Notch Ligand Delta-like 4 Promotes Regulatory T Cell Identity in Pulmonary Viral Infection Free

This article is featured in In This Issue, p.1379

Notch is a highly conserved signaling pathway that contributes to cell differentiation and function. Engagement of Notch receptors (Notch 1–4) and Notch ligands (Delta-like ligand 1, 3, 4, Jagged 1, 2) initiates cleavage of the Notch intracellular domain. The Notch intracellular domain binds to recombination binding protein (RBP)-Jκ and activates the transcription of Notch target genes in association with a coactivator of the mastermind-like (MAML) family (MAML1–3). Notch signaling regulates various stages of T cell lineage development and differentiation, including early stages of thymocyte development, as well as CD4 Th differentiation (1, 2). In the priming stage of Th cell differentiation, Notch ligand Delta-like ligand 4 (Dll4) sensitized Ag responses and augmented T cell activation (3), whereas separate studies demonstrated that Dll4 and Notch could activate production of IFN-γ and the Th1 transcription factor T-bet (2–5). During Th2 differentiation, Jagged1 and intracellular Notch promoted Gata3 and Il4 expression (2, 6, 7), whereas Dll4 suppressed Il4, Il5, and Il13 (8, 9). Furthermore, Dll4 and Notch were reported to promote Th17 and Th9 differentiation by enhancing Il17a, Rorc, and Il9 expression (10, 11). Taken together, Notch can serve as an amplifier for persistent Th1, Th2, and Th17 cell differentiation (4), suggesting that it is not a skewing signal, but rather that it enhances coactivation. However, the role of Dll4 and Notch signaling in regulatory T (T_reg) cells remains unresolved. Jagged2 and specific receptors, Notch1 and Notch3, promoted the T_reg cell master transcription factor Foxp3 expression and T_reg cell survival (12–16), and RBP-Jκ was reported to directly bind the Foxp3 promoter and regulate Foxp3 transcription (16). In contrast, inactivating Notch signaling after Foxp3 is expressed enhanced T_reg cell numbers and promoted tolerance (17). Blockade of Notch receptors and Notch ligands expanded Foxp3⁺ T cell populations in vivo in experimental autoimmune encephalomyelitis, graft-versus-host disease (GVHD), and type 1 diabetes (18–21). However, the role of Notch ligands in T_reg cell development and their resistance to inflammation during infection has not been well defined.

Notch ligands can be induced on APC by pathogen-associated molecular patterns (2, 6). Pathogens themselves can also induce Notch ligands. Studies showed that respiratory syncytial virus (RSV) induced Dll4 expression on dendritic cells (DC) in vitro (8, 22), and Dll4 blockade exacerbated RSV-induced Th2 airway pathogenesis (8). Because T_reg cells are required to limit pulmonary inflammation and pathogenic Th2 responses during RSV infections (23–25), we hypothesized that initial exposure of Dll4 may modulate peripherally induced T_reg (iT_reg) cell differentiation, homeostasis, and stability to control the intensity of the immune response and lung pathology during RSV infection. In the present study, we report that Dll4 sustained CD62L^hiCD44^lo central T_reg (cT_R) cells and solidified iT_reg cell identity during infection. This study defines novel roles of Dll4 in iT_reg cell subset regulation and iT_reg cell stability.

Six- to eight-week-old female BALB/cJ and C57BL/6J mice were purchased from The Jackson Laboratory. Female CD45.1 (B6-Ly5.1/Cr) mice were purchased from Charles River Laboratories. Foxp3^eGFP mice (B6.Cg-Foxp3^tm2(EGFP)Tch/J, stock no. 006772) were bought from The Jackson Laboratory and bred in-house. CD4-specific double-negative MAML (DNMAML) mice (Cd4-Cre × R26^DNMAMLf) and CD4-specific Rbpj knockout mice (Cd4-Cre × Rbpj^f/f) were generated as described (26–28). All mice were housed in the University Laboratory Animal Facility under animal protocols approved by the Animal Use Committee at the University of Michigan.

RSV line 19 was a clinical isolate originally from a sick infant in the University of Michigan Health System to mimic human infection (29). BALB/cJ mice were anesthetized and infected intratracheally (i.t.) with 1 × 10⁵ PFU of line 19 RSV, as previously described (8). For Dll4 blockade in vivo, 2.5 mg of purified polyclonal anti-Dll4 Ab or control IgG were injected i.p. 2 h before RSV infection at day 0. The same doses of control or anti-Dll4 Ab were given on days 2, 4, and 6.

Left lobe of lung was fixed with 4% formaldehyde and embedded in paraffin, and 5 μm of section were stained with periodic acid–Schiff (PAS) to detect mucus.

RNA were extracted with TRIzol (Invitrogen) by following the manufacturer’s protocol, and 1 μg of total RNA was reverse transcribed to cDNA to determine gene expression using TaqMan gene expression primer/probe sets. Dll1, Dll4, Jag1, and Jag2 were detected by SYBR as described (2). Delta4 primers were 5′-AGGTGCCACTTCGGTTACACAG-3′ and 5′-CAATCACACACTCGTTCCTCTCTTC-3′. Muc5ac and Gob5 expression was assessed by custom primers as described (30). Detection was performed in an ABI 7500 real-time PCR system (Applied Biosystems). Gene expression was calculated using the ΔΔCt method and normalized with Gapdh as input control.

Mice lungs were chopped. Lung and mediastinal lymph nodes (mLN) were enzymatically digested using 1 mg/ml collagenase A (Roche) and 25 U/ml DNaseI (Sigma-Aldrich) in RPMI 1640 with 10% FCS for 45 min at 37°C. Tissues were further dispersed through an 18-gauge needle (10-ml syringe) and filtered through 100-μm nylon mesh twice. Cells (5 × 10⁵) from mLN cells were plated in 96-well plates and restimulated with 10⁵ PFU of RSV line 19 for 48 h. IFN-γ, IL-4, IL-5, IL-13, IL-17A, IL-10, and IL-9 levels in supernatants were measured with a Bio-Plex cytokine assay (Bio-Rad Laboratories).

Single-cell suspensions of lung and lymph nodes were stimulated with 100 ng/ml PMA, 750 ng/ml ionomycin, 0.5 μl/ml GolgiStop (BD Biosciences), and 0.5 μl/ml GolgiPlug (BD Biosciences) for 5 h (when mentioned). After excluding dead cells with Live/Dead fixable yellow stain (Invitrogen), cells were preincubated with anti-FcγR III/II (BioLegend) for 15 min and labeled with the following Abs from BioLegend: anti-B220 (RA3-6B2), CD3 (145-2C11), CD4 (GK1.5), CD8α (53-6.7), CD11b (M1/70), CD11c (N418), CD25 (PC61), CD44 (IM7), CD45 (30-F11), CD62L (MEL-14), CD69 (H1.2F3), CD103 (2E7), CD127 (SB/199), CCR7 (4B12), Dll1 (HMD1-3), Dll4 (HMD4-1), Gr-1 (RB6-8C5), I-A/I-E (M5/114.15.2), and ST2 (DIH9). For innate lymphoid cell staining, lineage markers were anti-CD3, CD11b, B220, Gr-1, and TER119. After 30 min of incubation at 4°C, cells were washed and proceed to intracellular staining.

For intracellular staining, cells were fixed and permeabilized with a transcription factors staining buffer set (eBioscience). Cells were labeled with Abs from eBioscience: Foxp3 (FJK-16s), IL-17A (eBio17B7), IL-13 (eBio13A), GATA3 (TWAJ), retinoic acid–related orphan receptor (ROR)γt (AFKJS-9), and GzmB (NGZB) for 30 min at room temperature. Flow cytometry data were acquired with an LSR II (BD Biosciences) or NovoCyte (ACEA Biosciences) flow cytometer and were analyzed with FlowJo software (Tree Star).

CD4⁺CD25⁻CD62L^hiCD44^lo naive T cells were enriched from spleen using a naive CD4 T cell isolation kit (Miltenyi Biotec) with >92% purity. Naive T cells were then plated and cultured in 24-well plates. Naive T cells (10⁶/0.5 ml) were stimulated with plate-bound anti-CD3 (2.5 μg/ml; eBioscience), soluble anti-CD28 (3 μg/ml; eBioscience), and plate-bound recombinant Dll4 (1.65 μg/ml or the dose mentioned; R&D Systems); to skew toward in vitro iT_reg cells, human TGF-β1 (2 ng/ml; R&D System) and mouse IL-2 (10 ng/ml; R&D Systems) were added at the same time; to restimulate toward in vitro–induced Th17 cells, mouse IL-6 (10 ng/ml; R&D Systems), human TGF-β1 (2 ng/ml; R&D Systems), anti–IFN-γ neutralizing Ab (10 μg/ml; eBioscience), anti–IL-4 neutralizing Ab (10 μg/ml; eBioscience), and anti–IL-12/23 p40 neutralizing Ab (10 μg/ml; eBioscience) were added at 5 × 10⁵/0.2 ml of viable iT_reg culture or 10⁵/0.2 ml sorted enhanced GFP (eGFP)⁺ iT_reg cells when mentioned.

Single-cell sorting was performed on a FACSAria II (BD Biosciences). DAPI⁻CD4⁺eGFP⁺ viable T_reg cells were sorted with >93% efficiency. A suppression assay was performed as described with small modifications (31). In brief, naive T cells isolated from CD45.1 mice were labeled with CellTrace Violet (CTV) (Invitrogen). Labeled CD45.1⁺ naive T cells (2.5 × 10⁴) were cocultured with serial-diluted eGFP⁺ T_reg cells in 96-well round-bottom plates. Dynabeads mouse T activator CD3/CD28 (0.625 μl; Invitrogen) was added to 0.2 ml of culture. After 72 h, cells were harvested and CD45.1⁺ responder cell proliferation was accessed by CTV dilution.

Data were analyzed by Prism 6 (GraphPad Software). Data presented are mean values ± SEM. Comparison of two groups was performed with an unpaired, two-tailed Student t test. Comparison of three or more groups was analyzed by ANOVA with a Tukey posttest. A p value <0.05 was considered significant.

Previously published data have shown that Dll4 expression was upregulated on DC after RSV infection in vitro (8, 22). To further characterize the expression of Dll4 during RSV infection in vivo, BALB/cJ mice were i.t. infected with 1 × 10⁵ PFU of RSV before harvesting of lung tissues at 2 and 6 d postinfection (dpi). Dll4 transcripts were upregulated significantly at 6 dpi, whereas transcripts for the other Notch ligands Dll1, Jag1, and Jag2 were not increased in the lung (Fig. 1A). After gating on MHC^hiCD11c⁺ DC (Fig. 1B), Dll4 expression level was increased consistently on MHC class II (MHC-II)^hiCD11c⁺ DC after 6 dpi (Fig. 1C). To further understand which type of DC expressed Dll4 after RSV infection, two populations of pulmonary DC were identified to determine Dll4 expression: MHC-II^hiCD11c⁺CD11b⁺CD103⁻ DC and MHC-II^hiCD11c⁺CD103⁺CD11b⁻ DC. CD11b⁺ DC expressed Dll4, whereas few of the CD103⁺ DC expressed Dll4 during RSV infection (Fig. 1D). When we examined the number of DC in lung and draining lymph node at 6 dpi, increasing numbers of Dll4⁺CD11b⁺ DC in both lung and mLN were detected as the infection progressed. In contrast, the number of Dll4⁺CD103⁺ DC was significantly lower and variable in lung or mLN postinfection (Fig. 1E). These data indicated that Dll4 was expressed on pulmonary DC, especially CD11b⁺ DC, after RSV infection in vivo.

Next, we explored the role of Dll4 in controlling RSV immunopathology by blocking Dll4 in vivo during RSV infection. We have previously demonstrated that our polyclonal anti-Dll4 Ab is specific for Dll4 and not other Notch ligands (8). The polyclonal anti-Dll4 Ab decreased the abundance of Notch target gene Hes1 transcripts in activated T cells (Th0) in coculture with fixed Dll4-expressing stromal cells (Supplemental Fig. 1A), indicating that anti-Dll4 Ab blocked Dll4-mediated Notch activation in vitro. We next aimed at blocking Dll4 in vivo by injecting 2.5 mg of purified anti-Dll4 Ab i.p. Mice were then infected i.t. with RSV 2 h after Ab injection at day 0, followed by i.p. injection of anti-Dll4 every other day. After 8 d, the abundance of Hey1–Notch target gene transcripts was decreased in the lung (Supplemental Fig. 1B) and sorted CD4 T cells (Supplemental Fig. 1C) of anti-Dll4–treated mice, suggesting that systemic administration of anti-Dll4 Abs inhibited Notch signaling activation in CD4 T cells in vivo. Dll4 blockade increased peribronchial infiltration by mononuclear cell clusters and drove mucus production in the airway as assessed in PAS-stained lung sections (Fig. 2A). As an indication of RSV-induced pathology, expression of the mucus-associated gene gob5 (mClca3) was substantially increased (Fig. 2B), but the viral protein mRNA expression measured with RSV-F and RSV-G were not significantly different at either 6 or 8 dpi (Fig. 2C, data not shown). Although Dll4 inhibition increased total CD4 T as described (Fig. 2D) (8), data also showed that more CD4 T cells acquired a CD44^hiCD62L^lo effector phenotype (Fig. 2E). Another activation marker CD25 was not changed in CD4 T cells at 8 dpi (data not shown). Thus, Dll4 inhibition significantly altered lung immunopathology during RSV infection. Besides CD4 T cells, RSV infection can also expand group 2 innate lymphoid cells (ILC2) between 4 and 6 dpi in lung to be mucogenic and immunopathogenic (32). In this study, we investigate whether Notch ligand Dll4 affects ILC2 homeostasis. Strikingly, Dll4 inhibition increased the number of lineage⁻CD45⁺CD127⁺CD90.2⁺ST2⁺CD25⁺ ILC2 in lung (Fig. 2F) and mLN (Fig. 2G) at 6 dpi. Thus, Dll4 blockade increased ILC2 in the lungs during RSV infection.

Excessive cytokine production is one of the key features leading to RSV pathogenesis. The IFN-γ promotes viral clearance whereas IL-4, IL-5, IL-13, and IL-17A are pathogenic (29–34). In this study, we determined the effect of Dll4 blockade on cytokine profiles in draining lymph nodes. Equal numbers of mLN cells were cultured and restimulated with 10⁵ PFU of RSV. In vivo Dll4 blockade during RSV infection led to significantly increased Th2 cytokines IL-5 and IL-13 as well as IL-17A whereas IFN-γ, IL-4, IL-9, and IL-10 were not significantly affected in RSV-rechallenged lymph node cells (Fig. 2H). To specify whether Dll4 regulated Th cell cytokine production in CD4 T cells, mLN cells were restimulated with PMA plus ionomycin and CD4 T cells were examined by flow cytometry. Dll4 blockade significantly increased IL-13⁺ and IL-17A⁺ CD4 Th cells in mLN at both 6 and 8 dpi (Fig. 2I, data not shown). These data indicated that Dll4 neutralization exacerbated RSV-induced immunopathology, as shown by elevated mucus production, activated CD4 T cells, and Th2-Th17 cytokine overproduction.

Several studies demonstrated that T_reg cells quickly accumulate in lymphoid tissues and in the airway to dampen activated T cell infiltration and Th2 responses (23, 24, 35), and RSV infection drove T_reg cells to Th2-like effector cells with impaired function (36). Our group showed that Dll4 inhibition exacerbated type 2 cytokine production and activated T cell infiltration, but little is known about the role of Dll4 in T_reg cell development, stability, and function. After targeting Dll4 as previously described, total T_reg cells did not change in mLN at 6 dpi (Fig. 3A). However, when gating on CD62L^hiCD44^lo central CD4 T cells (Fig. 3B), Dll4 inhibition decreased Foxp3⁺ cell frequency and expression (Fig. 3C, 3D) while decreasing expression of the CCR7 chemokine receptor on central T cells (Fig. 3E, 3F). CCR7-mediated signals recruit CD62L^hiCD44^lo T cells to the T cell zone, and CCR7 is a marker of cT_R cells (37). Dll4 neutralization decreased the cT_R cell populations (Fig. 3G), whereas effector T_reg (eT_R) cells were not altered in mLN (Fig. 3H). These data suggested that Dll4 specifically sustained cT_R cells in draining lymph node during RSV infection.

In contrast to its effects in lymph nodes, Dll4 neutralization enhanced the abundance of Foxp3⁺ T_reg cells in the lungs of infected mice (Fig. 4A) and CD44^hiCD62L^loFoxp3⁺ eT_R cells in lung at 8 dpi (Fig. 4B) without affecting cT_R cells (data not shown). Based on our previous finding that Dll4 blockade enhanced IL-13 and IL-17A production in CD4 T cells (Fig. 2I), we examined whether Dll4 can also perturb cytokine production in Foxp3⁺ CD4 T cells. Lung cells harvested at 8 dpi were restimulated and stained for intracellular Foxp3, IL-13, and IL-17A before flow cytometric analysis. In vivo Dll4 blockade significantly increased the abundance of IL-17A⁺Foxp3⁺ (Fig. 4C) and IL-13⁺Foxp3⁺ (Fig. 4D) but not IL-13⁻, IL-17A⁻, Foxp3⁺ T_reg cells (data not shown). To further characterize whether Dll4 regulates either Th17 or Th2 transcription factor coexpression in T_reg cells was examined. RORγt or GATA3 was costained with Foxp3 after restimulation. Dll4 neutralization increased the percentage of RORγt⁺Foxp3^lo cells and RORγt⁺Foxp3⁻ but not RORγt⁻Foxp3⁺ after 8 dpi (Fig. 4E, 4F). We did not detect significant changes in either GATA3⁺ T_reg cells or total GATA3⁺ CD4 T cells in both lung and lymph nodes after Dll4 inhibition (data not shown). These data suggested that Dll4 inhibition drove a Th17-like effector phenotype in T_reg cells during RSV infection in vivo.

To further investigate whether Dll4 regulated T_reg cell functional markers, several were examined, including ICOS, programmed cell death protein 1, granzyme B (GzmB), and neuropilin-1 (Nrp1). Dll4 inhibition decreased the expression of GzmB, whereas ICOS, programmed cell death protein 1, and Nrp1 expression was unchanged in Foxp3⁺ T_reg cells at 8 dpi (Fig. 4G, data not shown). It was previously shown that during RSV infection, GzmB was the critical functional molecule for T_reg cell function to limit the associated immunopathology (25). We found that Dll4 neutralization decreased the frequency of GzmB⁺ in Foxp3+ T_reg cells in the lung after 8 dpi (Fig. 4H), especially for GzmB⁺ eT_R cells (Fig. 4I). Our results indicate that Dll4 prevented Foxp3 T_reg cell acquisition of Th17-like effector phenotype, and it supported GzmB expression in T_reg cells in the lung during RSV infection.

Our data reveal differential effects of Dll4 in Foxp3⁺ T_reg cells in lymphoid versus nonlymphoid tissue during pulmonary infection. To further investigate Dll4 effects on T_reg cell differentiation in a controlled context, splenic CD4⁺CD25⁻CD62L^hiCD44^lo naive CD4 T cells were activated with anti-CD3/anti-CD28 with or without plate-bound recombinant Dll4, and skewed toward T_reg cells with TGF-β and IL-2. We first confirmed that Dll4 was able to activate expression of the Notch target gene Hes1 in naive CD4 T cells and Th0 and T_reg cells (Supplemental Fig. 2A–C) in a dose-dependent manner (Supplemental Fig. 2C) and activated intracellular domain of Notch1 cleavage (Supplemental Fig. 2D). Subsequently, naive T cells were skewed toward Th0 or iT_reg cells. Foxp3 expression was significantly increased with Dll4 activation under iT_reg cell skewing conditions (Fig. 5A). Importantly, Dll4 increased CD25⁺Foxp3⁺ iT_reg cells in a dose-dependent manner only in the presence of TGF-β (Fig. 5B). Furthermore, Foxp3 expression was significantly decreased with homozygous ROSA26-driven expression of the pan–Notch inhibitor DNMAML1 (Fig. 5C), and with inactivation of the Rbpj gene, encoding RBP-Jκ (Fig. 5D). As MAML1 and RBP-Jκ mediate transcriptional activation from Notch signaling, these data revealed that Dll4-activated Foxp3 expression was dependent on canonical RBP-Jκ/MAML–dependent Notch signaling. To further characterize Foxp3⁺ cells in this differentiation system, primary Dll4-activated iT_reg cell cultures were rested in IL-2 for 3 d without Dll4 (Fig. 5E). Dll4-treated iT_reg cells consistently maintained a higher percentage of Foxp3⁺ (Fig. 5F) and higher Foxp3 expression (Fig. 5G). To investigate whether Dll4 affected homeostatic features on T_reg cells during in vitro skewing, CD62L and CD44 expression was evaluated before or after the IL-2 rest phase. Naive CD4 T cells that were exposed to Dll4 during primary iT_reg cell skewing maintained more CD62L^hiCD44^loFoxp3⁺ cells consistently independent of the dose of IL-2 during the rest period (Fig. 5H) and at multiple time points after primary skewing (Fig. 5I). These data suggest that the presence of Dll4 during iT_reg cell activation and differentiation stabilized Foxp3 expression and sustained more T_reg cells with the CD62L^hiCD44^lo phenotype in vitro.

Our data showed that Dll4 inhibition decreased T_reg cell functional markers during RSV infection (Fig. 4), and Dll4/Notch directly sustained Foxp3 expression and cT_R cells (Fig. 5). These findings imply that Dll4 may change T_reg cell function. To further evaluate the function of iT_reg cells after Dll4 exposure, we sorted viable, Foxp3-eGFP⁺ iT_reg cells from iT_reg cultures on day 6 and cocultured these cells with CD45.1⁺ naive CD4 T cells as responders. CD45.1⁺ naive T cells were labeled with CTV and stimulated with anti-CD3/anti-CD28 mouse T activator for 72 h. Viable CD45.1⁺ cells were gated to examine CTV dilution as readout of cell proliferation (Fig. 6A). Responder cells that were cocultured with Dll4-treated iT_reg cells were less proliferative, especially at the 1:1 and 1:2 cell ratio of T_reg/T_naive cells (Fig. 6B), suggesting that Dll4-treated iT_reg cells were more suppressive than iT_reg cells without Dll4. Intriguingly, CD45.1⁺ responder cells were less viable when cocultured with Dll4-exposed iT_reg cells (Fig. 6C). Because one of the functions of T_reg cells is to consume survival signals from other T cells, we evaluated IL-2Rα (CD25) expression after 6 d of iT_reg cell skewing. Costimulation of iT_reg cells with Dll4 enhanced CD25 expression on Foxp3⁺ T_reg cells in vitro (Fig. 6D). These data indicated that Dll4-exposed iT_reg cells were more functional in vitro.

The data above showed that Dll4 inhibition increased Th17-like T_reg cells during RSV infection. To examine whether Dll4 and Notch signaling could inhibit acquisition of a Th17 effector phenotype in iT_reg cells, naive CD4 T cells were purified and cultured in iT_reg cell skewing conditions. Il17a expression was examined during iT_reg cell skewing. Inactivation of Notch increased Il17a expression during iT_reg cell skewing (Fig. 7A), whereas Foxp3 transcripts were decreased (Fig. 7B). Both DNMAML1 expression and Rbpj inactivation in CD4 T cells led to increased IL-17A in iT_reg⁺ Dll4-skewed cultures (Fig. 7C). These data suggest that intrinsic Notch activation inhibited Il17a expression while enhancing Foxp3 expression in iT_reg cells during differentiation. To further investigate whether Dll4 affects IL-17A production after T_reg cell differentiation, an equal number of cells were challenged under IL-6 plus TGF-β Th17 conditions for 3 more days after the standard 6 d iT_reg cell skewing culture. In Th17 restimulation, Dll4-treated iT_reg cells secreted significantly less IL-17A (Fig. 7D) with associated decreased RORγt⁺ cells (Fig. 7E). Furthermore, Dll4-exposed iT_reg cells retained a higher percentage and expression level of Foxp3 (Fig. 7F, 7G). These data suggest that Dll4-mediated signals during iT_reg cell differentiation led to less inflammatory Th17 skewing while maintaining T_reg cell commitment. To clarify whether Dll4-educated Foxp3⁺ iT_reg cells had less plasticity to become RORγt⁺, 10⁵ viable Foxp3-eGFP⁺ iT_reg cells were sorted after 6 d of skewing and cultured in Th17 conditions. Approximately 40% of eGFP⁺ iT_reg cells without Dll4 acquired RORγt⁺ expression, whereas Dll4-skewed eGFP⁺ iT_reg cells expressed significantly less RORγt⁺Foxp3⁺ (Fig. 7H). Although a similar amount of eGFP⁺ cells lost their Foxp3 expression to be Foxp3⁻ in either non-Dll4 or Dll4-educated iT_reg cells (data not shown), Dll4-treated iT_reg cells still retained more Foxp3⁺RORγt⁻ cells during Th17 skewing (Fig. 7I). These data suggest that Dll4-educated iT_reg cells had less plasticity in an inflammatory Th17 environment.

The objective of this study was to determine whether the Notch ligand Dll4 influenced Foxp3⁺ T_reg cell development, homeostasis, and function during pulmonary infection. Notch signaling has been suggested to be instrumental in Foxp3⁺ T_reg cell formation and maintenance (17, 38). Notch ligands—Dll1, Jagged1, and Jagged2—have been shown to regulate T_reg cell expansion and elicit suppressive activity (13, 16, 39). The present study suggests that Dll4 mechanistically regulates the stability of iT_reg cells. Previous studies showed that primary RSV infection enhanced Foxp3⁺ T_reg cells in the airway (24, 35). Our data indicate that Dll4 sustained Foxp3 expression especially in cT_R cells and altered homeostatic features and functional markers of T_reg cells. In vivo Dll4 inhibition allowed the acquisition of effector cytokine production in T_reg cells, increased their effector phenotype (CD44^hiCD62L^lo), and decreased functional T_reg cell markers (GzmB) during infection. In vitro skewing of iT_reg cells costimulated with Dll4 from naive T cells had enhanced regulatory function, retained their cT_R cell phenotype, and were more stable with less susceptibility to become RORγt⁺ Th17 cells. Importantly, stable and functional T_reg cells prevent autoimmunity and enhance immunity against chronic infection (40). Thus, identifying Dll4 as one stabilizing factor in autologous T_reg cells could potentially further optimize clinical efficacy of treatment protocols (41, 42).

Dll4/Notch signaling appears to have differential functions in various disease models. Using systemic blockade, Dll4 inhibition ameliorated GVHD, type 1 diabetes, and experimental autoimmune encephalomyelitis (18, 19, 21, 43, 44). In contrast, Dll4 blockade leads to exacerbated allergic airway disease and pulmonary RSV infection responses, as well as to increased bacillus Calmette–Guérin-induced granuloma formation (8, 45). These different results could be reconciled with a revised model that Notch signaling fine-tunes T cell responses to different environmental cues (4), and thereby plays a distinct role in different diseases, such as Ag-driven/infectious versus autoimmune or allogeneic responses. Recently, a novel study demonstrated that intrinsic Notch signaling limited peripheral T_reg cell maintenance during steady-state and in GVHD (17). In contrast to these latter findings, our results suggest that Dll4 and canonical Notch signaling sustained iT_reg cell differentiation and stability to prevent immunopathogenesis of pulmonary infection. Recent evidence supports that iT_reg cells are important for mucosal homeostasis, suppression of immune responses to environmental and food Ags, and to impede mucosal inflammation. In contrast, naturally occurring T_reg (nT_reg) cells prevent autoimmunity and alter the threshold of immune activation (46, 47). Notch signaling and Dll4 restrain nT_reg development in vivo (19), and GVHD was mainly prevented by nT_reg (48, 49) but not iT_reg cells due to lineage instability (50). To answer the question whether Dll4 differentially affects nT_reg versus iT_reg cells, we used Nrp1 to distinguish nT_reg versus iT_reg cells in vivo as described (51–54). Our preliminary findings showed that Dll4 inhibition decreased Nrp1⁻ T_reg cells during RSV infection but not Nrp1⁺ T_reg cells in the spleen (data not shown). Therefore, we speculate that Dll4/Notch may have a differential role on nT_reg versus iT_reg cells, thereby limiting nT_reg cells in GVHD but enhance iT_reg cells in RSV infection. Additional research needs to be done to elucidate this intriguing hypothesis.

RSV infection can induce IL-17A production (34, 55), and IL-17A facilitates mucus production to promote Th2-driven immunopathology (34). Furthermore, T_reg cells could be reprogrammed to Th17-like in the presence of appropriate inflammatory cytokines (56, 57). iT_reg cell plasticity is critical for conversion to IL-17A–producing Foxp3⁺ and RORγt⁺Foxp3⁺ populations under multiple microenvironments and diseases (58–60). In the present study, Dll4 prevented IL-17A⁺ T_reg cell development in infected mice. Moreover, Dll4-“educated” Foxp3-GFP⁺ iT_reg cells retained higher Foxp3 expression and resisted skewing toward Th17 in vitro. These results indicate that Dll4 modulates T_reg cell stability and plasticity toward Th17 in inflammatory environments.

Pathogenic RSV infection also exhibits a type 2–biased immune response (61). Our group has shown that Dll4 prevented Th2 cytokine secretion in CD4 T cells during RSV infection. Because both Notch signaling and T_reg cells have been reported to be involved in Th2 programs, it raised the question of whether Dll4 regulates Th2 cytokine production directly or indirectly through T_reg cell effects. Surprisingly, our data did not demonstrate a significant perturbation in GATA3⁺ Th2 lineage after Dll4 inhibition (data not shown). Instead, our data presented a dysfunctional T_reg cell and Th17-like effector phenotype after Dll4 inhibition. Because IL-17A can enhance Th2 responses in RSV infection and other pulmonary inflammation, these results may suggest that the Th2 response is a result of an indirect effect of reduced T_reg cell function and increased IL-17.

Besides Th2 cells, recent studies have demonstrated that RSV infection enhances ILC2 with a concurrent IL-13 increase (32). The present study indicated that Dll4 inhibition increased the number of ILC2 corresponding with elevated IL-5 and IL-13 levels. Notch activation has been shown to support ILC2 progenitor development in vitro (38, 62), but the in vivo role of Dll4 in ILC2 biology during infection has not been described. Our novel findings further highlight the importance of Dll4 in ILC2 homeostasis in vivo, but it is not clear whether the regulation is direct or indirect. Because iT_reg cells inhibit ILC2 and attenuate pulmonary inflammation (63), future studies will be necessary to specify the indirect contribution of iT_reg cells in restraining ILC2 during pulmonary infection.

The homeostatic features of Foxp3⁺ T_reg cells were characterized with CD62L, CD44, and CCR7. CD62L^hiCD44^loCCR7⁺ cT_R cells are quiescent, long-lived, and dependent on IL-2, whereas CD44^hiCD62L^loFoxp3⁺ eT_R cells appear to be shorter lived (37, 64). These two subpopulations favored distinct compartments, as cT_R cells homing to secondary lymphoid organs were found to be suppressive for autoimmunity (65). Lymph node sensitization is critical to mount a proper immune response because insufficient lymph node formation led to impaired Foxp3 T_reg cell function, promoted IL-17A–mediated pulmonary pathology, and initiated lymphoid cell clusters in the lung (66, 67). In the present study, data showed that Dll4 sustained the cT_R cell population in secondary lymphoid organs while preventing Th17 effector phenotype during RSV infection, representing a novel role for Dll4 in cT_R cell maintenance.

In addition to the alteration of the T_reg cell cytokine phenotype, the present study further suggests that Dll4 sustained T_reg cell function by influencing GzmB in vivo and CD25 in vitro. Intriguingly, GzmB in Foxp3⁺ T_reg cells controlled RSV-induced T cell infiltration in airways (25), and GzmB has been postulated to be a better functional marker of iT_reg cells during viral infection (68). Further investigations are needed to characterize the function of T_reg cell subsets during RSV infection. In the present study, our findings with regard to GzmB may provide a mechanistic explanation for Dll4 inhibition in exacerbating RSV-induced CD4 and CD8 T cell infiltration (8).

Taken together, our study demonstrates that Dll4/Notch promotes iT_reg cell development, homeostasis, stability/plasticity, and function in vitro and in vivo during RSV infection. Dll4 prevented RSV pathogenesis and supported iT_reg cell development in vitro. The effects of Dll4 on T_reg cells are an example of how Notch signaling can promote T cell homeostasis and disease regulation. By extending our understanding of Dll4 in T_reg cell stability and function, our results may provide a road for future translational research leading to a better therapeutic strategy.

We thank Dr. M.A. Schaller, Dr. D.E. de Almeida Nagata, Dr. Catherine Ptaschinski, and Dr. Ivan P. Maillard for helpful discussions.

The authors have no financial conflicts of interest.