Diffuse alveolar hemorrhage (DAH) is a rare but severely life-threatening complication observed in patients with systemic lupus erythematosus (SLE) and other autoimmune diseases including anti-neutrophil cytoplasmic antibody (ANCA) vasculitis (1, 2). DAH is driven by injury to the alveolar capillary bed basement membrane leading to accumulation of red blood cells in the alveolar space. What causes injury to the basement membrane is specific to the systemic disease that underlies DAH. Specifically in SLE- and ANCA-associated vasculitis, immune complex deposition, complement activation, and neutrophil infiltration of the alveolar wall are the key histopathologic findings (3).

Type I IFN-producing neutrophils have come to the forefront in driving both inflammation and exacerbating tissue pathology in lupus (4–7). Termed low density neutrophils (LDNs), they respond to stimuli such as immune complexes with a much higher propensity to release neutrophil extracellular traps (NETs), a process called NETosis, compared to normal density neutrophils (8, 9). NETosis is a critical physiological response to bacterial and fungal infections, aiding to trap, immobilize, and help prevent the spread of infection (10). NETs are also triggered by cytokines, immune complexes and autoantibodies (11). The enzyme peptidylarginine deiminase 4 (PAD4) is expressed in mouse and human neutrophils and catalyzes the citrullination of histones leading to chromatin decondensation, a hallmark of NETosis. PAD4 has been proposed as a critical enzyme in the formation of NETs and its deletion or inhibition leads to an abrogation of NETosis in various contexts including bacterial infection, atherosclerosis, and the MRL/lpr SLE model (11–14). In addition, myeloid-specific deletion of PAD4 results in an inability of neutrophils from these mice to induce NETosis in response to the calcium ionophore A23187 and a reduction in inflammation and atherosclerosis in the context of ApoE-deficiency (15). The release of DNA and nuclear histones during NETosis drives inflammation and type I IFN production through activation of RNA/DNA sensing pathways (4). Although these mechanisms are vital for anti-microbial defenses, inappropriate activation exacerbates SLE pathology and acute lung injury (16). Recent evidence using the pristane-induced lupus model where mice develop DAH demonstrates that NETosis is an important contributor to lung injury (11). In this model, neutrophils accumulate in the lungs of mice following a single intraperitoneal injection of the hydrocarbon oil pristane. Enhanced NETosis is observed in the lungs of pristane-treated mice and treatment with recombinant DNase I reduced NET formation in the lung with a subsequent reduction in lung injury in C57BL/6 mice (17). Thus, NETs may play key roles in driving pristane-induced lung injury (17).

Additional pathways have also been shown to contribute to acute and chronic lung injury. For example, evidence from the monogenic autoimmune disorder COPA syndrome, which presents frequently as pulmonary hemorrhage, suggests that endoplasmic reticulum (ER) stress may be a key driver of lung injury in autoimmune disease (18). ER stress is sensed by ER-resident transmembrane proteins such as IRE1, ATF6 and PERK. For example, unfolded proteins that initiate ER stress bind to BIP which then dissociates from IRE1-α and PERK to promote their phosphorylation and activation. IRE1-α and PERK activation triggers cell protective mechanisms such as the synthesis of the ER chaperone protein CHOP via the generation of spliced Xbp1 transcripts, downstream of IRE1-α and/or ATF6 activation. Increased expression of Bip or Chop and higher spliced to unspliced Xbp1 transcript ratios are therefore useful markers of enhanced ER stress. Recent evidence suggests that activation of the ER stress response can act as a danger signal to augment the inflammatory responses in bronchial epithelial cells (19). Importantly, inhibiting ER stress can alleviate LPS-induced lung injury in mice (20), suggesting the importance of this pathway in inflammatory lung disease. IRE1-α signaling has widely been shown to be activated by numerous TLRs, including TLR7, the driver behind pristane-driven autoimmunity (21–23). Inhibition of IRE1-α activity during pristane administration abolishes lupus nephritis, autoantibody levels, and cytokine production. Additionally, in the bleomycin pulmonary fibrosis model, IRE1-α inhibition reduces alveolar epithelial apoptosis and prevented fibrosis (24, 25). Recently, crosstalk between NETs and ER stress has been observed in a model of sepsis-induced intestinal injury, with inhibition of NETosis and ER stress reducing intestinal epithelial cell monolayer barrier disruption, suggesting that neutrophils and NETosis contribute to ER stress in intestinal epithelial cells (26).

We hypothesized that NETs and NETosis directly drive ER stress in lung epithelial cells and contribute to acute lung injury in the pristane mouse model. In keeping with this we observed that expression of ER stress genes (Bip and Chop) increased in a time-dependent manner and closely matched the influx of neutrophils into the lungs of pristane-treated mice. To assess the importance of NETosis in driving lung inflammation in this model, we generated mice with myeloid specific deletion of PAD4 (LysMCrePAD4^fl/fl mice) and found that these mice showed reduced lung pathology and expression of ER stress and ISGs when treated with pristane compared to C57BL/6 (WT) mice. Co-culturing precision-cut lung slices from WT mice with neutrophils from pristane-treated PAD4^fl/fl mice drove expression of both ER stress markers and ISGs (11–14) whereas neutrophils from LysMCrePAD4^fl/fl mice failed to do so, supporting NETosis as a mechanism for driving ER stress. Because low-density neutrophils from SLE patients undergo spontaneous NETosis ex vivo (9), we hypothesized that neutrophils from SLE patients would drive ER stress in alveolar epithelial cells when compared with neutrophils from healthy controls (HC). As expected, SLE neutrophils, but not HC neutrophils, cocultured with an alveolar epithelial cell line induced high levels of ER stress and genes associated with inflammation in the lung such as Il8 and Il16, suggesting an important role for neutrophils in driving ER stress and inflammation in lung epithelial cells in the context of SLE. Taken together our results demonstrate the importance of neutrophils and NETs during lung injury in the pristane mouse model and in their implications in human SLE-related DAH.

Increasing evidence implicates release of neutrophil extracellular traps (NETs), NETosis, may drive lung injury, specifically ER stress in epithelial cells (9, 26, 31, 32). Our study shows that not only are ER stress markers increased in the pristane model of IFN-inducible SLE, but that inhibition of NETosis using LysMCrePad4^fl/fl mice in this model prevented pristane-induced ER stress and, more importantly, lung inflammation. Coculture experiments demonstrated that neutrophils isolated from pristane-treated mice could directly upregulate ER stress markers in lung slices in addition to driving IFN-stimulated gene expression. Importantly, we found that neutrophils isolated from SLE patients highly upregulated inflammation markers and induced both ER stress markers and ISG expression in alveolar epithelial cells, suggesting that ER stress may be an important pathological mechanism in SLE.

Recently neutrophils and NETosis have been shown to play important roles in lung inflammation. In autoimmune lung inflammation for example, NETs have been shown to activate lung fibroblasts and promote their proliferation and differentiation into myofibroblasts and thus contribute to interstitial lung disease in patients with poly- and dermato-myositis (33, 34). In the pristane-induced DAH model, treatment of mice with recombinant DNase I reduced NET formation and reduced lung pathology (17). In keeping with this pathogenic role for NETosis in pristane-induced lung injury, we have observed that myeloid-specific deletion of Pad4 results in reduced lung inflammation and ISG expression. Other evidence for NETosis playing a pathogenic role in lung injury models have found that inhibition of NETosis as well as disintegration of NETs by DNase I treatment was highly protective against ventilator-induced lung injury (VILI) (35). NETs were also observed in both patients and a murine model of primary graft dysfunction, with DNase I treatment again reducing lung injury (36), suggesting enhanced NETosis and the presence of NETs to play a pathologic role in lung injury. However, another study using conventional knockout mice for Pad4 in the pristane model found that constitutive deletion of Pad4 results in higher levels of anti-nuclear antibodies (ANAs) and inflammatory mediators (37). However, in this study mice lung inflammation was not assessed and more importantly, mice were sacrificed approximately 4 months post pristane injection, whereas DAH and lung inflammation in these mice occurs day 10-14 and resolves after approximately 28 days. Also in our short-term study, we employed myeloid specific deletion of Pad4, in contrast to the aforementioned study, as conventional knockout mice for Pad4 display enhanced expression of hematopoietic multipotent progenitor numbers which may affect biology and response to agonists (38). Thus, although the potential exists that NETosis may indeed be important in helping to resolve inflammation in murine lupus models as the disease progresses, our data and data of other suggests that restricting NETosis prevents pristane-induced lung inflammation.

Type I IFNs are known to be central to the pathology of pristane-driven autoimmune disease and lung inflammation in C57BL/6 mice, with pristane-driven autoimmune disease absent in IFNAR knockout mice (27). Importantly, NETs have been shown to be important for induction of ISG expression in murine models of SLE and ex vivo analysis of SLE patient neutrophils (6, 9). The presence of dsDNA in the NETs activates DNA sensors in both immune and non-immune cells triggering IFN induction and subsequent ISG expression. Our results show that prevention of NETosis using LysMCrePad4^fl/fl mice results in reduction of IFN-stimulated gene expression in the lungs of pristane-treated mice. NETs have also been shown to trigger ER stress in gut epithelial cells in a model of septic shock (14). In this study, ER stress activation by NETs in intestinal epithelial cells was mediated by TLR9, suggesting that DNA in the NETs was directly triggering ER stress in these cells. Our study shows that neutrophils can directly induce ER stress in epithelial cells, with Pad4-deleted neutrophils unable to drive ER stress in coculture experiments. Interestingly, in the aforementioned study, TLR9-driven ER stress increased gut barrier permeability, potentially suggesting a mechanism by which ER stress is contributing to lung inflammation and damage in the pristane model. ER stress has been shown to contribute to lung inflammation through modulation of the NFκB/HIF-1α pathways (20). ER stress is also known to drive hepatic, endothelial and epithelial cell apoptosis via reactive oxygen-dependent pathways and contribute to disease pathology in various models (39–41). Because IRE1-α activity has been shown in several models to be critical in regulating pristane-induced autoimmunity as well as pulmonary disease (21–25), we focused on sXbp1 driven ER stress in the lung after pristane administration and in keeping with previous studies we found that IRE1-α-dependent pathways to be involved. However, it should be noted that we did not fully explore ATF4 and ATF6 in this study and that a potential role for these pathways cannot be ruled out. It is also important to note that ER stress has also been shown to regulate production of, and sensing of, type I IFNs, resulting in exacerbated IFN-driven responses (42). The IRE1/XBP1 arm of the unfolded protein response has been largely shown to augment type I IFN production via activation of additional stress-responsive proteins such as protein kinase R (PKR), STING and IRF3, or directly by spliced XBP1 binding to the promoter of Ifnb1 (43–46). In keeping with this, we have found that treatment of alveolar epithelial cells not only drives induction of ER stress markers but also induces the expression of IFN-stimulated genes directly. Thus, although the specific mechanism is currently under investigation, our data suggests that ER stress can directly drive IFNβ production and that NET-induced ER stress may be contributing to IFN induction in the pristane model of SLE.

In summary, the current study shows that ER stress markers are upregulated in the lungs of pristane-treated mice. Targeted deletion of the enzyme PAD4 in myeloid cells reduced lung inflammation and markers of ER stress and IFN-driven responses. Neutrophils isolated from lupus patients and pristane-treated mice could directly upregulate ER stress markers in alveolar epithelial cells and murine lung slices, respectively. Collectively, our data supports role for NETosis and ER stress in driving autoimmune lung inflammation.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Cedars-Sinai Institutional Review Board. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Cedars-Sinai Medical Center Institutional Animal Care and Use Committee.

GT, DEL and JC conceived and carried out experiments. GS, LAand EM contributed to sample preparation. YW and RM conducted analysis and helped with preparation of manuscript. MI, SV, LF, DW recruited patients for the study. BS assisted with design of experiments. CJ took the lead in experimental design, project supervision, and manuscript preparation. All authors contributed to the article and approved the submitted version.

The following funding bodies supported this work: Selkowitz Family Foundation (CJ, BRS); Arthritis Foundation award AF2017-433570 (CJ) and the Leon Fine Translational Award, CSMC (CJ).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.