouse strains. Grohmann and colleagues reported that a different CTLA4-Ig isotype (CTLA4-IgG3) partially blocked T cell-mediated rejection of pancreatic islet allografts transplanted into B6 mice, and showed that this reagent induced functional IDO expression in isolated B6 DCs (6). With our CTLA4-Ig preparation (CTLA4-IgG2a) we found that DCs from B6 mice were unresponsive, while DCs from CBA, BALB/c and 129/SvJ mice responded by up-regulating IDO (our unpublished data). Thus, the CTLA4-Ig reagent we used may have failed to induce IDO in B6 mice for technical reasons, perhaps related to the Fc domain structure. In this regard, it may be important that a mutant isotype of CTLA4-IgG2a, engineered to reduce complement factor C1q and FcR binding, also failed to induce IDO. Since CD19+ DCs constituted the principal cell subset mediating IDO-mediated suppression in tumor-bearing B6 mice (11), and CD19+ DCs were detected in comparable proportions in CBA, 129/SvJ and F1[CBA × B6] mice (our unpublished data), it is likely that the presence of CD19+ DCs is not strain dependent.
Though we identified CD19+ DCs as the principal DC population that mediated functional T cell suppression in the present study, other DCs might also express non-functional immunoreactive IDO protein following B7 ligation, or other treatments. Consistent with this, the proportion of MACS-enriched (CD11c+) DCs containing activated intra-nuclear P-STAT1 (∼30–50%) following B7 ligation was higher than the proportion of CD19+ DCs (∼10%). Previously, we detected IDO expression in several different DC subsets, including DCs co-expressing CD8α, B220 and the NK-DC marker DX5, all of which expressed immunoreactive IDO protein by immunohistochemistry after B7 ligation in vivo (7). However, these earlier studies did not include assays to measure T cell stimulatory functions of sorted DC subsets. It is known that IDO can be expressed in non-functional form in both murine and human DC subsets (3, 24). Thus, the functional analyses of T cell stimulatory functions performed in the current study were critically important in identifying biologically relevant populations of IDO-expressing DCs.
Several recent reports revealed that IDO enzyme activity in DCs has potent inhibitory effects on T cell responses in vitro and in vivo (5, 7, 8, 11, 25). In mice, IDO expression was first associated with CD8α+ DCs in response to IFNγ treatment (5, 24). More recently, we identified B220+ DCs in spleen and TDLNs as potent mediators of IDO-dependent T cell suppression (8, 11). B220+ and CD8α+ DC subsets may overlap to some extent as CD8α is expressed by some plasmacytoid DCs (20–22), as discussed above. However, in our system, the CD19 marker gave the best segregation of IDO-dependent T cell suppressor functions, STAT1 activation and IFNα production in distinct populations of splenic DCs.
The role of IFNα in our system was unexpected. Plasmacytoid DCs are known to produce IFNα in response to microbial infections (17), most likely via signals transmitted through Toll-like receptors, but IFNγ is known to be a more potent IDO inducer than IFNα (19). However, we found that IFNα signaling was required to induce IDO expression in CD19+ DCs and that recombinant IFNα could re-capitulate the response to B7 ligation, leading to IDO-dependent T cell suppression.
The unique signaling processes that confer the highly selective link between B7 ligation and IDO induction in distinct DC populations are not fully defined. Grohmann and colleagues showed that IFNγ was an essential upstream ligand required for IDO induction in unfractionated splenic CD11c+ DCs following CTLA4-Ig treatment in vitro (6). However, we developed different experimental approaches to address the specific question of which DC populations were principally responsible for IDO-mediated suppression when DCs were exposed to CTLA4-Ig in vivo. We found that IFNγ signaling was not essential for this process, while IFNα signaling was essential. STAT1 activation appears to be an obligate event preceding IDO expression since IDO was not induced in STAT1-deficient mice (6, 26). Previous reports have also shown that IFNα can induce IDO expression via STAT1-dependent signaling, though IFNα is considerably less potent as an IDO inducer than IFNγ in most cell types studied (9, 19). Our findings are consistent with the hypothesis that STAT1 activation is a selective response to IFNα by specific DCs, including minor DC populations expressing CD19.
Mechanisms that confer selective IDO expression exclusively in CD19+ DCs have not been defined. Presumably, selective induction of IFNα expression following B7 ligation is controlled by factors in DCs that modulate downstream signals generated following B7 ligation, though the nature of these mechanisms is not known. Similarly, selective IFNα-mediated STAT1 activation in CD19+ DCs is probably controlled by factors downstream of IFNαβRs, since many cell types express these receptors. One speculative possibility is that IFN regulatory factors (IRFs), such as IRF-2 and IRF-7, which are differentially expressed in distinct DC populations, might regulate responses to IFNs differently in distinct DC subsets (10, 27, 28). Elucidating these signaling mechanisms will be critical for understanding why CD19+ DCs selectively produce IFNα in response to B7 ligation and express IDO in response to IFNα, while most DCs do not respond in this way, even though they express B7 molecules. The key point to emerge from the current study, however, is that certain minor populations of splenic DCs, best identified by the expression of CD19 in our system, are selectively programmed to respond to B7 ligation by inducing IDO, and acquiring potent T cell regulatory functions as a consequence.
These authors contributed equally to this study.
Transmitting editor: E. Simpson
We thank the manager of the MCG Flow Cytometry Core facility, Jeanine Pihkala, for expert assistance with flow cytometry and Anita Wylds, Doris McCool and Erika Thompson for technical assistance with multiple aspects of studies reported here. This work was supported by NIH grants to A.L.M. (HD41187, AI063402) and D.H.M. (CA103320, CA096651).
References
Mellor, A. L. and Munn, D. H.
Seo, S. K., Choi, J. H., Kim, Y. H. et al.
Munn, D. H., Sharma, M. D., Lee, J. R. et al.
Grohmann, U., Fallarino, F. and Puccetti, P.
Grohmann, U., Fallarino, F., Silla, S. et al.
Grohmann, U., Orabona, C., Fallarino, F. et al.
Mellor, A. L., Baban, B., Chandler, P. et al.
Mellor, A. L., Chandler, P. R., Baban, B. et al.
Taylor, M. W. and Feng, G.
Levy, D. E. and Darnell, J. E., Jr.
Munn, D. H., Sharma, M. D., Hou, D. et al.
Tarazona, R., Sponaas, A.-M., Mavria, G. et al.
Baban, B., Chandler, P., McCool, D., Marshall, B., Munn, D. H. and Mellor, A. L.
Simpson, E. and Chandler, P.
Julius, M. H., Simpson, E. and Herzenberg, L. A.
Fung, M. C., Sia, S. F., Leung, K. N. and Mak, N. K.
Asselin-Paturel, C., Boonstra, A., Dalod, M. et al.
Asselin-Paturel, C., Brizard, G., Pin, J. J., Briere, F. and Trinchieri, G.
Hassanain, H. H., Chon, S. Y. and Gupta, S. L.
Corcoran, L., Ferrero, I., Vremec, D. et al.
Martin, P., Del Hoyo, G. M., Anjuere, F. et al.
O'Keeffe, M., Hochrein, H., Vremec, D. et al.
Bilsborough, J., George, T. C., Norment, A. and Viney, J. L.
Fallarino, F., Vacca, C., Orabona, C. et al.
Fallarino, F., Asselin-Paturel, C., Vacca, C. et al.
Gurtner, G. J., Newberry, R. D., Schloemann, S. R., McDonald, K. G. and Stenson, W. F.
Honda, K., Yanai, H., Mizutani, T. et al.
Author notes
1Immunotherapy Center and 2Department of Medicine, 3Department of Surgery and 4Department of Pediatrics, Medical College of Georgia, 1120, 15th Street, Augusta, GA 30912, USA