Inducible MHC class II expression by mast cells supports effector and regulatory T cell activation

Abstract

In addition to their well-established role as regulators of allergic response, recent evidence supports a role for mast cells in influencing the outcome of physiologic and pathologic T cell responses. One mechanism by which mast cells influence T cell function is indirectly through secretion of various cytokines. It remains unclear, however, whether mast cells can directly activate T cells through antigen presentation, as the expression of MHC class II by mast cells has been controversial. In this report, we demonstrate that in vitro stimulation of mouse mast cells with LPS and IFNγ induces the expression of MHC class II and co-stimulatory molecules. Although freshly isolated peritoneal mast cells do not express MHC class II, an in vivo inflammatory stimulus increases numbers of MHC class II-positive mast cells in situ. Expression of MHC class II granted mast cells the ability to process and present antigens directly to T cells with preferential expansion of antigen-specific regulatory T cells over naive T cells. These data support the notion that, in the appropriate setting, mast cells may regulate T cell responses through the direct presentation of antigen.

Introduction

Mast cells (MCs)³ are tissue-resident cells of the immune system that are primarily located at the host-environment interface, making them one of the first cell types to encounter environmental threats. MCs were once believed to participate solely in allergy, owing to their abundant intracellular granules that are rapidly released upon crosslinking of their high affinity IgE receptor. However, the importance of MCs extends far beyond allergic disease, a notion that was initiated by the discovery that MCs are critical effectors in host defense against parasitic infections. Although the mechanisms are not fully understood, MCs contribute to protection against pathogens such as Leishmania majo r(1), Giardia lambia (2), and intestinal helminthes (3, 4).

MCs also play a pathologic role in the development of T cell-mediated hypersensitivity disorders such as delayed type contact hypersensitivity (5, 6) and asthma (7, 8), and in the induction of autoimmune mouse models of inflammatory bowel disease (9) and multiple sclerosis (10, 11). T cells play a vital role in these mouse models, suggesting that MCs may influence T cell activation. In at least some of these models, a direct correlation between the activation of T cells and the presence of MCs has been established (1, 12), as attenuated activation of T cells was observed in MC-deficient mice. The effect of MCs on T cell responses may also be inhibitory under certain circumstances, as MCs were recently shown to be vital for T cell-mediated skin allograft tolerance (13).

It has been previously suggested that MCs act as APCs and directly stimulate T cells. Both rodent (14, 15) and human (16, 17) MCs have been reported to constitutively express MHC class II (MHCII), present antigen to T cell hybridomas, and initiate antigen-specific responses in vivo. Moreover, induction of co-stimulatory molecules CD80 and CD86 by treatment of MCs with GM-CSF has been observed (18). However, follow-up studies demonstrated that MHCII is found only to a limited extent at the cell surface and resides mainly in intracellular lysosomal compartments (19). The initially described activation of antigen-specific T cell responses by MCs was later attributed to the release of immunologically active MC-derived exosomes, since the activation of T cells still occurred by MHC haplotype-mismatched MCs (20). This understanding has led to re-examination of whether MHCII are expressed at all in MCs, as we and others have demonstrated that resting or FcεRI-activated MCs do not express MHCII on the cell surface or intracellularly (21, 22). We did find, however, that MCs can indirectly promote T cell activation by internalizing antigens through FcεRI, undergoing apoptosis, and subsequently providing antigens to other professional APCs (22).

In this report, we extend our analysis of how MCs may regulate T cell responses under specific conditions. We demonstrate that although MHCII is not detected on resting MCs, stimulation of MCs with LPS and IFNγ induces robust expression of MHCII. The expression of MHCII conferred MCs the ability to process and present antigens directly to previously activated CD4⁺ T cells and to a limited extent to naive CD4⁺ T cells. Furthermore, we show that MCs preferentially expand antigen-specific regulatory T cells (Treg) over naive T cells, possibly shedding light on one of the mechanisms that governs allograft tolerance induction by MCs. These data suggest that one of the mechanisms by which MCs regulate T cell responses could be through the direct presentation of antigen.

Materials and methods

Mice

C57BL/6 (B6), BALB/c, MHCII^−/−, RAG^−/−, and H-2DM^−/− mice were obtained from Jackson Laboratories (Bar Harbor, ME) and bred in the animal care facility at the University of Pennsylvania. OT-II.2a/Rag1 mice (Mouse Line #4234, Taconic Emerging Model) were obtained through the NIAID Exchange Program (23, 24). Myeloid differentiation factor 88 (MyD88)^−/− and B6-Eα (25) mice were generous gifts from Drs. S. Akira and R.A. Flavell, respectively. TS1, TS1×HA28, and HACII mice were bred and maintained as previously described (26). All animal care and work was in accordance with national and institutional guidelines and the Institutional Animal Care and Use Committee at the University of Pennsylvania.

Chemicals and tissue culture reagents

All chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise specified. All cytokines and cell culture reagents were purchased from Peprotech (Rocky Hill, NJ) and Invitrogen (Carlsbad, CA), respectively. The α chain of I-E^d (amino acids 46–74) inserted into pTrcHis2-TOPO vector in frame with RFP (red fluorescence protein) was generously provided by Dr. M.K. Jenkins (27). Eα-RFP fusion protein was purified from bacterial lysate and TNP-conjugated with picrylsulfonic acid (TNBS) (pH 8.5) overnight at 4°C and removing excess TNBS by size exclusion columns (GE Healthcare, Upsalla, Sweden).

Flow cytometry

All antibodies used for flow cytometry were purchased from BD Biosciences (Franklin Lakes, NJ) except for YAe and anti-FcεRI (eBioscience, San Diego, CA). Anti-MHCII antibody (clone Y3P) was purified from HB183 hybridoma supernatant and FITC-conjugated. Biotinylated 6.5 TCR clonotype-specific antibody against TS1 T cells was previously described (26). Cells were blocked with anti-CD16/32 antibody, stained with specified antibodies (anti-CD117-allophycocyanin, anti-FcεRI-PE, anti-IA^b-biotin (clone KH74), anti-CD4-allophycocyanin, anti-CD69-PE, anti-CD80-biotin, anti-CD86-biotin, anti-programmed death ligand 1(PD-L1)-PE, anti-PD-L2-PE, YAE-biotin) followed by streptavidin-PE or -allophycocyanin when using biotinylated antibodies. The fluorescence intensity was measured on a FACSCalibur™ flow cytometer (BD Biosciences) and analyzed using Cell Quest (BD Biosciences) or FlowJo (FlowJo, Ashland, OR) software.

Generation of bone marrow-derived MCs (BMMCs), spleen-derived MCs, peritoneal MCs, and bone marrow-derived dendritic cells (BMDCs)

To generate MCs (28), spleen or bone marrow (BM) cells of mice were cultured in MC medium (MCM; RPMI1640, 15% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, 2.9 mg/ml glutamine, 50 mM 2-mercaptoethanol, 1 mM sodium pyruvate, 1× non-essential amino acids, 10 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid; HEPES) containing IL-3 (10 ng/ml) and stem cell factor (SCF; 12.5 ng/ml) for 6–8 weeks, replenishing with fresh media twice weekly, and used when >95% of cells expressed high homogeneous levels of FcεRI and CD117. BMDCs (29) were generated by culturing BM cells in dendritic cell (DC) medium (DMEM, 15% FBS, penicillin, streptomycin, glutamine) containing IL-4 (10 ng/ml) and GM-CSF (10 ng/ml) for 7 days, and purified by magnetic cell sorting (MACS) using CD11c beads (Miltenyi Biotec, Auburn, CA). Peritoneal MCs were obtained by peritoneal lavage of mice using 10 ml of PBS containing 2mM EDTA.

MC stimulation and RT-PCR analysis

MCs were stimulated with LPS (E. coli O127:B8) and/or IFNγ in 96-well U-bottom plates in MCM containing IL-3 (10 ng/ml), and the expression of surface molecules was measured on CD117⁺FcεRI⁺ MCs by flow cytometry. For antigen processing experiments, MCs were stimulated with LPS/IFNγ for 72 h in the presence or absence of TNP-Eα protein (50 μg/ml), Eα_52–66 peptide (ASFEAQGALANIAVDKA), or anti-TNP IgG₁ (10 μg/ml). In some experiments the MCs were pretreated with anti-TNP IgE (1 μg/ml) for 24 h before adding the TNP-Eα protein.

For RT-PCR analysis, MCs were FACS-sorted by Moflo cell sorter (Dako, Carpenteria, CA) using CD117 and FcεRI antibodies, stimulated with or without LPS (10 μg/ml) and IFNγ (10 ng/ml) for 24 h in MCM containing IL-3, washed, and RNA was extracted using RNEasy kit (QIAGEN, Valencia, CA). RT-PCR was performed using OneStep RT-PCR kit (QIAGEN). The primer sets used were: IA^b-α sense: 5′-GAAGACGACATTGAGGCCGACCACG-3′, anti-sense: 5′-TAAAGGCCCTGGGTGTCTGGAGGTG-3′ (product size: 748bp) (30); IA^b-β sense: 5′-GCGACGTGGGCGAGTACC-3′, anti-sense: 5′-CATTCCGGAACCAGCGCA-3′ (product size: 220bp) (31); H-2DMα sense: 5′-AAGGTATGGAGCATGAGCAGAAGT-3′, anti-sense: 5′-GATCAGTCACCTGAGCACGGT-3′ (product size: 768bp) (32); H-2DMβ sense: 5′-TGAATTTGGGGTGCTGTATCC-3′, anti-sense: 5′-TGCTGAACCACGCAGGTGTAG-3′ (product size: 395bp ) (30); CIITA sense: 5′-TGCAGGCGACCAGGAGAGACA-3′; anti-sense: 5′-GAAGCTGGGCACCTCAAAGAT-3′ (product size: 488bp); IL-3 sense: 5′-ATAGGGAAGCTCCCAGAACCTGAACTC-3′; anti-sense: 5′-AGACCCCTGGCAGCGCAGAGTCATTC-3′ (product size: 206bp) (33); β-actin sense: 5′-TTCTTTGCAGCTCCTTCGTTGCCG-3′, anti-sense: 5′-TGGATGGCTACGTACATGGCTGGG-3′ (product size: 450bp).

MC, DC, T cell co-cultures

1 × 10⁵ spleen-derived MCs were stimulated with LPS (10 μg/ml) and IFNγ (10ng/ml) in MCM containing IL-3 (10 ng/ml) in the presence or absence of OVA protein (Grade V), live or heat-inactivated influenza PR8 or J1 virus (1:1000 titer), OVA peptide (ISQAVHAAHAEINEAGR), or influenza S1 peptide (SFERFEIFPK) for 72 h in 96-well U-bottom plates. Heat-inactivation of influenza virus was performed by incubating the virus at 56oC for 30 min. The peptides were added to the MCs 48 h after LPS stimulation. After extensive washing, the MCs were co-cultured in 96-well U-bottom plates with T cells in MCM containing IL-3. 1 × 10⁵ FACS-sorted Thy1.2⁺ T cells from OT-II, TS1, or TS1/HA28 mice were used as a source of T cells. In some experiments, the CD4⁺CD25⁺ and CD4⁺CD25⁻ fraction of TS1/HA28 T cells were FACS-sorted prior to co-culture. For proliferation assays, the TS1 and TS1/HA28 T cells were pre-labeled with carboxyfluorescein succimidyl ester (CFSE). To obtain antigen-experienced cells, OT-2 T cells were expanded for 6 days by culturing OT-2 spleen cells with OVA peptide (1 μM). In experiments involving DCs, 5 × 10⁴ BMDCs pulsed with OVA peptide were added to the T cells. For CD69 expression, the co-cultures were incubated for 48 h. For detection of IFNγ, the co-cultures were incubated for 6 h in the presence of brefeldin A (10μM) and intracellularly stained with anti-IFNγ-PE antibody (BD Pharmingen). Proliferation of T cells was analyzed by CFSE dilution 4 days after co-culture. When measuring the proliferation of TS1/HA28 Treg, cells were stained with anti-CD4-PerCPCy5.5 and 6.5-biotin antibody, followed by streptavidin-PE, and intracellularly stained for Foxp3 using the mouse regulatory T cell staining kit (eBioscience), and visualized by flow cytometry.

LPS and Leishmania major inoculation of mice

C57BL/6 and RAG^−/− mice were injected subcutaneously in both flanks with 25 μg/flank of LPS (S. minnesota R595 Re, AXXORA platform, San Diego, CA) in 100 μl of PBS. Bilateral inguinal lymph nodes (LN)s were harvested at various time points post LPS injection, and FcεRI⁺CD117⁺ MCs were enumerated and analyzed for MHCII, CD80, CD86, PD-L1, and PD-L2 expression by flow cytometry. Leishmania infection was performed by inoculation C57/BL6 mice subcutaneously in the right hind footpad with 2 × 10⁶ late stationary phase Leishmania major promastigotes. At 7 days post-infection, the popliteal LNs ipsilateral and contralateral to the site of infection were harvested. Sections were prepared from LN fixed in 10% formalin, mounted on glass slides, and stained with toluidine blue to visualize MCs. Representative microscope images were obtained using a Leica DMLB microscope (Leica Microsystems, Bannockburn, IL) equipped with a SPOT Insight color camera (Diagnostic Instruments, inc., Sterling Heights, MI) and incorporated using Photoshop computer software (Adobe Systems, San Jose, CA). Statistical analysis was performed by ANOVA using Microsoft Excel computer software (Microsoft, Redmond, WA).

Results

Induction of MHCII on MCs by stimulation with LPS/IFNγ

BMMCs do not express MHCII constitutively or after stimulation through FcεRI (21, 22). Despite lack of MHCII expression, we previously demonstrated that BMMCs stimulate antigen-specific CD4⁺ T cell responses in an MHCII-independent manner by incorporating antigens through FcεRI and transferring them to DCs (22). In follow-up studies, we observed experiment-to-experiment variation within our T cell/MC co-culture assays. We mapped this variability to commercial sources of antigen and found that endotoxin-contaminated antigen increased the activation of T cells in an FcεRI-independent manner, suggesting that endotoxin might influence the way MCs present antigen. To test this notion, we stimulated BMMCs with LPS and found that a fraction of BMMCs expressed MHCII, a response that was potentiated dramatically by the addition of IFNγ (Fig. 1A). Bona fide MHCII expression was confirmed with a second anti-MHCII antibody (clone Y3P; data not shown) and by the failure to observe this effect in BMMCs lacking the MHCII gene (Fig. 1A). The induction of MHCII expression was observed maximally at 72 h post stimulation (Fig. 1B) and occurred at physiological concentrations of both LPS (~10 ng/ml) and IFNγ (~0.1 ng/ml) (Fig. 1C, D). Of note, the addition of polymyxin B to LPS/IFNγ-treated MCs completely abrogated MHCII upregulation, suggesting that the effect was not secondary to potential contaminants in our LPS preparation.

Figure 1. LPS and IFNγ stimulation induces MHCII expression by MCs.

(A) WT B6 BMMCs were left unstimulated (top left plot) or stimulated with LPS (10 μg/ml; top middle plot), LPS + IFNγ (10 ng/ml; top right plot), or IFNγ alone (bottom left plot). MHCII^−/− (bottom middle plot) or MyD88^−/− (bottom right plot) BMMCs were treated with LPS + IFNγ. Seventy-two hours later, MHCII expression was measured by flow cytometry on FcεRI⁺ BMMCs. (B) WT B6 BMMCs were treated with LPS + IFNγ. The fraction of MHCII-expressing BMMCs was assessed and plotted against time or plotted against various concentrations of (C) IFNγ (with fixed LPS at 10 μg/ml) or (D) LPS (with fixed IFNγ at 10 ng/ml) at 72 h post stimulation. (E) WT B6 and MyD88^−/− BMMCs were co-cultured and treated with LPS + IFNγ. The BMMCs were distinguished from each other by labeling either the MyD88^−/− (top plot) or WT (bottom plot) with CFSE (green). MHCII expression was measured 72 h post stimulation on FcεRI⁺ BMMCs. (F) RNA was extracted from FcεRI⁺CD117⁺ FACS-sorted B6 BMMCs stimulated with (S) or without (U) LPS + IFNγ for 24 h. The presence of IA^b-α, IA^b-β, H2-DMα, H2-DMβ, CIITA, and β-actin transcripts were detected by RT-PCR. All results are representative of at least 2 independent experiments.

Signaling through TLR4 by LPS occurs through two distinct pathways that involve either MyD88 or Toll-IL-1 receptor (TIR) domain-containing adapter inducing IFN-β (TRIF) (34). In BMMCs, the induction of MHCII required activation of the MyD88-dependent signaling pathway by LPS, since MHCII expression was not detected on WT MCs stimulated with IFNγ alone or on MyD88-deficient MCs treated with LPS/IFNγ (Fig. 1A). One potential explanation for the LPS effect on MHCII induction is that LPS was indirectly stimulating MHCII expression via the elaboration of cytokines. To test this possibility, MyD88-deficient and WT BMMCs were labeled with dye (CFSE) to distinguish their genotypes, co-cultured, and stimulated with LPS/IFNγ. MHCII expression was observed only on WT MCs (Fig. 1E), suggesting that the induction of MHCII occurred through direct TLR4 stimulation of the MCs. Since IL-1 and IL-18 also signal through MyD88 and LPS-stimulated MCs produce IL-1β (35), it was still possible that these cytokines contributed to MHCII expression by MCs. However, neutralization of IL-1β and IL-18 did not have any effect on LPS/IFNγ-stimulated MHCII expression by MCs, ruling out the involvement of these cytokines (data not shown).

We next tested whether other stimuli could upregulate MHCII on MCs. Since enhancement of MHCII expression by IL-4 and GM-CSF was shown in other cell types (36, 37), we tested the ability of these cytokines to enhance the effects of LPS on BMMCs. Unlike IFNγ, IL-4 was incapable of increasing MHCII expression (Fig. 2A). Moreover, GMCSF had a slight but limited effect on MHC class II expression compared to IFNγ. MCs have been reported to express other TLRs including TLR2, TLR3, TLR5, TLR7, and TLR9 (38). Among the TLR stimuli tested (TLR2, TLR3, TLR9), only TLR1/2 stimulation (peptidoglycan and Pam3Cys) showed increased MHCII expression on MCs, albeit to a lesser extent than observed with TLR4 stimulation (Fig. 2B). These data suggest that maximal induction of MHCII expression by MCs occurs by signaling through TLR4 and IFNγ receptors.

Figure 2. Effects of other cytokines and TLR stimuli on MHCII expression by MCs.

(A) BMMCs were stimulated with LPS (10 μg/ml), LPS + IL-4 (10 ng/ml), LPS + GM-CSF (10 ng/ml), or LPS + IL-4 + GM-CSF (plots from left to right). (B) BMMCs were stimulated with IFNγ (10 ng/ml) and peptidoglycan (10 μg/ml), FSL-1 (TLR2/6 agonist; 100 ng/ml), Pam3Cys (TLR1/2 agonist; 300 ng/ml), Poly I:C (10 μg/ml), and CpG DNA (1 μM). Seventy-two hours later, MHCII expression was measured by flow cytometry on FcεRI⁺ BMMCs. Results are representative of 2 independent experiments.

To extend our studies on the induction of MHCII expression and to test whether other MHCII-associated molecules necessary for antigen presentation were expressed in BMMCs upon LPS/IFNγ stimulation, RT-PCR analysis was performed. LPS/IFNγ stimulation of BMMCs induced mRNA expression of MHCII chains IA^b-α and IA^b-β (Fig. 1F), suggesting that surface expression of MHCII was due to de novo synthesis of MHCII rather than relocalization of internal stores. H2-DM, which is required for efficient peptide exchange on MHCII (39), and CIITA, the master regulator of MHCII and MHCII-associated genes, were also upregulated by LPS/IFNγ stimulation (Fig. 1F), indicating that LPS/IFNγ-stimulated BMMCs possess the necessary molecules to present antigens on MHCII.

MCs are poor stimulators of naive CD4⁺ T cells

To examine whether MHCII-bearing MCs stimulate naive CD4⁺ T cells, OVA peptide-pulsed MHCII-expressing MCs were co-cultured with FACS-sorted naive T cells from OVA peptide-specific TCR transgenic (OT-2) mice. Spleen-derived MCs were used in these experiments, since a larger proportion of spleen-derived MCs (50–60%) express MHCII compared to BMMCs (20–30%) when stimulated with LPS/IFNγ. Upon co-culture with OVA peptide-pulsed DCs, OT-2 T cells were strongly activated as measured by CD69 expression. In contrast, co-culture of OT-2 T cells with peptide-pulsed MCs showed no effect above background (Fig. 3A). To test whether the lack of naive T cell activation by MCs was due to defective peptide binding by MHCII, we used two approaches: first, binding of biotinylated OVA peptide, and second, staining with an MHCII/peptide conformation-specific antibody known as YAe, which specifically recognizes MHCII (I-A^b) bound to a peptide derived from the α chain of the I-E molecule (Eα). With both approaches, peptide binding was detected on LPS/IFNγ-stimulated WT but not MHCII-deficient MCs (Fig. 3B, C), suggesting that MHCII on MCs bind to peptides.

Figure 3. LPS/IFNγ-stimulated MCs cannot support naive T cell proliferation and poorly express co-stimulatory molecules.

(A) B6 MCs were stimulated with LPS/IFNγ for 72 h. For the last 24 h, the MCs were pulsed with (left plot) or without (middle plot) OVA peptide (10μM). FACS-sorted OT-2 T cells were co-cultured with the MCs or with peptide-pulsed BMDCs (right plot) for 48 h. CD69 expression on CD4⁺ T cells was analyzed by flow cytometry. (B) MHCII^−/− (left plot) or WT B6 (right plot) MCs were stimulated with LPS/IFNγ for 72 h and pulsed with biotin-OVA peptide (10 μM) or (C) Eα peptide (10 μM) for the last 24 h. Biotin peptide and Eα peptide binding were detected by streptavidin-PE and YAe antibody staining, respectively, by flow cytometry. (D) WT B6 MCs were stimulated with (dotted line) or without (solid line) LPS/IFNγ for 72 h and CD80, CD86, PD-L1, and PD-L2 (plots from left to right) were analyzed by flow cytometry. The numbers represent the percent co-stimulatory molecule positive MCs or mean fluorescence intensity (for PD-L1 only) treated with (right number) or without (left number) LPS/IFNγ. (E) LPS/IFNγ-treated B6 MCs (solid line) and freshly isolated splenic B6 B cells (dotted line) were compared for expression of MHCII, CD80, CD86, and PD-L1 (plots from left to right) by flow cytometry. All results are representative of at least 2 independent experiments.

Naive CD4⁺ T cell activation not only requires TCR activation by cognate MHCII/peptide complexes but also is dependent on co-stimulatory signals provided by the APC. Therefore, we examined the expression of several B7 family co-stimulatory molecules on MCs, using B cells as a positive control. In contrast to B cells, little to no expression of CD80 or CD86 was observed on either resting or LPS/IFNγ-stimulated MCs (Fig. 3D, E). MHCII expression was also lower on MCs compared to B cells. In addition, compared to B cells, MCs constitutively expressed higher levels of the inhibitory B7 family member PD-L1, which was upregulated further by LPS/IFNγ (Fig. 3D, E). The expression pattern of co-stimulatory molecules by MCs could potentially explain the lack of naive T cell activation, despite proper peptide loading of MHCII on MCs.

MCs restimulate previously activated CD4⁺ T cells

We next examined whether MCs could restimulate antigen-experienced T cells, since previously activated cells do not require the same co-stimulatory signals as naive cells. Peptide-pulsed WT but not MHCII-deficient MCs induced IFNγ production by previously activated OT-2 T cells (Fig. 4A), suggesting that MCs could participate in the reactivation of antigen-experienced T cells. However, OVA protein-treated MCs failed to stimulate IFNγ production by these T cells, suggesting that MCs may lack the ability to process whole antigens and present them on MHCII.

Figure 4. Peptide-pulsed but not protein-pulsed MCs support stimulation of previously activated T cells.

(A) WT B6 (left two plots) or MHCII^−/− (third plot from left) MCs were stimulated with LPS/IFNγ for 72 h and pulsed with or without (left plot) OVA peptide (10 μM) for the last 24 h. Some of the WT MCs were treated with LPS/IFNγ and OVA protein (50μg/ml) for 72 h (right plot). The MCs were washed and co-cultured with previously activated OT-2 T cells for 6 h, and IFNγ production was measured by intracellular staining and flow cytometry. (B) WT B6 (left three plots) or MHCII^−/− (right plot) MCs were stimulated with LPS/IFNγ and TNP-Eα protein (50 μg/ml) for 72 h. Some of the WT MCs were co-treated with anti-TNP IgG1 (20 μg/ml; second plot from left) or pretreated with anti-TNP IgE (1 μg/ml for 24 h; third plot from left). (C) WT B6 (left plot) or H-2DM^−/− (right plot) MCs were stimulated with LPS/IFNγ for 72 h and TNP-Eα protein. All results are representative of at least 2 independent experiments.

To test antigen processing and presenting ability, specific MHCII/peptide complexes on TNP-conjugated Eα (TNP-Eα) protein-treated MCs were examined. YAe staining was detected on TNP-Eα-treated WT but not MHCII-deficient MCs in an H-2DM-dependent manner (Fig. 4B, C), suggesting that MCs were able to process and present protein antigens on MHCII. To examine whether antigen uptake through Fc receptors would positively impact antigen processing and presenting ability, TNP-Eα was incorporated into MCs by TNP-specific IgG₁ or IgE. Neither YAe staining intensity nor percent-positive fraction increased through antigen incorporation by TNP-specific IgG₁ or IgE (Fig. 4B), suggesting that internalization by receptor-mediated endocytosis may divert the antigen to compartments that are distinct from macropinocytosis. Endogenously derived proteins were also processed and presented on MCs, since YAe staining was detected on LPS/IFNγ-stimulated MCs derived from mice expressing the Eα-transgene (B6-Eα) without the addition of exogenous TNP-Eα protein (Fig. 5A). The MHCII/Eα peptide complexes were derived from an endogenous source in B6-Eα MCs, since WT MCs mixed with B6-Eα MCs did not stain with YAe antibody (Fig. 5B).

Figure 5. Endogenous Eα protein is presented on MHCII by MCs.

(A) WT B6 (left plot) or B6-Eα (right plot) transgenic MCs were stimulated with LPS/IFNγ for 72 h. MHCII/Eα peptide complexes were detected on FcεRI⁺ BMMCs by YAe staining and flow cytometry. (B) WT B6 and B6-Eα BMMCs were co-cultured and treated with LPS + IFNγ. The BMMCs were distinguished from each other by labeling the B6-Eα BMMCs with CFSE. MHCII/Eα peptide complexes were measured 72 h post stimulation on FcεRI⁺ BMMCs. Results are representative of 2 independent experiments.

MCs preferentially activate Treg

To explore whether MCs could activate other subsets of CD4⁺ T cells, we tested the ability of MCs to activate antigen-specific Treg, since MCs were recently implicated in potentiating allograft tolerance through interaction with Treg(13). To obtain large numbers of antigen-specific Treg, we used TS1×HA28 mice, which express the influenza virus PR8 HA as a neo-self peptide and co-express the TS1 TCR that is specific for the PR8 HA determinant S1. Treg (CD4⁺CD25⁺Foxp3⁺) comprise ~50% of all HA-specific CD4⁺ T cells from TS1×HA28 mice (26). When FACS-sorted TS1×HA28 CD4⁺ T cells were co-cultured with influenza peptide-pulsed splenocytes, a similar proportion of proliferating Foxp3⁺ and Foxp3⁻ CD4⁺ T cells was observed (Fig. 6A). In contrast, proliferation of TS1×HA28 CD4⁺ T cells was heavily skewed towards the Foxp3⁺ fraction after co-culture with influenza peptide-pulsed LPS/IFNγ-stimulated MCs. A similar effect was observed when the LPS/IFNγ-stimulated MCs were pretreated with live or heat-inactivated PR8 virus but not with influenza virus (J1) lacking the S1 epitope, suggesting that intact influenza-derived proteins could be processed and presented to TS1/HA28 Treg by MCs (Fig. 6B). Endogenously derived HA protein was also presented by MCs, since LPS/IFNγ-stimulated MCs derived from HACII mice, which express full-length HA protein driven by the I-Eα promoter, stimulated the proliferation of TS1×HA28 Treg (Fig. 6C).

Figure 6. MCs preferentially expand regulatory T cells over naive T cells.

(A) WT BALB/c MCs were stimulated with or without (left plot) LPS/IFNγ for 72 h and pulsed with (third plot from left) or without (second plot from left) S1 peptide (10 μM) for the last 24 h. CFSE-labeled FACS-sorted TS1×HA28 T cells were co-cultured with the MCs or with irradiated BALB/c splenocytes and S1 peptide (right plot) for 4 days. (B) WT BALB/c MCs were stimulated with LPS/IFNγ and live PR8, live J1, heat-inactivated PR8, or heat-inactivated J1 virus (plots from left to right) for 72 h and co-cultured with CFSE-labeled FACS-sorted TS1×HA28 T cells for 4 days. (C) HACII MCs were stimulated with (right plot) or without (left plot) LPS/IFNγ for 72 h and co-cultured with CFSE-labeled FACS-sorted TS1×HA28 T cells for 4 days. (D) WT BALB/c MCs were stimulated with LPS/IFNγ and live PR8 for 72 h and co-cultured with CFSE-labeled FACS-sorted CD4⁺CD25⁺ (right plot) or CD4⁺CD25⁻ (left plot) TS1×HA28 T cells for 4 days. (E) WT BALB/c MCs were stimulated with LPS/IFNγ for 72 h and pulsed with S1 peptide (10 μM). CFSE-labeled FACS-sorted Thy1.2⁺ TS1 T cells were co-cultured with the MCs (left plot) or with irradiated BALB/c splenocytes and S1 peptide for 4 days (right plot). (F) WT B6 MCs were stimulated with LPS/IFNγ for 72 h and pulsed with OVA peptide (10 μM). CFSE-labeled FACS-sorted Thy1.2⁺ OT-2 T cells were co-cultured with the MCs with (left plot) or without IL-2 (50 U/ml; right plot) for 4 days. The proliferation of FoxP3⁺ and FoxP3⁻ CD4⁺ T cells was analyzed by CFSE dilution and flow cytometry. Plots A–E and F are gated on CD4⁺6.5⁺ T cells (TS1 TCR clonotype-positive T cells) and CD4⁺ T cell, respectively. All results are representative of at least 2 independent experiments.

The expansion of Treg by MCs could have resulted from the induction of Foxp3 in previously Foxp3⁻ T cells. To test this possibility, MCs were co-cultured with TS1×HA28 T cells that had been sorted into CD25⁺ and CD25⁻ fractions, which correlated well with Foxp3 expression (data not shown). MCs did not induce Foxp3 expression in Foxp3⁻ T cells, since Foxp3⁺ T cell proliferation was only observed with FACS-sorted CD25⁺ TS1×HA28 T cells but not with FACS-sorted CD25⁻ T cells (Fig. 6D).

MCs were able to stimulate the proliferation of isolated Foxp3⁻ T cells from TS1×HA28 mice (Fig. 6D), although MCs preferentially stimulated Foxp3⁺ T cells over Foxp3⁻ T cells when both subsets were present in the same culture. In the latter situation, Treg-mediated active suppression exerted on Foxp3⁻ T cells may be contributing. In support of this, MCs induced the proliferation of Foxp3⁻ naive T cells from TS1 mice, which lack the neo-self HA and are much less enriched for clonotypic FoxP3⁺ T cells (Fig. 6E). These results differ from experiments using OT-2 T cells, which showed no proliferation of naive T cells by MCs (Fig. 6F). However, the proliferation of OT-2 T cells could be slightly induced by MCs in the presence of IL-2 (Fig. 6F). It is possible that the activation of the TS-1 TCR may be less stringent than OT-2 T cells, due to differences in TCR affinity for their cognate antigens. Therefore, depending on the TCR expressed by the T cell, MCs may also be able to prime naive T cells.

LN-localized MCs increase upon inflammation and express MHCII and co-stimulatory molecules

We next asked whether MCs expressed MHCII in vivo. Similar to cultured MCs, freshly isolated peritoneal MCs were virtually devoid of MHCII expression (Fig. 7A). We predicted that peritoneal MCs might express MHCII if stimulated by LPS. However, when mice were injected intraperitoneally with LPS, MCs were no longer recovered from the peritoneal cavity (data not shown), suggesting that MCs might have migrated from the peritoneal cavity to secondary lymphoid organs upon TLR stimulation. To test this possibility, mice were treated with LPS subcutaneously, and the draining LNs were examined for the presence of MCs. Although only few MCs could be seen residing in the LNs of unchallenged mice, the number of MCs significantly increased after LPS injection peaking at ~11 days post LPS challenge (Fig. 7C). The increase in MCs was specific to LNs, since no increase in MC numbers were observed in the spleen (Fig. 7B). All LN-localized MCs expressed MHCII and PD-L1 (Fig. 7D, E). Moreover, the positive co-stimulatory B7 family members CD80 and CD86 were also expressed on LN MCs.

Figure 7. LPS increases numbers of LN MCs that express MHCII and co-stimulatory molecules.

(A) Freshly isolated peritoneal MCs from B6 mice were stained with isotype control (mouse IgG2a; left plot) or MHCII antibody (clone Y3P; right plot), and analyzed by flow cytometry. Plots are gated on CD117⁺FcεRI⁺ MCs. (B) B6 mice were injected subcutaneously with LPS (25 μg/flank), and the inguinal LNs were harvested 0, 3, or 5 days post-inoculation (plots from left to right). Spleen cells were harvested on day 3 post-inoculation (right plot). The cells were analyzed for the presence of CD117⁺FceRI⁺ MCs by flow cytometry. (C) Total numbers of CD117⁺FcεRI⁺ MCs were enumerated in inguinal LNs at various days post-subcutaneous LPS inoculation of B6 mice (n = 3 mice/time point). * indicates significance of p < 0.01 compared to mast cell numbers on Day 0 by ANOVA. (D) LN CD117⁺FcεRI⁺ MCs from B6 mice 5 days after challenge with subcutaneous LPS were stained with isotype control or anti-MHCII antibody and analyzed by flow cytometry. (E) LN CD117⁺FcεRI⁺ MCs from B6 mice 11 days after challenge with subcutaneous LPS were analyzed for expression of CD80 (top left plot), CD86 (top right plot), PD-L1 (bottom left plot), or PD-L2 (bottom right plot) by flow cytometry. Plots are all gated on CD117⁺FcεRI⁺ MCs. All results are representative of at least 2 independent experiments.

The few MCs in the LN of unchallenged mice localized to the subcapsular and trabecular sinuses (Fig. 8A). Since it was difficult to examine MC localization due to massive B cell hyperplasia and consequent distortion of LN architecture after LPS-challenge, LNs from LPS-challenged RAG-deficient mice were analyzed for the presence of MCs. LN architecture was preserved after LPS challenge of RAG-deficient mice, and although there was an increase in MC numbers, the localization of MCs was unchanged compared to unchallenged RAG-deficient mice (Fig. 8B).

Figure 8. MC numbers increase in LN during Leishmania major infection and reside in LN sinuses.

(A) Inguinal LNs were harvested from LPS-treated B6, (B) unchallenged, or (C) LPS-challenged RAG^−/− mice. Sections were taken from fixed LN, stained with toluidine blue, and visualized with light microscopy (magnification = 200×). (C) B6 mice were inoculated with L. major subcutaneously in the right footpad, and the ipsilateral (right LN) and contralateral (left LN) popliteal LN were harvested separately on 7 days post-inoculation. Total MC numbers in the LN were enumerated by flow cytometry (n = 4 mice). * indicates significance of p < 0.01 by ANOVA. (D) Sections from popliteal LN on day 0 (right image) and 10 (left image) post L. major inoculation were stained with toluidine blue and visualized with light microscopy (magnification = 200×). All results are representative of at least 2 independent experiments.

We next tested whether a more physiologic inflammatory stimulus provided by a pathogen would yield similar results to that of LPS. Leishmania major was chosen, since the cutaneous infection remains localized with defined lymphatic drainage. Therefore, mice were challenged with Leishmania major subcutaneously in one footpad, and the draining popliteal LNs were examined for the presence of MCs. Similar to the findings after treatment with LPS, a significant increase in MC numbers was found in ipsilateral LN compared to LN contralateral to the infected footpad (Fig. 8C). Again, localization of MCs was restricted to LN sinuses (Fig. 8D). Collectively, these data suggest that upon inflammation, MCs accumulate in draining LNs and express both MHCII and co-stimulatory molecules necessary for antigen presentation.

LPS protects against MC death

We have recently reported that IgE crosslinking by cognate antigen induces apoptosis of MCs (22). The antigen-incorporated apoptotic MCs then serve as a source of antigen to be presented to T cells by DCs. However, for MCs to be involved in the direct presentation of antigen on MHCII, the MCs must survive. Thus, we tested the effects of LPS/IFNγ on mast cell survival. In contrast to FcεRI-crosslinking, LPS/IFNγ stimulation protected against MC apoptosis in a MyD88-dependent manner (Fig. 9A). The cytoprotective effect did not require IFNγ and was not a direct effect of LPS on MCs, since co-culture of WT MCs with MyD88^−/− MCs protected MyD88^−/− MCs from apoptosis (Fig. 9B). Upon further investigation, LPS was found to induce IL-3 from MCs (Fig. 9C). Blockade of IL-3 by neutralizing antibody reversed the cytoprotective effect of LPS on MCs (Fig. 9D), suggesting that LPS protects MCs against apoptosis by inducing IL-3. Therefore, we propose that there may be two distinct ways that MCs could be involved in antigen presentation: one involving antigen-incorporated apoptotic MCs through FcεRI-crosslinking and another involving LPS-induced survival and MHCII expression by MCs.

Figure 9. LPS promotes survival of MCs by induction of IL-3.

(A) WT (left graph) or MyD88^−/− (right graph) BMMCs were cultured in IL-3-free MCM in the presence or absence (open circles) of LPS (closed circles) or LPS + IFNγ (open triangles). MC apoptosis was measured by AnnexinV staining on days 0, 2, and 3 post-culture. (B) WT (left graph) or MyD88^−/− (right graph) cells were mixed and co-cultured in IL-3-free MCM in the presence (closed circles) or absence (open circles) of LPS. The BMMCs were distinguished from each other by labeling the MyD88^−/− BMMCs with CFSE. (C) RNA was extracted from FcεRI⁺CD117⁺ FACS-sorted B6 BMMCs stimulated with LPS for 0, 6, 24, 48, or 72 h. The presence of IL-3 and β-actin transcripts was detected by RT-PCR. (D) WT (left graph) or MyD88^−/− (right graph) cells were mixed and co-cultured in IL-3-free MCM in the presence (closed circles) or absence (open circles) of LPS with (open and closed triangles) or without anti-IL-3 antibody. The BMMCs were distinguished from each other by labeling the MyD88^−/− BMMCs with CFSE. All results are representative of at least 2 independent experiments.

Discussion

We demonstrate herein that cultured MCs express MHCII after stimulation with LPS/IFNγ. Concomitant expression of MHCII-associated molecules as well as the inhibitory co-stimulatory molecule PD-L1 was observed, while positive co-stimulatory B7 family members CD80 and CD86 were not detected. MHCII-bearing mast cells stimulated antigen-specific naive T cells in certain situations, as MCs activated TS-1 but not OT-2 TCR transgenic naive CD4⁺ T cells. However, MCs were fully capable of stimulating previously activated T cells as well as Treg.

MHCII expression by MCs has been controversial. Earlier reports claimed that MHCII is constitutively expressed on cultured MCs (14, 40), while more recent studies have failed to observe MHCII expression on resting cultured MCs (21, 22). Futhermore, OVA peptide-pulsed BMMCs were found to be poor stimulators of OT-2 T cells (41). The study described here sheds light on this controversy by demonstrating that MHCII can be induced on MCs when activated with appropriate stimuli such as LPS/IFNγ. One may speculate that the discrepancies among reports resulted from the potential use of endotoxin-contaminated reagents in some studies. Indeed, an earlier report demonstrated that IFNγ was contained in WEHI conditioned medium used to grow MCs and that together with LPS further enhanced the constitutive expression of MHCII by MCs (14). It is unclear why MCs require both LPS and IFNγ for expression of MHCII. IFNγ receptor was constitutively expressed on mast cells and mast cells functionally responded to IFNγ by increasing MHC class I expression (data not shown), suggesting that LPS is not required for IFNγ responsiveness.

Previous studies have reported that IL-4 and GM-CSF enhance while IFNγ decreases the antigen presenting capability of MCs (40). This outcome is in disagreement with our present results, as MHCII was not observed in MCs cultured with IL-4 and GM-CSF in the absence of LPS. Previous studies have also argued that incorporation of antigen by IgE converts IFNγ-treated MCs into potent APCs (42). However, we found that antigen incorporation by IgE does not facilitate presentation of antigens on MHCII. In fact, antigens incorporated by IgE/FcεRI were protected against proteolytic degradation and were preserved as an intact protein much longer than those antigens acquired by macropinocytosis (T. Zou, unpublished observations). Our findings are supported by a recent report demonstrating that Escherichia coli incorporated through IgE/FcεRI are protected from proteolysis and remain viable in MCs (43). Perhaps the enhancement of antigen presentation by IgE in IFNγ-treated MCs occurs in an MHCII-independent manner, e.g. by transferring IgE-incorporated antigens to DCs by exosomes or as apoptotic bodies (20, 22).

It is puzzling why MCs presented TNP-Eα and influenza but not OVA protein on MHCII. This discrepancy may be explained by the presence of a specific uptake mechanism for those proteins that were presented on the MHCII of MCs. Since Eα protein was produced as a recombinant protein in bacteria, TNP-Eα could potentially be coupled to endotoxin or other bacteria-derived products that may facilitate incorporation into MCs through TLRs (44). The influenza virus can be incorporated by receptor-mediated endocytosis using HA/sialic acid interactions. In contrast, the OVA used in our experiments contained low levels of endotoxin (Grade V OVA) and possesses no other means of incorporation into MCs other than macropinocytosis. Thus, it is possible that a specific uptake mechanism of antigens, perhaps those associated with pathogen recognition such as TLRs, is required for efficient processing and presentation of exogenous antigens on MHCII by MCs.

MHCII expression by MCs was not merely an in vitro phenomenon. Compared to unchallenged mice, LNs of LPS- or Leishmania-inoculated mice contained significantly increased numbers of MCs, all expressing high levels of MHCII and B7 co-stimulatory family members (CD80, CD86, PD-L1). The ratio of MCs to total LN cells was only slightly increased, suggesting that the increase in MC numbers was proportional to LN hyperplasia. LN of unchallenged mice contained fewer MCs but all expressed MHCII and co-stimulatory molecules (data not shown). These MCs may represent those recruited from skin or mucosal sites after stimulation by normal flora or environmental irritants. Upon infection, more MCs may be recruited to LN leading to increased numbers of MCs. In support of this argument, MCs were detected in the popliteal LN of Leishmania-infected MC-deficient Wsh/Wsh mice that have been reconstituted with MCs in their footpads (Kambayashi et al., unpublished observations). Furthermore, previous studies by others have demonstrated migration of MCs to LN under allergic and bacterial inflammation (45, 46). Alternatively, MCs found in the LN may represent a subtype of MCs that are LN-resident, have antigen-presenting capability, and expand upon inflammation. In support of this hypothesis, all MCs in the LN of LPS-injected mice incorporated BRDU suggesting that these MCs have expanded by proliferation (Kambayashi et al., unpublished observations). However, the possibility that the expansion of MCs took place at a remote site and later migrated to the LN cannot be excluded.

The central question that remains is the function of MHCII on MCs in vivo. Given the potency of DCs in stimulating naive T cells, it is unlikely that MCs play a major role in initiating primary T cell responses. Indeed, MCs were poor stimulators of naive T cells in vitro, most likely resulting from the absence of co-stimulatory molecules. It is more likely that MCs participate in the reactivation or propagation of activated T cells, as MHCII/peptide-bearing MCs stimulated the production of IFNγ from antigen-experienced T cells, which do not require co-stimulation for reactivation. However, given that LN-localized MCs express CD80 and CD86, some contribution of MCs to naive T cell priming in vivo cannot be excluded.

The role for MHCII expression on MCs may be to activate Treg and dampen the immune response or avoid self-reactivity. MCs stimulated the antigen-specific proliferation of Treg and favored their activation over naive T cells in mixed co-cultures. Activation of Treg by MCs may contribute to the protective effect of MCs on skin allografts, a process that was proposed to involve IL-9 production by Treg to recruit mast cells to the graft site (13). Bidirectional communication may take place between MCs and Treg, of which one involves antigen presentation by MCs to Treg. Endogenous proteins were presented well on MHCII of MCs, and thus many of the bound peptides may be self-derived, which would favor the notion that MCs activate Treg. The interaction of T cells and MCs could take place in LNs where MCs are situated to encounter cells that drain through the lymphatic sinuses.

How MCs preferentially stimulate Treg is uncertain. A recent study demonstrated that PD-L1 is necessary for the generation of adaptive Treg by antigen-primed DCs (47). Adaptive Treg differ from natural Treg in that they are conventional CD4⁺ T cells that have post-thymically acquired Foxp3. PD-L1 appears not to be involved in our system involving natural Treg, since blockade of PD-L1 by anti-PD-L1 antibody had no effect on the proliferation of Treg from TS-1×HA28 mice (data not shown). However, it is possible that PD-L1 on MCs is involved in conversion of CD4⁺ T cells into adaptive Treg under certain conditions. Further studies involving graft rejection models or infectious disease models will be required to understand how the acquisition of antigen presenting capability by MCs contributes to the overall function of MCs in physiological and pathologic states.