Suppression of Dendritic Cell Maturation and T Cell Proliferation by Synovial Fluid Myeloid Cells from Mice with Autoimmune Arthritis

Abstract

Objective

To determine whether myeloid cells (such as granulocytes) present in the synovial fluid (SF) of arthritic joints have an impact on adaptive immunity. Specifically, we investigated the effects of SF cells, harvested from the joints of mice with proteoglycan (PG)-induced arthritis (PGIA), on dendritic cell (DC) maturation and antigen-specific T-cell proliferation.

Methods

We monitored DC maturation (MHC class II and CD86 expression) by flow cytometry upon co-culture of DCs with SF or spleen myeloid cells from mice with PGIA. The effects of these myeloid cells on T-cell proliferation were studied using T cells purified from PG-specific T cell receptor transgenic (PG-TCR-Tg) mice. Phenotypic analysis of myeloid cells was performed employing immunostaining, RT-PCR, Western blot, and biochemical assays.

Results

Inflammatory SF cells significantly suppressed the maturation of DCs upon co-culture. PG-TCR-Tg T cells cultured with antigen-loaded DCs showed dramatic decreases in proliferation in the presence of SF cells. Spleen myeloid cells from arthritic mice did not have suppressive effects. SF cells were unable to suppress CD3/CD28-stimulated proliferation of the same T cells, suggesting a DC-dependent mechanism. SF cells exhibited all of the characteristics of myeloid-derived suppressor cells (MDSCs), and exerted suppression primarily through production of nitric oxide and reactive oxygen species by granulocyte-like cells.

Conclusion

SF in the joints of mice with PGIA contains a population of granulocytic MDSCs that potently suppress DC maturation and T-cell proliferation. These MDSCs have the potential to limit the expansion of autoreactive T cells, thus breaking the vicious cycle of autoimmunity and inflammation.

Granulocytes (mainly neutrophils) are abundantly present in the synovial fluid (SF) of inflamed joints in patients with rheumatoid arthritis (RA) (1,2), and these cells also constitute a major population of joint-infiltrating cells in murine models of RA including proteoglycan (PG)-induced arthritis (PGIA) (3–7). Neutrophils can inflict considerable damage to joint tissues via secretion of proteinases, reactive oxygen species (ROS), cytokines, and chemoattractants (8). In addition, they can interact with other cell types such as dendritic cells (DCs) that are also present in arthritic joints (9,10). Effects of these neutrophils on joint-resident DCs, which migrate to the joint draining lymph nodes (JDLNs) (11) and may present joint-derived autoantigens (autoAgs) to T cells, could be of importance. In both human and murine systems, activated neutrophils have been shown to induce the maturation of DCs via cell-cell contact and secretion of DC-activating cytokines (12–14). Such an interaction between joint-resident neutrophils and DCs could increase the autoimmune response through enhancement of both the migration of DCs and their capacity to present joint-derived autoAgs to T cells in the JDLNs. Conversely, if suppressive subsets of neutrophils or other myeloid lineage cells are present in an inflamed joint, they may prevent the spreading of arthritis to other joints by inhibiting the maturation of DCs, thus limiting the activation of autoreactive T cells in the JDLNs.

A recently described cell population, termed myeloid-derived suppressor cells (MDSCs), has been implicated in the suppression of T cell activation. MDSCs are a heterogeneous group of cells that belong to the CD11b⁺ myeloid lineage (15). First identified in cancer patients and tumor-bearing animals, MDSCs were later found to be enriched under the conditions of infection, organ transplantation, and autoimmunity (reviewed in (16)). In mice, two major populations of MDSCs (Ly6-G^hiLy6-C^int/lo and Ly6-G^neg/loLy6-C^hi) have been distinguished (15,17). Morphologically, the Ly6-G^hiLy6-C^int/lo subset resembles granulocytes (neutrophils), whereas the Ly6-C^hiLy6-G^neg/lo subset consists of monocyte-like cells. The suppressive activities of MDSCs have been mechanistically linked to their upregulation of arginase 1, inducible nitric oxide (NO) synthase (iNOS), and production of ROS (16,18–20).

Although the presence of large populations of joint-infiltrating granulocytic cells in RA and PGIA has long been known, the possibility of joint-resident neutrophils activating DCs (thus potentially enhancing autoimmunity) or acting as MDSCs (thus suppressing autoimmunity), has not been considered. Therefore, the primary goal of this study was to determine whether SF (and spleen) cells in mice with PGIA contained a population that could promote or suppress DC maturation, with a potential to affect the DC-mediated activation of T cells.

MATERIALS AND METHODS

Mice, immunization, and assessment of arthritis

Adult female BALB/c mice were purchased from the National Cancer Institute (Frederick, MD). BALB/c mice, expressing a T-cell receptor (TCR) specific for an arthritogenic epitope within the G1 domain of human cartilage PG (PG-TCR-Tg), have been characterized previously (21). Mice expressing enhanced green fluorescent protein knocked in the myeloid cell-specific lysozyme M promoter (EGFP-LysM KI) (22) were back-crossed to BALB/c for 12 generations (7).

PG was extracted from human cartilage as described in detail elsewhere (7,23–26). Collection of cartilage from joint replacement surgery was approved by the Institutional Review Board of Rush University Medical Center. Recombinant human PG G1 domain (rhG1) protein was produced as described previously (27). To induce PGIA, mice were immunized intraperitoneally with PG (100 μg protein) (23,24) emulsified in dimethyl-dioctadecyl-ammonium bromide adjuvant (Sigma-Aldrich, St Louis, MO) two or three times, 3 weeks apart (7,26). Mice were inspected for arthritis symptoms (swelling and redness) regularly after the second PG immunization. The degree of arthritis was scored visually on a scale of 0 to 4 for each limb (7,26,27). All experiments involving animals were approved by the Institutional Animal Care and Use Committee of Rush University Medical Center.

Flow cytometry, cytospin preparations, and immunocyto/histochemistry

Cells were isolated from the blood, spleens, JDLNs, bone marrow (BM) of naïve and arthritic mice. SF was harvested from arthritic ankles (at the peak of acute inflammation; score of 3 or 4) and swollen knees of euthanized mice after puncturing the joints under aseptic conditions (7). Fc receptors were blocked with Fc Block (BD Biosciences, San Diego, CA) prior to the specific staining with fluorescence-conjugated monoclonal antibodies (mAbs) against Gr-1, CD11b, Ly6-C, Ly6-G, CD11c, MHC II (I-A/I-E), CD86, CD3, or CD4, or with IgG isotype controls (Abs from BD Biosciences or eBioscience, San Diego, CA). Flow cytometry was performed using a BD FACS Canto II instrument and FACS Diva software (BD Flow Cytometry Systems, San Jose, CA). Occasionally, immunostained cells were examined by fluorescence microscopy. Cell samples were also spun onto glass slides and stained with Wright-Giemsa (Sigma-Aldrich). The cytospin preparations were examined by light microscopy (Nikon, Melville, NY). Immunohistochemistry with fluorescence-labeled mAbs to Ly6-G and Ly6-C (and isotype controls) was performed as described (7) on frozen sections of inflamed joints and spleens of mice with PGIA.

T-cell enrichment and spleen myeloid cell selection

T cells were enriched from the spleens of naïve PG-TCR-Tg BALB/c mice (21) using a T-cell negative selection kit (StemCell Technologies, Vancouver, Canada) (7). The purity of T cells, assessed by flow cytometry, was typically 95% or greater. Myeloid (CD11b⁺) cells were positively selected from the spleens of arthritic mice with anti-CD11b mAb-conjugated microbeads followed by AutoMacs column separation (Miltenyi Biotec, Auburn, CA), or negatively selected by depletion of non-myeloid cell populations using a biotin selection kit (StemCell).

Generation of BM-derived DCs

DCs were generated from BM as described (28). Briefly, BM was harvested from femurs and tibias of naïve BALB/c mice, and cultured at a density of 2 × 10⁵ cells per ml Dulbecco’s Modified Eagle Medium (DMEM; Sigma-Aldrich) containing 10% fetal bovine serum (FBS) (Hyclone, Logan, UT) and 200 U/ml of recombinant murine granulocyte macrophage colony-stimulating factor (rmGM-CSF; Sigma-Aldrich). On day 9 or 10 of culture, non-adherent cells could be used as DCs on the basis of their CD11c expression. The BM-DCs exhibited a myeloid (or conventional) DC phenotype similar to the DCs found in SF, and DCs from both sources expressed MHC II (see Figure 1D).

Figure 1.

Synovial fluid (SF) cells, but not spleen myeloid cells, from arthritic mice suppress the maturation/activation of dendritic cells (DCs) in vitro. A, DCs were cultured alone or in the presence of SF cells in direct contact or separated by filter inserts for 24 h. B, DCs were also cultured for 96 h alone or in the presence of SF cells or spleen myeloid cells. In all cases, the DC:SF or DC:spleen myeloid cell ratio was 1:3, and MHC II and CD86 expression on the DCs was determined by flow cytometry. The results are expressed as the ratios of mean fluorescence intensity (MFI) values (changes relative to the MFI of the positive control [DCs cultured alone]). Data shown are the means ± SEM from three independent experiments. * = P < 0.05. C, Flow histograms of MHC II and CD86 expression levels in DCs cultured alone or with SF cells. D, MHC II levels were compared in bone marrow-derived DCs and inflammatory SF-derived DCs (both isolated from EGFP-LysM KI BALB/c mice that express EGFP in myeloid cells (7,22)) by flow cytometry (upper panels) and fluorescence microscopy (lower panels). Blue circles depict MHC II high DCs in the flow panels.

Co-culture of DCs, myeloid cells, and T cells

BM-DCs were cultured in DMEM containing 10% FBS either alone or with SF or spleen myeloid cells at a DC:myeloid cell ratio of 1:3 in 24-well plates. Occasionally, DCs were separated from the other cells by 0.4 μm pore-size filter inserts (Millipore, Billerica, MA). Expression of the maturation markers MHC II and CD86 in DCs was determined by flow cytometry.

For Ag-specific T-cell proliferation assays, DCs were first incubated with the recombinant G1 domain of human PG (rhG1, 75 μg/ml) (27) in the absence or presence of SF cells or spleen myeloid cells for 24 h in 96-well culture plates. T cells, purified from naïve PG-TCR-Tg mice, were added to the DCs or DC-SF or DC-spleen myeloid cell co-cultures, and cultured for another 84–96 h. The ratios of T cells, DCs and SF (or spleen myeloid) cells were 3:1:1, 3:1:2, or 3:1:3. Occasionally, the monocytic subset was removed from SF or spleen myeloid cells by immonomagnetic depletion with an anti-Ly6-C mAb prior to culture with DCs and T cells. In a limited set of experiments, DCs were isolated from the SF using a positive CD11c selection kit (StemCell). SF-DCs were loaded with rhG1, and cultured with PG-TCR-Tg T cells for 96 h. For anti-CD3/CD28-induced cell proliferation, 96-well culture plates were coated with purified anti-CD3 and anti-CD28 mAbs (1 μg/well each; eBioscience). PG-TCR-Tg T cells were cultured in the coated wells in the absence or presence of SF or spleen myeloid cells as described above. The effect of MHC II blockade on the Ag-presenting function of BM-DCs was examined by culturing rhG1-loaded DCs with PG-TCR-Tg T cells with a serially-diluted blocking anti-MHC II mAb (clone M5/114.15.2; eBioscience) (29). Cultures were pulsed with [³H]thymidine (Perkin Elmer, Boston, MA) at 1 μCi/well 18 hours prior to harvest. Isotope incorporation was measured in a scintillation counter.

Inhibitors of MDSC products, added to the co-cultures (alone or in combination) were as follows: the arginase inhibitor Nω-hydroxy-nor-arginine (nor-NOHA; 0.5 mM), the iNOS inhibitor N5-[imino(methylamino)methyl]-L-ornithine citrate (L-NMMA; 2 mM), the ROS scavenger catalase (1000 U/ml), and the apoptosis (caspase) inhibitor Z-VAD-FMK (0.1 mM). Drug-induced DC apoptosis control was set up by culturing DCs alone with the apoptosis-inducing topoisomerase inhibitors camptothecin (10 μM) and etoposide (1 μM) (30,31) in the absence or presence of Z-VAD-FMK. Inhibitors were purchased from Calbiochem (Gibbstown, NJ) or Sigma-Aldrich.

Reverse transcription-polymerase chain reaction (RT-PCR)

Total RNA was isolated from cells using TRI reagent (Sigma-Aldrich) according to the manufacturer’s instruction. cDNA was synthesized using a SuperScript First Strand kit (Invitrogen, Carlsbad, CA), and PCR was performed with HotStart Taq Plus enzyme (Qiagen, Carlsbad, CA) in 35 cycles (95°C/20 sec, 57°C/30 sec, and 72°C/45 sec) with a final extension at 72°C for 10 min in a C1000 Thermal Cycler (Bio-Rad, Hercules, CA). The sequences of the arginase 1 (Arg-1) primers were as follows: Arg-1 forward 5′-CAGAAGAATGGAAGAGTC AG-3′, Arg-1 reverse 5′-CAGATA TGCAGGGAGTCACC-3′ (18). An iNOS-specific primer pair (Cat. No.: RDP-101) was purchased from R&D Systems (Minneapolis, MN), and the housekeeping (β-actin) gene-specific primer sequences were as follows: Act-b forward 5′-TGACAGGATGCAGAAGGAGA-3′, Act-b reverse 5′-GCTGGAAGGTGGACAGTGAG-3′. After amplification, reaction products were loaded onto 1.5% agarose gel.

Western blot

Cells were lysed in cold RIPA buffer containing protease inhibitors. Cell lysates were loaded onto 12% SDS-PAGE gel (20 μg protein per lane), separated under reducing conditions, and transferred to nitrocellulose membranes. The membranes were probed with an anti-arginase 1 Ab (Santa Cruz, Santa Cruz, CA), or with an anti-iNOS Ab (Abcam, Cambridge, MA). Horseradish peroxidase (HRP)-conjugated rabbit anti-goat IgG (Invitrogen) was used as a secondary Ab. The protein bands were visualized using enhanced chemiluminescence (Amersham, Arlington Heights, IL). The membranes were stripped and re-probed with a HRP-conjugated Ab to β-actin (Abcam) to confirm equal sample loading.

Measurement of arginase and iNOS activity and ROS production

Arginase activity was measured in cell lysates on the basis of urea production from L-arginine, as described elsewhere (32). iNOS activity was determined by measuring the amounts of NO in cell culture supernatants with a nitrate/nitrite colorimetric assay kit (Cayman Chemical, Ann Arbor, MI) according to the manufacturer’s instructions. Intracellular ROS was detected using 2′2′-dichloro-dihydro-fluorescein diacetate (Invitrogen), a cell-permeant non-fluorescent compound that is converted to a fluorescent product in the presence of ROS (33).

Statistical analysis

Statistical analysis was performed using SPSS software (SPSS, Chicago, IL). Data from two groups were compared using the independent samples Student’s t-test or the non-parametric Mann-Whitney U-test. Multiple group comparisons were made using ANOVA with post-hoc Dunnett’s t-test. P values of less than 0.05 were accepted as statistically significant.

RESULTS

Synovial fluid cells suppress the maturation of DCs

To study the possible effect of SF cells on DC maturation, we examined the magnitude of MHC II and CD86 expression by DCs after culturing them alone or in the presence of SF cells. Based on studies reporting enhancement of DC maturation by activated neutrophils (12–14), we anticipated that SF cells (containing >90% Gr-1⁺CD11b⁺ myeloid cells with neutrophil morphology) would increase both MHC II and CD86 expression in DCs upon co-culture. Surprisingly, DCs showed significantly decreased levels of MHC II and lower expression of CD86 after 24 h of culture in the presence of SF cells compared to culture without SF cells (Figure 1A). To determine whether this suppressive effect of SF cells on DC maturation required cell-cell contact, we performed the same experiments, but with SF cells separated from DCs by 0.4 μm pore-size filter inserts. Although the SF cells were unable to contact DCs in this system, they retained their ability to significantly reduce the expression of MHC II and CD86 on the DCs (Figure 1A).

As joint-resident DCs are likely exposed to SF cells for more than 24 h in vivo, we examined the effects of SF cells on DCs after a longer (96 h) co-culture. Since CD11b⁺ (myeloid) cells in the spleen have been found to expand in tumor models, as well as in experimental autoimmune encephalomyelitis (EAE) (16,34), and they exhibit a phenotype similar to SF cells in PGIA (see Figure 3A), we also investigated the possible effects of spleen myeloid cells on DC maturation. While DCs cultured with SF cells for 96 h further reduced DC maturation as compared to the 24 h co-cultures, spleen myeloid cells showed no significant suppression (Figure 1B). Analysis of the histograms of MHC II and CD86 staining indicated that the loss in expression was due to the disappearance of an MHC II^hi and CD86^hi DC population, rather than the result of a global reduction of MHC II and CD86 levels (Figure 1C). Interestingly, comparison of MHC II expression levels in BM-derived DCs and DCs isolated from the SF also revealed markedly reduced proportion of MHC II^hi cells among SF-DCs (Figure 1D).

Figure 3.

SF and spleen myeloid cells of arthritic mice contain populations with phenotypes characteristic of myeloid-derived suppressor cells (MDSCs). A, SF and spleen myeloid cells from arthritic mice were examined by flow cytometry for CD11b, Gr-1, Ly6-G, and Ly6-C expression. Subsets of Ly6-C^hiLy6-G^lo and Ly6-G^hiLy6-C^lo cells were detected among the myeloid (CD11b⁺Gr-1⁺) cells in both the SF and myeloid-enriched spleen cell populations. B, To assess cell morphology, cytospin preparations were stained with Wright-Giemsa and examined by light microscopy for the proportions of polymorphonuclear (granulocyte-like) and mononuclear (monocyte-like) cells. Representative samples of 7 SF and 5 spleen myeloid cell preparations are shown. C, To determine the tissue localization of granulocytic and monocytic cells, frozen sections, prepared from the inflamed hind limbs and spleens of arthritic mice (n = 4), were immunostained with a red fluorescence-conjugated anti-Ly6-G and a green fluorescence-labeled anti-Ly6-C antibody. The left-side panel is the immunohistochemistry image of an arthritic ankle joint section. ST = synovial tissue, SL = synovial lining, SF = synovial fluid. The right-side panel is the immunohistochemistry image of a spleen section.

SF cells suppress Ag- and DC-dependent proliferation of T cells in vitro

To study the effect of SF or spleen myeloid cells on the proliferation of Ag-specific T cells, we cultured Ag (rhG1)-loaded DCs with cognate T cells (isolated from naïve PG-TCR-Tg mice) in the presence of absence of SF or spleen myeloid populations. As shown in Figure 2A, T cells cultured with DCs and rhG1 (G1) in the presence of SF myeloid cells, but not in the presence of spleen myeloid cells (at a DC:myeloid cell ratio of 1:3), showed a dramatic decrease in proliferation as compared to T cells cultured with DCs and G1 only. Significant suppression of T-cell proliferation was still observed at lower DC:SF cell ratios (1:2 or 1:1), suggesting a high suppressive potency of SF cells. T cells cultured with G1 or with G1 and SF cells, but without DCs, showed negligible uptake of [³H]thymidine, and so did SF and spleen myeloid cells when cultured alone or with G1 (Figure 2A), indicating that these myeloid populations had neither Ag-presenting function nor proliferative capacity.

Figure 2.

SF cells, but not spleen myeloid cells, exert a profound suppressive effect on antigen-and DC-dependent T-cell proliferation. A, DCs alone (black bar) or mixed with SF cells (at the indicated ratios; dark gray bars) or spleen myeloid cells (light gray bar) from arthritic mice, were cultured with recombinant PG G1 domain (G1) for 24 h. T cells purified from PG-TCR-Tg mice were co-cultured with the DCs or CD-SF cell or DC-spleen myeloid cell mixtures for 96 h. Open bars represent the negative controls. The results are expressed as changes in [³H]thymidine incorporation relative to the positive control (black bar) (n = 4 experiments). B, The poor T-cell proliferation, mediated by SF cell-suppressed bone marrow-derived DCs (DC) or SF-derived DCs (SF-DC), was reproducible with a blocking anti-MHC II mAb. Data show [³H]thymidine incorporation (cpm) (n = 4 wells per treatment). C, Proliferation of CD3/CD28-stimulated PG-TCR-Tg T cells in the absence (black bar) or presence of SF cells (dark gray bar) or spleen myeloid cells (light gray bar). Open bar indicates proliferation of unstimulated T cells. Results (n = 3 experiments) are expressed as in panel A. Values shown are the means ± SEM. * = P <0.05.

DCs isolated from SF (SF-DC) showed some Ag-presenting capacity, but T-cell proliferation was much weaker in the presence of SF-DCs than in the presence of unsuppressed BM-derived DCs (Figure 2B). MHC II expression levels on DCs likely played a crucial role in mediating T-cell proliferation in this system, as the poor proliferation of T cells observed either in the presence of SF cell-suppressed BM-DCs or in the presence of SF-DCs could be “mimicked” by addition of small amounts of a blocking anti-MHC II mAb (29) to the BM-DC-T cell co-cultures (Figure 2B).

Unexpectedly, neither SF cells nor spleen myeloid cells had the ability to suppress the anti-CD3/CD28-induced (DC-independent) proliferation of PG-TCR-Tg T cells (Figure 2C). In fact, spleen myeloid cells added to the CD3/CD28-stimulated T cell cultures enhanced T-cell proliferation (Figure 2C), which was probably due to the activation and expansion of “contaminating” T cells (initially ~3–5%) in the spleen myeloid cell preparation.

SF and spleen cells of mice with PGIA contain Gr-1⁺CD11b⁺ cells that show cell surface expression of MDSC-associated markers

The suppressive activity of SF cells of arthritic mice prompted us to determine if SF and the spleen contained cell populations that exhibit the phenotypic characteristics of MDSCs. As reported before (7), SF collected from the arthritic joints of mice was dominated by Gr-1⁺CD11b⁺ cells, and so was the purified myeloid population of the spleens from the same animals (Figure 3A, upper panels). We could identify both Ly6-G^hiLy6-C^int/lo (granulocyte-like) and Ly6-C^hiLy6-G^neg/lo (monocyte-like) (17) subsets in the SF and spleen myeloid populations of arthritic mice (Figure 3A, lower panels). The majority of SF cells was Ly6-G^hiLy6-C^int/lo (granulocyte-like), although it should be noted that it is not possible to phenotypically distinguish such cells from mature (terminally differentiated) neutrophils. The Ly6-C^hiLy6-G^neg/lo (monocytic) subset was smaller in the SF (~1% of all cells) than among spleen myeloid cells (~4%) (Figure 3A). Although SF cells and spleen myeloid cells exhibited very similar profiles in terms of Ly6-G and Ly6-C expression, cell morphology on Wright-Giemsa stained cytospin preparations (Figure 3B) revealed a predominant population of polymorphonuclear (granulocytic) cells in the SF, and a mixed population of polymorphonuclear and mononuclear (monocytic) cells in the spleen myeloid cell preparations. Immunohistochemistry on frozen sections of inflamed hind limbs showed abundant Ly6-G staining in the synovial tissue and SF of ankle joint, and a distinct Ly6-C staining of synovial lining cells, while Ly6-G⁺ and Ly6-C⁺ cells were sparsely detected in the spleen sections of the same mice (Figure 3C).

SF cells of arthritic mice exhibit marked upregulation of MDSC-related genes and proteins and exhibit MDSC-like activities

To screen for expression of MDSC-related genes such as arginase 1 (Arg-1) and iNOS (Nos-2), we examined various tissues from naïve and arthritic mice for mRNA expression by RT-PCR. As shown in Figure 4A, both Arg-1 mRNA and Nos-2 mRNA were found in the SF and, in a lesser amount, in the spleen of arthritic mice, but these transcripts were essentially undetectable in the spleen of naïve mice. The other organs tested did not show elevated transcription of these genes. As revealed by Western blot, SF from arthritic joints contained large amounts of both arginase 1 and iNOS proteins (Figure 4B), while spleen from the same arthritic mice contained neither. Interestingly, arthritic mice did have a positive band in all tested organs, but at a lower molecular mass (~50–55 kDa) than was expected of iNOS (~120 kDa) (Figure 4B). While the low molecular-mass band may represent a cleavage fragment of iNOS, identification of this protein requires further investigation.

Figure 4.

SF cells and spleen cells of arthritic mice express genes and proteins characteristic of MDSCs. A, Screening of blood, SF, bone marrow (BM), lymph node (LN), and spleen cells from arthritic (arth) and naïve mice for arginase-1 (Arg-1), and iNOS (Nos-2) mRNA showed high expression of both transcripts in SF cells, low expression in arthritic spleen cells, and negligible levels in cells isolated from blood, BM, or LNs. The housekeeping gene (Act-b) encoding β-actin, was expressed at equal levels. B, Western blot for arginase-1 and iNOS showed expression of both proteins (arrowheads) almost exclusively in SF cells, although a lower molecular-mass protein was detected in all samples from arthritic (but not naïve) mice on the iNOS blot. The β-actin control demonstrated nearly equal sample loading. Representative results from one of two replicate experiments are shown.

Arginase and iNOS activity, as well as ROS production, were predominantly detected in SF cells, although they were also found at lower levels in spleen myeloid cells (data available from the authors).

iNOS activity and ROS are involved in the SF cell-mediated suppression of both DC maturation and Ag- and DC-dependent T-cell proliferation

To determine the potential mechanisms of SF cell-mediated down-regulation of the maturation and Ag-presenting capacity of DCs, we co-cultured DCs with SF cells in the absence or presence of the arginase inhibitor nor-NOHA, the iNOS inhibitor L-NMMA, the ROS scavenger catalase, or the apoptosis/caspase inhibitor Z-VAD-FMK. The arginase inhibitor did not rescue DC maturation, but the iNOS inhibitor and ROS scavenger almost fully reversed the suppressive effect of the SF cells (Figure 5A). Whereas Z-VAD-FMK protected DCs from an apoptosis-related drop in both MHC II and CD86 levels (induced by treatment with camptothecin and etoposide), Z-VAD-FMK was ineffective in preventing the SF cell-mediated loss of the DCs’ MHC II and CD86 expression.

Figure 5.

Inhibition of iNOS and ROS reverses the SF cell-mediated suppression of DC maturation and partially restores T-cell proliferation. A, DCs were cultured in the absence or presence of SF cells for 96 h with or without the arginase inhibitor nor-NOHA, the iNOS inhibitor L-NMMA, the ROS scavenger catalase, or the apoptosis inhibitor Z-VAD-FMK. As an additional control, apoptosis was induced in DCs (cultured alone) with camptothecin and etoposide (Campto+Etopo) in the absence or presence of Z-VAD-FMK. MHC II and CD86 expression levels in the DCs were determined by flow cytometry. The results are the means ± SEM and are expressed as ratios of MFI values. Data from 3–4 experiments are shown. * = P < 0.05. B, G1-loaded DCs were cultured with PG-TCR-Tg T cells without SF cells (black bar), with SF cells, or with SF cells depleted in the Ly6-C⁺ population (Ly6-C-depl.), in the absence or presence of the indicated inhibitors (gray bars). T-cell proliferation was determined in 96-h co-cultures. The results shown are the means ± SEM from 3–5 experiments, and are expressed as ratios of [³H]thymidine incorporation relative to the positive control (black bar). * = P < 0.05.

Just as in the case of the down-regulation of DC maturation, iNOS and ROS appeared to be the dominant factors in suppressing T-cell proliferation, although their inhibitors did not fully restore the proliferation of T cells either alone or in combination (Figure 5B). The SF cell-mediated suppression of T cells remained unchanged in the presence of the caspase inhibitor (Figure 5B), excluding the possibility that suppression was due to apoptosis of T cells or any other cell type in the co-culture. Importantly, SF cells depleted in the Ly6-C⁺ monocytic subset still maintained their suppressive capacity, which was also significantly reversed by the iNOS inhibitor L-NMMA (Figure 5B).

DISCUSSION

In this study, we have identified MDSCs in the SF of arthritic joints of mice with PGIA. MDSCs have been described in cancer patients and tumor-bearing animals (15,16,35), and in mice with EAE (34). However, the presence of MDSCs has not been reported in patients with RA or in animal models of RA.

We show that the Gr-1⁺CD11b⁺ population that dominates the SF of the arthritic joints of mice with PGIA, upon in vitro co-culture with DCs, significantly reduces the expression levels of MHC II and the co-stimulatory molecule CD86, both of which play pivotal roles in Ag presentation by DCs (11,36). Importantly, SF cells also exhibit a profound suppressive effect on the DC- and Ag-dependent proliferation of PG-specific T cells, and this effect is maintained after depletion of the monocytic subset from the SF. Spleen Gr-1⁺CD11b⁺ cells from the same arthritic mice contain a mixture of monocyte- and granulocyte-like cells with a larger Ly6-C^hi monocytic subset than SF cells. However, spleen myeloid cells do not show the same suppressive capability as SF cells towards DCs or T cells. These findings have led us to the conclusion that MDSCs with neutrophil-like phenotype exist within the Gr-1⁺CD11b⁺ population of SF cells in mice with PGIA.

The lack of suppression by SF Gr-1⁺CD11b⁺ cells on CD3/CD28-stimulated T-cell proliferation suggests that SF MDSCs have no direct effect on T cells when the T cells are activated through CD3/CD28 engagement, and that SF MDSC-mediated down-regulation of MHC II and CD86 on DCs is, at least in part, responsible for the suppression of Ag-specific T-cell proliferation. Although DCs exposed to SF cells still express MHC II and CD86, the DC population expressing these molecules at the highest levels practically disappears in the presence of SF MDSCs. The stability of the DC-T cell interaction and the subsequent robustness of activation through the TCR are thought to be strongly dependent on the density of MHC on DCs (36,37). Therefore, DCs with the highest density of MHC II are likely the driving force behind the robust proliferation of PG-TCR-Tg T cells, and this robustness is lost due to the loss of a MHC II^hi population. This notion is supported by our observation that the poor T-cell proliferation in the presence of SF-suppressed BM-DCs (which have lost a MHC II^hi population) or SF-derived DCs (which lack a MHC II^hi population) can be reproduced with a mAb that blocks MHC II function.

MDSCs exert their suppressive effects through depletion of L-arginine (via arginase 1 expression), production of NO (via iNOS), and generation of ROS (18,20,33,38,39). In our case, all these MDSC characteristics were found predominantly in SF cells, although spleen myeloid cells from arthritic mice also showed evidence of arginase, iNOS, and ROS activity. Inhibitors of iNOS or ROS (but not arginase) almost completely reversed the negative effect of SF cells on DC maturation, and the same inhibitors also significantly antagonized the suppressive activity of the SF cells towards DC-dependent T-cell proliferation. We found no evidence of a role for either suppressive cytokines (such as interleukin-10 or transforming growth factor β) or regulatory T cells in the SF cell-mediated suppression of T-cell proliferation (unpublished data).

The monocyte-like subset of MDSCs has been reported to suppress T-cell proliferation through iNOS activity in tumor models (17) and in EAE (34). In PGIA, we have also found iNOS to be the principal mediator of suppression, but by granulocyte-like rather than monocyte-like SF cells. Although iNOS expression and NO production have been detected in mouse (40), rat (41), and human (42) granulocytes under various conditions, this is the first report implicating iNOS as a mediator of the suppressive activity of granulocytic MDSCs in the SF of arthritic joints.

Since granulocytic MDSCs and mature neutrophils share similar phenotypic traits, they are likely recruited in the arthritic joints in response to the same chemoattractants (8), although the lack of cell surface markers strictly specific for granulocyte-like MDSCs (15) does not allow for distinction between these two cell types. After entering the inflamed joints, granulocytic MDSC precursors may acquire a maturation-resistant phenotype and suppressor function. Similarly, monocytic cells (which appear to localize to the lining layers of synovial tissue in PGIA) may differentiate into MDSCs under the conditions of inflammation. These are distinct possibilities, as rheumatoid joint effusions contain bioactive GM-CSF (43) that supports MDSC function (44,45), and GM-CSF is also detectable in the SF of arthritic joints in PGIA (our unpublished results). Such a supportive microenvironment is not likely to be present in the spleen, which may explain, at least in part, the lack of suppressor activity in spleen myeloid cells in PGIA.

Although the ability of SF cells to suppress DC- and Ag-dependent T cell proliferation is clearly indicated by our in vitro results, the in vivo impact of these cells and the strategies they employ, remain to be elucidated. SF MDSCs might have a negative effect on joint-resident DCs prior to the migration of these DCs to the JDLNs. DCs present in the SF in PGIA resemble MDSC-suppressed DCs in that they lack a population expressing MHC II at high levels. Such DCs might represent a tolerogenic population that renders T cells unresponsive to autoAgs (46,47) within the joint or in the JDLN. Alternatively, SF MDSCs themselves may migrate to the JDLNs. There is some precedent for neutrophils migrating from sites of inflammation to the LNs and interfering with DC-T-cell interactions (48). In PGIA, mice that experience full-fledged inflammation in a single joint after two intraperitoneal PG injections are quite resistant to the development of arthritis in other joints after a third PG immunization (our unpublished observation), suggesting that MDSCs present in the SF may play a role in protecting other joints from inflammation. Studies are currently under way in our laboratory to determine whether intravenous transfer of arthritic SF MDSCs to recipient mice with early arthritis can inhibit disease progression through suppression of autoimmunity in the recipients.

MDSCs with granulocytic phenotype likely exist in the inflamed joints of RA patients as well. Deciphering the differences between destructive neutrophils and granulocyte-like MDSCs in the SF could result in a better definition of MDSCs, potentially leading to therapies that exploit the potent suppressor activity of these cells to prevent the expansion of autoreactive T cells in RA.