NETs decorated with bioactive IL-33 infiltrate inflamed tissues and induce IFN-α production in patients with SLE

IL-33, a nuclear alarmin released during cell death, exerts context-specific effects on adaptive and innate immune cells, eliciting potent inflammatory responses. We screened blood, skin, and kidney tissues from patients with systemic lupus erythematosus (SLE), a systemic autoimmune disease driven by unabated type I IFN production, and found increased amounts of extracellular IL-33 complexed with neutrophil extracellular traps (NETs), correlating with severe, active disease. Using a combination of molecular, imaging, and proteomic approaches, we show that SLE neutrophils, activated by disease immunocomplexes, release IL-33–decorated NETs that stimulate robust IFN-α synthesis by plasmacytoid DCs in a manner dependent on the IL-33 receptor ST2L. IL33-silenced neutrophil-like cells cultured under lupus-inducing conditions generated NETs with diminished interferogenic effect. Importantly, NETs derived from patients with SLE are enriched in mature bioactive isoforms of IL-33 processed by the neutrophil proteases elastase and cathepsin G. Pharmacological inhibition of these proteases neutralized IL-33–dependent IFN-α production elicited by NETs. We believe these data demonstrate a novel role for cleaved IL-33 alarmin decorating NETs in human SLE, linking neutrophil activation, type I IFN production, and end-organ inflammation, with skin pathology mirroring that observed in the kidneys.


Introduction
Systemic lupus erythematosus (SLE) is the prototype systemic autoimmune disease characterized by dysregulated innate and adaptive immunity leading to generation of autoantibodies and formation of immune complexes (ICs) deposited to afflicted tissues such as the skin and kidneys (1). Transcriptome studies (2,3) have implicated neutrophils in SLE, which is further corroborated by genetic (4,5) and functional (6) evidence. A cardinal disease feature is unabated production of type I IFN (2), which exerts pleiotropic effects on hematopoietic and nonhematopoietic cells, thus propagating inflammation (7,8). One of the primary sources of IFN is plasmacytoid DCs (pDCs) especially in response to TLR-7 and -9 signaling (8). Genetic depletion (9) or inactivation (10) of pDCs diminishes IFN and ameliorates lupus-like disease in animal models, and likewise, therapeutic targeting of pDCs and/or type I IFN has yielded promising results in human SLE (11,12). Various molecules can instruct IFN production by pDCs, such as neutrophil extracellular traps IL-33, a nuclear alarmin released during cell death, exerts context-specific effects on adaptive and innate immune cells, eliciting potent inflammatory responses. We screened blood, skin, and kidney tissues from patients with systemic lupus erythematosus (SLE), a systemic autoimmune disease driven by unabated type I IFN production, and found increased amounts of extracellular IL-33 complexed with neutrophil extracellular traps (NETs), correlating with severe, active disease. Using a combination of molecular, imaging, and proteomic approaches, we show that SLE neutrophils, activated by disease immunocomplexes, release IL-33-decorated NETs that stimulate robust IFN-α synthesis by plasmacytoid DCs in a manner dependent on the IL-33 receptor ST2L. IL33-silenced neutrophil-like cells cultured under lupus-inducing conditions generated NETs with diminished interferogenic effect. Importantly, NETs derived from patients with SLE are enriched in mature bioactive isoforms of IL-33 processed by the neutrophil proteases elastase and cathepsin G. Pharmacological inhibition of these proteases neutralized IL-33-dependent IFN-α production elicited by NETs. We believe these data demonstrate a novel role for cleaved IL-33 alarmin decorating NETs in human SLE, linking neutrophil activation, type I IFN production, and end-organ inflammation, with skin pathology mirroring that observed in the kidneys.
IL-33 is a chromatin-bound alarmin that exists in the nucleus as full-length isoform (31 kDa), whereas it is externalized mainly during necrotic cell death of nonhematopoietic cells as a shorter (18 kDa) cytokine-like isoform that signals through the ST2L receptor (17). Immune cells express and release IL-33 upon stimulation by death-or damage-related agents. Notably, full-length IL-33 ([fl]IL-33) has modest bioactivity extracellularly, which, nevertheless, is profoundly induced after cleavage by neutrophil and mast cell serine proteases or allergen-related cysteine proteases (18).
Previous work (19,20) has suggested context-specific expression and function of IL-33. Although IL-33 supports regulatory T cell (21) and nonpathogenic Th2 responses (22), it can also drive T cells to become effectors (23). Moreover, IL-33 may confer autoimmunity by augmenting innate immune responses (24). In SLE, elevated serum concentrations of IL-33 and the soluble form of the IL-33 receptor ST2L have been reported (25)(26)(27). Genetic studies have also implicated IL33 gene polymorphisms in susceptibility to SLE (28,29). In animal studies, exogenous IL-33 induced the B cell activating factor (30), a growth factor implicated in SLE, and MRL/lpr lupus-prone mice treated with anti-IL-33 antibody have reduced renal inflammation and serum autoantibodies (31). However, the in vivo expression and function of IL-33, as well as its contribution to immune cell activation and tissue inflammation in SLE, remain ill defined.
Herein, driven by our striking finding of extracellular IL-33 complexed with NETs in lupus-afflicted tissues, we reasoned that NETting neutrophils may represent an important source of IL-33 alarmin contributing to excessive type I IFN. In addition to skin and kidneys, screening of serum samples from patients with active SLE revealed IL-33-complexed NETs, thus reinforcing the disease pathogenesis and end-organ injury roles of NETs. Through genetic, imaging, and proteomic assays, we confirmed that ex vivo, cultured, human SLE neutrophils and their NETs express significant amounts of IL-33. Our studies denote a critical role for IL-33 as a chromatin-bound alarmin mediating the IFN-inducing effect of SLE NETs through interaction with the ST2L receptor on pDCs. Importantly, we implicate neutrophil proteases for the production of cleaved interferogenic IL-33 isoforms on lupus NETs, thus suggesting that interference with IL-33 maturation and/or signaling could be therapeutically exploited.

Results
IL-33-decorated NETs were detected in inflamed tissues of patients with active SLE. We focused on the role of IL-33, a nuclear alarmin released during cell death, in SLE. To gain initial insights, we monitored biopsy specimens from the kidney and skin, which represent 2 frequently affected organs in SLE, of patients with active disease (1). We identified extracellular structures in SLE kidneys, especially at the tubulointerstitium, where IL-33 and NET markers were colocalized, indicative of IL-33-containing NETs ( Figure 1A). IL-33-decorated chromatin structures were also detected in lupus-affected dermis (corroborated by H&E staining), contrary to healthy skin, which exhibited no inflammation or IL-33 expression ( Figure 1B).
Based on previous work linking blood neutrophil activation and NETosis in active severe SLE (2, 32-34), we screened for IL-33 NETs in the serum of patients with SLE. For this, we developed 2 sandwich ELI-SA systems based on an anti-IL-33 coating antibody and a detection antibody specific for either dsDNA, the most prevalent nucleic acid in SLE ICs, or neutrophil myeloperoxidase (MPO). Using both assays in independent patient cohorts, serum samples from patients with SLE exhibited increased levels of IL-33-complexed NETs as compared with those of healthy counterparts (Figure 2, A and B). Serum IL-33 NET concentration correlated significantly with patient disease activity assessed by the validated SLE Disease Activity Index (SLEDAI) (35) (Figure 2, A and B), and longitudinal reduction in serum IL-33/MPO complexes was noted in patients with good clinical response to belimumab treatment (36) (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.147671DS1). To validate our technique, and in agreement with previous reports (32,37), we also detected increased NETs containing neutrophil MPO (MPO/dsDNA complexes) in SLE versus healthy sera (Supplemental Figure 1B).
Lastly, we purified blood neutrophils and used confocal microscopy to characterize spontaneous NETs release. Consistent with the serum data, SLE neutrophils generated IL-33 NETs, whereas healthy neutrophils did not release NETs and displayed minimal IL-33 expression ( Figure 2C). We also detected IL-33 decoration on NETs from SLE low-density granulocytes (Supplemental Figure 1C), a subset of pathogenic neutrophils with enhanced NETosis potential. In line with the immunofluorescence observations, IL33 mRNA levels were elevated in neutrophils from patients with SLE versus healthy individuals ( Figure 2D). Collectively, SLE neutrophils release IL-33 NETs in the blood and other inflamed tissues, such as the kidneys and skin, suggesting their implication in the disease.
SLE ICs were potent inducers of IL-33-containing NETs. Soluble factors in the serum contribute to the NETotic potential of SLE neutrophils (14,32). By culturing healthy neutrophils in medium supplemented with serum derived from patients with SLE who were positive for anti-dsDNA and anti-ribonucleoprotein autoantibodies, we noted production of IL-33-decorated NETs and upregulation of IL33 mRNA (Supplemental Figure 2, A-C). Within lupus serum, ICs induce NETs through activation of neutrophil Fcγ receptors (FcRs) and TLR-7 receptors (14). To address this, neutrophils were pretreated with FcR-blocking reagent or the TLR-7 antagonist IRS661, followed by incubation with lupus serum. Inhibition of either receptor led to reduced production of IL-33-containing NETs and decreased IL33 mRNA levels (Supplemental Figure 2, A-C), consistent with a mediating effect of ICs.
Next, we sought to examine the effect of ICs directly on neutrophils and in relation to the production of IL-33 NETs. By culturing SLE neutrophils with nucleic acid-containing ICs generated as previously described (38), we observed increased NET production (Supplemental Figure 2D). In confirmation of our experimental setup, IC-induced NETs stained positive for 8-Oxo-2′-deoxyguanosine (8-OH-dG; a marker of oxidized DNA), TOMM20 (an outer-membrane mitochondrial protein), and citrullinated histone 3 (a marker indicative of the NOX-independent pathway of NETosis), which are all hallmarks of SLE NETosis (16) (Supplemental Figure 3, A and B). In contrast, resting (unstimulated), healthy neutrophils displayed minimal IL-33 and 8-OH-dG expression, and TOMM20 staining was cytoplasmic (Supplemental Figure  3A). SLE neutrophils exposed to ICs exhibited increased intracellular IL33 mRNA and protein levels ( Figure  3, A and B), notwithstanding the low IL-33 abundance in myeloid cells (39,40). Accordingly, both confocal microscopy and immunoblotting in precipitated NETs revealed elevated IL-33 protein levels in IC-released versus spontaneously released SLE NETs (Figure 3, C and D), whereas SLE ICs, per se, contained no measurable IL-33 (Supplemental Figure 4A). Contrary to SLE neutrophils, healthy neutrophils produced IL-33 NETs only after priming with exogenous recombinant IFN-α followed by administration of ICs (Supplemental Figure 4B) (13,14), which reiterates the role of the SLE milieu in determining the neutrophil NETotic potential. Furthermore, when comparing NOX-dependent (namely, phorbol 12-myristate 13-acetatetreated [PMA-treated] healthy neutrophils) versus NOX-independent (namely, monosodium urate-treated [MSU-treated] healthy neutrophils, IC-treated SLE neutrophils) NETs, only the latter exhibited significant IL-33 decoration (Supplemental Figure 3B), thus suggesting a mechanistic link between NOX-independent pathway of NETosis and IL-33 release. Conclusively, nucleic acid-containing IC stimulation of SLE neutrophils recapitulates typical lupus NETosis and is associated with increased IL-33 expression.
SLE NETs activated pDCs to produce type I IFN in an ST2L-dependent manner. In SLE, IC-treated neutrophils externalize NETs that trigger pDCs to produce type I IFN (14)(15)(16). Besides the nucleic acid content, NETs interferogenic capacity is also regulated by the protein cargo, thus the NET complex is a potent mediator of the characteristic dysregulated immune response in lupus (41). We examined whether IL-33 contributes to the interferogenic effect of IC-induced SLE NETs.
First, we assessed the expression of ST2L on the membrane of peripheral blood CD123 + pDCs by flow cytometry and confirmed that healthy and SLE pDCs are already receptive of IL-33 signaling (Supplemental Figure 5, A and B). We administered NETs derived from IC-treated SLE neutrophils to healthy pDCs that had been pretreated with an inhibitory anti-ST2L antibody (aST2L) to block IL-33 signaling. pDCs were pretreated with FcR blocker to minimize any IC carryover or nonspecific aST2L effects. In agreement with findings in previous reports (14)(15)(16), SLE IC NETs induced robust IFNA mRNA and protein expression, which was significantly ameliorated upon ST2L blockade ( Figure 4A). A similar effect was noted with the mRNA levels of IRF7 (IFN regulatory factor 7), a characteristic IFN-stimulated genes ( Figure 4B). Likewise, SLE IC NET-containing supernatants pretreated with anti-IL-33 blocking antibody displayed significantly reduced inteferogenic capacity (Supplemental Figure 5C), thus further supporting the role of the IL-33/ST2L axis on NET-mediated IFN-α production by pDCs. Of note, PMA-induced NETs released from healthy neutrophils induced a less profound type I IFN response (such as IFNA and IRF7 mRNA expression) by pDCs, which was not significantly altered by ST2L blockade (Supplemental Figure 5D).
IFN production by pDCs is mediated through activation of TLR-7 and TLR-9, with cytosolic receptors playing an auxiliary role (14)(15)(16). To gain insights into the intracellular events linking IL-33-complexed NETs with IFN-α production by pDCs, we examined the levels of phosphorylated IRF7 (p-IRF7), which is essential for the expression of type I IFN genes downstream to TLR signaling. Treatment of pDCs with SLE IC NETs caused upregulation of p-IRF7, which was reversed following ST2L blockade ( Figure 4C), therefore reiterating the role of the IL-33/ST2L axis in SLE NETs' interferogenic effect. To discern the relative effects of NET-associated DNA and IL-33, we cultured pDCs using cytochalasin D, which blocks endocytosis of NETs, or chloroquine, disrupting endosomal TLR trafficking. With either treatment, IFN-α response by pDCs was attenuated (Supplemental Figure 5E), suggesting that the NET DNA scaffold is essential for the immunostimulatory effects of IL-33. In line with this, treatment of pDCs with an exogenous cytokine isoform (18 kDa) of IL-33 alone failed to induce significant IFN-α production (Supplemental Figure 5F). These results indicate that IL-33 associated with SLE NET DNA may interact with ST2L on pDCs to regulate type I IFN through IRF7.
Loss of IL-33 impaired the interferogenic potential of SLE-like NETs. To provide direct evidence for the implication of IL-33 in the interferogenic capacity of SLE NETs, we pursued gene-silencing experiments. Due to the short lifespan of primary neutrophils, we differentiated the HL-60 myeloid cell line into neutrophil-like cells using a retinoic acid-based protocol (42). Differentiated HL-60 cells expressed characteristic neutrophil markers (Supplemental Figure 6A), and following priming with recombinant IFN-α and treatment with SLE ICs, they produced NET-like structures, as previously described (43), which were decorated with IL-33 ( Figure 5A). We performed IL33 gene silencing in neutrophil-like HL-60 cells, which resulted in efficient knockdown of IL-33 protein (Supplemental Figure 6, B and C) without affecting the release and DNA content of IC-induced NETs (Supplemental Figure 6D). Notably, NETs derived from HL-60 cells transfected with scramble siRNA induced significant amounts of IFNA mRNA and protein by healthy pDCs in an ST2L-dependent manner, thus recapitulating the effects of SLE NETs ( Figure 5B). In contrast, NETs from IL33-silenced neutrophil-like cells lost their interferogenic capacity, and ST2L blockade had no additional effect ( Figure 5B). These data support that the IL-33/ST2L axis is critical for the interferogenic potential of SLE IC NETs.
SLE IC-induced NETs were enriched in IL-33 isoforms with putative enhanced bioactivity. Our findings suggest that NET-complexed IL-33 from IC-treated lupus neutrophils drives IFN-α production by pDCs. IL-33  ) were performed to determine mRNA and protein expression or secretion of IFN-α, respectively, by healthy pDCs treated overnight with IC-induced SLE NETcontaining supernatants (IC NETs) (25% v/v). The contribution of the IL-33/ST2L axis on IFN-α response was assessed by pretreating pDCs with an antibody against ST2L (a-ST2L, 3 μg/mL). FcR blocking reagent was used to avoid any IC-carry over effect or non-specific a-ST2L binding. Each dot represents a different pDC donor (n = 12) and bar plots show the mean ± SEM expression. *P < 0.05; ***P < 0.001 (2-tailed, repeated measures ANOVA with Holm-Sidak correction). (B) Real-time PCR for IRF7 mRNA expression in pDCs (n = 12 healthy donors) stimulated with IC NETs (25% v/v), with or without pre-treatment with FcR blocking agent and a-ST2L, as described for A. Quantification was performed using the 2 -ΔΔCT method, where ΔCt = IRF7 Ct minus GAPDH Ct. **P < 0.01 (2-tailed, repeated measures ANOVA with Holm-Sidak correction). (C) Intracellular p-IRF7 staining was performed 4 hours after stimulation of purified pDCs with IC NETs (25% v/v), with or without pretreatment with FcR blocking agent and a-ST2L, as described for A. Each dot (open circle, full circle, triangle) corresponds to the kinetics of p-IRF7 in 9 independent donors, and bar plots show the mean ± SEM expression. *P < 0.05 exists in chromatin-associated, full-length nuclear form that can be cleaved by neutrophil serine proteases into highly bioactive forms (18,44,45). We asked whether IC SLE NETs contain cleaved IL-33 isoforms not detectable by immunoblotting, due to their low abundance. We performed parallel reaction monitoring mass spectrometry in NET-precipitated proteins from untreated and IC-treated SLE neutrophils, as well as from healthy neutrophils undergoing PMA-induced NETosis (n = 6-8 individuals in each group) (Supplemental Tables 2-4). In line with the immunofluorescence data, IC-treated SLE NETs demonstrated the highest intensity of the IL-33 proteotypic peptide VDSSENLCTENILFK (aa 251-265) ( Figure 6).
We next designed the same focused proteomic analysis using pooled NET proteins to obtain a richer proteomic profile and gain additional insights on possible NET protease-mediated IL-33 cleavage and activation. We collected NET-associated proteins from a new cohort of the same 3 groups of neutrophils followed by targeted IL-33 peptide quantification (normalized over MPO intensity to allow for comparative analyses). Through this approach, we identified 2 additional IL-33 proteotypic peptides, namely MLM-VTLSPTK (aa 180-189) and DNHLALIK (aa 243-250), which displayed increased levels in IC-induced versus untreated SLE NETs and PMA-treated healthy NETs ( Figure 6B). Importantly, all 3 IL-33 peptides found to be upregulated in SLE IC NETs were localized near the C-terminus, which exists in the bioactive isoforms of IL-33 and accounts for their signaling potential (17,18), and not in the N-terminus, which is found only in the full-length protein ( Figure 6C). Together, these results suggest that IC-induced SLE NETs may be enriched in cleaved IL-33 isoforms with putative enhanced bioactivity. NET proteases were implicated in the generation of interferogenic NET-complexed IL-33. IC SLE NETs are enriched in serine proteases, including elastase and cathepsin G, which are potent IL-33 activators (14,18). Driven by our proteomic data, we hypothesized that NETs may act as a platform for neutrophil proteases to target NET-bound IL-33, thus generating bioactive isoforms that augment the interferogenic potential of SLE NETs. To evaluate this possibility, and due to the low endogenous level of IL-33, we administered recombinant (fl)IL-33 in cultures of IC-treated NETotic neutrophils from patients with SLE. As a control, we used spontaneously NETotic SLE neutrophils and PMA-or MSU-induced NETotic neutrophils from healthy donors, all supplemented with the same amount of (fl)IL-33. Immunoblotting in the supernatants revealed an approximately 19 kDa protein band in SLE NETting neutrophils, especially the IC-treated ones, which might correspond to protease-generated IL-33 isoforms (18) ( Figure 7A).
To address more directly the implication of neutrophil proteases in the interferogenic properties of NET-bound IL-33, we blocked neutrophil elastase using the selective inhibitor sivelestat, optimizing the inhibitor concentration and timing of administration to minimize any interference with the NETotic process (Supplemental Figure 7A). We repeated the (fl)IL-33 plus NETs mixture assay using IC-stimulated SLE NETting neutrophils, which were pretreated with sivelestat. Immunoblotting of the sivelestat-treated supernatants showed reduced abundance of the approximately equal to 19 kDa protein band of putative bioactive IL-33 isoforms (Figure 7B), thus suggesting diminished elastase-mediated cleavage of (fl)IL-33. Following treatment with sivelestat, IC-treated SLE NETs abolished their capacity to induce IFN-α response by pDCs (Figure 7, C and D), implying that NET-bound IL-33 signaling is abrogated due to lack of protease-mediated activation. A similar effect was observed using an inhibitor of cathepsin G (Supplemental Figure 7B), another NET-associated neutrophil protease that mediates IL-33 cleavage.
To further assess IL-33 bioactivity, we administered SLE NET-cleaved or unprocessed (fl)IL-33 to cultures of pDCs, which were stimulated with the TLR-9 ligand CpG-A. Supernatants derived from the incubation of IC-treated SLE NETting neutrophils with (fl)IL-33 caused a significant, ST2Ldependent increase in p-IRF7 levels of pDCs, suggesting enrichment in bioactive IL-33, whereas an equivalent dosage of uncleaved (fl)IL-33 had the opposite effect ( Figure 7E). Furthermore, SLE NET-processed IL-33-containing supernatants promoted IFNA and IRF7 mRNA expression by CpG-A-activated pDCs (Supplemental Figure 7C).
Finally, to recapitulate in vitro the elastase and IL-33 interaction presumed to occur on SLE NETs, we incubated (fl)IL-33 with recombinant elastase to obtain supernatants enriched in bioactive IL-33 (Supplemental Figure 7D). Notably, CpG-A-treated pDCs supplemented with bioactive (elastase-treated) IL-33-containing supernatants showed significant upregulation of IFNA and IRF7 mRNA expression,

Discussion
A poorly explained feature of SLE is unabated type I IFN signaling that spurs autoreactive responses and persists even during clinical remission (2,8). pDCs are a major source of IFN-α, and lupus-prone mice with defective pDC-mediated IFN response have reduced autoantibody formation, reduced lymphadenopathy, and prolonged survival (46). Perpetual production of IFN-α in SLE is triggered by self-derived nucleic acids complexed with autoantibodies or immunostimulatory proteins, such as in the form of NETs (13)(14)(15). We focused on IL-33, a chromatin-bound alarmin with context-specific immunomodulatory effects, and evaluated its role in human SLE. Driven by our observation that neutrophils infiltrate and release IL-33bearing NETs in the blood and other inflamed tissues of patients with SLE, we herein demonstrate that lupus neutrophils are prone to producing IL-33 NETs, which induce a robust IFN-α response by pDCs through the ST2L receptor. Importantly, our NETs proteome data, coupled with results of ex vivo inhibition assays, implicate NET proteases in the generation of bioactive IL-33 of high interferogenicity, thus offering what we believe are novel mechanistic insights linking neutrophil activation and NETosis with IFN-α and end-organ injury in human SLE.
Extracellular IL-33 augments immune responses during tissue inflammation and injury; however, its precise role in autoimmunity remains elusive. We show that IL-33 is externalized on SLE NETs and contributes to their interferogenic capacity through ST2L on pDCs. This is in line with findings of a previous study suggesting that chromatin binding regulates ST2-mediated bioactivity of IL-33 (47). Possible explanations for the adjuvant effect of NET-complexed IL-33 include protecting the NET structure from degradation (15) or facilitating NET uptake through ST2L (48). IL-33/ST2L axis can both enhance (49) and inhibit (50) TLR signaling in a context-specific fashion, and IL-33 induces IRF7 expression in innate lymphoid cells (51), which might represent a mechanism by which NET-complexed IL-33 renders pDCs more responsive to immunostimulatory DNA. IL-36, another IL-1 family cytokine, induces IFN-α production by facilitating endosomal TLR trafficking in pDCs (52). Of interest, previous studies have indicated cross-talk signaling between cyclic GMP-AMP synthase-stimulator of interferon genes (cGAS-STING), known to be triggered by NET DNA (53), and IL-33/ST2L in the context of allergic airway inflammation (54,55), therefore raising the hypothesis that NETassociated IL-33 might also amplify non-TLR cytoplasmic nucleic acid-sensing signaling pathways.
IL-33 complexed with or processed by SLE NETs, but not the recombinant cytokine form, was capable of inducing robust IFN-α production by pDCs. This raised the possibility that IL-33 may be modified by NETs to gain biological activity. Indeed, IL-1 family cytokines require proteolytic processing for activation, and IL-33 can be cleaved into highly bioactive isoforms by neutrophil proteases (18,44,45). However, the topology of this process in the setting of autoimmunity has not been demonstrated. To this end, our proteomic analysis of SLE IC NETs revealed upregulation of 3 distinct IL-33 peptides exclusively localized near the C-terminal cytokine domain, which exists in mature IL-33 isoforms (17,18). In this regard, SLE neutrophils and their NETs exhibit a potent serine protease signature (14); moreover, our proteomic analysis revealed the absence of a major endogenous inhibitor of elastase (SerpinB1) from IC SLE NETs (Supplemental Tables 2-4). Accordingly, inhibition of elastase and cathepsin G in NETting SLE neutrophils abrogated the IL-33-mediated interferogenic effect of NETs. Our data are in line with findings implicating NET protease-processed IL-1β in gout attacks (56) and of neutrophil protease-generated IL-33 in acute lung injury (18), pointing to neutrophil proteases generating NET-associated IL-33 with high interferogenic activity in the context of lupus.
Limited studies have detected proteolytically processed forms of IL-33 in biological samples because characterization of human extracellular IL-33 has been challenging due to its low abundance. Genetic evidence supports that even small changes in IL-33 expression are implicated in disease susceptibility (57). Under inflammatory conditions, neutrophils are rapidly recruited into afflicted tissues, and the expression level of alarmins may increase significantly, reaching locally high concentrations. Intriguingly, under high neutrophil densities, NETs may accumulate (58) and, presumably, act as platforms for the extracellular scavenging and processing of IL-33 released by neighboring damaged epithelial and endothelial cells (18,45).
Previous work has suggested that oxidation of IL-33 limits its bioactivity, due to formation of disulfide bridges that impede the IL-33/ST2L interaction (59). Since neutrophil activation and NETosis in SLE occur JCI Insight 2021;6(21):e147671 https://doi.org/10.1172/jci.insight.147671 under oxidative stress driven by mitochondrial ROS (16,60), it is of interest to discern how NET-associated IL-33 may retain its bioactivity. Notably, oxidized isoforms have not been detected in nuclear (DNA-bound) IL-33 (59), which could indicate that the NET-DNA scaffold or histones are protective against oxidation. Also, according to published data (61) and our proteomics analysis, SLE NETs are decorated with molecules that display reducing capacity, including thioredoxin reductase, peroxiredoxin-2, and glutathione S-transferase, which might counteract IL-33 oxidation. Finally, if IL-33 abolishes NET protection extracellularly, it is possible that it exerts its biological effects acutely before its oxidation, pertinent to the close proximity of NETting neutrophils and pDCs in lupus-inflamed tissues (62,63).
Our group (2,32) and others (13,33,64) have associated neutrophils and aberrant NET formation with lupus nephritis, but the implication of specific alarmins externalized on SLE NETs to the kidney disease is not well understood. Our ex vivo functional assays, coupled with evidence that pDCs infiltrate the inflamed kidneys in SLE (62), raise the possibility that IL-33-decorated NETs trigger intrarenal type I IFN production in lupus nephritis, as recently suggested by single-cell transcriptomic studies (65). Interestingly, we found that IL-33 NETs were predominantly localized within the tubulointerstitium, and tubulointerstitial inflammation has been correlated with pDC infiltrates (66) and poor prognosis in lupus nephritis (67). IL-33 exhibits myeloid cell chemoattractant properties (68), thus it might also attract neutrophils to orchestrate a neutrophil/NETs IL-33/IFN-α feedback loop within lupus-inflamed kidneys.
Considering that IL-33-decorated NETs were also detected in SLE cutaneous lesions, it is possible that these structures can be sensed by skin-infiltrating pDCs, thus contributing to the profound type I IFN signature observed in lupus skin (69). This finding corroborates those of previous studies showing that kidney and skin tissue of patients with SLE share common genomic perturbations, including upregulated type I IFN-inducible genes (65).
Although we focused on the role of IL-33 in regulating the interferogenic capacity of NETs, it is possible that IL-33 might contribute to SLE through other mechanisms. Thus, IL-33-bearing NETs might interact with other ST2L-expressing immune cell types such as Th2, Tregs, or NK cells (17,19). In addition, the IL-33/ST2L axis has been shown to promote fibrosis under inflammatory conditions (70).
Our results have potential implications for the design of novel therapeutics to counteract aberrant IFN-α production in SLE and other relevant pathologies such as infection-induced autoinflammatory pneumonitis (71). Since IL-33 is released by activated neutrophils undergoing NETosis, targeting IL-33/ ST2L might be beneficial in selectively neutralizing excessive IFN-α response without inducing generalized immunosuppression. Anti-IL-33 and/or aST2L antibodies are currently under clinical development (72); in a preliminary report, administration of anti-IL-33 antibody in MRL/lpr lupus-prone mice reduced anti-dsDNA and IC levels, kidney inflammation, and proteinuria (31).
In conclusion, we provide evidence that SLE NET-derived IL-33 processed by neutrophil proteases may contribute to disease pathogenesis by augmenting IFN-α production by pDCs. IL-33-bearing NETs infiltrating lupus end organs such as the kidneys and skin suggest they play an important role in regulating local autoimmune inflammation and tissue injury. Accordingly, IL-33 and/or its maturation process represent druggable targets toward ameliorating excessive IFN-α production and SLE pathology.

Methods
Patients. We included patients with SLE followed at the Rheumatology Clinic, University Hospital of Heraklion (Heraklion, Greece), who met the 1997 American College of Rheumatology revised classification criteria (73) (Supplemental Table 1). Active SLE was defined according to an SLEDAI 2000 score (SLE-DAI-2K) equal to or greater than 6 (74) on the day of sampling. Clinical response to treatment was evaluated according to the SLE Responder Index 4 (75). Patients abstained from their medications for at least 12 hours prior to blood-sample collection.
In NET-mediated cleavage experiments, recombinant glutathione-S-transferase-tagged, (fl)IL-33 (100 nM; catalog H00090865; Abnova) was added after 2 hours of neutrophil stimulation for 1 hour. PMA (100 nM; catalog P1585; Sigma-Aldrich) and MSU crystals (100 μg/mL; catalog tlrl-msu; Invivogen) were used as controls. Both cell supernatants and NETs were precipitated for Western blot analysis. To quantify NET-containing supernatants, we used the SYTOX Green Nucleic Acid Stain Kit (catalog S7020; Invitrogen) according to the manufacturer's instructions. NETs were quantified as previously described (77). Briefly, decondensed nuclei or filamentous DNA structures stained for both DAPI and MPO were considered NETs. We used the FIJI software (ImageJ2; NIH) and assessed 5 randomly selected coverslip fields for each treatment and 3 biological replicates (independent experiments) to calculate the proportion of NET-releasing cells in the total number of cells.
Paraffin-embedded renal and skin sections from patients with active SLE and healthy donors were cut at 3 μm, mounted, and dried on Superfrost Plus slides (Thermo Fisher Scientific). After dewaxing and rehydration, the sections were incubated in citrate-based HIER (heat-induced epitope retrieval) buffer at 60°C in a water bath for 60 minutes. After antigen retrieval, the sections were left in the respective citrate buffer at room temperature to cool below 30°C, rinsed with deionized water 3 times, then with TBS once at pH 7.4, and permeabilized for 8 minutes with 0.2% Triton X-100 in TBS, followed by 3 rinsing steps with TBS and blocking with 5% w/v BSA/TBS. Immunostaining was conducted as detailed previously. The same tissue sections were also stained with H&E and evaluated using a Nikon Eclipse E-400 light microscope (×400 magnification).
NET protein precipitation for mass spectrometry. Neutrophils (4-7 × 10 6 ) were stimulated using PMA (100 nM; Sigma-Aldrich) for 3 hours at 37°C. SLE neutrophils were stimulated using ICs (as described in Generation, isolation, and quantification of NETs) or left unstimulated (for 3 hours at 37°C) to release NETs spontaneously. Supernatants were carefully removed, cells were washed twice, and NETs were harvested using DNase (20 U/mL; catalog 4716728001; Roche) in HBSS (catalog 14025; Thermo Fischer Scientific) for 20 minutes. Supernatant-containing NETs were isolated via centrifugation (400g, 4°C, 10 minutes). NETs proteins were precipitated using 80% v/v acetone overnight at -20°C. Precipitated pellet was obtained via centrifugation at 10,000g for 30 minutes. Then, precipitated proteins were washed twice using an acetone-based buffer (80% v/v acetone, 10 mm Tris-HCl at pH 8). Protein pellets were stored at -80°C until mass spectrometry analysis.
Mass spectrometry. Detailed descriptions of sample preparation, mass spectrometry and data analysis, including total proteome and targeted proteomic analyses, are presented in Supplemental Methods.
Quantitative real-time PCR. Total RNA from neutrophils or pDCs was collected using the TRIzol (catalog 15596026; Invitrogen) extraction protocol followed by Turbo DNase (catalog AM2238; Ambion) treatment JCI Insight 2021;6(21):e147671 https://doi.org/10.1172/jci.insight.147671 according to the manufacturer's instructions. cDNA was prepared using PrimeScript 1st Strand cDNA Synthesis Kit (catalog RR037A; Takara Bio). Transcripts were quantified by incorporation of the KAPA SYBR FAST qPCR Kit Master Mix (catalog KK4602; Kapa Biosystems) in a CFX Connect Real-Time PCR System (BIO-RAD). Expression levels were normalized to GAPDH or HPRT1 and calculated by the change-in-threshold method; that is, 2 -ΔΔCT . Primer sequences are listed in Supplemental Table 5.
ELISA. NET-associated IL-33/dsDNA complexes were quantified by a modified capture ELISA. Specifically, 1 μg/mL mouse anti-human IL-33 ("Nessy") antibody (Enzo) was coated overnight in 96-well microtiter plates. After blocking with 1% w/v BSA and 1% v/v donkey serum, human serum (1:6 dilution) was added together with a peroxidase-labeled anti-DNA monoclonal antibody in the Cell Death ELISAPLUS Kit (1:25; catalog 11774425001; Roche). The plate was incubated for 2.5 hours, shaking at 300 rpm at room temperature. After 3 washes with PBS, peroxidase substrate (ABTS) was added to incubate at room temperature in the dark for 20 minutes. Absorbance at 405 nm wavelength was measured (490 nm was used as a reference filter). IL-33/MPO complexes were quantified using a goat anti-human IL-33 antibody (R&D) for coating (1:200 dilution) and a mouse anti-human MPO antibody (catalog HM1051; Hycult Biotech) for detection (1:100 dilution). After 3 washes with PBS, anti-mouse HRP antibody (1:10,000 dilution; catalog AP308P; Millipore) was added for 1 hour at room temperature. After 5 washes with PBS, 3,3',5,5;-tetramethylbenzidine (TMB; catalog 002023; Invitrogen) was added and light absorbance was measured at 450 nm (540 nm was used as a reference filter). Detection of IFN-α (Human IFN Alpha Multi-Subtype ELISA Kit; catalog PBL41110-1; PBL Assay Science) in culture supernatants was performed by precoated sandwich ELISA. Light absorbance at 450 nm was measured using the ELx800 microplate reader (Biotek). Background signal was subtracted. All samples were assessed in duplicate.
Statistics. Data were analyzed using GraphPad Prism 8 software. Results are presented as the mean ± SEM and/or dot plots. Data normality was assessed by the Shapiro-Wilk's test to guide selection of parametric or nonparametric tests. In assays including multiple conditions or treatments, repeated measures 1-way ANOVA was used, followed by the post hoc Holm-Sidak test to correct for multiple comparisons. The association between IL-33 serum complexes (i.e., IL-33/dsDNA, IL-33/MPO) and SLE disease activity (determined using the SLEDAI-2K) was assessed by the Spearman's rank correlation coefficient. P < 0.05 (2-tailed) was considered statistically significant.
Study approval. The study was approved by the Ethics Committee of the University Hospital of Heraklion, and all participants gave written informed consent (protocol no. 5944/14-6-2017).

Author contributions
SG designed and performed experiments, interpreted and analyzed the data, generated the figures and wrote parts of the manuscript. KG, GP, ED, DTB, and PS provided critical feedback, helped shape the research plan and reviewed and edited the manuscript. MLE and LR offered critical reagents and provided useful feedback on the manuscript. HG provided human paraffin-embedded tissue samples from patients with lupus nephritis. JZ analyzed the proteomics experiments and provided feedback on the manuscript. MM conducted and analyzed the proteomics experiments and provided feedback on the manuscript. GK and EB conducted and analyzed the proteomic experiments. PV co-supervised the project, designed experiments, interpreted data, and provided critical feedback on the manuscript. GB designed and supervised the study, recruited and evaluated patients with SLE, interpreted and analyzed data, generated the figures, and wrote the manuscript.