Two subsets of circulating Ly6Clo monocytes distinguished by CD138 (syndecan-1) expression and Nr4a1 dependence in pristane-treated mice

Chronic peritoneal inflammation following pristane injection induces lupus with diffuse alveolar hemorrhage (DAH) and pulmonary capillaritis in C57BL/6 mice. The pathogenesis involves pristane-induced microvascular lung injury. BALB/c mice are resistant to endothelial injury and DAH. Lung disease in C57BL/6 mice is abolished by depleting monocytes/macrophages. The objective of this study was to define the role of myeloid subsets in DAH. Hemorrhage and vasculitis were abolished in Ccr2-/- mice, indicating involvement of bone marrow-derived monocytes/macrophages. Along with Ly6Chi monocytes, we found two subsets of circulating Ly6Clo monocytes: one CD138- and a novel CD138+ subset. Nr4a1-dependent patrolling Ly6Clo monocytes maintain vascular integrity after endothelial injury. Circulating Ly6CloCD138+ monocytes were associated with DAH and were absent in mice without DAH. They also were absent in Nr4a1-/- mice, whereas Ly6CloCD138- monocytes were unaffected. However, Nr4a1-/- mice were susceptible to pristane-induced DAH and lung vasculitis, suggesting that disease onset does not require Ly6CloCD138- monocytes. Peritoneal Ly6CloCD138+ Mϕ were unchanged in Nr4a1-/- mice, indicating that they are not derived from Ly6CloCD138+ monocytes. We conclude that pristane-induced lung microvascular lung injury stimulates a wave of Nr4h1-dependent Ly6CloCD138+ patrolling monocytes in an ineffectual effort to maintain vascular integrity in the face of ongoing endothelial damage.

Pristane or mineral oil (MO) injection causes sterile peritoneal inflammation and an influx of Ly6C hi monocytes, which become Ly6C hi inflammatory peritoneal Mϕ (6). The chronic inflammatory response evolves into lupus after several months (7). In C57BL/6 (B6) mice, but not other strains, the onset of lupus is heralded by lung microvascular injury culminating in diffuse alveolar hemorrhage (DAH) (8) (H Zhuang, Submitted).
In contrast to the striking predominance of peritoneal Ly6C hi Mϕ in pristane-treated mice, Ly6C hi Mϕ are replaced by Ly6C lo Mϕ in MO-treated mice and lupus does not ensue. A subset of the Ly6C lo peritoneal Mϕ expresses CD138 (syndecan-1) (9). Ly6C lo CD138 + Mϕ from MOtreated mice have an anti-inflammatory phenotype and are highly phagocytic for dead cells. The significance of CD138 expression by Mϕ is unclear and it is not known whether it is expressed by other myeloid cells, such as monocytes.
Non-classical (Ly6C lo ) monocyte development requires the transcription factor nuclear receptor subfamily 4 group A member 1 (Nr4a1) and Notch signaling (2,10). Although both Nr4a1-independent Ly6C hi monocytes and Nr4a1-dependent Ly6C lo monocytes are derived from Ly6C + precursors, they have different functions. In contrast to the role of Ly6C hi monocytes as mediators of inflammation, Nr4a1-dependent monocytes patrol the vascular endothelium and promote the TLR7-dependent removal of damaged cells (5,11). However, they also can migrate across the endothelium into inflamed tissues (12,13).
We examined myeloid cell CD138 expression in the blood and inflamed peritoneum.
Ly6C lo CD138 + and Ly6C lo CD138 -Mϕ both were Nr4a1-independent and Ccr2-dependent, suggesting that they are derived from Ly6C hi monocytes recruited to the peritoneum.
Unexpectedly, circulating Ly6C lo monocytes also consisted of two subpopulations: CD138 + and CD138 -. Only the Ly6C lo CD138 + monocyte subset was Nr4a1-dependent. Compared with the Ly6C lo CD138and Ly6C hi subsets, Ly6C lo CD138 + monocytes expressed higher levels of Triggering Receptor Expressed on Myeloid Cells Like 4 (Treml4), an innate immune receptor that amplifies TLR7 signaling and binds to late apoptotic and necrotic cells (14)(15)(16)(17). Treml4 expression was regulated by Nr4a1, consistent with a role in the Ly6C lo monocyte-mediated, TLR7-dependent, removal of damaged endothelial cells. Circulating Nr4a1-dependent Ly6C lo CD138 + monocytes increased substantially after pristane, but not MO, treatment, and were more abundant in B6 vs. BALB/c mice, suggesting that they may represent an ultimately ineffectual effort by the organism to repair pristane-induced chronic endothelial injury.

Results
Ly6C hi monocyte recruitment to the peritoneum following pristane or MO injection is Ccr2dependent (6). In the peritoneum of pristane-treated mice, Ly6C hi Mϕ are more abundant than CD138 + Mϕ, whereas CD138 + Mϕ are more abundant in MO-treated mice. We examined CD138 expression on circulating and peritoneal myeloid cells from mice treated with pristane or MO.
As classical monocytes are CD115 + Ly6C + CD43 lo whereas non-classical monocytes are CD115 + Ly6C -CD43 hi TremL4 + (5), we concluded that the circulating R1 cells were classical (Ly6C hi ) monocytes and the circulating R2 and R3 cells were subsets of non-classical monocytes. The circulating R2 subset was heterogeneous, with one population expressing high CD115, CD43, and TremL4 and a smaller population with low expression (Fig. 2F). R1 also was heterogeneous, with two distinct populations expressing high and low levels of TremL4.
Together the data suggest that peripheral blood of pristane-treated mice contains classical monocytes along with two subtypes of non-classical monocytes distinguished by the presence/absence of CD138 staining.
Circulating monocyte subsets varied with time after pristane treatment (Fig. 4A). Ly6C hi monocytes (R1) appeared first, about 5 days after pristane or MO treatment, while at the same time, the percentage of Ly6C lo CD138monocytes (R2) decreased. Circulating Ly6C lo CD138 + monocytes (R3) appeared in pristane-treated mice at around day-9 but only small numbers were seen in MO-treated mice (Fig. 4A).
In pristane-treated Ccr2-/-mice, circulating R1 and R3 cells were greatly reduced, but not absent (Fig. 4B). In contrast, the percentages of circulating R2 cells were similar in Ccr2-/-vs. wild-type mice. In PECs, R1 and R3 cells were absent and R2 cells were present at lower levels in Ccr2-/-mice vs. wild-type (Fig. 4B). Thus, the egress of R1 (Ly6C hi ) and R3 (Ly6C lo CD138 + ) monocytes from the BM was Ccr2-dependent, whereas the circulating R2 monocyte subset (Ly6C lo CD138 -) was unaffected by the absence of Ccr2. R1 and R3 monocytes appeared in the circulation with different kinetics. All three peritoneal Mϕ subsets were reduced in Ccr2-/-mice.

Pristane-induced DAH is unaffected by absence of Nr4a1. Induction of DAH by pristane is
abolished by eliminating monocytes and Mϕ with clodronate liposomes (8). Pristane-treated Nr4a1-/-mice developed DAH at a frequency comparable to wild-type controls (Fig. 5E), suggesting that circulating CD138 + (R3) monocytes are dispensable for the induction of DAH. In contrast, DAH was abolished in Ccr2-/-mice suggesting that, as in the inflamed peritoneum, migration of Ly6C hi monocytes from the BM to the lung is involved in the pathogenesis of DAH.
Consistent with that possibility, pristane and MO treatment both increased the fluorescence intensity of Nr4a1 as well as TremL4 in circulating CD11b + Ly6Gcells (Fig. 6E).
Since circulating Ly6C lo CD138 + monocytes were present in pristane-treated (develop DAH) but not MO-treated (do not develop DAH) B6 mice, we asked whether these Nr4a1-dependent cells are associated with susceptibility to pristane-induced DAH. Circulating Ly6C lo CD138 + monocytes were substantially lower in BALB/c mice (resistant to DAH induction) vs. B6 (DAHsusceptible) and there was a similar trend in pristane-treated B-cell-deficient (μMT) B6 mice, which also fail to develop DAH (8) (Fig. 6F). Intracellular Nr4a1 staining (flow cytometry) was lower in circulating Ly6C hi and Ly6C lo monocytes from pristane-treated BALB/c vs. B6 mice ( Fig. 6G). Thus, high circulating, Nr4a1-dependent, Ly6C lo CD138 + monocytes and a high level of Nr4a1 protein in blood monocytes both were associated with the development of DAH.
Finally, we examined whether inflammation could increase TremL4 expression in PECs.

Discussion
Pristane-induced lupus in B6 mice is complicated by severe lung disease closely resembling DAH with antineutrophil cytoplasmic antibody (ANCA)-negative pulmonary vasculitis that is seen in SLE patients (8,25). DAH and small vessel vasculitis are abolished by depleting monocytes/Mϕ with clodronate liposomes and are absent in mice lacking immunoglobulin or C3, suggesting that immune complexes are involved (8). Disease appears to result from pristaneinduced lung microvascular injury (H Zhuang, submitted). Here, we examined the role of myeloid cell subsets in the pathogenesis of pristane-induced DAH. Lung disease was abolished in Ccr2-/-mice, suggesting that it depends on the influx of circulating BM-derived monocytes rather than activation of resident alveolar Mϕ. Circulating monocytes in pristane-treated mice included classical Ly6C hi (inflammatory) monocytes and non-classical Ly6C lo monocytes. Two subsets of Ly6C lo monocytes were identified: CD138and CD138 + . Ly6C lo CD138 + monocytes were dependent on the transcription factor Nr4a1, whereas the Ly6C lo CD138subset was Nr4a1independent. Induction of DAH was associated with a wave of circulating Ly6C lo CD138 + monocytes 9-days after pristane treatment that was nearly absent in MO-treated B6 mice or pristane-treated BALB/c mice, neither of which develop DAH. However, Nr4a1-/-mice still developed DAH and vasculitis. In view of the important role of Nr4a1-dependent "patrolling" monocytes in monitoring vascular integrity (2), our data suggest that Ly6C lo CD138 + monocytes are generated in response to lung endothelial cell injury resulting from the influx of Ly6C hi Ccr2 + inflammatory monocytes.
Two subsets of Ly6C lo monocytes. Ly6C lo (non-classical) monocytes develop from BMderived Ly6C hi precursors and are thought to be a single cell population (26,27) that monitors the vascular endothelium and promotes the TLR7-dependent removal of damaged cells (11,12). Development of this lineage requires the transcription factors Nr4a1 (2) and C/EBPβ (27).

What is the role of CD138?
The function of CD138 (syndecan-1) in myeloid cells is uncertain. Although CD138 was expressed on a subset of Ly6C lo monocytes and Mϕ, Sdc1 is not among the genes differentially expressed in Ly6C hi vs. Ly6C lo monocytes from non-pristanetreated-mice (27). There are several plausible explanations. First, the anti-CD138 monoclonal antibody we used might cross-react with another target in myeloid cells. However, clone 281-2 is a standard antibody for flow cytometry of mouse CD138 that recognizes the extracellular domain of CD138 on plasma cells, epithelial cells, and early B cells (30). Although transcriptional profiling of Ly6C lo CD138 + monocytes was not done, flow-sorted CD138 + Mϕ express high CD138 is a heparan sulfate-containing membrane proteoglycan that binds extracellular matrix (ECM) proteins and promotes integrin activation, wound healing, cell adhesion/migration, endocytosis, and fibrosis (34). Its functions in normal and neoplastic epithelial, endothelial, and stromal cells have been studied extensively, but little is known about its role in myeloid cells.
The extracellular domain binds type IV collagen and laminins (34)(35)(36)(37), which are components of the basement membrane separating the alveolar capillary endothelium and the epithelium (38). In chronic inflammation, monocytes attach to the vascular endothelium, migrate to the subendothelial space, and reversibly attach to basement membrane laminin via integrins (39,40).
It will be of interest to investigate whether expression of CD138 by Ly6C lo monocytes promotes migration and/or attachment to the basement membranes of damaged blood vessels.

CD138 + monocytes are not precursors of peritoneal CD138 + Mϕ. Migration of Ly6C hi
monocytes to the peritoneum in pristane-treated mice requires Ccr2 (6). Consistent with a previous report (11), circulating Ly6C hi monocytes were reduced by about 50% in pristanetreated Ccr2-/-mice (Fig. 4B). However, in the peritoneum, the absence of Ccr2 eliminated both Ly6C hi and Ly6C lo CD138 + Mϕ (Fig. 4B). The inability of the remaining circulating Ly6C hi monocytes to enter the peritoneum suggests that both BM egress and migration to the peritoneum are Ccr2-dependent. Following monocyte/Mϕ depletion with clodronate liposomes, the peritoneum is re-populated by Ly6C hi monocytes, which can differentiate into Ly6C lo Mϕ (6).
The absence of both Ly6C hi and CD138 + Mϕ in the peritoneum of Ccr2-/-mice suggests that CD138 + Mϕ are derived from Ly6C hi monocyte precursors.
Our data suggest that Treml4 is one of the genes regulated by Nr4a1. Nr4a1 correlated with Treml4 mRNA and protein in pristane-treated mice and in TLR ligand-or pristane-treated RAW264.7 cells (Fig. 6-8). Conversely, Nr4a1 knockdown decreased Treml4 expression in RAW264.7 cells and Treml4 expression was lower in adherent Mϕ from Nr4a1-/-mice vs. controls (Fig. 8). Although Nr4a1 appears to transcriptionally regulate Treml4, additional factors are likely to be involved, in light of the expression of low levels of Treml4 in Ly6C lo CD138 + and Ly6C lo CD138monocytes and Mϕ from Nr4a1-/-mice (Fig. 6).
Relevant to the role of Ly6C lo patrolling monocytes in monitoring for endothelial damage, TremL4 binds late apoptotic and necrotic cells and sensitizes cells to TLR7 signaling by recruiting MyD88 to endosomes (14,15). Treml4-/-MRL/lpr mice have lower autoantibody and interferon-α production and improved survival vs. Treml4+/+ controls (15), suggesting that Treml4 plays a causal role in lupus nephritis by upregulating TLR7-driven interferon production.
There also is evidence that patrolling monocytes mediate glomerular inflammation in murine lupus nephritis (43).
Genetic polymorphisms that increase TREML4 mRNA expression in human peripheral blood cells are associated with the progression and extent of coronary atherosclerotic lesions (44). However, a causal relationship has not been established and most evidence suggests that patrolling monocytes are protective in atherosclerosis (20,45). It is possible that upregulation of TLR7 signaling by Treml4 has a dual effect, on one hand promoting vascular integrity by enhancing the removal of damaged endothelial cells, and on the other enhancing vascular inflammation if the endothelial cell damage cannot be resolved.
In summary, we identified two subsets of Ly6C lo non-classical monocytes in pristane-treated mice, one CD138 + and Nr4a1-dependent and the other CD138and Nr4a1-independent.
Although CD138 expression has not been reported previously on Ly6C lo monocytes, the Nr4a1 dependence of Ly6C lo CD138 + cells strongly suggests they are patrolling monocytes involved in monitoring vascular integrity. We conclude that pristane-induced lung microvascular lung injury stimulates the production of these cells in an ineffectual effort to maintain vascular integrity in the face of ongoing endothelial damage. It remains to be determined whether Ly6C lo CD138 + monocytes can be detrimental when vascular damage is unresolved.   Table S1.

RAW264.7 cell culture with pristane.
Pristane or mineral oil (1 ml) was added to PBS (9 ml) containing 100 mg/ml BSA in a 15-ml polypropylene tube and rotated for 48-h at 4°C. The surface layer of unincorporated hydrocarbon oil was aspirated at the end of the incubation. The amount of pristane incorporated using this method was calculated as described (8)  Mini Kit (Qiagen) and cDNA was synthesized using the Superscript II First-Strand Synthesis kit (Invitrogen). SYBR Green qPCR analysis was performed using the CFX Connect Real-Time system (Bio-Rad). Gene expression was normalized to 18S RNA and the expression level was calculated using the 2 -∆∆Ct method. Primer sequences were as above.
Peritoneal cell culture with TLR ligands. PECs were harvested from untreated wild-type B6 and Nr4a1-/-mice and allowed to adhere to plastic wells (6-well plates, 2 × 10 6 cells/well) for 1 hour in AIM V serum free medium (ThermoFisher Scientific, Waltham, MA). Non-adherent 20 cells were washed off with PBS and the adherent cells were cultured for 3-hrs in complete DMEM + 10% FBS medium containing LPS (1 µg/ml in PBS) or PBS alone. Nr4a1 and Treml4 expression was quantified by qPCR as above.
ns, not significant. ns, not significant.