Mitochondrial One-Carbon Metabolism Drives CD34-Lineage Cells to Differentiate Into T Follicular Helper Cells to Form Tertiary Lymphoid Organs in Transplant Arteriosclerosis

Abstract

BACKGROUND:

Allograft arteriosclerosis, a significant cause of graft failure, is linked to the formation of tertiary lymphoid organs. T follicular helper (Tfh) cells are a vital subset of helper T cells that control the formation of the germinal center in tertiary lymphoid organs. Thus, understanding the origins and regulatory mechanisms of Tfh cells in allograft arteriosclerosis is essential for developing targeted therapies.

METHODS:

We used a lineage-tracing strategy to track Tfh cell fate in mouse models. Single-cell RNA sequencing, flow cytometry, and immunofluorescence staining were employed to analyze cell populations in remodeled arteries 2 and 4 weeks after transplantation. Additionally, we used VEGFR-3 inhibitors and lymph node dissection to suppress lymphatic vessel formation. Metabolic signatures and flux in different cell types were investigated using ultrahigh-performance liquid chromatography and high-resolution mass spectrometry–based metabolomics. CD4⁺ T cell–specific MTHFD2 knockout mice were used to corroborate our hypothesis about the role of mitochondrial one-carbon metabolism in Tfh cell differentiation. Mechanisms discovered in vivo were also tested ex vivo.

RESULTS:

CD34-lineage cells were found to be the major source of cells differentiating into T cell populations in allograft arteries. CD34-lineage cells mainly originated from the thymus, with drainage through lymphatic vessels, and differentiated into effective T cells around grafting arteries. Using CD34 lineage-tracing mice and single-cell RNA sequencing, we identified a Tfh cell population derived from CD34-lineage CD4⁺ T cells. Untargeted and targeted metabolomics revealed distinct upregulation of one-carbon metabolism during CD4⁺ T-to-Tfh cell differentiation. Supplementation of amino acids essential for one-carbon metabolism, such as serine, methionine or glycine, facilitated differentiation from CD4⁺ T to Tfh cells. Using deuterium-labeled serine, we found that the mitochondrial one-carbon pathway is predominant. Inhibition of the mitochondrial one-carbon metabolic enzyme MTHFD2 by administration of DS18561882 or generating CD4⁺ T cell–specific MTHFD2 knockout mice, significantly inhibited the numbers of Tfh cells and tertiary lymphoid organ formation as well as vascular remodeling.

CONCLUSIONS:

This study provides insights into the critical role of mitochondrial one-carbon metabolism and MTHFD2 in governing the differentiation of CD34-lineage cells into Tfh cells, which contributes to tertiary lymphoid organ formation in transplant vasculopathy, offering potential therapeutic targets to enhance transplant outcomes.

Clinical Perspective.

What Is New?

• CD34-lineage cells are a major contributor to T follicular helper (Tfh) cell populations and drive the formation of tertiary lymphoid organs and vascular remodeling in allograft arteries.

• CD34-lineage Tfh cells involved in vascular remodeling primarily originate from the thymus and are transported via lymphatic networks.

• Targeting the mitochondrial one-carbon metabolic enzyme MTHFD2 hinders CD4⁺ T-to-Tfh cell differentiation, which confers therapeutic benefits in transplant vasculopathy.

What Are the Clinical Implications?

• CD34-lineage derived Tfh cells drive tertiary lymphoid organ formation and vascular remodeling in allograft arteries, making them a potential target for preventing transplant vasculopathy.

• Inhibiting the MTHFD2-dependent mitochondrial one-carbon metabolic pathway in CD4⁺ T cells offers a novel strategy to suppress Tfh cell differentiation and mitigate chronic allograft rejection.

Organ transplantation has emerged as a life-saving intervention for individuals with end-stage organ failure.¹ Transplant-related arteriosclerosis, characterized by concentric and diffuse narrowing of the vascular lumen,² represents a significant contributor to graft failure.³ For instance, cardiac allograft vasculopathy is a primary cause of post–heart transplantation graft failure, with incidence increasing over time (29.3% at 5 years, 47.4% at 10 years). Therefore, mitigating allograft arteriosclerosis is crucial for improving transplant outcomes.

Accumulating evidence has shown that tertiary lymphoid organs (TLOs), atypical lymphoid annexes, aggregate during chronic inflammation in vasculopathy.^4–7 Artery TLOs, discovered by Habenicht et al in arterial vessels during atherosclerosis,⁸ have been identified in various vascular diseases. Our work recently revealed that TLOs emerge in atherosclerosis, abdominal aortic aneurysms, intimal hyperplasia (IH), and isograft and allograft vessels based on integrated single-cell RNA sequencing (scRNA-seq) databases and immunological staining.⁴ However, mechanisms that support TLO formation in different models of vasculopathy are heterogeneous, and the specific pathophysiological impact of TLOs remains inconsistent. TLOs may exert varied effects, either promoting or inhibiting disease progression, depending on the immune environment and specific conditions.^4,9 In allograft arteriosclerosis, TLOs are detrimental by activating proinflammatory effectors and promoting autoantibody production.^10,11 Habenicht and colleagues provided evidence that there are interactions between arterial TLOs and the autonomous nervous system and that these can contribute to atherogenesis.⁵ However, previous work by the authors suggested that TLOs protect against atherosclerosis.⁶ Endeavors using in TLOs revealed that TLOs harbor large numbers of naïve and effector T cell subsets with various activities at different stages of differentiation.⁸ Modulating the balance between the effective and regulatory subsets of T cells in TLOs potentially limits vasculopathy.⁸

T follicular helper (Tfh) cells are a vital subset of helper T cells that control the formation of germinal centers (GCs) in TLOs and drive the maturation of long‐lived, high‐affinity B cells into memory B cells and plasmacytes to produce immunoglobulin G.⁸ Thus, controlling Tfh cell activation and differentiation is a critical checkpoint to avoid B cell overactivation and the development of autoimmune pathologies in chronic inflammatory responses. Here, we focused on the origin and regulatory mechanisms of CD4⁺ T cell to Tfh cell differentiation in TLOs during the progression of allograft arteriosclerosis. Understanding the distinct and key molecular targets that impact the development of Tfh cells holds promise for offering novel strategies for mitigating TLO-associated autoimmune and vascular complications in transplanted organs.

METHODS

The data, protocols, and study materials will be made available to other researchers for purposes of reproducing the results or replicating the procedures. Detailed methods and data are available in the Supplemental Material.

Animals

All animal procedures in this study were approved by the Animal Care and Use Committee of Central South University. More detailed information on the breeding conditions and treatments in mouse strains is available in the Supplemental Material.

Statistical Analysis

Statistical analyses were performed with a standard unpaired Student t test for comparisons between 2 groups and with an ANOVA followed by a Tukey test for comparisons between >2 groups. Differences with P<0.05 were considered statistically significant. All data are presented as the mean±SD. All statistical analyses were performed using SPSS. The statistical graphs were generated by GraphPad Prism 8 or R software.

RESULTS

CD34-Lineage Cells Differentiate Into T Cells and Enhance TLO Formation in Allografts

Our previous studies have shown that TLO participates in the pathological process of allograft arteriosclerosis.^4,10 To investigate the function of CD34-lineage cells in TLOs and vascular remodeling, we generated Cd34-CreER^T2; Rosa26-tdTomato mice to receive aortic arteries from BALB/c mice (Figure 1A; Figure S1A through S1C). Graft tissues were collected 4 weeks after transplantation. Hematoxylin-eosin staining showed that the graft artery developed significant IH and highly organized structures in perivascular areas 4 weeks after transplantation (Figure 1B). Three-dimensional fluorescence imaging revealed that a large number of tdTomato⁺ CD34-lineage cells infiltrated into the graft artery, taking up 53.7% of the surface area of the vascular adventitia (Figure 1C; Figure S1D). However, tdTomato⁺ cells were not evenly distributed, with concentrations in some areas and sparsity in others (Figure 1C and 1D). Moreover, immune fluorescence staining demonstrated that these organized structures were lymphoid-enriched compartments with aggregated CD3⁺ T cells (green) and B220⁺ B cells (purple) (Figure 1D). Notably, a large proportion of CD3⁺ T cells was colocalized with tdTomato⁺ (red) cells, indicating that ectopic lymphoid structures partly consisted of CD34-lineage T cells (Figure 1D and 1E).

Figure 1.

CD34-lineage cells differentiate into T cells and enhance tertiary lymphoid organ formation in allograft artery. A, Experimental scheme depicting the tamoxifen-induced CD34-lineage cell labeling, allograft murine model, and tissue harvesting after 4 weeks of allograft. B, Representative hematoxylin-eosin images of mouse arteries, showing a healthy artery (AG0w) and a 4-week remodeled artery after allograft (AG4w). C, Whole-mount bright-field (left) and fluorescence (tdTomato; right) views of a mouse remodeled artery 0 week and 4 weeks after allograft. D, Representative three-dimensional whole-mount immunofluorescence (IF) views of a remodeled mouse artery at baseline and 4 weeks after allograft (green, CD3⁺ T cells; red, tdTomato⁺; purple: B220⁺ B cells). Left, An artery at baseline. Middle, A periarterial nonstereotyped lymph node. Right, Tertiary lymphoid organs (TLOs). E, Representative images showing the distinct populations of CD34-lineage T cells (tdTomato⁺CD3⁺) and B cells (B220⁺) via IF staining in the indicated groups. F, Experimental scheme depicting tamoxifen induced CD34-lineage cell labeling, allograft murine model, tissue harvesting labeled cell isolation for single-cell RNA sequencing, whole tissue for IF staining, FC , and cytometry by time-of-flight assay. G, Uniform manifold approximation and projection plot displaying the major tdTomato⁺ cell types and 9 color-coded cell clusters in the 2-week (AG2w) and AG4w remodeled arteries after allograft. There were 7248 cells (including monocyte-derived macrophages, resident-like macrophages, T cells, B cells, mesenchymal stem cells, dendritic cells, plasmacytoid dendritic cells, natural killer cells, and endothelial cells) in the AG2w group and 8923 cells in the AG4w group. H, Bar charts displaying the proportion of each cell type in overall tdTomato⁺ vascular cell populations in the AG2w and AG4w groups, as determined by single-cell RNA sequencing. I, Quantification of the ratio of CD3⁺ T cells and tdTomato⁺CD3⁺ T cells (top) and F4/80⁺ macrophages and tdTomato⁺F4/80⁺ macrophages (bottom) among total cells by FC in the indicated groups (n=5 per group). J, Representative images (left) and quantification (right) showing the distinct populations of T cells (CD3⁺) or macrophages (F4/80⁺) and CD34-lineage T cells (tdTomato⁺CD3⁺) or macrophages (tdTomato⁺F4/80⁺) in the indicated groups by IF staining. K, Experimental scheme depicting CD34-lineage cell ablation by diphtheria toxin intraperitoneal injection for Cd34-CreER^T2;Rosa26-tdTomato-iDTR mice 4 weeks before allograft and tissue harvesting 4 weeks after allograft. L, Representative hematoxylin-eosin images (left) and quantification (right) of a mouse AG4w remodeled artery of WT and Cd34-CreER^T2;Rosa26-tdTomato-iDTR mice. M, Representative three-dimensional whole-mount IF (red, tdTomato; green, LYVE-1) views (left) of a mouse AG4w remodeled artery of wild-type and Cd34-CreER^T2;Rosa26-tdTomato-iDTR mice and quantification of TLO counts (middle) and LYVE-1⁺ area (right) via three-dimensional whole-mount scanning in the indicated groups (n=11 mice per group). N, Scatterplot showing the Pearson correlation coefficients between the TLO counts and tdTomato⁺ area (detected by three-dimensional whole-mount scanning; n=11). O, Schematic illustrating the involvement of CD34-lineage cells in formation of TLOs during vascular remodeling in allogeneic transplantation. CD34-lineage cells differentiate into TLO-forming cells, contributing to the formation of TLOs and vascular remodeling. Data shown are mean±SD (I, J, L, and M). For L and M, normality of the data was analyzed by D’Agostino-Pearson omnibus test, and data that passed the normality tests were tested with standard unpaired Student t test with Welch’s correction. For I and J, data were first analyzed for normality test by a Shapiro-Wilk test, followed by 1-way ANOVA with Tukey test. A Pearson correlation analysis was employed to evaluate the linear relationship between the 2 continuous variables for N. P<0.05 was considered to be statistically significant. A indicates adventitia; Cytof, cytometry by time-of-flight; DC, dendritic cell; DT, diphtheria toxin; EC, endothelial cell; FC, flow cytometry; I, intima; M, media; Mø, macrophage; Mono, monocyte; MSC, mesenchymal stem cell; NK, natural killer; PALN, para-aortic lymph node; pDC, plasmacytoid dendritic cell; Res-like, resident-like; scRNA-seq, single-cell RNA sequencing; Tam, tamoxifen; UMAP, uniform manifold approximation and projection; and WT, wild-type.

We conducted morphological and immunohistochemical analyses to further distinguish true TLOs from para-aortic lymph nodes.⁶ High-resolution three-dimensional fluorescence imaging and collagen III connective tissue staining confirmed that TLOs are nonencapsulated lymphoid aggregates within the adventitia. Para-aortic lymph nodes are encapsulated by collagen III connective tissue capsules and located farther from the artery (Figure S1E and S1F). The absence of a collagen capsule is a defining feature of TLOs, as illustrated in Figure S1F. In addition, TLO numbers positively correlated with IH (r=0.6428, P=0.0030), whereas para-aortic lymph nodes s showed no such correlation (r=0.2758, P=0.4530) (Figure S1G). This highlights the unique contribution of TLOs to vascular remodeling.

To obtain a comprehensive understanding of the role of CD34-lineage cells in allograft arteriosclerosis, we performed scRNA-seq of isolated tdTomato⁺ CD34-lineage cells and flow cytometry (FC) and cytometry by time-of-flight for single-cell suspension of whole allograft arteries 2 and 4 weeks after transplantation (Figure 1F). The data from scRNA-seq revealed that tdTomato⁺ cells mainly differentiated into myeloid cell populations 2 weeks after transplantation, whereas >80% of tdTomato⁺ CD34-lineage cells differentiated into T cells at 4 weeks (Figure 1G and 1H; Figure S1H and S1I). FC, cytometry by time-of-flight, and immunofluorescence (IF) staining analysis also showed consistent results (Figure 1I and 1J; Figure S1J through S1M). tdTomato⁺ CD34-derived T cells account for up to 83% of total CD3⁺ T cells 4 weeks after vessel transplantation, as evidenced by FC, cytometry by time-of-flight, and IF (Figure 1I and 1J; Figure S1J through S1M). Because CD34-lineage cells contribute to a remarkable proportion of immune cells, we further generated Cd34-CreER^T2; Rosa26-tdTomato-iDTR mice to explore the effect of CD34-lineage cell depletion on TLO formation and vascular remodeling (Figure 1K; Figure S1N and S1O). Deletion of the CD34-lineage derived cell population significantly reduced IH (Figure 1L). Furthermore, the depletion of CD34-lineage cells distinctly suppressed the formation of TLOs, though it had no discernible effect on the formation of lymphatic vessels (LYVE-1⁺ area) (Figure 1M). Notably, the quantity of TLOs in allograft vessels exhibited a positive correlation with the area occupied by tdTomato⁺ CD34-lineage cells (r=0.7552, P=0.0072) (Figure 1N). These findings provide valuable evidence showing that CD34-lineage T cells orchestrate TLO formation and allograft arteriosclerosis (Figure 1O).

Graft CD34-Lineage T Cells Mainly Originate From the Thymus and Are Drained by a Lymphatic Network

We next sought to address the origin of CD34-lineage T cells involved in allograft arteriosclerosis. First, we determined whether CD34-lineage T cells originated from the recipient or donor. We observed that >90% of CD3⁺ T cells were costained with CD34 in Cd34-CreER^T2; Rosa26-tdTomato mice that received an artery from BALB/c mice (Figure S2A through S2H). However, CD34-lineage cells progressively diminished in the BALB/c model that received an artery from Cd34-CreER^T2; Rosa26-tdTomato mice (Figure S2A through S2H). Next, we investigated whether recipient CD34-lineage T cells originated from recipient bone marrow or non–bone marrow tissues. Chimeric mice were created by transferring bone marrow cells from Cd34-CreER^T2; Rosa26-tdTomato mice into irradiated wild-type C57BL/6J mice, followed by allograft transplantation (Figure S3A). In mice with tdTomato-labeled bone marrow, only 4.5 % of CD3⁺ T cells of the arteries colocalized with tdTomato⁺ cells (Figure S3B through S3H). These results indicate that the CD34-lineage T cells participating in allograft arteriosclerosis predominantly originate from non–bone marrow tissues (Figure S3I).

Because the thymus is the primary organ responsible for T cell development, we further tested whether CD34-lineage T cells derived from the thymus tissue.¹² First, we observed that thymus weight and thymus weight/body weight significantly increased by 4 weeks after allograft surgery (Figure S4A). The number of total thymocytes and the proportions of intrathymic early T cell progenitors (CD4⁻CD8⁻CD25⁻CD44⁺CD117⁺),¹³ CD3⁺ T cells, CD4⁻CD8⁻ double-negative T cells, CD4⁺CD8⁺ double-positive T (DPT) cells, CD4⁺ T cells, and CD8⁺ T cells among total thymocytes increased significantly compared with the baseline controls (Figure 2A; Figure S4B). In the thymus, the percentages of CD34-lineage early T cell progenitors, CD3⁺ T cells, double-negative T cells, DPT cells, CD4⁺ T cells, and CD8⁺ T cells increased to 3.9%, 68.6%, 6.2%, 33.7%, 21.4%, and 8.6%, respectively, 4 weeks after allograft surgery (Figure 2A; Figure S4B). Almost no early T cell progenitors were detected in lymph nodes (LNs) (Figure 2B). The evidence showed that the thymus was activated after allograft and participated in CD34-lineage T cell development.

Figure 2.

Graft CD34-lineage T cells mainly originate from the thymus and are drained by a lymphatic network. A, The proportion of early T cell progenitors (CD4⁻CD8⁻CD25⁻CD44⁺CD117⁺) among total thymocytes in the indicated groups via FC. B, The proportion of early T cell progenitors among total lymph node (LN) cells via FC. C, Schematic of the experiment design. Cd34-CreER^T2;Rosa26-tdTomato mice were administered tamoxifen for 3 weeks, followed by transplantation of allogeneic grafts and thymectomy (allogeneic graft + thymectomy) or sham surgery (allogeneic graft + sham). Samples were collected 4 weeks after surgery for FC analysis (n=6 per group). D, Quantification of T cell subsets within an allograft artery in the indicated groups (n=3 per group). E, Representative immunofluorescence (IF) images of allografted arteries in the indicated groups stained for CD3 (green), B220 (grey), tdTomato (red), and 4’,6-diamidino-2-phenylindole (DAPI; blue) to assess T cell localization in tertiary lymphoid organs. F, Quantification of tdTomato⁺CD3⁺ T cells (left) in total arteries via IF staining (n≥30 fields per group from 6 mice, each spot represents the average of one mouse), quantification of tertiary lymphoid organ counts (middle) in allograft arteries via IF staining, and intima-to-media ratio (right) in allograft arteries in the indicated groups. G, Schematic illustrating the study design for LN dissection with or without lymphangiogenesis blockage in the allograft model. I, C57BL/6J mice underwent LN dissection and served as allograft recipients for vascular transplantation. The VEGFR-3 inhibitor was continuously administered, and tissues were harvested 4 weeks after vascular transplantation. II, C57BL/6J mice underwent LN dissection and served as allograft recipients for vascular transplantation, with tissues harvested 4 weeks after vascular transplantation. III, C57BL/6J mice served as allograft recipients for vascular transplantation, with tissues harvested 4 weeks after vascular transplantation. H, Quantification of the tertiary lymphoid organ counts in the indicated groups (n=6). I, Quantification of the ratio of CD3⁺ cells among total cells of a remodeled artery via IF staining in the indicated groups (n≥30 fields per group from 6 mice, each spot represents the average of one mouse). J, Schematic illustrating that Cd34-CreER^T2;Rosa26-tdTomato mice with tdTomato fluorescence-labeled cells were used as receptors for bone marrow transplantation; bone marrow cells were cleared by irradiation, and C57BL/6J mice served as donors. The chimeric mice served as recipients for vascular transplantation. Throughout the procedure, tamoxifen was maintained to fluorescently label CD34-lineage cells, and the VEGFR-3 inhibitor was administrated to suppress lymphangiogenesis, with tissue harvested 4 weeks after transplantation for analysis. K, The proportion of early T cell progenitors (CD4⁻CD8⁻CD25⁻CD44⁺CD117⁺) among total thymocytes in the indicated groups via FC. L, Quantification of the ratio of tdTomato⁺ cells among total cells of a remodeled artery via FC in the indicated groups (n=6 per group). M, Representative images showing the distinct populations of CD34-lineage cells (tdTomato⁺), T cells (CD3⁺), and B cells (B220⁺) via IF staining in the indicated groups. N, Quantification of the T cells (top) or B cells (bottom) among total cells of a remodeled artery via IF staining in the indicated groups (n≥30 fields per group from 6 mice, each spot represents the average of one mouse). O, Schematic illustrating that CD34-lineage cells from LNs, transported by lymph vessels, contributed to vascular remodeling. Data shown are mean±SD (A, B, D, F, H, I, K, L, and N). For A, B, D, F, and K, normality of the data was analyzed by D’Agostino-Pearson omnibus test, and data that passed the normality tests were tested with standard unpaired Student t test with Welch’s correction. For H, I, L, and N, data were first analyzed for normality test by a Shapiro-Wilk test, followed by 1-way ANOVA with Tukey test. P<0.05 was considered to be statistically significant. A indicates adventitia; AG, allogeneic graft; BM, bone marrow; BMT, bone marrow transplantation; ETP, early T cell progenitor; FC, flow cytometry; I, intima; LN, lymph node; LND, lymph node dissection; M, media; PBMC, peripheral blood mononuclear cell; Tam, tamoxifen; Thy, thymocyte; TLO, tertiary lymphoid organ; and VEGFR, vascular endothelial growth factor receptor.

Then, we performed thymectomy in mouse models to test whether T cells in TLOs can still develop and function normally. Cd34-CreER^T2; Rosa26-tdTomato mice were treated with tamoxifen 3 weeks before BALB/c aortic graft transplantation. Thymectomy was performed at the time of grafting. Four weeks after allograft surgery, grafts were harvested, and changes in T lymphocytes before and after thymectomy were analyzed using FC and IF (Figure 2C). After thymectomy, T cell proportions in the allograft artery, including CD3⁺ T cells, double-negative T cells, DPT cells, CD4⁺ T cells, and CD8⁺ T cells decreased by >60% and CD34-lineage T cells were diminished (Figure 2D). CD34-lineage T cell subsets were also significantly decreased in the LNs of the allograft mouse model with thymectomy (Figure S4C). CD3⁺ T cell infiltration and allograft-associated TLOs were significantly reduced in mice with thymectomy (Figure 2E and 2F). Allograft-induced arterial remodeling, as measured by the intima-to-media (I/M) ratio, was also suppressed by thymectomy (Figure 2F; Figure S4D).

As previous reports emphasized the significance of lymphatic vessels in transporting immune cells to the vascular transplant,¹⁴ we suppressed lymphatic vessel formation using a VEGFR-3 inhibitor in the allograft model (Figure S4E). Three-dimensional fluorescence imaging showed that the VEGFR-3 inhibitor led to a reduction of approximately 85% in the LYVE-1⁺ lymphatic vessel area in the adventitia of the allograft vessels (Figure S4F and S4G). Consistent with this, IF staining and FC analyses also showed that VEGFR-3 inhibition led to a substantial decrease of LYVE-1⁺ cells and tdTomato⁺ cells by 90% 4 weeks after transplantation (Figure S4H through S4K). Furthermore, IF staining showed that inhibition of lymphatic vessels significantly reduced the proportion of CD3⁺ T cells and B220⁺ B cells in the allograft vessels by >90% (Figure S4L and S4M). Consistent with these results, >75% of the T cells participating in transplant-associated arteriosclerosis colocalized with tdTomato⁺ cells, whereas B cells did not colocalize with tdTomato⁺ signaling (Figure S4L and S4M). In parallel, the ratio of tdTomato⁺ cell area in graft vessels and severity of IH decreased by 90% and 70%, respectively, with lymphatic vessel inhibition (Figure S4N through S4Q). A combination of LN dissection and blocking lymphangiogenesis was conducted (Figure 2G). TLO formation and IH in allograft vessels were blocked by LN dissection or combined LN dissection and VEGFR-3 inhibitor (Figure 2H and 2I; Figure S5A through S5C). Moreover, thymus weight and the proportions of CD34-lineage T cell subsets, including double-negative T cells, DPT cells, CD4⁺ T cells, and CD8⁺ T cells, were not significantly changed in the thymus of the allograft mouse model after LN dissection (Figure S5D and S5E). The evidence suggests that the accumulation of graft CD34-lineage cells for T cell differentiation requires a lymphatic network, which contributes to the progress of allograft arteriosclerosis.

To eliminate the potential impact of circulating tdTomato⁺ CD34-lineage cells on T cell accumulation after transplantation, we further transferred bone marrow cells from wild-type C57BL/6J mice to irradiated Cd34-CreER^T2; Rosa26-tdTomato mice (Figure 2J). Four weeks after bone marrow transplantation, bone marrow and peripheral blood mononuclear cells were reconstituted by cells without tdTomato labeling (Figure S5F through S5H). Then, allograft transplantation was carried out in such chimeric mice and followed by VEGFR-3 inhibitor administration (Figure 2J). The proportions of tdTomato⁺ cells in both bone marrow and peripheral blood mononuclear cells were limited to less than 2% 4 weeks after bone marrow transplantation (Figure S5F and S5G), whereas in LNs, it remained about 22% (Figure S5H). FC also showed reduced tdTomato⁺ cell proportions in the bone marrow and peripheral blood mononuclear cells in the chimeric mice with VEGFR-3 inhibitor (Figure S5I). This confirms the successful elimination of tdTomato labeling on CD34-lineage cells only in the circulation. Increased early T cell progenitors in the thymus after allograft in such chimeric mice were not tdTomato⁺ (Figure 2K). Four weeks after allograft transplantation, FC and IF analyses demonstrated active participation of tdTomato⁺ CD34-lineage cells in transplanted arteries (Figure 2L through 2N). IF staining revealed the presence of approximately 40% T cells in the adventitia, with >75% of them being tdTomato⁺ cells, which are noncirculating cells, whereas B cells in grafts were not CD34-lineage derived (Figure 2M and 2N). The increase in CD34-lineage T cells was markedly suppressed by blocking lymphangiogenesis (Figure 2M and 2N). These findings strongly support the notion that CD34-lineage T cells from the thymus involved in transplantation atherosclerosis were primarily transported by lymphatic vessels.

In sum, these experiments suggest that graft CD34-lineage T cells mainly originate from the thymus and are drained by a lymphatic network (Figure 2O).

CD34-Lineage Cells Give Rise to Effector T Cells and Are Essential for TLO Formation

The results so far provide evidence supporting a significant role of CD34-lineage T cells in TLO formation and vascular remodeling. Using scRNA-seq analysis of tdTomato⁺ CD34 cells, we found that tdTomato⁺ CD34 cells give rise to DPT cells and effector T cell populations, including CD4⁺ T cells, CD8⁺ T cells, toxicity T cells, T helper 17 (Th17) cells, naïve T cells, Tfh cells, regulatory T (Treg) cells, and memory T cells. Two weeks after allograft transplantation, double-positive T cells and CD4⁺ T cells were the dominant cell types, and T cells further differentiated into various effector subsets 4 weeks after transplantation (Figure 3A and 3B).

Figure 3.

CD34-lineage cells give rise to effector T cells and are essential for tertiary lymphoid organ formation. A, Uniform manifold approximation and projection plot displaying the major natural killer and T cell subtypes (including CD4⁻CD8⁻ T, CD4⁺CD8⁺ T, CD4⁺ T, CD8⁺ toxicity T, naïve T, memory T, regulatory T, T helper 17, and T follicular helper [Tfh] cells) and 10 color-coded cell clusters in the 2-week (AG2w) and 4-week (AG4w) remodeled artery after allograft. There were 357 cells in the AG2w group and 7688 cells in the AG4w group. B, Bar plots showing the percentage distribution of various T cell subsets in arteries of indicated groups analyzed by single-cell RNA sequencing (left) and cytometry by time-of-flight (right). The subsets include CD4⁻CD8⁻ T, CD4⁺CD8⁺ T, CD4⁺ T, CD8⁺ toxicity T, natural killer, naïve T, memory T, regulatory T, T helper 1, T helper 17, and Tfh cells. C, Quantification of the ratio of ICOS⁺ cells in total tdTomato⁺CD3⁺ cells by FC (left) and cytometry by time-of-flight (right) in the indicated groups (n=5 per group). D, Representative FC plots (left) and quantification (right), showing the percentage of tdTomato⁺ and tdTomato⁻ cells among total Tfh cells (CD4⁺ICOS⁺) of a mouse AG4w remodeled artery (n=6). E, Quantification of the percentage of tdTomato⁺ and tdTomato⁻ cells in memory T, regulatory T, and T helper 17 cells of a mouse AG4w remodeled artery by cytometry by time-of-flight (n=6). F, Representative FAC plots (left) and quantification (right), showing the percentage of tdTomato⁺ cells in total germinal center B cells (B220⁺GL7⁺) of a mouse AG4w remodeled artery (n=6). G, quantification of the percentage of tdTomato⁺ and tdTomato⁻ cells in memory B, plasma B, and follicular B cells of a mouse AG4w remodeled artery by cytometry by time-of-flight (n=6). H, Heatmap showing the interaction score of T cell subtypes (columns) and B cell subtypes (rows), as determined by single-cell RNA sequencing. I, Quantification of the ratio of Tfh (left) and germinal center B (right) cells among total cells of remodeled arteries via FC in the indicated groups (n=5 per group). J, Quantification of the ratio of memory T, regulatory T and T helper 17 cells among total CD3⁺ cells of remodeled arteries via FC in the indicated groups (n=5 per group). K, Quantification of the ratio of memory B, plasma B, and follicular B cells among total B220⁺ cells of remodeled arteries via FC in the indicated groups (n=5 per group). L, Schematic illustrating the conclusion that CD34-lineage CD4⁺T cells differentiate into Tfh cells, which then interact with B cells to promote lymphocyte recruitment, proliferation, and tertiary lymphoid organ formation. Data shown are mean±SD (C through G and I through K). For D through G and I through K, normality of the data was analyzed by D’Agostino-Pearson omnibus test, and data that passed the normality tests were tested with standard unpaired Student t test with Welch’s correction. For C, data were first analyzed for normality test by a Shapiro-Wilk test, followed by 1-way ANOVA with Tukey test. P<0.05 was considered to be statistically significant. APB indicates antigen-presenting B cell; CTL, toxicity T cell; Cytof, cytometry by time-of-flight; DNT, double-negative T cell; DPT, double-positive T cell; DZ B, dark zone B cell; FC, flow cytometry; GC, germinal center; NK, natural killer; LZ B, light zone B cell Th1, T helper 1; Th17, T helper 17Tm, memory T; Tn, naïve T; Treg, regulatory T; TLO, tertiary lymphoid organ; and UMAP, uniform manifold approximation and projection.

Among these T cell subsets, Tfh cells are essential for TLO formation by interacting with B cells to promote the development of a GC.⁵ FC and cytometry by time-of-flight analysis showed that Tfh cells amount to 4.93% of tdTomato⁺ CD34-lineage T cells (Figure 3C). More than 90% of Tfh cells originated from tdTomato⁺ CD34-lineage T cells (Figure 3D). tdTomato⁺ CD34-lineage T cells also contribute to the memory T, Treg, and Th17 cell generation after transplantation (Figure 3E). However, GC B cells were not derived from the CD34-lineage (Figure 3F and 3G). Combining scRNA-seq analysis of tdTomato⁺ CD34 cells and previous whole-allograft artery cells from GSE140812.¹⁰ The analysis of cell interactions revealed a pronounced interplay between tdTomato⁺ CD34-lineage Tfh cells and various B cell subgroups within vascular grafts (vertical), particularly with GC B cells, including dark zone B and light zone B cells (Figure 3H). These results suggest that CD34-lineage Tfh cells are closely involved in TLO formation and allograft arteriosclerosis. Consistent with this, depletion of CD34-lineage cells resulted in reduced Tfh cells (Figure 3I) and other T cells, including memory T, Treg, and Th17 cells (Figure 3J) in the grafts. Because the generation of GCs requires close interaction of Tfh and B cells, GC B cells, memory B cells, plasma B cells, and follicular B cells in grafts were largely reduced upon CD34-lineage cell depletion (Figure 3I and 3K). These findings demonstrate that CD34-lineage T cells give rise to effector T cells, especially Tfh cells, and are essential for B cell interaction and TLO formation in allograft arteriosclerosis (Figure 3L).

One-Carbon Metabolism Is Crucial for CD34-Lineage CD4⁺T Cell Differentiation into Tfh Cells

Because Tfh cells interact with B cells determining GC formation, we sought to identify the key mechanism that differentiates CD34-lineage Tfh cells from other T cell subsets. First, we employed pseudotime analysis and RNA velocity to investigate the differentiation paths in CD34-lineage T cell subtypes (Figure 4A through 4C). Generally, we observed that CD34-lineage T cells followed a pattern of differentiation from CD4⁻CD8⁻ T cells to CD4⁺CD8⁺ T cells and then to effective T cells in the vascular graft (Figure 4A and 4C). Immature T cells (state 1, including CD4⁻CD8⁻ T and CD4⁺CD8⁺ T cells) differentiate into CD4⁺ T cells (state 2, including CD4⁺ T, Treg, Tfh, and naïve T cells) and CD8⁺ T cells (state 3, including memory T and CD8⁺ toxicity T cells) (Figure 4A, 4B, and 4D). Using pseudotime analysis, we compared the differentially expressed genes in CD4⁻CD8⁻ T cells and CD4⁺CD8⁺ cells during their differentiation toward CD4⁺ or CD8⁺ T cells. From state 1 to state 2 or state 3, there is an initial downregulation of Rag1, followed by a sustained decrease in the expression levels of the early markers Ccr9 and Cd24a. Additionally, there is transient expression of the T cell receptor-driven activation genes Cd69, Egr1, Nr4a1, and Itm2a during this transition and late upregulation of the maturation markers Klf2, S1pr1, and Sell (Figure 4B and 4H).¹⁵ From state 1 to state 2, Cd4 and interferon production genes (Il12rb2 and Jak2), costimulatory molecular genes (Icos and Cd28), and early exhaustion markers (Ctla4 and Pdcd1) were upregulated (Figure 4B, 4D, 4E, and 4H). From state 1 to state 3, Cd8 and cytotoxicity genes (Prf1, Ifng, Nkg7, Gzmb, Gzma, and Tnfsf10), costimulatory molecular genes (Tnfrsf25 and Cd226), and the tumor necrosis factor receptor superfamily gene Cd27 were upregulated (Figure 4B, 4D, 4E, and 4H).¹⁶ Compared with Treg or Th17 cells, CD4⁺ T cells during differentiation into Tfh cells showed initial downregulation of Rag1, Ltb, Ccr7, and Sell and notable upregulation of T cell activation genes (Cxcr3 and Tnfrsf4) and one-carbon metabolism–related genes (Cth, Shmt1, and Shmt2) (Figure 4F, 4G, and 4I).

Figure 4.

One-carbon metabolism is crucial for CD34-lineage CD4⁺T cell differentiation into T follicular helper cells. A, Monocle pseudotime trajectory of all CD34-lineage T cells. Cells are labeled by pseudotime (left) and cell subtypes (right). B, Heatmap of the significantly changed genes (P<0.01) discovered by the BEAM function from Monocle in branchpoint 1. C, RNA velocity analysis distinguished velocity vectors across the pseudotime axis. D, Monocle pseudotime trajectory of CD34-lineage CD4⁻CD8⁻ T, CD4⁺CD8⁺ T, CD4⁺ T, and CD8⁺ T cells of a mouse 2-week remodeled artery after allograft. Cells are labeled by pseudotime (left) and cell subtypes (right). E, monocle pseudotime trajectory of CD34-lineage CD4⁻CD8⁻ T, CD4⁺CD8⁺ T, CD4⁺ T, and CD8⁺ T cells of a mouse 4-week remodeled artery after allograft. Cells are labeled by pseudotime (left) and cell subtypes (right). F, Monocle pseudotime trajectory of CD34-lineage CD4⁺ T, regulatory T, T follicular helper, and T helper 17 cells of a mouse 2-week remodeled artery after allograft. Cells are labeled by pseudotime (left) and cell subtypes (right). G, Monocle pseudotime trajectory of CD34-lineage CD4⁺ T, regulatory T, T follicular helper, and T helper 17 cells of a mouse 4-week remodeled artery after allograft. Cells are labeled by pseudotime (left) and cell subtypes (right). H, Volcano plots show the genes with different expression between CD4⁺ T and CD8⁺ T cells of a mouse 2-week (left) and 4-week (right) remodeled artery after allograft. Differentially expressed genes (DEGs) are highlighted in red or blue and were determined by Wilcoxon rank-sum test. I, Volcano plots showing the genes with different expression between CD4⁺ T and T follicular helper cells. DEGs are highlighted in red or blue and were determined by Wilcoxon rank-sum test. J, Heatmap summary of the expression of select metabolic pathway activation–related genes (rows) in the indicated cell types (columns), as determined by single-cell RNA sequencing. K, Heatmap summary of the expression of select metabolic pathway activation–related genes (rows) in the indicated T cell subtypes (columns), as determined by single-cell RNA sequencing. L, Schematic illustrating the conclusion that several metabolic pathways are activated during differentiation of CD34-lineage CD4⁺ T cells into T follicular helper cells. Th17 indicates T helper 17; Tm, memory T; Tn, naïve T; and Treg, regulatory T.

Compass analysis also showed that these metabolic pathways are highly activated during differentiation from the CD34⁺ progenitor state to the effective T cell state, such as CD4⁺ T cells and Tfh cells (Figure 4J). It is noteworthy that, when comparing the gene expression in other effector CD4⁺ T subgroups, such as CD4⁺ T cells, Th17 cells, and Treg cells, Tfh cells have particularly high expression of one-carbon metabolism–related genes (Figure 4K). During differentiation of CD34-lineage T cells into Tfh cells, various metabolic pathways (e.g., anaerobic glycolysis, carbohydrate-active metabolism, and phosphoglyceride metabolism) were gradually activated, with one-carbon metabolism being the most prominent (Figure 4J and 4K). The significant activation of one-carbon metabolism also distinguished CD34-lineage Tfh cells from other CD4⁺ T cell subtypes in terms of metabolic pathways (Figure 4K). In summary, these results suggest that metabolic regulations are essential for CD34-lineage T cell differentiation, particularly one-carbon metabolism, and highly involved in differentiation of CD34-lineage T cells into Tfh cells during allograft arteriosclerosis (Figure 4L).

Mitochondrial One-Carbon Metabolism Drives CD34-Lineage CD4⁺ T Cells to Tfh Cell Differentiation

To further elucidate metabolic pathways instructing CD4⁺ T cells to differentiate into Tfh cells, we performed broad-spectrum metabolomics profiling on OT-II mouse primary CD4⁺ T cells and their differentiated Tfh cells. The differentiation rate of CD4⁺ T cells into Tfh cells (identified as CXCR5⁺ICOS⁺ cells) was approximately 85% by FC (Figure S6A and S6B). Reverse transcription polymerase reaction further confirmed that Tfh cells generated in our induction system exhibited high expression of Cxcr5, Il6st, and Bcl6 (Figure S6C), distinguishing them from Th1 cells with elevated expression of Ifng and Tbx21, Th2 cells with expression of Il4 and Gata3, and Th17 cells with Il17a and Il22 expression (Figure S6D). Ultrahigh-performance liquid chromatography/high-resolution mass spectrometry–based metabolomics showed that the metabolic pattern of Tfh cells differs significantly from that of CD4⁺ T cells (Figure 5A), and approximately 75% of metabolic pathways were significantly enhanced in Tfh cells (Figure 5B). Notably, metabolites related to cell proliferation (pattern 1: histidine metabolism, and so forth), energy metabolism (pattern 2: tricarboxylic acid cycle, glycolysis, and so forth), and amino acid and one-carbon metabolism (pattern 3: tyrosine , cysteine, folate, methionine metabolism, and so forth) were significantly upregulated (Figure 5C). In contrast, metabolites related to arachidonic acid metabolism (pattern 4) and lipid metabolism (pattern 5) were significantly downregulated in Tfh cells (Figure 5D). By integrated analysis, the top 3 (with the highest pathway impact score) metabolic pathways most significantly impacting Tfh cell differentiation were related to one-carbon metabolism (Figure 5E).

Figure 5.

Mitochondrial one-carbon metabolism drives CD34-lineage CD4⁺ T cells to T follicular helper cell differentiation. A, Principal-component analysis score plot demonstrating metabolic differences between CD4⁺ T cells and T follicular helper (Tfh) cells via ultrahigh-performance liquid chromatography/high-resolution mass spectrometry UHPLC-HRMS -based metabolomics. B, Heatmap displaying all differential metabolites between CD4⁺ T cells and Tfh cells. Each row represents a differential metabolite, and each column represents a sample, as determined by metabolomics. C, Heatmap illustrating the correlation between metabolites with higher concentrations in Tfh cells based on Pearson correlation analysis, with each row and column representing a differential metabolite, as determined by metabolomics. Pathway summaries for metabolites in different patterns are provided on the right of the heatmap. D, Heatmap illustrating the correlation between metabolites with lower concentrations in Tfh cells, as determined by metabolomics. E, Scatterplot summarizing the pathways enriched with metabolites exhibiting higher concentrations in Tfh cells via metabolomics. F, Bar plot illustrating the concentrations of different amino acids in CD4⁺ T cells and Tfh cells, as determined by ultrahigh-performance liquid chromatography/high-resolution mass spectrometry–based untargeted metabolomics (n=4 per group). G, Violin plot displaying the expression of transporter-related genes for different substances (including glucose, fatty acids, amino acids, folic acid, and choline; top) and amino transporter–related genes for different amino acids (bottom) in CD4⁺ T cells and Tfh cells, as determined by single-cell RNA sequencing. H, Schematic of one-carbon metabolism with amino acid uptake–related enzymes highlighted (top) and heatmap depicting the expression of amino acid uptake–related enzyme genes (bottom) in the indicated cell types, as determined by single-cell RNA sequencing. I, Bar plot illustrating the concentrations of metabolites in CD4⁺ T cells and Tfh cells, as determined by targeted metabolomics (n=4 per group). J, Schematic of the ex vivo experiments designed to demonstrate the amino acid that is vital for the differentiation of CD4⁺ T cells into Tfh cells. K, Quantification of the percentage of Tfh cells detected via FC in the indicated groups (n=4 per group). L, Schematic of one-carbon metabolism and violin plot showing expression of metabolic enzyme–related genes in CD4⁺ T cells and Tfh cells based on single-cell RNA sequencing. M, Dot plot depicting the log2(fold change) values of expression for various metabolic enzyme–related genes in one-carbon metabolism (fold change was determined by the ratio of Tfh cells and CD4⁺ T cells), colored according toy cytoplasmic and mitochondrial pathways. N, Barplot depicting the log2(fold change) values of expression for metabolic enzyme–related genes in one-carbon metabolism between cytoplasmic and mitochondrial pathways. O, Metabolic tracing strategy using D³-serine to differentiate flux through the mitochondrial versus the cytosolic arm of one-carbon metabolism by monitoring the labeling pattern of thymidylate. P, Tfh cells produce predominantly the m+1 isotopomer of deoxythymidine monophosphate (dTMP), indicative of mitochondrial rather than cytosolic flux. Q, Cytoplasmic/mitochondrial flux via computational metabolic flux analysis. Data shown are mean±SD (F, I, K, and N). For F, I, and N, normality of the data was analyzed by D’Agostino-Pearson omnibus test, and data that passed the normality tests were tested with standard unpaired Student t test with Welch’s correction. For K, data were first analyzed for normality test by a Shapiro-Wilk test, followed by 1-way ANOVA with Tukey test. P<0.05 was considered to be statistically significant. Ala indicates alanine; Arg, arginine; CBS, cystathionine beta-synthase; CTH, cystathionine gamma-lyase; Cys, cysteine; DHF, dihydrofolate; DHFR, dihydrofolate reductase; FC, flow cytometry; Gln, glutamine; Gly, glycine; HCY, homocysteine; IL, interleukin; Leu, leucine; Met, methionine; MTHFD, methylenetetrahydrofolate dehydrogenase; MTR, methionine synthase; SAH, S-adenosylhomocysteine; SAM, S-adenosyl methionine; Ser, serine; SHMT, serine hydroxymethyltransferase; THF, tetrahydrofolate; and UHPLC-HRMS, ultrahigh-performance liquid chromatography/high-resolution mass spectrometry.

Maintaining adequate amino acid levels, particularly methionine and serine, is crucial for supporting one-carbon metabolism.¹⁷ Metabolic profiling revealed that the levels of methionine, serine, and histidine were significantly higher in Tfh cells than in CD4⁺ T cells (Figure 5F). Transcriptomics analysis also indicated significant increases in genes related to amino acid metabolism (Figure 5G). In line with the observations in metabonomics analysis, the gene expressions of system SAT 1 (A1 amino acid transporter 1), ASCT2 (alanine, serine, cysteine–preferring transporter 2), LAT1 (L-type amino acid transporter 1), and CAT-1 (cationic amino acid transporter 1) were upregulated, including Slc38a1, Slc1a5, Slc7a5, Slc3a2, Slc7a1, and Slc7a11 (Figure 5G).

Furthermore, combining amino acid uptake and the expression of intracellular enzymes, we found that the genes of glycine, methionine, and serine uptake–related one-carbon metabolism pathways (GNMT and SARDH) were highly expressed in Tfh cells (Figure 5H).¹⁸ Consistent with this, targeted metabolic profiling of one-carbon metabolism revealed significant activation of one-carbon metabolism in Tfh cells, with notable accumulation of methionine, glycine, serine, and histidine in Tfh cells (Figure 5I). Increased histidine in T cells is a critical factor in cell proliferation and differentiation.¹⁹ Therefore, Tfh cells actively take up histidine, which can be seen in both transcriptomic and metabolomic results. These findings indicated that one-carbon metabolism, particularly involving methionine and serine as substrates, plays a pivotal regulatory role in the differentiation of CD4⁺ T cells into Tfh cells, which was further validated through cell starvation experiments (Figure 5J, 5K; Figure S6E). Compared with methionine and serine starvation, glycine starvation has a weaker effect on Tfh differentiation, likely because glycine can be synthesized intracellularly from methionine, histidine, and threonine (Figure 5K; Figure S6E).¹⁸

scRNA-seq analysis showed that the key one-carbon metabolic enzymes were remarkably upregulated during CD4⁺ T-to-Tfh cell differentiation (Figure 5L). Notably, CD4⁺ T-to-Tfh cell differentiation appears to depend primarily on mitochondrial one-carbon flux, as evidenced by higher expression of mitochondrial one-carbon enzyme genes compared to their cytosolic counterparts (Figure 5L through 5N). Mitochondrial Mthfd2 in particular exhibited an 8-fold increase in Tfh cells compared with that in CD4⁺ T cells (Figure 5M). To confirm the relative contribution of the cytosolic versus mitochondrial one-carbon metabolism to Tfh cell differentiation, we performed isotope tracing in primary CD4⁺ T cells with stimulation. Primary CD4⁺ T cells were fed (2,3,3-²H₃)-serine, and the incorporation of deuterium labeling in synthesized thymidine triphosphate was monitored (Figure 5O). Thymidine with 2 deuterium atoms (M+2) is synthesized via the cytosolic folate cycle and with 1 deuterium (M+1) via the mitochondrial path (Figure 5P). We found that the shift in the contribution of main mitochondrial one-carbon metabolism to pyrimidine biosynthesis is associated with primary CD4⁺ T cell-to-Tfh cell differentiation (synthesized thymidine triphosphate M+1/M+2>1) (Figure 5Q).

Targeting the Mitochondrial Enzyme MTHFD2 Confers Therapeutic Benefits in Transplant Vasculopathy

Considering the predominant contribution of mitochondrial one-carbon flux to the differentiation of CD4⁺ T cells to Tfh cells, we tested whether blocking a mitochondrial one-carbon metabolic enzyme, MTHFD2, hinders Tfh cell differentiation and compared its effects with cytoplasmic one-carbon metabolic inhibition. Ovalbumin-specific CD4⁺ T cells and CD34-lineage CD4⁺ T (tdTomato⁺) cells were isolated from OT-II mice and Cd34-CreER^T2; R26-tdTomato mice, respectively. Isolated CD4⁺ T cells were cultured with irradiated splenic antigen-presenting cells and stimulated to generate Tfh cells (Figure 6A). DS18561882 (100 μM), an MTHFD2 inhibitor, reduced Tfh differentiation by 85% in CD4⁺ T cells from OT-II mice (Figure 6B) and tdTomato⁺CD4⁺ T cells from Cd34-CreER^T2; R26-tdTomato mice (Figure 6C) to Tfh cells differentiation. In contrast to MTHFD2 inhibitor, carolacton (50 μM, MTHFD1 inhibitor) reduced Tfh differentiation by <50% in CD4⁺ T cells (Figure 6B) and tdTomato+CD4⁺ T cells (Figure 6C). The production of interleukin-21 was inhibited by 94.4% with MTHFD2 inhibition and by 22.2% with MTHFD1 inhibition (Figure 6D). Inhibition of MTHFD2 also suppressed Tfh cell–induced B cell activation and immunoglobulin G production by >90% (Figure 6E and 6F). MTHFD1 suppressed Tfh cell–induced B cell activation by 38.9% and immunoglobulin G production by 17.6% (Figure 6E and 6F). Therefore, we conclude that mitochondrial MTHFD2 is essential for CD34-lineage Tfh cell differentiation in vitro.

Figure 6.

Targeting the mitochondrial enzyme MTHFD2 confers therapeutic benefits in transplant vasculopathy. A, Schematic of the ex vivo experiments designed to demonstrate that MTHFD1 and MTHFD2 mediate differentiation of CD4⁺ T cells into T follicular helper cells. B, Quantification of the percentage of CXCR5⁺ICOS⁺ cells among total CD4⁺ cells via FC in the indicated groups isolated from OT-II mice (n=4 per group). C, Quantification of the percentage of CXCR5⁺ICOS⁺ cells among total CD4⁺ cells via FC in the indicated groups isolated from Cd34-CreER^T2;Rosa26-tdTomato mice (n=4 per group). D, Quantification of the percentage of interleukin-21⁺ cells among total CD4⁺ cells via FC in the indicated groups (n=4 per group). E, Schematic of the ex vivo experiments designed to demonstrate that T follicular helper cells activate B cells. F, FC analysis showing percentage of CD38⁺ cells among CD19⁺ B cells (left) and relative immunoglobulin G levels (right) in the indicated groups (n=5 per group). G, Schematic of the in vivo experiments designed to demonstrate that MTHFD1 and MTHFD2 inhibitors suppress vascular remodeling and tertiary lymphoid organ formation. H, Quantification of the intima/media ratio (top) and lumen area (bottom) in the indicated groups (n≥30 fields per group from 6 mice, each spot represents the average of one mouse). I, Representative images showing the populations of CD34-lineage cells (tdTomato⁺), T cells (CD3⁺), and B cells (B220⁺) of a mouse 4-week remodeled artery after allograft in the indicated groups. J, Quantification of ratio of tdTomato⁺ cell among total cells via immunofluorescence staining in the indicated groups. K, Quantification of the ratio of CD3⁺ cells among total cells via immunofluorescence staining in the indicated groups. L, Quantification of the ratio of B220⁺ cells among total cells via immunofluorescence staining in the indicated groups (J through L, n≥30 fields per group from 6 mice, each spot represents the average of one mouse). M, Quantification of tertiary lymphoid organ counts via pathological section and immunofluorescence staining in the indicated groups (n≥30 levels/slides per group from 6 mice, each spot represents the average of one mouse). N, Scatterplot showing the Pearson correlation coefficients between the ratio of CD3⁺tdTomato⁺ cells and intima/media ratio (detected by pathological section and immunofluorescence staining; n=18). O, Strategy for MTHFD2 knockout specifically in CD4⁺ cells in CD4⁺ T cell-specific MTHFD2 conditional knockout mice (CD4⁺ T-Mthfd2-KO; top) and schematic of the in vivo experiments designed to demonstrate that MTHFD2 conditional knockout in CD4⁺ T cells suppresses vascular remodeling (bottom). P, Quantification of the I/M ratio (top) and lumen area (bottom) in the indicated groups (n≥30 fields per group from 6 mice, each spot represents the average of one mouse). Q, Representative images showing the populations of T cells (CD3⁺) and B cells (B220⁺) of a mouse 4-week remodeled artery after allograft in the indicated groups. R, Quantification of the ratio of CD3⁺ cells among total cells via immunofluorescence staining in the indicated groups. S, Quantification of the ratio of B220⁺ cells among total cells via immunofluorescence staining in the indicated groups (n≥30 fields per group from 6 mice in R and S, each spot represents the average of one mouse). T, Quantification of tertiary lymphoid organ counts via pathological section and immunofluorescence staining in the indicated groups (n≥30 levels/slides per group from 6–7 mice, each spot represents the average of one mouse). Data shown are mean±SD (B through D, F, H, J through M, P, and R through T). Data were first analyzed for normality test by a Shapiro-Wilk test, followed by 1-way ANOVA with Tukey test. A Pearson correlation analysis was employed to evaluate the linear relationship between the two continuous variables for N. P<0.05 was considered to be statistically significant. A indicates adventitia; FC, flow cytometry; I, intima; IL-21, interleukin-21; M, media; MTHFD, methylenetetrahydrofolate dehydrogenase; Tam, tamoxifen; and TLO, tertiary lymphoid organ.

We further verified the predominant effect of MTHFD2 inhibition in vivo (Figure 6G). Following allograft transplantation, mice were randomly assigned to 3 treatment groups: (1) vehicle injection, (2) the MTHFD1 inhibitor carolacton (50 mg/kg), or (3) the MTHFD2 inhibitor DS18561882 (100 mg/kg). They were placed on a 5-day-per-week schedule until sample harvest. Inhibition of MTHFD2 reduced the ratio of I/M ratio and luminal area by approximately 50% (Figure 6H; Figure S7A). Simultaneously, IF staining revealed a pronounced decrease of tdTomato⁺ CD34-lineage T cell (yellow) and B220⁺ B cell (grey) infiltration by 81.1% and 89.5%, respectively (Figure 6I through 6L), and 100% suppression of TLO formation (Figure 6M) within graft vessels. The ratio of CD3⁺tdTomato⁺ cells was correlated positively with the vascular I/M ratio, which could be suppressed by DS18561882 (Figure 6N). However, treatment with the MTHFD1 inhibitor carolacton resulted in only a 24.7% reduction in the vascular I/M ratio (Figure 6H; Figure S7A). IF staining revealed the carolacton reduced infiltration by only 35.8% for tdTomato⁺ CD34-lineage T cells and 36.8% for B220⁺ B cells (Figure 6I through 6L). Nevertheless, approximately 50% of the TLOs still formed with carolacton (Figure 6M). Thus, mitochondrial MTHFD2 plays a critical role in regulating TLO formation and allograft vasculopathy.

Consistent with this, conditional knockout of Mthfd2 in CD4⁺ T cells (CD4⁺ T-Mthfd2-KO mice) also reduced the number of CD4⁺CXCR5⁺ICOS⁺ Tfh cells (Figure S7B) and the I/M ratio and increased the luminal area by approximately 70% (Figure 6O and 6P; Figure S7C; Tables S1 and S2). Notably, IF staining revealed a marked inhibition of CD3⁺ T cell (green) and B220⁺ B cell (red) infiltration (Figure 6Q through 6S; Figure S7D) as well as TLO formation (Figure 6T) within graft vessels. In sum, this evidence revealed that blocking the mitochondrial one-carbon metabolism pathway by MTHFD2 hinders CD34-lineage CD4⁺ T cell-to-Tfh cell differentiation and TLO formation, blunting arterial atherosclerosis after transplantation (Figure S8).

DISCUSSION

Tfh cells are a class of helper T cells that control the formation of GCs in TLOs and have a pivotal role in modulating local immunity. In this study, we elucidated CD34-lineage cells originating from the thymus and transported by the lymphatic network, which is the major source of Tfh cells around grafting arteries. Metabolomics analysis revealed that mitochondrial one-carbon metabolism drives CD34-lineage CD4⁺ T-to-Tfh cell differentiation (Figure S8). Targeting the mitochondrial enzyme MTHFD2 of one-carbon metabolism revealed its essential role as a controller that inhibits TLO formation and local immunity targeting allograft vessels, providing mechanistic insight and a potential therapeutic strategy for allograft arteriosclerosis.

TLOs are specialized ectopic lymphoid structures that develop at sites of chronic inflammation or tissue damage, serving as dynamic hubs for immune cell interaction and modulation.^20,21 Such localized immune surveillance facilitates efficient immune responses tailored to the specific microenvironment, contributing to tissue homeostasis and repair.²¹ Although our understanding of the formation and function of TLOs has improved, their role as mediators of protective or pathological immune responses in cancer,^21–23 chronic infection, autoimmune diseases,^20,24 and transplant rejection^25–27 remains enigmatic. For example, TLOs can play a protective role during infection with Mycobacterium tuberculosis as well as fungal and viral pathogens in the lungs. TLOs have also been implicated in facilitating pathological responses during respiratory syncytial virus infections by inducing hypersecretion of mucus, exacerbating inflammation, and leading to asthma exacerbation.²⁸ The role of TLOs in promoting pro- or antitumoral immune responses is disease specific and depends on their cellular composition and location. For instance, TLOs enriched in Treg cells have been associated with a worse prognosis by suppressing tumor-specific T cell immunity, which promotes ovarian carcinoma growth in humans.²⁹ However, TLOs enriched in dendritic cells and CD8⁺ T cells are associated with improved survival in patients with non–small cell lung cancer and rectal cancer.^30,31 The recent discovery of TLOs within vascular tissues has unveiled profound implications for vascular diseases, including atherosclerosis, abdominal aortic aneurysm, IH, isografts, and allografts.^4,6,32 The correlation between immune responses and clinical deterioration rates in various vascular pathologies,³³ along with the association of TLO structural cells with the inflammatory response, underscores the clinical translational value of targeting TLOs to mitigate vascular disease progression and reduce restenosis rates in transplant and atherosclerotic settings.⁴

Tfh cells and their interaction with B cells are vital for GC generation in TLO.³⁴ This study reveals CD34-lineage Tfh cells drive allograft-induced atherosclerosis and reveals their progenitor cells originate predominantly in the thymus, as demonstrated by lineage-tracing technology. Transferring tdTomato-labeled bone marrow cells into the allograft model confirmed that CD34⁺ progenitor cells in the thymus were not replenished from the bone marrow after transplantation. Our thymectomy experiments provide evidence that these T cells are thymus derived. However, this approach does not clarify whether thymic CD34⁺ progenitor cells are thymus residential or originate from hematopoietic stem cells that migrated to the thymus during embryonic development. Given the role of early hematopoietic progenitors in thymopoiesis, future studies are needed to trace their developmental trajectory. Additionally, studies have identified a secondary thymus, a smaller ectopic thymic tissue that may contribute to T cell development.³⁵ Its precise role in T cell homeostasis remains unclear; therefore, although our findings primarily emphasize the essential role of the primary thymus in generating T cells after allograft, future studies should consider the potential involvement of the secondary thymus in T cell development and vascular infiltration. Because the current study focused on the mechanism regulating CD4⁺ T cell-to-Tfh cell differentiation, the origin of CD34 progenitor cells in the thymus warrants investigation in future in-depth studies.

Using multiple single-cell trajectory analyses, Tfh cells are derived from CD4⁺ T cells in the CD34-lineage. Therefore, understanding the differentiation mechanism of CD4⁺ T cells into Tfh cells is paramount for discovering specific targets for halting Tfh differentiation and GC formation. Previous studies have delineated that Tfh cell differentiation requires CD4⁺ T cells receiving TCR stimulation in concert with costimulation signals and cytokines.^36,37 Such stimulation initiates the expression of Bcl-6 in CD4⁺ T cells, promoting the expression of the hallmark Tfh cell molecules CXCR5, CXCR4, and PD-1.^38–40 In contrast, the transcription factor Blimp-1, an antagonist of Bcl6, inhibits Tfh differentiation and GC generation.³⁹ Other mechanisms regulating Tfh differentiation have been reported in intratumoral TLOs. Transforming growth factor-β–mediated silencing of the genomic organizer SATB1 promotes Tfh cell differentiation and the formation of intratumoral TLOs.⁴¹ Whereas initial studies focused on the immune receptors and transcriptional regulators involved in T cell quiescence and activation, recent findings have highlighted cell metabolism as a crucial regulator of these processes.^42,43 One-carbon metabolism supports multiple physiological processes, including purine and thymidine biosynthesis, amino acid homeostasis, epigenetic maintenance, and redox defense.¹⁸ Studies have reported that one-carbon metabolism engages in T cell receptor stimulation and activation,^44,45 supports effective T cell expansion,⁴⁶ and facilitates Th17 cell differentiation.⁴⁷ In addition to the transcription factor Bcl6 in CD4⁺ T cells being necessary for Tfh cell differentiation, we revealed that metabolic modulation is robustly engaged in Tfh differentiation, specifically for one-carbon metabolism in vessel remodeling. Using in vivo conditional knockouts and pharmaceutical inhibition to block the mitochondrial one-carbon metabolism enzyme MTHFD2, we uncovered significant protection in transplant vasculopathy. MTHFD2 is upregulated during several highly proliferative processes, such as embryogenesis, hematopoiesis, and angiogenesis, but it has little to no expression in most adult tissues.⁴⁸ A previous study revealed that mitochondrial MTHFD2 is a metabolic checkpoint of T cell activation during heart transplant rejection,⁴⁹ which is partly consistent with our discovery. Our study identified that CD34-lineage CD4⁺ T-cell differentiation into Tfh cells was driven by mitochondrial one-carbon metabolism, offering potential for targeted therapeutic intervention. Interestingly, Sugiura et al reported that depletion of MTHFD2 in CD4⁺ T cells leads to the enhancement of Treg cell responses and suppression of Th17 cell responses in multiple autoimmune conditions.⁴⁷ In allograft arteries, although Th17 cells also showed elevated one-carbon metabolism, compared with Th17 and other T cell subtypes, Tfh cells have the highest MTHFD2 expression and one-carbon metabolism. This may explain why Tfh cells are particularly susceptible to MTHFD2 depletion. Whether CD4⁺ T-MTHFD2 deletion in the allograft atherosclerosis model also modulates Treg/Th17 cell imbalance and local immune response in TLOs warrants further investigation.

Limitations of the Study

One potential limitation of this study is that we did not fully assess the contribution of the secondary thymus to the generation of T cells.³⁵ Because we did not specifically evaluate the presence or activity of the secondary thymus during allograft, future studies may consider addressing this aspect through targeted investigations. Although one-carbon metabolism displays a distinct role in modulating Tfh cell differentiation, we also observed multiple metabolic mechanisms robustly involved in this process. More definitive strategies need to be developed to manipulate T cell subpopulations and better characterize the effect of T cell subtype deficiencies on disease pathogenesis. The interrelated metabolic mechanism underlying cell phenotypic transition requires further investigation. Future studies should employ in vivo tracking with sequential immunophenotyping to fully delineate the differentiation pathway from CD34⁺ progenitors to CD4⁺ T cells and subsequently to Tfh cells. This study was conducted using male recipient mice. Although this design minimizes hormonal variability within groups, it also limits the generalizability of our findings. Future studies including both male and female recipients will be important to determine whether MTHFD2-dependent Tfh differentiation exhibits sex-specific differences. Although our findings provide critical mechanistic insights into allograft-associated vascular complications, translating these discoveries into clinical therapies necessitates further investigation.

ARTICLE INFORMATION

Acknowledgments

X.S. performed and validated the major experiments; analyzed the single-cell RNA sequencing, metabolomics, and mass cytometry data; organized figures, and wrote the initial draft. J.W. performed staining experiments and bioinformatics and statistical analyses, and organized revised figures. T.H collected samples for single-cell RNA sequencing and mass cytometry and helped with the mouse bone marrow transplantation model. M.Y. and L.Q. performed the in vitro experiments and helped with animal experiments. C.W. conducted pathological experiments and isolated single cells from the aortic grafts. L.P. performed the in vitro experiments and constructed the mouse model of allograft and lymph node dissection. Q. Xiao, Y.L., and H.Y. provided data interpretation, constructive suggestions, and material support. Q. Xu and J.C. conceived the idea, designed projects, performed a critical review of the manuscript, supervised the study and approved it for publication, and provided funding for the study.

Sources of Funding

This work was supported by the Noncommunicable Chronic Diseases-National Science and Technology Major Project (2023ZD0503300 to J.C.), the National Natural Science Foundation of China (32330046 to Q. Xu, 82170436 to J.C., and 82400499 to J.W.), a Hunan Natural Science Foundation key grant (2024JJ3043 to J.C.), the Science and Technology Innovation Program of Hunan Province (2022RC3073 to J.C. and 2023RC3088 to J.W.), the Jiangxi Key Research and Development Program (20243BCC31005 to J.C.), the Hainan Province Key Research and Development Program (ZDYF2024SHFZ040 to J.C.), and the Independent Exploration and Innovation Project for Graduate Students of Central South University (1053320210374 to X.S.). This work was supported in part by the High-Performance Computing Center of Central South University.

Disclosures

None.

Supplemental Material

Methods

Tables S1–S5

Figures S1–S8

ARRIVE Checklist for Animal Studies

References 50–68

Supplementary Material

cir-152-537-s001.pdf ^{(10.2MB, pdf)}

cir-152-537-s002.pdf ^{(98.2KB, pdf)}