Type I IFN exacerbates disease in tuberculosis-susceptible mice by inducing neutrophil-mediated lung inflammation and NETosis

Nov 5, 2020Nature communications

Type I Interferon may worsen tuberculosis in mice by triggering neutrophil-driven lung inflammation and cell trap formation

AI simplified

PubMed ↗DOI ↗OA ↗

Abstract

Transcriptional signatures associated with type I IFN signalling and neutrophilic inflammation correlate with increased bacterial loads in tuberculosis.

Blockade of GM-CSF leads to exacerbated pathology during Mycobacterium tuberculosis infection.
Loss of GM-CSF signalling or genetic susceptibility results in formation of neutrophil extracellular traps (NETs) that may enhance bacterial growth.
NETs are found in lung lesions of tuberculosis patients who respond poorly to antibiotic treatment.
These findings suggest a cytokine-based immune network plays a critical role in the outcome of tuberculosis infection.

AI simplified

Tuberculosis (TB) is a leading cause of mortality due to infectious disease, but the factors determining disease progression are unclear. Transcriptional signatures associated with type I IFN signalling and neutrophilic inflammation were shown to correlate with disease severity in mouse models of TB. Here we show that similar transcriptional signatures correlate with increased bacterial loads and exacerbate pathology during Mycobacterium tuberculosis infection upon GM-CSF blockade. Loss of GM-CSF signalling or genetic susceptibility to TB (C3HeB/FeJ mice) result in type I IFN-induced neutrophil extracellular trap (NET) formation that promotes bacterial growth and promotes disease severity. Consistently, NETs are present in necrotic lung lesions of TB patients responding poorly to antibiotic therapy, supporting the role of NETs in a late stage of TB pathogenesis. Our findings reveal an important cytokine-based innate immune effector network with a central role in determining the outcome of M. tuberculosis infection.

Full Text

What this is

This research investigates the role of type I interferon (IFN) in tuberculosis (TB) pathogenesis, particularly in mice susceptible to TB.
The study focuses on how type I IFN signaling exacerbates disease through neutrophil-mediated lung inflammation and the formation of neutrophil extracellular traps (NETs).
Findings indicate that blocking GM-CSF, a cytokine important for TB resistance, leads to increased NET formation and worsened disease outcomes.

Essence

Type I IFN signaling exacerbates tuberculosis in susceptible mice by promoting neutrophil activation and NET formation, leading to increased bacterial loads and lung inflammation.

Key takeaways

Blocking GM-CSF during Mycobacterium tuberculosis infection enhances neutrophil recruitment and activation, resulting in increased bacterial growth and lung pathology.
NETs, which are formed in response to type I IFN, are associated with severe lung lesions in both TB-susceptible mice and patients with pulmonary TB.
The study reveals that type I IFN-induced contributes to disease severity, suggesting a potential therapeutic target for improving TB treatment outcomes.

Caveats

The study primarily uses mouse models, which may not fully replicate human TB pathogenesis and responses to treatment.
While the findings are significant, the exact mechanisms linking NET formation to disease exacerbation require further investigation.

Definitions

NETosis: A form of cell death in neutrophils that leads to the release of extracellular traps composed of DNA and antimicrobial proteins.

AI simplified

Introduction

Tuberculosis (TB) remains a major health problem worldwide with 1.2 million reported deaths in 2018¹. Despite its clinical significance, there are still significant gaps in our understanding of the mechanisms underlying protection or pathogenesis following Mycobacterium tuberculosis infection that could be exploited to improve TB treatment and thus reduce disease severity and mortality^2–4.

Transcriptional studies of human TB have been instrumental in unveiling new mediators of the immune response to M. tuberculosis infection⁵. Blood transcriptional signature of active TB revealed a dominance of type I IFN-inducible genes⁶ (and reviewed in⁷), which correlated with the extent of radiographic lung disease and was diminished upon successful treatment^6,8,9. These studies highlighted an important role for type I IFN in driving TB pathogenesis. Consistently, studies in mouse models of TB demonstrated that induction of high and sustained levels of type I IFN by different mechanisms result in increased host susceptibility to M. tuberculosis infection^7,10–15. Transcriptomic analyses in human and mouse studies have also revealed an over-abundance of type I IFN and granulocyte-associated genes correlating with lung disease severity in both active TB patients and TB-susceptible mice¹⁶. In humans, neutrophils are the predominant cell type infected with M. tuberculosis in the airways and are detected within inflammatory lung lesions of patients with active disease^17,18. In mice, neutrophils are abundant in the lung lesions of M.tuberculosis-infected susceptible mice and have also been reported to contribute to TB progression promoting both tissue damage and bacterial replication^18–20. However, the mechanisms of neutrophil-mediated pathology in TB-susceptible mice and patients with active TB remain unclear.

Macrophages and CD4⁺ T cells, and several cytokines such as IL-12, IFN-γ, TNF and IL-1 have been shown to play a major role in protection against M. tuberculosis infection^{2–4,13,21–24}. More recently, GM-CSF has also been increasingly recognised as a key mediator for TB resistance^25–31, and is in the pipeline for host-directed therapy as an adjuvant treatment of TB³². The presence of GM-CSF specific autoantibodies that block GM-CSF function have been linked to cryptococcal disease and some cases of pulmonary TB²⁸. In addition, GM-CSF treatment of mouse and human macrophages restricts intracellular mycobacterial growth in vitro^25,26,29,30 and mice lacking GM-CSF are highly susceptible to M. tuberculosis infection^27,31. Impaired innate immune responses and decreased recruitment of protective IFN-γ producing CD4⁺ T cells into the site of infection have been proposed as mechanisms for the increased susceptibility following M. tuberculosis infection in mice deficient in GM-CSF^27,31. However, these studies are difficult to interpret owing to the developmental defects in alveolar macrophages and abnormalities in surfactant recycling in mice lacking GM-CSF from birth, leading to the development of a lung disease that resembles human pulmonary alveolar proteinosis (PAP)^33,34. Therefore, the mechanisms underlying the pathology resulting from an absence of GM-CSF signalling during M. tuberculosis infection remain poorly understood.

To further investigate the mechanisms underlying the pathology and disease exacerbation resulting from an absence of GM-CSF signalling, without the confounding developmental lung pathologies resulting from gene deletion, here we instead administer blocking antibodies against GM-CSF during M. tuberculosis infection. We show that increased expression of genes associated with neutrophil recruitment and activation, as well as type I IFN-inducible genes, in blood and lungs of M.tuberculosis-infected mice in the absence of GM-CSF signalling, accompanies enhanced mycobacterial growth and lung pathology. Disease exacerbation driven by GM-CSF blockade is neutrophil-dependent and associates with exacerbated neutrophil extracellular trap (NET) formation at the site of infection. Mechanistically, our findings reveal that increased type I IFN signalling induces pulmonary NETosis and promotes mycobacterial growth in the absence of GM-CSF. We show that type I IFN-induced NETosis is also associated with disease severity in TB-susceptible C3HeB/FeJ mice, supporting a more generalised role of type I IFN-induced NETosis in TB pathogenesis over and above GM-CSF blockade. Importantly, we demonstrate for the first time that NETs are present in necrotic lung lesions from patients with nonresolving pulmonary TB, supporting the clinical revelance of NETs in TB pathogenesis.

Results

GM-CSF blockade increases host susceptibility to TB

Since increased susceptibility to M. tuberculosis infection in GM-CSF-deficient mice has been associated with an impaired T cell-mediated IFN-γ-response at the site of infection²⁷, we evaluated the effects of αGM-CSF treatment on T cells in the lungs of infected mice. We observed a slight but not significant decrease in the percentage of CD4⁺ and CD8⁺ T cells in the lungs of infected mice treated with αGM-CSF compared to Ctrl mAbs, with no effect on T cell numbers (Supplementary Fig. 1a, b). Moreover, similar percentages and numbers of IFN-γ-producing CD4⁺ and CD8⁺ T cells were detected ex vivo from both αGM-CSF and Ctrl Ab treated mice (Supplementary Fig. 1a, b), demonstrating that blockade of GM-CSF did not impact IFN-γ responses in infected mice. Blockade of GM-CSF during M. tuberculosis infection also resulted in increased lung mycobacterial loads in IFN-γ receptor-deficient (IFNγR^−/−) mice (Supplementary Fig. 1c) and in WT mice co-treated with anti-IFN-γ mAbs (Supplementary Fig. 1d), compared to their respective infected Ctrl Ab treated mice. Our findings thus reveal an additional in vivo role for GM-CSF in host resistance to M. tuberculosis infection that is independent of IFN-γ.

Fig. 1

GM-CSF controls lung bacterial growth and necrotic inflammation. Schematic of experimental design for GM-CSF blockade duringinfection: intraperitoneal (i.p.) injection of 1 mg anti-GM-CSF (αGM-CSF; open circles) or isotype control (labelled as Ctrl Ab; closed circles) monoclonal antibodies (mAbs) the day prior to aerosol infection with HN878 strain, followed by i.p. injections of 0.3 mg of αGM-CSF or Ctrl mAbs, twice weekly for 3 weeks.Change in body weight at various days post-infection (= 8 mice/group; mean ± SD). ***< 0.0001.Viable bacterial loads determined in the lungs at day 21 post-infection. **= 0.0079.Representative hematoxylin and eosin (H&E) staining of infected lung sections (= 3 mice/group). Scale bars from top to bottom: 1 mm; 200 µm; 50 µm, 10 µm.Relative lung lesion burden from H&E stained sections (Supplementary Data). 0 = no lesions, 1 = focal lesion, 2 = multiple focal lesions, 3 = one or more focal severe lesions, 4 = multiple focal lesions that are extensive and coalesce, 5 = extensive lesions that occupy the majority of the lung lobe. ***< 0.0001.Relative necrosis (left) and intra-alveolar necrotic debris (right) scores from H&E stained lung sections (Supplementary Data). 0 = not present, 1 = minimal, 2 = mild, 3 = moderate, 4 = marked. ***< 0.0001.Representative Ziehl–Neelsen staining of infected lung sections (= 3 mice/group). Scale bars, 20 µm. Data representative (–,) or pooled (,) of five biological experiments. Represented is the mean ± SD; each dot represents an individual mouse:= 5 mice/group () or= 15 mice/group (,). Source data are provided as a Source Data file. Statistical analysis was performed using two-way ANOVA () or two-tailed Mann–Whitney test (,,). a b c d e f g b d g e f c e f b c e f M. tuberculosis n P P n P P n n n 1 1

GM-CSF controls blood transcriptional TB signatures

To investigate pathways underlying the pathogenesis resulting upon M. tuberculosis infection in the absence of GM-CSF signalling, we performed RNA-Seq of whole blood from uninfected and HN878-infected mice treated with αGM-CSF or Ctrl mAbs. Principal component (PC) analysis depicted distinct global transcriptional signatures in blood samples from the different groups (Supplementary Fig. 2a), with PC1 explaining ~73% of the variance and segregating the uninfected from HN878-infected samples. On the other hand, PC2 segregates the infected samples according to treatment (αGM-CSF versus Ctrl Ab, ~13% of the variance explained) (Supplementary Fig. 2a, b). We identified 7010 genes differently expressed in the blood of HN878-infected Ctrl Ab mice compared to uninfected Ctrl Ab treated mice (Supplementary Data 2 and Supplementary Fig. 2c, left panel). While we observed negligible transcriptional changes due to αGM-CSF treatment in the blood from uninfected mice (Supplementary Fig. 2c, middle panel), GM-CSF blockade during infection triggered major transcriptional changes in the blood, with 4479 differently expressed genes detected between αGM-CSF versus Ctrl Ab treated infected mice (Supplementary Data 2 and Supplementary Fig. 2c, right panel).

Fig. 2

GM-CSF controls transcriptional signatures associated with TB pathogenesis. WT mice were infected and/or treated as in Fig.. Blood and lungs were collected fromHN878-infected and uninfected mice for RNA-Seq analysis (= 4–5 mice/group).Blood modules of co-expressed genes derived using WGCNA from human TB datasetsare shown for blood RNA-Seq datasets obtained from infected mice treated with αGM-CSF or Ctrl mAbs, compared to uninfected Ctrl Ab treated mice (Supplementary Data).Lung modules of co-expressed genes derived using WGCNA from mouse lung disease modulesare shown for lung RNA-Seq datasets obtained from infected mice treated with αGM-CSF or Ctrl mAbs, compared to uninfected Ctrl Ab treated mice (Supplementary Data).,Fold enrichment scores derived using QuSAGE are depicted, with red and blue indicating modules over- or under-abundant, compared to the controls. Colour intensity and size of the dots represents the degree of perturbation, indicated by the colour scale, with the largest dot representing the highest degree of perturbation within the plot. Only modules with fold enrichment scores with FDR-value < 0.05 were considered significant and depicted here. Module name indicates biological processes associated with the genes within the module.,Boxplots depicting the module eigengene expression, i.e., the first principal component for all genes within the module, are shown for uninfected Ctrl Ab (light green) or αGM-CSF (light purple) and infected Ctrl Ab (dark green) or αGM-CSF (dark purple) treated groups (= 4–5 mice/group), for () blood modules: Inflammasome/Granulocyte (HB3), Innate/hemopoietic mediators (HB5), Innate Immunity/PRR/C’/Granulocyte (HB8), Interferon/PRR (HB12) and Interferon/C’/Myeloid (HB23); and () lung modules: Type I IFN//(L5), Myeloid/granulocyte function (L10), IL-17 pathway/granulocytes (L11), Inflammation/IL-1 signalling/myeloid cells (L12), and Myeloid cells//(L13). Gene expression profiles are shown using boxplots, where the upper and lower box limits show the interquartile range (limits of second and third quartiles), a thick horizontal bar within the box shows the median, whiskers show the minimum and maximum values, and each dot represents an individual mouse (= 4–5 mice/group). C’, complement.; PRR, pathogen recognition receptor. GCC, glucocorticoid; K-channel, potassium channel; TM, transmembrane; Ubiq, ubiquitination. 1a 3 5 M. tuberculosis n P n Ifit Oas Il1b Tnf n a c a c b d b d [35] [36]

GM-CSF controls lung transcriptional TB signatures

To determine the transcriptional response at the site of infection, RNA-seq data were obtained from whole lungs of the same mice used for the blood data from Fig. 2a. Sample distribution of lung RNA-Seq data PC analysis highly resembled blood RNA-Seq data, with the biggest transcriptional variance explained by the infection (PC1; Supplementary Fig. 4a, b) and with 11,532 genes differentially expressed between the lung samples from HN878-infected Ctrl Ab treated mice versus the lung samples from uninfected Ctrl Ab treated mice (Supplementary Data 4 and Supplementary Fig. 4c, left panel). Although with a smaller explained variance (~3%), PC2 clearly segregated the infected samples according to αGM-CSF or Ctrl Ab treatment (Supplementary Fig. 4a, b). GM-CSF blockade triggered transcriptional changes in the lungs from both uninfected and infected mice (Supplementary Data 4 and Supplementary Fig. 4c, middle and right panel). In uninfected mice, we detected 3 significantly upregulated and 327 downregulated genes between αGM-CSF versus Ctrl Ab groups (Supplementary Data 4 and Supplementary Fig. 4c, middle panel). Among the genes downregulated by GM-CSF blockade at steady-state, we identified genes associated with alveolar macrophages³⁷ (Supplementary Data 4 and Supplementary Fig. 4c, middle panel), in keeping with a key role for GM-CSF in the maintenance of these cells^38,39. During infection, GM-CSF blockade affected the lung expression of 2119 genes, with alveolar macrophage-associated genes among the top downregulated and neutrophil-related genes among the top upregulated genes (Supplementary Data 4 and Supplementary Fig. 4c, right panel).

To identify GM-CSF driven changes in co-expressed sets of genes across the infected lungs, we next tested the previously published mouse lung disease modular signature³⁶ on our lung TB RNA-Seq data (Supplementary Data 5 and Fig. 2c). We observed a similar modular signature in the lungs of HN878-infected mice treated with Ctrl Ab (left) or αGM-CSF (right) mAbs, but with distinct pattern in the abundance of certain modules (Fig. 2c). The Type I IFN/Ifit/Oas module (L5) was over-abundant in the infected lungs but slightly more enriched in infected mice treated with αGM-CSF compared to Ctrl Ab (Fig. 2c, d). This is consistent with an increased expression of IFN-inducible genes resulting from GM-CSF blockade during infection (Supplementary Data 4 and Supplementary Fig. 4c, cluster III). Modules associated with innate inflammation (L10–L13) were also over-abundant in infected lungs but further over-abundance was observed in lungs from infected mice treated with αGM-CSF mAbs compared to infected Ctrl Ab treated mice, also shown quantitatively by Eigengene expression (Fig. 2c, d). The elevated expression of inflammatory genes is consistent with the exacerbated lung pathology observed in HN878-infected αGM-CSF treated mice compared to infected Ctrl Ab treated mice (Fig. 1d–f and Supplementary Data 1). In contrast, the modules Ifng/Gbp/antigen presentation (L7) and Cytotoxic/T cells/ILC/Tbx21/Eomes/B cells (L35), which have been associated with TB resistance¹⁶, were slightly less over-abundant in infected lungs from infected mice treated with αGM-CSF versus Ctrl Ab (Supplementary Data 5 and Fig. 2c). In all, the transcriptional profile of lungs from HN878-infected αGM-CSF treated mice resembled that of infected TB-susceptible mice¹⁶_.

GM-CSF prevents NETosis duringinfection M. tuberculosis

As neutrophils were required for disease exacerbation (Fig. 3d) and their activation profile by gene expression appeared altered following GM-CSF blockade during infection (Fig. 3b), we examined neutrophil responses in vivo by immunofluorescence microscopy. We found an increased accumulation of neutrophils, detected by myeloperoxidase staining (MPO⁺), in the lesions of infected αGM-CSF treated mice compared to the lung lesions from infected Ctrl Ab treated mice (Fig. 3e). These findings prompted us to determine whether NETs were also present by examining the colocalization of DAPI (DNA staining) and citrullinated H3 (CitH3), a specific marker for NET formation^41,42. NETs were detected in surprisingly high amounts in the lung lesions from αGM-CSF treated mice but were scarcely found in infected Ctrl Ab treated mice, as shown quantitatively by measuring the percentage of CitH3 staining areas within the MPO positive areas in the lung lesions (Fig. 3f). These data indicate that neutrophil-driven disease exacerbation induced by GM-CSF blockade during M. tuberculosis infection correlates with excessive NETosis in the infected lungs.

Fig. 3

Disease exacerbation upon GM-CSF blockade is neutrophil-dependent and correlates with NETosis. GSEA of immune cell-type associated gene setsshowing normalized enrichment scores (NES) for lung RNA-Seq data from infected αGM-CSF versus infected Ctrl Ab treated mice.Heatmap showing relative expression of genes in the neutrophil-associated gene set for individual lung samples from uninfected Ctrl Ab (light green) or αGM-CSF (light purple) and infected Ctrl Ab (dark green) or αGM-CSF (dark purple) treated mice (= 4–5 mice/group). Gene expression values were averaged and scaled across the row to indicate the number of standard deviations above (red) or below (blue) the mean, denoted as row Z-score. The dendrogram shows unsupervised hierarchical clustering of genes.WT mice were infected and treated with Ctrl Ab (closed circles) or αGM-CSF (open circles) as in Fig.. Lung cell suspensions were prepared and stained for the detection of myeloid cells (Supplementary Fig.). Cell numbers for neutrophils (CD11bLy6G), Ly6Cmonocytes (CD11bLy6GLy6C) and alveolar macrophages (CD11bCD11c) are shown. *= 0.0317.WT mice were infected and treated with Ctrl Ab (closed symbols) or αGM-CSF (open symbols) as in Fig.with the exception that from day 7 post-infection, mice also received 0.2 mg of anti-Ly6G (αLy6G; squares) or isotype control (labelled as Isotype; circles) mAbs every other day by i.p. injection. Lung viable bacterial loads were determined. **= 0.0079.Representative images of lung sections stained with citrullinated histone H3 (citH3; red), MPO (green), and DAPI (blue). NETs are visualized by colocalization of citH3 and DAPI staining (merged images;= 3 mice/group). Bottom rows show detail from the top rows for each group. Scale bars, 50 µm.Percentage of NET area in the lung normalized to MPO positive signal. **= 0.0019. Data representative of one (,), two () or five (,) biological experiments; or data pooled of three independent experiments (). Represented is the mean ± SD (,) or mean ± SEM (); each dot represents an individual mouse:= 5 mice/group (,) or= 9 mice/group (). Source data are provided as a Source Data file. Statistical analysis was performed using two-tailed Mann–Whitney test. a b c d e f a b d c e f c f d c d f [40] + + + + − + low + n P P n P n n 1a 5 1a

NETs are detected in TB-susceptible mice and human TB

Fig. 4

NETs are detected in lung lesions of TB-susceptible C3HeB/FeJ mice. C3HeB/FeJ mice were aerosol infected withHN878 treated by i.p. injection with 0.2 mg of anti-Ly6G (αLy6G; open circles) or isotype control (labelled as Isotype; closed circles) mAbs every other day, from day 7 post-infection until harvest on day 26 post-infection. Lung viable bacterial loads were determined. **= 0.0079. Represented is the mean ± SD; each dot represents an individual mouse:= 5 mice/group. Source data are provided as a Source Data file. Statistical analysis was performed using two-tailed Mann–Whitney test.C3HeB/FeJ mice were aerosol infected withHN878 and their lungs harvested at day 35 post-infection. Formalin-fixed paraffin-embedded lung sections were stained with citrullinated histone H3 (citH3; red), MPO (green), and DAPI (blue). NETs are visualized by colocalization of citH3 and DAPI staining (merged image;= 3 mice/group). Scale bars, 50 μm. Data from one experiment. a b M. tuberculosis P n M. tuberculosis n

Fig. 5

NETs are detected in necrotic lung lesions of human pulmonary TB. Formalin-fixed paraffin-embedded lung sections from a total of thirteen patients with pulmonary TB (1 section per TB patient; Supplementary Table) were labelled with antibodies specific for citrullinated histone H3 (citH3; red) MPO (green) and DAPI (blue). NETs, visualized by colocalization of citH3 and DAPI staining (merged images, left), are shown for 6 TB patients. Scale bars, 50 µm. (All thirteen slides were stained and scanned together. See also Supplementary Fig.). 1 6a

Type I IFN signalling induces NETosis upon GM-CSF blockade

Fig. 6

IFNAR deletion abrogates disease exacerbation prompted by GM-CSF blockade during infection. WT (circles) and(triangles) mice were infected and treated with Ctrl Ab (closed symbols) or αGM-CSF (open symbols) as in Fig..Viable bacterial loads determined in the lungs at day 21 post-infection. **< 0.0079. Represented is the mean ± SEM; each dot represents an individual mouse:= 5 mice/group. Source data are provided as a Source Data file. Data representative of two biological experiments. Statistical analysis was performed using the two-tailed Mann–Whitney test.Representative H&E staining of infected lung sections (= 2–4 mice/group). Scale bars from top to bottom: 1 mm; 200 µm; 50 µm; 10 µm.Relative lung lesion burden from H&E stained sections (Supplementary Data). 0 = no lesions, 1 = focal lesion, 2 = multiple focal lesions, 3 = one or more focal severe lesions, 4 = multiple focal lesions that are extensive and coalesce, 5 = extensive lesions that occupy the majority of the lung lobe. ***< 0.0001.Relative necrosis and intra-alveolar necrotic debris scores from H&E stained lung sections (Supplementary Data). 0 = not present, 1 = minimal, 2 = mild, 3 = moderate, 4 = marked. ***< 0.0001; *= 0.0294.,Data pooled from four biological experiments. Represented is the mean ± SD; each dot represents an individual mouse:= 11–13 mice/group. Source data are provided as a Source Data file. Statistical analysis was performed using one-way ANOVA Tukey’s multiple comparisons test.Representative Ziehl–Neelsen staining of infected lung sections (= 2–4 mice/group). Scale bars, 20 µm. Data representative of four biological experiments (,). Ifnar P n n P P P n n -/- 1a 1 1 a b c d c d e b e

Fig. 7

IFNAR deletion abrogates NET formation prompted by GM-CSF blockade during infection. WT (circles) and(triangles) mice were infected and treated with Ctrl Ab (closed symbols) or αGM-CSF (open symbols) as in Fig..Representative images of lung sections stained with citrullinated histone H3 (citH3; red), MPO (green), and DAPI (blue). NETs are visualized by colocalization of citH3 and DAPI staining (merged images;= 3 mice/group). Scale bars, 50 µm. Data representative of two biological experiments.Percentage of NET area in the lung normalized to MPO positive signal. **= 0.0022. Data pooled from two biological experiments. Represented is the mean ± SD; each dot represents an individual mouse:= 6 mice/group. Source data are provided as a Source Data file. Statistical analysis was performed using the two-tailed Mann–Whitney test. Ifnar n P n −/− 1a a b

Neutrophil over-activation by type I IFN promotes TB

Fig. 8

Type I IFN activates neutrophils to promote bacterial growth. MRP8-Cre negative (Cre; circles) or positive (Cre; squares) mice were infected and treated with Ctrl Ab (closed symbols) or αGM-CSF (open symbols) as in Fig..Viable bacterial loads determined in the lungs at day 21 post-infection. **< 0.0022. Data representative of two biological experiments. Represented is the mean ± SEM; each dot represents an individual mouse:= 4–6 mice/group/experiment.Representative images of lung sections stained with citrullinated histone H3 (citH3; red), MPO (green), and DAPI (blue). NETs are visualized by colocalization of citH3 and DAPI staining (merged images). Scale bars, 50 µm. Data representative of two biological experiments (= 3–4 mice/group/experiment).Percentage of NET area in the lung normalized to MPO positive signal. Data pooled from two biological experiments. Represented is the mean ± SD; each dot represents an individual mouse:= 7–8 mice/group. Source data are provided as a Source Data file. Statistical analysis was performed using two-tailed Mann–Whitney test. Ifnar P n n n fl/fl neg pos 1a a b c

IFNAR signalling induces NETosis in TB-susceptible mice

Fig. 9

IFNAR signalling is required for severity of lung NETosis in C3HeB/FeJ mice. Schematic of experimental design: C3HeB/FeJ mice were treated by intraperitoneal (i.p.) injection of 0.5 mg of anti-IFNAR (αIFNAR; open circles) or isotype control (labelled as Ctrl Ab; closed circles) mAbs the day prior low dose HN878 aerosol infection, followed by i.p. injections of 0.5 mg of αIFNAR or Ctrl mAbs, three times per weekly until harvest on day 26 post-infection.Viable bacterial loads were determined. **= 0.0079.Representative hematoxylin and eosin (H&E) staining of infected lung sections (= 3 mice/group). Scale bars from top to bottom: 1 mm; 200 µm; 50 µm, 10 µm.Relative lung lesion burden from H&E stained sections (Supplementary Data). 0 = no lesions, 1 = focal lesion, 2 = multiple focal lesions, 3 = one or more focal severe lesions, 4 = multiple focal lesions that are extensive and coalesce, 5 = extensive lesions that occupy the majority of the lung lobe. *= 0.0152.Relative necrosis (left) and intra-alveolar necrotic debris (right) scores from H&E stained lung sections (Supplementary Data). 0 = not present, 1 = minimal, 2 = mild, 3 = moderate, 4 = marked. **= 0.0023.Representative Ziehl–Neelsen staining of infected lung sections. Scale bars, 20 µm.Representative images of lung sections labelled with citrullinated histone H3 (citH3; red), MPO (green), and DAPI (blue). NETs are visualized by colocalization of citH3 and DAPI staining (merged images). Scale bars, 50 μm.Percentage of NET area in the lung normalized to MPO positive signal. **< 0.0012.,,,Data representative of two biological replicate experiments.,,Data pooled from two biological experiments. Represented is the mean ± SD; each dot represents an individual mouse:= 5 mice/group () or= 6–7 mice/group (,,). Source data are provided as a Source Data file. Statistical analysis was performed using the two-tailed Mann–Whitney test. a b c d e f g h b c f g d e h b d e h P n P P P n n 1 1

Discussion

GM-CSF has been increasingly appreciated as an essential mediator of protective immunity against TB^25–31. However, the mechanisms underlying the pathology observed in the absence of GM-CSF signalling during M. tuberculosis infection remain poorly understood, particularly since past studies performed in GM-CSF-deficient mice had confounding lung pathologies^33,34, hampering interpretation of the findings. Here, using blocking mAbs, we describe an IFN-γ-independent mechanism of pathogenesis in the absence of GM-CSF signalling during M. tuberculosis infection. Transcriptomic analyses revealed that blockade of GM-CSF during M. tuberculosis infection results in increased expression of genes associated with neutrophil activation and a type I IFN response, which have been previously associated with disease severity in mouse models and human TB¹⁶. Confirming their role in TB pathogenesis, we show that disease exacerbation driven by either GM-CSF blockade during M. tuberculosis infection or during infection of susceptible C3HeB/FeJ mice was dependent on neutrophil over-activation by type I IFN leading to exacerbated NETosis at the site of infection. NETs were also found in necrotic lung lesions from patients with nonresolving pulmonary TB, further supporting a role for NETosis in TB pathogenesis.

Previous studies addressing the in vivo effects of GM-CSF on the immune response against M. tuberculosis infection have been performed using GM-CSF-deficient mice^27,31. However, in the steady-state, these mice display dramatic defects on alveolar macrophages and lung function, including features of human PAP^33,34,38,39 and deficiencies in other immune cells^52,53. This makes it difficult to distinguish the mechanisms of pathogenesis resulting from M. tuberculosis infection in the absence of GM-CSF from the developmental defects reported in the GM-CSF-deficient mice. Studies using GM-CSF-deficient mice^27,31 have suggested that GM-CSF is required for the recruitment of protective IFN-γ-producing T cells into the site of infection following M. tuberculosis infection. Conversely, our findings showed that GM-CSF blockade during infection did not compromise IFN-γ-producing T cell numbers and instead resulted in increased transcriptional IFN-γ responses in infected lungs. Moreover, impaired control of mycobacterial growth driven by GM-CSF blockade during infection was also observed in the absence of IFN-γ signalling, revealing an IFN-γ-independent mechanism for the resulting disease exacerbation during M. tuberculosis infection in the absence of GM-CSF. Developmental defects in both alveolar macrophages and other immune cells such as dendritic cells^52,53, reported in GM-CSF-deficient mice, may account for the differences in T cell recruitment to the infected lungs observed in the GM-CSF-deficient mice^27,31, but not in the current scenario of lungs of adult mice where GM-CSF was only blocked during infection.

In contrast to the small impact of GM-CSF blockade on the numbers of T cells, neutrophils and monocytes, we found that GM-CSF blockade reduced the numbers of alveolar macrophages in the infected lungs, in keeping with the importance of GM-CSF in the maintenance and function of these cells^38,39. It has been shown that the depletion of alveolar macrophages reduced M. tuberculosis burden in vivo⁵⁴, suggesting that alveolar macrophages are permissive to mycobacterial growth. Therefore, the reduced numbers of alveolar macrophages observed may not explain the increased M. tuberculosis burden observed upon GM-CSF blockade.

We found that GM-CSF blockade perturbed the expression of different networks of genes associated with protective versus damaging immunity in the lungs and blood during M. tuberculosis infection. We have recently shown that transcriptional dominance of a type I IFN response and neutrophil activation, together with a loss of B cell, NK and T cell effector responses contribute to the pathogenesis of M. tuberculosis infection¹⁶. Using distinct but complementary transcriptomic analyses, here we show that GM-CSF blockade boosted transcriptional responses reflective of type I IFN-inducible signalling and neutrophil activation during M. tuberculosis infection, promoting disease exacerbation. The effect of GM-CSF blockade to increase bacterial load was abrogated in the absence of IFNAR on neutrophils, providing an explanation to the link between type I IFN and neutrophil-driven transcriptional signatures. In contrast to the increased type I IFN and neutrophil-driven transcriptional signatures, GM-CSF blockade during infection, decreased the abundance of modules associated with NK, T and B cells, similarly to the decrease in these signatures seen in the blood of TB patients which correlated with advanced disease¹⁶. We also observed a decreased over-abundance in the blood Cell cycle module upon GM-CSF blockade during infection, which may be linked to the increased type I IFN signalling and the anti-proliferative effects of type I IFN⁵⁵.

Neutrophils have been emerging as key players of TB pathogenesis and their influx into the site of infection is associated with poor outcomes^16–20. However, the mechanisms linking neutrophils and increased TB susceptibilty are still unclear⁵⁶. We observed enhanced neutrophil recruitment to the site of infection and NETosis following M. tuberculosis infection in the absence of GM-CSF signalling, in agreement with previous reports of increased neutrophilic infiltration into the lungs of GM-CSF-deficient mice following bacterial infection⁵⁷. Transcriptional analysis revealed that the expression of genes associated with neutrophil recruitment but also activation was increased upon GM-CSF blockade in the blood and lungs from infected mice, suggesting that GM-CSF blockade during M. tuberculosis infection affects not only the recruitment but also the activation status of neutrophils in circulation and at the site of infection, similarly to what was observed during infection of TB-susceptible C3HeB/FeJ mice¹⁶. We found that neutrophils were required for disease exacerbation induced in TB resistant C57Bl/6 mice by GM-CSF blockade during M. tuberculosis infection but additionally in TB-susceptible C3HeB/FeJ mice, and that neutrophil-driven disease exacerbation correlated with excessive NETosis at the site of infection in both models. NETs were originally described as a host antimicrobial defence mechanism against pathogens⁴¹, but recent studies reported that NETs can modulate immune responses and may also induce damaging inflammation in various infections and other inflammatory diseases^{41,42,58–62}. NETs have been shown to be released in vitro by M.tuberculosis-infected mouse⁶³ and human^64,65 neutrophils, however, NETs did not prevent M. tuberculosis replication in vitro⁶³. Consistently, the presence of the NET component citH3 in induced sputum⁶⁵ and neutrophil-related inflammatory proteins associated with NETs in sputum⁶⁵, serum^18,66,67 and lung biopsies^18,65 have been shown to reflect active TB and TB-related lung damage. We now show unequivocally that NETs are induced in human TB lung lesions showing a poor resolution after antibiotic therapy, further supporting a role for NETs in human TB pathogenesis and suggesting that although initially perceived as antimicrobials, NETs may promote TB infection and may limit the ability of antibiotics to access the infected site. Although future mechanistic studies are required to fully decipher the contribution of NETosis to disease progression, these observations suggest that inhibiting NET release in conjunction with antibiotic therapy could be clinically beneficial and prevent long-term morbidity due to lung destruction in TB patients⁶⁸, particularly in cases of MDR or XDR TB where current anti-mycobacterial drugs are of limited use.

Our data are consistent with a critical role for type I IFN in the regulation of NETosis, and in turn exacerbated NETosis serving as a leading effector of type I IFN-induced pathology. In our infection models of increased susceptibility to TB, NETosis was completely abrogated in the absence of type I IFN signalling, which correlated with better control of infection and less tissue damage. Similarly, previous in vitro studies have reported that neutrophils produce NETs when exposed to plasma samples from patients with autoimmune disorders showing elevated levels of type I IFN or exogeneous IFN-α^69,70. These data suggested that type I IFN primes neutrophils for the formation of NETs. Several mechanisms have been proposed for the role of sustained and increased levels of type I IFN in exacerbating M. tuberculosis infection^{7,10,11,13–15,55,71,72}. Our data indicate that this mechanism, that was originally identified in autoimmune and inflammatory conditions, is also critically relevant in certain infection scenarios. Type I IFN appear to be key regulators of neutrophil antimicrobial strategy decisions in both sterile disease and infection, although the mechanisms may vary given that type I IFN signalling acts directly on neutrophils in autoimmune conditions. Although the contribution of type I IFN signalling in other cells to TB disease exacerbation has been previously reported^6,11,71,72, we now demonstrated that increased type I IFN signalling also promotes TB pathogenesis through direct stimulation of neutrophils in vivo during infection, leading to uncontrolled mycobacterial growth at the site of infection. The selectivity of NET release in response to different microbes suggests that NETosis does not always result in an antimicrobial response, such as phagocytosis, to control certain pathogens⁴¹, but that in other infections such as TB contributes to disease exacerbation.

Together, our data from TB patients and mouse models of increased susceptibility to TB by infection of susceptible C3HeB/FeJ mice, or resistant C57Bl/6 mice where GM-CSF signalling is abrogated during infection, indicate that excessive formation of NETs at the site of infection correlates with poor disease outcomes. Our findings provide an insight as to how uncontrolled neutrophil infiltration and activation may exert a detrimental effect during M. tuberculosis infection. Moreover, we show that this process is controlled by type I IFN signalling as demonstrated by reduced NETosis and mycobacterial replication when IFNAR activity was abrogated by using IFNAR-deficient mice or anti-IFNAR mAbs during infection of TB-susceptible mice. These findings provide a mechanism to explain how the reported type I IFN-induced neutrophil signatures in human TB promote disease exacerbation.

Methods

Human samples

The human samples used for this study came from the collection obtained within the SH-TBL project (Supplementary Table 1; 10.17632/knhvdbjv3r.1), led by the Experimental Tuberculosis Unit (UTE) and conducted in collaboration with the National Center for Tuberculosis and Lung Diseases of Georgia (NCTLD); and registered at the ClinicalTrials.gov database under code NCT02715271. The project was reviewed and approved by both the Ethics Committee of the NCTLD (IRB00007705 NCTLD Georgia #1, IORG0006411) and the Germans Trias I Pujol Hospital (IGTP) ethics committee (EC: PI-16-171). Written informed consent was obtained for the collection of biological material and data from all study participants before being enroled.

Experimental animals and ethics

C3HeB/FeJ, C57BL/6 wild-type (WT), IFN-γ receptor-deficient (Ifngr1^tm1Agt; IFNγR^−/−) and type I IFN receptor-deficient (Ifnar1^tm1Agt; Ifnar^−/−) mice were bred and housed in specific pathogen-free facilities at The Francis Crick Institute. Ifnar^flox/flox (Ifnar1^tm1Uka; Ifnar^fl/fl) mice^73,74, which have loxP sites flanking exon 10 of the Ifnar1 gene, were crossed with MRP8-Cre (MRP8 also known as S100a8; Tg^{(S100A8-cre,-EGFP)1Ilw}) mice^50,51, to delete Ifnar in neutrophils. Ifnar^fl/fl-MRP8-Cre^pos mice and Ifnar^fl/fl-MRP8-Cre^neg littermate control mice were used in experiments. All protocols for breeding were performed in accordance with Home Office (U.K.) requirements and the Animal Scientific Procedures Act, 1986. Age-matched females were used in experiments. Experiments were performed in accordance with Home Office (U.K.) requirements and the Animal Scientific Procedures Act, 1986, and with recommendations of the European Union Directive 2010/63/EU and approved by Portuguese National Authority for Animal Health – Direção Geral de Alimentação e Veterinária (DGAV-Ref.0421/000/000/2016). Mice were kept under specific-pathogen-free conditions, with controlled temperature (20–24 °C), humidity (45–65%) and light cycle (12-h light/dark). Mice were kept with food and water ad libitum and humanely euthanized by CO₂ asphyxiation.

Experimental infection

M. tuberculosis experiments were performed under ABSL-3 conditions. M. tuberculosis HN878 bacilli (clinical isolate) were grown to midlog phase in Middlebrook 7H9 broth supplemented with 10% oleic acid albumin dextrose complex (Difco), 0.05% Tween 80, and 0.5% glycerol before being quantified on 7H11 agar plates and stored in aliquots at −80 °C. Aliquots frozen at −80 °C were then thawed (6 aliquots) and quantified, to determine the concentration of the stored inocula. Mice were infected via the aerosol route using an inhalation exposure system, calibrated to deliver ∼200 CFUs to the lung. The infection dose was confirmed by determining the number of viable bacteria in the lungs of five mice immediately after the aerosol infection. Infected mice were monitored regularly for signs of illness such as wasting, piloerection and hunching. Mice were euthanized by CO₂ inhalation and blood and/or lung samples from each group were collected from individual mice at day 21 post M. tuberculosis infection, unless otherwise indicated in the figure legend. Blood and/or lung samples from age-matched uninfected mice were collected at the same time and used as controls. Determination of lung bacterial load was performed by plating serial dilution of the organ homogenate on Middlebrook 7H11 agar supplemented with 10% oleic acid albumin dextrose complex plus PANTA to prevent contamination with other bacteria. CFUs were counted after 3 weeks of incubation at 37 °C, and the bacterial load per lung was calculated.

Antibody-mediated GM-CSF blockade

Anti-mouse-GM-CSF monoclonal antibody (αGM-CSF mAb: clone MP1-22E9, Lot A1102882) and isotype control mAb (rat IgG2a, clone GL117, Lot A1102882; labelled as Ctrl Ab throughout the manuscript and Figs. 1–3 and 6–8) were gifts from DNAX (now Merck, USA). Mice were injected intraperitoneally (i.p.) with 300 μl containing 1 mg of either αGM-CSF or isotype control mAbs diluted in PBS the day prior to M. tuberculosis infection. After infection, mice received twice weekly i.p. injections of 200 μL containing 0.3 mg αGM-CSF or isotype control mAbs for the course of the experiment. Similar results were obtained with mAbs purchased from BioXCell (αGM-CSF mAb: clone MP1-22E9, Lot 683718A1; isotype control mAb: rat IgG2a, clone 2A3, Lot 686318A2).

For IFN-γ blockade experiments, mice received 0.5 mg of either anti-mouse-IFN-γ (αIFN-γ) mAb (clone XMG1.2, Lot 5543-3/5543/0715; from BioXCell) or isotype control mAb (rat IgG1, clone GL113, Lot A2022916; gift from DNAX; labelled as Isotype in Supplementary Fig.) together with αGM-CSF or isotype control mAbs for the course of the experiment. 1d

Antibody-mediated neutrophil depletion

Anti-mouse-Ly6G (αLy6G) mAb (clone 1A8, Lot 695418J3) and isotype control (rat IgG2a, clone 2A3, Lot 71671801; labelled as Isotype in Figs. 3d and 4a) were purchased from BioXCell. M. tuberculosis-infected mice received a total of 7 i.p. injections containing 0.2 mg of αLy6G or isotype control mAbs every other day from day 7 post-infection.

Antibody-mediated IFNAR blockade

Anti-mouse-IFNAR1 (αIFNAR) mAb (clone MAR1-5A3, Lot 0419L555) was purchased from Leinco Technologies, Inc. Isotype control (rat IgG1, clone GL113, Lot A2022916; labelled as Ctrl Ab in Fig. 9) was a gift from DNAX (now Merck, USA). Mice were injected i.p. with 200 μL containing 0.5 mg of either αIFNAR or isotype control (Ctrl Ab) mAbs the day prior to M. tuberculosis infection and then three times per week after infection for the course of the experiment.

Flow cytometry

Single-cell homogenates prepared from harvested lungs from infected mice were stained according to manufacturer’s instructions to exclude dead cells using a Live/Dead fixable red dead cell stain kit (Invitrogen). Cells were pre-treated for 10 min with anti-FcγRI/FcγRII (anti-CD16/CD32; clone 24G2, Harlan; 1:100) Ab and then stained with Abs against specific extracellular markers to identify myeloid cells or lymphocytes. Myeloid cell panel included Thy1.2 (clone 53-2.1, eBioscience; 1:400), Ly6G (clone 1A8, BD; 1:100), Ly6C (clone HK1.4, eBioscience; 1:200), CD11c (clone HL3, BD; 1:100), CD11b (clone M1/70, BD; 1:100). Lymphoid cell panel included Thy1.2 (clone 53-2.1, eBioscience; 1:400), CD3 (clone 145-2C11, eBioscience; 1:200), CD4 (clone RM4-5, eBioscience/BD; 1:200) and CD8 (clone 53-6.7, BD; clone Ly-2, eBioscience; 1:400). For detection of IFN-γ production, cells were restimulated ex vivo with M. tuberculosis tuberculin purified protein derivative (PPD; 20 μg/ml; Statens Serum Institute) and anti-CD28 (2 μg/ml; clone 37.51, Harlan) for 20 h. Brefeldin A (10 μg/ml; Sigma-Aldrich) was added during the last 4 h. After extracellular staining, cells were fixed and treated with permeabilization buffer (BD) according to manufacturer’s instructions and stained with anti-IFN-γ (clone XMG1.2, eBioscience; 1:100) or isotype control (clone eBRG1, eBioscience; 1:100) Abs. All stained samples were fixed with stabilizing fixative (BD) and refrigerated in the dark overnight before being acquired on a CyAN ADP analyser (Dako, Ely, U.K.) using Summit software v4.4.0 (Cytomation). Data were analysed using FlowJo software v10.3 (Tree Star). Myeloid cell populations were analysed after exclusion of dead cells and Thy1.2⁺ cells; neutrophils were gated as CD11b⁺Ly6G⁺, Ly6C⁺ monocytes as CD11b⁺Ly6G⁻Ly6C⁺ and alveolar macrophages as CD11b^lowCD11c⁺. T cell populations were analysed after the exclusion of dead cells and gating on Thy1.2⁺CD3⁺ cells.

Histopathological analysis of lung samples

Lung tissues from M. tuberculosis-infected mice were perfused and fixed in 10% neutral-buffered formalin followed by 70% ethanol, processed and embedded in paraffin, sectioned at 4 µm and stained with hematoxylin and eosin (H&E) or Ziehl-Neelsen (ZN). A single section from all lung lobes was viewed and scored as a consensus by three board-certified veterinary pathologists (S.L.P., E.H. and A.S-B.) blinded to the groups (Supplementary Data 1, which contains data from all original experiments contributing to the study). The relative lesion burden scoring (0–5 points) was determined using the following scale: 0 = no lesions, 1 = focal lesion, 2 = multiple focal lesions, 3 = one or more focal severe lesions, 4 = multiple focal lesions that are extensive and coalesce, and 5 = extensive lesions that occupy the majority of the lung lobe. A semi-quantitative scoring (0–4 points) method was devised to assess the following histological features: inflammatory cells (neutrophils, lymphocytes, macrophages), necrosis, and pleuritis; using the following scale: 0 = not present, 1 = minimal, 2 = mild, 3 = moderate, and 4 = marked changes. The presence of M. tuberculosis bacilli detected by ZN positive staining was scored as paucibacillary or multibacillary according to their abundance in the tissue. Representative images of each group were acquired on an Olympus BX43 microscope using an Olympus SC50 camera and cellSens Entry software (Ver. 1.18).

Immunofluorescence detection of NETs

To identify NETs from lung tissues, sections from M. tuberculosis-infected mice or from TB patients were de-waxed and re-hydrated before staining. Sections were treated with Dako Target Retrieval Solution, pH 9 Ready-to-Use (Agilent) at 97 ^oC for 45 min for antigen retrieval, and permeabilized in PBS 0.5% triton x-100. Samples were incubated with a blocking buffer (PBS with 2% BSA and 2% of donkey serum (Sigma-Aldrich)) for 1 h at room temperature (RT) and stained with rabbit anti-mouse antibodies directed against citrullinated histone H3 (ab5103 from Abcam; 1:500) and goat anti-mouse antibodies directed against MPO (AF3667, R&D systems; 1:40) overnight at 4 ^oC. After washing samples with PBS, sections were incubated with secondary donkey anti-rabbit IgG antibodies conjugated with Alexa Fluor 568 (A10042, Invitrogen; 1:200), donkey anti-goat IgG antibodies conjugated with Alexa Fluor 488 (A11045, Invitrogen; 1:200) and DAPI (Invitrogen; 1:1000) for 2 h in the dark at RT. Glass coverslips were mounted onto the slides using Prolong Gold (Invitrogen). Controls were stained with secondary antibody only, and nonspecific fluorescent staining was not detected when secondary antibodies were tested alone. Slides were analysed using Olympus VS120-L100 Slide Scanner and the OLYMPUS OlyVIA software (v2.9). Representative images for each group were exported using QuPath software (v0.2.0-m5)⁷⁵.

The percentage of NET area normalized to MPO signal was calculated by exporting three regions of interest of 2000 × 2000 μm representative of each lung sections from Qupath (version 0.2.0) and analysed with FIJI (v 2.0.0). The images were split into the different channels; the DAPI channel was subjected to the threshold Triangle, a mask was generated to account for the total amount of tissue present in the region of interest. Default and Triangle thresholding was used with the MPO and the CitH3 channel respectively. The percentage of the area was calculated for the different channels based on the generated masks. The measured percentage of CitH3 signal was normalized to the measured percentage of MPO signal and the % CitH3/MPO values obtained from the three different regions were averaged and plotted. Statistical analysis of the plots was obtained by performing a nonparametric Mann–Whitney test.

RNA extraction and pre-processing for RNA-Seq

Blood was collected in Tempus reagent (Life Technologies) at 1:2 ratio. Total RNA was extracted using the Tempus™ Spin RNA isolation kit (Thermo Fisher Scientific). Globin RNA was depleted from total RNA (1.5–2 µg) using the GLOBINclear™-Mouse/Rat Globin mRNA removal kit (Thermo Fisher Scientific). Lungs were collected in TRI-Reagent (Sigma-Aldrich). Total RNA was extracted using the RiboPure™ Kit (Ambion). RNA quantity was verified using NanoDrop™ 1000/8000 spectrophotometers (Thermo Fisher Scientific). Quality and integrity of the total and the globin-reduced RNA were assessed using a BioAnalyzer 2100 (Agilent).

RNA-Seq data analyses

Biological replicate libraries were prepared from total/globin-reduced RNA (200 ng) using the KAPA mRNA polyA Hyper Prep kit (Illumina) and sequenced on an Illumina HiSeq 4000 sequencer platform generating ~30 million 101 bp paired-end reads per sample. RNA-seq data were subjected to quality control using FastQC and MultiQC. Fastq files were trimmed for Illumina adapter sequences using Trimmomatic (version 0.36) prior to alignment. The RSEM package (version 1.3.30)⁷⁶ in conjunction with the STAR alignment algorithm (version 2.5.2a)⁷⁷ was used for the mapping and subsequent gene-level counting of the sequenced reads with respect to Ensembl mouse GRCm.38.89 version genes. Normalisation of raw count data and differential expression analysis was performed with the DESeq2 package (version 1.18.1)⁷⁸ within the R programming environment (version 3.4.3)⁷⁹. The top 500 genes with the highest variance across each normalised dataset were used to generate the PC analysis plot. The first 10 principal components association was tested against Infection (samples from HN878-infected versus uninfected mice) and αGM-CSF (samples from αGM-CSF versus isotype Ctrl mAbs treated mice) status using linear modelling. Barchart shows the % variance described by the first 10 PCs. The tiled plot shows the significance of each PC with αGM-CSF and Infection status as assessed by an F-test of the associated regression model. Differentially expressed genes were defined as those showing statistically significant differences, P values were adjusted for multiple testing using the Benjamini-Hochberg method (Supplementary Data 2 and 4). Volcano plots showing the logFC and adjusted P value (FDR) were plotted using ggplot2 (version 3.3.2) for differentially regulated genes (FDR < 0.05). Genes were colour coded based on up or downregulated relative to either uninfected Ctrl Ab or infected Ctrl Ab. Differentially expressed genes were plotted using pheatmap (version 1.0.12) using normalised log2 expression values, scaled per row indicating the standard deviation from the mean (Z score). Genes were clustered using a Euclidean distance matrix and clustered using Ward’s hierarchical clustering algorithm⁸⁰. Gene lists ranked by the Wald statistic were used to look for pathway and biological process enrichment using the Broad’s GSEA software (version 2.1.0) with genesets from MSigDB (version 6)⁸¹. Gene lists ranked by the Wald Statistic comparing lung RNA-Seq data from infected αGM-CSF versus infected Ctrl Ab treated mice were also used to determine cell-type enrichment using cell-specific signatures previously curated⁴⁰ and GSEA. Bar plot shows the normalised enrichment score (NES) for specific cell signatures in the lungs from infected αGM-CSF versus infected Ctrl Ab treated mice, where the FDR < 0.05.

Cellular deconvolution analysis for quantification of relative levels of distinct cell types on a per sample basis was carried out on normalized counts using CIBERSORT⁸². CIBERSORT estimates the relative subsets of RNA transcripts using linear support vector regression. Mouse cell signatures for 25 cell types were obtained using ImmuCC and grouped into 9 representative cell types based on the application of ImmuCC cellular deconvolution analysis to the sorted cell RNA-seq samples from the ImmGen ULI RNA-seq dataset (ImmGen Consortium: GSE109125; http://www.immgen.org) as previously described³⁶.

For module enrichment analysis, the Bioconductor package qusage (version 2.18.0)⁸³ was used to determine the enrichment of gene sets identified from human TB blood datasets³⁵ and mouse lung disease datasets³⁶. Fold enrichment results were plotted as a dotplot where red and blue colouring represents up and downregulation of cumulative abundance of module genes relative to controls. The size of the dots represents the relative change in cumulative abundance. Only modules with an adjusted P value < 0.05 are shown. Eigengene values per module were calculated on the VST normalised expression using the module Eigengenes function from WGCNA package.

Reporting summary

Further information on research design is available in thelinked to this article. Nature Research Reporting Summary

Type I IFN exacerbates disease in tuberculosis-susceptible mice by inducing neutrophil-mediated lung inflammation and NETosis

Abstract