Main

The immune system needs to balance immune responses to be able to control infectious challenge but at the same time avoid excessive inflammation that could generate host tissue damage3,4. Neutrophils, the most abundant immune cell type in circulation, contribute to the first line of defence against a large number of pathogens and are critical in this fine balance as their antimicrobial responses must be tightly regulated to maintain homeostasis5,6. Neutrophils are activated based on pro-inflammatory signals that trigger various effector functions depending on their surface receptor composition and intracellular protein content7,8. For instance, neutrophils can respond to microorganisms via phagocytosis, generation of reactive oxygen species (ROS), and degranulation or the release of NETs2. NETs are defined as extracellular structures composed of DNA and cytosolic, granular and nuclear proteins assembled on a scaffold of decondensed chromatin1. NETs contain, neutralize and kill microorganisms including fungi, bacteria and parasites, but they can also be released in response to other stimuli, including crystals and immune complexes9,10,11. Of note, neutrophils and, in particular, NETs that are released following cell activation are associated with autoimmune pathogenesis, such as in rheumatoid arthritis or systemic lupus erythematosus (SLE)12,13. Moreover, NETs have been linked to the development of autoantibodies that are also associated with progression and severity of autoimmune diseases6,7. In inflammatory diseases, such as COVID-19 infection, neutrophils can acquire a persistent inflammatory signature that leads to increased NET release, which is associated with disease severity14,15,16.

NETs function as immune stimulants, acting as damage-associated molecular patterns that induce local inflammation and tissue damage7. NETs contain various molecules recognized by immune cell receptors. For example, following internalization of NETs by macrophages and dendritic cells, the cytosolic sensor cyclic GMP–AMP synthase (cGAS) recognizes the DNA backbone of NETs17. In addition, the cell-surface receptor TLR9 also recognizes NET–DNA in a process that is facilitated via two NET proteins: LL-37, which potentiates a type I IFN induction, and HMGB1, which acts through the receptor for advanced glycation products (RAGE)18,19,20,21,22. Recently, it has been shown that TLR4 recognizes and is activated by NET-associated histones, whereas histone containing chromatin DNA regulates TLR4 recruitment to endosomes23. However, the mechanisms that regulate cellular responses to NETs are still incompletely understood23.

Loss or mutation of inhibitory receptors are often associated with unchecked inflammation and destructive autoimmunity24. Recent work from our group has indicated that MICL (also known as CLEC12A) is one such receptor, which can regulate the pathogenesis of rheumatoid arthritis25. In contrast to wild-type (WT) mice, we showed that MICL-deficient animals (Micl−/−) undergoing collagen antibody-induced arthritis (CAIA) presented with exacerbated joint inflammation that did not resolve25. Through investigating the underlying mechanism, we discovered that MICL functions as an essential pattern recognition receptor (PRR) for NETs on neutrophils. Recognition of NETs by MICL inhibits neutrophil activation and further NET formation, regulating a positive-feedback cycle that, on the one hand, protects from excessive inflammation during multiple autoimmune diseases, but on the other hand, increases susceptibility to invasive infections.

MICL regulates neutrophil responses

Micl/− mice exhibit enhanced and non-resolving joint inflammation compared with WT mice during CAIA25 (Extended Data Fig. 1a). To understand the underlying mechanism, we first determined the cell type (or types) responsible for this phenotype by analysing the cellular infiltrate in the hind paws of arthritic mice by flow cytometry. Day 7 was selected for analysis, representing the peak clinical score in WT mice undergoing CAIA (Extended Data Fig. 1a). We found a significant increase in neutrophils (defined as CD45+Ly6G+CD11b+ cells) in the joints of Micl−/ animals, whereas other myeloid populations were unaffected (Fig. 1a,b and Extended Data Fig. 1b). Analyses of an earlier time point (day 5) showed that neutrophil numbers were higher in the joints of MICL-deficient mice even during early stages of inflammation, when there was no apparent clinical difference between the two groups of mice, again without affecting other myeloid populations (Extended Data Fig. 1c). Of note, MICL deficiency also exacerbated joint inflammation in another effector phase model of rheumatoid arthritis: the K/BxN serum transfer model26 (Extended Data Fig. 2a). As in CAIA, Micl−/− mice in the K/BxN serum transfer model presented higher numbers of neutrophils than WT mice (Extended Data Fig. 2b). There were no alterations in the frequency of cellular populations in the bone marrow or blood of naive Micl−/− mice (Extended Data Fig. 2c,d).

Fig. 1: MICL is required for control of neutrophil responses.
figure 1

a, tSNE plots of CD45+ myeloid populations in the inflamed ankle joint during CAIA displayed as CD11b+ cells (grey), neutrophils (blue), Ly6Chigh cells (green), Ly6Clow cells (orange) and remaining antigen-presenting cells (APCs; pale violet). b, Myeloid cell populations (defined as shown in the gating strategy in Extended Data Fig. 1b) in the inflamed ankle joint during CAIA at day 7 are represented as a percentage of total live cells (pooled data from two independent experiments with four mice per group per experiment). WT versus Micl−/− neutrophils P < 0.0001. c, Schematic representation of the anti-Ly6G-mediated neutrophil depletion strategy in the CAIA model. D0, day 0; LPS, lipopolysaccharide. d, Quantification of neutrophils (CD45+CD11b+F4/80SSChigh) in the peripheral blood at day 9 by flow cytometry (n = 1 experiment with 3 mice per group). Isotype versus anti-Ly6G WT P = 0.0035 and Micl−/− P = 0.0021. e, Severity scoring of WT and knockout (Micl−/−) mice treated with isotype or anti-Ly6G antibodies (Abs), as indicated (n = 1 experiment with 5 mice per group). f, Schematic representation of neutrophil adoptive transfer during CAIA in WT mice. KO, knockout. g, CAIA severity scoring in WT mice that received adoptively transferred WT or knockout neutrophils, as indicated (pooled data from two independent experiments with five mice per group per experiment). Day 8 WT versus Micl−/− P = 0.0013. h, Representative Safranin O-stained sections of the tarsal joints of WT mice that received adoptively transferred WT or knockout neutrophils, as indicated at day 8 (left). Synovial inflammation (black asterisks) is indicated. Scale bars, 500 µm. The histological arthritis severity score is also shown (right; nine mice per group). WT versus Micl−/− P = 0.0281. Data are represented as mean ± s.d. (b,d,e,g,h). Statistical significance was determined by two-way analysis of variance (ANOVA) with Bonferroni’s multiple comparisons test (b,d,e,g). Data were analysed using an unpaired two-tailed Student’s t-test (h). *P < 0.05, **P < 0.01 and ****P < 0.0001. Schematics in panels c,f were created using BioRender (https://biorender.com).

Source Data

We next analysed the expression of key neutrophil adhesion and activation molecules (CD18, CD11b and CD62L) and chemotactic receptors (CXCR2 (ref. 27), CCR1 and C5aR28,29) by flow cytometry. Neutrophils in the joints of Micl−/− mice under CAIA exhibited increased expression of CD18 and CD11b, and decreased expression of the lymphoid organ homing receptor CD62L, compared with WT animals, indicative of a heightened state of activation (Extended Data Fig. 2e). This activated phenotype was only observed in cells isolated from the joints and not on neutrophils concomitantly isolated from the peripheral blood of mice undergoing CAIA (Extended Data Fig. 2e). Analysis of receptors controlling neutrophil migration revealed increased expression of C5aR and the chemokine receptor CXCR2 on neutrophils isolated from the joints of Micl−/− mice, compared with WT animals, although the increase of the former was not statistically significant (Extended Data Fig. 2f). By contrast, expression levels of CXCR2 and C5aR were comparable between neutrophils isolated from the peripheral blood of WT and Micl−/− mice (Extended Data Fig. 2f). The expression of CCR1 on neutrophils was unaffected by MICL deficiency (Extended Data Fig. 2f). MICL-deficient animals in the K/BxN serum transfer model of arthritis showed similar changes in neutrophil activation (Extended Data Fig. 2g). There were no differences in the expression of key adhesion and activation molecules and chemotactic receptors in the bone marrow or blood neutrophils of naive animals (Extended Data Fig. 2h,j). Together, these data demonstrate that MICL deficiency is characterized by an increased number and an enhanced activation of neutrophils in the joints of mice during CAIA and in the K/BxN serum transfer model of arthritis.

Neutrophils contribute significantly to the pathology of CAIA30. To investigate whether these cells were solely responsible for the exacerbated disease in Micl−/− mice, we successfully depleted circulating neutrophils via anti-Ly6G administration following the onset of CAIA (Fig. 1c,d and Extended Data Fig. 3a). Of note, clinical disease was significantly reduced following neutrophil depletion, and to an equivalent level, in both WT and Micl−/− mice (Fig. 2e). A similar reduction in disease severity occurred in both groups of mice following neutrophil depletion using the less-specific marker GR-1 (Extended Data Fig. 3b–e). These data show that the elevated pathology in the Micl−/− mice undergoing CAIA was stemming from the neutrophil compartment.

Fig. 2: NET formation drives CAIA severity in Micl−/− mice.
figure 2

a, ROS generation by MSU, PMA, zymosan or A. fumigatus hyphae-stimulated bone marrow neutrophils, depicted as relative light units (RLU), over time. Data are a representative example of n = 4 independent experiments and mean ± s.d. performed in triplicate. Area under curve was analysed using unpaired two-tailed Student’s t-test. WT versus Micl−/− MSU P < 0.0001, A. fumigatus P = 0.0056 and not significant (NS). b, Sytox green fluorescent images of MSU-induced NETs and NET formation percentage by thioglycollate-elicited neutrophils (3 fields of view per condition) 4 h post-stimulation with PMA or MSU, or medium alone (–). Data are represented as mean ± s.d. (n = 3 independent experiments performed in triplicate). Scale bars, 500 µm. Statistical significance was determined by two-way ANOVA with Bonferroni’s multiple comparisons test. WT versus Micl−/− MSU P = 0.0304. c, Representative confocal immunofluorescence microscopy of NETs in CAIA WT and Micl−/− day 11 synovial sections (as per Extended Data Fig. 1a; n = 1 experiment with 3 mice per group). GR-1 (purple), DAPI (blue), citrullinated histone 3 (cit-H3; yellow) and DNA/H1 (green) are shown. NETs are defined as GR-1+cit-H3+DNA/H1+-stained cells. Scale bars, 100 µm. d, Schematic of the PAD4 inhibitor CAIA treatment regime. e,f, Image stream quantification of NET-positive cells (e) and neutrophils isolated from arthritic ankle joints (f) at day 11 during PAD4 inhibitor (BB-CL-amidine) treatment. Pooled data are from 2 independent experiments with n = 7 biologically independent mice per group, represented as mean ± s.d. Statistical significance was determined using two-way ANOVA with Tukey’s multiple comparisons test. WT versus Micl−/− P = 0.0063 and Micl−/− versus Micl−/− + PAD4 inhibitor P = 0.0295 (e). g, Severity scoring of WT and Micl−/− mice during CAIA treated with vehicle or the PAD4 inhibitor (GSK484). The black arrow indicates the start of treatment. Data shown are a representative example of n = 2 experiments with 4 mice per group, represented as mean ± s.d. Statistical significance was determined by two-way ANOVA with Tukey’s multiple comparisons test. Days 8–10 Micl−/− versus Micl−/− + GSK484 P < 0.0001. *P < 0.05, **P < 0.01 and ****P < 0.0001. The diagram in panel d was created using BioRender (https://biorender.com).

Source Data

We then investigated whether the aberrant neutrophil response in the Micl−/− mice resulted from a cell-intrinsic or cell-extrinsic defect. For this, we isolated neutrophils from the bone marrow of naive WT or Micl−/− mice (Extended Data Fig. 3f) and performed a single adoptive transfer of these cells into arthritic WT mice at day 7 post-induction of CAIA (Fig. 1f). We detected transferred cells in the joints even 48 h post-transfer (Extended Data Fig. 3g). Although transfer of WT neutrophils did not significantly alter disease severity (Fig. 1g,h), transfer of MICL-deficient neutrophils into WT mice induced a significant increase in joint inflammation, determined by both clinical score and histological changes (Fig. 1g,h). Together, these data reveal that aberrant neutrophil function underlies the pathology of Micl−/− mice during CAIA.

MICL controls NET formation

We then determined how loss of MICL was affecting neutrophil function. A previous report has suggested that MICL negatively regulates the respiratory burst in neutrophils in response to specific ligands, such as uric acid crystals (monosodium urate (MSU))31. Indeed, we recapitulated this observation and showed that Micl−/− neutrophils induced a significantly elevated ROS in response to MSU, but not to phorbol-12-myristate-13-acetate (PMA) or zymosan, a potent inducer of ROS and a ligand for a related C-type lectin (Dectin-1)32, respectively (Fig. 2a). Micl−/− neutrophils also produced higher levels of ROS in response to A. fumigatus hyphae, which are also known to induce ROS33 (Fig. 2a). As ROS production is involved in induction of NETs34, we examined this programmed cell death response and found that neutrophils from Micl−/− mice exhibited significantly increased NET formation in vitro following MSU stimulation (Fig. 2b and Extended Data Fig. 4,b). NET formation was unaltered following PMA stimulation (Fig. 2b and Extended Data Fig. 4a). The ability of diphenyleneiodonium chloride (DPI) to inhibit NET formation in cells from MICL-deficient animals revealed the dependence of this response on NADPH oxidase (Extended Data Fig. 4c). MICL transduces intracellular inhibitory signals through SHP1/2 (refs. 35,36); we found that inhibition of SHP1/2 increased NET formation in WT neutrophils, but not in MICL-deficient neutrophils, indicating that inhibitory signalling from MICL was required to regulate this response (Extended Data Fig. 4d). Of note, we detected increased NET formation (defined as GR-1+cit-H3+DNA/H1+-stained cells) in the joints of Micl−/− mice undergoing CAIA at day 11 by immunofluorescence (Fig. 2c and quantification in Extended Data Fig. 4e) and by imaging flow cytometry (Fig. 2e and Extended Data Fig. 4f).

In addition to ROS, the formation of NETs can require histone citrullination through protein arginine deaminase 4 (PAD4)37. Using two different PAD4 inhibitors, BB-CL-amidine38 and GSK484 (ref. 39), complete inhibition of NET formation in vitro was shown (Extended Data Fig. 5a). In vivo (Fig. 2d), imaging flow cytometry revealed that treatment with these inhibitors led to a significant reduction in the number of NETs detectable in the joints of Micl−/− but not WT mice during CAIA (Fig. 2e). There was no effect of the inhibitor on the number of neutrophils recruited into the joints of these mice (Fig. 2f). Treatment of Micl−/− mice with both PAD4 inhibitors reduced the severity of the disease back to WT levels (Fig. 2g and Extended Data Fig. 5b). There was no significant effect of the inhibitors on disease development in WT mice (Fig. 2g and Extended Data Fig. 5b), as expected40. Specific PAD4 inhibition also reduced disease severity in MICL-deficient mice during the K/BxN serum transfer model of arthritis (Extended Data Fig. 5c). Administration of DNase I to degrade NETs in vivo ameliorated the enhanced clinical severity in the Micl−/− mice undergoing CAIA, but had no effect on disease severity in WT mice (Extended Data Fig. 5d). Thus, these data show that aberrant regulation of NET formation underlies the pathology of MICL-deficient mice during models of arthritis.

Anti-MICL antibodies link to NET diseases

We had previously shown that administration of antibodies targeting MICL was able to exacerbate CAIA in WT mice, recapitulating the phenotype of MICL-deficient animals25. Here we found that treatment of WT neutrophils with anti-MICL antibodies resulted in elevated ROS production in response to MSU, but not zymosan, suggesting that these antibodies were blocking receptor functionality (Extended Data Fig. 6a). To gain more insight into the effect of anti-MICL antibodies on neutrophil activation, we administered anti-MICL monoclonal antibodies to WT mice during CAIA in the presence of NET inhibitors. As previously observed, administration of anti-MICL antibodies exacerbated CAIA pathology in WT mice (Extended Data Fig. 6b). Administration of a PAD4 inhibitor completely reduced disease symptoms to the levels seen in the isotype control-treated mice (Extended Data Fig. 6b), revealing that aberrant NET formation was underlying the effect of the anti-MICL antibodies on disease pathology.

To further substantiate the receptor-blocking effect of anti-MICL antibodies, we tested them using an alternative model: collagen-induced arthritis (CIA). This is an important T cell-dependent model of arthritis that recapitulates both the pathophysiology and the histological presentation of human disease, although C57BL/6 mice, the same background on which our knockout model was based, are resistant to this model of disease41. However, the ability of anti-MICL antibodies to alter MICL function enabled us to explore the role of this receptor during CIA induced in DBA/1 mice. As before, administration of anti-MICL antibodies to DBA/1 mice exacerbated CIA disease and increased the number of neutrophils in the joints (Extended Data Fig. 6c). These results confirm that antibodies to MICL interfere with key functions of this receptor and its ability to regulate neutrophil activation, exacerbating arthritic inflammation in different mouse models.

As we previously detected the presence of anti-MICL antibodies in a small cohort of patients with rheumatoid arthritis25, we explored whether these antibodies could affect neutrophil activation. Here we found that anti-human MICL antibodies were able to enhance the respiratory burst and NET formation in human neutrophils in vitro in response to MSU, but not to PMA (Fig. 3a,b). Moreover, serum samples from patients with rheumatoid arthritis were able to increase ROS production in human neutrophils following stimulation with MSU (Fig. 3c and Extended Data Fig. 6d); this effect directly correlated with the anti-MICL serum titres, demonstrating that patient anti-MICL antibodies interfere with MICL function.

Fig. 3: Anti-MICL antibodies modulate neutrophil function and correlate with disease.
figure 3

a, ROS generation by MSU or PMA of human neutrophils (hPMNs) in the presence or absence of anti-MICL antibodies (anti-hMICL), depicted as RLU over time. Anti-hMICL versus isotype MSU P = 0.0084. b, Fluorescence of NET-bound Sytox green of MSU or PMA-induced NETs in hPMNS in the presence or absence of anti-hMICL. MSU anti-hMICL versus isotype P = 0.0043. RFU, relative fluorescence units. c, ROS generation by MSU of hPMNs in the presence of serum from patients with rheumatoid arthritis (RA) and healthy controls (HC). HC1 versus RA38 P < 0.0001, HC1 versus RA40 P < 0.0001, HC2 versus RA38 P < 0.0001 and HC2 versus RA40 P < 0.0001. d, Level of anti-MICL autoantibodies detected in serum samples from patients with rheumatoid arthritis (n = 199) and healthy controls (n = 132). Abs450, absorbance at 450 nm. e, Correlation of MICL autoantibody titres with rheumatoid factor levels in SERA cohort serum from patients with rheumatoid arthritis. f, Correlation of MICL autoantibody titres with CCP levels in SERA cohort serum from patients with rheumatoid arthritis. g, Level of anti-MICL autoantibodies detected in serum from patients with SLE (n = 40) and healthy controls (n = 25). h, Level of anti-MICL autoantibodies detected in serum from patients with mild/moderate (mCOVID-19; n = 25) and severe (sCOVID-19; n = 67) COVID-19 and healthy controls (n = 36). sCOVID-19 versus mCOVID-19 P = 0.0472 and sCOVID-19 versus healthy controls P = 0.0284. i, ROS generation and area under curve (AUC) of hPMNs stimulated with MSU in the presence of serum from patients with SLE pre-incubated with Fc–hMICL. Data are a representative example of n = 3 (a,b) or n = 2 (c,i) independent experiments, and mean ± s.d. performed in triplicate. The AUC was analysed by one-way ANOVA with Tukey’s multiple comparisons test. Box plots extend from the 25th to 75th percentile, including the median, and whiskers extend from the minimum to maximum value (d,g,h). Data were analysed by unpaired two-tailed Student’s t-test (d,g), a Spearman correlation and two-sided t-test (e,f), or using one-way ANOVA with Kruskal–Wallis multiple comparisons test (h). *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001.

Source Data

To determine whether anti-MICL antibodies were influencing clinical severity, we further analysed the prevalence of these antibodies in patients with rheumatoid arthritis, using an ELISA-based assay25. For this, we interrogated 200 serum samples from patients, along with matched controls, obtained from the Scottish Early Rheumatoid Arthritis (SERA) cohort42. We found that the majority of patients with rheumatoid arthritis presented with significantly elevated titres of anti-MICL autoantibodies (Fig. 3d). High levels of autoantibodies were also detected in some healthy controls without rheumatoid arthritis. Although we did not find an association between the level of anti-MICL autoantibodies and DAS28 or Sharp erosion scores (data not shown), after correcting for age and sex, we detected an association between the level of anti-MICL antibodies and the level of rheumatoid factor, an autoantibody targeting the Fc region of IgG antibodies, which is used to diagnose, classify and predict development of rheumatoid arthritis43,44 (Fig. 3e). We also found a significant correlation between the level of anti-MICL autoantibodies and the level of anti-cyclic citrullinated peptide (CCP) antibodies in patients with rheumatoid arthritis (Fig. 3f), indicative of a direct link of serum anti-MICL antibodies with NETosis and disease severity45,46. Together, these results show that autoantibodies targeting MICL interfere with the function of this receptor and correlate with severity of rheumatoid arthritis in patients.

We next determined whether anti-MICL autoantibodies were associated with any other NET-linked inflammatory diseases. Of note, we found high titres of anti-MICL antibodies in patients with SLE (Fig. 3g) and severe COVID-19 (Fig. 3h), both of which are inflammatory disorders where disease severity has been linked to NETosis15,47. Moreover, as we found in rheumatoid arthritis, serum samples containing anti-MICL antibodies from patients with SLE or severe COVID-19 were able to modulate neutrophil function, as demonstrated by increased ROS production following MSU stimulation (Extended Data Fig. 6e,f). We demonstrated the specificity of this response through the addition of a soluble chimeric protein containing the C-type lectin-like ___domain (CTLD) of human MICL, which abrogated the serum effect on the neutrophil response (Fig. 3i and Extended Data Fig. 6g). A related control CTLD had no effect on these responses (Extended Data Fig. 6g). Thus, autoantibodies that modulate MICL-mediated neutrophil functions are present in patients with a wide variety of NET-related pathologies.

MICL is a PRR for NETs

Given that NETs can activate pro-inflammatory functions of naive neutrophils48, we wondered whether antibody-mediated interference or loss of MICL function was affecting neutrophil responses to NETs themselves. We observed that following exposure to preformed NETs, MICL-deficient neutrophils induced significantly increased levels of ROS and higher formation of NETs compared with the response of WT neutrophils (Fig. 4a,b and Extended Data Fig. 7a,b). ROS production in MICL-deficient cells was unaltered when neutrophils were stimulated with NETs treated with polymyxin B (Extended Data Fig. 7c). Following stimulation of cells from both groups of mice with preformed NETs, the ability of both ROS (DPI) and NET (PAD4) inhibitors to prevent NET formation shows that this process is dependent on NADPH and PAD4 activity (Extended Data Fig. 7d). To show that human MICL was functioning similarly, we generated MICL-knockout human neutrophils derived from CD34+ haematopoietic progenitors49 (Extended Data Fig. 7e) and found increased ROS production in response to preformed NETs compared with control cells (Extended Data Fig. 7f). Of note, we also demonstrated that normal neutrophils isolated from healthy human controls had significantly elevated ROS responses to human NETs in the presence of anti-human MICL antibodies (Fig. 4c).

Fig. 4: NET–MICL interaction regulates inflammation.
figure 4

a, ROS generation by bone marrow neutrophils stimulated with preformed NETs, depicted as RLU over time. Data are a representative example of n = 4 independent experiments, depicted as mean ± s.d. performed in triplicate. mNET, preformed mouse NETs. b, Immunofluorescence staining for MPO (yellow), cit-H3 (magenta) and DNA (DAPI; grey) of neutrophils stimulated with preformed NETs. Scale bars, 200 µm. Quantification is in Extended Data Fig. 7b. c, ROS generation by human neutrophils stimulated with preformed NETs in the presence or absence of antibodies targeting MICL. Data are a representative example of n = 2 independent experiments, depicted as mean ± s.d. performed in duplicate. hPMNs versus anti-hMICL P = 0.0011 and anti-hMICL versus isotype P = 0.0016. d, Fc–MICL recognition of untreated NETs (NETs), proteinase K-treated (+prot K) or DNase-treated (+DNase I) NETs by ELISA. e, MICL-expressing BWZ reporter cell recognition of untreated NETs, proteinase K-treated or DNase-treated NETs. OD, optical density. Pooled data from two independent experiments, depicted as mean ± s.d. performed in triplicate (d,e). f, Neutrophil infiltration (CD45+CD11b+Ly6G+ cells) 4 h after preformed NETs or LPS injection in the peritoneum of WT and Micl−/− mice. Pooled data are from two independent experiments (n = 9 mice per group), depicted as mean ± s.d. NETs WT versus Micl−/− P = 0.0384. *P < 0.05, **P  < 0.01 and ****P < 0.0001.

Source Data

To determine whether MICL was functioning as a PRR for NETs themselves, we examined the ability of a soluble chimeric protein consisting of the CTLD of mouse MICL fused to the Fc region of human IgG1 (Fc–mMICL) to directly recognize NETs. Using this Fc–protein and a structurally related control protein, we showed that MICL binds directly to NETs using an ELISA-based assay (Fig. 4d). Moreover, recognition of NETs by MICL could also be demonstrated in a cellular context using MICL-expressing reporter cells50 (Fig. 4e). MICL recognition of NETs was blocked in the presence of antibodies targeting MICL (Extended Data Fig. 7g), showing that antibodies to MICL block the ability of the receptor to recognize its ligand. To determine which component of NETs is required to interact with this receptor, we treated NETs with DNase I or proteinase K, and found that treatment with DNase abolished the ability of MICL to recognize NETs (Fig. 4d,e). Furthermore, MICL-deficient neutrophils stimulated with DNase-treated NETs induced significantly lower levels of ROS (Extended Data Fig. 7h). Consistent with these observations, MICL was able to recognize genomic DNA, and this interaction could be blocked in the presence of anti-MICL antibodies (Extended Data Fig. 7i).

To show a direct role for MICL in response to NETs in vivo, we injected preformed NETs into the peritoneum of Micl−/− and WT mice. Neutrophil recruitment was significantly higher in Micl−/− mice treated with NETs, but similar to WT in mice administered with LPS (Fig. 4f), revealing a specific negative regulatory function of MICL following recognition of NETs. Moreover, there was a trend to higher levels of cell-free DNA measurements following NET administration in MICL-deficient mice (Extended Data Fig. 7j). Collectively, our data show that MICL functions as a PRR for NETs, regulating activation and NET formation in both mouse and human neutrophils.

MICL regulates fungal-induced NETs

To validate our hypothesis that NET sensing by MICL functions as a universal regulator of neutrophil activation and NET formation, we examined neutrophil responses elicited by A. fumigatus. NET formation is critical in the immune response to A. fumigatus, restraining fungal growth and preventing tissue dissemination51,52. We demonstrated that NETs are released in response to A. fumigatus hyphae53,54 (Fig. 5a), and both MICL-deficient neutrophils and human neutrophils treated with anti-MICL antibodies induce higher levels of NET formation in vitro in response to this pathogen (Fig. 5a and Extended Data Fig. 8a). Of note, MICL-deficient mice were significantly more protected than WT mice following intravenous infection with A. fumigatus conidia (Fig. 5b). This increased susceptibility of WT mice was associated with increased fungal burdens in the brains of these animals at day 2 post-infection (Fig. 5c and Extended Data Fig. 8b). Analysis of inflammatory responses showed that serum levels of IL-6 and brain levels of G-CSF were significantly higher in infected WT than in MICL-deficient mice (Fig. 5d,e), showing that MICL functions to restrain systemic inflammation during invasive aspergillosis. Furthermore, we confirmed that the increased survival of MICL-deficient mice was associated with higher NET formation, as the treatment of these animals with a PAD4 inhibitor (GSK484), but not with the vehicle, increased the susceptibility of the knockout mice to A. fumigatus infection to levels similar to those observed in WT mice (Fig. 5f and vehicle control shown in Extended Data Fig. 8c). Moreover, administration of the inhibitor increased fungal burdens in the brains of MICL-deficient mice (Fig. 5g). There was no effect of the NET inhibitor or vehicle control on disease development and fungal burdens in WT mice53 (Fig. 5g and Extended Data Fig. 8c). Thus, dysregulation of NET formation in MICL-deficient mice results in increased resistance to a systemic fungal infection.

Fig. 5: MICL regulates NET formation during fungal infection.
figure 5

a, Fluorescence of A. fumigatus hyphae-induced NETs bound to Sytox green in bone marrow-isolated neutrophils from WT and MICL-deficient mice or human neutrophils in the presence of antibodies to MICL. The fluorescence background signal from the unstimulated controls was subtracted from values. Data are a representative example of n = 3 independent experiments, depicted as mean ± s.d. performed in triplicate. The AUC was analysed using an unpaired two-tailed Student’s t-test. WT versus Micl−/− P = 0.0272 and anti-hMICL versus isotype P = 0.0032. b, Survival of mice following intravenous infection with 106 A. fumigatus conidia (n = 15 mice per group). Pooled data are from two independent experiments, analysed by log-rank test; P = 0.0005. ce, Brain fungal burdens (c), and serum (d) and brain (e) cytokine levels of mice 2 days after intravenous infection with 106 A. fumigatus conidia. Values are mean ± s.e.m. of pooled data from two independent experiments. n = 12 (c) or n = 8 (d,e) biologically independent mice, analysed using an unpaired two-tailed Student’s t-test. IL-6 WT versus Micl−/− P = 0.0185 (d) and G-CSF WT versus Micl−/− P = 0.0492 (e). CFU, colony-forming unit. f, Survival of mice following intravenous infection with 106 A. fumigatus conidia and treated with GSK484 (PAD4 inhibitor). Pooled data are from two independent experiments; n = 14 biologically independent mice, analysed by log-rank test. g, Brain fungal burdens of mice 2 days after intravenous infection with 106 A. fumigatus conidia treated with GSK484 (PAD4 inhibitor) or vehicle control. Values are mean ± s.e.m. of 1 experiment with n = 6 biologically independent mice, analysed by one-way ANOVA with Tukey’s multiple comparisons test. WT vehicle versus Micl−/− vehicle P = 0.0378, WT + GSK4848 versus Micl−/− vehicle P = 0.0356 and Micl−/− vehicle versus Micl−/− + GSK484 P = 0.0242. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001.

Source Data

Discussion

The integration of signalling between diverse receptors helps to refine and modulate immune responses55. Inhibitory receptors regulate immune cell activation, avoiding excessive inflammation and host tissue damage. Our data revealed that MICL, an inhibitory C-type lectin receptor, is vital for the regulation of key neutrophil functions. We discovered that MICL is a PRR for NETs, and that this interaction is essential for controlling key neutrophil responses, including the respiratory burst and further induction of NETs themselves. This regulation is critical for limiting NET-mediated inflammatory conditions, such as in arthritis, SLE or severe COVID-19 infections, in which MICL prevents the formation of a positive-feedback loop that leads to uncontrolled inflammation (Extended Data Fig. 9). By contrast, this regulation of NET formation supresses robust control of invasive infections, promoting disease susceptibility.

In the context of arthritis, we showed that loss of MICL functionality increases NET formation and neutrophil activation by NETs, triggering an inflammatory chain reaction that leads to increased joint inflammation in three mouse models of arthritis. We also found that anti-MICL antibodies block the receptor function, and that the presence of these antibodies influences the severity of disease in mouse models and in patients with rheumatoid arthritis. In fact, we demonstrated that serum samples from patients with rheumatoid arthritis possessing anti-MICL autoantibodies, and antibodies to mouse or human MICL, are able to recapitulate MICL-deficient neutrophils, in terms of dysregulation of ROS production and NET release. NETs are released locally in the inflamed joints of patients with rheumatoid arthritis, and the citrullinated neo-epitopes generated during this process promote the formation of anti-CCP antibody that contribute to the perpetuation of the disease6,56. Of note, we found that the level of autoantibodies to MICL correlate with the level of anti-CCP antibodies in patients, indicating a direct link between MICL function and NET formation in patients with rheumatoid arthritis. Moreover, we showed that autoantibodies capable of blocking MICL function are also present in other autoinflammatory conditions, including SLE and severe COVID-19, in which NETs are linked to disease pathology15,47. Thus, MICL represents a universal novel autoregulatory pathway that prevents aberrant neutrophil activation and subsequent tissue damage in autoimmunity.

Neutrophils also have a crucial role in protection against fungal pathogens, such as A. fumigatus57. Protective neutrophil responses to A. fumigatus hyphae involve ROS production and NET formation51,52,54. Although PAD4 is required for NET formation in response to A. fumigatus, previous studies using Pad4-knockout mice have suggested no role for this enzyme during infection53. Indeed, we found that PAD4 inhibition does not affect disease severity in WT mice. Here we discovered that the PAD4 pathway is stringently regulated by MICL, but that mice lacking this receptor use this pathway to restrict fungal infection through increased NET formation. Pharmacological inhibition of NET formation using a PAD4 inhibitor blocked this pathway in MICL-deficient mice, reverting their susceptibility to infection back to WT levels. Our data show that similar MICL-mediated control of the PAD4 pathway also occurs during autoimmune disease. Indeed, the PAD4 pathway has previously been shown not to be required for effector phase responses during arthritis58, as we found in our WT mice (Fig. 2g and Extended Data Fig. 5b). Of note, our data show that in the absence of MICL (or inhibition of receptor function by anti-MICL antibodies), the PAD4 pathway becomes activated during autoimmune disease, leading to dysregulated NET formation and aberrant inflammation.

In conclusion, we defined a key mechanism by which MICL regulates neutrophil activation and NET formation. MICL directly recognizes NETs, inducing intracellular signalling that dampens neutrophil responses and preventing the formation of a positive-feedback loop leading to further NET formation. In patients with rheumatoid arthritis, SLE and severe COVID-19, this feedback loop becomes dysregulated by autoantibodies to MICL, leading to a worsened disease outcome. It is likely that autoantibodies to MICL are also associated with the severity of other NET-mediated autoimmune disorders, such as anti-neutrophil cytoplasmic antibody-associated vasculitis. Given that we also observed elevated levels of anti-MICL antibodies in a proportion of healthy individuals, future work should explore the possibility that the presence of such autoantibodies are associated with, or predispose towards, the development of autoimmune disease. Our data also reveal that NET sensing by MICL can have different clinical outcomes depending on the context. In this sense, our results suggest therapeutic targeting of MICL with antibodies to increase ROS and NET formation, contributing to fungal clearance, could be used to treat disseminated forms of invasive infection. Conversely, blocking of the antibody–receptor interaction could be of therapeutic benefit for NET-mediated inflammatory disease. In summary, our observations revealed a novel and fundamental function of inhibitory pathways underlying infectious and non-infectious disease pathogenesis.

Methods

Mice

Micl−/− and C57BL/6J mice were bred and maintained under specific pathogen-free (SPF) conditions at the University of Aberdeen, University of Exeter and Charles River Laboratories. Mice were housed in separate groups with bedding exchanges between cages every 2 days for 1 week before commencement of experiments, and maintained on a 12 h–12 h dark–light cycle (07:00–19:00) at 20–24 °C and relative humidity of 55 ± 15%. Mouse experiments were performed by random assignation of age-matched (6–8 weeks old) and sex-matched mice in experimental or control groups at the beginning of each experiment; females were co-housed and experiments were not blinded. All experiments conformed to the ethical review committee of the University of Aberdeen, University of Exeter and the UK Home Office regulations (project license numbers: P79B6F297 and P6A6F95B5).

CAIA

Male mice received intraperitoneal (i.p.) injections of 2 mg of ArthritoMab monoclonal antibody cocktail (MD Bioproducts) on day 0, followed by 50 μg of lipopolysaccharide (LPS) i.p. (MD Biosciences) on day 3. Joint inflammation was scored visually using a scale of 0 (no visible signs of redness or swelling) to 4 (extensive swelling with signs of deformity). To achieve neutrophil depletion during CAIA, 500 μg of rat anti-mouse Ly6G (clone 1A8, Bio-X-Cell) or isotype control (rat anti-mouse IgG2a; Bio-X-Cell) was administered i.p. to mice every 48 h from day 5 onwards. Alternatively, 50 μg of rat anti-mouse GR-1 (clone RB6-8C5) or isotype control (rat anti-mouse IgG2b) was administered i.p. every 48 h from day 5 onwards. Mice were culled on day 5, 7, 10, 11 or 13, as indicated in the text. For adoptive transfer experiments, bone marrow neutrophils were isolated as detailed below and stained with cell proliferation dye eFluor 670 (eBioscience). Labelled cells (5 × 106) were transferred intravenously (i.v.) to WT mice on day 7 following induction of CAIA. Mice were culled on day 9. To investigate the role of NET formation, mice undergoing CAIA were injected i.p. with 2 mg kg−1 daily of BB-CL-amidine (Cayman Chemical), 4 mg kg−1 daily of GSK484 (Cambridge Biosciences), 75 U per animal of DNase I (Merck) or vehicle (5% DMSO in 10% cyclodextran for BB-Cl-amidine or ethanol 99.9% diluted 1:50 in 0.9% NaCl for GSK484) from day 7 to day 10. Mice were culled on day 11.

K/BxN serum transfer model of arthritis

Male mice were administered i.p. with 100 μl of serum from transgenic K/BxN mice26. The development of clinical symptoms were monitored daily and mice were culled on day 10. Joint inflammation was scored visually using a scale of 0 (no visible signs of redness or swelling) to 4 (extensive swelling with signs of deformity). To investigate the role of NET formation, mice were injected i.p. with 4 mg kg−1 daily of GSK484 or vehicle (ethanol 99.9% diluted 1:50 in 0.9 % NaCl) from day 7 to day 10. Mice were culled on day 11.

CIA

Male DBA/1 mice were purchased from Inotiv and maintained at the University of Exeter. Mice were treated subcutaneously with 100 μl of Immunization Grade Chick Type II Collagen (Chondrex; 200 µg per mouse) in Complete Freund’s Adjuvant (Sigma-Aldrich) at day 0 and with 100 μl of Immunization Grade Chick Type II Collagen (200 µg per mouse) in Incomplete Freund’s Adjuvant (Sigma-Aldrich) at day 14, followed by an i.p. injection of 50 µg of LPS at day 26. Of anti-MICL antibodies, 0.7 mg of isotype control antibodies or PBS were administered i.p. every 48 h from day 17 to day 33. Mice were culled at day 35. In our animal facility, naive DBA/1 animals did not develop any spontaneous form of arthritis by 13 weeks of age (the latest timepoint in our experiments).

Histology

For histology, paws were fixed in 4% paraformaldehyde (PFA) overnight at 4 °C and decalcified in 10% EDTA for 3–4 weeks. Decalcified paws were embedded in optimal cutting temperature (OCT) and cryosectioned or embedded in paraffin wax and sectioned before staining with haematoxylin and eosin, or Safranin O, haematoxylin and fast green. Scoring of histological sections was performed blinded using an arthritis severity score as previously described18.

Immunofluorescence staining protocol

Frozen tissue sections were thawed for 30 min at room temperature and washed with PBS before permeabilization with 0.25% Triton X-100 or 0.5% saponin in PBS for 10 min. After permeabilization, sections were washed with PBS again and blocked with 3% BSA at room temperature for 30 min. Tissue sections were stained with primary antibodies (anti-cit-H3 (Abcam), anti-DNA/H1 (Merck) and anti-GR-1 (produced in-house59)) diluted in blocking buffer (3% BSA in PBS) for 1 h 30 min in a humidified chamber, after which they were washed with PBS-Tween-20 (0.05%) and stained with secondary antibodies for 1 h at room temperature in the dark. Sections were washed as before and counterstained with 1 μg ml−1 DAPI and mounted in Vectashield antifade mounting medium for fluorescence (Vectorlabs). Coverslips were sealed with nail polish and slides were stored in the dark at 4 °C until imaging. Fluorescence was visualised using the Zeiss confocal LSM700 microscope and Zen Black software (Zeiss). The quantification of positive area (area+ %) for each channel was conducted using Fiji software. Image preprocessing included utilizing the built-in ‘Moments’ algorithm for thresholding each channel.

Neutrophil isolation

Bone marrow neutrophils were isolated using Histopaque (Merck) by a density gradient centrifugation method or using the EasySep Mouse Neutrophil Enrichment Kit (StemCell Technologies) according to the manufacturer’s guidelines. Flow cytometry analysis confirmed an approximately 90% pure neutrophil population (the remaining population consisting primarily of monocytes). To isolate thioglycollate-elicited neutrophils, mice were injected i.p. with 1 ml of 3% thioglycollate broth. After 4 h, mice were culled and neutrophils were harvested by peritoneal lavage with PBS 5 mM EDTA.

Human neutrophils from the blood of healthy donors were purified using a Ficoll-Paque density centrifugation method60 or using the EasySep direct human neutrophil isolation kit (StemCell Technologies) as per the manufacturer’s instructions. Samples were obtained from consenting healthy donors with the approval of the Faculty of Health and Life Sciences ethics review board, University of Exeter (eCLESBio000371) and the College of Life Sciences and Medicine ethics review board, University of Aberdeen (CERB number 1243).

CD34+ cell isolation and CRISPR-mediated knockout

Peripheral blood mononuclear cells (PBMCs) were isolated from apheresis blood waste (NHSBT) with ethical approval from NHS Research Ethics Committee (REC 18/EE/0265) by density centrifugation using Histopaque-1077 (Sigma-Aldrich) according to the manufacturer’s instructions. After washing, cells were resuspended in red cell lysis buffer (55 mM NH4Cl, 0.137 mM EDTA and 1 mM KHCO3, pH 7.5) and incubated on ice for 10 min. Next, to enrich haematopoietic stem cells, CD34+ cells were isolated from PMBCs using a human CD34 Microbead Kit (Miltenyi Biotec) according to the manufacturer’s protocol. Isolated cells were cultured in Iscove’s modified Dulbecco’s medium (IMDM) supplemented with 10% (v/v) FBS and 1% (v/v) penicillin–streptomycin at 37 °C in 5% CO2. Cytokines were added at the indicated concentrations and days of culture: stem cell factor (SCF; 50 ng ml−1; day 0–5 of culture), Flt-3 ligand (50 ng ml−1; day 0–5 of culture), interleukin-3 (IL-3; 10 ng ml−1; day 0–5 of culture), granulocyte–macrophage colony-stimulating factor (GM-CSF; 10 ng ml−1; day 3–7 of culture) and granulocyte colony-stimulating factor (G-CSF; 10 ng m−1; day 7–14 of culture). All functional assays were completed between day 17 and day 18 of culture49.

CRISPR-mediated knockout was completed using the AmaxaTM 4D-Nucleofector (Lonza) using a P3 Primary Cell 4D-NucleofectorTM X Kit S (Lonza) and TrueCut Cas9 Protein v2 (Invitrogen) according to the manufacturers’ instructions. In brief, day 3 cultured neutrophils were resuspended in the mixture of P3 Primary Cell Solution and supplement 1 containing 50 pmol Cas9 and 125 pmol of guide RNA (62.5 pmol of two guides with the same gene target). Cells were transferred into nucleocuvette strip and electroporated using the EO-100 program. After electroporation cells were transferred into six-well plates containing StemSpan medium (StemCell Technologies) supplemented with FBS and penicillin–streptomycin and containing SCF, Flt-3 and IL-3. From day 5 onwards, cells were cultured as outlined above in IMDM. All functional assays were completed between day 17 and day 18 of culture. Guide RNAs were designed using Knockout Guide Design (Synthego). The following single guide RNAs were used (Synthego, modified single guide RNA with EZ scaffold): CLEC12A + 9979452: UGAAUAUCUCCAACAAGAUC and CLEC12A-9979406: GUUGUAGAGAAAUAUUUCUC; negative control scrambled #1: GCACUACCAGAGCUAACUCA and negative control scrambled #2: GUACGUCGGUAUAACUCCUC. Cells were stained with anti-CD66-PE/Dazzle (clone G10F5; diluted 1:50), anti-CD15-AF700 (clone Hi98; diluted 1:50), anti-CD16-APC (clone 3G8; diluted 1:50) and anti-hMICL or isotype control antibodies at 10 µg ml−1 to confirm loss of MICL expression.

ROS

Bone marrow neutrophils and human neutrophils were isolated as described above and resuspended in OptiMEM (Thermo Fisher Scientific) supplemented with 5% FCS. Cells (5 × 105) were added to each well of a white 96-well flat-bottomed plate and stimulated in triplicate with 200 μg ml−1 MSU crystals (Invivogen), 25 μg ml−1 Zymosan (Molecular Probes), 100 nM PMA (Sigma), isolated NETs (100 μg ml−1 based on NET–DNA concentration) or A. fumigatus hyphae (1 × 104 conidia per well incubated for 12 h at 37 °C) in the presence of 100 μM luminol (Sigma). Chemiluminescence was measured every 3 min for 2 h in a FLUOStar Optima microplate reader (BMG Labtech) or a Spark Cyto (Tecan) at 37 °C with 5 % CO2.

NETs

Bone marrow and thioglycollate-elicited neutrophils were isolated as described above. Cells were resuspended in RPMI (without phenol red; Thermo Fisher Scientific) supplemented with 2% DNase−/− mouse serum and seeded in an eight-well iBidi μ-slide (iBidi). Cells were stimulated with 100 μg ml−1 of MSU crystals (Invivogen), 100 nM PMA (Sigma), isolated NETs (100 μg ml−1 based on NET–DNA concentration) or A. fumigatus hyphae and incubated at 37 °C with 5% CO2 for 4 h. In some experiments, as indicated in the text, DPI (10 μM; Sigma) or NSC-87877 (5 μM; Cayman Chemical) were included in the assays. Extracellular DNA was visualized with 5 μM Sytox Green (Invitrogen). In experiments without fixation, cells were previously stained with Draq5 (Thermo Fisher Scientific) and cell-impermeable Sytox Green, and images were acquired on an inverted Zeiss AxioObserver Z1 using a PlanApo ×20/0.75 NA dry lens (Carl Zeiss) and a Hamamatsu Fusion sCMOS camera with an attached incubation chamber (PeCon GmbH) at 37 °C. Fluorescent images were analysed in Image J, and NETs defined as Sytox Green-positive cells showing extrusions were counted.

NET (%) = (total Sytox Green-positive cells extruding NETs/total cells counted) × 100

For kinetic curves of NET formation, neutrophils were seeded in a 96-well black plate with a transparent bottom and the cells were left to adhere for 30 min in a cell culture incubator. Cells were stimulated with the defined reagents and 5 μM Sytox Green. Fluorescence signal (504-nm excitation and 523-nm emission) was measured every 10 min for 7 h in a Spark Cyto (Tecan) at 37 °C with 5% CO2.

NETs were isolated as previously described61. In brief, bone marrow-isolated or purified human neutrophils were plated in a six-well plate at a density of 1 × 106 cells per well in RPMI without phenol red. Following stimulation with 100 nM PMA for 4 h at 37 °C, the culture medium was removed and NETs were partially digested by application of a restriction enzyme mix combining the enzymes BseRI, PacI, NdeI and AfIII (New England Biolabs) at a concentration of 5 U ml−1 in NEB buffer for 1 h at 37 °C. Supernatants were collected and centrifuged at 300g for 10 min at 4 °C. NET supernatants were transferred to a fresh tube and stored at −80 °C until used. When indicated, NETs were treated with DNase I (Thermo Fisher Scientific) or proteinase K (Roche) for 1 h at 37 °C.

Recognition of isolated NETs by MICL was assessed by ELISA. A 96-well plate (Nunc Maxisorp) was coated with NETs diluted in PBS overnight at 4 °C. Wells were washed and blocked with blocking buffer (5% BSA) at room temperature. Fc–mMICL or Fc–mCLEC12B at 1 μg ml−1 in blocking buffer was added and incubated for 2 h at room temperature. Bound Fc-fusion proteins were detected with horseradish peroxidase-conjugated goat anti-human IgG (Jackson Immunoresearch) diluted 1:10,000 for 1 h. TMB substrate was added, and absorbance was measured using a plate reader (Tecan). Purification of Fc–MICL and Fc–mCLEC12B, from transduced HEK293T cells (originally purchased from the American Type Culture Collection (ATCC), but not tested for Mycoplasma contamination for the experiments detailed in this paper), was performed as previously described32.

To analyse the recognition of isolated NETs by MICL using a MICL-expressing cell line, a 96-well plate was coated with NETs diluted in PBS overnight at 4 °C. Wells were washed, and BWZ.36 NFAT-LacZ cells (provided by W. Yokoyama; not tested for Mycoplasma contamination for the experiments detailed in this paper) expressing the CD3ζ chain fused to the transmembrane and carbohydrate recognition ___domain (CRD) of mMICL or mCLEC12B were added (2 × 105 cells per well) for 18 h at 37 °C. After stimulation, cells were centrifuged at 800g for 2 min and washed two times with PBS. Of CPRG substrate buffer, 100 µl was added per well. The reaction was stopped 4 h later with the addition of 50 µl of glycine-EDTA buffer, and the absorbance was measured using a plate reader62 (Tecan). Antibody crosslinking with the appropriate receptor antibody was used to confirm the functionality of the chimeric receptor constructs.

To analyse neutrophil infiltration by isolated NETs, 6–8-week-old Micl−/− and C57BL/6J mice were injected i.p. with 300 μg NETs or 50 μg LPS. Four hours later, mice were euthanized, and their peritoneal cells were counted and analysed by flow cytometry. Cell-free DNA was evaluated using the Quant-iT PicoGreen dsDNA Assay (Invitrogen) following the manufacturer’s instructions.

Immunofluorescent staining of in vitro samples

Neutrophils (5 × 104) were plated in an eight-well iBidi μ-slide (iBidi) 1 h before stimulation with isolated NETs, PMA or MSU. Four hours after stimulation, cells were fixed for 20 min at 37 °C with 2% PFA and permeabilized with 0.5% Triton X-100 in PBS. Samples were blocked with 3% (v/v) normal goat serum, 1% (w/v) BSA in PBS and incubated with anti-cit-H3 (Abcam) and anti-myeloperoxidase (R&D) antibodies. The secondary antibodies donkey anti-rabbit Cy5 and rabbit anti-goat Alexa Fluor 488 were used. Finally, the samples were stained with DAPI.

Fluorescence quantification was conducted by segmenting the DAPI channel using QuPath software (version 0.4.3) with the Cellpose extension (https://github.com/BIOP/qupath-extension-cellpose). Before segmentation, preprocessing of the DAPI channel was carried out using Fiji software to mitigate background noise, involving background subtraction and enhanced contrast built-in functions.

Preprocessing of the GFP and Texas Red channels involved a background subtraction of the mean grey value and two times the standard deviation of an empty region (background noise). The quantification of each channel was conducted using the measurement function of Fiji software, utilizing the mask generated by Qupath.

Flow cytometry and monoclonal antibodies

Cells were isolated from the hind paw ankle joint of arthritic mice using the protocol previously described20. Peripheral blood was collected by cardiac puncture or tail nicking in the presence of EDTA and red blood cell lysis performed using PharmLyse (BD Biosciences). Single-cell suspensions were stained with fixable viability dye eFluor 780 (eBioscience) and further stained with conjugated antibodies for same-day acquisition or fixed in 2% PFA. Conjugated antibodies used in these experiments included: anti-CD45-FITC (clone 102), anti-CD45-PerCP-cyanine5.5 (clone 102), anti-CD11b-BUV395 (clone M1/70), anti-CD11b-PE-Cy7 (clone M1/70), anti-GR-1-APC (clone RB6-8C5), anti-MHC-II-FITC (clone 2G9), anti-MHC-II-BUV496 (clone 2G9), anti-C5aR-BV510 (clone 20/70), anti-Ly6G-BV421 (clone 1A8), anti-Ly6G-Spark Blue 550 (clone 1A8), anti-Ly6G-APC (clone 1A8), anti-CD62L-BV510 (clone MEL-14), anti-CD18-BV650 (clone C71/16), anti-CD18-APC (H155-78), anti-F4/80-AF700 (clone BM8), anti-F4/80-PE-Cy7 (clone BM8), anti-Ly6C-PE-Cy7 (clone HK1.4), anti-Ly6C-Brilliant Violet 570 (clone HK1.4), anti-CCR1-PE (clone 643854), anti-CXCR2-APC (clone SA045E1), anti-CD11c-BV711 (clone HL3), anti-B220-BV605 (clone RA3-6B2) and anti-CD3-Alexa Fluor 647 (clone 17A2). All were purchased commercially from eBioscience, R&D systems or BioLegend and used diluted 1:600. Anti-mCLEC12A-biotinylated21, anti-hCLEC12A63 and isotype control AFRC MAC 49 (ECACC 85060404; isotype for anti-MICL) were generated in-house and used at 10 μg ml−1. Anti-mCLEC12A-biotinylated21 and anti-hCLEC12A63 antibodies were validated using mouse and human CLEC12A-transduced NIH3T3 fibroblast (originally purchased from the ATCC; not tested for Mycoplasma contamination for the experiments detailed in this paper). Cells were acquired using a BD LSR II Fortessa flow cytometer (BD Biosciences), BD LSRFortessa X-20 flow cytometer (BD Biosciences) or Cytek Aurora Spectral cytometer (Cytek). See Extended Data Fig. 1b for the cellular gating strategy. Data were collected using BD FACSDiva v8.0.3 (BD Biosciences) or SpectroFlo software v3.2.1 (Cytek) and analysed using FlowJo v10 software (BD Biosciences).

Imaging flow cytometry

Single cells isolated from the arthritic joint, as detailed above, were fixed in 1% PFA for 30 min at 37 °C. Cells were washed with PBS 2 mM EDTA for 5 min at 500g and stained with conjugated antibodies (detailed above), anti-DNA/histone 1 (Merck Millipore; 1.4 μg ml−1) and anti-cit-H3 (Abcam; diluted 1:300) for 1 h at room temperature. The secondary antibodies goat anti-rabbit APC (Molecular probes) and goat anti-mouse AF488 (Invitrogen) at 1 μg ml−1 were added for 1 h at 4 °C. Cells were washed and resuspended in 50 μl of PBS 2 mM EDTA before acquisition using the Amnis ImageStreamX MKII (Luminex) INSPIRE acquisition software. Files were analysed using IDEAS software v6.2 (Luminex).

Single cells were selected by plotting the area feature of brightfield channel 1 (BF1) versus the aspect ratio parameter of the same brightfield channel, which is the ratio of minor axis to the major axis of the applied mask, and describes the shape of the mask applied to the cells (Extended Data Fig. 7f). Focused cells were then selected by plotting the ‘Gradient RMS’ feature of BF1 against the BF1 contrast parameter. Cells with high-gradient and high-contrast value were more in focus and chosen for further analysis. All focused cells were used in the analysis for the presence of the cellular fluorescence parameters. Fluorescence parameters were measured in channel 2 (DNA/histone H1-AF488), channel 11 (cit-H3-APC) and channel 7 (Ly6G-BV421) with magnification set to ×40.

Serum collection from patients with rheumatoid arthritis

Serum samples from patients with rheumatoid arthritis were obtained from the SERA cohort42 (Extended Data Table 1). Sera from healthy controls were obtained by collection of whole blood in a BD vacutainer serum collection tube (BD Biosciences). The blood was allowed to clot for 15–30 min at room temperature, after which the tubes were centrifuged at 2,000g for 10 min in a 4 °C. Sera were removed, aliquoted and stored at −80 °C. Serum samples were obtained from consenting healthy donors with the approval of the College of Life Sciences and Medicine ethics review board, University of Aberdeen (CERB number 1243).

Serum collection from patients with COVID-19

Blood from 121 patients diagnosed with SARS-CoV-2 infection at Weill-Cornell Medicine between March and July 2020. Research on patients with COVID-19 was reviewed and approved by the Institutional Review Board of Weill-Cornell Medicine (New York Presbyterian and Lower Manhattan hospitals; #IRB 20-03021645 and #IRB 20-03021671). Informed consents were obtained from all enrolled patients and health-care workers by trained staff, and records were maintained in our research database for the duration of the study. All patients were classified as mild/moderate (n = 25) and severe (n = 66) disease according to oxygen requirements with mild/moderate disease defined as SARS-CoV-2 infection and less than 6 l of non-invasive supplementary oxygen to maintain SpO2 > 92%, and severe disease was defined as SARS-CoV-2 infection requiring hospitalization and received 6 l or more supplementary oxygen or mechanical ventilation. For controls, we used blood samples from 36 SARS-CoV-2-negative individuals collected by the JRI IBD Live Cell Bank Consortium at Weill-Cornell Medicine. Heparinized plasma and serum samples were aliquoted, heat-inactivated at 56 °C for 1 h and then stored at −80 °C.

Serum collection from patients with SLE

All patients with SLE in the study met the revised American College of Rheumatology criteria64 and the SLICC criteria65. Some patients had a history of biopsy-proven nephritis according to the International Society of Nephrology/Renal Pathology Society classification. Healthy female volunteers (with no family history of autoimmune disease) served as age-matched and ethnicity-matched controls. All patients provided informed consent, and samples used in this research project were obtained from the Imperial College Healthcare Tissue Bank (ICHTB). The ICHTB is supported by the National Institute for Health Research (NIHR) Biomedical Research Centre based at Imperial College Healthcare NHS Trust and Imperial College London. The ICHTB is approved by Wales REC3 to release human material for research (22/WA/0214), and the samples for this project (ref: R13010a) were issued from sub-collection reference number IMM_MB_13_001.

ELISAs from human patients

Sera from patients with rheumatoid arthritis, COVID-19 and SLE and healthy controls were screened for the presence of MICL autoantibodies using a modified ELISA method25. In brief, Nunc Maxisorp 96-well plates were coated with equivalent amounts of the Fc-fusion proteins Fc–hMICL and Fc–mCLEC12B diluted in PBS overnight at 4 °C. Plates were blocked with 10% BSA and 10% HI goat serum (Merck) in PBS for 1 h. Sera from patients with rheumatoid arthritis, COVID-19 and SLE and healthy control were diluted to a 1:256 dilution in PBS and added to the pre-blocked plate and incubated for 2 h. Bound autoantibodies were detected with horseradish peroxidase-conjugated goat anti-human F(ab′)2 fragment (Jackson Immunoresearch) diluted 1:50,000 in PBS for 1 h. TMB substrate was added, and absorbance was measured at 450 nm in a plate reader (Tecan).

A. fumigatus systemic infection model

Micl−/− and C57BL/6J female mice were injected intravenously with 106 A. fumigatus ATCC 13073 conidia, as previously described66. Mice were culled when they lost 20% body weight or had become moribund. To investigate the role of NET formation, mice were injected i.p. with 4 mg kg−1 daily of GSK484 (Cambridge Biosciences) from day 1 to day 5. Organs were homogenized in PBS and used for the determination of fungal burdens and levels of inflammatory cytokines. Fungal burdens were determined by serial dilution onto potato dextrose agar plates and normalized to organ weights. Cytokines were measured by ELISA (BD Biosciences), as described by the manufacturer, and normalized to protein concentration.

Statistical analysis

Data are represented as mean ± s.d., unless otherwise indicated. All statistical analyses were performed using GraphPad Prism (v9, GraphPad Software) and depicted in the respective figure legends. For all experiments with two groups, two-tailed unpaired Student’s t-tests (equal variances) or two-tailed Mann–Whitney tests were used. One-way or two-way ANOVA (with equal variances) with correction for multiple comparisons was performed for experiments with more than two groups. All P < 0.05 were considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.