Article Text

Original research
CARD9 in neutrophils protects from colitis and controls mitochondrial metabolism and cell survival
  1. Camille Danne1,2,3,
  2. Chloé Michaudel1,3,
  3. Jurate Skerniskyte4,
  4. Julien Planchais1,3,
  5. Aurélie Magniez1,3,
  6. Allison Agus1,3,
  7. Marie-Laure Michel1,3,
  8. Bruno Lamas1,3,
  9. Gregory Da Costa1,3,
  10. Madeleine Spatz1,3,
  11. Cyriane Oeuvray2,3,
  12. Chloé Galbert2,3,
  13. Maxime Poirier1,3,
  14. Yazhou Wang1,3,
  15. Alexia Lapière1,3,
  16. Nathalie Rolhion2,3,
  17. Tatiana Ledent2,
  18. Cédric Pionneau5,
  19. Solenne Chardonnet5,
  20. Floriant Bellvert6,
  21. Edern Cahoreau6,
  22. Amandine Rocher6,
  23. Rafael Rose Arguello7,
  24. Carole Peyssonnaux8,
  25. Sabine Louis8,
  26. Mathias L Richard1,3,
  27. Philippe Langella1,3,
  28. Jamel El-Benna9,
  29. Benoit Marteyn4,10,11,
  30. Harry Sokol1,2,3
  1. 1Université Paris-Saclay, INRAE, AgroParisTech, Micalis Institute, Jouy-en-Josas, France
  2. 2Sorbonne Université, INSERM UMRS-938, Centre de Recherche Saint-Antoine, CRSA, AP-HP, Hôpital Saint-Antoine, Service de Gastroentérologie, F-75012 Paris, France
  3. 3Paris Center For Microbiome Medicine (PaCeMM) FHU, Paris, France
  4. 4CNRS, UPR 9002, Université de Strasbourg, Institut de Biologie Moléculaire et Cellulaire, Architecture et Réactivité de l'ARN, Strasbourg, France
  5. 5Sorbonne Université, INSERM, UMS PASS, Plateforme Postgénomique de la Pitié Salpêtrière (P3S), Paris, France
  6. 6MetaToul-MetaboHUB, National Infrastructure of Metabolomics & Fluxomics (ANR-11INBS-0010), 31077 Toulouse, France
  7. 7Aix Marseille Univ, CNRS, INSERM, CIML, Centre d'Immunologie de Marseille-Luminy, Marseille, France
  8. 8Institut Cochin, INSERM, CNRS, Université de Paris, Laboratoire d'excellence GR-Ex, Paris, France
  9. 9Université de Paris, INSERM-U1149, CNRS-ERL8252, Centre de Recherche sur l'Inflammation (CRI), Laboratoire d'excellence Inflamex, Faculté de Médecine Xavier Bichat, Paris, France
  10. 10University of Strasbourg Institute for Advanced Study (USIAS), Strasbourg, France
  11. 11Institut Pasteur, Université de Paris, Inserm 1225 Unité de Pathogenèse des Infections Vasculaires, 28 rue du Dr. Roux, 75724 Paris Cedex 15, France
  1. Correspondence to Dr Camille Danne, Université Paris-Saclay, INRAE, AgroParisTech, Micalis Institute, Jouy-en-Josas, France; camille.danne{at}; Dr Harry Sokol; harry.sokol{at}


Objectives Inflammatory bowel disease (IBD) results from a combination of genetic predisposition, dysbiosis of the gut microbiota and environmental factors, leading to alterations in the gastrointestinal immune response and chronic inflammation. Caspase recruitment domain 9 (Card9), one of the IBD susceptibility genes, has been shown to protect against intestinal inflammation and fungal infection. However, the cell types and mechanisms involved in the CARD9 protective role against inflammation remain unknown.

Design We used dextran sulfate sodium (DSS)-induced and adoptive transfer colitis models in total and conditional CARD9 knock-out mice to uncover which cell types play a role in the CARD9 protective phenotype. The impact of Card9 deletion on neutrophil function was assessed by an in vivo model of fungal infection and various functional assays, including endpoint dilution assay, apoptosis assay by flow cytometry, proteomics and real-time bioenergetic profile analysis (Seahorse).

Results Lymphocytes are not intrinsically involved in the CARD9 protective role against colitis. CARD9 expression in neutrophils, but not in epithelial or CD11c+cells, protects against DSS-induced colitis. In the absence of CARD9, mitochondrial dysfunction increases mitochondrial reactive oxygen species production leading to the premature death of neutrophilsthrough apoptosis, especially in oxidative environment. The decreased functional neutrophils in tissues might explain the impaired containment of fungi and increased susceptibility to intestinal inflammation.

Conclusion These results provide new insight into the role of CARD9 in neutrophil mitochondrial function and its involvement in intestinal inflammation, paving the way for new therapeutic strategies targeting neutrophils.

  • IBD

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • In inflammatory bowel disease (IBD), alterations of the immune response lead to exacerbated and chronic intestinal inflammation.

  • Caspase recruitment domain 9 (Card9), one of the IBD susceptibility genes, is protective against intestinal inflammation and fungal infection; but the cell types and cellular mechanisms involved remain unknown.


  • Different knock-out mice showed that CARD9 expression in neutrophils, but not in lymphocytes, epithelial cells or CD11c+ cells, protects against intestinal inflammation.

  • In the absence of CARD9, neutrophils die prematurely through apoptosis, due to a dysfunction of their mitochondria that produce excessive mitochondrial reactive oxygen species.

  • The survival defect of neutrophils might explain the increased susceptibility to intestinal inflammation and fungal infection in a context of Card9 deficiency in vivo.


  • Understanding the role of CARD9 and neutrophils in chronic inflammation could lead to innovative therapeutic strategies targeting these key immune cells for various complex diseases.


Inflammatory bowel disease (IBD) results from a combination of genetic predisposition, dysbiosis of the gut microbiota and environmental factors, leading to alterations in the gastrointestinal immune response and chronic inflammation.1 2 Especially, the innate compartment of the immune system has been involved in IBD development, with a role for dendritic cells, macrophages and neutrophils.3–6 Neutrophils, one of the most abundant and important mediators of innate immunity, are professional phagocytes that mount the acute inflammatory response and act as the first line of defence against invading pathogens.7 8 An impaired neutrophil function may result in limited pathogen clearance and fuel a chronic inflammatory response with excessive lymphocyte activation. Patients with congenital disorders in neutrophil function such as chronic granulomatous disease (CGD) often develop IBD-like phenotypes.9–12 Moreover, functional defects have been observed in neutrophils from patients with IBD, including impaired chemotaxis, migration, phagocytosis or reactive oxygen species (ROS) production.4 5

CARD9, one of the numerous IBD susceptibility genes, encodes an adaptor protein that integrates signals downstream of pattern recognition receptors.13–18 Especially, CARD9 is involved in the host defence against fungi via C-type lectin sensing.19 20 CARD9 polymorphisms in humans are associated with multiple susceptibilities including IBD,21 whereas loss-of-function mutations are associated with invasive fungal infections caused by species such as Candida albicans.21–24 CARD9 was shown to mediate its protective functions, at least in part, through the induction of adaptive Th17 cell responses.22 23 25 Card9−/− mice are more susceptible to colitis due to impaired IL-22 production and have an increased load of gut-resident fungi.25 Indeed, CARD9 affects the composition and function of the gut microbiota, altering the production of anti-inflammatory microbial metabolites.26 27 However, the cell types involved in the CARD9 protective role against intestinal inflammation remain unknown.

In this work, we show that CARD9 expression in neutrophils, but not in epithelial or CD11c+cells such as dendritic cells, protects against dextran sulfate sodium (DSS)-induced colitis in mice. The absence of CARD9 impacts neutrophil capacity to contain fungal dissemination, notably by impairing neutrophil mitochondrial function and survival. Indeed, Card9 deletion induces a basal overactivation of mitochondria, increasing the generation of mitochondrial ROS (mtROS) and leading to neutrophil apoptosis. These results provide new insight into the role of CARD9 in neutrophil mitochondrial function and its consequences in intestinal inflammation.


Lymphocytes have no intrinsic role in the Card9-/- susceptibility to colitis

CARD9 was previously reported to be mainly expressed in myeloid cells, especially dendritic cells, macrophages and neutrophils.14 28 Using qRT-PCR analyses in various C57BL/6 mouse organs, we confirmed that Card9 is mainly expressed in immune organs such as bone marrow, spleen and distal small intestine (ileum), but is low at baseline in proximal and mid small intestine, caecum, colon, stomach and liver, and not detectable in Card9-/- tissues (online supplemental figure S1A). Consistently, western-blot analysis showed the expression of CARD9 protein in bone marrow, spleen and distal small intestine of WT mice (online supplemental figure S1B). To dissect Card9 expression at the cellular level, we sorted immune cell populations from spleen and bone marrow of WT mice. Card9 is highly expressed in neutrophils (Ly6G+CD11b+ cells), macrophages (CD11b+F4/80+ cells), CD11c+ cells, including dendritic cells, and monocytes (CD11bhiF4/80+ cells); but the expression is lower in innate or adaptive lymphocytes (CD3+TCRγδ+ lymphoid cells, CD3+CD4+ and CD3+CD8+ T cells, CD3-CD19+ B cells) (online supplemental figure S1C). Thus, CARD9 likely plays a major role within the myeloid immune compartment.

Supplemental material

Figure 1

Lymphocytes have no intrinsic role in the Card9-/- phenotype in colitis models. (A) Weight and (B) Disease Activity Index (DAI) score of DSS-exposed Rag2-/- or Rag2-/- x Card9-/- mice. (C) Colon length and (D) histological score of colon sections at day 7. Data points represent individual mice. One representative experiment out of three. (E) Representative H&E-stained images of mid-colon cross-sections from DSS-exposed Rag2-/- (left) and Rag2-/- x Card9-/- (right) mice at day 7. Scale bars, 500 µm. (F) Weight of mice receiving naive T cells for adoptive transfer of colitis experiment. Rag2-/- received either WT or Card9-/- lymphocytes. Rag2-/- x Card9-/- mice received WT lymphocytes. Data are mean±SEM of two independent experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 as determined by two-way analysis of variance with Sidak’s post-test (A, B, F) and Mann-Whitney U test (C, D). DSS, dextran sulfate sodium; LT, lymphocytes T.

Previous studies from our group and others started to investigate the role of Card9 in murine models of experimental colitis, showing that Card9 deletion increases colitis susceptibility.25 26 29 In order to decipher the respective roles of lymphocytes and myeloid cells in the Card9 susceptibility to intestinal injury and inflammation, we first induced colitis with DSS in Rag2-/- and Rag2-/-xCard9-/- mice that are deficient in functional T and B cells. Mice were euthanised after receiving 3% DSS in drinking water for 7 days, as the severity limit was reached for the Rag2-/-xCard9-/- group. Indeed, desease severity (defined by weight loss, DAI (Disease Activity Index) score, colon length and histological score) was strongly increased in Rag2-/-xCard9-/- mice compared with Rag2-/- mice (figure 1A–E). In an adoptive transfer model of colitis, in which Rag2-/- mice lacking functional lymphocytes received T cells either from WT or Card9-/- mice, no difference was observed on colitis severity (figure 1F), meaning that Card9 expression in T cells does not impact colitis susceptibility. However, the transfer of WT T cells into Rag2-/-xCard9-/- recipient mice did aggravate colitis compared with Rag2-/- simple KO recipients, with a significantly stronger weight loss (figure 1F). These results demonstrate that Card9 mediates its protective role against colitis through the innate immunity compartment, although its role in intestinal epithelial cells cannot be ruled out.

Card9 expression in neutrophils, but not in epithelial or CD11c+ cells, protects against colitis

Based on these findings, we generated conditional KO mice using the cre-lox technology, and obtained mouse strains defective for Card9 either in epithelial cells only (VillincrexCard9lox line), CD11c-expressing cells only, including dendritic cells, macrophages and monocytes (CD11ccrexCard9lox line), or neutrophils only (Mrp8crexCard9lox line). To validate their phenotypes, we isolated epithelial cells from the colonic lamina propria (LP) of Card9Villincre and Card9Villinwt mice, or used MACS separation columns to isolate either CD11c+ or Ly6G+ cell fractions from spleen or bone marrow of Card9CD11cwt and Card9CD11ccre mice or Card9Mrp8wt and Card9Mrp8cre mice, respectively. We performed qRT-PCR on these cell fractions (online supplemental figure S2A), and western-blot analyses on the Ly6G+ and Ly6Glow/- fractions of Card9WT, Card9-/-, Card9Mrp8wt and Card9Mrp8cre mice (online supplemental figure S2B). Results confirmed that Card9 deletion is restricted to the expected cell types (online supplemental figure S2A,B). Purity of Ly6G+CD11b+ neutrophils isolated from the bone marrow of C57Bl/6 mice reached 95% by flow cytometry (online supplemental figure S2C).

Supplemental material

We then assessed the susceptibility of these newly generated mice strains in a model of intestinal inflammation. DSS was administered for 7 days, followed by additional 5 days in which DSS was discontinued. The deletion of Card9 in epithelial or CD11c+ cells did not affect mouse susceptibility to colitis (figure 2A,B and online supplemental figure S2D). However, the deletion of Card9 in neutrophils aggravates colitis compared with WT littermate controls, with significantly increased weight loss from day 8 (figure 2C), DAI score from day 5 (figure 2C), and histological score (figure 2E,F), as well as decreased colon length (figure 2D). Thus, the expression of Card9 in neutrophils plays a crucial role in the protection against intestinal inflammation. The expression of myeloperoxidase (MPO), an anti-microbial enzyme abundantly expressed in neutrophils, was increased in Card9Mrp8cre compared with Card9Mrp8wt colon tissue at day 12, indicating a more important presence or activation of neutrophils in the absence of Card9 (figure 2G). Similarly, the expression of the inflammatory marker lipocalin (Lcn2) was increased in Card9Mrp8cre colon tissue at day 12 (figure 2H).

Figure 2

Card9 expression in neutrophils, but not in epithelial or CD11c+ cells, protects against colitis. (A) Weight and DAI score of DSS-exposed Card9Villinwt or Card9Villincre mice. (B) Weight and DAI score of DSS-exposed Card9cd11cwt or Card9cd11ccre mice. (C) Weight and DAI score of DSS-exposed Card9Mrp8wt or Card9Mrp8cre mice. (D) Colon length, (E) histological score of colon sections and (F) representative H&E-stained images of mid colon cross-sections from DSS-exposed Card9Mrp8wt (left) and Card9Mrp8cre (right) mice at day 12. Scale bars, 500 µm. (G) Myeloperoxidase (MPO) concentration in total colon tissue at day 12. (H) Lipocalin (Lcn2) expression by qRT-PCR in total colon tissue at day 12, normalised to Gapdh. Data points represent individual mice. Data are mean±SEM of three independent experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 as determined by two-way ANOVA with Sidak’s post-test (A–C) and Mann-Whitney U test (D–H). ANOVA, analysis of variance; DAI, Disease Activity Index; DSS, dextran sulfate sodium; ns, not significant.

These results suggest that Card9-deficient neutrophils are efficiently recruited to the inflamed tissue but likely exhibit functional defects preventing them from adequately controlling microbial invaders and thus maintaining inflammation within the intestinal mucosa. We previously showed that the gut microbiota of total Card9-/- KO lineage mice exhibit an altered production of AhR agonists compared with WT.26 However, no difference was observed between Card9Mrp8wt and Card9Mrp8cre mice (online supplemental figure S2E), suggesting that this aspect of the phenotype is not intrinsically related to the role of CARD9 in neutrophils.

Card9 deletion affects the number of activated neutrophils in the inflamed colon

To investigate neutrophil function in WT mice during DSS-induced inflammation, we analysed RNA expression of the neutrophil-specific genes Lcn2, Cxcr2 and S100A8 in colon tissue at days 0, 4, 7, 9, 12 and 16 of DSS-induced colitis (figure 3A and online supplemental figure S2F). The neutrophil recruitment was maximal at day 9, corresponding to the peak of clinical inflammation, and remained high up to day 16 (figure 3A). Histological sections of the distal colon confirmed these findings (figure 3B). The fact that inflammation was higher in Card9Mrp8cre than in Card9Mrp8wt mice (online supplemental figure S2E), and that Lcn2, Cxcr2 and S100a8 expression at day 9 in colon tissue were similar in both genotypes (figure 3C) excludes the hypothesis of a defect of neutrophil recruitment in Card9Mrp8cre mice. We then examined the immune cell populations recruited to the colon LP of Card9Mrp8cre vs Card9Mrp8wt mice at day 9 of colitis. Surprisingly, although the total number of neutrophils was similar in the two genotypes, the percentage and count of mature and activated Ly6G+CD11b+ neutrophils was decreased in the colon LP of Card9Mrp8cre (figure 3D,E). Moreover, the expression levels of both Ly6G and CD11b surface proteins, two major neutrophil maturation and activation markers, were significantly reduced in the overall neutrophil population, as shown by decreased MFIs (for Mean Fluorescence Intensity) (figure 3F,G). These findings suggest a structural or functional defect in neutrophils deleted for Card9.

Figure 3

Card9 deletion affects the number of mature neutrophils in the inflamed colon. (A) Relative expression of Lcn2, Cxcr2 and S1008a in distal colon tissue of C57BL/6 WT mice during a DSS colitis model relative to Gapdh. (B) Representative H&E-stained images of mid colon cross-sections from DSS-exposed WT mice at days 0, 4, 7, 9, 12 and 16 after DSS exposure. Scale bars, 50 µm. (C) Lcn2, Cxcr2 and S100A8 expression in total colon tissue from Card9Mrp8wt and Card9Mrp8cre mice at day 9 of DSS colitis by qRT-PCR analyses, normalised to Gapdh. (D) Representative flow cytometry plots of Ly6G+CD11b+ cells (neutrophils) in the colon lamina propria (LP) of DSS-exposed Card9Mrp8wt and Card9Mrp8cre mice. (E) Percentage and count of Ly6G+CD11b+ cells in the LP of DSS-exposed Card9Mrp8wt and Card9Mrp8cre mice. (F) Ly6G and (G) CD11b expression (MFI, mean fluorescence intensity) of Ly6G+CD11b+ neutrophils from Card9Mrp8wt and Card9Mrp8cre mice. Data points represent individual mice. Data are mean±SEM of two independent experiments. *p<0.05, **p<0.01, ***p<0.001 as determined by Mann-Whitney U tests. DSS, dextran sulfate sodium.

Candida albicans killing capacities are impacted by Card9 deletion in neutrophils

To investigate the defect caused by Card9 deletion in Card9Mrp8cre compared with Card9Mrp8wt neutrophils, we developed several in vitro assays with Ly6G+ neutrophils purified from mouse bone marrow using MACS separation columns. Immunofluorescence, scanning electron microscopy (SEM) and transmission electron microscopy (TEM) analyses did not reveal noticeable structural differences between both genotypes (online supplemental figure 3). On the functional side, we tested the neutrophils ability to kill microorganisms, especially fungi, as CARD9 plays a crucial role in host defence against fungal infection in both humans and mice.20 25 Indeed, C. albicans killing capacities are strongly affected by Card9 deletion in neutrophils (figure 4). An endpoint-dilution survival assay in 96-well plates revealed that twice as many C. albicans cfu (colony forming unit) do survive after 24 hours co-incubation with Card9-/- or Card9Mrp8cre neutrophils compared with Card9WT or Card9Mrp8wt controls, respectively (figure 4A,B). A killing assay using the cfu counting method on agar plates confirmed that Card9Mrp8cre neutrophils have impaired abilities to kill C. albicans compared with Card9Mrp8wt (figure 4C). Card9 deletion in neutrophils did not impact phagocytosis per se, as shown by flow cytometry experiments using fluorescein-isothiocyanate (FITC)-conjugated zymosan (from Saccharomyces cerevisiae cell wall), or cultures of live C. albicans-GFP or E. coli-GFP (online supplemental figure S4A,B). Moreover, we did not observe a difference in the levels of reactive oxygen species (ROS) production over time between neutrophils of the two genotypes in response to phorbol myristate acetate (PMA), a PKC-dependent neutrophil activator or zymosan, a fungal stimulus (online supplemental figure S4C,E). Even though Card9 is involved in autophagy,30 this cellular process did not seem to be affected by Card9 deletion in neutrophils in vitro, as shown by the normality of p62 and LC3BII/I ratio on western-blots (online supplemental figure 4F). Similarly, in our conditions, we did not see clear differences in NETosis pattern between neutrophils deleted or not for CARD9 in response to zymosan (online supplemental figure 4H).

Supplemental material

Supplemental material

Figure 4

Candida albicans killing capacity is impacted by Card9 deletion in neutrophils. (A) Representative images of C. albicans cfu number at the endpoint dilution after infection of Card9WT, Card9-/-, Card9Mrp8wt or Card9Mrp8cre neutrophils for 24 hours, using a microscope (top) or directly showing C. albicans cfu in the 96-well plate, with a photography from above (middle picture) or a scan from below the plate (bottom picture). (B) C. albicans cfu after infection of neutrophils for 24 hours. Cfu counting was performed in 96 well plates using a microscope. Data points represent individual mice. Data are mean±SEM of three independent experiments. (C) C. albicans cfu number after infection of Card9Mrp8wt or Card9Mrp8cre neutrophils for three or 24 hours. Cfu counting was performed after plating on YEPD agar plates. Data are mean±SEM of two independent experiments. (D) Experimental design of antibiotic treatments and C. albicans inoculation in mice treated with 3% DSS. (E) weight, (F) DAI and (G) colon length of DSS-exposed Card9Mrp8wt and Card9Mrp8cre mice after colonisation with C. albicans. (H–K) Fungal burden in the faeces (H), caecal content (I), caecal and colon tissue (J) and liver, spleen and kidney (K) of DSS-exposed Card9Mrp8wt and Card9Mrp8cre mice after colonisation with C. albicans. Data are mean±SEM of two independent experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 as determined by one-way ANOVA with Tukey’s post-test (A), two-way ANOVA with Sidak’s post-test (E, F) and Mann-Whitney test (C, G, I–K). Genta, gentamycin. ANOVA, analysis of variance; DSS, dextran sulfate sodium.

We, thus, followed the track of fungal killing to investigate whether the absence of Card9 expression in neutrophils drives a general impairment of the immune system to control infection. During a DSS-induced colitis model, we detected a slight but non-significant increase in total fungi/bacteria DNA ratio at days 7 and 12 in the faeces of Card9Mrp8cre compared with Card9Mrp8wt mice (online supplemental figure S4G). These data suggest a reduced ability of Card9Mrp8cre mice to contain fungi expansion in the inflamed intestine. However, this effect might be difficult to point out due to the very low fungal abundance in SPF mouse microbiota. Thus, to go further and reveal a potential phenotype, we induced colitis in mice treated with a broad-spectrum antibiotic and antifongic cocktail and gavaged with C. albicans to expand the intestinal fungal load (figure 4D). The increased colitis severity in Card9Mrp8cre mice was maintained in this setting (figure 4E–G). C. albicans load was slightly increased (although not statistically significant) in the caecal content (figure 4H,I), colon and caecal tissues (figure 4J) of Card9Mrp8cre mice, and the number of cfu recovered from liver, spleen and kidney were significantly higher compared with Card9Mrp8wt mice (figure 4K). These results show that Card9 expression in neutrophils is crucial to control the fungal load in the inflamed gut, the direct translocation of C. albicans from the gut to the liver, and to avoid its systemic dissemination.

The absence of Card9 impacts neutrophils survival by increasing apoptosis

To explore the mechanisms underlying the impaired capacity of Card9Mrp8cre neutrophils to kill C. albicans despite intact phagocytosis, autophagy and ROS production, we investigated neutrophils survival rates by flow cytometry using an AnnexinV-FITC assay coupled to Live/Dead staining (figure 5). AnnexinV reveals ongoing apoptosis, whereas live/dead only stains permeable dead cells. Interestingly, after 1-hour incubation at 37°C, Card9Mrp8cre neutrophils showed a significant increase in AnnexinV-FITC MFI compared with Card9Mrp8wt neutrophils (figure 5A,B). Consistently, an increase in percentages of apoptotic (Q1, AnnexinV+LD-) and dead (late apoptotic/necrotic) neutrophils (Q2, AnnexinV+LD+), and a decrease in viable neutrophils (Q4, AnnexinV-LD-), were observed for the Card9Mrp8cre genotype (figure 5C). Moreover, the surface expression of the CD62L marker (a surface protein that is lost on cell activation) was reduced in Card9Mrp8cre, with a lower percentage of CD62L+ cells and a higher percentage of CD62L- cells (figure 5D). These results suggest an excessive basal activation of Card9Mrp8cre compared with Card9Mrp8wt neutrophils. Similar results were obtained with Card9-/- versus Card9WT neutrophils (online supplemental figure S5A–D), confirming the impact of Card9 on neutrophil survival.

Supplemental material

Figure 5

The absence of Card9 impacts neutrophils survival by increasing apoptosis. (A) Representative flow cytometry plots of Ly6G+CD11b+ neutrophils purified from bone marrow of Card9Mrp8wt (left) and Card9Mrp8cre (right) mice, co-stained with AnnexinV and a Live/Dead marker. (B) AnnexinV MFI of Ly6G+CD11b+ neutrophils from Card9Mrp8wt and Card9Mrp8cre mice, incubated for 1 hour at 37°C. (C) Percentage of apoptotic neutrophils (Q1: AnnexinV+LD- cells), dead (late apoptotic/necrotic) neutrophils (Q2: AnnexinV+LD+ cells) and viable neutrophils (Q4: AnnexinV-LD- cells) among the Ly6G+CD11b+ population. (D) Percentage of CD62L+ and CD62L- neutrophils among the Ly6G+CD11b+ population. Data represent one out of two independent experiments. (E) Histogram representing Gene Ontology biological processes significantly enriched among proteins downregulated (blue) or upregulated (red) in Card9Mrp8cre compared with Card9Mrp8wt neutrophils in the unstimulated condition. (F) Morpheus heat-map representing mitochondria-related proteins significantly downregulated or upregulated in Card9Mrp8cre compared with Card9Mrp8wt neutrophils in the unstimulated condition. (E, F) Obtained from proteomics data analysis. *p<0.05, **p<0.01, ****p<0.0001 as determined by Mann-Whitney U test (B) and two-way ANOVA with Sidak’s post-test (C–D). ANOVA, analysis of variance; MFI, mean fluorescence intensity.

To go further, we performed a proteomic analysis on Card9Mrp8cre vs Card9Mrp8wt neutrophils after 1-hour incubation at 37°C. This approach revealed that a large number of proteins related to cellular metabolism pathways were differentially regulated between both genotypes. Indeed, a gene ontology (GO) functional analysis showed the enrichment in numerous cellular metabolic processes among proteins statistically downregulated or upregulated in Card9Mrp8cre compared with Card9Mrp8wt neutrophils (figure 5E and online supplemental figure S5E). Especially, we observed a high prevalence of mitochondrial proteins among the differentially regulated candidates between the two genotypes (figure 5F), suggesting that neutrophil mitochondrial functions are impacted by the absence of Card9.

Card9 controls neutrophil survival by affecting mitochondrial function

Subsequently, we analysed neutrophil mitochondrial function using MitoTracker Green (reflecting mitochondrial mass) and MitoTracker Red (reflecting mitochondrial membrane potential) markers in a flow cytometry assay (figure 6A). No difference was observed in terms of MFIs, but we observed a significant increase in the percentage of MitoGreen+MitoRed- neutrophils (with dysfunctional or metabolically inactive mitochondria) in the Card9Mrp8cre compared with the Card9Mrp8wt genotype (figure 6A). Tetramethylrhodamine methyl ester (TMRM) assay evaluating the mitochondrial membrane potential confirmed the increase in apoptotic, metabolically stressed cells (TMRM-) among Card9Mrp8cre neutrophils (figure 6B). These results suggest that the survival defect of Card9Mrp8cre neutrophils may be due to an altered energetic metabolism. Real-time bioenergetic profile analysis using Seahorse technology showed that Card9Mrp8cre neutrophils have a higher basal Oxygen Consumption Rate during a cell mito stress assay (figure 6C), indicating a higher oxidative phosphorylation activity compared with Card9Mrp8wt neutrophils. Basal respiration and ATP production rate were both highly increased in Card9Mrp8cre neutrophils (figure 6D). Moreover, a Seahorse real-time ATP rate assay demonstrated that Card9Mrp8cre neutrophils have an increased mitochondrial ATP (mitoATP) production rate, whereas glycolytic ATP (glycoATP) production rate is only mildly decreased, as shown by the ATP rate index and the energetic map (figure 6E). A Seahorse glycolytic rate assay confirmed the moderate decrease of the basal Extracellular Acidification Rate of Card9Mrp8cre neutrophils (online supplemental figure S6A). However, metabolomics analysis on neutrophil supernatants incubated for 24 hours revealed a decreased in lactate production, leading to a reduced lactate/glucose ratio in Card9Mrp8cre neutrophils (online supplemental figure S6B). Similar results were obtained when we compared Card9WT and Card9-/- neutrophils in the assays described above (online supplemental figure S6C–E). Thus, Card9 deletion in neutrophils induces an overactivation of mitochondria and tends to reduce glycolytic activity, which is the major energy source of normal neutrophils. Blocking glycolysis with 2DG (2-Deoxy-D-glucose) increased the apoptotic rate of Card9Mrp8wt but not of Card9Mrp8cre neutrophils; showing that glycolysis is more essential to Card9Mrp8wt than Card9Mrp8cre neutrophils (figure 6F, left). Conversely, blocking the mitochondrial respiratory chain with oligomycineA increased the apoptotic rate of Card9Mrp8cre but not of Card9Mrp8wt neutrophils, demonstrating the crucial role of mitochondria as an energy source in Card9Mrp8cre neutrophils (figure 6F, right). Thus, contrary to Card9Mrp8wt neutrophils that mainly rely on glycolysis as a source of energy, Card9Mrp8cre neutrophils present an altered metabolism with an overactivation of mitochondria associated with a dysfunctional state, leading to apoptosis. Indeed, we showed by MitoSOX staining that the production of mitochondrial ROS (mtROS) is increased in Card9Mrp8cre compared with Card9Mrp8wt neutrophils (figure 6G, left). This difference is increased by 2-DG treatment as mitochondrial respiration is promoted (figure 6G, right). Conversely, blocking mitochondrial respiration with oligomycinA decreases the differences between Card9Mrp8cre and Card9Mrp8wt neutrophils (figure 6G, right). These results show that the increased mitochondrial respiration in Card9Mrp8cre neutrophils generates increased mtROS production, major determinant of cell apoptosis.31 Intestinal inflammation is associated with a high degree of oxidative stress. To evaluate the effect of oxidative stress on the phenotype of Card9Mrp8cre neutrophils, they were treated for 1 hour with H2O2. The increased apoptosis and necrosis rates observed in Card9Mrp8cre neutrophils were stronger in oxidative stress than in basal condition (figure 6H). These results show that the role of CARD9 in the survival capacity of neutrophils, especially in oxidative conditions, is mediated by effects on mitochondrial functions. Consistently, blood neutrophils from patients with IBD carrying the CARD9 single-nucleotide polymorphism most commonly associated with IBD (rs10781499, homozygous state) had an increased mortality rate compared with age-matched and sex-matched control patients with IBD and healthy subjects (figure 6I).

Supplemental material

Figure 6

Card9 controls neutrophil energetic metabolism by affecting mitochondrial function. (A) Representative flow cytometry plots and precentages of MitoGreen+MitoRed- cells (corresponding to cells with dysfunctional mitochondria) among Ly6G+CD11b+ neutrophils. Neutrophils were purified from the bone marrow of Card9Mrp8wt or Card9Mrp8cre mice, and incubated for 1 hour at 37°C. (B) Representative flow cytometry plots and percentages of TMRM- cells (corresponding to cells with dysfunctional mitochondria/apoptotic or metabolically inactive cells) among Ly6G+CD11b+ neutrophils. (C) Oxygen consumption rate (OCR) of Card9Mrp8wt and Card9Mrp8cre neutrophils measured during a Seahorse Cell Mito Stress assay. (D) Basal respiration (late rate measurement before oligomycin injection (t3) - non-mitochondrial respiration rate (t12)) and ATP production rate (late rate measurement before oligomycin (ATP synthase blocker) injection (t3) – minimum rate measurement after oligomycin injection (t6) obtained from the Seahorse Cell Mito Stress assay. (E) ATP rate index and energetic map of Card9Mrp8wt and Card9Mrp8cre neutrophils obtained from the Seahorse Real-time ATP rate assay. (F) Percentage of apoptotic cells among the Ly6G+CD11b+ neutrophil population of Card9Mrp8wt and Card9Mrp8cre genotypes after treatment with 2-DG 10 mM (left) or oligomycin 1.5 μM (right) for 1 hour. (G) Representative flow cytometry plots and precentages of MitoSOXhigh cells (corresponding to cells with high mitochondrial ROS) among Card9Mrp8wt and Card9Mrp8cre neutrophils incubated for 1 hour at 37°C. (H) Percentage of apoptotic and necrotic cells among the Ly6G+CD11b+ neutrophil population of Card9Mrp8wt and Card9Mrp8cre genotypes after addition of H2O2 0.01 mM for 1 hour to increase oxydative stress. (I) Death rate of human blood neutrophils from patients with IBD carrying the CARD9 single-nucleotide polymorphism most commonly associated with IBD (rs10781499, homozygous state, n=8) and age-matched and sex-matched control patients with IBD (n=11) and healthy subjects (n=6). Data are mean±SEM of at least two independent experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 as determined by Mann-Whitney test (A, B, D, E, I) or two-way ANOVA with Sidak’s post-test (F, G, H). Norm. Unit, normalised unit; FCCP, Carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone; Rot/AA, Rotenone/AntimycinA; 2DG, 2 deoxy-glucose; OligoA, oligomycinA. ANOVA, analysis of variance; IBD, inflammatory bowel disease; ROS, reactive oxygen species.


Altogether, this study reveals that, in the absence of CARD9, mitochondrial dysfunction in neutrophils leads to their premature death through apoptosis, especially in an oxidative environment. The decrease of functional neutrophils in the gut affects fungal containment and increases susceptibility to intestinal inflammation. CARD9 polymorphisms are associated with IBD. Mouse studies showed the contribution of CARD9 to host defence and intestinal barrier, notably through the production of IL-22 and the modulation of the gut microbiota metabolic activity.25 26 However, the role of CARD9 in disease pathogenesis has not been elucidated at the cellular level. Our study reveals that Card9 deletion in neutrophils, contrarily to epithelial or CD11c+ cells, increases the susceptibility to DSS-induced colitis. We thus focused on studying the role of CARD9 expression in neutrophil functionality. Indeed, neutrophils have been studied in different models of IBD and fungal infection, but their direct contribution to pathogenesis and the role of CARD9 in these mechanisms remain poorly understood.4 5 32 33

Human CARD9 deficiency results in impaired neutrophil fungal killing, leading to a selective defect to contain invasive fungal infection.34 In both humans and mice, CARD9 is required in microglia for neutrophil recruitment and control of fungal infection in the central nervous system.35 36 CARD9 signalling was also involved in neutrophil phagocytosis and NETosis (neutrophil extracellular traps) functions, enhancing mouse survival to a lethal dose of C. albicans.37 Moreover, CARD9 expression in neutrophils promotes autoantibody-induced arthritis and dermatitis in mice,38 and inflammation levels in a mouse model of neutrophilic dermatosis.39 In line with these studies, we found that Card9 deletion affects the capacity of neutrophils to kill fungi in vitro and in vivo, but we found no impact on the neutrophil structure, ROS production, autophagy, phagocytosis or NETosis. We have not examined the role of CARD9 in chemotaxis because, in the context of fungal infection in patients with CARD9 deficiency, neutrophil-intrinsic chemotaxis was not affected.24 In DSS-induced colitis, Card9 deficiency in neutrophils does not impact their recruitment to the colon, but reduces the number of mature neutrophils. We do not think that impaired antimycotic function explains the increased susceptibility to colitis but rather reveals that CARD9-deficient neutrophils have a major functional defect. Indeed, we discovered that the absence of CARD9 increases mortality in both mouse and human neutrophils. This premature death is caused by mitochondria overactivation and the increased production of mtROS. Indeed, mtROS are generated as byproducts of mitochondrial respiration and are major determinants of cell apoptosis.31 Other CARD proteins mediate apoptotic signalling through CARD-CARD domain interactions.13 In addition to mediating inflammation, CARD9, a member of the CARD proteins family, was recently shown to inhibit mitochondria-dependent apoptosis of cardiomyocytes under oxidative stress.40 Here, we show that CARD9 mediates mitochondrial function and apoptosis in neutrophils, especially in an oxidative environment. Interestingly, mitochondrial dysfunction was observed in human in the context of active UC.41 Further investigation is required to fully elucidate the impact of mitochondria overactivation on neutrophil function, especially in vivo. The impact of Card9 on neutrophil mitochondrial function might not fully explain the increased susceptibility of Card9Mrp8cre mice to DSS colitis. Moreover, other cell types are likely involved in the Card9-/- mice phenotype, explaining the impact of Card9 deletion on microbiota metabolic activity, as we show that this is not dependent on neutrophils.25 26

Neutrophils contain very few mitochondria compared with other leukocytes, and depend mainly on glycolysis to produce ATP, which is essential to perform their designated tasks. This allows energy generation in a low-oxygen environment and keeps oxygen available for neutrophil effector functions.42 Thus, the dependence of neutrophils on glycolysis could be an adaptation to allow oxygen to be used in the anti-microbial response rather than entering oxidative phosphorylation.42 Here, we show that in the absence of CARD9, mitochondria overactivation, leads to the premature death of neutrophils through apoptosis and the loss of their anti-microbial functions. DSS and anti-CD40 colitis models performed with the addition of anti-Ly6G antibody, which specifically depletes neutrophils, show that the absence of neutrophils strongly exacerbates gut inflammation and impacts mouse survival.43 In humans, neutropenia is associated with severe intestinal inflammation, in the case of neutropenic enterocolitis44 and IBD-like phenotype in phenotype in patients with CGD45 for instance. Excessive oxidative phosphorylation and mitochondrial ATP production in neutrophils can also damage the surrounding tissues and exacerbate inflammation. This could explain the increased susceptibility to intestinal colitis in the absence of Card9, especially in inflammatory environments where neutrophil activity is normally tightly controlled by the low oxygen pressure. In humans, the « glycogen storage disease type Ib » induces a functional defect of neutrophils due to glycolysis dysfunction and impaired energy homeostasis.46 This disease is associated with IBD-like phenotypes, highlighting the importance of neutrophil metabolism in intestinal health.46 47 Immunometabolism is a central concept as immunity and metabolism impact each other in both ways: (1) energetic metabolism impacts immune function, which is well documented for lymphocytes and macrophages,48 but not for neutrophils; and (2) we show that a protein known for its roles in innate immunity, CARD9, can also impact cell metabolism. Further investigation is required to understand how CARD9 does interact with mitochondrial function in a direct or indirect manner.

Neutrophils are involved in various diseases, including infection, cardiovascular diseases, inflammatory disorders and cancer, which makes them exciting targets for therapeutic intervention.49 Despite the complex implication of neutrophils in disease, various therapeutic approaches aim to enhance, inhibit or restore neutrophil function, depending on the pathology. In inflammatory diseases with excessive neutrophil activity, their attenuation could be desired, even though killing functions against microorganisms might still need to be preserved. Our work shows that increased apoptosis in neutrophils does not alleviate intestinal inflammation, even though these cells are known to contribute to disease development. Recent studies indicate substantial phenotypic and functional heterogeneity of neutrophils.49 Thus, targeting a specific subpopulation may allow the attenuation of harmful aspects of neutrophils without compromising host defence. Some neutrophil-targeted therapeutic strategies have gained interest, notably in the context of IBD with positive effects of the growth factor granulocyte colony-stimulating factor (G-CSF).50 Recently, anti-GM-CSF autoantibodies have also been shown to precede Crohn’s disease onset by years.51 It opens promising options for numerous complex pathologies.

Material and methods


Card9-/-, Rag2-/-xCard9-/-, Rag2-/-, Card9loxxMrp8cre, Card9loxxVillincre and Card9loxxCD11ccre mice on C57BL/6J background were obtained from the Saint-Antoine Research Centre and housed at the IERP (INRAE, Jouy-en-Josas) under specific pathogen-free conditions. Animal experiments were performed according to the institutional guidelines approved by the local ethics committee of the French authorities, the ‘Comité d’Ethique en Experimentation Animale’ (COMETHEA, CEEA45).

Induction of DSS colitis and colonisation with C. albicans

Mice were administered drinking water supplemented with 2–3% (wt/vol) DSS (MP Biomedicals) for 5–7 days (depending on colitis severity of each experiment), and then water only for 5 days. Animals were monitored daily for weight loss. For C. albicans colonisation, mice were treated with 0.4 mg/mL streptomycin, 300 U/mL penicillin G and 0.125 mg/mL fluconazole as indicated in figure 4.


Colon samples were fixed, embedded in paraffin and stained with H&E. Slides were scanned and analysed to determine the histological score (online supplemental table S1).25

Supplemental material

Faecal DNA extraction and total bacteria and fungi quantification

Faecal DNA was extracted as previously described.26 Luna Universal qPCR Master Mix (New England Biolabs) was used for quantification of fungal ITS2 sequences and TaqMan Gene Expression Assays (Life Technologies) for quantification of bacterial 16S rDNA sequences.

Cell preparation and stimulation

Epithelial cells were isolated from colonic tissue using a dithiotheritol (DTT)/ethylenediaminetetraacetic (EDTA) buffer. Neutrophils were purified from mouse bone marrow using anti-Ly6G MicroBeads UltraPure and MACS separation columns (Miltenyi Biotec). CD11c+ cells were purified from mouse spleen using anti-CD11c MicroBeads UltraPure and MACS separation columns (Miltenyi Biotec). Purity checks and cell counts were performed using a BD Accuri C6 flow cytometer (BD Biosciences). After purification, neutrophils were seeded in 96-well suspension plate (Sarstedt), rested for 30 min in RPMI 1640 Medium (Gibco, ThermoFisher Scientific) before addition of 2% heat-inactivated Fetal Bovine Serum (FBS), 2-DG 10 mM, oligomycine 1.5μM or H2O2 0.01M and incubated at 37°C as indicated. Isolation of LP immune cells was performed as previously described.52

Gene expression analysis using quantitative RT-PCR

Total RNA was isolated from colon samples or cell suspensions using RNeasy Mini Kit (Qiagen), and quantitative RT-PCR performed using QuantiTect Reverse Transcription Kit (Qiagen) and Luna Universal RT-PCR Kit (New England Biolabs) in a StepOnePlus apparatus (Applied Biosystems) with specific mouse oligonucleotides (online supplemental table S2). We used the 2−ΔΔCt quantification method with mouse Gapdh as an endogenous control and the WT group as a calibrator.

Supplemental material

Immunoblot Analysis. Mouse tissue or cell suspensions were lysed using Laemmli buffer, loaded on a SDS-PAGE and analysed with antibodies against CARD9 (A-8:sc-374569, Santa Cruz Biotechnology), or β-ACTIN (D6A8, CST).

Flow cytometry, cell sorting and functional assays

Flow cytometry was carried out by using LSR Fortessa X-20 (BD) and cell sorting FACS Aria machines (BD). For apoptosis assay, 4×105 neutrophils were stained with AnnexinV-FITC in Binding Buffer (Miltenyi Biotec) and Live/Dead (Zombie Aqua Fixable, BioLegend). For mitochondria analyses, MitoTracker Green and MitoTracker Red FM or MitoSOX Red Mitochondrial Superoxide Indicator, were added to neutrophils for 15 min at RT in MACS buffer or PBS, respectively (ThermoFisher Scientific). Alternatively, neutrophils were incubated with TMRM in RPMI for 20 min at 37°C (Abcam). For phagocytosis assay, 105 neutrophils stimulated with zymosan-FITC (50μg/ml, Fluorescein zymosanA BioParticles conjugates, FisherScientific), C. albicans-GFP (MOI1:1) or Escherichia coli-GFP (MOI1:10) for 45 min. Cells were stained with surface antibodies in MACS buffer (online supplemental table S2).

Endpoint dilution survival assay

Isolated mouse neutrophils were seeded at 105 cells/well in 96-well plates in RPMI+2% FBS and infected with C. albicans (serial fourfold dilutions of an OD=1 solution were added by raws (row1, MOI20:1; row2, MOI5:1…). After 24 hours at 37°C, colonies were visualised with a Nikon TMS inverted microscope and counted at the lowest dilutions (row7-8).

Killing assay

Purified neutrophils were seeded in 96-well plate at 106 cells/well and stimulated with PBS, C. albicans (MOI1) or E. coli (MOI10) for 90 min at 37°C. Cells were washed and lysed with 200μl TritonX100 0.025%. Serial dilutions were plated on YEPD or LB plates.

Oxydative burst

Experiments were performed on a TriStar LB942 Reader using 105 neutrophils in 200 µL Hank's Balanced Salt Solution (HBSS, ThermoFisher Scientific) and luminol 80μM (Sigma) and stimulated with PMA (0.1 µg/mL; Sigma) or opsonised zymosan (20 mg/mL zymosan A from Saccharomyces cerevisiae; Sigma). The indexed maximal relative luminescence (in relative light units (RLU)) was calculated as follow: indexed RLU max=((haemochromatosis patient maximal RLU)/(healthy donor maximal RLU))×100. Alternatively, superoxyde dismutase (5 units/mL) and catalase A (10Units/mL) were added to differentiate total and intracellular ROS production. Absorbance was measured at 550 nm for 30 min.

Real time bioenergetic profile analysis

Mito Stress Test, Glycolytic Rate Assay and Real-time ATP assay were performed on a XF96 Extracellular Flux Analyzer (Seahorse Biosciences). Mouse neutrophils were seeded at 2×105 cells/well in RPMI+2% FBS in 0.01% poly-L-lysine pre-coated plate. After 2-hour rest, cells were washed in Seahorse RPMI medium and incubated for 1 hour at 37°C without CO2. In the analyzer, oligomycin 1.5 µM, FCCP 1 µM, rotenone+antimycinA 0.5 µM and 2DG 50 µM were injected at the indicated times. Protein standardisation was performed after each experiment, with no noticeable differences in protein concentration and cell phenotype.

Electron microscopy

Purified neutrophils were stimulated with C. albicans (MOI 2) or E. coli (MOI 10) for 1 hour at 37°C, washed in PBS and fixed. Samples preparation and SEM/TEM analyses were performed at the Microscopy and Imaging Platform MIMA2 (Université Paris-Saclay, INRAE, AgroParisTech, Jouy-en-Josas, France,


The 5 µg protein extracts were submitted to in-gel digestion. Desalting was performed as described before.53 Peptides were analysed on a nanoElute-timsTOF ProLC-MS/MS system (Bruker).54 Raw files were analysed using MaxQuant V. database UP000000589_10090 (21 994 entries, 17 June 2020). Data filtering, imputation and statistical analysis were performed with ProStar Zero V. Proteins with False Discovery Rate (FDR <5% (Pounds method) were significant with a fold change >1.2.

Nuclear magnetic resonance

The 200 µL culture media were analysed by 1D 1H-NMR. All NMR spectra were recorded on a Bruker AvanceIII 800MHz spectrometer equipped with a QPCI 5 mm cryogenic probe head. Spectra were acquired and processed using the Bruker Topspin V.4.0 software. Quantification of glucose and lactate was performed using addition of 25% TSPd4 in D2O as internal standard.

Patients and samples collection

A diagnosis of IBD was defined by clinical, radiological, endoscopic and histological criteria. Patient characteristics are presented in online supplemental table S3. Human blood neutrohpils were isolated using the technique of dextran sedimentation and Ficoll density gradient, and stained with AnnexinV-FITC (Miltenyi Biotec) and a Live/Dead marker (Zombie Aqua Fixable, BioLegend). Data were acquired using a CytoFlex cytometer (Beckman).

Supplemental material

Statistical analysis

Statistical analysis was performed using GraphPad Prism V.7 software.

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

Ethics statements

Patient consent for publication

Ethics approval

Approval was obtained from the local ethics committee (Comite de Protection des Personnes Ile-de-France IV, IRB 00003835, Suivitheque study, registration number 2012/05NICB). Participants gave informed consent to participate in the study before taking part.


Supplementary materials


  • Twitter @h_sokol

  • Contributors CD and HS conceived and designed the study, performed data analysis and wrote the manuscript and both act as guarantor; CD designed and conducted all experiments, unless otherwise indicated; CM, JP, AM, AA, M-LM, BL, GDC, MS, CG, MP, YW and AL provided technical help for the in vitro and/or in vivo experiments; JS and BM performed the immunofluorescence microscopy experiments; CéP and SC conducted the proteomics analyses; FB, EC and AR performed the metobolomics analysis; TL provided mice and CO performed genotyping; CaP, SL and JE-B helped with ROS production assays; RRA provided Scenith kits for tests with neutrophils; CD, CM, JP, MLR, PL, NR, JE-B, BM and HS discussed experiments and results.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.