Neuropilin-1 (Nrp-1) is a well described marker molecule for CD4+Foxp3+ thymus-derived regulatory T cells (Tregs). In addition, a small population of CD4+Foxp3− conventional (conv) T cells expresses Nrp-1 in naive mice, and Nrp-1 expression has been described to be upregulated on activated CD4+ T cells. However, the function of Nrp-1 expression on CD4+ non-Tregs still remains elusive. In this study, we demonstrate that Nrp-1 expression was induced upon stimulation of CD4+Foxp3− T cells in vitro and during an ongoing immune response in vivo. This activation-induced Nrp-1+CD4+ T cell subset (iNrp-1+) showed a highly activated phenotype in terms of elevated CD25 and CD44 expression, enhanced production of proinflammatory cytokines, and increased proliferation compared with the Nrp-1−CD4+ counterpart. In contrast, Nrp-1+CD4+Foxp3− conv T cells from naive mice (nNrp-1+) were dysfunctional. nNrp-1+CD4+ conv T cells upregulated activation-associated molecules to a lesser extent, exhibited impaired proliferation and produced fewer proinflammatory cytokines than Nrp-1−CD4+ conv T cells upon stimulation in vitro. Moreover, the expression of PD-1 and CTLA-4 was significantly higher on nNrp-1+CD4+Foxp3− T cells compared with iNrp-1+CD4+Foxp3− T cells and Nrp-1−CD4+Foxp3− T cells after stimulation and under homeostatic conditions. Strikingly, transfer of Ag-specific iNrp-1+CD4+ conv T cells aggravated diabetes development, whereas Ag-specific nNrp-1+CD4+ conv T cells failed to induce disease in a T cell transfer model of diabetes. Overall, our results indicate that Nrp-1 expression has opposite functions in recently activated CD4+ non-Tregs compared with CD4+ non-Tregs from naive mice.
This article is featured in Top Reads, p.1223
Neuropilin-1 (Nrp-1) is a transmembrane glycoprotein that mainly acts as a coreceptor for semaphorins (Sema), including Sema 3A and Sema 4A, and for members of the vascular endothelial growth factor (VEGF) family. In addition, homophilic interaction of Nrp-1 has been described (1, 2). Studies with Nrp-1–deficient mice, which die in the embryonic stage, revealed its importance in several physiological processes, such as development, axonal guidance, and angiogenesis, including endothelial cell migration and vascular sprouting (3–5). However, Nrp-1 also plays a crucial role in the immune system and has been described to be involved in development, migration, and communication between different immune cells (6).
Nrp-1 is expressed by various immune cells, including mast cells, macrophages, dendritic cells (DCs), and T cells. On DCs, Nrp-1 seems to promote the formation of the immunological synapse. Tordjman et al. (1) detected Nrp-1 expression on human mature DCs and resting T cells. They proposed homophilic binding of Nrp-1 to promote intercellular adhesion, resulting in the initiation of T cell activation. Indeed, preincubation of DCs or T cells with Nrp-1 blocking Abs inhibited DC-mediated T cell proliferation. In mice, similar homotypic Nrp-1 interaction has been suggested for long-lasting interaction of immature DCs with regulatory T cells (Tregs) under steady-state conditions (2). We identified Nrp-1 to be highly expressed on immunosuppressive murine thymus-derived Tregs (tTregs), thereby representing a useful surface marker molecule for murine Tregs (7). Besides development within the thymus, Tregs can also be induced from naive CD4+Foxp3− T cells under certain conditions in the periphery (8–11). Two studies have reported that these peripherally induced Tregs (pTregs) do not express Nrp-1, and thereby, the authors proposed Nrp-1 as suitable surface molecule to distinguish tTregs from pTregs (12, 13). However, Nrp-1 expression was detected in pTregs isolated from highly inflamed tissues, including spinal cord of mice suffering from experimental autoimmune encephalitis (EAE) and lungs of mice with chronic asthma but not in the secondary lymphoid organs of the same mice (13). Hence, Nrp-1 alone seems not to be sufficient to distinguish pTregs and tTregs under all conditions, in particular during ongoing inflammation in vivo.
Mechanistically, Nrp-1 expression on Tregs was proposed to be involved in the suppressive function of Tregs. During EAE, characterized by the infiltration of Ag-specific T cells into the brain, resulting in a local inflammatory response, T cell–specific ablation of Nrp-1 expression aggravated EAE severity. In addition, Nrp-1–deficient Tregs exhibited reduced suppressive capacity, at least in in vitro experiments (14). Our studies revealed that Nrp-1 expression mediates migration of CD4+Foxp3+ Tregs into tumor tissue in response to tumor-derived VEGF. T cell–specific ablation of Nrp-1 expression, which mostly affects Tregs, resulted in fewer Treg numbers within the tumor and elevated antitumoral CD8+ T cell responses accompanied by significantly reduced tumor growth (15). In addition, Nrp-1 expression has been described to be important for the stability and function of Tregs. Sema 4A binding to Nrp-1 initiated a signaling cascade in Tregs that led to stabilized Foxp3 expression and potentiated survival of Tregs (16). Specific depletion of Nrp-1 expression from intratumoral Tregs induced a functional fragile phenotype characterized by elevated IFN-γ expression accompanied by increased antitumoral CD8+ T cell responses and reduced tumor burden (17).
Although Nrp-1 expression is widely used to identify Tregs, a small population of Nrp-1+CD4+Foxp3−CD25− T cells from naive mice and Nrp-1–expressing activated CD4+ conventional (conv) T cells have been described. Stimulation of human CD4+ T cells from peripheral blood of healthy donors resulted in the induction of Nrp-1 in vitro (18, 19). Among tumor-infiltrating lymphocytes, Nrp-1+ non-Tregs were detected in human colorectal liver metastasis (20) and human non–small cell lung cancer, as well as in different transplanted murine solid tumors (18). Moreover, during allograft rejection in a skin transplantation model and in mice suffering from atherosclerosis, Nrp-1 expression was induced in CD4+CD25− T cells or CD4+Foxp3− T cells, respectively (21, 22). Mechanistically, the authors assumed that Nrp-1 expression facilitates the migration of effector T cells into the aorta, where they exhibit their pathogenic function during atherosclerosis (22). In contrast, Nrp-1+CD4+CD25− T cells from healthy mice seem to have a more suppressive function because adoptive transfer of these cells prolonged graft survival, in particular in synergy with rapamycin (23). Hence, the role of Nrp-1 expression in CD4+ conv T cells during steady-state conditions and upon stimulation still remains elusive and therefore was analyzed in more detail in vitro and in vivo in the current study.
Materials and Methods
BALB/c wild-type (WT) mice were obtained from Envigo (Rossdorf, Germany). Foxp3eGFP mice (C.Cg-Foxp3tm2Tch/J; The Jackson Laboratory, Bar Harbor, ME) express GFP in Foxp3+ Tregs (24). TCR-HA/Thy.1.1 transgenic mice (BALB/c) express an αβ TCR specific for the peptide110–120 from influenza hemagglutinin (HA) presented by I-Ed class II MHC molecules (25) and were backcrossed to Thy1.1 mice. In these mice, ∼10–15% of peripheral CD4+ T cells express the transgenic TCR. TCR-HA/Foxp3eGFP (BALB/c) mice were bred in house. INS-HA/Rag2 knockout (KO) transgenic mice (BALB/c) express the HA protein under control of the insulin promoter (26) and are deficient for Rag2 expression. Nrp-1tdTomato mice (C57BL/6N NTac-Nrp-1-tm4609(T2A-tdTomato)Tac) were generated by Taconic Biosciences (Cologne, Germany). In short, the sequence for the T2A and the open reading frame of tdTomato were inserted between the last amino acid and the translational termination codon in exon 17 of the nrp-1 gene. The positive selection marker (puromycin resistance [PuroR]) was flanked by flippase recombination target sites and inserted into intron 16. The targeting vector was generated using BAC clones from the C57BL/6J RPCI-23 BAC library and transfected into the Taconic Biosciences C57BL/6N Tac embryonic stem cell line. Homologous recombinant clones were isolated by using positive (PuroR) and negative (thymidine kinase [Tk]) selections. The constitutive knock-in allele was obtained after in vivo Flp-mediated removal of the selection marker. This allele expresses a chimeric transcript of the nrp-1 gene fused to the T2A and the tdTomato sequence, resulting in coexpression of the Nrp-1 and tdTomato proteins under the control of the endogenous Nrp-1 promoter (Supplemental Fig. 1). Mice were housed and bred in the animal facility at the University Hospital Essen under specific pathogen–free conditions. All animal experiments were carried out in accordance with the guidelines of the German Animal Protection Law and were approved by the State Agency for Nature, Environment and Customer Protection, North Rhine– Westphalia, Germany.
Flow cytometric expression analysis of proteins was performed by using fluorochrome-labeled anti-CD4, anti-CD25, anti-CD44, anti–CTLA-4, anti–IFN-γ (BD Biosciences, Heidelberg, Germany), anti-Foxp3, anti-Ki67 (eBioscience; Thermo Fisher Scientific, Langenselbold, Germany), anti-CD69, anti–PD-1, anti–TNF-α (BioLegend, San Diego, CA), and anti–Nrp-1 (R&D Systems, Minneapolis, MN) Abs. The 6.5 (anti–TCR-HA) mAb was purified from hybridoma supernatant and labeled with Alexa Fluor 647. Dead cells were identified by staining with the Fixable Viability Dye eFluor 780 (eBioscience; Thermo Fisher Scientific, Langenselbold, Germany). Intracellular staining for Foxp3, CTLA-4 and Ki67 was performed with the Foxp3 Transcription Factor Staining Kit (eBioscience; Thermo Fisher Scientific, Langenselbold, Germany) according to the manufacturer’s protocol. Cytokine expression was measured by stimulating splenocytes with 10 ng/ml PMA and 100 µg/ml ionomycin (both Sigma-Aldrich, Munich, Germany) for 4 h in the presence of 5 µg/ml brefeldin A, followed by treatment with 2% paraformaldehyde and 0.1% IGEPALCA-630 (Sigma-Aldrich, Munich, Germany), and staining with the respective Ab for 30 min at 4°C. Flow cytometric analyses were done on an LSR II or a BD FACSCelesta with Diva Software (BD Biosciences, Heidelberg, Germany).
Cell isolation and stimulation
Single-cell suspensions of splenocytes were generated by rinsing spleens with erythrocyte lysis buffer and washing with PBS supplemented with 2% FCS and 2 mM EDTA. CD4+CD25− T cells were isolated by using the CD4+ T Cell Isolation Kit (Miltenyi Biotec, Bergisch-Gladbach, Germany), according to the manufacturer`s recommendation, and by addition of biotinylated anti-CD25 Ab (BD Biosciences, Heidelberg, Germany). For isolation of Nrp-1+CD4+CD25− and Nrp-1−CD4+CD25− T cells from TCR-HA/Thy1.1 mice or Nrp-1+CD4+Foxp3−(eGFP−)CD25− and Nrp-1−CD4+Foxp3−(eGFP−)CD25− T cells from Foxp3eGFP reporter mice, splenocytes were stained with anti–Nrp-1 (FAB566; R&D Systems, Minneapolis, MN), anti-CD4 and anti-CD25 and cell sorted by using a BD FACSAria II Cell Sorter (BD Biosciences, Heidelberg, Germany). Alternatively, activation-induced Nrp-1+(iNrp-1+)CD4+Foxp3−(eGFP−) and Nrp-1−CD4+Foxp3−(eGFP−) T cells were cell sorted from in vitro–stimulated CD4+ T cells. For isolation of Nrp-1+(tdTomato+)CD4+CD25− and Nrp-1−(tdTomato−)CD4+CD25− T cells from Nrp-1tdTomato reporter mice, splenocytes were incubated with fluorochrome-labeled anti-CD4 and anti-CD25 Abs and cell sorted by using a BD FACSAria III Cell Sorter (BD Biosciences, Heidelberg, Germany). Murine T cells were stimulated with 1 µg/ml anti-CD3 (BD Biosciences, Heidelberg, Germany) in the presence of irradiated splenocytes in IMDM culture supplemented with 10% heat-inactivated FCS, 25 mM β-Mercapthoethanol, and antibiotics (100 U/ml penicillin, 0.1 mg/ml streptomycin).
T cells were labeled with the cell proliferation dye CFSE (Thermo Fisher Scientific, Langenselbold, Germany) according to the manufacturer`s protocol and stimulated for 3 d with 1 µg/ml anti-CD3 (BD Biosciences, Heidelberg, Germany) in the presence of irradiated splenocytes. Proliferation was assessed as loss of the proliferation dye by flow cytometry.
Quantification of cytokines
Cytokine concentrations in cell supernatants were quantified by polystyrene bead–based Luminex technology (R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions. The assay was measured on a Luminex AtheNA Multi-Lyte System using Luminex IS software (Luminex Corporation, Austin, TX).
Adoptive T cell transfer
Percentages of TCR-HA+ T cells in sorted nNrp-1+CD4+CD25− and Nrp-1−CD4+CD25− T cells from naive TCR-HA/Thy1.1 mice, Nrp-1+CD4+Foxp3−(eGFP−)CD25− and Nrp-1−CD4+Foxp3−(eGFP−)CD25− T cells from TCR-HA/eGFP mice, or in iNrp-1+CD4+Foxp3− and Nrp-1−CD4+Foxp3− T cells sorted from splenocytes of TCR-HA/eGFP mice stimulated with the HA-peptide110–120 for 48 h were determined by flow cytometry. The absolute cell number was calculated, and an equivalent of 1 × 105 TCR-HA+CD4+ T cells were adoptively transferred i.v. into INS-HA/Rag2KO mice. Blood glucose concentrations were monitored using the Stat Strip Xpress-i (Nova Biochemical, Runcorn, Cheshire, UK). Mice were considered diabetic when glycaemia was >200 mg/dl.
Statistical analyses were performed with Student t test or one-way ANOVA with Tukey multiple comparisons test. Statistical significance was set at the levels of *p < 0.05, **p < 0.01, and ***p < 0.001. All analyses were calculated with GraphPad Prism Software (GraphPad Software, La Jolla, CA).
Nrp-1 expression is upregulated on activated CD4+ conv T cells in vitro and in vivo
Nrp-1 is a widely used marker molecule for Foxp3+ tTregs. However, we also detected significantly elevated percentages of Nrp-1–expressing cells upon stimulation of sort-purified Nrp-1−CD4+Foxp3−CD25− T cells from Foxp3eGFP reporter mice (Fig. 1A). Relative numbers of Nrp-1+CD4+Foxp3− T cells were significantly enhanced until day 3 after stimulation of CD4+Foxp3−CD25− T cells isolated from Foxp3eGFP mice (Fig. 1A). To investigate whether activated CD4+ conv T cells also upregulate Nrp-1 expression during an ongoing immune response in vivo, we made use of a murine model for autoimmune diabetes. For this purpose, we adoptively transfer-sorted Nrp-1−CD4+Foxp3−CD25− T cells from Foxp3eGFP/TCR-HA mice (Fig. 1B, before transfer day 0) to INS-HA/Rag2KO mice, resulting in the development of diabetes. The spleen and pancreas of diabetic mice were isolated, and HA-specific (TCR-HA+) CD4+Foxp3− T cells were analyzed for the expression of Nrp-1 (Fig. 1B, diabetes day 9). Approximately 10% and 1% of transferred HA-specific Nrp-1−CD4+Foxp3−CD25− T cells upregulated Foxp3 expression in the spleen and pancreas of diseased mice, respectively. The majority of these induced CD4+Foxp3+ Tregs expressed Nrp-1 on their surface, as expected (Fig. 1B, representative dot plots). Strikingly, the percentage of Nrp-1–expressing reisolated HA-specific CD4+Foxp3− T cells significantly increased in the spleen and pancreas of diseased mice, defined as mice with blood glucose levels ≥200 mg/dl (Fig. 1B). Relative numbers of Nrp-1–expressing HA-specific CD4+Foxp3− T cells from diabetic mice (Fig. 1B) were higher (∼20%) than those of Nrp-1+CD4+Foxp3− T cells induced by in vitro stimulation (∼10%) (Fig. 1A). This might reflect differences between the in vitro and in vivo situation, including continuous Ag expression and Ag-specific stimulation in the diabetes model, in contrast to anti-CD3 stimulation in vitro. Overall, our results indicate that activation of CD4+Foxp3−CD25− T cells resulted in significantly enhanced relative numbers of Nrp-1–expressing cells, referred to as iNrp-1+CD4+ T cells.
Stimulation-induced iNrp-1+CD4+ T cells exhibit a highly activated phenotype
To get further insights into the phenotype and function of activation-induced Nrp-1–expressing (iNrp-1+) CD4+ T cells, we isolated CD4+Foxp3−CD25− conv T cells from spleens of naive Foxp3eGFP mice (see cell sorting strategy in (Fig. 1A, top) and stimulated them in vitro with anti-CD3. At different time points postactivation, we analyzed the expression of different activation-associated molecules on gated iNrp-1+ and Nrp-1− CD4+Foxp3− conv T cells by flow cytometry. As expected, stimulation of CD4+Foxp3−CD25− T cells resulted in a significant upregulation of CD25 and CD44 expression (Fig. 2A). Interestingly, the expression of CD25 and CD44 was significantly higher on gated iNrp-1+CD4+Foxp3− T cells than on Nrp-1−CD4+Foxp3− T cells at 24 h and 48 h after stimulation (Fig. 2A). Next, we asked whether the iNrp-1+ subpopulation of stimulated CD4+Foxp3− T cells also differs from Nrp-1−CD4+Foxp3− T cells in terms of functional properties, and therefore we analyzed the cytokine profile and proliferative activity by flow cytometry. For this purpose, CD4+Foxp3−CD25− T cells were sorted from Foxp3eGFP mice and activated with anti-CD3. Flow cytometry analysis after 24 h and 48 h of stimulation revealed that the percentage of IFN-γ– as well as TNF-α–producing cells was significantly elevated within the gated iNrp-1+CD4+Foxp3− T cell subset compared with the Nrp-1−CD4+Foxp3− T cell population (Fig. 2B). Moreover, we analyzed the expression of Ki67 in unstimulated Nrp-1−CD4+Foxp3− T cells and upon stimulation as a measure for the proliferative capacity. As expected, Nrp-1−CD4+Foxp3− T cells upregulated the expression of Ki67 after stimulation with anti-CD3 (Fig. 2C). However, iNrp-1+CD4+Foxp3− T cells expressed significantly higher levels of Ki67 than stimulated Nrp-1−CD4+Foxp3− T cells, indicating enhanced proliferative activity after 48h of stimulation. These results suggest that Nrp-1–expressing CD4+Foxp3− T cells, which were induced upon stimulation of CD4+Foxp3−CD25− T cells, represent a highly activated subset of CD4+ effector T cells.
Impaired function of nNrp-1+CD4+Foxp3− T cells isolated from naive mice
We have demonstrated that the percentage of Nrp-1–expressing CD4+ conv T cells is significantly enhanced upon stimulation in vitro and in vivo. However, a small population of CD4+Foxp3−CD25− T cells from naive mice already expressed Nrp-1 under homeostatic conditions. To investigate whether this population is also present in organs other than the spleen, we analyzed the Nrp-1 expression of CD4+Foxp3−CD25− T cells and CD4+Foxp3+CD25+ Tregs (Fig. 3A, left panel) from different organs of naive BALB/c mice. We detected between 1 and 6% Nrp-1+CD4+Foxp3−CD25− T cells (nNrp-1+CD4+ T cells) in the colon, lung, thymus, blood, and spleen of mice, independent of their age (Fig. 3A, middle panel). As expected, the majority of CD4+Foxp3+CD25+ Tregs in different organs from naive mice expressed Nrp-1 (Fig. 3A, right panel). Next, we asked, whether the subpopulation of Nrp-1–expressing CD4+Foxp3−CD25− T cells from naive mice differs from activation-induced iNrp-1+CD4+Foxp3− T cells in their molecular and functional phenotype. To study this issue in more detail, we sorted Nrp-1+ and Nrp-1− CD4+Foxp3−CD25− T cells from the spleen of naive Foxp3eGFP mice according to the cell sorting strategy depicted in (Fig. 1A and analyzed the expression of activation-associated molecules upon anti-CD3 stimulation in vitro. As already shown in (Fig. 1A, activation of sorted Nrp-1−CD4+Foxp3−CD25− T cells resulted in the induction of Nrp-1 expression (Fig. 3B, left panel). The percentage of Nrp-1–expressing T cells remained stable within sorted nNrp-1+CD4+Foxp3− conv T cells (Fig. 3B, left panel). Interestingly, the expression of CD25 as well as CD69 was significantly lower in activated nNrp-1+CD4+Foxp3− T cells than in activated Nrp-1−CD4+Foxp3− T cells (Fig. 3B, middle and right panel). These results suggest an impaired activation of nNrp-1+CD4+ conv T cells from naive WT mice.
To elucidate whether Nrp-1 expression on CD4+ conv T cells from naive mice also influences the functional activity, we analyzed the proliferative capacity and cytokine production upon stimulation. nNrp-1+CD4+Foxp3− T cells and Nrp-1−CD4+Foxp3− T cells were isolated from the spleen of naive Foxp3eGFP reporter mice, labeled with CFSE, and stimulated for 72 h. The loss of CFSE was analyzed by flow cytometry as a measure for proliferation. As depicted in (Fig. 3C, stimulated nNrp-1+CD4+Foxp3− T cells showed a significantly reduced proliferation in comparison with activated Nrp-1−CD4+Foxp3− T cells (Fig. 3C). In contrast, the proliferative capacity did not significantly differ between Nrp-1−CD4+ T cells and CD4+ conv T cells that induced Nrp-1 expression (iNrp-1+) upon stimulation.
For cell sorting of nNrp-1+CD4+Foxp3− T cells, we used a fluorochrome-labeled anti–Nrp-1 Ab. To exclude that binding of this Ab to the Nrp-1 surface receptor might have an influence on the proliferation, we generated a Nrp-1tdTomato reporter mouse, allowing for sorting of Nrp-1+(tdTomato+) T cells without the use of an additional Nrp-1 binding Ab. Staining of Nrp-1 on CD4+ T cells from Nrp-1tdTomato reporter mice with an anti–Nrp-1 Ab revealed that almost all Nrp-1–expressing cells coexpress tdTomato (Fig. 3D).
Nrp-1+(tdTomato+)CD4+CD25− and Nrp-1−(tdTomato−)CD4+CD25− T cells were sorted from Nrp-1tdTomato reporter mice, labeled with CFSE, and left unstimulated as control or stimulated in the presence or absence of anti–Nrp-1 Abs for 72 h. nNrp-1+CD4+ conv T cells exhibited a significantly reduced proliferation in comparison with Nrp-1−CD4+ conv T cells upon stimulation, irrespective of the presence of a Nrp-1 binding Ab (Fig. 3D). Hence, the interaction of Nrp-1 with the anti–Nrp-1 Ab used for cell sorting in this study has no impact on the proliferation of nNrp-1+CD4+ conv T cells. In addition, we determined cytokine concentrations within the supernatants harvested from the different stimulated populations analyzed in (Fig. 3D. In line with impaired proliferation, we detected less IL-2, IFN-γ, and TNF-α in the culture medium of stimulated nNrp-1+CD4+ conv T cells than in medium of stimulated Nrp-1−CD4+ conv T cells (Fig. 3E). The presence of an anti–Nrp-1 binding Ab had no influence on the production of proinflammatory cytokines (Fig. 3E).
These data suggest that nNrp-1+CD4+ conv T cells from naive mice are not able to respond to TCR stimulation to the same extent as their Nrp-1−CD4+ T cell counterparts, and more importantly, they significantly differ from the iNrp-1+CD4+ conv T cell subset induced upon stimulation in vitro.
Elevated expression of exhaustion-associated molecules on nNrp-1+CD4+ conv T cells from naive mice
To gain further insights into the molecular differences between nNrp-1+CD4+ conv T cells from naive mice and iNrp-1+CD4+ conv T cells that recently acquired Nrp-1 expression due to stimulation in vitro, we analyzed the expression of exhaustion-associated molecules. For this purpose, we sorted nNrp-1+CD4+Foxp3−CD25− T cells and Nrp-1−CD4+Foxp3−CD25− T cells from splenocytes of naive Foxp3eGFP mice, left them untreated, or stimulated them with anti-CD3 for 24 h or 48 h. We determined the expression of CTLA-4 and PD-1 of sorted nNrp-1+CD4+Foxp3−CD25− T cells and Nrp-1−CD4+Foxp3−CD25− T cells from naive mice and of gated iNrp-1+ and Nrp-1−CD4+Foxp3− T cells upon activation of sorted Nrp-1−CD4+Foxp3−CD25− T cells by flow cytometry. Stimulation of nNrp-1+CD4+Foxp3−CD25− T cells as well as Nrp-1−CD4+Foxp3−CD25− T cells resulted in upregulation of CTLA-4 and PD-1 expression (Fig. 4A, 4B). However, nNrp-1+CD4+ conv T cells expressed significantly higher levels of both exhaustion-associated molecules than Nrp-1−CD4+Foxp3− T cells. Interestingly, this was true for stimulated cells but also at the unstimulated stage (0 h). Strikingly, although induction of Nrp-1 expression in recently activated iNrp-1+CD4+ conv T cells was accompanied by elevated expression of CTLA-4 and PD-1 compared with Nrp-1−CD4+ conv T cells that did not upregulate Nrp-1, the expression levels were significantly lower than in nNrp-1+CD4+ conv T cells from naive mice (Fig. 4A, 4B). Hence, our in vitro analyses provide evidence, that nNrp-1+CD4+Foxp3−CD25− T cells from naive mice have a different phenotype than iNrp-1+CD4+Foxp3−CD25− T cells that gained Nrp-1 expression in response to recent stimulation. Impaired capacity to become fully activated and enhanced expression of exhaustion-related molecules suggest a dysfunctional phenotype of nNrp-1+CD4+Foxp3−CD25− T cells.
nNrp-1+CD4+ conv T cells from naive mice do not induce diabetes, in contrast to the iNrp-1+ subset of recently activated CD4+ T cells
Finally, we asked whether nNrp-1+CD4+ conv T cells and iNrp-1+CD4+ conv T cells also differ in their functional properties during a more complex in vivo situation. Therefore, we made use of the established T cell transfer diabetes model. In this model, adoptive transfer of CD4+ T cells from TCR-HA mice into INS-HA/Rag2KO mice that express HA in the pancreas results in the development of diabetes (27). In TCR-HA mice, 10–15% of peripheral CD4+ T cells express the transgenic TCR, and HA-specific CD4+Foxp3− T cells showed comparable Nrp-1 expression as Ag-unspecific CD4+Foxp3− T cells that do not express the HA-specific TCR in these mice (data not shown). Upon transfer, Ag-specific T cells become activated and infiltrate the pancreas. Depending on the functional properties of transferred cells, diabetes development is delayed or induced more rapidly (28).
We sorted nNrp-1+CD4+CD25− T cells and Nrp-1−CD4+CD25− T cells from naive TCR-HA/Thy1.1 mice and adoptively transferred equal numbers of HA-specific CD4+ T cells i.v. into INS-HA/Rag2KO. Whereas, transfer of HA-specific Nrp-1−CD4+CD25− T cells resulted in the development of autoimmune diabetes from day 8 posttransfer on, mice that received HA-specific nNrp-1+CD4+CD25− T cells stayed healthy until day 9 posttransfer (Fig. 5A). Further analysis revealed significantly elevated relative numbers of Ag-specific CD4+ T cells expressing higher levels of Ki67 within the spleen and pancreas of INS-HA/Rag2KO mice transferred with HA-specific Nrp-1−CD4+CD25− T cells, in contrast to mice that received HA-specific nNrp-1+CD4+CD25− T cells (Fig. 5A). These results suggest that nNrp-1+CD4+CD25− T cells show an impaired response to Ag-specific stimulation in vivo, presumably due to impaired proliferative activity.
To further clarify, whether iNrp-1+CD4+Foxp3−CD25− T cells exhibit an enhanced activation state not only upon stimulation in vitro but also in the in vivo situation, we adoptively transferred sort-purified Nrp-1−CD4+Foxp3−CD25− T cells from TCR-HA/Foxp3eGFP mice (see cell sorting strategy in (Fig. 1B) into INS-HA/Rag2KO mice. The expression of activation-associated molecules on gated HA-specific Nrp-1−CD4+Foxp3− and HA-specific iNrp-1+CD4+Foxp3− T cells was analyzed in spleen and pancreas of diabetic mice (Fig. 5B). Consistent with the in vitro results (Fig. 2), we detected elevated expression of CD69, CD25, and CD44 on gated HA-specific iNrp-1+CD4+Foxp3− T cells in comparison with stimulated HA-specific CD4+Foxp3− T cells that did not upregulate Nrp-1 expression in diseased mice (Fig. 5B). Finally, to study the direct impact of Nrp-1 expression in recently activated CD4+Foxp3− T cells (iNrp-1) on diabetes development, we stimulated splenocytes from TCR-HA/eGFP mice with the cognate HA-peptide for 48 h and sorted iNrp-1+CD4+Foxp3− T cells and Nrp-1−CD4+Foxp3− T cells. Equal numbers of HA-specific CD4+ T cells from the respective cell subsets were adoptively transferred i.v. into INS-HA/Rag2KO mice. Well in line with our prior results, we detected elevated relative numbers of HA-specific T cells in the spleen and pancreas, accompanied by a more rapid development of diabetes in mice that received HA-specific iNrp-1+CD4+Foxp3− T cells compared with mice adoptively transferred with control HA-specific Nrp-1−CD4+Foxp3− T cells (Fig. 5C). In summary, these results demonstrate the functional impairment of Ag-specific nNrp-1+CD4+ conv T cells from naive mice to elicit an autoimmune response in vivo in contrast to activation-induced iNrp-1+CD4+ conv T cells, which aggravated the disease.
It is well established that the majority of murine CD4+ Tregs express Nrp-1, and therefore this surface receptor is widely used as maker for tTregs (7, 12, 13). In addition, a small subset of CD4+Foxp3− conv T cells from naive mice stably expresses Nrp-1 (14, 23), and it becomes more and more evident that T cell activation induces Nrp-1 expression in a subpopulation of non-Tregs (21, 22). However, the function of Nrp-1 in different T cell subsets is still discussed controversially.
Nrp-1 induction was detected on polyclonal tumor-infiltrating CD4+ T cells and CD8+ T cells in tumor-bearing mice and patients suffering from non–small cell lung cancer (18, 29). Interestingly, intratumoral Nrp-1+ T cells coexpressed molecules associated with an exhausted phenotype, such as PD-1, CTLA-4, and Tim-3 (18). Functional studies revealed that Nrp-1 expression interfered with the migration and cytotoxic function of Nrp-1+PD-1+ tumor-infiltrating CD8+ T cells, suggesting that Nrp-1 might act as an inhibitory receptor (18). However, a recent study demonstrated that CD8+ T cell–specific ablation of Nrp-1 has no impact on primary tumor growth but mediated protection from tumor rechallenge (29). During the acute phase of persistent viral infection, Nrp-1 expression was induced in CD8+ T cells, but ablation of Nrp-1 in CD8+ T cells had no impact on viral clearance. Nevertheless, depending on the time point, CD8+ T cell–specific ablation of Nrp-1 might affect the antiviral secondary response upon rechallenge (30).
In the current study, we demonstrated that Nrp-1 might have inhibitory as well as activating functions in CD4+ conv T cells, depending on the cellular origin. Whereas, nNrp-1+CD4+ non-Tregs from naive mice exhibited impaired proliferation and reduced expression of activation-associated molecules upon stimulation in vitro, activation-induced iNrp-1+CD4+ T cells showed a more effector-like phenotype in terms of proliferation and production of proinflammatory cytokines. Strikingly, we also observed these functional differences in the more complex in vivo situation as analyzed by the use of the T cell transfer diabetes model. Transfer of low numbers of HA-specific CD4+ T cells into INS-HA/Rag2KO mice expressing HA under control of the rat insulin promotor was sufficient to induce destruction of the pancreatic β cells, resulting in the development of type I diabetes (27). The onset and severity of the disease is dependent on the phenotype of transferred CD4+ T cells. HA-specific T cells with a more activated status induced diabetes at earlier time points (28). In accordance with results from our in vitro studies, transfer of HA-specific nNrp-1+CD4+ conv T cells from naive mice did not induce diabetes compared with transfer of HA-specific Nrp-1−CD4+ conv T cells. In contrast, mice that received activation-induced iNrp-1+CD4+ T cells developed diabetes more rapidly than INS-HA/Rag2KO mice adoptively transferred with preactivated Nrp-1−CD4+ T cells. In line with our results, Gaddis and colleagues (22) detected elevated relative numbers of Nrp-1–expressing CD4+Foxp3− T cells during atherosclerosis that exhibited a more activated, inflammatory phenotype compared with Nrp-1−CD4+Foxp3− T cells, and T cell–specific ablation of Nrp-1 expression ameliorated the disease. In contrast, transfer of nNrp-1+CD4+ conv T cells isolated from naive mice prolonged cardiac allograft survival (23) in immunocompetent mice and ameliorated the onset and severity of EAE (14). Hence, nNrp-1+CD4+ conv T cells from naive mice and activation-induced iNrp-1+CD4+ T cells clearly differ in their molecular and functional properties in vitro and in vivo.
Nrp-1 binds to a broad variety of different ligands, accounting for diverse biological functions. Nrp-1 on Tregs has been described to interact with Sema 4A, resulting in inhibition of the AKT signaling pathway and maintenance of Treg stability and function (16). Interaction of Sema 3A with T cells inhibited the proliferative response and production of proinflammatory cytokines by blocking TCR polarization, ZAP-70 phosphorylation (31), and the MEK/ERK1/2 pathway (32). Interestingly, this effect was not observed upon TCR-independent stimulation with PMA, indicating that the TCR signaling pathway is essential for Sema 3A–induced effects on T cell function (32). For CD8+ T cells, an impaired cytotoxic function of human CD8+ tumor-infiltrating lymphocytes in the presence of recombinant Sema 3A as well as a negative impact of Sema 3A on the migratory capacity of CD8+ CTLs in response to CXCL12 was described (18). Although VEGF binds to a different extracellular domain of Nrp-1 than Sema, the ligand also contributes to the migration of Nrp-1+ T cells. We demonstrated that Nrp-1+CD4+Foxp3+ Tregs infiltrate the tumor tissue in response to tumor-derived VEGF (15). Ablation of Nrp-1 from T cells or VEGF expression from tumor cells resulted in reduced numbers of Tregs within the tumor that were accompanied by elevated antitumoral immune response and reduced tumor growth (15). Moreover, Nrp-1+CD4+ T cells exhibited elevated migratory capacity toward VEGF in vitro, and activation-induced Nrp-1+CD4+ conv T cells showed an enhanced migration to the atherosclerosis-involved organs that express VEGF in diseased mice (22). Hence, the immunological context and cell type seem to determine the function of Nrp-1 within different T cell subsets.
Interestingly, in the current study, we demonstrated that, although both nNrp-1+CD4+ conv T cells from naive mice and activation-induced iNrp-1+CD4+ T cells express Nrp-1, they exhibited a completely different functional phenotype under the same stimulatory conditions and environment in vitro and in vivo. However, nNrp-1+CD4+ conv T cells expressed significantly increased levels of PD-1 and CTLA-4 in the naive state as well as upon TCR stimulation than iNrp-1+CD4+ conv T cells that were induced by recent activation. These molecules are upregulated upon stimulation of T cells and if expressed at the high levels widely used to identify exhausted T cells (33, 34). The high expression of PD-1 and CTLA-4 on nNrp-1+CD4+Foxp3− T cells suggests that these cells are dysfunctional. Indeed, we showed that nNrp-1+CD4+Foxp3− T cells respond insufficiently to TCR stimulation in terms of expression of activation-associated molecules, proliferation, and cytokine production.
It has been proposed that Nrp-1 could be transferred from DCs or monocytes to resting and activated human T cells by trogocytosis, an active cell membrane transfer mechanism (35). In addition, Campos-Mora and colleagues (36) speculate about release of Nrp-1+ extracellular vesicles from Nrp-1+ Tregs that modulate the phenotype and function of CD4+ conv T cells in an Nrp-1– dependent manner. Transfer of Nrp-1 from other cells might explain the presence of nNrp-1+CD4+ conv T cells in naive mice. However, stimulation of Nrp-1−CD4+Foxp3− T cells with anti-CD3 and anti-CD28 in the absence of any other immune cell population also resulted in Nrp-1 upregulation (data not shown). In accordance, purified human CD4+ T cells upregulated Nrp-1 expression upon stimulation, and the expression was even higher than in the presence of monocytes (19). These results argue against trogocytosis as an underlying mechanism for activation-induced Nrp-1 expression in CD4+ conv T cells. Hence, further analysis is necessary to clarify the mechanism of Nrp-1 induction in CD4+ non-Tregs.
Overall, our data indicated that stimulation of Nrp-1−CD4+ conv T cells results in the induction of Nrp-1 expression. This stimulation-induced iNrp-1+CD4+ T cell subset exhibited a highly activated phenotype in vitro and in vivo, in contrast to nNrp-1+CD4+ conv T cells from naive mice that showed a dysfunctional phenotype. Hence, Nrp-1 expression seems to have different functions in CD4+ conv T cells, depending on the cellular origin.
We are grateful to Christian Fehring and Witold Bartosik for excellent cell sorting and Christina Liebig for technical assistance.
This work was supported by Deutsche Forschungsgemeinschaft grants to W.H., A.M.W., and J. Buer (RTG1949).
The online version of this article contains supplemental material.
Abbreviations used in this article
experimental autoimmune encephalitis
conv T cell from naive mice
peripherally induced Treg
regulatory T cell
vascular endothelial growth factor
The authors have no financial conflicts of interest.