In the K/BxN mouse model of inflammatory arthritis, T cells carrying a transgenic T cell receptor initiate disease by helping B cells to produce arthritogenic anti–glucose-6-phosphate isomerase (anti-GPI) autoantibodies. We found that lethally- irradiated lymphocyte-deficient C57BL/6 (B6).g7 (I-Ag7+) recombinase-activating gene–deficient (Rag−/−) mice reconstituted with K/BxN hematopoietic stem and progenitor cells exhibit arthritis by week 4. In contrast, healthy B6.g7 recipients of K/BxN hematopoietic stem and progenitor cells show only mild arthritis, with limited extent and duration. The objective of this study was to investigate the factors responsible for the attenuation of arthritis in B6.g7 recipients.
Antibody responses were measured by enzyme-linked immunosorbent assay. Fluorescence-activated cell sorting analyses were performed for testing chimerism, expression of markers of activation and suppression, tetramer binding, and intracellular cytokines in CD4+ T cells. Suppressive activity of CD4+ T cells was studied by adoptive transfer.
Titers of anti-GPI antibodies in reconstituted B6.g7 mice were ∼60-fold lower than in reconstituted B6.g7 Rag−/− mice. Examination of chimerism in the reconstituted B6.g7 mice showed that B cells and myeloid cells in these mice were donor derived, but CD4+ T cells were primarily host derived and enriched for cells expressing the conventional regulatory markers CD25 and FoxP3. Notably, CD4+CD25−FoxP3− T cells expressed markers of suppressive function (CD73 and folate receptor 4), and delayed disease after adoptive transfer. Activation of donor-derived CD4+ T cells was reduced, and thymic deletion of these cells appeared increased.
Despite myeloablation, host CD4+ T cells having a regulatory phenotype emerge in these mice and attenuate autoimmunity.
Autoimmune diseases occur when tolerance to self antigen fails and the immune system initiates attack against self tissues. Rheumatoid arthritis (RA) is an autoimmune disease in which T cells have been proposed to recognize autoantigen and participate in effector pathways (1, 2). Initial bone marrow transplant experiments in mice demonstrated that the ability to transfer autoimmune arthritis rests within the hematopoietic compartment (3). These findings led to the idea of using bone marrow transplantation as therapy for RA (4). Autologous hematopoietic stem cell transplantation (HSCT) has been tried as treatment for severe autoimmune diseases in humans (4). However, fatal infections due to insufficient recovery of T cells and relapses of autoimmunity likely due to the persistence of autoreactive clones have limited the use of this approach (5).
Studies in C57BL/6 (B6) mice demonstrated that following lethal irradiation and transplantation of bone marrow (BM), the myeloid leukocytes were almost completely donor derived, but significant numbers (25%) of CD4+ T cells were recipient derived (6). In another study, although host-derived cells were the major constituent (60–80%) of the Treg cell compartment 5 weeks following autologous BM transplantation (7), donor-derived Treg cells were detectable ∼2–3 weeks posttransplant and became the major source of Treg cells by 8 weeks posttransplant. The initial predominance of the host-derived Treg cells in these mice was due to their proliferative expansion during the first 5 weeks posttransplant. The presence of host-derived Treg cell–enriched CD4+ T cells in these studies has raised the hope of devising a cell-based strategy to inhibit relapse of autoimmunity in human HSCT. However, more detailed information on the emergence and function of these host-derived Treg cells is needed.
Here, we describe experiments with a novel HSCT-based model of autoimmune disease. We used stem cells from the spontaneous K/BxN model of autoimmune arthritis in which class II–restricted, transgenic T cell receptors (TCRs) drive disease. K/BxN mice are a cross of KRN mice with the NOD strain; KRN mice carry a transgenic TCR that recognizes a glucose-6-phosphate isomerase (GPI) peptide bound to the NOD class II molecule I-Ag7 of the major histocompatibility complex (MHC). (KRN × NOD)F1 mice show severe distal joint inflammation, with onset at age ∼4–5 weeks. The severe symmetric polyarthritis in these mice is dependent on expression of the KRN TCR (8) and T cell help for B cells that make pathogenic anti-GPI antibodies (9). The anti-GPI antibodies form immune complexes with GPI, triggering a joint-specific inflammatory response mediated by neutrophils, macrophages, natural killer cells, and Th17 cells (9, 10). In the K/BxN model, CD25+FoxP3+ Treg cells are selected in the thymus and enriched in the spleen (∼20%) and draining lymph nodes (LNs) during arthritis. The Treg cells mediate suppressive function in vitro but do not inhibit severe arthritis in the mice (11). However, Treg cell depletion in K/BxN mice results in multiorgan inflammation with persistent arthritis (12).
The crucial role of I-Ag7 in the K/BxN arthritis model was shown by breeding experiments: B6 or BALB/c mice carrying the I-Ag7 gene, when crossed into B6 mice carrying the KRN TCR, yield offspring that develop arthritis similar to that in K/BxN mice (13). We predicted that transfer of hematopoietic stem and progenitor cells from K/BxN mice into myeloablated B6.g7 mice (I-Ag7 expression on B6 background) would result in autoimmune arthritis; surprisingly, however, autoimmunity in recipient mice was highly attenuated. Here, we identify immunosuppressive roles for host-derived CD4+ T cells with regulatory phenotypes, including CD25−FoxP3−CD73+ cells.
MATERIALS AND METHODS
NOD mice and B6.g7 mice derived from B6 mice expressing the NOD class II MHC molecule I-Ag7 were obtained from The Jackson Laboratory. Recombinase-activating gene (Rag)–deficient B6.g7 (B6.g7 Rag−/−) mice lack T and B cells (Beilhack GF, Shizuru JA: unpublished observations). KRN mice with a transgenic TCR against GPI were kindly provided by Dr. Diane Mathis (Harvard Medical School, Boston, MA). The arthritic K/BxN mice were obtained by crossing KRN mice with NOD mice, as is standard. All animals were bred, housed, and cared for in the Stanford Veterinary Service Center under the approval of the Administrative Panel for Laboratory Animal Care, protocol no. 15867.
Hematopoietic stem and progenitor cell isolation.
Hematopoietic stem and progenitor cells were isolated as described, with some modifications (14). BM cells were harvested and selected for c-Kit+ cells using the MACS system (Miltenyi Biotec). The c-Kit+ BM cells were then sorted in a FACSAria cell sorter (BD Biosciences) for c-Kit+Sca-1highLin− cells after staining with antibodies (further information is available at http://med.stanford.edu/genetherapy/research/related_sites.html). All antibodies were obtained from eBioscience. Fluorescence-activated cell sorting (FACS) was performed at the Stanford Shared FACS Facility at Stanford University.
Recipient mice (B6.g7 or B6.g7 Rag−/−) ages 3–4 months were lethally irradiated with a Phillips Unit irradiator (250 kV, 15 mA) at 980 rads (in 2 doses with a 4-hour interval). Irradiated mice were reconstituted with hematopoietic stem and progenitor cells from 3–5-month-old donors (K/BxN mice) 3 hours after irradiation. Hematopoietic stem and progenitor cells (10,000–12,000) in 200 μl phosphate buffered saline (PBS) were given to each recipient by tail vein injection. The reconstituted mice were maintained on trimethoprim/sulfamethoxazole (suspension; Hi-Tech Pharmacal) in drinking water up to 3 months posttransplant. Control mice that received irradiation only (without transplantation) died within 2 weeks.
Antibody and tetramer staining and flow cytometry analyses.
Thymus, spleen, and popliteal LNs were harvested from mice. Surface staining was done with various antibodies directly conjugated to fluorochromes (further information is available at http://med.stanford.edu/genetherapy/research/related_sites.html). For biotin-labeled antibodies, Pacific Orange–conjugated streptavidin (Invitrogen) was added and incubated for an additional 30 minutes. For intracellular FoxP3 staining, cells were treated with fixation/permeabilization reagent (eBioscience) followed by staining with antibodies (eBioscience). For cytokine stimulation and intracellular staining, spleen cells were suspended in complete Iscove's modified Dulbecco's medium (10% fetal bovine serum, 1% L-glutamine, 0.1% β-mercaptoethanol) and incubated for 4 hours at 37°C with or without Leukocyte Activation Cocktail (BD Biosciences). Cell viability was assessed by staining with Live Dead Aqua (Invitrogen). Subsequently, cells were permeabilized and stained with fluorescein isothiocyanate–conjugated anti–interleukin-17 (anti–IL-17) antibody (TC11-18H10), phycoerythrin (PE)–conjugated anti–interferon-γ (anti-IFNγ) antibody (XMG1.2), and allophycocyanin-conjugated anti-FoxP3 antibody (FJK-16s).
For tetramer staining, the single-cell suspension was blocked with 0.5 mg/ml Fc block for 10 minutes on ice. Cells were then washed and stained with PE-labeled MHC/GPI peptide tetramer (kindly provided by Dr. Luc Teyton, The Scripps Research Institute, La Jolla, CA) at a final concentration of 10 μg/ml for 1 hour at room temperature (15). PE-labeled MHC/hen egg lysozyme peptide tetramer was used as negative control in all experiments. Costaining of surface markers was done as mentioned above. In all antibody staining experiments, data were collected on an LSRII flow cytometer (BD Biosciences) and analyzed with FlowJo software (Tree Star), gating on single cell populations. All FACS analyses were performed at the Stanford Shared FACS Facility at Stanford University.
Evaluation of autoimmune disease.
Mice were assessed by clinical arthritis index and circulating anti-GPI antibody titer. The clinical arthritis index was tallied from the degree of inflammation of each ankle, based on redness and swelling measured with a pocket thickness gauge (Mitutoyo). Scores were as follows: 0 = none; 0.25 = mild; 0.5 = moderate; 1 = severe, with 4 being the maximum score in mice with full-blown arthritis (16). The mean ± SEM score per animal group was calculated. GPI titer was determined by enzyme-linked immunosorbent assay (ELISA). Specifically, ELISA plates were coated with GPI protein (1–4 μg) and blocked with PBS containing 0.05% Tween 20 and 1% fat-free powdered milk followed by addition of diluted serum at 100 μl/well. After incubation at 37°C for 2 hours, the plates were washed and incubated with horseradish peroxidase–conjugated goat anti-mouse IgG antibody (1:5,000 dilution; Millipore) for 1 hour before adding tetramethylbenzidine substrate (BD OptEIA kit; BD Biosciences). Results were read out with a SpectraMax 190 Absorbance Microplate Reader (Molecular Devices) after stopping the reaction with 50 μl H2SO4 (1M stock). For joint histopathology, ankle joints were fixed in 4% phosphate buffered formaldehyde and then sectioned and stained with hematoxylin and eosin by Premier Laboratory. Microscopy was performed with an Axioplan 2 instrument (Carl Zeiss Microscopy).
Spleen cells and LN cells were harvested from K/BxN→B6.g7 mice 3 months after reconstitution of hematopoietic stem and progenitor cells. Cells were resuspended in Hanks' balanced salt solution buffer with 2% fetal calf serum and surface stained with antibodies against CD45.2, CD45.1, CD4, CD25, folate receptor 4 (FR4), and CD73 to identify host- and donor-derived regulatory cells. Host-derived CD4+CD25high cells and CD4+CD25−CD73+FR4+ cells were sorted using a FACSAria cell sorter and injected separately (5 × 105 cells per mouse) into B6.g7 Rag−/− recipients that had been reconstituted with K/BxN hematopoietic stem and progenitor cells 20 days earlier. Mice were then monitored for onset and progression of arthritis as above.
The groups were compared using Student's unpaired 2-tailed t-test. GraphPad Prism software, version 5 was used for statistical analysis.
B6.g7 mice reconstituted with hematopoietic stem and progenitor cells from K/BxN mice have mild, transient disease.
In an effort to create an HSCT model for autoimmune arthritis, lethally irradiated B6.g7 Rag−/− mice were reconstituted with hematopoietic stem and progenitor cells (c-Kit+Sca-1highLin−) isolated from K/BxN mice. The recipients were followed up for development of arthritis. All developed severe arthritis (Figure 1A) and had high-titer anti-GPI autoantibodies (Figure 1B) comparable to those in transgenic K/BxN mice. In contrast, lethally irradiated B6.g7 mice reconstituted with hematopoietic stem and progenitor cells from K/BxN mice developed very mild arthritis that disappeared by week 14 (Figure 1A) and had very low-titer, circulating anti-GPI autoantibodies (Figure 1B). Joint histopathology showed minimal inflammatory cell infiltrates in the K/BxN→B6.g7 mice, whereas the K/BxN→B6.g7 Rag−/− chimeras showed prominent cell infiltration and severe bone destruction (Figure 1C). Thus, transfer of K/BxN hematopoietic stem and progenitor cells into immunocompetent B6.g7 mice resulted in attenuated autoimmunity compared to transfer into immunodeficient B6.g7 Rag−/− mice.
Reduced donor-derived CD4+ T cells in reconstituted B6.g7 mice compared to reconstituted B6.g7 Rag−/− mice.
In the K/BxN mouse, CD4+ T cells bearing the KRN TCR and antigen-presenting cells expressing the class II MHC molecule I-Ag7 are required for the initiation and progression of autoimmunity. To determine if disease resistance in the B6.g7 mice was due to incomplete reconstitution of T or B cells, we analyzed splenocytes from B6.g7 recipients 12 weeks after transplant of hematopoietic stem and progenitor cells. Compared to K/BxN mice, the K/BxN→B6.g7 chimeras showed comparable ratios of CD4+:CD8+ T cells, and comparable proportions of B220+ B cells, with a lower proportion of CD11b+ cells (neutrophils), presumably reflecting the reduced inflammatory response in these mice (further information is available at http://med.stanford.edu/genetherapy/research/related_sites.html). However, compared to B6.g7 mice, the proportion of CD4+ T cells was low in both the chimera and K/BxN mice, likely due to negative selection of the self-reactive transgenic TCR (13).
To distinguish the host and donor hematopoietic cells in the chimera, we used the congenic markers CD45.1 and CD45.2 (further information is available at http://med.stanford.edu/genetherapy/research/related_sites.html). Approximately 98% of B cells and neutrophils were of donor origin (Figure 2A), but only ∼10% of splenic CD4+ T cells were of donor origin (Figure 2B). As pathogenic autoreactive CD4+ T cells undergo expansion in the draining LNs of the inflamed target organs in the K/BxN mice (13), we assessed chimerism in draining LNs. The joint-draining popliteal LNs of the K/BxN→B6.g7 chimera showed percentages of donor-derived CD4+ T cells comparable to those in spleen (Figure 2C). The absolute numbers of donor-derived splenic CD4+ T cells (Figure 2D) and CD4+CD25−FoxP3− T cells (Figure 2E) were also reduced in the B6.g7 recipients compared to the B6.g7 Rag−/− recipients. Thus, insufficient numbers of KRN TCR+, donor-derived CD4+ Teff cells is one possible explanation for attenuated autoimmunity in K/BxN→B6.g7 chimeras. However, the appearance of autoantibodies and joint infiltrates in the K/BxN→B6.g7 chimeras indicates that Teff cell numbers were sufficient to break tolerance.
K/BxN→B6.g7 chimeras have reduced proportions of transgenic TCR+ or I-Ag7/GPI tetramer+ donor-derived CD4+ T cells.
We next analyzed the thymi of the K/BxN→B6.g7 chimeras and found that the majority of CD4+ single-positive T cells were host derived and absolute numbers of donor-derived CD4+ single-positive T cells were reduced compared to thymocytes in the K/BxN→B6.g7 Rag−/− chimeras (Figures 3A and B). The β-chain of the KRN TCR in the K/BxN mice is Vβ6 (13). Selection of KRN TCR–expressing CD4+ single-positive T cells can be evaluated by the expression of Vβ6 compared to the expression of other variable regions from endogenous TCR loci, like Vβ8. In the thymus and spleen of the K/BxN→B6.g7 Rag−/− mice, we found that ∼50% of the CD4+ T cells were Vβ6+Vβ8− and ∼15% of T cells were Vβ6intermediateVβ8+ (Figures 3C–E), an expression pattern similar to that in K/BxN mice (13). K/BxN→B6.g7 mice showed substantially lower proportions of each subset, including among CD4+ single-positive thymocytes (Figures 3C–E). To enumerate circulating cells expressing the KRN TCR based on their recognition of I-Ag7 molecules with bound GPI282–294 peptide (17), we used I-Ag7/GPI282–294 peptide tetramers to stain CD4+ T cells in the blood obtained from K/BxN→B6.g7 and K/BxN→B6.g7 Rag−/− mice. The frequency of donor-derived, CD4+CD25−Vβ6+tetramer+ cells was significantly lower in the K/BxN→B6.g7 mice compared to the K/BxN→B6.g7 Rag−/− mice (mean ± SEM 2.53 ± 0.6% versus 12.26 ± 1.3%; P = 0.0006) (Figure 4A).
Donor-derived CD4+CD25−FoxP3− cells in K/BxN→B6.g7 mice are less activated than those in K/BxN→B6.g7 Rag−/− mice.
The peripheral CD4+ T cell compartment in K/BxN mice is activated, as evidenced by expression of the activation/memory marker CD44 (18). In the arthritic K/BxN→B6.g7 Rag−/− chimeras, similar to the K/BxN mice, a high proportion of the donor-derived CD4+CD25−FoxP3− cells were CD44high (mean ± SEM 62.5 ± 3.1%). In contrast, significantly fewer donor-derived CD4+CD25−FoxP3− cells in the K/BxN→B6.g7 chimeras were CD44high (mean ± SEM 28.7 ± 2.8%) (P < 0.0001) (Figure 4B). To further characterize the activation state of donor-derived CD4+ T cells, we measured their expression of the inflammatory cytokines IFNγ and IL-17 after stimulation with phorbol myristate acetate (Figures 4C and D). These cytokines are known to promote arthritis in K/BxN mice (10). A higher proportion of donor-derived splenic CD4+ T cells expressed IFNγ and IL-17 in K/BxN→B6.g7 Rag−/− mice than in K/BxN→B6.g7 mice (for IFNγ, 21.5 ± 6.8% versus 4.19 ± 1.5%; P < 0.0002) (for IL-17, 2.28 ± 0.17% versus 0.36 ± 0.37%; P < 0.013). Taken together, these data indicate reduced activation of potentially arthritogenic T cells in the K/BxN→B6.g7 chimeras.
Host-derived CD4+ T cells are enriched for CD25+FoxP3+ Treg cells.
The presence of host-derived CD4+ T cells in the spleen and thymus of K/BxN→B6.g7 chimeras indicates a possible role of these cells in attenuating autoimmunity. To investigate whether these cells had regulatory properties, we assessed levels of CD4+CD25+FoxP3+ Treg cells in the host and donor-derived CD4+ T cells in the K/BxN→B6.g7 reconstituted mice (Figure 5A). Host-derived splenic CD4+ T cells were 15.2 ± 2.6% CD25+FoxP3+ compared to 6.7 ± 0.85% CD25+FoxP3+ among donor-derived CD4+ T cells (P < 0.0059); the latter value is similar to the Treg cell portion of CD4+ T cells in normal B6.g7 mice (8 ± 0.5%) (Figure 5A). The proportion of CD25+FoxP3+ cells in the donor-derived CD4+ T cells is not characteristic of CD4+ T cells expressing the transgenic KRN TCR. On the contrary, in K/BxN mice, Treg cells constitute 30% of the CD4+ T cell population (11). Similarly, in K/BxN→B6.g7 Rag−/− chimeras, the donor-derived CD4+ T cells included 26.5 ± 1.8% CD25+FoxP3+ T cells (Figure 5A).
To test their immunosuppressive capacity, we transferred host-derived CD4+CD25high T cells from K/BxN→B6.g7 mice (3 months after transfer of hematopoietic stem and progenitor cells) into K/BxN→B6.g7 Rag−/− mice 20 days after transfer of K/BxN hematopoietic stem and progenitor cells. Because FoxP3 is an intracellular protein and cannot be used as a marker for sorting Treg cells, we used CD4+CD25high T cells as a surrogate for CD4+CD25+FoxP3+ T cells (19, 20). The K/BxN→B6.g7 Rag−/− mice were free of disease at the time of transfer. The mice were monitored for onset and progression of arthritis. We found that injected host-derived (CD45.2) CD4+CD25high T cells delayed onset of disease when compared to control K/BxN→B6.g7 Rag−/− mice that received PBS (Figure 5B). However, by day 48 after T cell transfer, arthritis scores in both groups were comparable. At this time point, 55 ± 14.3% of the remaining host-derived CD45.2 cells were CD4+CD25+ cells (not shown).
Host-derived CD4+CD25−FoxP3− cells show a memory-like phenotype and express the suppression markers CD73 and FR4.
As another potential source of suppressive host-derived CD4+ T cells, we assessed CD4+CD25−FoxP3− T cells for evidence of a regulatory phenotype. Radiation-resistant CD4+ T cells undergoing lymphopenia-induced proliferation in irradiated hosts acquire a memory-like phenotype, characterized by high expression of CD44 (21). These memory-like CD4+ T cells have suppressive effects on naive T cell proliferation (22). We found a higher percentage of host-derived CD4+CD25−FoxP3−CD44high splenic T cells in K/BxN→B6.g7 mice compared to the percentage of this subset in B6.g7 mice (Figure 5C).
FR4 and CD73 are associated with regulatory function (23, 24). Notably, the host-derived splenic CD4+CD25−FoxP3− cells in K/BxN→B6.g7 mice showed a significantly increased frequency of cells expressing both FR4 and CD73, compared to the donor-derived CD4+CD25−FoxP3− cells in K/BxN→B6.g7 mice (P < 0.021) or the CD4+CD25−FoxP3− cells in normal B6.g7 mice (P < 0.0129) (Figure 5D). The host-derived CD4+CD25− cells also showed higher expression (mean fluorescence intensity) of CD73 and FR4 than did the donor-derived CD4+CD25− T cells (Figures 5E and F). We tested the immunosuppressive capacity of these cells in adoptive transfer. Host-derived CD4+CD25−CD73+FR4+ cells from K/BxN→B6.g7 mice (3 months after transfer of hematopoietic stem and progenitor cells) also delayed disease onset when transferred into K/BxN→B6.g7 Rag−/− mice (Figure 5B). Thus, these host-derived FoxP3− cells could be contributing to protection against the manifestations of autoimmunity. These cells also persist (87 ± 9.19% of CD45.2 cells) in the recipient mice at day 48 (not shown), although their efficacy is overcome.
The phenotype of host-derived CD4+ T cells in the thymi of reconstituted mice.
In K/BxN→B6.g7 mice, a high proportion (47.2 ± 2.36%) of host-derived CD4+ single-positive thymic cells were CD25+FoxP3+ (Figure 6A). Further, a higher percentage of host-derived CD4+ single-positive CD25− thymic cells were CD44high, compared to donor-derived CD4+ single-positive CD25− thymic cells (Figure 6B), whereas CD4+ single-positive thymocytes typically are CD44low. We also measured CD103, an αEβ7 integrin previously shown to be expressed on Treg cells (25) and on a unique subset of CD4+CD25− T cells that are suppressive in function (26). We found high CD103 expression on host-derived CD4+CD25−FoxP3− T cells in the thymi of K/BxN→B6.g7 mice compared to normal B6.g7 mice or to donor-derived CD4+CD25−FoxP3− T cells (Figure 6C). The host-derived and donor-derived splenic CD4+CD25−FoxP3− T cells expressed low levels of CD103 (data not shown). The expression of CD103 on the host-derived CD4+CD25−FoxP3− T cells in the thymus suggests a regulatory phenotype for these cells.
Transfer of hematopoietic stem and progenitor cells from K/BxN mice into B6.g7 Rag−/− mice results in disease similar to that in K/BxN mice. However, transfer of hematopoietic stem and progenitor cells into B6.g7 mice results in significant attenuation of arthritis. What regulates autoimmunity in the reconstituted mice? Arthritis in K/BxN mice is thought to be initiated by the extravasation of circulating anti-GPI–GPI immune complexes into joints, followed by complement-induced tissue destruction (27). Anti-GPI autoantibody titers in reconstituted K/BxN→B6.g7 chimeras are much lower than in K/BxN mice and K/BxN→B6.g7 Rag−/− chimeras. This is likely a key explanation for reduced arthritis in K/BxN→B6.g7 mice. This, in turn, implicates defect(s) in the T and/or B cell compartments of these mice. Data on chimerism in K/BxN→B6.g7 mice indicate that B cells and neutrophils are >98% donor (K/BxN) derived, and the fact that low levels of anti-GPI antibodies are produced demonstrates that the B cell compartment is competent to develop this specificity. In contrast, only ∼10% of the peripheral CD4+ T cells in K/BxN→B6.g7 mice are donor derived, and the rest are host-derived CD4+ T cells that emerge despite irradiation, even at high (1,100-rad) doses (not shown).
The few donor-derived CD4+ T cells in the thymus and periphery of K/BxN→B6.g7 mice include a lower proportion of tetramer+Vβ6+ cells than in K/BxN→B6.g7 Rag−/− mice and fewer thymic Vβ6intermediateVβ8+ cells than in B6.g7 Rag−/− chimeras. This reduced level of donor-derived KRN TCR–expressing T cells may provide only limited amounts of help to GPI-specific B cells. In K/BxN mice, Vβ6intermediateVβ8+CD4+ T cells escape negative selection due to lower TCR affinity compared to Vβ6+CD4+ T cells (13). The reduced proportion of donor-derived thymic Vβ6intermediateVβ8+CD4+ T cells in K/BxN→B6.g7 mice suggests increased negative selection of these thymocytes. Reduced frequency of circulating tetramer+CD4+ T cells in these mice also points to a process of negative selection. The mechanism of these phenomena is under investigation (also see below).
The surviving donor-derived CD4+ T cells are further restrained by immunosuppressive mechanisms. They are less activated in B6.g7 than in B6.g7 Rag−/− recipients, as reflected in a lower proportion that is CD44high, and they have reduced levels of IFNγ and IL-17, 2 cytokines that are important for the development of arthritis in K/BxN mice (10, 28). Key contributors to this suppression are likely the CD25+FoxP3+ cells that are enriched among the host-derived CD4+ T cells in the K/BxN→B6.g7 chimeras, persist even 3 months after transplant, and function to delay arthritis in adoptive transfer. Consistent with our results, residual, host-derived Treg cells allow survival of myeloablated B6 recipient mice following syngeneic BM transfer, whereas B6 Rag−/− recipients succumb to lethal syngeneic graft-versus-host disease (29). Interestingly, host-derived CD4+ T cells in K/BxN→B6.g7 mice do not expand to populate the lymphopenic CD4+ T cell compartment to achieve homeostatic proportions. The cause of this limitation needs to be investigated.
CD44, an activation and memory marker, is expressed at higher levels in the host-derived CD4+CD25−FoxP3− Teff cell population than in the donor-derived CD4+CD25−FoxP3− cells or in B6.g7 mice. This result also resembles observations in BM transplantation of syngeneic or Rag-deficient animals, where memory-like (CD44+) T cells are enriched in the host-derived CD4+ T cell compartment (6, 21, 22). Like conventional CD4+CD25+ Treg cells, a subset of memory-like CD4+CD44high T cells can also restrain activation and proliferation of naive T cells (23, 24).
As a first step toward investigating immunosuppressive mechanisms of these cells in our model, we analyzed the CD4+CD25−FoxP3− cells for other markers of suppressive capacity, CD73 and FR4. CD73, a GPI-linked surface protein with ecto-5′-nucleotidase activity, mediates suppression by CD25+FoxP3+ T cells (23, 30). CD39, an ecto-ATPDase that is also expressed by Treg cells and neutrophils, rapidly converts circulating ATP and ADP to 5′-AMP (31). Through its ectonucleotidase activity, CD73 converts 5′-AMP to adenosine (32). Adenosine mediates its immunoregulatory activities through various adenosine receptors expressed on T cells, B cells, neutrophils, and macrophages (33–35). In addition to Treg cells, a subset of memory-like CD4+ T cells expresses high levels of CD73 and is suppressive by producing adenosine (23). FR4 is one of the receptor subtypes for folic acid. FR4 is expressed at high levels on both natural and transforming growth factor β–induced Treg cells. Host- and donor-derived conventional Treg cells (CD25+FoxP3+) in K/BxN→B6.g7 chimeras showed high expression of CD73 and FR4 (not shown). Strikingly, the conventional Teff cell or CD4+CD25−FoxP3− compartment also contains CD73+FR4+ cells, more so among host-derived cells, by percentages of positive cells and expression levels.
Martinez et al (36) recently showed that adoptive transfer of KRN T cells into lymphopenic TCRα−/− I-Ag7+ hosts, but not I-Ag7+ wild-type mice, results in arthritis. They demonstrate that protection against arthritis is due to induction of anergy of KRN Teff cells via enhanced CD73 and FR4 expression. That study differed from ours in that we observed host-derived CD4+ T cells that exhibited a CD73+FR4+ phenotype. Adoptive transfer of these host-derived CD4+CD25−CD73+FR4+ cells from K/BxN→B6.g7 mice into K/BxN→B6.g7 Rag−/− mice also delayed onset of arthritis. To our knowledge, this is the first demonstration of high levels of CD73+FR4+ cells with possible regulatory properties, in the Teff cell compartment, persisting in mice after lethal irradiation.
Adoptive transfer of host-derived CD25high cells or CD73+FR4+ T cells into K/BxN→B6.g7 Rag−/− mice leads only to delayed onset of disease and not to complete protection. This could be because additive or synergistic effects of these 2 regulatory populations are required for long-term control (both are present in K/BxN→B6.g7 mice).
The K/BxN→B6.g7 chimeras have a high frequency (65%) of host-derived thymic CD4+ T cells with a mature phenotype (CD44high), which is uncharacteristic of CD4+ single-positive thymocytes. The latter are CD44low and acquire the CD44high phenotype in the periphery after encountering antigen. Thus, these CD44highCD4+ T cells could have re-entered the thymus from the periphery. CD4+ T cells are capable of re-entering the thymus from the periphery under pathologic and lymphopenic conditions (22, 37–39), and most of the re-entering cells are of the activated phenotype (39). The CD4+CD25−FoxP3− host-derived thymic CD4+ T cells in our model express CD103, an integrin that is induced at sites of inflammation and involved in the retention of T cells at these sites (40). CD103 expression also increases in response to IL-2 (41). Expression of CD103 on host-derived CD4+ T cells may mediate their retention in the thymi of the K/BxN→B6.g7 chimeras. Retention of circulating CD4+ T cells in the thymus has been shown to contribute to tolerance induction, specifically negative selection (37, 39, 42). Recirculating peripheral T cells also contribute to proper thymic organization and regeneration of the medullary thymic epithelium in an SCID mouse model (43). These cells may be responsible, at least in part, for the increased negative selection of KRN TCR–expressing CD4+ single-positive thymocytes in our model.
An alternative origin for host-derived CD4+ single-positive T cells in our model can be proposed, based on data from syngeneic BM transplant. Here, evidence suggested that the host-derived CD4+ T cells arose mainly from a transient wave of differentiation from radiation-resistant DN2 (CD4−CD8−CD25+CD44+) precursors within the thymus (6, 44). The host-derived thymic CD4+ T cells of the K/BxN→B6.g7 chimera are also enriched for Treg cells. In a recent BM transplantation model, host-derived Treg cells required IL-2 for proliferation and IL-7 for survival (45). FoxP3 expression by the host-derived CD4+ single-positive cells in the K/BxN→B6.g7 chimera may result from increased thymic production of IL-2 due to robust negative selection (46).
In summary, persistent, host-derived CD4+ T cells with a high proportion of CD25+FoxP3+ cells and novel CD4+CD25−FoxP3−CD73+FR4+ cells are likely critical in protecting mice with autoimmune potential from full-blown disease. Recently, we found similar results using transfer of hematopoietic stem and progenitor cells carrying a transgenic, diabetogenic TCR into B6.g7 mice, indicating that the phenomena we observe are not unique to the KRN TCR and the arthritis model (Rajasekaran N, Mellins ED: unpublished observations). Our models will facilitate further study of the origin of host-derived CD4+ T cells and the mechanisms by which they mediate tolerance. Enrichment of these cell types is a potential therapeutic strategy in RA.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Mellins had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Rajasekaran, Mellins.
Acquisition of data. Rajasekaran, Wang, Truong, Macaubas, Beilhack.
Analysis and interpretation of data. Rajasekaran, Wang, Rinderknecht, Macaubas, Shizuru, Mellins.
We thank Drs. Diane Mathis and Christophe Benoist (Harvard Medical School, Boston, MA) for kindly providing the KRN mouse line, Dr. Luc Teyton (The Scripps Research Institute, La Jolla, CA) for kindly providing the MHC/GPI tetramers, and Dr. Andreas Hadjinicolaou (University of Cambridge, Cambridge, UK) for his assistance in assessment of arthritis in mice. We thank Jeanette Baker, Swati Acharya, and Suparna Dutt (Stanford University, Stanford, CA) for their useful discussions.