Lymph Node Stromal Cell-Intrinsic MHCII Expression Promote MHCI-Restricted CD8 T Cell Lineage Conversion to Regulatory CD4 T Cells

Abstract

MHC class I-restricted CD4⁺ T cells have long been discovered in the natural repertoire of healthy humans as well as patients with autoimmune diseases or cancer, but the exact origin of these cells remains to be fully characterized. In mouse models, mature peripheral CD8⁺ T cells have the potential to convert to CD4⁺ T cells in the mesenteric lymph nodes (MLN). This conversion can produce a unique population of MHCI-restricted CD4⁺ T cells including Foxp3⁺ regulatory T cells termed MHCI-restricted CD4⁺Foxp3⁺ T (CI-T_reg) cells. In this study we examined the cellular and molecular elements which promote CD8-to-CD4 lineage conversion and the development of CI-T_reg cells in mice. Using adoptive transfer and bone marrow chimera experiments, we found that the differentiation of CI-T_reg cells was driven by lymph node stromal cell (LNSC)-intrinsic MHCII expression as opposed to transcytosis of MHCII from bone marrow-derived APC. The lineage conversion was accompanied by Runx3 versus ThPOK transcriptional switch. This finding of a new role for LNSC in vivo led us to develop an efficient tissue culture method using LNSC to generate and expand CI-T_reg cells in vitro. CI-T_reg cells expanded in vitro with LNSC effectively suppressed inflammatory tissue damage caused by pathogenic CD4⁺ T cells in mouse models of colitis. This study identified a novel role of MHCII expressed by LNSC in immune regulation and the potential utilization of LNSC to generate novel subsets of immune regulatory cells for therapeutic applications.

Introduction

In a well-established immunological paradigm, the CD8 and CD4 T cells recognize antigens presented by the MHC class I (MHCI) and MHC class II (MHCII) molecules, respectively (1, 2). However, the existence of MHCI-restricted CD4 T cell clones has also long been known in the natural repertoire of healthy humans (3). The mismatch of MHCI restriction of TCR and CD4 co-receptors has been thought to be due to thymic mis-selection (4), although the origin and biological effect of the cross-differentiation have long been a debate.

CD4 and CD8 T cell lineages develop in the thymus in a dichotomic process of lineage specification by transcriptional and epigenetic programing (5–7). As such, it is generally thought that CD8 versus CD4 lineage commitment is stable. Nevertheless, a number of recent studies have shown that there is plasticity between these two major T cell lineages, in terms of conversion from the CD8 cytotoxic T cells to MHCI-restricted CD4⁺Foxp3⁺ T (CI-T_reg) cells and CD4⁺Foxp3⁻ CD4 T cells, or CD4 T_H or effector cells to MHCII-restricted cytotoxic CD8 T cells (8–11). For CD8-to-CD4 lineage conversion, our previous studies with athymic or thymectomized mice indeed confirmed that the lineage conversion occurred independently of the thymus (10). Kinetic evidence and data from experiments with lymphadenectomy showed that the lineage conversion originated from the mesenteric lymph nodes (MLN) (10). It is unclear what exact molecular and cellular pathways drive the development and differentiation of MHCI-restricted CD4⁺ T cells and CI-T_reg cells within the MLN, and how exactly their differentiation processes are different from thymic differentiation of MHC class II (MHCII)-restricted CD4⁺ T cells and CD4⁺Foxp3⁺ regulatory (CII-T_reg) cells.

The lymph node is composed of hematopoietic and non-hematopoietic cells and is one of the primary locations of lymphocyte activation and differentiation in the periphery (12, 13). The primary signal for T cell activation and differentiation is derived from MHC-mediated antigen presentation (14). Previous studies have shown that genetic ablation of MHCII in the OT1⁺Rag^o model abrogated the development of both MHCI-restricted CD4⁺ T cells and CI-T_reg cells (10). MHCII is mainly expressed by professional antigen presenting cells (APC) in the hematopoietic lineage, such as dendritic cells and macrophages. These cells provide the MHCII signal for T cell activation and differentiation in a typical immune response (15).

MHCII expression on professional APC has been studied extensively (16, 17); however, recent studies have found an important role of MHCII expression on non-hematopoietic cells in lymphoid organs (18–20). In the lymph node, the non-hematopoietic population is commonly referred to as lymph node stromal cells (LNSC). LNSC compose approximately 1% of the cells found in the lymph node and can be divided into different subsets (13). LNSC subsets include fibroblastic reticular cells (FRC), lymphatic endothelial cells (LEC), and blood endothelial cells (BEC). These populations are commonly identified based on differential expression of podoplanin (PDPN) and PECAM1 (CD31). FRC are characterized as PDPN⁺CD31⁻, LEC are PDPN⁺CD31⁺, and BEC are PDPN⁻CD31⁺. Additionally, there is a small population of CD45⁻ cells which are double negative for PDPN and CD31 (DN) (12, 13, 21).

Traditionally, LNSC are believed to provide structural and nutritional support for lymphocytes within the lymph node, but more recently several new roles for LNSC have been characterized. These functions include but are not limited to roles in fluid drainage, lymphocyte trafficking, nutrient delivery (21), antigen presentation (18, 22–25), tolerance induction (26, 27), immune suppression (28, 29), innate immune response, and angiogenesis (25, 30–32).

In this study we examined the role of hematopoietic APC and the non-hematopoietic stromal cell compartment in driving CD8-to-CD4 lineage conversion and CI-T_reg cell differentiation. We uncovered a novel role of LNSC in immune suppression and tolerance induction, showing MHCII expression by LNSC promoted the development of CI-T_reg cells and CI-T_reg cells expanded in vitro with LNSC had the ability to protect against immune-mediated tissue damage in vivo.

Material and Methods

Mice

The following transgenic and knockout mice were described previously: OT1 (33), MHCII knockout (MHCII^o) (34), Foxp3^FIR (35), RAG1 knockout (Rag^o) (36), CD8cre (37), and Rosa26YFP (which carries a yellow fluorescent protein (YFP) reporter) (38). These mutant strains were obtained from The Jackson Laboratory (Bar Harbor, ME). These strains were intercrossed to generate combinatorial mutant strains for studies of CD8 T cell lineage fate at monoclonal or polyclonal levels. Animals were maintained in specific pathogen free facility. In our experimental system, we did not observe a difference between male and female animals. Therefore, both sexes with littermate controls were used in this study. The studies were approved by the Institutional Animal Care and Use Committee at the University of Miami.

Flow cytometry and cell sorting

Flow cytometry analyses and cell sorting were conducted using a standard procedure. Single cell solutions were prepared and blocked for non-specific binding of antibodies with anti-CD16/CD32 antibodies and normal mouse serum. Cells were stained with fluorescent antibody conjugates to determine cell phenotype. The following antibody conjugates were used: PacBlue-conjugated anti-CD8 (53–6.7) from Biolegend (San Diego, CA), V500-conjugated anti-CD4 (RM4–5) from BD Bioscience (San Diego, CA), eF660-conjugated anti-CD4 (GK1.5) from Thermo Fisher (Waltham, MA), PerCPcy5.5-conjugated anti-TCR (H57–597), BV785-conjugated anti-TCR (H57–597) (Biolegend), BV510-conjugated anti-CD103 (M290) (BD Bioscience), APC-conjugated TIGIT (GIGD7) (Thermo Fisher), PerCPcy5.5-conjugated anti-PD1 (29F.1A12), PE-Cy7-conjugated anti-CD25 (PC61.5), AL700-conjugated anti-CD45 (30-F11), FITC-conjugated anti-CD102 (3C4) (Biolegend), AF523-conjugated anti-CD45 (30-F11) (Thermo Fisher), PE-conjugated anti-PDPN (8.1.1) (Biolegend), APC-conjugated anti-MHCII (M5/114.15.2) (Thermo Fisher), and APC-R700 conjugated anti-CD31 (MEC 13.3) (BD Bioscience). Foxp3 expression was identified by a “knockin” RFP reporter (35) or intracellular staining. ThPOK and Runx3 intracellular staining was performed based on the manufacture’s guidelines (BD Bioscience). Fixable Viability Dye eFlour 780 (Thermo Fisher) was used to exclude dead cells from analyses. “Doublets” were gated out before further analysis of marker expression by single cells. Samples were analyzed with a Cytoflex from Beckman Coulter (Brea, CA) or Sony SP6800 from Sony Biotechnology (San Jose, CA). Cell purification was done with FACS Aria sorter (BD Bioscience).

Lymph node enzymatic digestion

MLN from Balb/c, C57BL/6 (B6) and MHCII^o mice were collected and placed in RPMI-1640 from Sigma-Aldrich (St. Louis, MO) containing 2% FBS on ice. Once collected, RPMI was removed and 3mL of freshly made enzyme digestion buffer were added which contains 0.8 mg/ml of dispase (Sigma-Aldrich), 0.2 mg/ml of Collagenase P (Sigma-Aldrich), and 0.1 mg/ml of DNAse I from Worthington Biochemical Corporation (Lakewood, NJ) in RPMI-1640 containing 2% FBS. MLN digestion mixture was incubated at 37°C for 15 minutes. After 15 minutes MLN were gently disrupted using a pipette. The supernatant was collected and added to 2mL of cell collection media containing 2% FBS and 5 mM EDTA in PBS and left on ice. After collecting the supernatant, 2 mL of enzyme digestion buffer was added to the remaining lymph nodes and incubated at 37°C for 10 minutes. Lymph nodes were disrupted every 5 minutes in order to aid in the release of cells. After 10 minutes, the supernatant was collected and added to the lymph node collection media. This was repeated until the lymph node is completely digested and no cells appear to be released when disrupted (digestion media is clear). Once the digestion is complete, tubes containing the supernatant and cell collection media were centrifuged for 3 minutes at 4°C at 300 g. The supernatant was removed, and cells were resuspended in 1 mL of RPMI-1640 containing 2% FBS. Cells were filtered through a 70 micron mesh strainer.

Magnetic beads-based cell isolation

A single cell suspension was prepared and labeled with biotinylated antibodies according to standard procedures. LNSC isolation was done with biotinylated anti-CD45 (30-F11) (Biolegend) antibodies and followed with magnet depletion of CD45⁺ cells with streptavidin-conjugated magnetic beads from Miltenyi Biotec (Auburn, CA). A cocktail of biotinylated antibodies including anti-Ter119 (TER-119) (Biolegend), anti-CD3 (145–2C11), anti-CD4 (RM4–5), anti-CD8 (53–6.7), and anti-TCRb (H57–597) (Thermo Fisher) was used for preparation of T-cell depleted bone marrow. A cocktail of biotinylated antibodies including anti-Ter119 (TER-119), anti-B220 (RA3–6B2), anti-CD11b (M1/70) (Biolegend), anti-CD11c (N418), anti-Ly6G (RB6–8C5), anti-F4/80 (BM8), anti-CD49b (HMa2), and anti-CD4 (RM4–5) (Thermo fisher) were used for negatively selecting CD8 T cells for adoptive transfer experiment (purity >90%). A cocktail of biotinylated antibodies including anti-Ter119 (TER-119), anti-B220 (RA3–6B2), anti-CD11c (N418), anti-Ly6G (RB6–8C5), anti-F4/80 (BM8), anti-CD49b (HMa2), anti-CD11b (M1/70), anti-CD25 (PC61.5) (Thermo Fisher) and anti-CD8 (53–6.7) were used for negatively selecting CD4⁺CD25⁻ pathogenic effector T (T_eff) cells in colitis induction experiments.

Quantitative RT-PCR

LNSC were isolated from the MLN of B6 mice using magnetic beads-based negative selection based on CD45 expression. Dendritic cells were isolated from the spleen of both B6 and MHCII^o mice by magnetic beads-based CD11c positive selection. Cells were immediately dissolved in Trizol reagent (Thermo Fisher) after isolation. The mRNA isolation and cDNA synthesis were performed according to a standard procedure. Quantitative PCR was performed using H2-Ab specific primers: 5’ GTCCTGGTCATGCTGGAGAT 3’ (forward) and 5’ TCCTGTGACGGATGAAAAGG 3’ (reverse) with SYBR® Green qPCR mastermix from Qiagen (Hilden, Germany). Relative units of mRNA were calculated against the expression level of a housekeeping gene, Hprt.

Immunofluorescence microscopy

MLN were collected and fresh-frozen in embedding media. The cryopreserved tissue blocks were sectioned with cryostat at 6 um thickness, fixed with acetone, and then stained with labeled antibodies. Immunofluorescence microscopy was conducted using a standard protocol using a fluorescence microscope from Keyence (Itasca, IL). The following antibody conjugates were used: APC-conjugated anti-CD4 (GK1.5) (Biolegend), biotinylated anti-CD8 (53.67) (Thermo Fisher), biotinylated anti-CD31 (390) (Thermo Fisher), AF488-conguated streptavidin (Thermo fisher), PE-conjugated PDPN (8.1.1) (Biolegend), APC-conjugated anti-MHCII (M5/114.15.2) (Thermo Fisher).

In vitro CI-T_reg cell expansion with LNSC

Lymph nodes were collected from OT1⁺Rag^o or CD8cre-R26YFP mice. For co-culture of lymphocytes and LNSC, the lymph nodes were disrupted with enzymatic digestion as previously described and the whole mixtures of lymphocytes and LNSC were plated in 24 well plates, in comparison to culture of lymph node cells prepared by gentle mechanic disruption and filtration and thus were devoid of LNSC. DMEM (Sigma-Aldrich) containing 10% FBS from Omega Scientific (Tarzana, CA), 25 mM HEPES, L-glutamine, Sodium Pyruvate, Pen Strep, and nonessential amino acids was used. Cells were cultured for 7–10 days, when a population of CD4⁺Foxp3⁺ cells in the OT1 model or YFP⁺CD4⁺Foxp3⁺ in the CD8Cre-YFP model were observed. CD4⁺Foxp3⁻ cell population from the CD8 lineage was also present, although its population size was less consistent from experiment to experiment than that of CI-T_reg cells or the original CD8 T cells. At day 10 the suspension of the cell culture was collected and processed for cell sorting. Sorted CD4⁺Foxp3⁺ cells or CD8⁺ cells were transferred to Rag^o mice five days prior to CD4 T effector cell transfer.

Colitis models and histopathology analysis

To test the function of CI-T_reg cells, we used a standard colitis model, adoptive transfer of pathogenic CD4⁺CD25⁻ cells into immunodeficient (Rag^o) mice. Five days prior to the T_eff cell transfer, some groups of the Rag^o recipients had also received sorted CI-T_reg cells or the control CD8 T cells which was the original precursor of CI-T_reg cells. The spacing of T_reg and T_eff cell transfers was set partially due to logistic limitation of sorter access and the time needed for tail-vein recovery after the first injection. The intended T_reg and T_eff ratio was maintained in vivo and this schedule tested the effect of a pre-existing T_reg cell population. After T_eff cell transfer, animals were monitored for weight maintenance and sign of diarrhea. Large intestine samples were collected from euthanized animals and fixed in 10% formalin solution. Paraffin-embedded sections and H&E staining were done by the Pathology Research Resource Core at the University of Miami. Histopathology was documented by microscopy. Pathology scoring was performed in a blinded fashion to assess the extent of inflammatory damage according to the criteria previously described by Tajima and colleagues (39).

Statistics

Student’s t tests or Mann-Whitney tests were used for single comparisons and ANOVA was used for multiple group analyses. When analyzing nonparametric multiple groups, Kruskal-Wallis analyses was performed with false discovery rate adjustment. * P<0.05; **, P<0.01; ***, P<0.001; ns, not statistically significant.

Results

Conversion of adoptively transferred mature polyclonal CD8 T cells to CD4 T cells mediated by host MHCII expression

Previous studies have shown that MHCII deficiency in the OT1⁺Rag^o mice abrogated the cross differentiation from the CD8 clone to CD4 T cells (10). To examine the role of MHCII expression on CD8 lineage plasticity in a mature CD8 population possessing a natural TCR, mature peripheral CD8⁺ T cells (CD8^YFP) were isolated from CD8cre-R26YFP mice and transferred into either Rag^oMHCII⁺ or Rag^oMHCII^o mice. The CD8cre-R26YFP model uses an YFP reporter to mark the CD8 lineage in mature peripheral CD8 T cells, thus allowing for CD8 lineage plasticity to be traced (37, 38). After approximately 10 weeks post transfer, the T cell population from the mesenteric lymph nodes (MLN) and the large intestine lamina propria (LILP) were analyzed using flow cytometry. CD4⁺YFP⁺ populations were identified as the MHCI-restricted CD4 T cells converted from CD8 T cell lineages.

As shown in Figure 1, in both the MLN and the LILP the percentage and number of total YFP⁺ cells were unchanged in the MHCII⁺ and MHCII^o groups (Fig 1A, 1B), indicating the MHCII-deficiency in the host did not affect the homeostasis of adoptively transferred CD8⁺ T cells. Interestingly, the frequency and number of YFP⁺CD4⁺ T cells were diminished in the MHCII^o group compared to the MHCII⁺ group. The differentiation into CI-T_reg cells was abrogated in the MHCII^o group as well. These results were consistent in both the MLN and the LILP (Fig 1C, 1D). In addition to the YFP⁺CD4⁺ and YFP⁺CD8⁺ populations, there was a population of double positive cells (CD4⁺CD8⁺YFP⁺) which also exhibited reduced frequencies in the MHCII^o model. This population of cells may serve as a transition state during conversion from the CD8 lineage to the CD4 lineage. These observations suggest that MHCII expression by host antigen presenting cells has a critical role for CD8-to-CD4 lineage conversion.

Figure 1. Requirement of MHCII by host antigen presenting cells for conversion of mature CD8 T cells with natural T cell repertoire in adoptive transfer.

Lineage marked CD8 T cells were isolated from CD8cre-R26YFP mice and adoptively transferred into Rag-deficient mice with MHCII (Rag^oMHCII⁺) or without MHCII (Rag^oMHCII^o) and analyzed ~10 weeks later. (A) Representative flow cytometry plots of the MLN followed by a summary of percentages and cell counts. (B) Representative flow cytometry plots of the LILP followed by a summary of percentages and cell counts. (C) Representative flow cytometry plots of the MLN followed by a summary of cell counts (Gated on CD45⁺ CD8cre-R26YFP⁺). (D) Representative flow cytometry plots (Gated on CD45⁺ CD8cre-R26YFP⁺) of the LILP followed by a summary of cell counts. Cell counts are normalized to counts per million cells. The number in the flow cytometry plots is the percentage of the gated population. Each data point represents one animal (mean ± SEM, n = 5–9), *p < 0.05, **p < 0.01, ***p<0.001.

MHCII expression in the non-hematopoietic compartment facilitates the development of CI-T_reg cells

Following the observation that MHCII expression by host APC drove lineage plasticity of adoptively transferred CD8⁺ T cells, we further examined which cell type was playing a role on this process. Dendritic cells and macrophages in the hematopoietic compartment have long been known as the main type of cells for MHC-restricted antigen presentation (16, 17), but more recently studies have shown that non-hematopoietic cells can also express MHCII and function as antigen presenting cells (18, 22–25). To determine the relative contribution of the hematopoietic versus the non-hematopoietic compartment in CD8 lineage conversion, we generated bone marrow chimeric mice using the OT1 model and tracked the lineage fate of the CD8 T cell clone in this chimeric setting, in the absence of its cognate antigen.

We first examined the role of MHCII expression by non-hematopoietic cells. Bone marrow was isolated from OT1⁺Rag^o mice, depleted of T cells, and injected into neonate Rag^oMHCII⁺ or Rag^oMHCII^o mice. The neonate bone marrow reconstitution, as done in previous studies (40),was used to enable the study of bone marrow chimera without irradiation that is typically needed for bone marrow reconstitution in adult animals. In chimeric mice wherein the MHCII was deficient in the non-hematopoietic compartment, there was a striking reduction in the development of CI-T_reg cells. However, interestingly, the conversion from the CD8 T cell lineage to the CD4⁺Foxp3⁻ cells still occurred in this setting (Fig 2A). The large reduction of CI-T_reg population was evident in both the MLN and the LILP (Fig 2B), indicating an important role for MHCII expression by non-hematopoietic cells in the development of CI-T_reg cells.

Figure 2. MHCII expression by non-hematopoietic versus hematopoietic antigen-presenting cells for CI-T_reg development.

(A-B) Bone marrow was isolated from OT1⁺Rag^o mice, depleted for T cells and injected into neonatal (<3 days of age) Rag^oMHCII⁺ or Rag^oMHCII^o mice (7×10⁶ cells each). The reconstituted animals were analyzed ~6 weeks later. (A) Representative flow cytometry plots of the MLN showing CD4⁺ and CD8⁺ T cell populations (Gated on CD45⁺TCR⁺ besides singlet and live cell gates as described in the methods). (B) A summary of MLN and LILP cell counts for CD4⁺Foxp3⁻ and CD4⁺Foxp3⁺ T cell populations. (C-D) Bone marrow was isolated from either OT1⁺Rag^oMHCII⁺ or OT1⁺Rag^oMHCII^o mice, depleted for T cells, and injected into neonatal (<3 days) Rag^oMHCII⁺ mice (7×10⁶ cells each). The reconstituted animals were analyzed ~6 weeks later. (C) Representative flow cytometry plots of the MLN showing CD4⁺ and CD8⁺ T cell populations (Gated on CD45⁺TCR⁺ besides singlet and live cell gates as described in the methods). (D) A summary of MLN and LILP cell counts for CD4⁺Foxp3⁻ and CD4⁺Foxp3⁺ T cell populations. Cell counts are normalized to counts per million cells. The number in the flow cytometry plots is the percentage of the gated population. Each data point represents one animal (mean ± SEM, n = 4–5), **p < 0.01.

Next, we examined the role of MHCII expression by hematopoietic cells. T cell-depleted bone marrow from OT1⁺Rag^o mice with or without MHCII (MHCII⁺ or MHCII^o) was injected into neonate Rag^o mice (MHCII⁺). As shown in Figure 2C and 2D, MHCII deficiency in the hematopoietic compartment did not significantly reduce the numbers of CI-T_reg cells or the CD4⁺Foxp3⁻ cells converted from the CD8 lineage in either the MLN or LILP (Fig 2C, 2D), suggesting the lack of a substantial role for MHCII expressed by hematopoietic cells in CI-T_reg cells generation and homeostasis. Taken together, the observations from the complementary sets of bone marrow chimera experiments indicate MHCII expression by non-hematopoietic cells is playing an important role in the development of CI-T_reg cells in vivo.

CD8 to CD4 T cell lineage conversion was reflected in the transcriptional switch to ThPOK programming

The CD8 and CD4 T cell lineages are driven by distinct transcriptional programs. To examine whether the conversion from the CD8 to the CD4 T cell lineage occurred with a transcriptional switch, we analyzed the expression of Runx3 and ThPOK using flow cytometry. The CD8 T cell lineage is driven by Runx3 programming while the CD4 lineage is driven by ThPOK programming (5–7). As shown in Figure 3, mature CD8⁺ T cells and CD4⁺ T cells from the lymph node had mutually exclusive expression of Runx3 and ThPOK, respectively (Fig 3A, 3B). To determine if a transcriptional switch occurred in the converted CD4⁺ T cells, we first examined T cells from the OT1⁺Rag^o model which represented T cell lineage conversion at the steady state. The CD4⁺ T cells converted from the CD8⁺ T cell clone indeed upregulated ThPOK and downregulated Runx3 (Fig 3C). Second, we examine Rag^o animals reconstituted with mature CD8⁺ T cells. Because of technical difficulties with the polyclonal CD8cre-R26YFP model in terms of the YFP signal being extinguished by the fixative used for the intracellular staining, we focused on the monoclonal model which the CD8 lineage origin was known. Thus, we isolated mature CD8⁺ T cells from the OT1⁺Rag^o model. The CD4⁺ T cells converted from adoptively transferred CD8⁺ T cells upregulated ThPOK and downregulated Runx3 (Fig 3D). Interestingly, the CD4⁺ T cells converted in the steady state OT1⁺Rag^o mice maintained a significant amount of Runx3 expression, but the ThPOK upregulation was on par with that of the naturally differentiated CD4⁺ T cells in B6 mice. On the other hand, in the adoptive transfer setting the CD4⁺ T cells expressed Runx3 and ThPOK at levels similar to the B6 CD4⁺ T cells (Fig 3E).

Figure 3. Transcriptional program of converted CD4⁺ T cells.

Analysis of transcriptional factors Runx3 and ThPOK was done by intracellular staining of cells from the MLN. (A) Isotype control was used for the intracellular staining of Runx3 and ThPOK. Representative flow cytometry plots of Runx3 and ThPOK expression by CD4 and CD8 T cells were shown for the following groups of mice: (B) B6, (C) OT1⁺Rag^o, (D) Rag^o reconstituted with CD8⁺ T cells isolated from OT1⁺Rag^o animals. The following gating strategy was used for all groups: single cells, live/dead, TCR. (E) Summary of median fluorescent intensity (MFI) of Runx3 and ThPOK of the converted CD4⁺ T cells compared to that of natural CD4⁺ T cells. The number in the flow cytometry plots is the percentage of the gated population. Each data point represents one animal (mean ± SEM, n=3–5), *p < 0.05.

Intrinsic MHCII expression by LNSC plays a major role in CI-T_reg development

The results from the bone marrow chimera experiments shown in Figure 2, taken together with our previous report that the generation of CI-T_reg cells originated in the MLN (10), indicated a critical role of MHCII expression by the non-hematopoietic compartment in the MLN. In the lymph node, the non-hematopoietic compartment is composed of LNSC.

We analyzed LNSC within the MLN for their MHCII expression. Although the mutant animals used in this study are all on the B6 genetic background, we analyzed the MHCII expression by stromal cells in MLN from both B6 and Balb/c mice, to assess the generality of the finding. MHCII^o mice were used as a rigorous negative control for the anti-MHCII antibody staining. We analyzed four main subsets of LNSC including FRC (PDPN⁺CD31⁻), LEC (PDPN⁺CD31⁺), BEC (PDPN⁻CD31⁺), and DN (PDPN⁻CD31⁻) or sometime referred to as epithelial cells (Fig 4A). A MHCII⁺ population was identified within all four subsets of LNSC from both B6 and Balb/c mice (Fig 4B, 4C). The percentage of MHCII⁺ cells within each subset was similar between the B6 and Balb/c LNSC, suggesting that MHCII expression by LNSC is not a unique property of a particular genetic background. Using immunofluorescence staining of MLN tissue sections from B6 mice, we detected the regional localization of LNSC subsets in the MLN which expressed MHCII with PDPN and CD31, the main markers used to define LNSC populations (Fig 4D).

Figure 4. MHCII expression by non-hematopoietic cells (stromal cells) in the lymph node.

MLN from B6, Balb/c, and MHCII^o mice were collected and subjected to enzymatic digestion. The CD45⁻ population (LNSC) was analyzed based on the following flow cytometry gating strategy: single cells, live/dead, CD45⁻. (A) Representative flow cytometry plots of the MLN defining four LNSC populations: FRC (Pdpn⁺CD31⁻), LEC (Pdpn⁺CD31⁺), BEC (Pdpn⁻CD31⁺), and DN (Pdpn⁻CD31⁻). (B) Representative flow cytometry plots of MHCII⁺ cells for each LNSC subset. Flow cytometry plots for each subset are as followed: FRC (top left), LEC (top right), BEC (bottom right), DN (bottom left). (C) Summary of the percentage of MHCII⁺ cells for each LNSC subset. (D) Immunofluorescence staining for PDPN (red), CD31 (green), MHCII (cyan), and DAPI (blue) of the MLN from B6 mice. The number in the flow cytometry plots is the percentage of the gated population. Each data point represents one animal (mean ± SEM, n = 5–6), *p < 0.05, **p < 0.01.

Although the four main LNSC populations have been well defined, there is a great amount of diversity that exist in each of these subsets. In order to gain a clue on the potential functional impact of the MHCII expressing LNSC, we analyzed MHCII expression in the context of Intercellular adhesion molecule 2 (ICAM2) or CD102 expression. CD102 is expressed on endothelium and binds LFA-1 to facilitate the extravasation of lymphocytes. It is highly expressed on resting endothelial cells (41). Therefore, it might promote T cell adhesion to LNSC and thus facilitate MHCII-driven conversion of the CD8 T cell lineage. This hypothesis was refuted as we did not detect MHCII expression in CD102⁺ subsets of LNSC (Fig 4B). Future studies are needed to identify the specific LNSC subset responsible for CD8 T cell conversion.

To further understand the role of MHCII by LNSC in the process of CD8 T cell lineage conversion, we investigated the mechanism of MHCII expression. Studies have shown two main mechanisms in which LNSC express MHCII. The first is through the process of transcytosis, in which the LNSC acquire the MHCII molecule from antigen presenting cells such as dendritic cells (18). Secondly, LNSC have been shown to endogenously express MHCII (13). To determine which mechanism was responsible for MHCII expression that drives CI-T_reg cells development, we generated bone marrow chimeric mice containing MHCII-deficiency in the non-hematopoietic cells. Bone marrow was isolated from OT1⁺Rag^o mice, depleted of T cells and injected into neonate Rag^oMHCII⁺ control mice or Rag^oMHCII^o experimental group. The latter group represents a chimeric setting of MHCII deficiency in the stromal but not the hematopoietic compartment and enabled us to examine potential transcytosis from hematopoietic cells to LNSC in the process of CD8 lineage conversion.

Six weeks after the bone marrow transfer, only a very small percentage of MHCII⁺ LNSC were detected in the MHCII^o group, compared to the MHCII⁺ group (Fig 5A). The largely absent MHCII expression in the chimera is consistent with the large reduction of CI-T_reg cells in the same chimeric setting, as opposed to the control chimera with MHCII⁺ stromal compartment which had generated a robust population of CI-T_reg cells (Fig 2) by six weeks after bone marrow transfer. These results indicate that the transcytosis mechanism unlikely plays a major role in the development of CI-T_reg cells, and endogenous expression of MHCII by LNSC is most likely a critical contributor to this process.

Figure 5. Transcytosis versus intrinsic MHCII expression by lymph node stromal cells.

(A) Bone marrow was isolated from OT1⁺Rag^o mice and the T cells were depleted. Following T cell depletion, 7×10⁶ cells were injected into neonatal (<3 days of age) Rag^oMHCII⁺ or Rag^oMHCII^o mice and analyzed ~6 weeks later. MLN were collected and subjected to enzymatic digestion in order to analyze LNSC. Representative flow cytometry plots followed by a summary of percentages and cell counts. The following flow cytometry gating strategy was used: single cells, live/dead negative, CD45⁻. (B) Relative MHCII (H2-Ab) mRNA expression by qRT-PCR of LNSC populations or dendritic cells (CD11c⁺). (C) MLN from B6 mice were collected and subjected to enzymatic digestion. Representative flow plots followed by percentages and MFI of MHCII⁺ populations for dendritic cells (CD45⁺CD11c⁺) or LNSC (CD45⁻CD11c⁻). The number in the flow cytometry plots is the percentage of the gated population. Each data point represents one animal (mean ± SEM, n = 5–6), *p < 0.05, **p < 0.01.

To verify the intrinsic MHCII expression by LNSC, we examined MHCII mRNA level by LNSC compared to that by dendritic cells. As shown in Figure 5B, quantitative RT-PCR detected a high level of MHCII mRNA (H-2Ab) in LNSC from the MLN, although it was lower than what was seen in dendritic cells (Fig 5B). Of note, in this experiment, the whole populations of LNSC were analyzed. Therefore, a lower percentage of MHCII expressing cells in LNSC could contribute to the lower readout of mRNA level compared to the dendritic cells (Fig 5C). Regardless, these results from Figure 5, taken together with the data from Figure 2, strongly suggest that intrinsic MHCII expression by LNSC, as opposed to transcytosis, plays the major role in driving the development of CI-T_reg cells.

Converted CD4⁺ T cells localize in the niche of the lymph node with MHCII expression by stromal cells

To better understand the stromal context associated with MHCII-driven conversion of CD8 T cell lineage, we used fluorescence microscopy of MLN tissue sections. MLN were isolated from either Rag^oMHCII⁺ or Rag^oMHCII^o mice reconstituted with CD8⁺ T cells from CD8cre-R26YFP mice (CD8^YFP). MLN sections were stained for CD4 and CD8 T cell markers or MHCII, PDPN, and CD31 in serial tissue sections of the same sample.

In the MLN of Rag^oMHCII⁺ mice reconstituted with CD8^YFP T cells, a clear population of both CD8⁺ T cells and CD4⁺ T cells were detected. When compared with the LNSC staining in the serial section, these T cell populations localized with a microenvironment of stromal cells with MHCII expression (Fig 6A). In contrast, in the Rag^oMHCII^o group, MHCII expression was not detected on LNSC. There was a rare population of MHCII⁺ cells that did not have the co-staining with LNSC markers and was most likely due to MHCII⁺ cells from adoptive transfer. The absence of MHCII in LNSC correlated with rarity of CD4⁺ T cells whereas CD8⁺ T cells were abundant in the microenvironment (Fig 6B). These findings suggest conversion of the CD8 T cell lineage to the CD4 T lineage likely occurred in the specific MLN microenvironments that were defined by the expression of MHCII by stromal cells.

Figure 6. Detection of converted CD4⁺ T cells in the microenvironment with MHCII expression by lymph node stromal cells.

MLN were collected and fresh-frozen in embedding media. The cryopreserved tissue blocks were sectioned with cryostat at 6 um thickness and then stained with labeled antibodies. The MLN tissue sections were stained for fluorescence microscopy. Left: Staining for DAPI (blue), CD8 (green), and CD4 (red). Right: Staining for DAPI (blue), Pdpn (red), CD31 (green), and MHCII (cyan). (A) Representative immunofluorescence images of MLN sections from Rag^oMHCII⁺ mice reconstituted with CD8⁺ T cells from CD8cre-R26YFP mice; (B) Representative immunofluorescence images of MLN sections from Rag^oMHCII^o mice reconstituted with CD8⁺ T cells from CD8cre-R26YFP mice (n = 5 for each group).

Lymph node stromal cells effectively expanded CI-T_reg cells in vitro

Prompted by the results of the in vivo experiments showing that LNSC play an important role in the conversion from the CD8 T cell lineage to CI-T_reg cells, we tested whether LNSC could be used for generating and/or expanding CI-T_reg cells in vitro for therapeutic testing, as a potential solution to overcome one of the major hurdles in T_reg cell therapy — the large number of cells required (42).

First, we tested LNSC expansion of monoclonal CI-T_reg cells with known specificity yet in the absence of cognate antigens. MLN were collected from OT1⁺Rag^o mice and subjected to either enzymatic digestion to prepare a mixture of LNSC and immune cells or to a standard method of gentle mechanic dissociation and filtration which leads to a cell preparation of immune cells without a substantial presence of LNSC. The cells were plated in culture without addition of the specific antigen for OT1 T cells. After one week, the culture was analyzed by flow cytometry. In the culture with the presence of LNSC, a population of CD4⁺ T cells, including CI-T_reg cells subsets, was detected. In the absence of LNSC the CD4⁺ T cells were not detected in the culture of MLN cells from OT1⁺Rag^o mice (Fig 7A), even though the starting population had a subset of converted CD4⁺ T cells (10). The results indicated a loss of those cells in the absence of LNSC in the culture. Therefore, in the in vitro culture, LNSC might help promote the survival of the MHCI-restricted CD4⁺ T cells including the CI-T_reg cell subset, although the technical limitation of the culture model precludes us from determining whether the CD4⁺ T cells and CI-T_reg cells detected in the LNSC culture were the product of in vitro conversion from CD8⁺ T cells, or the survival and/or expansion of a pre-existing population.

Figure 7. In vitro culture of CI-T_reg and CII-T_reg cells with lymph node stromal cells.

(A) MLN were collected from OT1⁺Rag^o mice, subjected to enzymatic digestion and then cultured in 24 well cell-culture treated plates as a whole mixture of hematopoietic cells and stromal cells (the “with stromal cell” group). Alternatively, the MLN cells were mechanically disassociated with a standard procedure of gentle grinding between glass slides, passing through syringe needles and filtration (the “no stromal cell” group), and then plated in the 24 well plates. The cell cultures were analyzed at day 7. Representative flow plots followed by a summary of percentages (Gated on CD45⁺TCR⁺). Each data point represents one well, with each data set representing pooled wells from three independent experiments. (B-C) MLN were collected from CD8cre-R26YFP mice and subject to enzymatic digestion and cultured in 24 well cell culture treated plates. Representative flow plots followed by a summary of percentages and MFI for PD-1, CD103, and TIGIT for the CD4⁺CD25⁺Foxp3⁺ and CD4⁺CD25⁺Foxp3⁻ populations. Each data point represents one well, with each data set representing pooled wells from three independent experiments. The flow cytometry gating strategy for MHCI-restricted cells was as follows: single cells, CD45⁺TCR⁺YFP⁺. The flow cytometry gating strategy for MHCII-restricted cells was as follows: single cells CD45⁺TCR⁺YFP⁻. The number in the flow cytometry plots is the percentage of the gated population (mean ± SEM, n = 5–8), *p < 0.05, **p < 0.01, ***p<0.001.

Next, we tested LNSC-immune cell co-culture for in vitro generation and/or expansion of polyclonal CI-T_reg cells. We used MLN cells from the CD8cre-R26YFP model to track the lineage fate of the YFP-marked CD8 lineage. Of note, in the steady state without any immune perturbation, CI-T_reg cells exists in CD8cre-R26YFP mice at extremely low frequencies (<0.1% of CD4 T cells) (10). This model also allows for the comparison of MHCI-restricted and MHCII-restricted CD4⁺ T cells within the same culture system, distinguished by the presence and absence of the YFP CD8 lineage marker, respectively. MLN from CD8cre-R26YFP mice were dissociated into single cell preparation by enzymatic digestion and plated in culture without antigenic stimuli. One week later, a population of CD4⁺ T cells were detected in the culture, including CD4⁺Foxp3⁺ subsets that carried the YFP marker for CD8 lineage origin (Fig 7B). Interestingly, in our culture model, we observed a relatively high percentage of CI-T_reg cells out of the converted CD4⁺ T cells (Fig 7B), although the total number of CI-T_reg cells and CII-T_reg cells differentiated/expanded in the cell culture were comparable (data not shown). This finding complements the previous report that stromal cells can drive CII-T_reg cells differentiation in IL-2 and antigen dependent mechanisms (43). Furthermore, our initial tests with a few markers identified some differences between the CI-T_reg cells and CII-T_reg cells differentiated/expanded in vitro. The CI-T_reg population exhibited higher levels of PD-1 and CD103 expression than the CII-T_reg population, while there was no significant difference in expression of TIGIT (Fig 7C). Both PD-1 and TIGIT are inhibitory receptors and associated with immunosuppression (44), while CD103 is involved in localization of cells to the gut microenvironment as well as other mucosal sites (45). Upregulation of PD-1 and CD103 may aid in the trafficking of CI-T_reg cells to mucosal regions to protect mucosal surfaces from inflammatory injury. Overall, the results from the in vitro experiments showed that LNSC co-culture can effectively expand CI-T_reg cells in vitro.

CI-T_reg cells expanded in vitro with LNSC suppressed immune damage by pathogenic CD4 T cells in the large intestine

To test the in vivo function of CI-T_reg cells expanded in vitro with LNSC co-culture, we used a standard model of colitis induced by the transfer of CD4 T effector (T_eff) cells into Rag^o mice (46). Lymph nodes from CD8cre-R26YFP mice were subjected to enzymatic digestion and cultured for 10 days. Suspension cells in the culture were collected for sorting CI-T_reg cells as well as the remaining CD8 cell population by flow cytometry. The purified subsets were then injected into Rag^o mice followed by T_eff cell transfer 5 days later at 1:5 ratios. Of note, the converted CD4⁺Foxp3⁻ cells were not sorted for this experiment due to the variation of its population size at day 10 culture and logistic constrains of recipient animal availability. However, our previous study showed MHCI-restricted CD4⁺Foxp3⁻ cells did not exhibit suppressive function in their physiological niche or in adoptive transfer but rather functioned as pathogenic or helper cells (10).

The adoptive transferred CI-T_reg cells indeed protected the animals from immune damage mediated by pathogenic CD4 T_eff cells. Animals which received CI-T_reg cells had a superior overall health as reflected in their weight maintenance, compared to the animals injected with control CD8 cells plus pathogenic CD4 T_eff cells or the pathogenic CD4 T_eff cell alone which exhibit obvious sign of diarrhea and weight loss (Fig 8A). The large intestines from those two groups also exhibited gross appearance of inflammatory damage including thickening and shortening, in contrast to that of the animals which received CI-T_reg cells (Fig 8B). Histopathological examination revealed that the animals which received the CD8 control cells with pathogenic CD4 T_eff cells or pathogenic CD4 T_eff cell alone showed extensive tissue damage in the large intestine including lymphocyte infiltration into the submucosa and the mucosa as well as the loss of goblet cells throughout the tissue (Figure 8C, 8D).

Figure 8. In vivo efficacy of CI-T_reg cells expanded in vitro with lymph node stromal cells.

Lymph nodes were collected from CD8cre-R26YFP mice and subjected to enzymatic digestion. The lymph node cells with LNSC were cultured for 10 days and then the following populations were sorted: YFP⁺CD4⁺Foxp3⁺ cells (CI-T_reg) and YFP⁺CD8⁺ cells. Sorted cell populations were transferred to Rag^o mice and followed by CD4 T_eff cell injections 5 days later. Cells were injected at a ratio of 1:5. (A) Representative weight loss curve. (B-D) Pathology assessment of colitis development. Each data point represents one animal (mean ± SEM, n = 5–6), *p < 0.05, **p < 0.01.

To examine the suppressive mechanism of the CI-T_reg cells, the activation phenotype of the CD4 T_eff cells was analyzed. We did not see a difference in the percentage of activated CD4⁺ T cells (CD44⁺CD62L⁻) in any of the groups (data not shown). In this standard model of colitis the T cell activation associated with homeostatic expansion plays a substantial role in the differentiation to pathogenic T cells (47) which may obscure the impact of T_reg cells in the early phase of T_eff cell activation and differentiation. Nevertheless, the animals which received CI-T_reg cells had substantially less inflammatory infiltrate in the intestine tissue in contrast to the CD8 T cell control or the CD4 T_eff cell alone group. This observation suggest that the CI-T_reg cells may directly or indirectly inhibit T cell infiltration into the target tissue (48), although the exact mechanism remains to be identified. Overall, these results indicate the potential of CI-T_reg cells expanded in vitro with LNSC to control immune-mediated tissue damage.

Discussion

The existence of MHCI-restricted CD4⁺ T cells has been demonstrated at clonal and population levels in healthy humans (3) and in patients with cancer and autoimmune diseases (49–55), although the functional relevance of those cells in humans remains unknown. Furthermore, the origin of those MHC-mismatched cells in humans has not been determined, e.g., from peripheral conversion (8–11), or due to “mis-selection” by thymic MHCII (4). In this study we found that MHCII expression by LNSC promoted the conversion from the CD8 lineage to the CD4 lineage and the development of a population of regulatory CD4⁺ T cells. These findings led to the development of a tissue culture model in which co-cultures between LNSC and T cells allowed for the efficient expansion of CI-T_reg cells in vitro. Adoptive transfer of the stromal-cell expanded CI-T_reg cells from in vitro culture suppressed inflammatory tissue damage in vivo in the large intestine.

Stromal cells have traditionally been thought to provide support and nutrients for lymphocytes within the lymph node. However, an accumulation of new evidence suggests that the relationship between stromal cells and lymphocytes is much more complex (12, 21, 25). Many new roles of LNSC have been describe in recent literature; these new roles include the immunological and metabolic regulation of T cells (56). In this study, we discovered a novel mechanism in which LNSC facilitated immune regulation through the induction of CD8-to-CD4 T cell lineage conversion. This lineage conversion generated a population of regulatory CD4⁺ T cells which are MHCI-restricted as opposed to MHCII-restricted.

LNSC facilitated the generation of CI-T_reg cells through the expression of MHCII. Importantly, it was the endogenous MHCII expression by LNSC, as opposed to transcytosis (18), that was primarily responsible for driving the CD8-to-CD4 linage conversion. It should be noted, however, our results are very limited and thus does not rule out a role of MHCII transcytosis by stromal cells in long-term homeostasis of CI-T_reg cells, or their functional adaptation in a particular microenvironment. Additionally, endogenous MHCII expression by LNSC or dendritic cells and acquired MHCII by stromal or immune compartments through transcytosis or even trogocytosis (57) could have the potential to elicit a combinatory effect on CI-T_reg cells generation and function with their distinct triggers in the context.

The stromal cell population within the lymph node is comprised of multiple cell subsets. How each of these subsets plays into immunity versus tolerance remains a question (13, 18, 58–61), with the respective role of MHCII in each subset remaining to be characterized. Furthermore, it is unknown how the mismatch of CD8 versus CD4 co-receptor with MHC affects CI-T_reg cells differentiation and function. It should be noted, however, that cross-reactivity of MHC I-restricted TCR with MHCII has been documented (62–65). In addition, CI-T_reg cells might engage both MHCI and MHCII on the same stromal cells, by the MHCI-restricted TCR and CD4, respectively (66). Of note, even for thymic CD4 vs. CD8 lineage commitment, the exact roles of MHC and co-receptors remain debated (67, 68). However, further studies characterizing how MHCII and CD4 promote CI-T_reg cells generation will help design future strategies to expand human CD4 CI-T_reg cells, especially with the encouraging results shown in this study using the CI-T_reg cells expanded with LNSC in vitro to suppress inflammation in vivo in animal models of colitis. Additionally, the conversion from the CD8 lineage to CD4⁺Foxp3⁺ CI-T_reg cells may occur via CD4⁺Foxp3⁻ intermediates. Further studies are needed to understand the differentiation process of these cells in the periphery with regards to the exact role of lymph node stromal cells and what may promote Foxp3 induction in those cells (69–71).

The results from this study identified a critical role of MHCII expression by LNSC in CI-T_reg cell development. However, it is still unclear what other stromal cell factors may be contributing to this process. Stromal cells have been shown to produce many factors which have the ability to metabolically reprogram T cells. These factors include but are not limited to cytokines, chemokines, and metabolites (29, 72). Further investigation into what other stromal cell factors may be contributing to the conversion from the CD8 T cell lineage to the CD4 T cell lineage could led to the development of novel therapeutics for inflammatory diseases.

Additionally, given the distinct role of LNSC on CI-T_reg cells and CII-T_reg cells shown in the in vitro culture, investigation into the changes in metabolic profiles and phenotype of converted CD4⁺ T cells due to stromal cell interactions could help to optimize the suppressive function of both CI-T_reg cells and CII-T_reg cells.

Furthermore, heterogeneity of LNSC subsets could also account for why CI-T_reg cells originally occurred in the gut microenvironment (10) although MHCII is expressed by LNSC from other locations (18, 73). The four main subsets of LNSC discussed in this study can be further divided into subsets with varying functions. Their functions are often affected by the geographical location of the cells. For example, various LEC populations can be found within the subcapsular and medullary sinus while the FRC population can be divided into subsets based on the T cell and B cell regions of the lymph node (74, 75). It is possible that within the gut microenvironment yet-to-be-identified factors, such as microbial or other factors from intestinal drainage, may allow for site-specific functions of the LNSC (76). For instance, alterations of cytokine and chemokine secretion, as well as changes in the metabolic program of the LNSC in the gut microenvironment could help facilitate the development of CI-T_reg cells (Fig 9).

Figure 9.

A proposed model of conversion from the CD8 T cell lineage to CD4 T cells and the development of CI-T_reg cells.

Since the discovery of CD4 regulatory T cells (77), which are programed by Foxp3 (78–80), T_reg-cell-based immune therapy has been regarded with great potential in suppressing immune-mediated damage. For example, extensive studies have demonstrated the potential of CII-T_reg cells in type 1 diabetes prevention and treatment, promoted by clear evidence that in the islet, CII-T_reg cells effectively halt ongoing autoimmune damage to β cells and curtail diabetes development (48, 81–84). Nevertheless, the antigen-specificity requirement for T_reg cells to control immune destruction (48, 81, 82) remains a major challenge for T_reg cell therapies in human autoimmune disorders and other conditions of immune-mediated damage (42).

Major efforts in the field have led to success in expanding human CII-T_reg cells to billions for therapeutic trials (85, 86). The limitation on cell quantity can now be readily overcome. However, CII-T_reg cells recognized mainly exogenous antigens presented by MHCII on APC. A study showed the possibilities of engineering human CD4 T_reg cells to recognize MHCI-presented antigens by transducing MHCI-restricted TCR into CD4 T_reg cells, thus “Turning T_regs into class I suppressors” (87, 88). Such an approach has the complication of multiple types of TCR in a T cell. On the other hand, naturally generated CI-T_reg cells maintain a single TCR clonotype (10). These cells recognize MHCI. MHCI is expressed by all nucleated cells, mainly presents intracellular antigens (89) and facilitates direct recognition of target tissue cells by T cells with MHCI-restricted TCRs. Thus, although CI-T_reg cells have yet to be identified in humans, this new subset of regulatory cells may substantially broaden the spectrum of antigen specificity in T_reg cell-based therapy.

Overall, CI-T_reg and CII-T_reg cells may synergize to enforce protection against immune-mediated tissue damage. Future studies are needed to examine whether and how CI-T_reg and CII T_reg cells may synergize, or at least cooperate with additive effect, to enhance immune regulation. Moreover, it remains unclear how CI-T_reg cells are involved in human diseases. Future studies are needed to develop reagents (e.g., tetramers) and methods to isolate and expand human CI-T_reg cells, and characterize their molecular signature. Although lentiviral TCR transduction of human T_reg cells was capable of “Turning T_regs into class I suppressors” (87, 88), one can envision in vitro expanded CI-T_reg cells being tested in combination with CII-T_reg cells for immune protection, taking advantage of their diversified recognition of MHCI- and MHCII-presented antigens.

key points.

Lymph node stromal cell MHCII drives conversion of CD8 T cells to CD4 CI-T_reg cells.

Lymph node stromal cell co-culture helps generate and expand CI-T_reg cells in vitro.

CI-T_reg cells expanded in stromal cell co-culture suppress colitis in mice.

Abbreviations used in this article:

B6

C57BL/6

BEC

blood endothelial cells

CI-T_reg

MHC class I-restricted regulatory T cells

CII-T_reg

MHC class II-restricted regulatory T cells

FRC

fibroblastic reticular cells

LEC

lymphatic endothelial cells

LILP

large intestine lamina propria

LNSC

lymph node stromal cells

MFI

median fluorescent intensity

MHCI

MHC class I

MHCII

MHC class II

MLN

mesenteric lymph nodes

T_eff

effector T cells

YFP

yellow fluorescent protein