CD4+Foxp3+T Regulatory Cells Promote Transplantation Tolerance by Modulating Effector CD4+ T Cells in a Neuropilin-1-Dependent Manner

Several mechanisms of immune suppression have been attributed to Foxp3+ T regulatory cells (Treg) including modulation of target cells via inhibition of cell proliferation, alteration of cytokine secretion, and modification of cell phenotype, among others. Neuropilin-1 (Nrp1), a co-receptor protein highly expressed on Treg cells has been involved in tolerance-mediated responses, driving tumor growth and transplant acceptance. Here, we extend our previous findings showing that, despite expressing Foxp3, Nrp1KO Treg cells have deficient suppressive function in vitro in a contact-independent manner. In vivo, the presence of Nrp1 on Treg cells is required for driving long-term transplant tolerance. Interestingly, Nrp1 expression on Treg cells was also necessary for conventional CD4+ T cells (convT) to become Nrp1+Eos+ T cells in vivo. Furthermore, adoptive transfer experiments showed that the disruption of Nrp1 expression on Treg cells not only reduced IL-10 production on Treg cells, but also increased the frequency of IFNγ+ Treg cells. Similarly, the presence of Nrp1KO Treg cells facilitated the occurrence of IFNγ+CD4+ T cells. Interestingly, we proved that Nrp1KO Treg cells are also defective in IL-10 production, which correlates with deficient Nrp1 upregulation by convT cells. Altogether, these findings demonstrate the direct role of Nrp1 on Treg cells during the induction of transplantation tolerance, impacting indirectly the phenotype and function of conventional CD4+ T cells.


INTRODUCTION
Foxp3+ T regulatory (Treg) cells are an important population of leukocytes that control immunity, mainly by dampening effector T cell responses. Many studies have described the mechanisms by which Treg cells carry out their function, such as IL-2 deprivation, secretion of cytotoxic granules (granzyme/perforin), metabolic disruption, secretion of anti-inflammatory cytokines, and release of extracellular vesicles (exosomes) (1,2).
In addition to their capacity to suppress immune responses, Treg cells had become an interesting target for cell therapy, due to the increasing number of diseases associated with malfunction and over-reactivity of the immune system, such as autoimmunity and transplant rejection (3,4). The current paradigm is based on the premise that immune tolerance to allogeneic transplant is broken by an imbalance of Treg cells over T effector cells. The infusion of Treg cells considerably increases graft survival in transplanted animals (4,5), and clinical trials of the administration of Treg cells into patients have demonstrated safety but variable efficacy (6). Understanding Treg cell biology and its mechanisms of immune suppression may improve the potential and use of Treg cells as therapeutic agents.
A few years ago, Neuropilin-1 (Nrp1) was described as a potential Treg cell marker. Nrp1 is a transmembrane coreceptor with affinity for a variety of ligands, all involved in physiological processes, such as angiogenesis, axonal guidance, or immune synapses (7). In the immune system, Nrp1 is expressed mainly by dendritic cells, Natural Killer (NK) and Treg cells (8)(9)(10)(11). Initially, the function described for Nrp1 was to stabilize the interaction between cells during antigen presentation through homotypic interactions (12,13). However, other studies suggested later that Nrp1 contributes to the function, phenotypic stability, and survival of Treg cells in tumors (14).
Several reports correlate Nrp1 expression on T cells with a state of immune tolerance (14)(15)(16)(17)(18)(19), which has been demonstrated in the transplantation context both in patient biopsies and experimental models (20)(21)(22). In addition, Nrp1deficient or Nrp1KO Treg cells are not capable of exerting suppressive function through a semi-porous membrane; and the same phenomenon was observed when using wild type Treg cells in the presence of anti-Nrp1 blocking antibodies (14).
We previously described that conventional CD4+ T cells (defined as CD4+CD25-Nrp1-Foxp3-cells or convT) upregulate Nrp1 expression during allograft rejection. Interestingly, in the tolerogenic condition in which Nrp1+Foxp3+ Treg cells are co-transferred with convT cells, a larger frequency of Nrp1+Eos+ convT cells was observed suggesting that Nrp1+Treg cells could modulate the phenotypic signature of convT cells (22), leading to the generation of T cells with modulatory effects.
Based on these antecedents, we hypothesized that convT cells gain Nrp1 and Eos in an Nrp1+Treg cell-dependent manner to favor immune suppression. Using Nrp1 conditional knocked out mice; we demonstrate that Nrp1KO Treg cells are deficient in exerting suppressive activity in a contact-independent manner. Even more, when Treg cells lack Nrp1, convT cells are unable to up-regulate Nrp1 and Eos expression favoring the appearance of type-1 T helper (Th1) cells. Accordingly, the frequency of IL-10+Treg cells is negatively affected, which correlates with the inability to induce long-term tolerance. Lastly, we demonstrate that Treg cells-modulated convT cells also gain the ability to suppress ex vivo T cell proliferation, which is affected if co-transferred Treg cells lack Nrp1. Hence, we demonstrate that Treg cells drive immune tolerance by modulating the phenotype and function of convT cells in an Nrp1-dependent manner.

RESULTS
The Lack of Nrp1 on Treg Cells Is Not Involved in Treg-Phenotypic Signature In 2015, our group reported that convT cells transferred into skin-transplanted animals gain Nrp1 expression (from 0 to ∼35%). This induction was highly increased when convT cells were co-transferred with Nrp1+Treg cells (22). To clarify whether this process depends on Nrp1 expressed specifically on Treg cells, we obtained Foxp3 Cre/YFP and Nrp1 flox/flox mice to generate Nrp1-deficient or Nrp1KO Treg cells, which are conveniently detectable by flow cytometry based on the expression of YFP (23,24). First, we tested the phenotype of T cells from different organs/tissue of Foxp3 Cre/YFP (wild type, wt control), Foxp3 Cre/YFP Nrp1 flox/+ (het) and Foxp3 Cre/YFP Nrp1 flox/flox (Nrp1KO Treg) offspring. As expected, we found that deletion of Nrp1 only occurs on Treg cells, as seen in peripheral lymph nodes (pLN), spleen (Spl), and blood cells (∼75% on wt and het, and <1% on Nrp1KO) ( Figure 1A). In the case of convT cells, we find a partial decrease in Nrp1 expression only in the Spl, which could correspond to antigen experienced ex-Treg cells (Foxp3-T cells), since convT cells were considered and gated as Foxp3-cells.
Even more, we performed high-dimensional single cell data analysis by visualization of t-Distributed Stochastic Neighbor Embedding (t-SNE) algorithm (viSNE) (37, 38) on pLN, Spl and blood cells populations from all three mice genotypes (Figures 2A,B), searching for main cell subsets based on the expression of CD8, CD4, Foxp3, and CD19 (Supplemental Figure 1B). viSNE heat maps confirmed that Nrp1 depletion only occurs in the Foxp3+Treg cell compartment of Nrp1KO animals, whereas expression of Nrp1, Eos, CD49b, and CD25 remain mostly unchanged among the identified populations (CD19+ B cells, CD4+Foxp3-convT, and CD8+ T cells) (Figures 2B,C). Additionally, by performing spanning-tree progression analysis of densitynormalized events (SPADE), we identified 11 spatially distinct immune cell populations with unaltered frequencies among wt, het, and Nrp1KO splenic cells (Figures 2D,E). Furthermore, the relative Mean Fluorescence Intensity (MFI) of each marker was calculated for each SPADEon-viSNE cell population studied, highlighting the lack of

Treg Cells Require Nrp1 Expression to Modulate Conventional CD4+ T Cell Phenotypic Signature
After confirming typical Treg cell phenotype in all genotypes, we performed a suppression assay to test the modulatory function of Treg cells and acquisition of Nrp1 by convT cells in vitro. For this experiment we used congenically marked antigen presenting cells (APC, CD45.2+), Treg cells (CD45.2+) and CTV-stained convT cells (CD45.1+) to facilitate the tracking of convT cells under proliferation (Supplemental Figure 2A). As shown in Figure 3A, wt Treg cells suppress the proliferation of convT cells at 1:1, 1:2, and 1:4 Treg:convT ratios (∼85, ∼65, and ∼45% of suppression, respectively); but both het and Nrp1KO Treg cells show slightly reduced suppressive activity, although these variations are not statistically different. These results suggest that Nrp1 is not fully required for Treg cells acting in a contact-dependent manner ( Figure 3B). On the other hand, we checked the phenotype of convT cells by looking at Nrp1 expression. As depicted in Figures 3C,D, convT cells acquired Nrp1 expression in a dose and Tregdependent manner as Nrp1KO Treg cells were able to allow convT cells to become Nrp1+ to a lesser extent (for instance, at 1:1 ratio we observe ∼5% of Nrp1+ conv T cells without Treg cells, ∼20% of Nrp1+ convT cells with wt Treg cells, ∼15% of Nrp1+ convT cells with het Treg cells and ∼12% of Nrp1+ convT cells with Nrp1KO Treg cells). This data supports our previous observation, in which convT cells up-regulated Nrp1 during allograft rejection but when convT cells were co-transferred with Treg cells, the expression of Nrp1 was enhanced in addition to the induction of transplant tolerance (22). At the same time, Nrp1-expressing Treg cells (wt and het) decrease Nrp1 expression in a dosedependent manner when co-cultured with convT cells at the different ratios, Figure 3E. Conversely, convT cells cocultured with wt Treg cells did not gain Eos expression in vitro, which could be a result of the experimental timing (3 days for in vitro assay vs. 20 days for in vivo assay) (Supplemental Figures 2B,C).
Since we did not observe a robust difference between the suppressive activity of wt, het, and Nrp1KO Treg cells, we repeated the suppression assay but using a contactindependent setting (transwell system) (Supplemental Figure 3). In this case, we observed differences in the suppressive function of wt and Nrp1KO Treg cells, where Nrp1KO Treg show reduced suppressive activity, indicating that Nrp1+ Treg cells produce factors with modulatory function, which are altered in Nrp1KO Treg cells, Figures 3F,G. Even more, convT cells also up-regulated Nrp1 expression when wt Treg cells, and not Nrp1KO Treg, were placed at the top chamber (∼17% vs. ∼9% of Nrp1+ convT cells, respectively, Figure 3H). Thus, the presence of Nrp1 on Treg cells modulates the phenotype of convT cells and contributes to T cell suppression in vitro.

Nrp1KO Treg Cells Cannot Induce Long-Term Transplant Tolerance and Fail to Modulate Conventional CD4+ T Cells in vivo
Next, we tested the role of Nrp1 on Treg cell function using an in vivo skin transplantation model, in which C57BL/6 × Balb/c (F1) skin is grafted onto RAG-KO mice previously administered with convT cells (Supplemental Figure 4A). Using this approach, we observed complete transplant rejection by day 20 post-surgery, but skin graft tolerance when convT cells are co-transferred with Nrp1+ Treg cells (22). In the current study, we carried out long-term skin transplant experiments using wt, het, and Nrp1KO Treg cells and found that wt Treg cells induce ∼60% of transplant survival and only ∼20% for the groups receiving het and Nrp1KO Treg cells (Figure 4).
Furthermore, we performed 20 days-long experiments, for which skin graft-draining lymph nodes (dLN) were removed and the number and phenotype of T cells was studied. All three Treg cell genotypes were able to reduce cellularity in the dLN of allografted animals (Supplemental Figure 4B). Additionally, we analyzed dLN cell phenotype by flow cytometry, discriminating clearly between convT cells (CD45.1+) and Treg cells (CD45.2+) (Supplemental Figures 4C,D). As depicted in Figures 5A,C, convT cells gained Nrp1 and Eos expression, which seems to occur in a Treg cell-dependent manner because ∼30% of convT cells became Nrp1+ when co-transferred with wt Treg cells in comparison with ∼15% or ∼12% when co-transferred with het or with Nrp1KO Treg cells, respectively, Figure 5B. Moreover, ∼20% of convT cells gained Eos expression when co-transferred with wt Treg cells in contrast to ∼10% for the case of het or Nrp1KO Treg cells, Figure 5D. Importantly, Eos expression remain unaltered regardless of Treg cells genotype, Supplemental Figures 4D-G. These observations confirmed our previous study in which we described that CD4+ T effector phenotype is modulated by Nrp1-expressing Treg cells.
To further investigate Nrp1KO Treg cells, we studied Treg cell cytokine production finding that ∼20% of either het or Nrp1KO Treg cells were IFNγ+ in contrast to∼10% of wt IFNγ+ Treg cells, Figures 6A,B. No differences were observed among the frequencies of wt, het, or Nrp1KO Treg cells either IL-17A+ or IFNγ+IL-17A+. When we tested for the production of the anti-inflammatory cytokine IL-10, we found that <5% of het and Nrp1KO Treg cells were IL-10+ compared with ∼8.5% of wt Treg cells, Figures 6C,D. Therefore, Nrp1deficiency on Treg cells during allograft rejection negatively affects IL-10 production but favors IFNγ production, which may explain the reduced % of transplant survival of grafted animals treated with Nrp1KO Treg cells. To corroborate this, we analyzed IL-10 levels on convT-and-Treg in vitro culture supernatants, observing decreased IL-10 production when convT were co-cultured with Nrp1KO Treg cells, compared with wt
Altogether, our findings indicate that Treg cells require Nrp1 to modulate the function of convT cell and to exert optimal suppressive activity in vivo.

DISCUSSION
Regulatory T cell malfunction has been widely associated with increased inflammatory immune responses. Understanding Treg cells biological processes and mechanisms of suppression are pivotal for recognizing new targets for therapy. Nrp1 has been previously proposed as a cell marker for thymic-derived Treg cells (39); but its expression has also been described on T cells during allogeneic skin graft rejection (22), sepsis (40), IL-10 deficiency (41), and anti-tumor immune responses (42,43). Moreover, Nrp1 deficiency on Foxp3+ Treg cells has been associated with lack of immune suppression, predominantly affecting tumor growth (14,15,44) and worsening EAE severity (45). In this work, we focused on analyzing the role of Nrp1 specifically on Treg cells during the induction of skin transplant tolerance. In one of our previous studies we demonstrated that Nrp1+ Treg cells drive transplantation tolerance, proposing the modulation of conventional CD4+ T cell phenotype as a possible mechanism (22). To get more insight regarding this observation, we obtained conditional knockout animals in which the lack of Nrp1 expression is restricted to Foxp3+ Treg cells (24). First, we extensively analyzed the phenotype of T cells in these genetically modified mice, including Foxp3+ Treg cells, finding no aberrant expression of Treg cell-associated markers in animals containing Nrp1KO Treg cells compared to controls (Figures 1, 2). Next, we designed in vitro experiments to evaluate the function of Nrp1KO Foxp3+ Treg cells and their capacity to modulate convT cell phenotype. The regulatory ability of Nrp1KO Foxp3+ Treg cells was evaluated in contact-dependent and-independent assays, observing that Foxp3+ Treg cells require Nrp1 to inhibit T cell proliferation in a contact-independent manner, suggesting that Nrp1+ Treg cells secrete modulators to exert their suppressive function (Figure 3). Because we obtained a defect in cell contactindependent suppression when using Nrp1KO Treg cells, and Delgoffe et al. reported that the inclusion of anti-Nrp1 blocking antibody in a transwell suppression assay negatively affected the suppressive capacity of Nrp1+ Treg cells (14), it is reasonable to propose that Nrp1+Foxp3+ Treg cells may secrete Nrp1+ extracellular vesicles (EV) to "deliver" Treg cells' modulatory effects to convT cells. In this regard, several authors have published that Treg cells are able to produce EV (containing modulatory molecules) as a novel manner to inhibit CD4+ T cell effector function (46)(47)(48). Furthermore, we also observed in the in vitro experiments that convT cells gained Nrp1 in a dose dependent manner according to the ratios used with Nrp1+ Treg cells (Figures 3C,D). In other words, we obtained higher % of Nrp1+ convT cells when co-cultured with higher number of Nrp1+ Treg cells, resulting in lower frequencies of Nrp1+ convT cells when Nrp1KO Treg cells were added. Even more, Nrp1+ Treg cells decreased Nrp1 expression while convT cells gained it (Figures 3D,E). This phenomenon was first reported by our group using an in vivo model of transplantation and in this current report we corroborated the observation in vitro in addition to proving that it occurs in Nrp1+Foxp3+ Treg cell-dependent manner (Figure 3). Hence, it is conceivable to propose that Nrp1+ Treg cells could secrete Nrp1+ EV to target convT cells; in this regard, data from our laboratory indicates that wt Treg cells secrete EV containing Nrp1 in their membrane and that these EV can modulate convT cells phenotype and function, in comparison to EV obtained from Nrp1KO Treg cells (manuscript in preparation).
Using a murine model for allograft transplantation, we demonstrated that Nrp1 is required by Foxp3+ Treg cells to facilitate long-term tolerance (Figure 4). Interestingly, when we studied the phenotype of wt and Nrp1KO Treg cells in this in vivo setting, we did not find changes in Foxp3 or Eos expression (Supplemental Figures 4D-F). The transcription factor Eos functions as a co-repressor of Foxp3, preventing the expression of convT cell-related genes in Treg cells (26); and it has been described that "Eos-labile" Treg cells display a more inflammatory/helper phenotype (25). On the contrary, Rieder et al. have reported that convT cells from Eos −/− animals produce less IL-2 and show a malfunctioning CD25/STAT5 signaling pathway upon in vitro activation, but under in vivo inflammatory conditions Eos −/− T cells become high producer of IL-17 (49). In this work, we found that Nrp1 deficiency did not modify Foxp3 or Eos expression on Foxp3+ Treg cells under homeostatic conditions and in our model of skin graft transplantation, supporting that graft rejection was not due to instability in the expression of Foxp3 or Eos. Complementing Treg cell phenotypic analysis, we found that Nrp1KO Treg cells up-regulate IFNγ (Figure 6), supporting that the Nrp1 signaling pathway is required by Treg cells to secrete the appropriate cytokines (44). Gao et al. showed that Nrp1 low CD25+ CD4+ Treg (Treg expressing low levels of Nrp1) purified from septic mice secreted lower amounts of IL-10 (17), and another study using tumor-harboring Il10KO (IL-10 deficient)mice showed decreased Nrp1 expression in tumor-infiltrating Treg, impaired Nrp1+ Treg tumor accumulation and tumor protection function (41). While RNA sequence observations by Delgoffe et al. previously showed decreased Il10 expression on Nrp1KO Treg (14), we have demonstrated here, for the first time, an impairment in IL10 production [both in vitro (in coculture suppression assay) and in vivo (during allograft rejection), Figure 6] on Treg cells lacking Nrp1 expression. The gaining of Nrp1 expression by convT cells failed in the presence of Nrp1KO Treg (Figure 3) but, intriguingly, it could be reversed by adding exogenous IL-10 to the culture media, suggesting that Treg-mediated upregulation of Nrp1 on convT cells (and possibly other immunomodulatory proteins) could be dependent on IL-10 signaling. On the other side, convT cells gained Nrp1 and Eos expression in vivo when co-transferred with wt Treg cells, which is in agreement with our previous findings (22). Notably, both het and Nrp1KO Treg cells are unable to favor Nrp1 and Eos expression on convT cells in vivo, and T cell suppression activity ex vivo, which correlates with increased IFNγ+ convT cell frequencies and poor allograft acceptance observed (Figures 4, 5, 7). These data support the hypothesis involving the secretion of Nrp1+ EV by Treg cells to modulate the phenotype/function of convT cells and the role of Nrp1 in controlling cytokine production of CD4+ T cells, although this control will depend on the subset of CD4+ T cells targeted (Treg vs. convT cells).
Altogether, our current results indicate that the expression of Nrp1 on Foxp3+ Treg cells is relevant for driving T cell suppression in vitro and in vivo by modulating convT cell proliferation, phenotype, cytokine production, and suppressive function.

Mice
Six to 8 week old male and female mice were used.

Skin Transplantation
Tail skin (∼1 cm 2 ) from C57Bl/6 (syngeneic) or F1 (allogeneic) donors was transplanted onto the dorsal area of RAG-KO recipient mice. Survival of skin allografts was evaluated twice per week and grafts were considered rejected when 80% of the initial graft had disappeared or become necrotic.

ELISA
Supernatants from co-culture assays were collected after 72 h and stored at −80 • C until cytokine quantification by sandwich ELISA. Briefly, 96-well flat-bottomed plates were coated overnight with 1 µg/mL purified anti-mouse IL-10 (clone JES5-2A5, Biolegend) capture antibody. After several washing steps with 1X PBS + 0.05% Tween 20 and a blocking step with 1% BSA in 1X PBS, samples were incubated at room temperature for 2 h. Biotinylated anti-mouse IL-10 (clone JES5-16E3, Biolegend) at 1 µg/mL in conjunction with HRP-avidin (Biolegend) were used for detecting immobilized cytokine and tetramethylbenzidine (TMB, ThermoFisher) substrate was added to detect HRP activity. Reaction was stopped by adding 2N H 2 SO 4 and absorbance was measured at 450 nm wavelength using a Tecan absorbance microplate reader.

ETHICS STATEMENT
Mice were maintained under pathogen-free conditions at the animal facility located at Facultad de Medicina, Universidad de los Andes. All procedures were carried out according to the bioethics committee guidelines from Universidad de los Andes, and National Commission of Science and Technology (CONICYT).  were sort-purified from Foxp3 GFP+ CD45.1+ animals, antigen presenting cells (or APCs, CD3-MHCII+CD45.2+), and Treg cells (CD4+Foxp3 YFP+ CD45.2+) were sort-purified from wt, het, and Nrp1KO-Foxp3 YFP+ animals. ConvT cells were stained with CellTrace TM Violet and cultured un-stimulated or activated with Mitomycin-C treated-APCs plus soluble anti-CD3 antibody, in the absence or presence of wt, het, or Nrp1KO Foxp3 YFP+ Treg cells. ConvT cell proliferation was measured by dye dilution using flow cytometry. (B) Representative dot plots show Eos expression on CD45.1+convT cells after 3 days of co-culture with wt Treg cells. (C) Accumulated frequency of Eos+ convT cells in the aforementioned conditions. For C, bars represent mean ± SEM, n = 2 independent experiments.

AUTHOR CONTRIBUTIONS
Supplemental Figure 3 | Contact-independent Treg cell suppression assay. Contact-independent suppressive assay strategy: responder convT, APCs, and Treg cells were obtained as detailed in Supplemental Figure 2. ConvT cells were stained with CTV and cultured in the bottom chamber un-stimulated or activated with Mitomycin-C treated-APCs plus soluble anti-CD3 antibody, in absence or presence of wt, het, or Nrp1KO Foxp3 YFP+ Treg cells placed in the top chamber (transwell). ConvT cell proliferation was measured by tracking dye dilution by flow cytometry.