Microvascular lung injury and endoplasmic reticulum stress in systemic lupus erythematosus-associated alveolar hemorrhage and pulmonary vasculitis

Abstract

Human COPA mutations affecting retrograde Golgi-to-endoplasmic reticulum (ER) protein transport cause diffuse alveolar hemorrhage (DAH) and ER stress (“COPA syndrome”). Patients with SLE also can develop DAH. C57BL/6 (B6) mice with pristane-induced lupus develop monocyte-dependent DAH indistinguishable from human DAH, whereas BALB/c mice are resistant. We examined Copa and ER stress in pristane-induced lupus. Copa expression, ER stress, vascular injury, and apoptosis were assessed in mice and COPA was quantified in blood from patients with SLE. Copa mRNA and protein expression were impaired in B6 mice with pristane-induced DAH, but not in pristane-treated BALB/c mice. An ER stress response (increased Hsp5a/BiP, Ddit3/CHOP, Eif2a, and spliced Xbp1) was seen in lungs from pristane-treated B6, but not BALB/c, mice. Resistance of BALB/c mice to DAH was overcome by treating them with low-dose thapsigargin plus pristane. CB6F1 mice did not develop DAH or ER stress, suggesting that susceptibility was recessive. Increased pulmonary expression of von Willebrand factor (Vwf), a marker of endothelial injury, and the chemokine Ccl2 in DAH suggested that pristane promotes lung microvascular injury and monocyte recruitment. Consistent with that possibility, lung endothelial cells and infiltrating bone marrow-derived cells from pristane-treated B6 mice expressed BiP and showed evidence of apoptosis (annexin-V and activated caspase-3 staining). COPA expression also was low in patients with SLE with lung involvement. Pristane-induced DAH may be initiated by endothelial injury, resulting in ER stress, apoptosis of lung endothelial cells, and recruitment of myeloid cells that propagate lung injury. The pathogenesis of DAH in SLE and COPA syndrome may overlap.

INTRODUCTION

Proteins and lipids are transported bidirectionally between the endoplasmic reticulum (ER) and Golgi (1). Newly synthesized secretory proteins exit the ribosome in an unfolded state and fold after translocation across the ER membrane. Additional folding occurs following coatomer coat protein complex I (COPII)-mediated (anterograde) transport to the Golgi (2). Conversely, proteins can be transported in a retrograde direction from the Golgi back to the ER by the coatomer coat protein complex I (COPI) composed of α-COP, β-COP, β’-COP, γ-COP, δ-COP, ε-COP, and ζ-COP (2). COPI is required for Golgi-to-ER transport of dilysine-tagged proteins (3, 4). The coatomer complexes COPII and COPI coat vesicles on the Golgi membrane involved in anterograde or retrograde transport of proteins, respectively, and are homologous to clathrin adapter proteins. Retrograde and anterograde transport are linked, as disruption of COPI-mediated transport with brefeldin A collapses COPII-mediated transport (1). Stringent quality control of protein folding is necessary to maintain ER homeostasis. Misfolding induces ER stress, activating the stress sensors IRE1, ATF6α, and PERK, which trigger the unfolded protein response (UPR) (5, 6). IRE1 promotes degradation of misfolded proteins and PERK arrests translation, whereas in the event of prolonged ER stress ATF6 and CHOP (DDIT3) activate apoptosis (7). See Table 1 for a list of gene symbols and corresponding proteins.

Table 1.

Gene symbols and their corresponding proteins

Protein	Gene
BiP (binding immunoglobulin protein)	Hsp5a (heat shock protein family A member 5)
CCL2 (C-C motif chemokine ligand 2)	Ccl2
CHOP (CCAAT/enhancer-binding protein homologous protein)	Ddit3 (DNA damage inducible transcript 3)
COPA (coatomer coat protein complex α protein)	Copa (COPI coat complex subunit α)
EIF-2A (eukaryotic translation initiation factor 2 A)	Eif2a
KLF4 (Kruppel-like factor 4)	Klf4
MxA (myxovirus resistance protein 1)	Mx1 (MX dynamin like GTPase 1)
STING (Stimulator of interferon response CGAMP interactor 1)	Sting1
VWF (Von Willebrand factor)	Vwf
XBP1 (X-box binding protein 1)	Xbp1

Gene names: Hsp5a (heat shock protein family a member 5, encoding BiP protein); Ccl2 (C-C motif chemokine ligand 2, encoding CCL2 protein); Ddit3 (DNA damage inducible transcript 3, encoding CHOP protein); copa (COPI coat complex subunit α, encoding COPA protein); Eif2a (eukaryotic translation initiation factor 2 A, encoding EIF-2A protein); Klf4 (Kruppel-like factor 4, encoding KLF4 protein); Mx1 (MX dynamin like GTPase 1, encoding MxA protein); Sting1 (stimulator of interferon response CGAMP interactor 1, encoding STING protein); vwf (von Willebrand factor, encoding VWF protein); Xbp1 (X-box binding protein 1, encoding XBP1 protein).

In the resting state, ER stress sensors are maintained in an inactive state through interactions with BiP (HSPA5). The accumulation of misfolded proteins in the ER lumen releases BiP from PERK, IRE1α, and ATF6 activating the UPR (8, 9). When ER stress cannot be resolved, the UPR promotes apoptosis via the upregulation of CHOP. Abnormal lipid composition in the ER membrane also induces the UPR (10).

The Golgi is a second checkpoint for misfolded proteins, which may be targeted to lysosomes for degradation (6) or undergo COPI-mediated retrograde transport back to the ER. Accumulation of misfolded proteins in the Golgi induces a “Golgi stress” response, which may lead to sustained MEK1/2 and ERK1/2 activation and cell death (11, 12). Consistent with the link between anterograde and retrograde transport, mutations of the human COPA gene (encoding α-COP) induce ER stress and upregulate cytokines, resulting in autoantibody production, polyarthritis, and severe lung disease (interstitial lung disease and diffuse alveolar hemorrhage, DAH) (13, 14). This has been termed “COPA syndrome.”

DAH also develops in ∼3% of patients with systemic lupus erythematosus (SLE) and is fatal in over half (15, 16). Clinical features include hemoptysis, a strong association with lupus nephritis, anti-neutrophil cytoplasmic antibody (ANCA)-negative pulmonary capillaritis, and pulmonary hemosiderin-laden macrophages (16, 17). C57BL/6 (B6) mice with pristane-induced lupus develop lung hemorrhage strongly resembling DAH in patients with SLE (18). In contrast, BALB/c mice are highly resistant to pristane-induced DAH. The onset of DAH requires monocytes/macrophages (Mϕ), complement, and immunoglobulin, but not type I interferon, TNFα, or IL-1 (19). B6 mice lacking IL-10 have more severe disease.

We asked whether the pathogenesis of lupus-associated DAH and COPA syndrome overlaps. Pristane treatment decreased Copa expression and induced endothelial injury and ER stress in the lungs of DAH-susceptible B6 mice but not resistant BALB/c mice. However, when mice were treated with a low dose of the ER stress inducer thapsigargin plus pristane, BALB/c mice developed DAH. Susceptibility to pristane-induced lung injury and DAH was a recessive trait. COPA expression also was low in peripheral blood cells from a subset of patients with SLE. Thus, the mechanisms leading to DAH in human and murine lupus and COPA syndrome may overlap.

MATERIALS AND METHODS

Mice

Mice were maintained under specific pathogen-free conditions. Female 8- to 12‐wk‐old C57BL/B6 (B6), BALB/cJ, (BALB/cJ X C57BL/6J) F1 (CB6F1/J), B6.129S4-Ccr2^tm1Ifc/J, B6.129S2-Il6^tm1Kopf/J (IL6⁻), C57BL/6J-Sting1^gt/J (Sting^Gt), and B6.Cg-Il17a/Il17f^tm1.1lmpr Thy1^a/J (Il17af⁻) mice and male B6 and BALB/cJ mice of the same age were from The Jackson Laboratory. Unless otherwise noted, all mice were female (age 8–12 wk). To induce lupus, 0.5 mL of pristane (Sigma‐Aldrich) was administered intraperitoneally. Controls were left untreated or received mineral oil (Sigma-Aldrich; 0.5 mL ip). Experiments were repeated at least twice. Peritoneal exudate cells (PEC) were collected 14 days after pristane treatment. DAH was assessed as described (19). Lungs were fixed in 10% neutral buffered formalin and embedded in paraffin. This study followed the recommendations of the Animal Welfare Act and US Government Principles for the Utilization and Care of Vertebrate Animals and was approved by the University of Florida Institutional Animal Care and Use Committee.

Patients

RNA was isolated from blood collected in PAXgene tubes (BD Biosciences) from 54 patients with SLE meeting the 2019 Consensus Criteria (20), 9 primary Sjogren’s syndrome patients meeting the ACR-EULAR Sjogren’s syndrome Consensus Criteria (21), and 22 healthy controls (Supplemental Table S1; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.21320592.v1). Doses of prednisone, hydroxychloroquine, methotrexate, mofetil mycophenolate, and azathioprine on the day of blood collection were recorded. SLE patients with nephritis (n = 23) had urine protein ≥0.5 g/day (or >3+) and/or a renal biopsy showing WHO class III, IV, or V lupus nephritis. SLE patients with lung disease (n = 10) had either interstitial lung disease on imaging or a history of DAH. Disease activity was determined using the SLEDAI (22). Active disease was defined as a SLEDAI ≥4. This study was carried out in accordance with the recommendations of the International Committee of Medical Journal Editors and was approved by the UF IRB. All subjects gave written informed consent in accordance with the Declaration of Helsinki.

Thapsigargin and Tunicamycin Treatment

BALB/c mice, age 8–10 wk, were treated with pristane with or without thapsigargin (low-dose 0.25 mg/kg body wt, high-dose 1 mg/kg, in DMSO-PBS, every 3 days; Cayman Chemical) or tunicamycin (1 mg/kg body wt, in DMSO-PBS, every 3 days; Cayman Chemical). DAH was assessed at 14 days. Tunicamycin is a bacterial nucleoside that induces ER stress by inhibiting N-glycosylation, resulting in the accumulation of misfolded proteins in the ER (23). Thapsigargin inhibits SERCA-type Ca²⁺ pumps and induces ER stress by promoting Ca²⁺ efflux from the ER resulting in the accumulation of misfolded proteins in the ER (24, 25). However, these drugs exhibit dose-dependent cytotoxic effects that may be independent of the UPR, necessitating careful consideration of the doses used in our studies based on the prior literature. We selected a dose of thapsigargin (0.25 mg/kg) below the standard in vivo dose of 1 mg/kg, and corresponding to a dose that does not induce cytotoxicity in cell culture (25).

Quantitative Real-Time PCR

RNA was isolated from 10⁶ mouse PEC or lung tissue using TRIzol (Invitrogen). cDNA was synthesized using the Superscript II First-Strand Synthesis Kit (Invitrogen). SYBR Green quantitative real-time PCR (qPCR) was performed using a CFX96 thermocycler (Bio-Rad). Gene expression was normalized to 18S RNA, and expression was calculated using the 2^−ΔΔCt method. For clinical samples, values were normalized to a standard control sample assigned a value of 1. Primer sequences are listed in Supplemental Table S2.

Immunohistochemistry of Lung Tissue

Paraffin sections (4 µm) of formalin-fixed lung tissue obtained 14 days after pristane treatment were stained with H&E or Prussian blue. For immunohistochemistry, paraffin sections were dried on slides for 2-h at 60°C. Slides were placed in a Ventana Medical Systems immunostainer and deparaffinized. Heat‐induced epitope retrieval was performed with Ventana Medical Systems CC1 retrieval solution (30 min at 95°C–100°C). Unconjugated primary antibodies specific for Copa (LifeSpan BioSciences), BiP (Cell Signaling), F4/80 (Abcam), or cleaved (activated) caspase-3 (BD PharMingen) were applied for 32 min at 37°C, followed by peroxidase‐conjugated goat anti‐mouse or goat anti‐rabbit secondary antibodies (30 min; Supplemental Table S3). The reaction product was visualized using an UltraView DAB Detection Kit (Ventana Medical Systems). Slides were counterstained with hematoxylin (Ventana Medical Systems). Positive cells were evaluated by a lung pathologist (LL) and percentages of Copa⁺ and BiP⁺ cells in lung sections from B6 mice 14 days after pristane treatment were quantified using the NIH ImageJ software (26). Positive cells were quantified in three representative sections per mouse and the data were and expressed as positive cells per high power field (Prussian blue) or % of cells positive (BiP, Copa).

Oxygen Saturation and Heart Rate

Oxygen saturation and heart rate were measured with a MouseOx Plus System (Starr Life Sciences) on conscious mice. Mice were rested for 5 min before taking measurements and then monitored for at least 5 min.

Flow Cytometry of Isolated Lung Cells

Single-cell suspensions of lung tissue were isolated from B6 and BALB/c mice treated with/without pristane ± thapsigargin. Lungs were dissected and rinsed with cold PBS to remove blood. A single-cell suspension was produced by dissociating lung tissue (Lung Dissociation Kit, MiltenyiBiotech) followed by tissue disruption in a gentleMACS Dissociator (MiltenyiBiotech) according to the manufacturer’s protocol.

One million lung cells were surface stained with anti-CD45-allophycocyanin (APC) and anti-CD146-peridinin-chlorophyll-protein (PerCP) antibodies (Biolegend) for 20 min in the dark. Then the cells were fixed with 100 μL 4% formaldehyde for 15 min at room temperature and washed with PBS. Fixed cells were permeabilized with ice-cold 100% methanol for 10 min, resuspended in 100-μL PBS, and stained intracellularly for 20 min with phycoerythrin (PE)-conjugated rabbit anti-BiP monoclonal antibodies (C50B12, Cell Signaling, Danvers, MA). In other experiments, lung cells were surfaced stained with APC-conjugated anti-CD45 and PerCP-anti-CD146 antibody for 20 min and washed twice with cold PBS followed by 1× Annexin-V binding buffer (BD Bioscience, San Diego, CA), before resuspending in 100 μL of binding buffer. FITC-annexin-V antibodies (5 μL) were added to each sample and the cells were stained for 15 min at room temperature. Samples were analyzed within 1 h of staining. Monoclonal antibodies are listed in Supplemental Table S3. Staining was analyzed using a BD Canto II flow cytometer, gating on live cells.

Statistical Analysis

Data are representative of at least two independent experiments and are presented as means ± SD. For normally distributed data, comparisons were performed using Student’s unpaired two‐tailed t test (GraphPad Prism software, version 5). When data were not normally distributed, comparisons were analyzed by the Mann–Whitney U test. Frequency data were analyzed by Fisher’s exact test. Correlations were analyzed using Spearman’s rank correlation coefficient. P values less than 0.05 were considered significant.

RESULTS

Pristane Causes ER Stress and Suppresses Copa Expression in Lung

Because human COPA mutations are associated with DAH, we examined Copa mRNA in lungs from pristane-treated mice. Basal expression was similar in untreated B6 and BALB/c mice (Fig. 1A). However, expression was lower in pristane-treated versus untreated B6 mice, whereas pristane had little effect on Copa in BALB/c lung (Fig. 1A).

Figure 1.

Reduced Copa expression and ER stress in pristane-induced DAH. A–E: B6 and BALB/c mice were treated with pristane (Pris) or thapsigargin (THA) alone, with pristane + THA, or were left untreated (−). Gene expression in lung tissue was determined by qPCR at 14 days. A: Copa expression. B: Eif2a expression. C: Hsp5a (BiP) expression. D: Ddit3 (CHOP) expression. E: expression of unspliced (top) and spliced (bottom) Xbp1. F: BALB/c mice were treated with thapsigargin alone and Copa, Hspa5, and Ddit expression in lung were measured by qPCR in comparison with lung from untreated B6 and BALB/c controls (−). Each symbol on the graphs (A–F) corresponds to one mouse. G: male and female B6 and BALB/c mice (5–19 per group) were treated with pristane alone or with pristane + thapsigargin and the % of mice with DAH was determined. Differences between groups were evaluated by χ². H, left: survival of B6 and BALB/c mice treated with pristane, THA, or pristane + THA. Numbers of mice are indicated below the bars (G) and in the legend (H). *P < 0.05; **P < 0.01; ***P < 0.001 (Student’s t test). #P = 0.001 (B6-Pris vs. BALB/c-Pris), P = 0.09 (B6-Pris vs. BALB/c-Pris+THA), Mantel-Cox log rank test; ##P = 0.002 (BALB/c-Pris vs. BALB/c-Pris+THA, Mantel-Cox log rank test. DAH, diffuse alveolar hemorrhage; ER, endoplasmic reticulum.

In patients with COPA syndrome, increased BiP/HSPA5 protein/mRNA is seen in alveolar epithelial cells and alveolar macrophages and patients’ B lymphoblastoid cell lines exhibit higher levels of thapsigargin-induced HSPA5 than controls (13). In pristane-treated mice, three genes involved in the ER stress response, Eif2a, Hsp5a (BiP), and Ddit3 (CHOP), were upregulated in B6, but not BALB/c, lung (Fig. 1, B–D). Splicing of Xbp1, which is regulated by Ire1, increased in lungs from pristane-treated versus control B6 mice (Fig. 1E). In contrast, Copa expression was reduced dramatically in B6, but not BALB/c, lung by pristane treatment (Fig. 1A).

Although pristane alone did not induce ER stress in BALB/c lung, Eif2a, Hsp5a, and Ddit3 expression was induced to high levels in mice receiving pristane plus low-dose thapsigargin (Fig. 1, B–D). Treatment with thapsigargin alone minimally affected Copa and Ddit3 expression but upregulated Hspa5 (Fig. 1F). Upregulation of Hspa5, but not Ddit3, may reflect the role of BiP in relieving ER stress (9) versus the role of CHOP in activating apoptosis when stress cannot be resolved (7).

The percentages of mice developing DAH were similar in male versus female B6 and BALB/c mice treated with pristane and in male versus female BALB/c mice treated with pristane + thapsigargin (Fig. 1G). BALB/c mice (both male and female) treated with pristane alone did not develop DAH. Mortality at 14 days was 35% in B6 mice versus 0% in BALB/c mice treated with pristane alone or thapsigargin alone (Fig. 1H). Mortality in BALB/c mice treated with pristane + low-dose thapsigargin was comparable to that in pristane-treated B6 mice. Mortality was accelerated in BALB/c mice receiving a higher dose of thapsigargin along with pristane, but overall mortality was ∼35%, regardless of whether the mice received high- or low-dose thapsigargin (Fig. 1H).

Lung tissue from BALB/c mice receiving pristane + low-dose thapsigargin showed DAH and small vessel vasculitis, which were absent in BALB/c mice receiving pristane alone (Fig. 2, A and B). BALB/c mice treated with pristane + tunicamycin, another ER stress inducer, also developed DAH and vasculitis (Fig. 2C). Mice receiving low-dose thapsigargin (0.25 mg/kg) alone did not develop DAH or vasculitis (Fig. 2D). Lungs from B6 mice and BALB/c mice treated with pristane + low-dose thapsigargin or tunicamycin showed large Prussian blue⁺ (hemosiderin-laden) macrophages at 14 days, which were absent in lungs from low-dose thapsigargin-treated or untreated BALB/c mice (Fig. 2, E–I). At 14 days, Prussian blue staining was more intense in pristane-treated B6 lungs than in lungs from BALB/c mice treated with pristane + tunicamycin or pristane + thapsigargin (Fig. 2, G–I), suggesting that the onset of DAH was earlier in B6 mice. The numbers of Prussian blue⁺ cells per high power field at 14 days were higher in BALB/c mice treated with pristane + tunicamycin or pristane + thapsigargin than in untreated BALB/c controls (Fig. 2J). However, Prussian blue⁺ cells were more numerous in pristane-treated B6 mice. Thus, although BALB/c mice were not susceptible to pristane-induced DAH, their resistance could be at least partially overcome by combining pristane treatment with an ER stress inducer (tunicamycin or thapsigargin).

Figure 2.

Lung pathology and hypoxemia. BALB/c mice were treated with pristane with/without ER stress inducers thapsigargin (THA) or tunicamycin (tuni) or were left untreated. Lung tissue was harvested at 14 days and stained with hematoxylin & eosin (A–D) or Prussian blue (E–H). Yellow arrow, small vessel vasculitis in a mouse treated with pristane + THA. Black arrows, hemosiderin-laden macrophages. Sections are representative of lung tissue from the mice described in Fig. 1. I: representative Prussian blue staining of lung from a pristane-treated B6 mouse. J: quantification of Prussian blue positive cells (3 mice per condition) at 14 days in BALB/c and B6 mice treated with pristane alone, pristane+THA, or pristane+tuni, or left untreated (number of positive cells per high power field). K: oxygen saturation ($\text{S}{\text{p}}_{{\text{O}}_2}$) by pulse oximetry in B6 and BALB/c mice treated with pristane alone, pristane + THA, or left untreated (Control). L: heart rate in B6 and BALB/c mice treated with pristane alone, pristane + THA, or left untreated (Control). *P < 0.05, **P < 0.01, ****P < 0.0001 (Student’s t test). ER, endoplasmic reticulum.

Consistent with the possibility that mortality was due to respiratory insufficiency, the peripheral capillary oxygen saturation ($\text{S}{\text{p}}_{{\text{O}}_2}$) at 14 days was reduced in B6 mice treated with pristane and in BALB/c mice treated with pristane + thapsigargin, whereas BALB/c mice treated with pristane alone had a normal $\text{S}{\text{p}}_{{\text{O}}_2}$ (Fig. 2K). Heart rate increased in pristane-treated B6, but not pristane + thapsigargin-treated BALB/c, mice (Fig. 2L).

Immunohistochemistry of lung tissue from untreated B6 mice showed strong Copa staining in the alveolar walls and bronchi, whereas the endothelial cells of larger pulmonary blood vessels exhibited little or no staining (Fig. 3A). Following pristane treatment, the normal alveolar architecture was obscured, but in comparison with untreated controls, there was little Copa staining (Fig. 3B). The number of Copa⁺ cells was lower in lung from mice with pristane (day 14) than in lung from untreated mice (Fig. 3B, right).

Figure 3.

Immunohistochemistry of Copa and BiP in pristane-treated mice. B6 mice were treated with pristane or left untreated and lung was harvested at 14 days for immunoperoxidase staining (brown reaction product) using HRP-conjugated anti-Copa, anti-BiP, and anti-F4/80 antibodies. A: representative Copa staining (untreated lung, ×200). B: representative Copa staining (pristane-treated lung, ×200). Right: quantification of Copa⁺ cells in untreated (−) vs. pristane-treated (Pris-14d) mice, expressed as a percentage of positive cells (number of positive cells/total cells ×100) per high power field. C: BiP staining (untreated lung, ×200). D: representative BiP staining (pristane-treated lung, ×200). Arrow indicates an enlarged lipid-laden macrophage stained positively for BiP). Right: quantification of BiP⁺ cells in untreated (−) vs. pristane-treated (Pris-14d) mice, expressed as a percentage of positive cells (number of positive cells/total cells ×100) per high power field. E: representative F4/80 staining (brown reaction product, untreated lung, ×400). Arrow indicates alveolar macrophages. F: BiP staining at higher magnification (pristane-treated lung, ×400). Arrow indicates an enlarged, lipid-laden macrophage stained positively for BiP. Each symbol on the graphs (B and D) corresponds to one mouse. **P < 0.01; ****P < 0.0001, Student’s t test. Alv, alveolus; Br, bronchus; BV, blood vessel.

The alveolar walls of untreated B6 mice were stained weakly by anti-BiP antibodies (Fig. 3C), but there was little staining 14 days after pristane treatment (Fig. 3D, left and right). However, the cytoplasm of large, lipid-laden Mϕ infiltrating the lungs of pristane-treated mice was stained by anti-BiP antibodies (Fig. 3D, arrow). At higher power, the lipid-laden Mϕ appeared considerably larger than F4/80⁺ alveolar Mϕ from untreated mice (Fig. 3, E and F). The abundant BiP⁺ lipid-laden Mϕ may represent bone marrow-derived myeloid cells recruited to the lung rather than resident alveolar Mϕ.

Pristane Causes Microvascular Lung Injury

Pristane can be detected in the lung and bone marrow by mass spectrometry after intraperitoneal injection (19) and induces apoptosis of lymphoid cells (27). We investigated whether the ER stress response originated from intrinsic lung cells or infiltrating bone marrow-derived myeloid cells using B6-Ccr2^−/− mice, in which bone marrow-derived monocytes are not recruited normally to sites of inflammation (28). Lipid-laden Mϕ were absent in pristane-treated B6-Ccr2^−/− mice, alveolar architecture was normal, and there was no DAH or vasculitis (Fig. 4A), consistent with prior observations (29). As expected, Ddit3 expression was higher in pristane-treated B6 versus BALB/c mice (Fig. 4B). Ddit3 expression also increased in pristane-treated B6-Ccr2^−/− mice, indicating that although there was little infiltration by myeloid cells and no DAH or vasculitis in Ccr2^−/− mice, pristane induced an ER stress response in the lung. Ccl2 (encoding MCP-1, a Ccr2 ligand) was induced by pristane in B6, but not BALB/c, mice (Fig. 4C). Ccl2 expression increased in pristane-treated B6 Ccr2-deficient mice (Fig. 4C), suggesting it was produced by intrinsic lung cells rather than infiltrating myeloid cells. Interestingly, in contrast to wild-type B6 mice, pristane did not downregulate Copa in Ccr2^−/− mice (Fig. 4D). As Copa was detected mainly in epithelial cells (Fig. 3A), this suggested that pristane may be more toxic to endothelial cells than epithelial cells. Consistent with that possibility, expression of Vwf, a specific marker of endothelial cell injury (30), was substantially higher in lungs from pristane-treated B6 versus BALB/c mice (Fig. 4E). As expected, Vwf transcripts were not detected in PECs.

Figure 4.

Lung microvascular injury in pristane-treated mice. A: representative H&E staining of lung from a pristane-treated B6 Ccr2^−/− mouse. Left: normal alveolar architecture without evidence of DAH or vasculitis (×200). Right: absence of infiltrating lipid-laden macrophages (×400). After 14 days, RNA was extracted from lung tissue from pristane-treated (Pris) and untreated (−) B6, BALB/c, and Ccr2-deficient B6 (Ccr2^−/−) mice. Expression of Ddit3 (B), Ccl2 (C), and Copa (D) was determined by qPCR. E: B6 and BALB/c mice were injected with pristane and Vwf mRNA was quantified from lung tissue and peritoneal exudate cells (PEC) at 14 days by qPCR. Each symbol on the graphs (B-E) corresponds to one mouse. *P < 0.05; **P < 0.01; ***P < 0.001, ****P < 0.0001 (Student’s t test). DAH, diffuse alveolar hemorrhage.

BiP expression also was examined in single-cell suspensions of lung cells (Fig. 5). BiP (Hspa5), also known as 78-kDa glucose-regulated protein (GRP78), is an abundant ER chaperone that promotes protein folding and is a key regulator of the UPR (31). In the presence of misfolded proteins in the ER lumen, BiP is released from PERK, IRE1α, and ATF6, activating the UPR (8, 9). BiP expression increases in response to ER stress (32, 33) and can be used to monitor ER stress (31, 33, 34). We examined intracellular BiP staining in CD45⁺ (bone marrow-derived) and CD146⁺ (endothelial) cells from the lung by flow cytometry.

Figure 5.

BiP expression and apoptosis in lung cells. Lungs from untreated and 14 days’ pristane- or pristine + thapsigargin (thap)-treated mice were disrupted using a gentleMACS Dissociator. Single-cell suspensions were stained with anti-BiP, CD45, CD146, and annexin-V antibodies and analyzed by flow cytometry. A: B6 mice were treated with/without pristane and lung cells were stained with anti-CD45 and anti-BiP antibodies. Left: representative flow cytometry data. Right: percentages of CD45⁻Bip⁺ cells and CD45⁺BiP⁺ cells. B: B6 mice were treated with/without pristane and lung cells were stained with anti-CD146 and anti-BiP antibodies. Left: representative flow cytometry data. Right: percentages of CD146⁺Bip⁺ cells. C: BALB/c mice were treated with/without pristane+thapsigargin or pristane alone and stained with anti-CD45 and anti-BiP antibodies. Left: representative flow cytometry data. Right: percentages of CD45⁻Bip⁺ cells. D: B6 mice were treated with/without pristane and lung cells were stained with anti-annexin-V and anti-CD45 antibodies. Left: representative flow cytometry data. Right: percentages of CD45⁻annexin-V⁺ cells and CD45⁺annexin-V⁺ cells. E: B6 mice were treated with/without pristane and lung cells were stained with anti-annexin-V and anti-CD146 antibodies. Left: representative flow cytometry data. Right: percentages of CD146⁺annexin-V⁺ cells. F: B6 mice were treated with pristane (right) or left untreated (left). Representative immunoperoxidase staining of lung tissue using anti-activated caspase-3 monoclonal antibodies is shown. Peroxidase⁺ cells along the alveolar walls are stained brown (red arrows). Each symbol on the graphs (A–E) corresponds to one mouse. *P < 0.05; **P < 0.01; ***P < 0.001 (Student’s t test).

Pristane treatment increased the percentages of BiP⁺ cells in both the CD45⁻ (10%–20% positive) and the CD45⁺ (3%–6% positive) populations (Fig. 5A). The percentage of CD146⁺ cells expressing BiP was higher in pristane-treated versus untreated B6 mice (Fig. 5B), indicating that pristane induced an ER stress response in endothelial cells. In pristane-treated BALB/c mice, few CD45⁻ or CD45⁺ cells were BiP⁺ (Fig. 5C). However, when low-dose thapsigargin was administered along with pristane, 10%–20% of CD45⁻ cells were BiP⁺ (Fig. 5C). There were still very few CD45⁺BiP⁺ cells Although Ddit mRNA expression was strongly induced by pristane treatment (Fig. 1D), flow cytometry with anti-CHOP antibodies was unsuccessful due to the low expression level of CHOP protein (not shown). Since CHOP is linked to the activation of apoptosis pathways, lung cells were stained with anti-annexin-V antibodies. As shown in Fig. 5D, the percentages of both CD45⁻annexin-V⁺ and CD45⁺annexin-V⁺ cells were higher in pristane-treated B6 mice versus untreated controls. In addition, the percentage of CD146⁺annexin-V⁺ endothelial cells was higher in pristane-treated versus untreated mice (Fig. 5E). CHOP can activate both mitochondrial (intrinsic) and death receptor-dependent (extrinsic) apoptosis pathways leading to caspase-3 activation (35–39). Immunohistochemistry of lung from pristane-treated B6 mice using a monoclonal antibody specific for the activated form of caspase-3 revealed cells entering into the caspase-3-dependent apoptosis “execution pathway” (Fig. 5F). Activated caspase-3⁺ cells were undetectable in lung from untreated mice. Thus, pristane damaged the lung, increased expression of the monocyte-attractive chemokine Ccl2, Vwf (a specific marker of endothelial injury), and BiP protein, ultimately leading to apoptotic cell death. Cells exhibiting evidence of ER stress and apoptosis included CD146⁺ endothelial cells as well as a subset of bone marrow-derived (CD45⁺) cells, most likely monocytes recruited to the lung by Ccl2 or other chemokines. As DAH induction requires monocytes/Mϕ, immunoglobulin, complement, and Ccr2/Ccl2, pristane-induced DAH and immune complex-mediated small vessel vasculitis may be initiated by endothelial injury followed by the recruitment of circulating monocytes, ER stress, and cell death.

B6 X BALB/c (CB6F1) Mice Do Not Develop DAH

Humans with heterozygous point mutations affecting the conserved 14-amino acid WD40 domain of COPA develop progressive interstitial lung disease with DAH, polyarthritis, a type I interferon signature, autoantibodies, and ER stress (13). The inheritance of COPA syndrome is autosomal dominant with incomplete penetrance. For unclear reasons, about one-third of COPA mutation carriers are unaffected clinically. We asked whether CB6F1 mice develop pristane-induced DAH. Consistent with prior observations (40), the susceptibility of B6 (8/10) and BALB/c (0/10) mice to pristane-induced DAH was markedly different (Fig. 6A). None of the 10 CB6F1 mice developed DAH, suggesting that susceptibility may be a recessive trait. Copa expression was lower in lungs from pristane-treated B6 mice versus CB6F1, but was not significantly different in PECs (Fig. 6B). Ddit3 transcripts exhibited the reverse pattern: high levels in lung from susceptible B6 mice and low in resistant CB6F1 (Fig. 6C). Expression of Klf4, a regulator of IL10 production (41), resembled Copa expression: lower in B6 versus CB6F1 lung and not significantly different, although highly expressed, in PECs from both strains (Fig. 6D). Lung pathology in pristane-treated CB6F1 and BALB/c was similar, with normal alveolar structure and no evidence of alveolar hemorrhage on H&E or Prussian blue staining (not shown).

Figure 6.

DAH is abolished in CB6F1 mice and independent of IL-17 and Sting. A–D: genetic susceptibility. B6, BALB/c, and CB6F1 mice were injected intraperitoneally with pristane and 14 days later, lungs were analyzed for the presence/absence of DAH by histological staining and gene expression by qPCR. A: prevalence of DAH. B–D: expression of Copa, Ddit3, and Klf4 mRNA, respectively, in lung and peritoneal exudate cells (PEC). E–H: role of IL17 and Sting in DAH. B6, B6 Il6^−/− (Il6⁻), B6 Il17a/f^−/− (Il17af⁻), and Sting^gt mice were injected intraperitoneally with pristane and 14 days later, lungs were analyzed for the presence/absence of DAH by gross pathology and histology and gene expression was determined by qPCR. E: prevalence of DAH. F–H: expression of Copa, Ddit3, and Klf4 mRNA, respectively, in lung tissue. Each symbol on the graphs (B–D, F–H) corresponds to one mouse. Numbers of mice are indicated for each group in bar graphs (A and E) *P < 0.05; **P < 0.01 (Student’s t test). DAH, diffuse alveolar hemorrhage.

Role of Cytokines

TLR7-driven type I interferon promotes autoantibody production and lupus nephritis in pristane-treated mice, but Ifnar^−/−, Tlr7^−/−, MyD88^−/−, Ticam^−/− (TRIF deficient), Tnfa^−/−, and Casp1^−/− mice all develop DAH (19, 40), suggesting that lung disease is independent of TLR, IFNAR signaling, and inflammasomes. DAH is more severe in Il10-deficient mice than wild-type controls, suggesting that IL10 is protective (19). IL17 (13) and the STING pathway (42–44) have been suggested as inflammatory mediators in COPA syndrome, though the evidence is indirect. We examined the role of Klf4 (regulates IL10), Il17a/Il17f, Il6 (regulates IL17), and Sting (regulates interferon) in pristane-induced DAH. Sting^gt (Goldenticket) mice have a missense mutation (I199N) rendering them deficient in Sting-dependent interferon responses to cyclic dinucleotides or Listeria infection (45). When treated with pristane, they developed DAH at a frequency similar to controls (Fig. 6E). Copa, Ddit3, and Klf4 expression levels were similar in the lung from Sting^gt and wild-type mice (Fig. 6, F–H).

The upregulation of ER stress markers and IL17-priming cytokines (IL1β, IL6, and IL23) along with increased numbers of IL17A-producing T_H17 cells in COPA syndrome patients suggest that IL17 is involved in lung disease (13). However, the frequency of DAH in pristane-treated IL6⁻ and IL17a/f⁻ mice was comparable to that of controls (Fig. 6E). Copa and Ddit3 expressions also were similar in pristane-treated IL6⁻ and IL17a/f⁻ mice versus controls (Fig. 6, F–H). Klf4 expression was lower in IL17a/f⁻ mice than in controls, but it was not associated with increased incidence (Fig. 6E) or severity (not shown) of DAH. Moreover, Tcrb/Tcrd mice (lacking αβ and γδ T-cell receptors and therefore T_H17 cells) are susceptible (40). Overall, the data suggest that neither Sting nor IL17 plays a critical role in the pathogenesis of pristane-induced DAH, whereas IL-10 is protective.

COPA Expression in Human PBMCs

Lung tissue from patients with SLE is not readily available, but many cell types express COPA, including human B lymphoblastoid lines (13). COPA mRNA was lower in blood cells from SLE patients with lung disease or nephritis versus healthy controls and a similar trend was seen in SLE patients without lung or kidney disease and SS patients (Fig. 7A). As expected, expression of the interferon-stimulated gene MX1 was higher in patients versus controls, opposite the pattern of Copa (Fig. 7B).

Figure 7.

Low COPA in SLE peripheral blood cells. Peripheral blood was collected in PAXgene tubes from patients with SLE with interstitial lung disease (SLE/ILD), lupus nephritis (SLE/Neph) or SLE without either ILD or nephritis (SLE), or from patients with primary Sjogren’s syndrome (Sjog) or healthy controls (HC). Expression of COPA (A) and MX1 (B) relative to 18S rRNA was measured by qPCR. C: comparison of COPA expression in patients with active (SLEDAI ≥ 4) or inactive (SLEDAI < 4). D: comparison of COPA expression in patients with SLE taking (+) or not taking (−) methotrexate (MTX), mofetil mycophenolate (MMF), hydroxychloroquine (HCQ), azathioprine (AZA), or prednisone (PRD). E: correlation of COPA expression with absolute lymphocyte and neutrophil counts and with lymphocyte percentages in patients with SLE. Each symbol on the graphs (A–C, E) corresponds to one patient. *P < 0.05; **P < 0.01; ***P < 0.001 (Student’s t test). SLE, systemic lupus erythematosus.

COPA expression was not significantly different in SLE patients with active (SLEDAI ≥ 4) versus inactive (SLEDAI < 4) disease (Fig. 7C) and did not correlate with the erythrocyte sedimentation rate (not shown). COPA expression also was not different in patients treated with/without methotrexate, mycophenolate mofetil, hydroxychloroquine, azathioprine, or prednisone (Fig. 7D).

RNA isolated from human blood cells collected in PAXgene tubes is derived from myeloid cells (mainly neutrophils and monocytes) plus lymphocytes. In patients with SLE, COPA mRNA negatively correlated with the percentage of lymphocytes and with the absolute lymphocyte count (Fig. 7E). However, COPA expression was lower in patients with SLE versus controls (Fig. 7A) and so were lymphocyte counts (mean 1,290 versus 2,130 per mL in SLE versus controls, P < 0.0001, Student’s t test). Absolute neutrophil counts were not significantly different (3,550 versus 4,100 per mL in SLE vs. controls). Thus, low COPA expression in patients with SLE is unlikely to be due to differences in the lymphocyte or neutrophil counts.

DISCUSSION

Patients with heterozygous loss of function mutations affecting the WD40 domain (residues 230–241) of COPA develop polyarthritis, severe pulmonary inflammation (interstitial lung disease and/or DAH), autoantibodies, IL17 and interferon production, and an ER stress response (“COPA syndrome”) (13, 14, 42–44). We found that B6 mice with pristane-induced DAH also had low Copa expression and developed ER stress, which was absent in DAH-resistant BALB/c mice. An ER stress response characterized by increased expression of Hsp5a/BiP, Ddit3/CHOP, Eif2a, and spliced Xbp1 was evident in lung tissue from pristane-treated B6 mice with DAH, but not in BALB/c mice. Consistent with lung microvascular endothelial injury, Vwf expression increased in pristane-treated B6, but not BALB/c, mice. CD146⁺ lung endothelial cells and infiltrating CD45⁺ (bone marrow-derived) cells from pristane-treated mice exhibited high levels of BiP protein expression and evidence of apoptosis (annexin-V and activated caspase-3 staining). The data suggest that DAH in pristane-induced lupus is initiated by endothelial injury, resulting in an ER stress response, apoptosis, and recruitment of myeloid cells that may further propagate injury and inflammation.

Susceptibility/Resistance to DAH Induction Is Strain Specific

Pristane-treated B6-BALB/c hybrids (CB6F1) did not exhibit an ER stress response or abnormal Copa expression and did not develop DAH. The resistance of BALB/c mice to DAH was overcome by treating them with pristane plus low doses of an ER stress inducer (thapsigargin or tunicamycin). The explanation for susceptibility (B6 mice) versus resistance (BALB/c mice) remains to be elucidated, but it is unlikely to reflect baseline Copa expression since polymorphic variants in B6 versus BALB/c are not reported. Basal Copa expression was similar in B6 and BALB/c (Fig. 1A), and in contrast to the dominant inheritance of human COPA syndrome (13), susceptibility to pristane-induced DAH was abolished in CB6F1 mice (Fig. 6A). Differential susceptibility of B6 versus BALB/c could reflect differences affecting either the target tissue (lung) or infiltrating inflammatory cells.

Role of Bone Marrow-Derived Cells

B6 is a prototypical “T_H1” mouse and BALB/c is a “T_H2 mouse” (46, 47). However, T-cell-deficient mice are susceptible to pristane-induced DAH (40), suggesting that differences in T-cell function are not major contributors to susceptibility. Pristane-induced DAH is prevented by depleting monocytes/Mϕ with clodronate liposomes and also is mitigated by IL-10 (19). Thus, increased expression of the monocyte-attractive chemokine Ccl2 in lungs from pristane-treated B6, but not BALB/c, mice (Fig. 4C) may be significant for disease pathogenesis. Monocyte egress from the bone marrow and recruitment to sites of inflammation are severely impaired in Ccr2-deficient mice (28, 48), and pristane does not induce DAH in B6 Ccr2^−/− mice (29). The high level of Ccl2 mRNA in the lung from B6 Ccr2^−/− mice (Fig. 4C) suggests it is produced by lung endothelial or epithelial cells rather than by infiltrating monocytes/Mϕ. Ccl2 is expressed constitutively by endothelial cells and is induced in lung microvascular endothelial cells by TNFα or IFNγ (49). Thus, like Vwf, Ccl2 expression may reflect the severity of lung endothelial injury.

Although the proinflammatory cytokines IL17 and IFNα have been implicated in the pathogenesis of COPA syndrome (13, 14, 42–44), neither IL17A/F nor the T_H17 cell-promoting cytokine IL6 had a major effect on susceptibility to pristane-induced DAH (Fig. 6). Similarly, although human COPA mutations promote ligand-independent STING activation, resulting in type I interferon overproduction (42, 43), the frequency of pristane-induced DAH in Sting^gt mice was similar to controls (Fig. 6A). Like Sting^gt mice, Ifnar1^−/− mice are highly susceptible to pristane-induced DAH (19). Interestingly, the TLR-mediated inflammatory response to pristane is exacerbated by STING deficiency, suggesting that STING has a negative regulatory role in pristane-induced lupus (50). Thus, although IL17 and interferon are overproduced in both COPA syndrome and pristane-induced lupus, they do not seem to play a central role in the pathogenesis of DAH.

Consistent with the importance of myeloid cells in DAH, B6 and BALB/c exhibit differences in M1/M1 Mϕ polarization. B6 mice are more susceptible than BALB/c to atherosclerosis (51), a monocyte/Mϕ driven disease (52), and granuloma formation is altered in B6 versus BALB/c mice (53). In addition, the regulation of eicosanoids and PPARγ differs in Mϕ from B6 versus BALB/c (54) and there are interstrain differences in iron metabolism that influence Mϕ polarization and the inflammatory response (55). Further studies will be necessary to determine whether these strain-specific differences in myeloid function affect DAH.

Susceptibility to Lung Injury

Susceptibility of the alveolar wall to pristane-induced injury also may help determine susceptibility versus resistance. Lung injury in mice is likely a toxic effect of pristane, which by mass spectrometry localizes to the lung after intraperitoneal injection, most likely via lymphatics, (19). The alveolar wall consists of type I and II alveolar epithelial cells and endothelial cells separated by a basement membrane rich in type IV collagen, laminins, and other proteins (56). Type I alveolar epithelial cells undergo extensive cell death during certain infections (e.g., influenza, COVID), resulting in acute lung injury/acute respiratory distress syndrome (ALI/ARDS), characterized by the influx of protein-rich fluid into the alveolar spaces, but not hemorrhage (56).

Endothelial injury, as indicated by increased BiP staining and apoptosis of CD146⁺ cells in pristane-treated B6 mice (Fig. 5), may be critical to the pathogenesis of DAH and pulmonary capillaritis/small vessel vasculitis (16, 18). Increased Vwf expression, a marker of endothelial injury (57), following pristane treatment of B6, but not BALB/c, mice (Fig. 4E), supports the idea that pristane injures the endothelium. VWF is synthesized only by endothelial cells and megakaryocytes and is stored in endothelial cell Weibel-Palade bodies and α-granules of platelets (58). It is released into the circulation after lung injury and high levels are associated with severe endothelial damage in COVID-19 infection and other forms of acute lung injury (57). Vwf transcripts also increase in lung tissue in response to microvascular lung injury in Coxsackievirus A2 infection (30). The increased Vwf in the lung from pristane-treated B6 versus BALB/c mice is consistent with the possibility that differential susceptibility to pristane-induced lung injury is involved in susceptibility to DAH.

Similarity of Pristane-Induced DAH to COPA Syndrome

The cytoplasmic domains of many ER-resident proteins contain di-lysine or -arginine sorting motifs recognized by the WD40 domains of α-COP (COPA) and β’-COP (59). The α-COP, β’-COP, and ε-COP subunits of COPI form a clathrin-like subcomplex that mediates retrograde transport of proteins and lipids from the Golgi to the ER, preventing cell surface expression of dysfunctional proteins (4, 60). COPI mutations impair the retrieval of these proteins causing leakage of ER proteins into the Golgi and other downstream compartments of the secretory pathway (13). Thus, like the ER, the Golgi is a checkpoint for misfolded proteins. A Golgi stress response is mediated by the ETS transcription factors ELK1, ETS1, and GABPA/B, resulting in MEK1/2 activation and in some cases cell death (12, 61). Our data raise the possibility that Golgi stress due to COPA deficiency acts synergistically with ER stress in pristane-induced DAH. But the nature of the interactions between Golgi retrograde protein transport and ER stress is not well understood. In view of the low Copa expression and activation of the UPR/ER stress, our data suggest that mice with pristane-induced DAH may have an acquired COPA-like syndrome caused by strain-specific susceptibility to lung microvascular injury leading to ER stress and cell death.

Pristane-Induced ER Stress Is Associated with Cell Death

Chronic activation of the UPR in endothelial cells leads to cell death (62, 63). DAH was associated with an ER stress response at both the mRNA and protein level (Figs. 1, 3, and 5) and enhancing the pristane-induced stimulus for ER stress response with low-dose thapsigargin led to the induction of DAH in “resistant” BALB/c mice (Fig. 1, G–H). The pathology of DAH in human and murine lupus is similar. Although humans are not exposed to pristane, most SLE patients with DAH show evidence of pulmonary infection. Respiratory viruses can activate ER stress pathways in the lung (64–67), suggesting that viral infection might take the place of pristane in human lupus-associated DAH. In influenza virus infection, ER stress is due to abnormal glycosylation of the hemagglutinin protein, resulting in IRE1α and PERK phosphorylation and activation of the MAP kinase-JNK pathway in lung epithelial cells (64) and apoptosis of lung epithelial cells (68). In both pristane-induced DAH and influenza-infected mice, lung damage is greatly attenuated in mice with defective Ccl2-Ccr2 signaling (29, 64), underscoring the importance of infiltrating myeloid cells. As in pristane-treated mice, COVID-19 infection increases BiP expression in the lung (66). Coronavirus infection of endothelial cells causes ER stress, inflammation, myelopoiesis, and activation of apoptosis, ferroptosis, and inflammasome-driven pyroptosis (65). ER stress and apoptosis of endothelial cells also contribute to acute lung injury in sepsis. BiP expression increases dramatically in the lungs from mice treated with LPS and inactivation of BiP reduces endothelial inflammation and attenuates LPS-induced acute lung injury (34), which is mediated, in part, by caspase-dependent apoptosis of endothelial cells (69).

Along with increased BiP expression, apoptotic cells staining for activated caspase-3 increased lungs from pristane-treated B6 mice with DAH (Fig. 5). Analysis of single-cell suspensions of the lung revealed apoptotic (annexin-V⁺) endothelial cells and bone marrow-derived cells, most likely infiltrating myeloid cells (Fig. 5). Unresolved ER stress leads to cell death via the CHOP, PERK/ATF4, and IRE1α/Bcl2 pathways (70–72). CHOP (Ddit3) expression increased significantly in pristane-treated B6 and pristane + thapsigargin-treated BALB/c mice (Fig. 1F). However, intracellular staining of lung cells with anti-CHOP antibodies was not sufficiently sensitive to detect CHOP protein by flow cytometry (data not shown).

BiP⁺CD45⁺ cells, most likely infiltrating monocytes/Mϕ, increased in the lungs from pristane-treated B6 mice, which also contained foam cells (Fig. 3, D and F). ER stress promotes foam cell development from myeloid cells, resulting in inflammatory cytokine production and apoptosis (73). This is central to the pathogenesis of atherosclerotic plaques. The clearance of apoptotic cells is impaired in peritoneal Mϕ from pristane-treated mice, leading to secondary necrosis and inflammation (74). Thus, in some respects, the inflammatory response in B6 mice with DAH resembles inflammation in atherosclerotic lesions.

Clinical Implications

About 3% of patients with SLE develop DAH with ANCA-negative small vessel vasculitis (arteriolitis, venulitis, and capillaritis) and half of these die (15, 16). However, autopsy studies demonstrate focal collections of erythrocytes in 30%–66% of patients with SLE, suggesting that subclinical disease is common (75). The presence of immune complexes in a granular distribution within the lungs distinguishes lupus DAH from Wegener’s granulomatosis or microscopic polyangiitis (pauci-immune) and Goodpasture’s syndrome (linear immunoglobulin deposits) (76). Lungs from mice with pristane-induced DAH and small vessel vasculitis have virtually identical pathological changes to those in human SLE (18, 40). Pristane-induced DAH requires immunoglobulin and complement, consistent with an immune complex pathogenesis (19). Thus, there may be similarities in the underlying disease mechanisms in mice and humans. It is interesting that peripheral blood cells from SLE patients with interstitial lung disease or lupus nephritis expressed lower levels of COPA than controls (Fig. 7), although further studies are needed to learn whether low COPA expression predisposes a subset of patients with SLE to DAH or subclinical hemorrhage or is secondary to lung damage. Lung infections are reported in many SLE patients with DAH (16, 77), suggesting that they may trigger the disease in a susceptible host. There is considerable evidence that pulmonary viral infections can induce ER stress (64, 78, 79) and we hypothesize that they could play a role in human DAH analogous to that of thapsigargin in BALB/c mice.

The role of ER stress uncovered here may have implications for the therapy of DAH, which at present is not highly effective. Medications that modulate the UPR have been developed (78), including small molecule activators of ATF6 (80, 81). These agents might be useful for preventing DAH in patients with SLE.

DATA AVAILABILITY

Data will be made available upon reasonable request.

SUPPLEMENTAL DATA

Supplemental Tables S1–S3: https://doi.org/10.6084/m9.figshare.21320592.v1.

GRANTS

This work was supported by National Institutes of Health (NIH) (National Institute of Arthritis and Musculoskeletal and Skin Diseases) Grant number R01-AR44731 (W.H.R.) and by Department of Medicine funding (H.Z.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

H.Z., S.H., and W.H.R. conceived and designed research; H.Z., E.H., S.H., R.D.A. and W.H. performed experiments; H.Z., E.H., S.H., R.D.A., W.H., L.L., and W.H.R. analyzed data; H.Z., E.H., S.H., L.L., and W.H.R. interpreted results of experiments; H.Z., E.H., S.H., R.D.A., W.H., L.L., and W.H.R. prepared figures; H.Z., E.H., S.H., R.D.A., and W.H.R. drafted manuscript; H.Z., S.H., and W.H.R. edited and revised manuscript; H.Z., E.H., S.H., L.L., and W.H.R. approved final version of manuscript.