CD4⁺CD25⁺ regulatory T cells circumstantially initiate transplant rejection through CXCL15/neutrophil axis

Abstract

Regulatory T cells (Tregs) are critical for immune tolerance. However, direct Treg therapies generate limited results. Recently, combinational treatment with Tregs and IL-2 to expand Tregs has become a hotspot and appeared to be promising. Here we unexpectedly find that infusion of CD4⁺CD25⁺ Tregs plus IL-2 in lymphocyte-deficient mice causes skin and heart allograft rejection, while IL-2 or Treg alone does not. Depletion of CD25⁺ Tregs in recipient mice receiving Tregs plus IL-2 abrogates allograft rejection. Gene profiling via qPCR arrays shows a tremendous increase in CXCL15 expression in allografts after the joint treatment. Its expression is augmented in Tregs from recipient mice. IL-2 augments CXCL15/IL-8 production by murine or human Tregs. Moreover, neutrophil infiltration is increased in allografts of mice receiving Tregs plus IL-2. However, allograft rejection or neutrophil infiltration does not occur when transferred Tregs lack CXCL15. Finally, the joint treatment provokes cardiac allograft rejection in pre-tolerized wild-type mice, while IL-2 or Treg alone does not. These results unveil hidden and detrimental effects of Tregs on transplant survival and somewhat challenge current dogma that Tregs are always immunosuppressive.

A Combined therapy of Tregs and IL-2 unexpectedly triggered allograft rejection in mice through CXCL15-mediated neutrophil infiltration, revealing a potentially harmful effect of Tregs on allograft survival.

Introduction

Regulatory T (Treg) cell therapy has shown considerable promise for inducing transplantation tolerance¹. CD4⁺CD25⁺Foxp3⁺ Tregs control immunological homeostasis and inhibit the responsiveness of effector T cells and other immune cells, resulting in self-tolerance². Removal of Treg cells in wild-type mice led to autoimmune diseases, whereas the expansion of endogenous Treg cells or infusion of exogenous Treg cells prohibited autoimmunity and transplant rejection^3,4. Therefore, some research centers have undertaken early-phase clinical trials for assessing both the safety and efficacy of Treg therapy in organ transplantation^5,6. However, Key issues to be addressed include the tissue distribution, stability, longevity, and function of adoptively transferred Tregs⁷.

Since Tregs require IL-2 for their homeostasis, survival, and suppressive activity^8,9, tremendous efforts have been placed on the expansion of endogenous/exogenous Tregs or the restoration of their function under immunosuppressive conditions using unmodified or engineered IL-2, which has recently become a hotspot. Simultaneous administration of Tregs and low-dose IL-2 or its equivalent was shown to enhance suppressive effects of Tregs on autoimmunity^9–11, while recombinant IL-2, IL-2 mutein, or IL-2R-engineered Tregs promoted transplant tolerance^12,13 or restored Treg function that otherwise would be suppressed by a calcineurin inhibitor^14,15. However, treatment with low-dose IL-2 might not always induce allograft tolerance, even if it could expand circulating Tregs¹⁶.

In the present study, originally based on a control group of transplant experiments, we unexpectedly found that infusion of Tregs and administration of IL-2 in Rag1^-/- recipient mice resulted in skin and heart allograft rejection. Our data demonstrated that allograft rejection under these conditions was associated with increased CXCL15 expression and infiltration of neutrophils in allografts. qPCR arrays showed a tremendous increase in CXCL15 gene expression, while in vitro experiments also proved that IL-2 stimulated Tregs to produce CXCL15. Furthermore, either neutrophil infiltration or allograft rejection mediated by Tregs plus IL-2 was mostly dependent on their expression of CXCL15. The joint treatment with Tregs plus IL-2 also induced cardiac allograft rejection in wild-type mice already tolerized by DST plus anti-CD154 Ab, whereas depletion of neutrophils abolished this effect. These results indicate that under certain circumstances, Treg therapy may be detrimental to an allograft.

Results

Infusion of CD4⁺CD25⁺ Tregs plus administration of IL-2 in Rag1^-/- mice results in skin allograft rejection

IL-2 is essential for the function and homeostasis of CD4⁺CD25⁺Foxp3⁺ Tregs^17,18. During the exploration of optimal strategies for Treg suppression of allograft rejection, we originally used Treg infusion in combination with IL-2 as control experiments. We unexpectedly found that infusion of CD4⁺CD25⁺ Tregs (2.5 ×10⁶) isolated from naive mice plus administration of recombinant IL-2 caused over 80% skin allograft rejection in Rag1 knockout (Rag1^-/-) mice lacking lymphocytes, as shown in Fig. 1A. Much less allograft grafts were rejected when 1.5 ×10⁶ Tregs plus IL-2 were used, while neither 0.75 ×10⁶ Tregs plus IL-2 nor 2.5 ×10⁶ Tregs alone rejected a skin allograft. Treatment with IL-2 alone (up to 2.5 ×10⁴ IU) failed to induce allograft rejection (Fig. 1A). However, 2.5 ×10⁶ Tregs plus ultra-low dose of IL-2 (0.4 ×10⁴ IU) caused the rejection in a small fraction of recipients, while the same number of Tregs plus 2.5 ×10⁴ IU of IL-2 resulted in the rejection in nine out of ten recipients. H&E staining demonstrated a significant increase in cellular infiltration in skin grafts from Rag1^-/- mice treated with Tregs (2.5 ×10⁶) and IL-2 compared to those treated with PBS (Control) or Tregs alone (Fig. 1B). Meanwhile, the immunofluorescence staining revealed higher expression of FoxP3 in skin grafts of groups treated with Tregs and IL-2 than with Tregs alone (Fig. 1C). In order to determine the purity of Foxp3⁺ Tregs and CD4⁺CD25⁺ cells before the cell transfer, flow cytometric analysis was performed and it was found that the purity of CD4⁺CD25⁺ Tregs was over 97% while that of FoxP3⁺ cells reached more than 96% (Fig. 1D). To examine the stability of transferred CD4⁺ Tregs, the percentages of FoxP3⁺ cells within CD45⁺CD4⁺ population in the lymph node (LN) and spleen were determined through flow cytometric analyses. As shown in Fig. 1E, around 95% of CD4⁺ T cells were confirmed to be FoxP3⁺ Tregs in both LNs and spleen of recipient mice treated with Tregs and IL-2, whereas the percentages of FoxP3⁺ Tregs in recipients treated with Tregs alone were around 75%. The absolute cell counts of FoxP3⁺ Tregs in LNs and spleen were also increased by IL-2 treatment at 1, 2 or 4 weeks post-transplantation (Fig. 1F), suggesting that IL-2 indeed is important for maintaining Treg stability.

Fig. 1. Infusion of CD4⁺CD25⁺ Treg with IL-2 in Rag1^-/- mice initiates skin allograft rejection.

A Shown is skin allograft survival time. Skin from BALB/c mice was grafted onto Rag1^-/- mice, which received PBS (control), IL-2, CD4⁺CD25⁺ Tregs alone, or various numbers of Tregs plus IL-2, as indicated (n = 9-13 mice/group). B H&E staining was performed on skin graft sections three weeks post-transplantation. Shown are representatives of H&E images (original magnification X100), while the area of inflammation in a skin graft is quantified as Average Optical Density (AOD) by Image J (n = 7-8 mice/group). C Immunofluorescence staining for the expression of Foxp3 and immunofluorescence images (original magnification X200) for each group are shown. The area fraction of Foxp3 in the skin allografts is quantified (n = 9-10 mice/group). Data are presented as the mean ± SD. D Dot plots present the gating strategy and purity of isolated CD4⁺CD25⁺Foxp3⁺ cells before the adoptive transfer. Shown is one set of representative dot plots. E Dot plots present the gating strategy and percentages of Foxp3⁺ cells within CD4⁺CD45⁺ cells from the lymph node (LN) and spleen (Sp) three weeks post-transplantation. Shown is one set of representative dot plots from five recipient mice per group. F Dot plots show the absolute numbers of CD4⁺ Foxp3⁺ cells from LNs and spleen of recipient mice 1, 2 and 4 weeks after transplantation (n = 5 mice/group). Data are presented as mean ± SD.

To further confirm that allograft rejection was indeed mediated by the adoptively transferred CD4⁺ Tregs, Rag1^-/- mice were transplanted with BALB/c skin and received Tregs from diphtheria toxin receptor (DTR) positive DEREG mice, while some recipients were administered with depleting chemical diphtheria toxin (DT). It was found that administration of Tregs or DTR⁺ Tregs plus IL-2 caused skin allograft rejection in Rag1^-/- recipients, whereas depletion of Tregs with DT abolished skin allograft rejection in mice receiving DTR⁺ Tregs + IL-2, but not WT Tregs + IL-2 (Supplementary Fig. 1A). The immunofluorescence staining revealed much lower expression of Ly6G in skin grafts of recipient mice treated with DTR⁺ Tregs plus IL-2 and DT compared to those mice treated with Tregs/DTR⁺Tregs plus IL-2 without DT (Supplementary Fig. 1B), suggesting that neutrophils are largely diminished by DTR⁺ Treg depletion. On the other hand, the purity of FACS-sorted CD4⁺CD25⁺ Tregs was not 100% (around 97%). To rule out the possibility that 3% contaminated cells were CD4⁺CD25^- conventional T cells that could cause rejection, we determined how many of these T cells are required to initiate allograft rejection in Rag1^-/- recipients. We found that 1×10⁵ cells plus IL-2 (1×10⁴ IU) did not cause any rejection, while 2.5 ×10⁵ cells plus IL-2 resulted in skin allograft rejection in most of the recipients (Supplementary Fig. 2). Since 1.5×10⁶ Tregs caused many rejections (Fig. 1A), 3% contamination would mean that 4.5×10⁴ CD4⁺CD25^- T cells were possibly included in the infused Tregs. Given that even 1×10⁵ CD4⁺CD25^- T cells plus IL-2 did not cause any rejection, allograft rejection in our model was indeed mediated by Tregs, but not by possibly contaminated CD4⁺ T cells.

Treatment with Tregs plus IL-2 increases the gene expression of CXCL15 in skin allografts, while IL-2 also induces CXCL15 production by Tregs in vitro

To explore the mechanisms underlying pro-inflammatory or pro-alloimmune effects of transferred CD4⁺CD25⁺ Tregs in vivo, we assessed the gene expression profiles of skin allografts or syngeneic grafts from mice treated with Tregs plus IL-2 three weeks after transplantation through qPCR arrays in 96-well plates. Overall, the expression of 16 genes, including Cxcl15, Pf4, Il1a, and Il6st, was significantly increased in skin allografts from mice treated with Tregs plus IL-2 compared to syngeneic skin grafts (Fig. 2A). CXCL15, also known as a murine homolog of human IL-8¹⁹, exhibits a neutrophil-chemotactic activity²⁰. Since it has been previously reported that CD4⁺Foxp3⁺ Tregs express IL-8 in the context of chronic hepatitis and ulcerative colitis^21,22, we speculated that there would be a role for CXCL15 in the allograft rejection mediated by Tregs plus IL-2. We first determined if Tregs could actually produce CXCL15/IL-8 in vitro. We found that IL-2 indeed increased CXCL15 or IL-8 production by cultured murine or human CD4⁺CD25⁺ Treg cells, while higher concentrations of IL-2 promoted Tregs to produce more CXCL15/IL-8 than the low concentration (Fig. 2B, C). Furthermore, as shown in Fig. 3A, it was found that Treg infusion plus IL-2 significantly increased the percentage of CXCL15⁺ cells within CD4⁺Foxp3⁺ Tregs of recipient mice with allografts compared to those with syngeneic grafts. As skin resident cells (e.g., keratinocytes, fibroblasts) can also produce CXCL15, we analyzed whether Tregs could indirectly promote CXCL15 expression by keratinocytes and fibroblasts using flow cytometry. In the mice with syngeneic grafts, we observed that the infusion of Treg plus IL-2 did not increase CXCL15 expression in both K19⁺ keratinocytes and S100A4⁺ fibroblasts from grafts (Supplementary Fig. 3). In the mice treated with PBS or Treg plus IL-2 infusion, the percentage of CXCL15⁺K19⁺ or CXCL15⁺S100A4⁺ cells was significantly higher in mice with allogeneic transplantation than those with syngeneic transplantation. However, in the mice with allotransplantation, no statistically significant increase was observed in the percentage of CXCL15⁺K19⁺ and CXCL15⁺S100A4⁺ cells from mice receiving Treg and IL-2 compared with those receiving PBS (Supplementary Fig. 3). These results indicate that allotransplant rejection or alloimmunity rather than Treg infusion induces CXCL15 expression in keratinocytes and fibroblasts. Immunofluorescence staining of Ly6G in skin grafts also showed that the number of neutrophils in the allografts was significantly higher than that in syngeneic grafts (Fig. 3B). These results indicate that the infusion of CD4⁺CD25⁺ Tregs with administration of IL-2 induces allograft rejection probably by recruiting neutrophils through CXCL15.

Fig. 2. CXCL15 expression is increased in an allograft of mice receiving Tregs plus IL-2 and in Treg culture in the presence of IL-2.

A Skin allografts in recipient mice treated with Tregs plus IL-2 were collected three weeks post-transplantation for RNA extraction and qPCR arrays were then performed. Shown is a heatmap of the expression levels (log₁₀(2^-△Ct)) of 16 upregulated genes in skin allografts (Allo-Tx) from mice treated with Tregs plus IL-2 vs. syngeneic grafts (Ctrl-Tx). Ranges of expression values are color-coded (red to blue gradient). Blue corresponds to lower expression and red to higher expression. Mouse (B) or human (C) CD4⁺ CD25⁺ Tregs (2 × 10⁵ cells/well) were cultured in 96-well plates and stimulated with anti-CD3 and anti-CD28 Abs for 4 days. The levels of CXCL15 or IL-8 in the supernatant were detected using ELISA. One of two separate experiments is shown (IL-2 low: 500 U/ml IL-2 and IL-2 high: 1000 U/ml IL-2, n = 4-5 per group). Data are presented as mean ± SD.

Fig. 3. Treatment with Tregs plus IL-2 increases the expression of CXCL15 and induces infiltration of neutrophils in an allograft.

A Skin graft-infiltrating cells and draining LN cells from transplanted mice receiving Tregs+IL-2 were isolated three weeks post-transplantation, and percentages of CXCL15⁺ cells in CD45⁺CD4⁺Foxp3⁺ Tregs were determined using FACS analysis. Column graphs show the percentage of CXCL15⁺ cells within CD4⁺Foxp3⁺ Treg cell population in skin grafts and LNs. Data are presented as the mean ± SD (n = 6-7 mice/group, Allo-Tx: allo-transplant and Ctrl-Tx: syngeneic transplant). B Immunofluorescence staining for the expression of Ly6G three weeks post-transplantation. Representatives of fluorescence images for both groups are shown (original magnification X200). The histogram shows the area fraction of Ly6G in the skin grafts (n = 12). Data are presented as the mean ± SD from three separate experiments.

Allograft rejection initiated by Tregs/IL-2 is overwhelmingly dependent on CXCL15

Given that Tregs could produce CXCL15/IL-8 based on our present data and previous studies by others and that skin allografts highly expressed CXCL15, we then determined a role for CXCL15 in allograft rejection in recipient mice receiving CD4⁺CD25⁺ Tregs and IL-2. To do this, Rag1^-/- mice were transplanted with BALB/c skin and infused with Tregs from B6 WT or CXCL15^-/- mice in combination with the administration of IL-2. Some recipient mice received IL-2 plus Tregs derived from FasL-deficient (gld) or perforin-deficient mice. As shown in Fig. 4A, the transfer of WT Treg plus IL-2 caused skin allograft rejection, while the allograft rejection was diminished by the infusion of CXCL15-deficient Tregs. On the other hand, a lack of FasL or perforin in Tregs did not significantly change their capacity to reject a skin allograft in the presence of IL-2. Moreover, H&E and immunofluorescence staining of skin allografts showed a significant decrease in cellular infiltration (Fig. 4B) and much fewer Ly6G⁺ cells (Fig. 4C) in mice that received CXCL15^-/- Tregs plus IL-2 compared to those that received WT Tregs plus IL-2. These findings indicate that CXCL15, but not FasL or perforin, in Tregs plays a critical role in Treg-mediated allograft rejection in this animal model.

Fig. 4. Allograft rejection mediated by Tregs and IL-2 is dependent on CXCL15.

A Rag1^-/- mice were transplanted with a BALB/c skin and received PBS (Control), wild type (WT) CD4⁺CD25⁺ Treg, FasL-deficient, perforin-deficient or CXCL15-deficient Tregs. All mice receiving Tregs were also administrated with IL-2. Skin allograft survival was observed (n = 9-11 mice). B Shown are representatives of H&E staining (original magnification X100) on skin sections three weeks post-transplantation. The area of cellular infiltration in a skin allograft is quantified as AOD for the images of H&E staining (n = 6-8 mice). C Representatives of immunofluorescence staining for the expression of Ly6G (original magnification X200). Recipient mice received PBS (Control), WT Tregs with IL-2, or CXCL15-deficient Tregs with IL-2. The histogram shown is the area fraction of Ly6G in the skin allografts (n = 8-9 mice). Data are presented as the mean ± SD from three separate experiments.

Treatment with Tregs + IL-2 in Rag1^-/- mice also precipitates cardiac allograft rejection

Further investigated were the effects of Treg transfer plus IL-2 administration on cardiac allograft survival in immunodeficient mice. Rag1^-/- mice were transplanted with a BALB/c heart and infused with Tregs, Tregs plus IL-2, or CXCL15-deficient Tregs plus IL-2. As shown in Fig. 5A, adoptive transfer of CD4⁺CD25⁺ Tregs with administration of IL-2, but not transfer of Tregs alone, resulted in heart allograft rejection (MST = 32.5 days) in the Rag1^-/- mice. However, allograft rejection did not occur when recipient mice were infused with CXCL15-deficient Tregs plus IL-2 treatment. Similarly, results from H&E and IHC stainings of Ly6G showed that treatment with CXCL15^-/- Treg plus IL-2 failed to increase overall cellular and neutrophil infiltration in heart allografts of recipient mice compared to treatment with WT Treg plus IL-2 (Fig. 5B, C). In addition, Treg infusion plus administration of IL-2 augmented the percentages of CXCL15⁺ cells within transferred CD4⁺Foxp3⁺ Tregs (Fig. 5D), but not CD4⁺Foxp3^- T cells (Fig. 5E, Supplementary Fig. 4), in the draining LNs of mice receiving a heart allograft compared to a syngeneic heart graft. These data demonstrated that Tregs plus IL-2 induced infiltration of neutrophils within a cardiac allograft through upregulation of CXCL15 in Tregs, leading to cardiac allograft rejection in Rag1^-/- recipient mice.

Fig. 5. Infusion of Tregs with IL-2 in Rag1^-/- mice results in heart allograft rejection that depends on CXCL15.

A Prior to heart transplantation, Rag1^-/- mice were treated with PBS (Control), WT CD4⁺CD25⁺ Tregs, WT Tregs plus IL-2, or CXCL15^-/- Tregs plus IL-2. Heart allograft survival was analyzed with Kaplan–Meier curves (n = 7-8). B H&E staining on heart sections three weeks after transplantation. Data are presented as means ± SD (n = 12). C Immunohistochemical staining of heart allografts for Ly6G expression (n = 9). Draining LN cells from recipient mice were isolated three weeks after heart transplantation and infusion of CD4⁺CD25⁺ Tregs or CD4⁺CD25^- T cells, while the percentages of CXCL15⁺ cells after gating on CD45⁺CD4⁺Foxp3⁺ Tregs (D) or CD45⁺CD4⁺Foxp3^- T cells (E) were determined using FACS analysis (n = 5/group). Data are presented as means ± SD. One of three separate experiments is shown.

Infusion of Tregs with IL-2 provokes CXCL15-mediated rejection of cardiac allografts in WT mice already tolerized by DST and anti-CD154 Ab

To confirm if the joint treatment with Tregs plus IL-2 can also cause allograft rejection in WT mice, C57BL/6 mice were transplanted with a BALB/c heart and treated with a standard tolerance-inducing protocol: DST plus anti-CD154 Ab, as described in the methods. Recipients were then infused with Tregs plus IL-2 treatment two weeks after cardiac transplantation. As shown in Fig. 6A, infusion of Tregs plus IL-2 provoked cardiac allograft rejection in WT mice that were otherwise tolerized by DST plus anti-CD154 Ab, while treatment with IL-2 alone failed to cause allograft rejection in this mouse model. However, infusion of CXCL15-deficient Tregs plus administration of IL-2 also did not provoke allograft rejection. Furthermore, depletion of neutrophils with anti-Ly6G Ab in the recipient mice abolished allograft rejection induced by Tregs plus IL-2. On the other hand, treatment with Tregs plus IL-2 resulted in significant infiltration of neutrophils in cardiac allografts that were otherwise tolerized by DST plus anti-CD154 Ab, while use of CXCL15-deficient Tregs or WT Tregs plus depleting anti-Ly6G Ab failed to do so (Fig. 6B, C). These findings indicate that the joint treatment with Tregs plus IL-2 can cause cardiac infiltration of neutrophils and allograft rejection in immune-competent WT animals that are even pre-tolerized, and that this rejection depends on CXCL15 expression in Tregs as well as neutrophils. To further confirm a role for neutrophils in allograft rejection mediated by Tregs plus IL-2, Rag1^-/- recipient mice receiving a skin allograft and Tregs plus IL-2 were also treated with anti-Ly6G Ab to deplete neutrophils. Similar results were observed that the depletion of neutrophils by anti-Ly6G Ab prohibited skin allograft rejection mediated by Tregs plus IL-2 (Supplementary Fig. 5), suggesting that this rejection is indeed dependent on neutrophils.

Fig. 6. Infusion of Tregs with IL-2 provokes rejection of a cardiac allograft in WT mice already tolerized by DST and anti-CD154 Ab.

A Shown is cardiac allograft survival time. Recipient mice were divided into six groups: (1) control group: C57BL/6 mice with heart transplantation without other pretreatment. (2) DST+ anti-CD154 group: C57BL/6 mice receiving DST (donor-specific splenocyte transfusion) and anti-CD154 Ab. (3) DST+ anti-CD154 + IL-2 group: C57BL/6 mice with DST and anti-CD154 Ab. Then IL-2 (500 U/ml) were administrated i.p. every other day for 14 days after transplantation. (4) DST+ anti-CD154 + WT Treg+ IL-2 group: C57BL/6 mice with DST and anti-CD154 Ab. Treg cells (2.5 ×10⁶ i.v.) and IL-2 were injected into recipient mice 14 days after transplantation. (5) DST+ anti-CD154 + CXCL15^-/- Treg+ IL-2 group: C57BL/6 mice were first given DST and anti-CD154 Ab. CXCL15^-/- Treg cells (2.5 ×10⁶, i.v.) and IL-2 were injected into the recipient mice 14 days after transplantation. (6) DST+ anti-CD154 + WT Treg+ IL-2+ anti-Ly6G group: B6 mice were treated first with DST plus anti-CD154 Ab and then WT Treg+ IL-2+ anti-Ly6G (n = 7-8 mice/group). B Shown are representatives of immunohistochemical staining of heart allografts for Ly6G expression. C The area of neutrophil infiltration in a heart allograft was quantified as AOD in the images of immunohistochemical staining. Data are presented as mean ± SD of AOD (n = 9).

Discussion

CD4⁺CD25⁺ Tregs play a key role in maintaining immune homeostasis and tolerance. Since IL-2 is essential for the stability and suppressive function of the Tregs²³, simultaneous administration of Tregs and low-dose IL-2 or its equivalent/mutein was shown to enhance suppressive effects of Tregs on autoimmunity or autoimmune diseases^9–11. For instance, it was reported that low-dose recombinant IL-2 selectively modulated the abundance of Treg, follicular helper T (TFH), and Th17 cells in SLE patients²⁴. Khoryati et al. showed that an engineered IL-2 mutant promoted Treg expansion amid ongoing autoimmunity and reduced agonism of effector T cells²⁵, while Yan et al. demonstrated that IL-2/anti-IL2 Ab complexes ameliorated lupus nephritis by expanding CD4⁺Fop3⁺ Tregs in lupus-prone mice²⁶. Early-stage clinical trials were also performed to evaluate the efficacy and safety of Tregs in various autoimmune diseases²⁷. In a multicenter phase II trial, low-dose IL-2 is beneficial in active SLE²⁸. However, it was also reported that low-dose IL-2 could expand both adoptively transferred Tregs and endogenous cytotoxic T cells in T1D patients²⁹. Taken together, low-dose IL-2 has been mostly beneficial for treating autoimmune diseases by expansion of exogenous or endogenous Tregs.

On the other hand, low-dose IL-2 has previously been utilized to suppress allograft rejection or GVHD. Administration of orthogonal IL-2 significantly promoted orthogonal IL-2R⁺Foxp3⁺CD4⁺ Treg cell proliferation without expanding other T cells and facilitated the long-term survival of heart allografts¹³. Moreover, IL-2 was associated with a lower incidence of viral infections and GVHD and expansion of Tregs in vivo³⁰, while an orthogonal IL-2/IL-2R system could also be used to prevent acute GVHD³¹. Interestingly, IL-2 reversed Treg dysfunction induced by a calcineurin inhibitor and prolonged allograft survival¹⁵. Engineering of IL-2 receptor also enhanced adoptively transferred Treg function suppressed by a calcineurin inhibitor¹⁴. In the present study, we unexpectedly found that infusion of Tregs plus IL-2 caused allograft rejection in Rag1^-/- mice lacking lymphocytes. The joint measure also provoked allograft rejection in WT recipient mice that would otherwise be tolerized by the regimen of DST plus anti-CD154 Ab. Thus, our studies have unveiled an unexpected finding that the treatment with IL-2/Treg may not always work as originally thought, paving the way for further improvement of Treg therapy. Our results are somewhat consistent with a previous finding that infusion of Tregs accelerated heart allograft rejection in lymphodepleted non-human primates³². On the other hand, our findings might not necessarily contradict studies by others showing that selective expansion of Tregs by IL-2 or its mutein promoted transplant survival, because we observed the effects of IL-2 and Tregs in an immunodeficient mouse model lacking Tconv cells. The interaction of IL-2, especially at differential doses, with Treg/Tconv cells in the real world is far more complicated than originally thought. However, our findings indicate that targeting CXCL15/IL-8 may improve Treg suppression of alloimmunity in the presence of IL-2.

In the present study, we demonstrated that infusion of 0.75×10⁶ Tregs plus IL-2 (1×10⁴ IU) failed to induce allograft rejection in Rag1^-/- recipient mice, while infusion of 1.5×10⁶ Tregs plus the IL-2 rejected allograft only in some of the recipients. However, injection of 2.5×10⁶ Tregs plus IL-2 resulted in allograft rejection in the vast majority of the recipients. Our findings indicate that it certainly requires a sufficient number of Tregs for the rejection to occur, and that an overdose of Tregs could do harm to an allograft. On the other hand, as low as 4×10³ IU of IL-2 (a low dose per mouse) plus sufficient Tregs (2.5×10⁶ cells) rejected allografts in a small percentage of the recipients, while treatment with 1×10⁴ IU of IL-2 (still a relatively low-dose) plus the Tregs caused allograft rejection in vast majority of the recipient mice, suggesting that IL-2, at low-doses, is sufficient to provoke Treg-initiated allograft rejection.

One may argue that homeostatic proliferation of Tregs in lymphopenic Rag1^-/- mice could interfere with the explanation of the results. Fortunately, we also found that the joint treatment with Tregs and IL-2 resulted in cardiac allograft rejection in wild-type mice that were pre-tolerized by DST plus anti-CD154 treatment. In particular, depletion of neutrophils in this WT mouse model also abolished allograft rejection mediated by Treg transfer and IL-2, suggesting that allograft rejection mediated by Tregs plus IL-2 is dependent on neutrophils. Importantly, lymphopenic conditions were frequently seen in lymphodepleted patients who experienced severe rejection and underwent intense treatment with lymphodepleting regimens. Thus, our findings may be implicated in clinical transplantation. On the other hand, it’s also possible that adoptively transferred Tregs in Rag1^-/- mice could become, at least in part, exTregs or exFoxp3 cells^33–35, which in turn reject an allograft³⁶. Indeed, Duarte et al. demonstrated that a small proportion of adoptively transferred Tregs in Rag2^-/- mice converted to exTregs³⁷. In our study, however, we administered IL-2 together with Tregs as IL-2 is essential for Treg stability. Only when IL-2 was supplied in Rag1^-/- mice were allografts rejected. Adoptive transfer of Tregs alone failed to provoke allograft rejection in Rag1^-/- mice even if some Tregs in these recipients lost Foxp3, with around 75% Tregs still expressing Foxp3, compared to ~95% Tregs derived from the recipients treated with both Tregs and IL-2 (Fig. 1E). Therefore, it’s unlikely that these Tregs in Rag1^-/- recipient mice had become exTregs that in turn rejected an allograft in our mouse models.

Methods

Study design

Although 6-10 mice per group were used for the vast majority of the experiments, at least 5 mice per group were utilized for each experiment, with the exact numbers mentioned in each figure legend. Mice that had very sick or moribund condition were sacrificed and excluded from analysis. Each experiment was separately carried out at least twice, with sufficient power to detect the efficacy or effect. Mice were randomly grouped based on their ages and body weights before experimentation. Experiments were generally performed without blinding. All endpoints were appropriately determined by experimental design, which was also based on our previous and preliminary studies.

Animals

BALB/c and C57BL/6 male mice (6-8 weeks old, 20 ± 2 g) were purchased from Guangdong Medical laboratory Animal Center (Guangdong, China). B6.Rag1-KO mice (Strain NO.T004753) were purchased from GemPharmatech (Nanjing, China). FasL^-/- and perforin^-/- mice (B6 background) were bought from Jackson Laboratory (Bar Harbor, ME, USA), while B6.CXCL15^-/- mice (Strain NO.T011260) were ordered from GemPharmatech (Nanjing, China). DTR⁺ DEREG mice (Cat. NO. NM-KI-190046) were purchased from Shanghai Model Organisms Center (Shanghai, China). All mice were bred and maintained under a specific pathogen-free conditions. All animal experiments were approved by the IACUC (Approval No. 2022073).

Murine skin transplantation

Skin or cardiac donors were 8-week-old male BALB/c mice, while recipients were Rag1^-/- C57BL/6 mice with similar age. Regarding skin transplantation, full-thickness trunk skin with an approximate size of 1 cm² was transplanted to the right dorsal flank area of recipient mice then secured with a bandage, which was removed 8 days after transplantation. Skin allograft rejection was defined as graft necrosis > 90%, as described in our previous studies³⁸.

Cervical heterotopic heart transplantation

Cervical heterotopic heart transplantation in recipient mice was performed. Briefly, the right cervical common carotid artery and the external jugular vein of recipients were cut, equipped with a tail cuff made by 24-G and 22-G intravenous catheter (BD Biosciences), and then connected to the ascending aorta and the pulmonary artery of a donor heart³⁹. Survival of the heart allograft was monitored by visual observation and palpation, and allograft rejection was defined as the cessation of cardiac contractions.

Treatment of Mice

Recipient mice received wild-type (WT) CD4⁺CD25⁺ Treg cells (i.v., 2.5 ×10⁶) with or without 1 ×10⁴ units of recombinant IL-2 (Proteintech, USA) i.p. one day prior to transplantation. IL-2 was administered every other day until graft rejection or sample collection. Some recipient mice received CXCL15-, FasL- or perforin-deficient Tregs (2.5 ×10⁶) in combination with IL-2. To deplete neutrophils, anti-mouse Ly6G (Clone: 1A8, Bio X cell, USA) and anti-rat Kappa immunoglobulin (Clone: MAR 18.5, Bio X cell, USA) Abs were used according to a combinatory strategy reported by Boivin et al⁴⁰. Briefly, anti-mouse Ly6G mAb was injected i.p. to recipient mice at 25ug per mouse daily for the first week following Treg infusion, and then 50ug daily. Meanwhile, anti-rat Kappa immunoglobulin was injected i.p. to recipient mice at 50 μg per mouse once every other day following Treg infusion. Both antibodies were administrated until graft rejection or sample collection.

Purification of CD4⁺CD25⁺ tregs

To isolate human CD4⁺CD25⁺ Treg cells, PBMC from healthy donors (Oribiotech, China) were isolated, while the human CD4⁺CD25⁺CD127^lowCD49d^- Regulatory T Cell Enrichment Kit (Stem Cell) was used according to the user’s guidelines. For murine Tregs, splenocytes from WT, CXCL15^-/-, FasL^-/- or perforin^-/- mice (all C57BL/6 background) were stained with anti-CD4-APC/anti-CD25-PE Abs (Invitrogen), and then CD4⁺CD25⁺ Tregs were sorted by FACSAria III (BD Biosciences). During sorting, small particles, such as normal debris and potentially exosomes, were observed in the lower-left region of the FSC-SSC plot. These small particles were typically not concerning and should not be considered evidence of cell impurity. The purity of the sorted cells was typically > 97%.

Flow cytometric analysis

Draining lymph node (LN) cells, splenocytes or skin-infiltrating cells were isolated from recipient mice and stained for surface markers with anti-CD4 (RM4-5 or GK1.5), and ani-CD45 (30-F11) (Biolegend). To analyze intracellular CXCL15, cells were cultured with the simulation of PMA and ionomycin in the presence of monensin for six hours before the intracellular staining. For intracellular FoxP3, K19, S100A4 and CXCL15 analysis, cells were fixed and permeabilized using Foxp3/Transcription Factor Fixation/Permeabilization Concentrate and Diluent Kits (eBioscience). Then, cells were stained with anti-FoxP3 (FJK-16s, Invitrogen), K19 (SA30-06, HuaBio) and S100A4 (Proteintech) antibodies. For CXCL15 staining, cells were stained with anti-CXCL15 (EPR14470, Abcam) followed by secondary antibody BV421-Donkey anti-rabbit IgG (Poly4064, Biolegend). Flow cytometric data were acquired on Novo Quanteon (Agilent) or CytmFlex S (Beckman, USA) and then analyzed using Flowjo software (Tree Star, USA).

Tolerance induction by donor-specific splenocyte transfusion (DST) and anti-CD154 antibody injection

Splenocytes from BALB/C mice were depleted of erythrocytes and intravenously injected into C57BL/6 mice (1 ×10⁷ cells/mouse) 7 days before heart transplantation. Anti-mouse CD154 mAb (MR1; 0.25 mg/mouse; Bio X cell, USA) was administered intraperitoneally on days −7, −4, 0, +4, +7, and +10 pre- and post- transplantation.

Histological analysis

Allografts of recipient mice were fixed with 4% paraformaldehyde overnight and then embedded in paraffin. Tissues were cut into 3-μm thick sections, which were then used for H&E and immunohistochemistry. For immunohistochemical staining, slides were incubated first with primary anti-Ly6G antibody (1:150, clone: E6Z1T, Cell Signaling Technology) at 4 °C overnight and then with secondary antibody HRP-anti-Rabbit IgG (Maxim, China) at room temperature for 30 min, followed by diaminobenzidine (DAB) color development. The inflammation area and average optical density (AOD) of Ly6G were measured using the Image J software.

Immunofluorescent staining

For immunofluorescence staining, slides with sections were incubated with primary anti-Foxp3 antibody (1:150, Cell Signaling Technology) or Ly6G (1:150, Cell Signaling Technology) at 4 °C overnight. Sections were then stained with a secondary anti-rabbit IgG Ab conjugated with FITC or Cy3 (Beyotime) and finally mounted using DAPI-Fluoromount-G clear mounting agents (Southern Biotech, USA). The sections were observed using a fluorescence microscope. For quantitative analyses, slides were imaged at a magnification of 200×, and the photographs were merged through Adobe Photoshop. The percentage of the positive area was measured using ImageJ software.

Quantitative qPCR arrays

Total mRNA was extracted from skin grafts using TRIzol reagents (Invitrogen) and transcribed to cDNA using Evo M-MLV RT Mix kit with gDNA Clean for qPCR (Accurate Biotechnology) according to the instructions of the manufacturer. Quantitative real-time PCR (qPCR) was performed on the ABI 7500 PCR System (Applied BioSystems) using SYBR Green Premix Pro Taq HS qPCR Tracking kits (Accurate Biotechnology) for the qPCR arrays of inflammatory cytokines/chemokines and their receptors (wc-mRNA0266-m, Wcgene Biotech, China), with primers pre-embedded in 96-well plates. The relative mRNA levels of genes were analyzed through the comparative 2^-△Ct method and normalized with the values of Actb and B2m.

Isolation of skin graft-infiltrating cells

To isolate graft-infiltrating cells in skin allografts, skin was cut into small pieces of 3–4 mm, and digested with 200 U/mL Collagenase Type I (Life Technologies) and 100 ug/mL DNAse (Roche) in HBSS at 37 °C for 4 hours. The cell suspension was then collected by filtering through a 40-um cell strainer, centrifugated and counted for further analysis.

Measurement of murine CXCL15 and human IL-8 via ELISA

Human or murine CD4⁺ Tregs (2 × 10⁵ cells/well) were cultured in 96-well plates in complete RPMI-1640 medium and stimulated with anti-CD3 (5 µg/ml) and anti-CD28 (1 µg/ml) Abs at 37 °C with 5% CO₂ for 2 days. In some groups, cells were also treated with recombinant IL-2 (Proteintech, USA). The levels of IL-8 or CXCL15 in the supernatant were measured using ELISA kits (ZCIBIO or Fine Biotech) according to the manufacturer’s instructions. The absorbance was read at 450 nm in a microplate spectrophotometer (Thermo Fisher Scientific).

Statistical analysis

Mean ± standard deviation (mean ± SD) was used to present the data. Normality was analyzed by the Shapiro-Wilk test. For normally distributed data involving more than two groups, one-way ANOVA was used when the assumption of homogeneity of variances was met. Otherwise, Welch’s ANOVA was performed. Kruskal–Wallis test was employed when the data did not meet the assumption of normality. If a statistically significant difference was found using any of the above tests, post-hoc analyses were conducted for pairwise comparisons. Tukey’s test was typically used following one-way ANOVA, and Dunnett’s T3 multiple comparisons were performed after a significant Welch’s ANOVA, while Dunn’s test was applied for pairwise comparisons following a significant Kruskal–Wallis test. Regarding non-normally distributed data, the Kruskal-Wallis test with Dunn’s post hoc test was used for three or more groups. The Mann-Whitney U test or two-tailed Student’s t-test was employed for two groups only. The analysis of graft survival was performed using the Kaplan–Meier method (log-rank test). The numbers of animals were indicated in each figure legend. Data were analyzed via GraphPad Prism 9 (GraphPad Software). A value of P < 0.05 was considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Feifei Qiu and Huazhen Liu: investigation, methodology, formal analysis and writing- original draft. Nianqiao Gong: conceptualization, supervision and writing- review & editing. Chun-Ling Liang: visualization and methodology. Yuchao Chen: formal analysis. Jonathan Bromberg and Zhenhua Dai: conceptualization, supervision and writing- review & editing.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Weishan Huang and Mengtan Xing.

Data availability

The original datasets are available upon request by readers. The source data for all figures presented in the manuscript are provided in Supplementary Data.

Competing interests

Authors declare that they have no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10542-z.