B6 WT (C57BL/6) and Tcrd ^-/- (B6.129P2-Tcrd ^tm1Mom/J) mice were purchased from the Jackson Laboratory and bred in our facility under specific pathogen-free (SPF) conditions. Il17a/f ^-/- (B6) mice were received from Dr. Charles D. Surh (POSTECH, Korea). CD45.1/2 B6 mice were received from Dr. Sin-Hyeog Im (POSTECH, Korea). All mice were used at the age of 6-12 weeks unless indicated, and age- and sex-matched animals were used as controls. Germ-free (GF) mice were bred and used as previously described (24). All mouse experiments were performed using protocols approved by the Institutional Animal Care and Use Committees (IACUC) of the POSTECH.

Mice were intranasally administered with 250 μg of particulate matter (PM) in saline or saline alone as controls. PM was obtained from Sigma (PM10-like ERMCZ100-1VL and ERMCZ120-1VL) and used as a 1:1 mixture. All intranasal administration (20 μl/nostril) was performed under anesthesia (i.p.) with ketamine (Yuhan)/xylazine (Rompun, BAYER) solution, as described (21).

We used a previously described house dust mite (HDM)-induced mouse asthma model with minor modification (25). HDM (Dermatophagoides pteronyssinus) extracts were purchased from Greer laboratories. Mice were intranasally administered with 20 μg of HDM daily for 4 days and challenged again with 20 μg of HDM for 4 days later. Fine dust was administered via the intranasal route with 250 μg of PM.

Five-week-old B6 CD45.1/2 and CD45.2/2 female mice were joined together by parabiosis for 2 or 7 weeks, as previously described (11). Weight-matched mice were anesthetized and shaved. An incision was made along the side of each mouse and the skin was connected using surgical clips.

Single-cell suspensions were isolated and stained with fluorescein-conjugated antibodies. For cytokine detection experiments, lymphocytes were stimulated with Cell Stimulation Cocktail and protein transport inhibitors (eBioscience) for 4 hours. Cells were washed twice in FACS buffer and stained with surface markers for 30 min at 4°C. For intracellular staining, single-cell suspensions were surface-stained, fixed, and permeabilized with the eBioscience Foxp3 staining buffer set. Following antibodies were used; anti-CD4-BUV395 (BD, GK1.5), anti-CD8α-BV650 (BD, 53-6.7), anti-SiglecF-PE (BD, E50-2440), anti-CD11b-PerCP-Cy5.5 (BD, M1/70), anti-CD11c-PE-Cyanine7 (eBioscience, N418), anti-TCRβ-PE-CF594 (BD, H57-597), anti-Ly6G-APCCy7 (BD, IA8), anti-CD45.2-BV605 (BD, 104), anti-B220-BV711 (BD, RA3-6B2), anti-CD8-BV510 (BD, 53-6.7), anti-CD44-redFluor 710 (TONBO, IM7), anti-TCRβ-APCCy7 (BD, H57-597), anti-CD11c-BV650 (BD, HL3), anti-Thy1.2-BV786 (BD, 53-2.1), anti-GL3-BV421 (BD, GL3), anti-CD45.2-BV650 (Biolegend, 104), anti-CD11c-BV711 (BD, HL3), anti-GATA3-PE (ebioscience, TWAJ), anti-Tbet-PE-Cyanine7 (eBioscience, 4B10), anti-RORγ-PE-CF594 (BD, Q31-378), anti-PLZF- Alexa Fluor 647 (BD, R17-809), Zombie-Aqua (Biolegend), anti-RORγt- PerCP-Cy5.5 (BD, Q31-378), anti-IL-17A-BV650 (BD, TC11-18H10), anti-IFNγ-BV786 (BD, XMG1.2), anti-CD11b-BV711 (BD, M1/70), anti- IFNγ-PE (Invitrogen, XMG1.2), anti-proIL-1β- PE-Cyanine7 (Invitrogen, NJTEN3), anti-CD11c-APC (Invitrogen, N418), anti- CD121a (IL-1R, Type I/p80)-PE (Biolegend, JAMA-147), anti-IFNγ-BV421 (BD, XMG1.2), anti-GL3- PE-CF594 (BD, GL3), anti-CD24-BV605 (Biolegend, M1/69), anti- Vγ2 TCR-BV786 (BD, UC3-10A6), anti-CD23p19-Alexa Fluor 488 (Invitrogen, fc23cpg), anti-Ki-67- PerCP-eFluor 710 (Invitrogen, SolA15), anti-GL3-PE (BD, GL3), anti-IL-17F- Alexa Fluor 488 (Biolegend, 9D3.1C8), anti-Vγ1.1 TCR-BV421 (BD, 2.11), anti- Vγ1.1+ Vγ1.2 TCR-PE (Biolegend, 4B2.9), anti-Vγ3 TCR-BV510 (BD, 536), anti-CD45.1- Pacific Blue (Biolegend, A20), anti-CD3-APCCy7 (BD, 145-2C11), anti-Vδ6.3/2 TCR (BD, 8F4H7B7). 17D1 hybridoma (anti-Vγ6 antibody) and biotinylated anti-Vγ7 antibody was used as previously described (21). Cells were analyzed on an LSR (Foretessa BD Biosciences) and data were processed using FlowJo software (Tree Star).

Biotinylated PBS57 loaded CD1d monomers and 5-OP-RU loaded MR1 monomers were obtained from the tetramer facility of the US National Institutes of Health (NIH). Biotinylated monomers were tetramerized using streptavidin-phycoerythrin (PE) (Prozyme), streptavidin-allophycocyanin (APC) (Prozyme), and streptavidin-PE-Cy7 (BD). For simultaneous enrichment of NKT, MAIT, and γδ T cells, single cell suspensions of lung were stained with PBS57-CD1d PE-Cy7, 5-OP-RU-MR1 PE, and anti-TCRγδ (GL3) PE-TR and enriched with anti-PE microbeads (Miltenyi) according to the manufacturer’s instructions.

Mice were sacrificed at the indicated time points and BAL fluids were collected in 1 mL PBS. To remove circulating cells, 15 ml of PBS was injected into the heart after incision of the abdominal aorta. Harvested lung tissues were minced by McIlwain Tissue Chopper and digested in 5 mL of RPMI-1640 containing collagenase D (400 Mandl Units; ROCHE) and DNase I (1 mg/ml; 9003-98-9) on a shaker at 37°C for 45min, followed by filtration through a 70 μm strainer and 40%, 70% Percoll (Merck) gradient centrifugation (20 min at 2,000 rpm at room temperature). To isolate LP cells from the intestine, we followed previous report (26). Single-cell suspensions were prepared and separated by Percoll gradient centrifugation. Adipose tissues were minced and digested with collagenase type IV (100 units, Gibco) and collagenase D (400 Mandl Units, ROCHE) on a shaker at 37°C for 45min. The ears were excised and cut into small pieces. The ear epithelial cell layer was removed by vigorous stirring in PBS containing 3% FBS, 20 mM HEPES, 100 U/ml penicillin, 100 μg/ml streptomycin, 1 mM sodium pyruvate, and 10 mM EDTA at 37°C for 20 min. The tissue samples were then digested in PRMI containing collagenase type V (1 mg/ml, Sigma) at 37°C for 45 min. Total cells were counted using a VI-CELL Cell Viability Analyzer (BECKMAN COULTER) and stained for FACS analysis.

For intracellular cytokine staining of proIL-1β, single cells were primed with lipopolysaccharide (LPS, 10 ng/ml; Sigma) for 2 hours in 10% FBS and 1x penicillin and streptomycin (P/S) containing RPMI media, as previously described (27, 28). Cells were co-incubated with 1x Monensin (Biolgened, 420701) and 1x Brefeldin A (Sigma). Intracellular cytokine staining was surface stained, fixed with IC fixation buffer (eBioscience), and permeabilized with the staining buffer (eBioscience).

Immunofluorescence staining was performed as described previously (29), with modifications. Briefly, tissues were fixed with 4% paraformaldehyde (PFA) for 1 hour and snap frozen. Five micrometer tissue sections were blocked with 5% bovine serum albumin and goat sera (Jackson Laboratory) for 1 hour at 25°C and stained with antibodies. Images were obtained using Leica DM6B with THUNDER system.

Prism software (GraphPad, Version 8.4.2) was used for statistical analysis, and all data were represented as mean ± SD. Unpaired two-tailed t-tests and one-way ANOVAs were used for data analysis and the generation of P values. P < 0.05 was considered significant.

To analyze γδ T cells systematically, we used a combination of transcription factors and surface markers as previously described (21), and newly defined naïve γδT (TγδN) cells as PLZF^lo/-RORγt ^– Tbet ^– CD44^lo cells in the thymus ( Figure 1A , upper panels). We further analyzed TCR Vγ usage using a panel of antibodies specific for TCR Vγ1, Vγ1/2, Vγ4, Vγ5, Vγ6, and Vγ7 in a single staining panel. In this way, we analyzed γδ T cells in the thymus and periphery, and phenotyped different subsets of γδ T cells with different TCR Vγ chain usages, except TCR Vγ3, which is a pseudogene.

As previously reported (30), Tγδ17 cells mainly consist of PLZF^loVγ4⁺ and PLZF^intVγ6⁺ cells both in the thymus and periphery, including the lung and skin ( Figure 1A and Supplementary Figure 1 ). In the thymus, all subtypes of γδ T cells were present and most intraepithelial lymphocytes (IEL) γδ T cells were TBET⁺ Tγδ1 cells. The majority of thymic TγδN and Tγδ1 cells consisted of Vγ1⁺ and Vγ4⁺ cells, whereas more than half of IEL Tγδ1 cells expressed TCR Vγ7, indicating that the Vγ TCR usage of γδ T cells varies depending on the tissue type, despite the same effector lineage. As shown in previous studies (31), in the skin, GL3^hi dendritic epidermal T cells (DETC) were TCR Vγ5⁺ and GL3^int dermal γδ T cells were RORγt⁺ Tγδ17 cells expressing TCR Vγ4 or Vγ6 ( Figure 1A , lower right panels).

In the thymus and periphery of SPF mice, naïve and effector subsets of γδ T cells were variably distributed, except Tγδ2 cells that were exclusively present in the thymus ( Figure 1B and Supplementary Figure 2 ). Because γδ T cells are affected by commensal microbiome (11, 32), we compared their subset distributions between SPF and GF mice ( Figure 1B and Supplementary Figure 2 ). Notably, GF condition most prominently affected the numbers and frequencies of Tγδ17 cells in the lung and small intestinal lamina propria (siLP), in which commensal or foreign micro-organisms are abundant ( Figure 1C ). TγδN and Tγδ1 cells were also slightly reduced in the spleen, mesenteric lymph node (mLN), and IEL of GF mice compared SPF control. Interestingly, there were decreased numbers of Tγδ2 cells, but not other subsets in thymi of GF mice ( Figure 1D ), indicating that commensal microbiota affects thymic development of γδ T cells. We further investigated the effect of microbiome on Vγ TCR chain usage; however, there were no noticeable differences in the thymus and periphery between SPF and GF mice ( Figure 1E , Supplementary Figure 3 and Supplementary Table 1 ). Overall, these findings indicate that TCR Vγ repertoire determined in the thymus is not affected by commensal microbiome in the periphery, suggesting that innate signaling rather than TCR engagement by specific antigens regulates the peripheral pool of γδ T cells.

Because maternal commensal microbiome affects the fetal immune system (33–35), we analyzed its effect on the development of Tγδ17 cells using new-born (day 1) mice, which have a fetal repertoire of γδ T cells. The numbers of thymic Tγδ17 cells were not different between SPF and GF mice at all ages ( Figure 2A ), and there were no substantial differences in their numbers and TCR Vγ usage of thymic immature Tγδ17 (CD24^hi RORγt⁺) and mature Tγδ17 (CD24^lo RORγt⁺) cells in the neonatal mice ( Supplementary Figures 4A, B ). In 3-week-old pre-weaned GF mice, there was increased usage of TCR Vγ1 in Tγδ17 cells compared SPF mice ( Figure 2B and Supplementary Figures 4C, D ). In the periphery, Vγ4⁺ or Vγ6⁺ peripheral Tγδ17 cells were variably decreased in GF mice compared to SPF mice ( Figure 2C ).

The development of mucosal associated invariant T (MAIT) cells is dependent on the microbiome, and later colonization of GF mice failed to reconstitute their development (36). We tested this in γδ T cells by cohousing 6-week-old GF mice with SPF mice for 6 weeks ( Figure 2D ). However, unlike MAIT cells, in these mice, we observed that not only peripheral Tγδ17 cells, including the lung and siLP ( Figure 2E ), but also thymic Tγδ2 ( Figure 2F ) and IEL Tγδ1 ( Figure 2G ) cells were all restored to equivalent levels of SPF mice. This features indicate that later colonization of the commensal microbiome is sufficient for the restoration of γδ T cells in adulthood.

γδ T cells are generally known to be tissue resident. To better understand the circulating dynamics of each subset of γδ T cells in the periphery, we generated a parabiosis model using C57BL/6 congenic mice and analyzed them 2 or 7 weeks later ( Figure 3A ). We first confirmed that 50% of B cells in the LN were from paired parabionts ( Figure 3B and Supplementary Figure 5A ) and analyzed γδ T cells. Interestingly, 50% of TγδN cells in most lymphoid and non-lymphoid organs, except thymus and siLP, were derived from paired parabionts, indicating that they are a circulating population similar to B cells. Tγδ17 cells were mostly tissue resident, especially in fat, ear skin, siLP, and lung. Tγδ1 cells were also tissue resident, especially in IEL and siLP. Generally, Tγδ1 cells showed a less tendency of tissue residency compared to Tγδ17 cells ( Figures 3C, D ). As known that Vγ5⁺ DETCs are only generated during the fetal period and reside in the skin (37–39), 97% of them were tissue resident at 2 and 7 weeks after parabiosis ( Supplementary Figure 5 ). Therefore, each subset of γδ T cells has different residential or circulatory characteristics with some variability depending on the tissues they localize. Consistent with a previous report (40), we also observed that invariant natural killer T (iNKT) cells, including both NKT1 and NKT17 cells, are tissue resident ( Supplementary Figure 6 ). However, there was not much difference in their tissue residency between NKT1 and NKT2 cells at the second week of parabiosis, and there were no naïve NKT cells. Taken together, unlike previous thoughts, these findings indicate that each subset of γδ T cells has unique pattern of tissue residency, that is Tγδ17 cells are mostly resident in the peripheral tissues compared to Tγδ1 cells, and TγδN cells are circulating cells.

We additionally compared tissue residency of γδ T cells using intravascular staining of anti-CD45 antibodies ( Figures 3E, F and Supplementary Figure 7 ) and found no significant differences between SPF and GF mice. Although intravascular staining does not necessarily differentiate tissue resident population as some cells are intravascular resident, this result suggests that the absence of microbiome does not affect circulating tendency of γδ T cells.

Tγδ17 cells are activated and rapidly produce IL-17 in response to IL-1 without TCR engagement (32, 41, 42). In the lung, γδ T cells express copious amounts of IL-1R and their over activation due to excessive IL-1 leads to poor control of lung adenocarcinoma (8, 43). Based on this, we further analyzed the expression pattern of IL-1R in γδ T cells and compared it with those in other types of innate T cells such as NKT, MAIT, conventional CD4 T cells, and innate lymphoid cells (ILCs) ( Figures 4A, B and Supplementary Figure 8 ). Frequencies of IL-1R expression in type 17 innate T cells including Tγδ17, NKT17, and MAIT17 cells and in ILC3s were comparable with one another at approximately 60–70%, whereas only about 20% of conventional Th17 cells expressed IL-1R. However, the number of IL-1R-expressing cells was highest in Tγδ17 cells occupying 68% ( Figure 4B ). In addition, pulmonary Tγδ17 cells expressed the highest level of IL-1R compared to those in thymus, spleen, and mediastinal lymph node (medLN) ( Figure 4C ). Together, these findings suggest that Tγδ17 cells would be the main population that responds to exogenous IL-1 and produces IL-17.

To investigate the pathogenic role of IL-1R⁺ lung-resident Tγδ17 cells in vivo, we used a mouse model of PM-induced acute airway inflammation. We intranasally administered mice with 250 μg of PM and analyzed at each time points after exposure. Lung epithelial cells are known to produce IL-1β in response to PM (14) and we further analyzed CD45⁺ leukocytes by flow cytometry. The mean fluorescence of intensity of IL-1β was sharply increased, as well as, the number of IL-1β-producing cells was increased after PM exposure ( Supplementary Figure 9A ). We analyzed the intracellular IL-1β in the CD45⁺ leukocytes using a gating strategy as depicted in Supplementary Figure 9B to include T cells, B cells, AM, interstitial macrophages (IM), neutrophils, and other undefined CD11b⁺ cells. After 4 hours of PM exposure, AM and neutrophils remarkably produced IL-1β ( Figures 5A, B ). Interestingly, the major cellular sources of IL-1β were neutrophils (33%) and CD11b⁺ cells (37%) in normal lungs, and neutrophils produced most of IL-1β (approximately 80%) in PM-exposed lungs ( Figure 5C ). Consistent with previous reports (18, 19), we found that PM causes an acute expansion of alveolar macrophages and strong neutrophilia in the airways ( Figure 5D ). The kinetics of neutrophil influx was similar to that of alveolar macrophages and peaked at 12 hours after PM administration ( Figure 5E ).

Since IL-17-producing Tγδ17 cells are associated with neutrophilia in the lung after bacterial or viral infection (44, 45), we next examined whether PM induces the production of IL-17 from γδ T cells. Using a gating strategy as shown in Supplementary Figure 10 , we found that the frequencies of IL-17-producing Tγδ17, MAIT17, and NKT17 cells significantly increased ( Figures 5F–H ), whereas there were no changes in IL-17 production from Th17 cells and ILC3s ( Figures 5I, J ) 24 hours after PM exposure. We also confirmed that the total number of IL-17-producing cells was approximately 2.68 times higher in PM-treated lungs compared to that in PBS-treated group ( Figure 5K , pie charts). Notably, we discovered that Tγδ17 cells produce 75% of IL-17 under both normal and inflammatory conditions ( Figure 5K ). PM exposure not only enhanced IL-17 production from Tγδ17 cells ( Figure 5F ), but also expanded their numbers upon its consecutive exposure for 4 days ( Supplementary Figures 11A, B ). However, numbers of Tγδ1 or TγδN cells were not increased and there were rather decreased IFNγ secretion from Tγδ1 cells ( Supplementary Figures 11C, D ). Taken together, these findings indicate that innate T cells, but not Th17 CD4 T cells or ILC3s, are the source of early IL-17 upon PM exposure.

We showed that homeostasis of Tγδ17 cells is dependent on commensal microbiomes ( Figures 1 , 2 ), and PM induces acute neutrophilia with Tγδ17 expansion ( Figure 5 ). Therefore, we tested whether GF mice have reduced neutrophilic inflammation upon PM exposure ( Figures 6A, B ). Indeed, GF mice had reduced infiltration of neutrophils and AMs upon PM exposure. Immunofluorescence staining of the lungs revealed that neutrophils clustered together with Tγδ17 cells around airways in SPF mice and GF mice had less infiltration of these cells ( Figure 6C ). Taken together, these results show that commensal microbiomes regulate neutrophilic inflammation upon PM exposure, which is likely to be mediated by Tγδ17 cells.

To obtain direct evidence that Tγδ17 cells are associated with the pathogenesis of PM-induced airway inflammation, we investigated and compared the severity of neutrophilic inflammation between B6 wild-type (WT) and TCRδ-deficient (Tcrd ^-/- ) mice 24 hours after PM administration. We found that Tcrd ^-/- mice showed significantly decreased neutrophilia ( Figures 7A, B ) without affecting the frequencies of IL-17-producing MAIT and iNKT cells compared to those of WT mice ( Figures 7C–E ). We and others have previously shown that MAIT cells expand in the absence of NKT or γδ T cells in the thymus and skin (21, 36). Consistent with these findings, the number of MAIT17 cells increased three times in lung of Tcrd ^-/- mice. However, they could not compensate the absence of γδ T cells and there was an average 6.7-fold reduction of IL-17-producing cells in the lung after PM exposure ( Figure 7F ). We also confirmed that neutrophilic inflammation was significantly relieved in Il17a/f-double knockout mice ( Supplementary Figure 12 ). Collectively, these findings indicate that Tγδ17 cells play a major role in acute neutrophilia induced by PM exposure.

We further analyzed the effect of γδ T cells in a chronic allergic asthma model induced by HDM and PM ( Figure 8A ). Previous reports showed that diesel dust converted allergic asthma from a Th2 to Th17-dominant inflammatory model (18, 19), and we also found that co-administration of HDM and PM induced the dominant expansion of RORγt⁺ CD4 T cells ( Figure 8B ). In Tcrd ^-/- mice, however, there was no decreased infiltration of neutrophils or other immune cells ( Figures 8C, D ), indicating that γδ T cells do not influence the chronic model of Th17-dominant inflammation.