γδ T Cells Mediate Protection against Neutrophil-associated Lung Inflammation in Pulmonary Melioidosis

Abstract

Pulmonary melioidosis is a severe tropical infection caused by Burkholderia pseudomallei and is associated with high mortality, despite early antibiotic treatment. γδ T cells have been increasingly implicated as drivers of the host neutrophil response during bacterial pneumonia, but their role in pulmonary melioidosis is unknown. Here, we report that in patients with melioidosis, a lower peripheral blood γδ T-cell concentration is associated with higher mortality, even when adjusting for severity of illness. γδ T cells were also enriched in the lung and protected against mortality in a mouse model of pulmonary melioidosis. γδ T-cell deficiency in infected mice induced an early recruitment of neutrophils to the lung, independent of bacterial burden. Subsequently, γδ T-cell deficiency resulted in increased neutrophil-associated inflammation in the lung as well as impaired bacterial clearance. In addition, γδ T cells influenced neutrophil function and subset diversity in the lung after infection. Our results indicate that γδ T cells serve a novel protective role in the lung during severe bacterial pneumonia by regulating excessive neutrophil-associated inflammation.

Pulmonary melioidosis is a tropical infection caused by the facultative intracellular pathogen Burkholderia pseudomallei (1). This severe pneumonia is frequently lethal, with mortality rates ranging from 22–50%, despite appropriate empiric antibiotics (2). Because of concern for its use as an aerosolized bioweapon, the Centers for Disease Control and Prevention (CDC) have classified B. pseudomallei as a Tier 1 select agent (3). Melioidosis is also emerging globally. For example, two autochthonous cases of pulmonary melioidosis were recently reported in the United States, resulting in the CDC declaring the pathogen to be endemic to the continental United States for the first time (4). As cases worldwide are likely underreported, novel therapeutic strategies are urgently needed for this burgeoning global health threat (2).

Innate-like lymphocytes, including γδ T cells, are small subsets of lymphocytes increasingly implicated in the early immune response to severe infections (5). γδ T cells, so named because of the components of the conserved TCR (T-cell receptor), can recognize microbially derived peptides, including phosphoantigens (6). Unlike adaptive αβ T cells, γδ T cells are often localized to the mucosa and can be activated rapidly after infection (7). In addition, although γδ T-cell antigen recognition does not depend on antigen-presenting cells (APCs), they can act as APCs for other lymphocytes, including γδ T cells (8, 9). The understanding of peripheral γδ T-cell differentiation is incomplete, but naive cells may be stimulated by an antigen, an APC, or costimulated through TLRs (Toll-like receptors) to develop an effector (CD45RA⁻,CCR7⁻) phenotype (10, 11). Effector cells may proliferate, home to sites of inflammation, and be activated by IL-23 and IL-1β to express either IL-17 or IFN-γ (12). However, this model has limited evidence in humans, and tissue subsets of γδ T cells may have variable roles during infection (6).

The role of γδ T cells in the mucosal immune response during lung infection is not completely understood, although they appear to be critical to successful bacterial clearance. For example, in models of Pseudomonas aeruginosa and Staphylococcus aureus pneumonia, mice lacking γδ T cells have worse survival and impaired bacterial clearance (13, 14). γδ T cells, together with TH17 cells, may be a primary producer of IL-17 during certain acute inflammatory conditions (15). Tissue-resident IL-17⁺ γδ T cells may respond quickly to IL-23 during a bacterial infection (16). In models of pneumococcal and P. aeruginosa pneumonia, however, IL-17 production may be pathogenic, leading to tissue damage (17, 18). These discrepancies suggest that the IL-17 γδ T-cell response within the lung is likely pathogen specific.

A postulated mechanism by which γδ T cells regulate pulmonary bacterial clearance is through IL-17–dependent neutrophil recruitment to the lung. In animal models of Nocardia asteroides pneumonia, for example, neutrophil recruitment in mice lacking γδ T cells is notably reduced (19). However, the exact mechanisms responsible for γδ T cell–neutrophil interactions during lung infection remain limited.

Little is known about the role of γδ T cells in pulmonary melioidosis. However, neutrophil recruitment appears critical to the pulmonary defense to B. pseudomallei (20, 21). Therefore, we hypothesized that γδ T cells are important mediators of the host response to pulmonary melioidosis by modulating neutrophil recruitment to the lung, in part by IL-17 production. To investigate this hypothesis, we first analyzed the association of γδ T-cell phenotypes with clinical outcomes in patients hospitalized with melioidosis. We next leveraged multiple in vivo models of pulmonary melioidosis to further understand the mechanisms by which γδ T cells regulate inflammation early in severe pulmonary infection.

Methods

Mice

Male and female C57BL/6 mice and Tcrd^−/− mice on a C57BL/6 background were purchased from Jackson Laboratory and bred in a barrier- and specific pathogen–free facility in the Department of Comparative Medicine at the University of Washington. Male and female homozygous mice between 8 and 12 weeks of age were used for experimental purposes.

Infection Model

Mice were infected in a whole-body aerosolization chamber using an aerosol generated by a Mini-heart Hi-Flo continuous nebulizer (Westmed) controlled by the Biaera AeroMP system (Biaera Technologies). Nebulizer airflow was maintained at 8 L/min over 15 minutes, followed by a 5-minute air purge. Bacterial deposition was determined by culture of homogenized left lung tissue from three to four mice immediately after infection. Mice were infected with either Burkholderia thailandensis E264 or B. pseudomallei 1026b. After infection, mice were examined daily for illness or death. Investigators blinded to mouse genotype performed daily monitoring of abdominal surface temperatures, weights, and a 7-point health score (consisting of 0–1 point for eye crusting, hunched posture, isolation, lack of resistance to handling, labored breathing, reduced activity, and ruffled fur). Animals with a surface temperature of <24°C, weight loss of >25% from baseline, or health score of <3 were killed. Experimental animals requiring organ procurement were killed by an intraperitoneal overdose of pentobarbital. All procedures were approved by the Institutional Animal Care and Use Committee of the University of Washington.

Mouse Tissue Procurement

At serial time points after infection, mice were killed, cardiac exsanguination performed, chest and trachea exposed, and the left lung tied off at the hilum and removed. After insertion of a catheter into the trachea, the right lung was lavaged with four 0.5-ml washes of 0.9% NaCl + EDTA. BAL fluid (BALF) supernatant was isolated after centrifugation and frozen at −80°C. The right lung was subsequently inflated to 15 cm of water pressure with 10% formalin and then immersed in the same solution. Spleens were also removed. Procured left lung and spleen were mechanically homogenized in PBS and bacterial counts determined by culture. Left lung samples were then diluted 1:1 with a lysis buffer containing 1% Triton X and protease inhibitors for 30 minutes on ice, and supernatants were frozen at −80°C.

Additional methods, including those related to mouse lung histology, biomarker measurements, mouse lung single-cell processing, mouse lung transcriptomics, human peripheral blood processing, and statistical analyses can be found in the data supplement.

Study Approvals

The recruitment and blood sampling of human subjects in northeastern Thailand was approved by the institutional review board/ethics committees of Faculty of Tropical Medicine, Mahidol University, and Udon Thani Hospital. The study was conducted according to the principles of the Declaration of Helsinki (2008) and the International Conference on Harmonization Good Clinical Practice guidelines. Written informed consent was obtained from all individuals enrolled in the study or their representatives. Animal experiments were conducted in accordance with and approved by the University of Washington Institutional Animal Care and Use Committee.

Results

Melioidosis Survival Is Associated with an Upregulation of Effector γδ T Cells

As γδ T-cell upregulation has previously been associated with bacterial infection, we first sought to determine whether patients hospitalized with melioidosis display a γδ T-cell response during the course of their illness. Using whole blood obtained at the time of enrollment from patients hospitalized in northeastern Thailand with bacteremic melioidosis, we performed flow cytometry to identify Vδ2⁺ γδ T-cell populations, the most common peripheral blood γδ T-cell subset in humans. In addition, we tested blood at 28 days after enrollment in patients surviving to that endpoint, as well as whole blood collected from healthy volunteers. We analyzed enrollment samples from 32 patients with melioidosis, of whom 11 died by 28 days. At enrollment, nonsurvivors had significantly lower concentrations of Vδ2⁺ γδ T cells and CD45RA⁺/CCR7⁺ Vδ2⁺ γδ T cells compared with 28-day survivors (P < 0.05; Figure 1A; see Figure E4E in the data supplement). Intriguingly, the concentration (P < 0.01) and percentage (P < 0.001) of CD45RA⁻/CCR7⁻ Vδ2⁺ γδ T cells, reflecting an effector phenotype, was significantly higher 28 days after enrollment in 28-day survivors (Figures 1B and E5E). Finally, when compared with healthy volunteers, survivors had a persistent increase in whole blood Vδ2⁺ γδ T-cell concentration (P < 0.001; Figure E4B) and percentage (P < 0.01; Figure E5G), as well as effector phenotype CD45RA⁻CCR7⁻ Vδ2⁺ γδ T-cell concentration (P < 0.01; Figure E4C) at 28 days after enrollment, despite a similar enrollment concentration of CD3⁺ cells (Figure E4A). Finally, we determined the intracellular expression of IL-17 and IFN-γ in Vδ2⁺ γδ T cells after stimulation with B. pseudomallei LPS (Bp-LPS). Both the concentration and relative frequency of IL-17⁺ Vδ2⁺ γδ T cells (Figures 1C and 1D) were increased in the peripheral blood of patients with melioidosis compared with healthy control subjects (Figures 1C and 1D; P < 0.01). The concentration of IL-17⁺ Vδ2⁺ γδ T cells remained persistently increased at 28 days after enrollment (Figure 1E). IFN-γ⁺ Vδ2⁺ γδ T cells were infrequent after Bp-LPS stimulation (Figure 1F).

Figure 1.

Melioidosis survival is associated with a persistent increase in peripheral blood γδ T cells with an effector phenotype. Whole blood samples were prospectively collected from adults hospitalized in northeastern Thailand with bacteremic melioidosis within 24 hours of culture positivity (n = 32). Repeat blood samples were collected at 28 days after enrollment in surviving subjects. Control whole blood was also collected from healthy adults recruited at the same hospital (n = 10). Red blood cells were lysed on collection and samples fixed and frozen until staining and cellular identification by flow cytometry. Concentrations of cell populations were calculated using counting beads. Concentration of Vδ2⁺ γδ T cells and CD45RA⁻/CCR7⁻ Vδ2⁺ γδ T cells from unstimulated enrollment whole blood samples from (A) 28-day nonsurvivors (D0-NS) and 28-day survivors (D0-S), as well as (B) Day 28 samples in survivors (D28-S). In samples stimulated with Burkholderia pseudomallei–LPS, intracellular IL-17 and IFN-γ were assessed by flow cytometry after gating for Vδ2⁺ γδ T cells. The concentration and percentage of (C–E) IL-17⁺ Vδ2⁺ γδ T cells and (F) IFN-γ⁺ Vδ2⁺ γδ T cells are shown in different populations, including healthy subjects (Healthy). Data are presented as median and interquartile range (IQR). For comparisons of more than two groups, the Kruskal-Wallis test was performed, followed by the Dunn’s test for multiple comparisons; comparisons of two groups used Mann-Whitney U (unpaired) or Wilcoxon tests (paired). *P < 0.05, **P < 0.01, and ***P < 0.001.

We next sought to better define the association of Vδ2⁺ γδ T cells with outcome in the melioidosis cohort by developing logistic regression models. Vδ2⁺ γδ T-cell concentration was inversely associated with the odds of death at 28 days in an unadjusted model (odds ratio [OR], 0.91; 95% confidence interval [CI], 0.82–1.00; P = 0.04), a model adjusted for the presence of shock (adjusted OR, 0.90; 95% CI, 0.82–1.00; P = 0.04), and a model adjusted for enrollment lymphocyte percentage (adjusted OR, 0.89; 95% CI, 0.81–0.99; P = 0.04; Table 1). In addition, a lower effector γδ T-cell concentration (CD45RA⁻/CCR7⁻) at enrollment trended toward a significant association with 28-day mortality when adjusted for the presence of shock. These findings suggest involvement of circulating γδ T cells in the human host immune response to melioidosis.

Table 1.

Association of Peripheral Blood γδ T Cells with 28-Day Mortality

	Unadjusted			Adjusted for Shock*			Adjusted for Lymphocyte %†
	OR	95% CI	P Value	OR	95% CI	P Value	OR	95% CI	P Value
γδ T cells, cells/μl‡	0.91	0.82–1.00	0.04	0.90	0.82–1.00	0.04	0.89	0.81–0.99	0.04
Effector γδ T cells, cells/μl§	0.26	0.06–1.04	0.06	0.25	0.05–1.30	0.05	0.2	0.34–1.24	0.08

Definition of abbreviations: CI = confidence interval; OR = odds ratio.

^*For model adjusted for shock, P values were determined using logistic regression models adjusted for the requirement of vasoactive medications at the time of enrollment.

^† Model adjusted for lymphocyte percentage; P values were determined using logistic regression models adjusted for lymphocyte percentage from hospital-obtained white blood cell differential count of lymphocytes obtained at the time of enrollment.

^‡ Concentration of CD3⁺ cells, Vδ2⁺ γδ T cells.

^§ Concentration of CD45RA⁻, CCR7⁻, Vδ2⁺ γδ T cells.

Pulmonary γδ T Cells Are Enriched in the Lung in a Murine Model of Pulmonary Melioidosis

As melioidosis frequently presents as pneumonia, and γδ T cells are often enriched at mucosal surfaces, we next sought to determine whether γδ T cells are enriched in the lungs in melioidosis. To investigate this question, we leveraged an established mouse model of pulmonary melioidosis including exposure to a whole-body aerosol of B. thailandensis, a surrogate for B. pseudomallei that does not require biosafety level 3 conditions but produces similar disease in mice (22). After aerosolizing a lethal inoculum (2–5 × 10⁴ colony-forming units [cfu]/lung) of B. thailandensis, we performed flow cytometry on homogenized lung tissue to identify γδ TCR⁺ lymphocytes in the lungs at serial time points after infection. Starting at 24 hours after infection, γδ T cell percentage (P < 0.01) and concentration (P < 0.01) were significantly enriched in the lung compared with uninfected control mice (Figure 2A). In addition, we next quantified IL-17⁺ and IFN-γ⁺ γδ T cells in the lung after PMA and ionomycin stimulation. Complementing our peripheral blood findings in patients with melioidosis, the percentage of IL-17⁺ γδ T cells in the lung increased significantly (P < 0.01) after pulmonary infection, but the IFN-γ⁺ γδ T-cell population remained quite low, suggesting an IL-17–predominant γδ T-cell response in the lung (Figure 2B). To assess γδ T-cell populations later during infection, we next exposed mice to a lower inoculum (3–8 × 10³ cfu/lung) of B. thailandensis, which causes illness but is typically survivable. At 6 days after infection, at which point mice are typically recovering from infection, we found that once again both the percentage (P < 0.001) and concentration (P < 0.001) of γδ TCR⁺ lymphocytes was significantly enriched in the lung compared with uninfected control mice (Figure 2C). These data are concordant with our human melioidosis findings that γδ T cells remain enriched in the peripheral blood after recovery from acute infection.

Figure 2.

IL-17⁺ γδ T cells are enriched in the lung in a murine model of pulmonary melioidosis. C57B6/J mice aged 8–12 weeks were infected with a whole-body aerosol of Burkholderia thailandensis. At specific time points after infection or in uninfected mice (naive), right ventricles were perfused with PBS and whole lungs procured. After homogenization, single-cell suspensions were stained, and cell populations were identified by flow cytometry. (A) Percentage and concentration of lung γδ TCR⁺ cells and CD3⁺ cells after infection. (B) Percentage of γδ TCR⁺, IL-17⁺, or γδ TCR⁺, IFN-γ⁺ lung cells after 3 hour stimulation using PMA/ionomycin and brefeldin A. (A and B) Represent lethal B. thailandensis infections with inoculum of 2–5 × 10⁴ colony-forming units (cfu)/lung. (C) Percentage and concentration of lung γδ TCR⁺ cells and CD3⁺ cells 6 days after sublethal infection of 3–8 × 10³ cfu/lung B. thailandensis. All data are presented as median and IQR. Concentration of cells is presented as cells/ml, calculated as percentage of target cell of live single cells multiplied by manual cell concentration of live cells. Comparisons of groups by Mann-Whitney tests. **P < 0.01 and ***P < 0.001. Data represent a combination of at least two independent experiments.

Mice Lacking γδ T Cells Have Worse Survival Related to Impaired Bacterial Clearance in a Murine Model of Pulmonary Melioidosis

Given the pulmonary enrichment of γδ T cells within 24 hours after infection, we next sought to investigate whether γδ T cells are critical to the immune response in the lung in melioidosis. Therefore, we infected wild-type mice as well as mice lacking γδ T cells (Tcrd^−/−) on the same C57BL/6 background with a typically sublethal inoculum of B. thailandensis and followed mice over 10 days, evaluating survival as well as several health metrics. Tcrd^−/− mice had significantly worse 10-day survival (P = 0.01) than wild-type control mice (Figure 3A). Tcrd^−/− mice also had evidence of impaired health as early as 24 and 48 hours after infection, with significantly reduced body temperatures compared with baseline (P < 0.01) and lower health scores (P < 0.01, 24 h; P < 0.001, 48 h). Knockout mice surviving infection also demonstrated more profound weight loss and slower return to baseline weight, suggesting greater severity of illness than the wild-type control mice. To confirm these findings, we next infected both Tcrd^−/− and wild-type mice with an aerosolized inoculum of B. pseudomallei that is typically sublethal to wild-type mice. After infection of 200 cfu/lung of B. pseudomallei, Tcrd^−/− mice again had significantly worse survival over 7 days (P = 0.01), worse health scores, and impaired temperature regulation compared with wild-type control mice (Figure 3B). Similar results were observed after a lethal inoculum of 1,600 cfu/lung of B. pseudomallei (Figure E6).

Figure 3.

Tcrd^−/− mice have worse survival and health outcomes than control mice in attenuated murine models of pulmonary melioidosis. C57B6/J mice (wild type/WT: solid black lines or open squares) and Tcrd^−/− mice (KO, dashed red lines or red circles) on a C57B6/J background were infected by whole-body aerosolization with either (A) B. thailandensis (5–8 × 10³ cfu/lung) or (B) B. pseudomallei (200 cfu/lung). Mice were followed for either 10 (A) or 7 (B) days after infection for survival, temperature change as a percentage of preinfection temperature, weight change as a percentage of preinfection weight, and a 7-point health score. All assessments were performed by blinded investigators. Killing occurred at predetermined targets, including a health score <3, body temperature of <24°C, or weight loss of ⩾25%. Data are presented as survival curves or median and IQR. Comparisons by log-rank test (survival) or Mann-Whitney U test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Data in A are representative of two independent infections, with n = 10 mice/group. Data in B are representative of one infection, with n = 8–12 mice/group. KO = knock out; WT = wild type.

To determine whether the impact of γδ T cells on survival after infection is related to impaired bacterial clearance, we next assessed the bacterial burden both locally in the lungs and systemically in the spleen at different time points after infection with a sublethal inoculum of B. thailandensis. Although bacterial burdens in the lung of Tcrd^−/− and wild-type mice were the same 5 hours after infection, Tcrd^−/− mice had significantly more lung bacteria starting at 24 hours after infection (Figure 4A), suggesting impaired local clearance of bacteria. Tcrd^−/− mice also had evidence of impaired systemic bacterial clearance in their spleens at 24 hours after infection (P < 0.01; Figure 4B).

Figure 4.

Bacterial clearance is impaired in Tcrd^−/− mice compared with control mice in an attenuated murine model of pulmonary melioidosis. C57B6/J mice (wild type/WT: open squares) and Tcrd^−/− mice (KO: red circles) on a C57B6/J background were infected by whole-body aerosolization with B. thailandensis (5–8 × 10³ cfu/lung). At selected time points after infection, (A) left lungs or (B) spleen (24 h) were procured, homogenized, and cultured for 24 hours. n = 10–12 mice/group/time point. Median and IQR are presented and comparisons by Mann-Whitney U test; *P < 0.05 and **P < 0.01. Data at each time point represent the combination of two independent experiments. ns = not significant.

Neutrophil Migration–associated Genes Are Differentially Expressed in Tcrd^−/− Lungs Compared with Wild Type after Infection

Given the evidence of a critical role of γδ T cells starting as early as 24 hours after infection, we next used an unbiased transcriptomic approach to elucidate key mechanistic pathways in the lung. We performed bulk RNA-seq on whole lungs from Tcrd^−/− and wild-type mice at 24 hours after infection as well as uninfected control mice. We initially compared the whole lung transcriptome between infected knockout and wild-type mice using a principal components analysis and found separation between the two conditions (Figure 5A), indicating that the absence of Tcrd induces large-scale transcriptional signals during B. thailandensis infection. This was confirmed by identification of more than 400 differentially expressed genes between the two genotypes after infection (false discovery rate [FDR] < 0.05). We next applied gene set enrichment analysis to identify enriched canonical pathways (FDR < 0.05) (Figure 5B). The majority of upregulated gene sets in the lungs of infected Tcrd^−/− mice compared with wild-type mice were related to the host immune response, with one of the most differentially expressed pathways related to leukocyte endothelial migration (FDR < 0.001), as well as processes involved in focal adhesion, interleukin signaling, and neutrophil degranulation (Figure 5B and Tables E2 and E3). To capture the transcriptional consequences of interaction between genotype (Tcrd^−/−, wild type) and infection (B. thailandensis, uninfected), we performed a likelihood ratio test in DESeq2 using RNA-seq data from all four groups. Importantly, of the top 16 identified genes with significant genotype × infection interaction (FDR < 0.01), 8 were related to neutrophil migration or activity. For example, Olfm4, the gene coding for olfactomedin 4, an important neutrophil-associated granule protein, had a 56 ± 11.5-fold (mean ± SD) increase in expression after infection in wild-type mice but a 384 ± 92-fold increase in expression after infection in the knockout mice (P < 0.001; Figure 5C) (23, 24). Taken together, these finding imply that γδ T cells may have an inhibitory effect on neutrophil-related gene activation in the lungs.

Figure 5.

Lung neutrophil–associated genes are differentially regulated in Tcrd^−/− mice in an attenuated murine model of pulmonary melioidosis. C57B6/J mice (wild type) and Tcrd^−/− mice on a C57B6/J background were infected by whole-body aerosolization with B. thailandensis (9 × 10³ cfu/lung). At 24 hours after infection or in uninfected Tcrd^−/− mice and wild-type mice, whole-lung RNA was extracted and bulk RNA-seq performed. (A) Principal components analysis of entire lung transcriptome separates infected Tcrd^−/− mice (n = 4, blue) from infected wild types (n = 4, red), indicating that a global transcriptional signal differentiates the two groups. (B) Volcano plot summarizing Gene Set Enrichment Analysis (GSEA) of differentially regulated pathways between Tcrd^−/− mice and wild-type mice at 24 hours after infection. (C) Neutrophil- and chemotaxis-associated genes with significant differential regulation based on the interaction of genotype × infection are shown. Normalized gene counts at 24 hours after infection were compared with uninfected controls of the same genotype to obtain fold/relative differences. Comparisons made by unpaired t test; n = 4/group. ***P < 0.001 and ****P < 0.0001. Open squares/WT: wild type; red circles/KO: Tcrd^−/−. Data represent a single experiment. FDR = false discovery rate.

Increased Neutrophil Enrichment in the Lung Is Independent of Bacterial Load in Tcrd^−/− Mice

γδ T cells have previously been postulated to be critical for pulmonary neutrophil recruitment to the lung in some pneumonia models (18, 19). However, our unbiased transcriptomic analysis suggested leukocyte migration and potentially neutrophil-associated migration were upregulated in mice lacking γδ T cells after infection. Therefore, we next assessed the leukocyte populations in the lung at various time points after infection. Using flow cytometry, we identified pulmonary neutrophils as Ly6G⁺, CD11b⁺, CD45⁺ live cells in a whole-lung single-cell suspension. Starting as early as 5 hours after infection, the Tcrd^−/− mice had a higher percentage and concentration of pulmonary neutrophils than wild-type control mice (both, P < 0.05; Figures 6A–6D). Notably, at 5 hours after infection, the two groups had similar bacterial burden in the lungs, suggesting that increased neutrophil recruitment to the lung in Tcrd^−/− mice is occurring independent of pulmonary bacterial load (Figure 4A). In addition, in uninfected mice as well as at other time points after infection, no differences were noted between groups of other pulmonary leukocyte populations (Figures 6B and E7). Recently, a specific subgroup of neutrophils, expressing the surface receptor Siglec-F (sialic acid–binding, Ig-like lectin F), have been implicated in the pathogenesis of several noninfectious inflammatory states (25, 26). Ear6, a gene associated with Ear6 (eosinophil-associated RNase A protein), was uniquely expressed by being significantly upregulated in the lungs of wild-type mice compared with Tcrd^−/− mice at 24 hours after infection (Figure E8A). In evaluating the Immunological Genome Project consortium’s mouse single-cell gene expression repository, Siglecf and Ear6 had relatively higher expression in eosinophils, alveolar macrophages, and neutrophils (27). Eosinophils and alveolar macrophages were not differentially enriched at different time points after infection (Figures E8B and E8C). Therefore, we assessed whether Siglec-F⁺ neutrophils were differentially recruited to the lungs of mice without γδ T cells after infection. Intriguingly, by 24 hours after infection, the percentage of Siglec-F⁺ neutrophils in the lungs of Tcrd^−/− mice began to rapidly decline compared with wild-type mice (P < 0.01; Figures 6E and 6F). Furthermore, by 48 hours after infection, the concentration of Siglec-F⁺ neutrophils in the lung was significantly higher (P < 0.05) in the wild-type compared with Tcrd^−/− mice. As Siglec-F⁺ neutrophils may have unique characteristics to develop neutrophil extracellular traps (NETs), we next measured neutrophil activity and NETosis within the lungs after infection. Although the myeloperoxidase concentration of BALF was significantly higher at 5 hours after infection in the Tcrd^−/− mice (P < 0.001), by 24 hours wild-type mice expressed higher concentrations (P < 0.001; Figure 6G). BALF and lung homogenate NETs, measured by myeloperoxidase–DNA complexes, increased over time after infection, although no difference was noted in the BALF NETs between Tcrd^−/− and wild-type mice (Figures 6H and E9).

Figure 6.

Lung neutrophil activity is regulated by γδ T cells before impairment of bacterial clearance in an attenuated murine model of pulmonary melioidosis. C57B6/J mice (wild type/WT: open squares) and Tcrd^−/− mice (KO: red circles) on a C57B6/J background were infected by whole-body aerosolization with B. thailandensis (5–8 × 10³ cfu/lung). At specific time points after infection or in uninfected mice (naive), right ventricles were perfused with PBS and whole lungs procured. After homogenization, single-cell suspensions were stained, and cell populations were identified by flow cytometry. (A) Percentage of lung neutrophils (Ly6G⁺, CD11b⁺ cells) at separate time points after infection and (B) lung inflammatory cells at 5 hours after infection. AM = alveolar macrophage; B = B cells; DC = respiratory dendritic cells; EOS = eosinophils; LYMPH = TCR-β⁺ lymphocytes; PMN = neutrophil). See data supplement for gating strategy and assessment of cell populations. (C) Concentration of lung neutrophils at separate time points after infection, including (D) 5 and 24 hours after infection. Concentration of neutrophils is presented as cells/ml, calculated as the Ly6G⁺ CD11b⁺ percentage of live, CD45⁺ cells multiplied by the manual cell concentration of live leukocytes. (E) Percentage and (F) concentration of lung Siglec-F⁺ neutrophils (Siglec-F⁺ Ly6G⁺ CD11b⁺ cells). (G) At 5 and 24 hours after infection, right lung BALF was obtained and supernatant myeloperoxidase (MPO) concentration or (H) MPO-DNA absorbance fold change relative to an uninfected control was assessed by ELISA. Data in A–F are presented as mean ± SEM and comparisons made by unpaired t test; n = 4–5/group. Data in G and H are presented as mean ± SEM and comparisons by unpaired t test; n = 7–9/group. *P < 0.05 and ***P < 0.001. Data at each time point in A–F are single experiments, which are representative of two independent experiments. Data in G and H represent the combination of two independent experiments.

Enhanced Lung Inflammation in Tcrd^−/− Mice Is Independent of Bacterial Burden after Infection

As we had identified a potentially pathologic enrichment of neutrophils in the lung in mice lacking γδ T cells, we next determined whether this increase in neutrophil activity was also associated with increased inflammation. A blinded pathologist scored histologic lung sections from Tcrd^−/− and wild-type mice at different time points after infection. At 5 hours after infection, before a difference in bacterial load, the lungs of Tcrd^−/− mice had significantly more histologic inflammation than wild-type control mice (Figure 7A). In addition, lung homogenate supernatants at 5 hours after infection had significantly higher concentrations of IL-6, IFN-γ, and IL-1β, as well as multiple chemokines, including IP-10, MCP-1, and MIP-1α (Figure 7B). In addition, by 48 hours after infection, lung homogenate concentrations of multiple inflammatory cytokines, including IL-17 and neutrophil-specific chemokines CXCL-1 and CXCL-2, were significantly higher in Tcrd^−/− mice than in wild-type mice (Table 2).

Figure 7.

Tcrd^−/− mice have evidence of increased inflammation before impairment of bacterial clearance in an attenuated murine model of pulmonary melioidosis. C57B6/J mice (wild type/WT: open squares) and Tcrd^−/− mice (KO: red circles) on a C57B6/J background were infected by whole-body aerosolization with B. thailandensis (5–8 × 10³ cfu/lung). (A) At selected time points after infection, right lungs were inflated with formalin, paraffin-embedded, stained with hematoxylin and eosin, and assessed by a blinded pathologist for inflammation. (B) At 5 hours after infection, left lungs were procured, homogenized, and supernatant analyte concentrations assessed by electrochemiluminescence assay. n = 10–12 mice/group/time point. (A) Data are presented as mean ± SEM and comparisons made by unpaired t test; (B) median and IQR are presented and comparisons by Mann-Whitney U test; *P < 0.05, **P < 0.01, and ***P < 0.001. Data at each time point represent the combination of two independent experiments.

Table 2.

Lung Homogenate Cytokine and Chemokine Concentrations Are Increased in Tcrd^−/− Mice after 48 Hours in an Attenuated Murine Model of Pulmonary Melioidosis

Analyte (pg/ml)	Tcrd−/−	Wild Type	P Value*
IFN-γ	26.7 (22.0–33.3)	26.2 (17.5–32.9)	ns
TNF	1,073.9 (867.0–1,232.7)	517.5 (314.6–812.7)	0.002
IL-1β	3,197.9 (2,432.3–5,889.5)	1,468.3 (687.5–2,673.3)	0.01
IL-6	18,035.7 (12,654.4–30,389.3)	3,078.5 (647.3–6,748.4)	<0.001
IL-10	49.2 (43.5–82.3)	20.7 (12.6–47.4)	0.02
IL-12p70	86.5 (71.7–132.8)	36.7 (25.1–49.7)	0.001
IL-17	53.4 (47.5–64.6)	28.1 (13.1–43.7)	0.01
IL-27p28	28.3 (26.7–41.5)	15.4 (2.9–33.2)	0.02
IL-33	2,245.8 (1,933.9–2,814.9)	2,778.6 (2,456.5–3,191.7)	ns
IP-10	9,901.9 (8,564.2–12,376.1)	5,630 (4,069.7–8,522.4)	<0.001
MCP-1	1,733.9 (1,471.6–2,858.6)	606.1 (405.4–1,091.8)	0.002
MIP-1α	4,593.7 (3,480.4–5,168.1)	1,810.6 (877.8–3,284.2)	0.02
CXCL-1	5,811.6 (3,385.6–9,603.0)	1,119.6 (242.0–2,499.4)	<0.001
CXCL-2	24,482.4 (18,842.4–27,725.3)	8,100 (2,649.5–16,593.1)	0.006

Definition of abbreviation: ns = not significant.

Data are presented as median (interquartile range).

^*P values <0.05 are listed in bold.

Discussion

Severe pneumonia is associated with significant global morbidity and mortality, although the exact contributions of the host immune response to these outcomes remains unknown (28). Pulmonary melioidosis, one example of severe pneumonia, is frequently fatal and associated with impaired health outcomes in survivors 1 year after discharge (1, 29). Pneumonic melioidosis models typically demonstrate a robust neutrophil response, which appears necessary for clearance of B. pseudomallei (20, 22, 30). However, the mechanism by which this neutrophil response is regulated remains poorly characterized. Although lymphocytes have been implicated in the host immune response to melioidosis, little is known about the role of innate-like lymphocytes, including γδ T cells (31, 32). In this study, we provide evidence that IL-17⁺ γδ T cells, a small population of lymphocytes, play a critical role in modulating the early, pathogen-driven inflammatory response in the lung. In addition, although γδ T cells have been reported to enhance neutrophil recruitment in other respiratory infections, we found that mice lacking γδ T cells experienced enhanced rather than reduced pulmonary neutrophil recruitment with subsequent impaired bacterial clearance (13, 14, 19).

Both pulmonary and systemic bacterial infections can activate and expand γδ T cells. Initially, γδ T cells were believed to be primarily activated through their TCR by microbial peptides, such as phosphoantigens. However, they also appear to have a wide array of cell surface receptors for pattern recognition receptors, such as TLRs, as well as natural killer cell receptors (33). Activation of γδ T cells can lead to the release of cytokines and recruitment of phagocytic cells, including neutrophils and monocytes. In addition, γδ T cells can directly eliminate infected cells through the production of granulysin, perforin, and other defensins (34). Both direct and indirect pathogen clearance could be critical during infection with a severe, intracellular organism such as B. pseudomallei.

In our study, we found evidence that peripheral blood IL-17⁺ γδ T-cell populations are activated and expand in melioidosis, and specific populations are associated with survival, consistent with other pneumonias. Complementing our human data, we report that in an experimental animal model of melioidosis, IL-17–producing γδ T cells in the lung expand as early as 24 hours after infection. Intriguingly, however, we found that mice lacking γδ T cells did not have impaired pulmonary neutrophil recruitment but rather enhanced neutrophil recruitment. In addition, this enhanced neutrophil recruitment to the lung was associated with impaired bacterial clearance, despite increased markers of neutrophil activity. B. pseudomallei uptake by mucosal monocytes, including alveolar macrophages, can directly elicit a profound neutrophil response (20). However, even as early as 5 hours after infection, when pulmonary bacterial loads were similar between groups, mice lacking γδ T cells not only had higher pulmonary neutrophils but also had corresponding evidence of increased lung inflammation. Neutrophils have also been implicated as drivers of deleterious inflammation during B. pseudomallei lung infection (35). These data therefore suggest that, during the early hours after a severe pulmonary infection such as B. pseudomallei, the most critical role of γδ T cells may be in modulating an excessive pathogen-driven inflammatory response.

Although driving a neutrophil-specific mucosal host response has been proposed as the primary mechanism for lung bacterial clearance of IL-17–producing γδ T cells, they may also have a role in tissue healing and homeostasis restoration. For example, after resolution of pulmonary Streptococcus pneumoniae infection, a significant increase in pulmonary γδ T cells was correlated with a corresponding decrease in alveolar macrophages (36). This finding mirrors both our human data indicating that an effector γδ T-cell response persists, as well as our animal model data demonstrating persistent enrichment of lung γδ T cells after acute disease resolution. In addition, γδ T cells may be particularly critical to reducing local inflammation in the setting of intracellular infection. For example, γδ T cells can inhibit intracellular replication through cell lysis during Listeria monocytogenes; however, the elimination of activated macrophages by γδ T-cell infection can also help prevent tissue damage related to excessive inflammation (37, 38).

Alternatively, the direct cytotoxic effects of γδ T cells have been explored as a potential mechanism for vaccine development, particularly for infections caused by evasive, intracellular pathogens. For example, γδ T cells appear to have a critical role in responding to Mycobacterium tuberculosis in humans and have been extensively evaluated as a vaccination target (39–41). During M. tuberculosis infection, γδ T cells are differentially enriched in the lung and are able to directly eliminate intracellular pathogens through granulysin and perforin release (42–45). In addition, γδ T cells are rapidly expanded in the peripheral blood of patients infected with Francisella tularensis, although not after vaccination (46, 47). Finally, stimulated nonhuman primate pulmonary γδ T-cell subsets were found to be able to inhibit growth of human monocytes infected with B. pseudomallei (48). Given the elevated, but ineffective, pulmonary neutrophil burden in our study in mice lacking γδ T cells, it is possible that the direct cytotoxic effects of γδ T cells are critical for bacterial control during the first 24 hours after infection, where much of the bacterial replication is occurring intracellularly.

Our findings also suggest that γδ T cells may uniquely influence neutrophil subset recruitment to the lungs, potentially offering an explanation for the impaired neutrophil response noted after infection. Specifically, we found that Siglec-F⁺ neutrophils were relatively downregulated in mice without γδ T cells as early as 24 hours after infection. Siglec-F⁺ neutrophils are a recently described subset of neutrophils, which have been implicated in driving deleterious inflammatory processes, including in some cancers and after myocardial infarction (25, 49). Siglec-F⁺ neutrophils have also recently been described as contributing to excessive inflammation in the lung after air pollutant–induced damage, a process believed to be related to increased NETosis activity by this neutrophil subset (26). In our study, we did note decreased airway neutrophil myeloperoxidase production after infection in mice without γδ T cells, despite higher lung neutrophil concentration. However, we did not observe a difference in myeloperoxidase-associated NETs in the airway after infection. NET production did increase in the lungs and airway after Burkholderia infection in our study, supporting prior reports suggesting NETs may induce host bacterial clearance in melioidosis (50). Although our findings do not directly implicate Siglec-F⁺ neutrophils as critical mediators of pulmonary bacterial clearance, further study is warranted into the role of this neutrophil subset in pneumonia.

Our study has multiple strengths. We report that γδ T cells may play a role in the host response to melioidosis across both human and animal studies. In addition, within our animal studies, we have confirmed a survival phenotype using both B. thailandensis and B. pseudomallei. Our animal models also leverage a whole-body aerosolization technique, which eliminates the need for anesthesia and mimics the inhalation route of infection of B. pseudomallei in endemic areas. Finally, our data included an early, unbiased transcriptomic pathway analysis with resulting verification using alternative methods.

Our study also has several limitations. Although B. thailandensis has a similar course to B. pseudomallei infection in mice, differences may exist, although a similar analysis was limited by CDC-related select agent restrictions of B. pseudomallei in our study. In addition, γδ T cells likely have different functions in the lung in humans and mice, limiting extrapolation of these findings (11). In addition, further study will be necessary to fully understand specific mechanisms influencing the relationship between γδ T cells and neutrophil function and recruitment in melioidosis.

In conclusion, our results implicate γδ T cells as a dynamic but important mediator of the pulmonary immune response during melioidosis. We show that in the few hours after pulmonary melioidosis infection, independent of bacterial burden, these cells play a crucial role in modulating the neutrophil response in the lung, while also regulating the broader inflammatory cascade. Further study is required to determine the specific pathways by which γδ T cells modulate neutrophil activity and intracellular bacterial replication in melioidosis, as well as how these processes translate to the human lung.