Work by our group and others has demonstrated a role for the extracellular matrix receptor CD44 and its ligand hyaluronan in CD4+CD25+ regulatory T cell (Treg) function. Herein, we explore the mechanistic basis for this observation. Using mouse FoxP3/GFP+ Treg, we find that CD44 costimulation promotes expression of FoxP3, in part through production of IL-2. This promotion of IL-2 production was resistant to cyclosporin A treatment, suggesting that CD44 costimulation may promote IL-2 production through bypassing FoxP3-mediated suppression of NFAT. CD44 costimulation increased production of IL-10 in a partially IL-2-dependent manner and also promoted cell surface TGF-β expression. Consistent with these findings, Treg from CD44 knockout mice demonstrated impaired regulatory function ex vivo and depressed production of IL-10 and cell surface TGF-β. These data reveal a novel role for CD44 cross-linking in the production of regulatory cytokines. Similar salutary effects on FoxP3 expression were observed upon costimulation with hyaluronan, the primary natural ligand for CD44. This effect is dependent upon CD44 cross-linking; while both high-molecular-weight hyaluronan (HA) and plate-bound anti-CD44 Ab promoted FoxP3 expression, neither low-molecular weight HA nor soluble anti-CD44 Ab did so. The implication is that intact high-molecular weight HA can cross-link CD44 only in those settings where it predominates over fragmentary LMW-HA, namely, in uninflamed tissue. We propose that intact but not fragmented extracellular is capable of cross-linking CD44 and thereby maintains immunologic tolerance in uninjured or healing tissue.
The CD4+CD25+ regulatory T cells (Treg)3 are a specialized subpopulation of T cells which suppress a range of effector cell types and thereby contribute to the maintenance of immune homeostasis (1, 2). Studies have demonstrated an increased frequency or severity of autoimmunity in the absence of Treg and that transfer of Treg is sufficient to protect from or reverse autoimmunity (3, 4, 5). Treg-mediated suppression is cell contact dependent but the immunosuppressive cytokines TGF-ß and IL-10 also play a role (6, 7, 8). Treg make negligible amounts of IL-2 but nonetheless require this cytokine to maintain their viability and suppressive function (1).
The development and function of Treg requires the transcription factor FoxP3. Spontaneous mutation of FoxP3 leads to widespread lymphocytosis and autoimmunity in the scurfy mouse and in humans with immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome (9, 10). It has been proposed that FoxP3 may function as a transcriptional repressor, potentially through the formation of both DNA-protein and protein-protein interactions with NFAT and also NF-κB (11, 12).
The relative expression of CD44 and particular CD44v isoforms was reported by ourselves and others to be associated with FoxP3 expression and Treg function (13, 14). CD44 is a cell surface molecule with important roles in activation, migration, and apoptosis (15, 16). The diverse roles of CD44 are thought to reflect different CD44v isoforms (17) as well as the range of other cell surface receptors with which CD44 complexes, including Fas (18) and the CD3 complex (19). As with other cell surface molecules reported to characterize Treg, CD44 expression is heightened on a range of activated cell types other than Treg.
A key ligand of CD44 is hyaluronan (HA), a repeating disaccharide of N-acetylglucosamine and d-glucuronic acid and a prominent component of inflamed tissues (20). The relative amount of HA as well as the size of HA molecules are highly contextual and of physiologic importance (21, 22). Low-molecular-weight forms of HA (LMW-HA) (<15 saccharides; <3 kDa molecular mass) predominate during injury and inflammation. LMW-HA breakdown products are generated from intact high-molecular-weight HA (HMW-HA) (>2000 saccharides; >400 kDa molecular mass) by endogenous catabolism, by bacterial hyaluronidases, and by mechanical forces and oxidative stress (23). LMW-HA promotes angiogenesis (24), inflammation (25, 26), maturation of APCs (27, 28), and cell migration (29). Consistent with this role LMW-HA has been shown to be a ligand of TLR4 (28). HMW-HA conversely predominates in steady-state conditions and in healing tissues (20). HMW-HA serves a variety of structural functions in joints (30, 31) and tissue repair (30, 31) and typically has been reported to be either inert or anti-inflammatory (32, 33, 34, 35). HA-binding competent CD44v isoforms are expressed on T cells only after activation via the TCR or with proinflammatory cytokines including TNF-α and IFN-γ (36, 37). Therefore, the ability of these cells to interact with the HA is intrinsically related to their activation state.
In our previous work, we asked whether CD44 and ligands such as HA exert direct immunomodulatory effects on Treg. We observed that intact HMW-HA promotes Treg-mediated suppression while LMW-HA does not. We demonstrated that CD44 expression is correlated with the expression of FoxP3 as well as with HA-binding efficiency (14).
Herein, we examine the mechanistic basis of these observations. We postulated that the extracellular matrix (ECM) displays tissue integrity signals through CD44 which are seen by Treg as the antithesis of danger signals. We theorized that these signals might be most relevant in tissues with minimal proinflammatory cues. Therefore, using GFP-FoxP3 knock-in mice, we asked whether this costimulatory signal is relevant to Treg maintenance and requirement for IL-2. We asked whether this costimulatory signal is relevant to Treg function and to production of known regulatory cytokines IL-10 and TGF-β. Finally, using human Treg, we explored the hypothesis that HMW-HA but not LMW-HA provides a costimulatory signal via cross-linking of CD44.
Materials and Methods
Human blood samples
Human peripheral blood samples were obtained from healthy volunteers with informed consent, participating in a research protocol approved by the institutional review board of the Benaroya Research Institute at Virginia Mason.
FoxP3-GFP C57BL/6 mice were a gift from Dr. A. Rudensky (University of Washington, Seattle, WA). CD44-deficient C57BL/6 (CD44−/−), IL-10-deficient C57BL/6 (IL-10−/−), and wild-type (WT) C57BL/6 mice were purchased from The Jackson Laboratory. FoxP3-GFP and CD44−/− mouse lines were intercrossed to generate FoxP3-GFP CD44−/− at our institution. All mice were maintained in a specific pathogen-free American Association for the Accreditation of Laboratory Animal Care-accredited animal facility at the Benaroya Research Institute and handled in accordance with institutional guidelines.
HA with a molecular mass of 1.5 × 106 kDa (HMW-HA) was provided by Genzyme. LMW-HA was prepared by digestion of HMW-HA with Streptomyces-derived hyaluronidase, followed by filtration through a Centricon microconcentrator (Amicon to produce fragments <3 kDa, as described previously (38). Cyclosporin A was obtained from Sigma-Aldrich.
Human T cell activation studies used the following Abs: CD3 (OKT3) and CD28 (CD28.2) from eBioscience, CD44 (G44-26) was from BD Biosciences, and CD44 blocking mAb (Bu75) from Ancell. rIL-2 was from Chiron and the anti-IL-2 Ab (MQ1–17H12) was from BD Biosciences. HMW-HA was conjugated to BSA using previously described methods (39) for plate-bound activation studies.
Mouse flow cytometry experiments used the following fluorochrome-labeled Abs: CD4 (RM4-5), CD25 (PC61.5), and CD44 (IM7) from BD Biosciences. FoxP3 (FJK.16a) Ab and staining reagents from eBioscience were used as per the manufacturer’s instructions. The same anti-human TGF-β1 polyclonal Ab was used for mouse staining as for the human samples.
Isolation of leukocyte populations
Human PBMC were prepared by centrifugation of peripheral blood over Ficoll-Hypaque gradients. CD4+ T cells were isolated using a Dynal CD4 Positive Isolation Kit (Invitrogen) as per the manufacturer’s instructions, then sorted for CD25 (and with CD127 where noted) on a FACSVantage Flow Cytometer Cell Sorter (BD Biosciences). The top 2.5–5.0% of CD25+ cells were used for Treg studies. Purity of the resulting cell fractions was reliably >98% CD4+CD25+. Cells were cultured in RPMI 1640 (Invitrogen) supplemented with 10% pooled human serum, 100 μg/ml penicillin, 100 U/ml streptomycin, and 1 mM sodium pyruvate (Invitrogen).
Mouse leukocyte populations were isolated from inguinal, axial, and brachial lymph nodes and spleen cells from 6- to 8-wk-old mice. CD4+CD25+ and CD4+CD25i T cell populations were isolated using a CD4+ T Regulatory Cell Isolation Kit (Miltenyi Biotec) per the manufacturer’s instructions. CD4+ FoxP3/GFP+ and CD4+ FoxP3/GFP− T cells were isolated by preselection with a Dynal CD4+ T cell Negative Isolation Kit (Invitrogen) and then sorted into both FoxP3/GFP+ and FoxP3/GFP− fractions using a FACSVantage Flow Cytometer Cell Sorter. Purity of the resulting cell fractions was reliably >99.9% FoxP3/GFP+. Cells were cultured in DMEM-10 (Invitrogen) supplemented with 10% FBS (HyClone), 100 μg/ml penicillin, 100 U/ml streptomycin, 50 μM 2-ME, 2 mM glutamine, and 1 mM sodium pyruvate (Invitrogen).
T cell activation and phenotyping assays
Human T cell experiments were performed as follows: 96-well flat-bottom tissue culture plates were precoated with CD3 Ab (0.5 μg/ml) and CD44 Ab (1 μg/ml) where relevant. Anti-CD28 Ab was not used in the human Treg experiments unless otherwise noted. Coating with HMW-HA was treated as a second step. Plates were washed with PBS and then either 100 μl of 100 μg/ml HMW-HA or 100 μl of 10% BSA in PBS was added and the plates were incubated at 37°C for 2 h. Plates were again washed with PBS before the addition of 150,000 T cells. No exogenous IL-2 was added unless otherwise noted. T cells cultures were analyzed by flow cytometry after 4 days. Other cell culture reagents were added at the initiation of the cultures as follows: human IL-2 Ab (10 μg/ml), soluble CD44 Ab (5 μg/ml) clone BU75 (blocks HA binding), soluble LMW-HA (20 μg/ml), recombinant human IL-10 (5 ng/ml), and IL-10 Ab (5 μg/ml). Where noted, cyclosporin A was added at 50 ng/ml. Where noted, rapamycin was added at 100 μM/ml.
Low-dose IL-2 experiments used freshly isolated human CD4+CD25+CD127− T cells activated with anti-CD3/28 beads (1 bead/10 cells; Invitrogen). Exogenous IL-2 was added at the concentrations indicated only at the initiation of the culture. After 36 h, the beads were removed with a magnet and the cells were transferred to wells with HMW-HA (20 μg/ml) or LMW-HA (20 μg/ml). After an additional 24 h, cells were analyzed by flow cytometry.
Mouse T cell activation experiments were performed as follows: 96-well flat-bottom tissue culture plates were precoated with CD3 (0.5 μg/ml) and CD44 (1 μg/ml) Abs. CD28 (0.2 μg/ml) Ab was added with 100,000 T cells/well in DMEM-10 complete medium and cultured for 3 days. The CD278 (ICOS) control costimulation was performed by precoating the well with CD278 Ab (1 μg/ml) in combination with anti-CD3. Other cell culture reagents were added at the initiation of the cultures as follows: mouse IL-10 (5 ng/ml), mouse IL-2 (100U/ml), human TGF-β (3 ng/ml), IL-10 Ab (5 μg/ml), TGF-β Ab (5 μg/ml), and soluble mouse IL-2Rα (5 μg/ml).
FACS samples were stained in medium on ice for 45 min, washed once, resuspended in FACS stain buffer (PBS containing 1% FBS and 0.1% sodium azide), and run on a FACSCaliber flow cytometer (BD Biosciences). Analysis was performed using CellQuest (BD Biosciences) and FlowJo (Tree Star) software.
Both mouse and human cell culture supernatants were analyzed for cytokines using Meso Scale Discovery MSD 96-well MultiArray Cytokine TH1/TH2 assays for humans and mice and read on a SECTOR Imager 2400 as per the manufacturer’s instructions.
Human suppression assays were performed as follows: CD4+CD25+ Treg were titrated into a combination of CFSE (0.8 μM; Molecular Probes)-labeled CD4+CD25− responder T cells (200,000 cells/well) and irradiated CD4− cells (600,000 cells/well) in 5 ml of FACS tubes (Falcon; BD Biosciences), then stimulated with soluble CD3 (5 μg/ml) and CD28 (2.5 μg/ml) Abs. After 4 days, T cells were analyzed by flow cytometry.
Mouse Treg suppression assays were performed as follows: purified CD4+CD25+ or CD4+ FoxP3/GFP+ T cells were titrated into a combination of CD4+CD25− or CD4+ FoxP3/GFP− T cells (4 × 104 cells/well) and irradiated (2000 rad) splenocytes (2 × 105/well) were depleted of CD4+ T cells, then stimulated for 72 h with 1 μg/ml Con A in 96-well round-bottom plates. During the final 16 h, cells were pulsed with 1μCi/well of [3H]thymidine, then harvested onto filter mats and counted on a 1450 Wallac microbeta scintillation plate counter (PerkinElmer).
Total RNA was harvested from freshly isolated and cultured mouse CD4+ FoxP3/GFP+ and CD4+ FoxP3/GFP− T cells using a RNeasy mini kit from Qiagen. cDNA was prepared from 350 ng of total RNA reverse transcribed in a 40-μl reaction mix with random primers using a High-Capacity cDNA Archive Kit according to the manufacturer’s instructions. Relative quantitation of TGF-β1 gene expression was performed using TaqMan Gene Expression Assay Mm03024053_m1 and eukaryotic 18S rRNA Endogenous Control part no.4333760. Briefly, 1.2 μl of cDNA was amplified in 1× TaqMan Fast Universal PCR Mix with 250 nM TaqMan probe in a 20-μl reaction using the Fast program for 50 cycles on an Applied Biosystems I7900HT thermocycler. (All quantitative PCR reagents were from Applied Biosystems). All samples were done in duplicate and data were analyzed using the comparative cycle threshold method with software from Applied Biosystems. Estimated copy numbers were generated from a standard curve created by using a selected reference cDNA template and TaqMan probe (40).
Statistical comparison of samples stimulated with either anti-CD3/28 alone vs in conjunction with anti-CD44 costimulation was made using the Wilcoxon-signed rank test. Values of p < 0.05 were considered significant.
CD44 costimulation promotes expression of FoxP3 and CD25 by GFP/FoxP3+ mouse Treg
To evaluate the effects of CD44 costimulation on Treg, we first isolated FoxP3/GFP+ Treg from CD4+ T cells (Fig. 1⇓A). These were then activated with plate-bound anti-CD3 Ab, soluble anti-CD28 Ab with or without plate-bound anti-CD44 Ab without exogenous IL-2 for 3 days before staining. We found that anti-CD44 costimulation promoted expression of FoxP3/GFP and CD25 (Fig. 1⇓B). Both anti-CD3/28- and anti-CD3/28/44-treated GFP/FoxP3+ Treg expressed GFP/Foxp3 to a greater degree than GFP/FoxP3− controls (supplemental Fig. 14). We found that anti-CD44 Ab alone did not promote GFP/FoxP3 expression; earlier or concomitant stimulation with anti-CD3 Ab was also required (Fig. 1⇓C). Notably, only plate-bound anti-CD44 had this effect while soluble anti-CD44 did not (data not shown). Although exogenous IL-2, added at 100 U/ml, had a similar effect, anti-ICOS-1 Ab treatment conversely did not (Fig. 1⇓, D and E). ICOS-1 was chosen because, similar to CD44, it is a costimulatory molecule with important roles in T cell activation and IL-2 production (41). It was possible to negate this effect with soluble rIL-2Rα protein. This reagent was used because it has the advantage of making IL-2 unavailable for binding to cell surface IL-2Rα without a propensity to form bioactive IL-2 dimers or signal through CD25. These data implicate a role for IL-2 in CD44 cross-linking effects on FoxP3 expression. Costimulation of mouse FoxP3/GFP+ Treg with anti-CD44 Ab led to enhanced Treg suppressive function (Fig. 1⇓F). HMW-HA, a natural ligand of CD44, also promoted expression of GFP/FoxP3 (Fig. 1⇓G). Interestingly, the capacity of HMW-HA to promote GFP/FoxP3 required supplementation with 5% mouse serum. Without this there was no effect (data not shown).
CD44 cross-linking promotes production of IL-2 by Treg in a manner which bypasses cyclosporin A treatment
Given the well-established role of IL-2 in Treg function and persistence, these data raised the obvious question of whether CD44 costimulation was promoting the availability of IL-2 or reducing the requirement of GFP/FoxP3+ Treg for this cytokine. Although small in magnitude, FoxP3/GFP+ Treg costimulated with anti-CD44 Ab demonstrated a modest but reproducible increase in IL-2 production over such cells stimulated with anti-CD3/28 alone (Fig. 2⇓A). Costimulation with anti-ICOS-A Ab did not promote IL-2 production (Fig. 2⇓B). Consistent with these data, mRNA harvested from FoxP3/GFP Treg pooled from five mice contained increased IL-2 message in CD3/28/44-activated cells relative to CD3/28-activated controls (Fig. 2⇓C). IL-2 mRNA was not detectable preactivation (data not shown). GFP/FoxP3+ Treg were used to exclude any contribution by FoxP3− cells to the effects described here. These data support the conclusion that CD44 cross-linking promoted modest IL-2 production by cells isolated on the basis of FoxP3/GFP expression.
Given that FoxP3 is thought to inhibit IL-2 production via inhibition of NFAT, it seemed reasonable to ask whether CD44 treatment promoted Il-2 production in a NFAT-independent manner. To ascertain this, we added cyclosporin A to the cultures on the grounds that this molecule inhibits IL-2 production via effects on calcineurin and hence NFAT activation. We found that the addition of cyclosporin A depressed GFP/FoxP3 mean fluorescence intensity (MFI) on Treg activated with anti-CD3/28, but had minimal effects on GFP/FoxP3 MFI on Treg activated with anti-CD3/28/44 (Fig. 3⇓A). Similar effects were observed vis-à-vis Il-2 production (Fig. 3⇓B). These data indicate that anti-CD44 treatment bypasses cyclosporin A-mediated inhibition of IL-2.
CD44 cross-linking and exogenous IL-10 promote cell surface TGF-β
The modest amounts of IL-2 produced and the fact that soluble IL-2R in excess did not completely abrogate the effects of CD44 costimulation (Fig. 1⇑D) prompted us to consider other mechanisms known to have salutary effects on Treg. One possibility was concomitant effects on TGF-β, a cytokine known to be important for Treg number and function (42, 43, 44). Costimulation through CD44 promoted a substantial increase in the amount of surface-bound TGF-β on the FoxP3/GFP Treg (Fig. 4⇓A). Interestingly, the addition of exogenous IL-10 also strongly promoted cell surface TGF-β. This was the case for normal (Fig. 4⇓B) as well as CD44−/− Treg (Fig. 4⇓C). The effect of CD44 cross-linking was not decreased by anti-IL-10 Ab, suggesting that both CD44 costimulation and IL-10 may independently promote cell surface TGF-β. CD44 cross-linking neither promoted TGF-β1 mRNA nor did it significantly alter the amount of soluble mature form of TGF-β present (data not shown). Together, these data suggest that the increase in TGF-β may be particular to the cell surface and may involve the conversion of latent TGF-β to an active form.
It should be noted that even without CD44 costimulation a small fraction of Treg do express activated TGF-β on their cell surface. The results here are generally consistent with previously published reports in terms of the percentage of Treg which are cell surface TGF-β+ (45, 46).
CD44 cross-linking and exogenous IL-2 promote production of IL-10
CD44 cross-linking strongly promoted IL-10 production (Fig. 5⇓A; p < 0.05). This effect appears to be fairly specific and furthermore does not support a departure from the phenotype of differentiated natural Treg as negligible amounts of other TH1 or TH2 cytokines were detected, including IFN-γ, IL-4, and IL-5, and minimal TNF-α or IL-1β was seen (data not shown). Treg from CD44−/− mice produced diminished levels of IL-10 upon activation with anti-CD3/28 (Fig. 5⇓A; p < 0.05). Consistent with these data, IL-10 mRNA from FoxP3/GFP+ Treg harvested 24 h after activation likewise demonstrated enhanced IL-10 production upon CD44 cross-linking (Fig. 5⇓B). Our data suggests that IL-10 production may occur in a partially IL-2-dependent manner since addition of exogenous IL-2 at the inception of the culture alone and in an additive manner with CD44 costimulation increased IL-10 production (Fig. 5⇓C). As with our earlier data on IL-2 production, the addition of cyclosporin A did not diminish IL-10 production in the setting of CD44 costimulation (Fig. 5⇓D).
Our data also suggest that there is an IL-2-independent component of CD44-mediated promotion of IL-10 production. The addition of soluble TGF-β did not promote IL-10 production (Fig. 5⇑C). Another candidate was mammalian target of rapamycin signaling, which has been reported to up-regulate IL-10 production via effects on the transcription factor STAT3 (47, 48). To evaluate the role of mammalian target of rapamycin in this system, we tested the effects of rapamycin addition. We found that CD44 up-regulation of IL-10 production also bypasses rapamycin treatment (supplemental Fig. 2).
To evaluate the functional role of IL-10 in anti-CD44 costimulation effects, we isolated CD4+CD25+ Treg from both IL-10−/− and WT mice and stimulated these with anti-CD3/28/44 (supplemental Fig. 3). The difference in Treg suppression observed can be attributed to differences in IL-10 production.
CD44−/− mice have functionally impaired Treg
We theorized that because HMW-HA cross-linking of CD44 on Treg promoted FoxP3 persistence, CD44−/− mice might have diminished numbers of FoxP3+ cells. This was not the case. CD44−/− mice did not have significantly diminished numbers of CD4+FoxP3+ cells in their lymph nodes and spleens in vivo (supplemental Fig. 4). However, equivalent numbers of Treg from CD44−/−animals demonstrated impaired regulatory function in vitro compared with their normal counterparts (Fig. 6⇓). These data suggested that the absence of CD44 expression adversely impacted the ability of Treg to suppress effector T cell proliferation.
Costimulation with either HMW-HA or CD44 Ab promotes persistent FoxP3 expression in Treg
We sought to ascertain whether the effects of CD44 costimulation observed in mice upon stimulation with Abs were also relevant to human Treg and natural ligands of CD44. As with our mouse activation protocol, we activated human CD4+CD25+ cells with plate-bound anti-CD3 Ab with or without] HMW-HA and controls without the addition of exogenous IL-2. After 4 days, cells were stained for CD25 and FoxP3 expression. From the time of isolation (Fig. 7⇓A) to 96 h afterward (Fig. 7⇓B) without the addition of exogenous IL-2, the level of FoxP3 and CD25 expression falls. However, FoxP3 expression can be rescued by costimulation with HMW-HA, anti-CD44 Ab, or anti-CD28 Ab. This was the case for Treg isolated from multiple individuals (Fig. 7⇓C). Interestingly, we found that with human Treg the incorporation of anti-CD28 does not seem to markedly impact FoxP3 expression in these assays (data not shown), whereas this Ab was essential in the experiments with mouse Treg. Whether this is due to a real difference in human/mouse biology or alternatively whether this reflects the greater relative purity of mouse GFP/FoxP3+ Treg is unclear.
The length of HA is of critical importance. HMW-HA, but not LMW-HA, promoted the FoxP3 expression (Fig. 8⇓A). LMW-HA was derived from the same HMW-HA used in these experiments and was therefore identical to HMW-HA except in terms of length. The maintenance of FoxP3 expression by HMW-HA costimulation can be inhibited by an anti-CD44 Ab known to inhibit HA binding, suggesting that HMW-HA is acting through this receptor. Consistent with a putative requirement for CD44 cross-linking, plate-bound but not soluble anti-CD44 Ab promoted FoxP3 expression (Fig. 8⇓B). We similarly observed in the mouse activation assays that plate-bound but not soluble anti-CD44 Ab promoted FoxP3 expression. Commensurate with this increase in FoxP3 expression and consistent with the analogous mouse Treg data, human Treg that received plate-bound anti-CD3/44 costimulation demonstrated increased suppression of effector T cell proliferation relative to that seen with Treg stimulated with anti-CD3 alone (Fig. 8⇓C).
HMW-HA treatment and CD44 cross-linking promote FoxP3 and CD25 expression in an IL-2-dependent manner
Both HMW-HA and anti-CD44 Ab had analogous effects on human Treg as those observed with mouse Treg vis-à-vis expression of both IL-10 (Fig. 9⇓A) and IL-2. (Fig. 9⇓B). Notably, minimal cell surface TGF-β production was observed on human Treg under these same conditions (data not shown). To assess the functional relevance of IL-2 production in this setting, we activated Treg with 0, 2, or 20 IU/ml IL-2 using anti-CD3/28-coated beads for 48 h. Cells were subsequently treated with HMW-HA or an equivalent volume of PBS for an additional 24 h before assessment of FoxP3 expression. Interestingly, differences in FoxP3 expression upon HMW treatment were seen in cells that received 0 or 2 but not 20 IU/ml IL-2 (Fig. 9⇓C). At the higher dose, IL-2 itself was sufficient for supporting Treg activity. HMW-HA effects were therefore most pronounced at low concentrations of IL-2.
To evaluate the effect of HMW-HA on Treg viability, we stained human Treg for annexin V/propidium iodide following treatment with HMW-HA treatment vs LMW-HA or PBS as controls. We did indeed observe enhanced viability of human Treg (supplemental Fig. 5).
The inflammatory milieu plays a major role in the generation and regulation of adaptive immune responses through a variety of mechanisms, yet the contribution of the ECM toward regulation of adaptive immunity is poorly understood. In earlier work, we demonstrated that the size and amount of HA modulate Treg function. In this study, we elucidate the salient mechanisms, focusing on the primary HA receptor CD44. We draw four conclusions from these data.
First, we conclude that CD44 signaling delivers a stimulatory signal to Treg in the context of TCR activation. In this setting, CD44 costimulation potently up-regulated CD25 expression and maintained expression of FoxP3. The requirement for a TCR signal suggests that Treg may interact with the ECM in inflamed and healing tissues in a manner which is relevant to particular Ags. CD44 costimulation has previously been suggested to promote T cell activation in the presence of a subthreshold stimulus (49); one way this might occur via the apposition of lck into the proximity of the TCR-CD3 complex (50).
Our data suggest that this observation is relevant to both mouse as well as human Treg. Furthermore, HMW-HA, a natural ligand of CD44, stimulates human and mouse Treg in an analogous, albeit less potent, manner to anti-CD44 Ab. Interestingly, the capacity of HMW-HA to promote GFP/FoxP3 in mouse Treg required the presence of mouse serum, as without this there was no effect. The identity of the permissive factor in mouse serum is as yet unclear. There are several candidate soluble factors which promote HA binding. These include inter-α-trypsin inhibitor and TSG6 (51, 52). Whatever the salient molecule is, it is evidently not present in commercially available FCS. This may be either because it is species specific or due to some aspect of processing.
Second, these data suggest that receptor cross-linking is necessary for these effects. Although both plate-bound anti-CD44 Ab and HMW-HA promoted FoxP3 persistence, neither soluble anti-CD44 Ab nor LMW-HA did so. This is consistent with previous reports of cross-linking being integral to several functions attributed to CD44 (53, 54, 55). We propose that the size of HA is directly related to the capacity of this molecule to cross-link spatially separated CD44 on the cell surface. This may occur via competitive exclusion, in which case the relative molar ratios of HMW-HA vs LMW-HA may determine the extent of CD44 cross-linking in a given environment. Given the relationship between HA size and the stage of inflammation, the extent of CD44 cross-linking by HMW-HA and other ligands may provide contextual cues to Treg and T cells regarding the inflammatory milieu.
Third, our data suggest CD44 cross-linking promotes modest IL-2 production. This was the case for cells previously isolated on the basis of FoxP3/GFP expression as well as for human Treg. These results are consistent with reports of enhanced IL-2 production by other T cell subsets and NKT cells upon CD44 costimulation (56, 57) and with reports of requirements for HA and CD44 in certain IL-2-dependent processes (58, 59, 60, 61). Because Treg are not known to produce substantial quantities of IL-2, it seems likely that at least some FoxP3/GFP+ cells acquire the capacity to produce IL-2 upon activation in the setting of CD44 costimulation.
These data raise the possibility that CD44 costimulation bypasses the negative regulatory influence of FoxP3 on IL-2 production. How might this occur? FoxP3 is known to down-regulate IL-2 production through inhibitory effects on NFAT (12, 62). We demonstrate that CD44-mediated IL-2 production by Treg occurs irrespective of cyclosporin A treatment, a potent inhibitor of NFAT. This suggests that CD44 costimulation has the capacity to bypass the inhibition of NFAT; whether CD44 signaling indeed bypasses FoxP3-mediated inhibition of NFAT and by what alternative signaling pathways are the subject of current investigation. If so, this may help explain how this highly IL-2-dependent cell type (63, 64) persists in healing or uninjured tissues where IL-2 concentrations are presumably low. This question is important because many models of autoimmunity invoke the inappropriate persistence of immune responses in former sites of inflammation.
It should be noted that although our sorting methodology yields purities >99.9% GFP/FoxP3+ cells, we cannot absolutely exclude the possibility that IL-2 is produced by a small number of FoxP3/GFP− cells conceivably contaminating our sorting protocol. However, we did not observe any expansion of GFP-negative cells over the course of the assay, such as would occur were there a quantifiable contaminating population of these cells (data not shown).
Fourth and finally, our data support a previously unreported role for CD44 cross-linking in enhancing expression of the anti-inflammatory cytokines IL-10 and TGF-β. CD44 cross-linking promotes IL-10 production in a partially IL-2-dependant manner. Furthermore, we found that CD44−/− Treg produce diminished amounts of this cytokine (47, 48). IL-10 production is reported to be one mechanism by which Treg maintain immune tolerance (65) and expression of this cytokine by CD4+CD25+ cells has been shown to be partially dependent on Il-2 (66). Enhanced IL-10 production may also reduce the dependency of Treg on exogenous IL-2 by independently promoting Treg viability (67). Consistent with a role for CD44 in Treg function, we find that Treg taken from CD44−/− mice demonstrated impaired regulatory function ex vivo.
CD44 cross-linking also promotes production of TGF-β. This cytokine has well-characterized roles in promoting FoxP3 expression (68, 69) as well as Treg function (42, 43, 44). Increased TGF-β+ in a Treg population has been shown to enhance suppression of activated T cell proliferation through induction of FoxP3 expression in responder cells (68). Our data suggest that both CD44 cross-linking and exogenous IL-10 promote cell surface TGF-β. A role for IL-10 in TGF-β production has been previously reported (70, 71, 72). HA has previously been demonstrated to promote CD44 interactions with TGF-β receptor I, thereby enhancing TGF-β1 signaling (73). Others have reported that CD44−/− mice have diminished levels of TGF-β (44).
In sum, our data support the conclusion that cross-linking of CD44 by HMW-HA and potentially other ECM components promote Treg persistence and function. Our findings are most consistent with a two-step model, in which the initial activation of Treg through the TCR is sufficient for FoxP3 induction, followed by a requirement for a costimulatory signal to support FoxP3 persistence. This latter step is heavily dependent on environmental cues, such that in the absence of high concentrations of IL-2, the CD44-IL-2-IL-10 pathway we describe can contribute to the support of peripheral immune tolerance. This mechanism may be one way in which viable Treg populations persist in healing or uninjured tissues to bring inflammatory processes to a close and maintain peripheral tissue tolerance. We propose that the HMW-HA molecular composition of the ECM conveys a tissue integrity signal which functions as the opposite of a danger signal, in that adaptive Treg receive an important costimulatory signal in situ, promoting immune homeostasis.
We thank Nathan Standifer and John Gebe for helpful comments and suggestions.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by grants from the National Institutes of Health (DK46635, HL18645, and DK53004) and the Juvenile Diabetes Research Institute (The Center for Translational Research at Benaroya Research Institute. P.L.B. is supported by National Institutes of Health K-08 Grant DK080178-01 and a National Institutes of Health LRP grant.
↵2 Address correspondence and reprint requests to Dr. Paul L. Bollyky, Benaroya Research Institute, 1201 Ninth Avenue, Seattle, WA 98101. E-mail address:
↵3 Abbreviations used in this paper: Treg, CD4+CD25+ regulatory T cell; ECM, extracellular matrix; HA, hyaluronan; HMW-HA, high-molecular-weight HA; LMW-HA, low-molecular-weight HA; WT, wild type.
↵4 The online version of this article contains supplemental material.
- Received January 21, 2009.
- Accepted June 5, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.