Research ArticleAutoimmunity

PD-1H (VISTA)–mediated suppression of autoimmunity in systemic and cutaneous lupus erythematosus

See allHide authors and affiliations

Science Translational Medicine  11 Dec 2019:
Vol. 11, Issue 522, eaax1159
DOI: 10.1126/scitranslmed.aax1159

Supporting a suppressor for lupus treatment

Autoimmunity can result when the intricate checks and balances in the immune system are disrupted. Han et al. discovered a potential treatment target for lupus erythematosus: the inhibitory receptor PD-1H. PD-1H knockout mice on a BALB/c background developed cutaneous lesions and were more susceptible to pristane-induced autoimmunity. Treating lupus-prone MRL/lpr mice with an antibody that activated PD-1H reduced skin symptoms and some markers of autoimmunity. The antibody was shown to act on myeloid cells and T cells and could potentially restore immune balance in patients with lupus.

Abstract

Systemic lupus erythematosus (SLE) and discoid lupus erythematosus (DLE) of the skin are autoimmune diseases characterized by inappropriate immune responses against self-proteins; the key elements that determine disease pathogenesis and progression are largely unknown. Here, we show that mice lacking immune inhibitory receptor VISTA or programmed death-1 homolog (PD-1H KO) on a BALB/c background spontaneously develop cutaneous and systemic autoimmune diseases resembling human lupus. Cutaneous lupus lesions of PD-1H KO mice have clustering of plasmacytoid dendritic cells (pDCs) similar to human DLE. Using mass cytometry, we identified proinflammatory neutrophils as critical early immune infiltrating cells within cutaneous lupus lesions of PD-1H KO mice. We also found that PD-1H is highly expressed on immune cells in human SLE, DLE lesions, and cutaneous lesions of MRL/lpr mice. A PD-1H agonistic monoclonal antibody in MRL/lpr mice reduces cutaneous disease, autoantibodies, inflammatory cytokines, chemokines, and immune cell expansion. Furthermore, PD-1H on both T cells and myeloid cells including neutrophils and pDCs could transmit inhibitory signals, resulting in reduced activation and function, establishing PD-1H as an inhibitory receptor on T cells and myeloid cells. On the basis of these findings, we propose that PD-1H is a critical element in the pathogenesis and progression of lupus, and PD-1H activation could be effective for treatment of systemic and cutaneous lupus.

INTRODUCTION

Systemic lupus erythematosus (SLE) is a chronic disease characterized by progressive autoimmunity against multiple organs and tissues that is debilitating and potentially lethal (14). SLE is clinically heterogeneous based on molecular classifications (5, 6), and such heterogeneity is one reason why it has been difficult to treat SLE. In the past 60 years, only one drug [belimumab, an anti–BAFF (B cell activating factor) antibody] has been approved by the U.S. Food and Drug Administration (FDA) for SLE (7). Discoid lupus erythematosus (DLE) is the most common type of cutaneous lupus, which is chronic and disfiguring (8, 9); there are no current FDA-approved treatments for DLE.

Patients with SLE and DLE often have alterations in T cell signaling, proliferation, cytokine production, and immunoregulatory function resulting in dysregulated T cell proliferation, activation, and pathogenic autoantibody production (3, 1012). In some instances, patients who are newly diagnosed with SLE were found to produce anti–programmed cell death 1 (PD-1) autoantibodies that enhance T cell function (13). Furthermore, inflamed tissues of autoimmune diseases such as rheumatoid arthritis and SLE have increased expression of inhibitory molecules such as B7-H1 (PD-L1), likely in an attempt to restrain ongoing inflammation and tissue damage (12, 14, 15). Mice deficient in immune inhibitory molecules such as PD-1 develop a lupus-like autoimmune disease (16). In addition to T cells, myeloid cells have also been implicated in lupus pathogenesis, including neutrophils, which are thought to be critical for autoantigen recognition (1719), and plasmacytoid dendritic cells (pDCs), which produce type I interferons (IFNs) (20).

We previously identified programmed death-1 homolog (PD-1H) [also called VISTA (V domain Ig suppressor of T cell activation), VSIR (V-set immunoregulatory receptor), Dies1, DD1α, and Gi24] as a cell surface inhibitory molecule of the B7/CD28 gene family expressed on T cells and myeloid cells (21, 22). PD-1H can function as an inhibitory ligand on antigen-presenting cells (APCs) and regulate T cell responses through an unknown receptor (23, 24). In addition, PD-1H can also function as an inhibitory receptor on T cells (23). For example, an agonistic PD-1H monoclonal antibody (mAb) markedly regulates antigen-specific CD4+ T cell responses and protects mice from graft-versus-host disease (GVHD) (21) and experimental hepatitis (23). Mice deficient in PD-1H on a C57BL/6 background [B6 PD-1H knockout (KO)] are more susceptible to autoimmune induction such as experimental autoimmune encephalomyelitis (25) and systemic lupus when backcrossed to a lupus-prone strain (26, 27). In one study, aged female B6 PD-1H KO mice spontaneously develop a severe autoimmune and inflammatory disorder with ulcerative dermatitis and elevated antinuclear antibodies (ANA) resembling lupus (28). Together, these results demonstrate that PD-1H is involved in peripheral immune tolerance and negatively regulates T cell activation by an unknown mechanism.

Here, we examined mice deficient in PD-1H on a BALB/c genetic background (hereafter referred to as PD-1H KO), the canonical murine model of lupus, MRL/Faslpr (referred to as MRL/lpr) (29), and samples from patients with human SLE and DLE. Our data suggest that immune inhibitory receptors such as PD-1H can be stimulated for the treatment of lupus and, possibly, other chronic inflammatory autoimmune diseases.

RESULTS

PD-1H–deficient mice develop spontaneous autoimmunity resembling lupus

We previously demonstrated that aged B6 PD-1H KO mice had increased numbers of effector memory CD4+ T cells in the spleen and liver, but no clinically detectable autoimmune disease (23). Wang and colleagues analyzed a different B6 PD-1H KO strain (VISTAKO) and reported that aged mice also do not develop overt organ-specific autoimmune disease despite the presence of chronic inflammation in some tissues (lung, liver, and pancreas) (25). Together, these data indicate that B6 PD-1H KO mice do not exhibit spontaneous autoimmune diseases in the absence of other predisposing factors (25, 26). However, Yoon et al. developed a B6 PD-1H KO strain (DD1α−/−) that, in contrast, did develop overt autoimmune diseases (28). Aged female mice in this strain spontaneously develop dermatitis, glomerulonephritis, and elevated autoantibodies resembling lupus. One possible explanation for the distinct phenotypes of the B6 PD-1H KO mouse strains could be the different breeding environments or genetic variants of mouse strains and substrains known to affect autoimmunity. To develop a mouse strain with more consistent autoimmune phenotype, we backcrossed B6 PD-1H KO mice onto different genetic backgrounds including BALB/c and DBA/1. PD-1H KO on a BALB/c genetic background developed spontaneous cutaneous and systemic autoimmune diseases resembling lupus that are distinct from previous reports.

By 6 months of age, 36% of female PD-1H KO mice (24 of 66) developed cutaneous plaques associated with alopecia, erythema, and scaling, as well as periorbital swelling and edema (Fig. 1, A and B). In contrast, male PD-1H KO (1 of 20, 5%) and female BALB/c wild-type (WT) littermate controls (1 of 62, 1%) rarely developed cutaneous lesions. Histological analysis of skin lesions shows hallmark features of human discoid lupus including acanthosis, hyperkeratosis, follicular plugging, lymphocytic infiltrate, and mild interface dermatitis (Fig. 1C) also found in MRL/lpr mice (30). PD-1H KO mice have greater histological evidence of disease as determined by inflammatory cellular infiltrate, epidermal hyperplasia, and epidermal ulceration when compared with age- and gender-matched (female) BALB/c WT littermate controls (Fig. 1D and table S1). In addition to cutaneous inflammation, about 50% of PD-1H KO mice (10 of 18 mice by 10 months of age) developed spontaneous pericardial calcification (Fig. 1, E and F). In contrast, pericardial calcification in age- and gender-matched WT BALB/c mice is rare (2 of 16 mice by 10 months of age) (Fig. 1F). Although we do not understand the mechanism underlying pericardial injury, it is tempting to speculate that it is a sequela of the systemic autoimmunity observed in PD-1H KO mice. About 20% of patients with SLE develop pericardial injury as a consequence of lupus serositis (31). However, histological evaluation of hearts from younger PD-1H KO mice (age, 4 to 6 months) did not show any evidence of cardiac or pericardial inflammation. In contrast to other murine models of systemic lupus, we did not observe proteinuria or lupus nephritis in PD-1H KO mice (fig. S1), suggesting that there is organ-specific autoimmunity occurring in the absence of PD-1H (32).

Fig. 1 PD-1H KO mice on a BALB/c background develop spontaneous systemic and cutaneous autoimmunity resembling lupus.

(A) The appearance of representative female WT BALB/c mice and female BALB/c PD-1H KO mice with spontaneous periorbital and dorsal nape dermatitis. (B) Incidence of spontaneous cutaneous lupus in female WT mice (n = 62) and female PD-1H KO mice (n = 66) over time. (C) Representative histological images of healthy skin from WT mice and affected skin from PD-1H KO mice, MRL/lpr mice, and human DLE (original magnification, ×400) stained with hematoxylin and eosin (H&E). Scale bars, 20 μm. (D) Skin biopsy score based on inflammation, acanthosis, and ulceration in female WT (n = 20) and PD-1H KO mice (n = 20). (E) Representative images of spontaneous pericardial calcification observed in 10-month-old PD-1H KO mice in comparison with 10-month-old WT mice. H&E of the pericardium from WT and PD-1H KO mice. Von Kossa insert (original magnification ×100 for histology). Scale bar, 100 μm. (F) Quantification of pericardial calcification as evaluated by the percentage of the area of the heart covered by calcium for 5-month-old female WT (n = 15) and PD-1H KO mice (n = 20) and 10-month-old female WT (n = 16) and PD-1H KO mice (n = 18) (0, none; 1, <10%; 2, 10 to 30%; 3, >30%). (G) Serum concentrations of ANA from 3-month-old female WT (n = 11) and PD-1H KO mice (n = 12) and 12-month-old female WT (n = 5) and PD-1H KO mice (n = 8). (H) Serum concentrations of anti-dsDNA IgG from 6-month-old female WT (n = 19) and PD-1H KO mice (n = 21). PD-1H KO mice are subdivided into those mice with cutaneous lupus lesions (lesion+; n = 11) and without cutaneous lupus lesions (lesion−; n = 10). (I) Concentrations of IL-6 and MCP-1 in sera of 9-month-old female WT (n = 15) and PD-1H KO mice (n = 29). Log-rank (Mantel-Cox) shown in (B). Bar graphs with mean ± SEM are shown in (D) and (F), and box and whisker plots are shown in (G), (H), and (I) analyzed with unpaired Student’s t test. *P < 0.05, **P < 0.01, ****P < 0.0001; ns, nonsignificant.

In addition, female PD-1H KO mice develop increasing ANA with age that are higher than that of age-matched female WT mice (Fig. 1G). Anti–double-stranded DNA (dsDNA) autoantibodies are elevated predominately in PD-1H KO mice with cutaneous disease (Fig. 1H) and, thus, track with disease activity similar to a subset of patients with SLE (33). Proinflammatory cytokines such as interleukin-6 (IL-6) and MCP-1, but not other inflammatory cytokines, are elevated in the sera of PD-1H KO mice (Fig. 1I and fig. S2). Using flow cytometry, we found that PD-1H KO mice have similar number of splenocytes but increased number of lymph node cells (Fig. 2A). In addition, PD-1H KO mice show immune dysregulation in immune cell subsets thought to be critical for lupus development, including increased number of CD4+ T cells, natural killer (NK) cells, CD11b+ myeloid cells, pDCs, and T follicular helper cells (TFH) within secondary lymphoid organs (Fig. 2, B and C). In addition, T cells have an increased effector memory T cell phenotype (CD44hiCD62L) as compared with age- and gender-matched WT controls (Fig. 2D), similar to B6 PD-1H KO mice (23). In our previous study, we found that the PD-1H pathway specifically boosts generation of inducible regulatory T cells (iTregs) but not natural Tregs (nTregs) (34). Consistent with our previous results, we demonstrate that PD-1H KO mice with cutaneous lupus have decreased peripheral Tregs (Fig. 2E). These results indicate a dysregulated and overactive immune system in the setting of PD-1H loss.

Fig. 2 Global immune overactivation of PD-1H KO mice.

(A) Total cell counts of lymph nodes (LN; pooled cell suspension of bilateral axillary, bronchial, and inguinal lymph nodes) and spleens (SPL) of female 5-month-old WT (n = 10) and PD-1H KO mice (n = 11). (B) Total number of CD3+ T cells, CD4+ T cells, CD8+ T cells, NKs, CD11b+ cells, and pDCs (CD11c+B220+PDCA-1+) in pooled lymph nodes of female 5-month-old WT (n = 6) and PD-1H KO mice (n = 7). (C) Percentages of TFH cells (CD3+CD4+CXCR5+PD-1+Bcl6+) in lymph nodes and spleens of female 5-month-old WT (n = 2) and PD-1H KO mice (n = 4). (D) Effector memory T cell percentages in lymph nodes and spleen of female 5-month-old WT (n = 15) and PD-1H KO mice (n = 13). (E) Percentages of Tregs in lymph nodes and spleen of female 5-month-old WT (n = 15) and PD-1H KO mice (n = 13). Percentages of Tregs are also compared between PD-1H KO mice with cutaneous lupus (lesion+) (n = 5) and without cutaneous lupus (lesion−; n = 8). Data are presented as box and whisker plots of measurements obtained in duplicate or triplicate experiments. *P < 0.05, **P < 0.01 by unpaired Student’s t test or Mann-Whitney U test for nonnormalized data as determined by Kolmogorov-Smirnov test for normality for CD11b+ cells in (B). ns, nonsignificant.

PD-1H modulates type I IFN production

Type I IFNs are thought to contribute to cutaneous lupus (35). To test whether the PD-1H pathway can regulate type I IFN response in vivo, we administered the Toll-like receptor 7/8 (TLR7/8) agonist R848 (ssRNA) to WT and PD-1H KO mice and measured circulating IFN-α. We found that IFN-α was elevated in PD-1H KO mice after R848 administration when compared with WT control mice (fig. S3A). In addition, isolated pDCs from PD-1H KO mice produced greater amounts of IFN-α in response to R848 in vitro as compared with pDCs isolated from WT control mice (fig. S3B). This suggests that the PD-1H pathway negatively regulates type I IFN release in response to proinflammatory triggers. We next performed a gene expression array to profile the expression of genes involved in type I IFN responses. We found up-regulation of a subset of IFN-stimulated genes (ISGs) in cutaneous lupus lesions of PD-1H KO mice when compared with normal skin of WT mice (fig. S3C), including Ccl4, Ccl2, Il6, Il10, Cd69, Irf3, Timp1, Cd80, and Sh2d1a. A subset of these genes was expressed more abundantly in normal skin from PD-1H KO mice compared with normal skin from WT mice, including Irf3, Ccl2, and Cd70 (fig. S3, D and E). This suggests that there could be subclinical inflammation in normal-appearing PD-1H KO skin compared with WT skin and may contribute, in part, to cutaneous lupus development. However, there was no increased expression of type I IFNs Ifna or Ifnb, suggesting that type I IFNs may not be the major pathway contributing to cutaneous lupus development in PD-1H KO mice. Together, these data indicate that type I IFN production is modulated, in part, by PD-1H.

Cutaneous lupus lesions of PD-1H KO mice have pDC clustering, and neutrophils infiltrate the skin before clinically evident disease

To determine the components of immune cell infiltrates within cutaneous lupus lesions from PD-1H KO mice, we first used immunohistochemistry (IHC) and found abundant staining for CD3+ T cells and CD11b+ myeloid cells (Fig. 3, A and B). Cutaneous lupus lesions of PD-1H KO mice show clusters of PDCA-1+ (CD317) pDCs (Fig. 3, A and B), which is a key histologic feature in DLE that is not seen in other murine of models of cutaneous lupus, including lupus lesions of MRL/lpr mice (3638). Therefore, PD-1H KO mice may represent a spontaneous, autochthonous model of cutaneous lupus that more accurately reflects the histopathological findings of human cutaneous lupus. However, like other models of murine cutaneous lupus, lesions of PD-1H KO mice have more abundant myeloid cells than typically observed in patients with DLE. To investigate in greater detail, we performed mass cytometry to study immune cell composition in established cutaneous lupus lesions in PD-1H KO mice. Notably, neutrophils (CD45+CD11b+Ly6G+Ly6C+) were the most abundant immune cell subset within cutaneous lupus lesions in PD-1H KO mice, representing up to 50 to 80% of total CD45+ cells infiltrating the skin (Fig. 3, C and D and fig. S4A). In addition, the number of infiltrating CD4+ T cells and CD8+ T cells in PD-1H KO mice with cutaneous lupus was also increased as compared with PD-1H KO mice without cutaneous lupus lesions, suggesting that T cells are also important for ongoing inflammation within cutaneous lupus (fig. S4B). To examine the earliest infiltrating immune cells within cutaneous lupus lesions, we also performed mass cytometry analysis on normal-appearing skin from 4-month-old WT and PD-1H KO mice. Before the development of cutaneous lupus, neutrophils selectively infiltrate skin from PD-1H KO mice, indicating that neutrophils are both the most abundant and earliest infiltrating immune cells within cutaneous lupus (Fig. 3D and fig. S4A). Other immune cell subsets present within established cutaneous lupus lesions from 6-month-old PD-1H KO mice such as CD4+ and CD8+ T cells are not increased in normal-appearing skin from 4-month-old PD-1H KO mice when compared with WT controls (fig. S4B). This suggests an important role for neutrophils in the initiation and pathogenesis of cutaneous lupus within mice. Furthermore, neutrophils isolated from PD-1H KO mice produce more inflammatory cytokines after stimulation by lipopolysaccharide (LPS), including IL-6, TNFα (tumor necrosis factor–α), and MCP-1 (Fig. 3E). Thus, neutrophils that lack PD-1H are more proinflammatory and release cytokines that have been implicated in cutaneous lupus pathogenesis, including IL-6 (39). Together, PD-1H plays a critical role in regulating the innate immune landscape within cutaneous lupus of mice and restrains neutrophil inflammation.

Fig. 3 Neutrophils infiltrate cutaneous lupus lesions of PD-1H KO mice and are proinflammatory.

(A) Representative images of immunohistochemistry (IHC) for CD3 (T cells), CD11b (myeloid cells), and CD317 (pDCs) from cutaneous lupus lesions of female PD-1H KO mice and cutaneous lupus lesions of MRL/lpr mice compared with age- and gender-matched WT controls. Bottom panel showing representative images of IHC for CD3 (T cells), CD68 (macrophages), and CD123 (pDC) from human DLE skin (magnification, ×400). Scale bars, 20 μm. (B) Quantification of IHC staining for CD3 and CD11b (top) and CD317 (bottom) in WT mice (n = 5) and lesional skin from MRL/lpr (n = 9) and PD-1H KO mice (n = 5) [hpf (high-power field), ×400]. (C) Representative viSNE graphs from mass cytometry analysis of healthy-appearing skin (single eyelid skin) from 6-month-old female WT or PD-1H KO mice as compared with established cutaneous lupus from female PD-1H KO mice (periorbital eyelid skin). (D) Quantification of mass cytometry data showing neutrophil percentages of infiltrated total CD45+ cells infiltrating periorbital (eyelid) skin from female 4-month-old WT (n = 3) and PD-1H KO mice (n = 3) and from female 6-month-old WT (n = 5) and PD-1H KO mice (n = 8). PD-1H KO mice are subdivided into those mice with cutaneous lupus lesions (lesion+; n = 4) and without cutaneous lupus lesions (lesion−; n = 4) in the third column (labeled as KO). (E) Neutrophils were enriched from bone marrow and pooled from three different female WT or PD-1H KO mice and stimulated by LPS (100 ng/ml) or not (no. sti). ELISA for IL-6, TNFα, and MCP-1 concentrations in culture medium are shown. Box and whisker plots are shown and analyzed by one-way ANOVA with Tukey post hoc analysis (B and D). Mean ± SEM shown in (E) and analyzed by unpaired Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001.

PD-1H–deficient mice develop more severe pristane-induced inflammatory arthritis and lupus

Given the proinflammatory phenotype of PD-1H–deficient neutrophils, we next investigated the pristane-induced model of SLE, which is predominately dependent on inflammatory monocytes and neutrophils (40). Pristane (2, 6, 10, 14-tetramethylpentadecane, or TMPD) is a hydrocarbon oil that when injected intraperitoneally into mice results in autoantibody induction, inflammatory arthritis, and glomerulonephritis with an elevated type I IFN signature (40). WT and PD-1H KO mice do not develop inflammatory arthritis spontaneously. After a single pristane injection, PD-1H KO mice developed more rapid and severe arthritis than WT mice. Five months after pristane treatment, about 70% of PD-1H KO mice developed inflammatory arthritis (joint swelling, erythema) as compared with 30% of WT mice (Fig. 4, A and B). Histological analysis of joints shows a severe inflammatory infiltrate and synovitis, predominately composed of cells that morphologically resemble neutrophils, in PD-1H KO mice treated with pristane as compared with WT mice (Fig. 4C). Using a clinical scoring system for collagen-induced arthritis model (41), PD-1H KO mice develop more severe arthritis as compared with WT controls (Fig. 4D). Neutrophils (CD11b+Ly6G+Ly6C+) are thought to be a key proinflammatory immune cell that contributes to the inflammatory cascade after administration of pristane (42, 43). At steady state, there are very few neutrophils that can be detected within the peritoneal cavity of WT or PD-1H KO mice (Fig. 4E). However, 4 days after pristane injection, a substantial number of neutrophils were recruited into the peritoneal cavity with more neutrophils recruited into the peritoneal cavity of PD-1H KO mice than WT mice, suggesting that PD-1H restrains neutrophil recruitment to sites of inflammation (Fig. 4E). Pristane also induces systemic autoimmunity resembling lupus in addition to inflammatory arthritis. Pristane-treated PD-1H KO mice developed elevated ANA and anti-histone autoantibodies as compared with WT mice (Fig. 4F), but not anti-dsDNA or anti-Sm autoantibodies (fig. S5, A and B). There were no differences in proteinuria or kidney disease in pristane-treated PD-1H KO or WT mice as both PD-1H KO and WT mice developed similar glomerulonephritis with immune complex deposition (fig. S5, C and D). Pristane-treated PD-1H KO mice do not develop enhanced cutaneous disease beyond baseline incidence of cutaneous lupus observed in untreated PD-1H KO mice (fig. S5E). WT mice treated with pristane had a modest increase in cutaneous lupus development (3 of 19 mice, 15%) compared with untreated WT mice (1 of 62, 1%) (fig. S5E). Our findings suggest a possible role of PD-1H in restraining neutrophil recruitment for murine lupus pathogenesis.

Fig. 4 PD-1H KO mice develop more severe pristane-induced inflammatory arthritis and lupus.

(A) Incidence of arthritis in female PD-1H KO (n = 12) and WT mice (n = 10) treated with pristane and representative of two independent experiments. Log rank (Mantel-Cox) analysis is shown. (B) Representative images of clinical appearance of front and hind legs from WT and PD-1H KO mice 6 months after pristane administration. (C) Representative images of H&E-stained hind limb ankle joint from WT and PD-1H KO mice 6 months after pristane administration. (D) Clinical arthritis score from female PD-1H KO (n = 12) and WT mice (n = 10) treated with pristane as determined by extent of joint swelling and inflammation. (E) Quantification of peritoneal neutrophils isolated from WT and PD-1H KO mice without treatment (WT, n = 3; KO, n = 3) or 4 days after pristane administration (WT, n = 6; KO, n = 5). (F) Serum concentrations of ANA and anti-histone Igs from female WT (n = 10) or PD-1H KO (n = 11) mice 6 months after pristane administration. Box and whisker plots are shown (D to F) with *P < 0.05 and ***P < 0.001 by unpaired Student’s t test.

Up-regulation of PD-1H in patients with SLE and DLE

To determine whether PD-1H is expressed in human lupus, we first performed IHC for PD-1H on biopsies from patients with DLE. All cases of DLE (n = 21) showed more PD-1H+ cells compared with control skin without DLE (n = 20) (Fig. 5, A and B, and fig. S6). Compared with control skin, DLE contains more CD3+ T cells, CD4+ T cells, CD8+ T cells, CD20+ B cells, CD68+ macrophages, and CD123+ pDCs (Fig. 5B and fig. S6), corroborating previous findings (44). The most abundant infiltrating immune cells within DLE are CD3+ T cells (45). Lichen planus (LP), another chronic autoimmune interface dermatitis with similar numbers of T cells and myeloid cells infiltrating the skin, only showed a modest increase in the number of PD-1H–expressing immune cells as compared with DLE and control skin (Fig. 5B and fig. S6). Therefore, increased PD-1H was likely not a result of general inflammation. Another difference is that CD123 pDCs are increased in DLE relative to other cutaneous autoimmune diseases (37). We also observed a notably increased number of myeloperoxidase (MPO)–positive neutrophils in DLE as compared with LP and control skin, but this was a minor population compared with other infiltrating immune cells (Fig. 5B). To determine the identity of PD-1H+ cells within DLE lesions, we next performed multiplexed quantitative immunofluorescence (mQIF) (46, 47). With this approach, PD-1H expression was quantified by automated quantitative analysis (AQUA) image system within CD3-, CD8-, CD11b-, and CD68-expressing cells. When compared with control skin, mQIF identified a greater number of CD3-, CD8-, CD11b-, CD68-, and PD-1H–expressing cells in DLE lesions similar to our IHC findings (fig. S7). Furthermore, PD-1H was more highly expressed by infiltrating CD3+ T cells, CD8+ T cells, CD11b+ myeloid cells, and CD68+ macrophages within DLE as compared with controls (Fig. 5C). To extend these findings to SLE, we analyzed PD-1H mRNA (C10orf54) expression using the RNA-sequencing datasets (OmicSoft DiseaseLand) (48) from circulating immune cells from patients with SLE and healthy controls. We found that patients with SLE (n = 111) have higher PD-1H gene (PD1H) expression in circulating immune cells than healthy controls (n = 25) (Fig. 5D). Elevated transcripts of other immune inhibitory receptors have also been reported to be up-regulated in affected autoimmune tissues during disease progression such as PD-L1 (B7-H1) (12, 14, 15). These data suggest that PD-1H is expressed on both infiltrating immune cells of human cutaneous lupus and circulating immune cells of SLE and represents a potential therapeutic target.

Fig. 5 Up-regulation of PD-1H in patients with SLE and DLE.

(A) Representative images of H&E and IHC of CD3, CD4, CD8, CD123 (pDCs), and PD-1H on DLE. (B) Quantification of IHC staining from DLE (n = 21), lichen planus (LP) (n = 11), and control skin (n = 20) FFPE tissue as displayed by average number of positive cells per high-powered field (×400). (C) Quantification of PD-1H expression within infiltrating CD3+ T cells, CD8+ T cells, CD68+ macrophages, and CD11b+ myeloid cells in DLE (n = 10 for CD3 and n = 6 for all others) and control (ctrl) skin (n = 10 for CD3 and n = 5 for all others) by mQIF using AQUA analysis algorithm. (D) Violin plot of human PD-1H mRNA expression (PD1H) as determined by fragments per kilobase millions (FPKM) of circulating immune cells (whole blood) from patients with SLE (n = 111) and healthy controls (n = 25). Box and whisker plots are shown. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by unpaired Student’s t test or one-way ANOVA with Tukey post-hoc analysis (B). ns, nonsignificant.

An agonistic PD-1H antibody suppresses autoimmune lupus in MRL/lpr mice

Similar to DLE, cutaneous lupus-like lesions of MRL/lpr mice also express PD-1H (fig. S8A). Previously, we have used an agonistic PD-1H mAb (clone MH5A) to treat allogenic GVHD and experimentally induced autoimmunity (21, 23, 49, 50). We have previously demonstrated that MH5A does not deplete cells in vivo (50). To assess whether PD-1H is a potential target for therapy in lupus, we treated MRL/lpr mice with our agonistic anti–PD-1H mAb. Compared with mice treated with control immunoglobulin G (IgG), MRL/lpr mice treated with agonist MH5A (200 μg of intraperitoneal injection weekly from weeks 6 to 16) developed fewer cutaneous lesions with delayed onset (Fig. 6, A and B). In addition, MRL/lpr mice treated with MH5A had less severe cutaneous lupus as determined by alopecia, erythema, erosions, or ulcerations, and extent of involvement at week 20 (Fig. 6, B and C) as determined by a previous clinical scoring system (51). Furthermore, cutaneous lupus lesions in MRL/lpr mice treated with MH5A had less severe histological evidence of disease as determined by inflammatory cellular infiltrate, epidermal hyperplasia, and epidermal ulceration (Fig. 6D and fig. S8B), including CD11b myeloid cell infiltration of the skin (Fig. 6E). There was also evidence of dampened systemic lupus in mice treated with the agonistic PD-1H mAb. For example, there were reduced circulating autoantibodies including ANA, anti-dsDNA, and anti-Ro52 in sera of MRL/lpr mice treated with MH5A (Fig. 6F).

Fig. 6 A PD-1H agonist reduces autoimmune lupus in MRL/lpr mice.

(A) Incidence of cutaneous lupus in female MRL/lpr mice treated weekly with either 200 μg of control hamster IgG (n = 10) or agonist PD-1H MH5A (n = 10) from 6 to 16 weeks of age. Data are representative to two independent experiments. Log rank (Mantel-Cox) analysis. Subsequent panels displaying analysis of mice used in (A). (B) Representative clinical images of MRL/lpr mice treated with either MH5A or control IgG. (C) Quantification of cutaneous lupus severity based on the clinical scores of erythema, alopecia, erosions/crust, and lesion size of MRL/lpr mice treated with either MH5A or control IgG at 20 weeks. (D) Quantification of skin biopsy score based on histological analysis of inflammation, acanthosis, and ulceration from skin of MRL/lpr mice treated with either MH5A or control IgG. (E) Quantification of IHC of CD3, CD11b, and CD317 of skin from MRL/lpr mice treated with either MH5A or control IgG at 20 weeks. (F) ANA, ds-DNA (week 20), and anti-Ro52 (week 20) autoantibodies detected in sera of MRL/lpr mice treated with either MH5A or control IgG. (G) Quantification of T cells from lymph nodes of MRL/lpr mice treated with either MH5A or control IgG at 20 weeks as determined by multiparameter flow cytometry. (H) Quantification of pDCs (CD11c+CD317+), neutrophils (CD11b+Ly6C+Ly6G+), and TFH (CD4+CXCR5+PDhi) cells from lymph nodes of MRL/lpr mice treated with either MH5A or control IgG at 20 weeks as determined by flow cytometry. (I) Proinflammatory cytokines/chemokines detected in sera of MRL/lpr mice treated with either MH5A or control IgG at week 20 using cytokine multiplex array. Mean ± SEM is shown (C and D) or box and whisker plots (E to I). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by unpaired Student’s t test.

We next investigated the effects of the PD-1H agonist on the severe lymphoproliferative disease of MRL/lpr. MH5A treatment reduced lymphadenopathy but had no effect on splenomegaly (fig. S8C). Immunophenotyping by flow cytometry revealed fewer total CD3+ T cells including CD4+ T cells, CD8+ T cells, and the double-negative (CD3+CD4CD8B220+) T cells in MRL/lpr mice (Fig. 6G and fig. S8D). Furthermore, there was a reduction in CD11b+Ly6G+Ly6C+ neutrophils, CD11c+PDCA-1+B220+ pDCs, and CD4+CXCR5+PD1hi TFH cells in the lymph nodes of mice treated with MH5A (Fig. 6H and fig. S8E). The numbers of B cells, monocytes, and conventional DCs were unaffected by MH5A treatment (fig. S8, F and G). Proinflammatory cytokines and chemokines such as IFN-α, IL-1α, IL-2, MIP-2, and M-CSF (macrophage colony-stimulating factor) were also significantly reduced in MRL/lpr mice treated with MH5A as determined by multiplex cytokine array (Fig. 6I and fig. S8H).

To determine the effect of PD-1H stimulation on lupus nephritis in MRL/lpr mice, we serially collected urine from MRL/lpr mice treated with MH5A or control IgG. We found that there was less proteinuria in MRL/lpr mice treated with MH5A as compared with controls (fig. S9A). However, histological analysis of glomeruli revealed no difference in the lupus nephritis score in mice treated with either MH5A or IgG (fig. S9, B and C). Furthermore, kidneys from MRL/lpr mice treated with MH5A did not show significant differences in infiltrating CD3+ T cells and CD11b+ myeloid cells (fig. S9D) or immune complex deposition (fig. S9E), suggesting that PD-1H agonism did not therapeutically alter kidney disease development. Together, stimulation of PD-1H reduces organ-specific (e.g., cutaneous) and systemic lupus autoimmunity in the canonical murine model of lupus.

PD-1H agonism inhibits T cell receptor signaling in T cells as well as pDC and neutrophil function

We and others have previously reported that PD-1H is broadly expressed on hematopoietic cells including CD4+ T cells, CD8+ T cells, CD11c+ DCs, CD11b+ myeloid cells, and CD11b+Gr-1+ neutrophils, but not B cells (21, 22). We extend these findings to pDCs, demonstrating that PD-1H is also highly expressed on B220+PDCA1+ pDCs from WT mice (Fig. 7A). Previous studies have reported that PD-1H functions as an inhibitory ligand on myeloid cells and as an inhibitory receptor on T cells (2124). However, the mechanism of PD-1H immunosuppression as either ligand or receptor remains largely unknown. Previous reports have demonstrated that myeloid cell PD-1H functions as an inhibitory ligand for T cell activation by attenuating T cell receptor (TCR) activation and reducing proliferation and cytokine production (22, 52). Using a similar approach, we demonstrate that activated murine CD4+ T cells incubated with PD-1H-Fc fusion protein resulted in decreased activation of mTORC1 (phosphorylation of p70 S6 kinase), mTORC2 (phosphorylation of Akt kinase), and mitogen-activated protein kinase (MAPK; phosphorylation of Erk1/2) signaling pathways (fig. S10A). These data indicate that PD-1H could act as a ligand to suppress downstream signaling after TCR activation via down-regulating mammalian target of rapamycin (mTOR) and MAPK pathways.

Fig. 7 PD-1H stimulation attenuates T cell signaling and myeloid cell function.

(A) Splenocytes from WT (red) and PD-1H-KO (gray) BALB/c mice were stained with PD-1H mAb or isotype control (open histogram) and cell surface markers. PD-1H expression in T cell (CD3+), pDC (B220+PDCA1+), myeloid cell (CD11b+), and B cell (B220+PDCA1) gates are shown. (B) Lymph nodes were isolated and pooled from three different female WT mice and stimulated with immobilized anti-CD3 (3 or 10 μg/ml) and PD-1H agonist MH5A (10 μg/ml) or IgG control (10 μg/ml). Thirty minutes after stimulation, cells were fixed, and phospho-flow cytometry was performed (percentages of phospho-Erk1/2+ CD4+ T cells are shown). (C) Purified pDCs from female WT (n = 3) or MRL/lpr mice (n = 3) were incubated with immobilized agonist PD-1H mAb MH5A or control IgG (10 μg/ml) and stimulated with R848 (50 μg/ml), and ELISA of culture media for IFN-α was performed. (D and E) Purified neutrophils from WT mice (n = 3 to 6) (D) or MRL/lpr mice (n = 3 to 6) (E) were incubated with immobilized agonist PD-1H mAb MH5A or control IgG (10 μg/ml) and stimulated with LPS (200 μg/ml), and cytokine array of culture media for IL-1β, TNFα, and MCP-1 was performed. (F) The granulocyte, monocyte, and macrophage numbers in the peritoneal cavity of WT BALB/c mice 4 days after pristane treatment. Mice were given two doses of 200 μg of either control IgG (n = 3) or anti–PD-1H MH5A mAb (n = 3) on days 0 and 2 after pristane treatment. Mean ± SEM is shown. *P < 0.05, **P < 0.01 by unpaired Student’s t test.

To investigate whether PD-1H can also function as a receptor on T cells, CD4+ T cells were stimulated with anti-CD3 and incubated in the presence of either the PD-1H agonistic mAb MH5A or control IgG and analyzed by phospho-flow cytometry. Engagement of PD-1H on T cells by agonist MH5A reduced phosphorylated Erk1/2, indicating inhibition of TCR-activated MAPK signaling pathway (Fig. 7B). However, there was no significant decrease in phosphorylation of mTORC1 or mTORC2 signaling pathways in CD4+ T cells treated with MH5A (fig. S10B). The decrease in MAPK activation is dependent on PD-1H expression by T cells because there was no decrease in phospho-Erk1/2 in PD-1H–deficient T cells treated with MH5A (fig. S10C). These results suggest that as a ligand, PD-1H may work as a universal regulator of multiple downstream TCR cell signaling pathways on T cells in a manner analogous to PD-L1 (53), whereas as a receptor, PD-1H seems to more specifically regulate MAPK signaling.

Currently, it is unknown whether PD-1H could function as a receptor on myeloid cells. To test this, neutrophils and pDCs were purified and subsequently treated with MH5A or control IgG in the presence of TLR ligands. Isolated pDCs had decreased CD86 and major histocompatibility complex II (MHC II) up-regulation in response to LPS stimulation when treated with MH5A as compared with control IgG, indicating that PD-1H agonism dampens pDC activation (fig. S10D). Furthermore, isolated pDCs from WT or MRL/lpr mice produce less IFN-α in the presence of MH5A when stimulated with R848 (Fig. 7C). In contrast, pDCs from PD-1H KO mice produced equivalent amounts of IFN-α in the presence of MH5A (fig. S10E). Agonist PD-1H MH5A–treated neutrophils isolated from WT mice (Fig. 7D) or MRL/lpr (Fig. 7E) produced less proinflammatory cytokines IL-1β, TNFα, and MCP-1 in response to LPS, indicating that the PD-1H agonist reduces neutrophil function by restricting proinflammatory cytokine secretion. Neutrophils deficient in PD-1H were unaffected by PD-1H agonism (fig. S10F). Last, WT mice treated with the PD-1H agonist MH5A had fewer infiltrating neutrophils and monocytes in the setting of pristane-induced peritoneal inflammation than control IgG (Fig. 7F). Together, PD-1H functions as an inhibitory receptor on myeloid cells including neutrophils and pDCs in addition to T cells.

DISCUSSION

In this report, we show that loss of PD-1H leads to strain-specific spontaneous autoimmunity and clinically evident organ-specific autoimmune disease. Autoimmune-prone MRL/lpr mice have up-regulated PD-1H within cutaneous lupus lesions, and treatment by an agonistic PD-1H mAb results in reduction in cutaneous and systemic lupus, but not lupus nephritis. TCR signaling and myeloid cell function including neutrophils and pDCs are shown to be attenuated by PD-1H agonism. Last, patients with SLE and DLE have up-regulated PD-1H on circulating or infiltrating immune cells, respectively. Together, our results support that PD-1H may function to restrain autoimmune inflammation and could be therapeutically targeted in both cutaneous and systemic autoimmune lupus.

Murine models of cutaneous and systemic lupus often poorly recapitulate human disease or have limited overlap (30, 54, 55). A feature of cutaneous lupus lesions from PD-1H KO mice is pDC clustering, which is a hallmark of DLE (37). Perhaps, the loss of PD-1H on pDCs results in pDC overactivation, liberation of type I IFNs, and subsequent clustering in cutaneous lupus lesions of PD-1H KO mice. Recently, pDC inhibition in patients with cutaneous lupus showed some benefit, suggesting that pDCs may be an important target for novel therapeutic strategies in patients with cutaneous lupus (56). These findings suggest that PD-1H KO mice may be a useful mouse model to dissect complex pathogenic factors, which determine the initiation and progression of cutaneous lupus.

Since the initial description of PD-1H, more effort has focused on the role of PD-1H as an immune checkpoint in tumor progression than autoimmunity (57). Recent studies have demonstrated that PD-1H is expressed predominately on myeloid cells including macrophages and, to a lesser degree, on T cells within the tumor microenvironment of several cancers (47, 58, 59). Current clinical trials that block PD-1H for cancer immunotherapy (NCT02812875) are in progress. In contrast, only one study has demonstrated that PD-1H is expressed within human autoimmune tissues and showed no difference between inflamed synovial tissue from patients with rheumatoid arthritis and synovial lining from control tissue (60). The function of PD-1H on normal human synovium and in the context of rheumatoid arthritis remains unexplored. In the current study, we found that immune cells within DLE and SLE both express high amounts of PD-1H. In contrast to the tumor microenvironment, expression of PD-1H is high on T cells as well as myeloid cells within autoimmune tissues of DLE. Thus, there may be differential expression patterns of PD-1H within cancer subtypes as compared with autoimmune tissues and between distinct autoimmune diseases. Delineating which autoimmune diseases have high expression of PD-1H may help predict potential organ-specific immune-related adverse events to future PD-1H blockade for the treatment of cancer.

It is well established that immune checkpoints such as the PD-1 axis are up-regulated in autoimmune tissues, likely in an attempt to restrain ongoing inflammation (12, 14, 15, 61). Furthermore, genetic deletion (e.g., PD-1) (62) of immune checkpoints in mice results in autoimmunity that is analogous to our study with PD-1H. Targeting immune inhibitory receptors in autoimmune diseases currently exist as ligand-Fc or receptor-Fc fusion proteins such as CTLA-4 receptor-Fc fusion protein (abatacept), which is FDA approved for rheumatoid arthritis (63). PD-1H-Fc fusion protein (VISTA.COMP) therapeutically reduced disease severity in an experimental model of acute inflammatory hepatitis (64). There are fewer reports of agonistic antibodies being used to stimulate immune inhibitory receptors for therapy in preclinical autoimmune models (65). Although no current agonist antibodies are approved for autoimmunity, several are in development (66). For example, a PD-1 agonist is now in clinical trials for the treatment of psoriasis, an autoimmune skin disease with up-regulated PD-1 expression on pathogenic IL-17A–producing CD4+ T cells (NCT03337022) (67).

A unique feature of PD-1H is that it can function as both receptor and ligand. As a ligand, it can deliver potent immunosuppressive signals within T cells by inhibiting both proximal and downstream TCR signaling pathways. For example, in this study and others, PD-1H-Fc results in decreased phosphorylation of LAT, SLP76, PLCγ-1, Akt, and Erk1/2, which is similar signaling attenuation seen in PD-1 engagement with B7-H1 (PD-L1) (52). Recently, PD-1H-Ig was found to be superior in suppressing T cell cytokine release (IFN-γ, TNFα) to PD-L1-Ig when cocultured with pancreatic tumor-infiltrating lymphocytes (59), suggesting that the PD-1H signaling axis is a powerful immunomodulatory pathway. The receptor on T cells that engages PD-1H as ligand remains to be identified. Using the agonistic PD-1H mAb MH5A, we demonstrate that PD-1H agonism on CD4+ T cells results in reduction in MAPK signaling but not mTORC1 or mTORC2 signaling. These data suggest that PD-1H as a receptor may more selectively limit downstream TCR signaling than as a ligand (PD-1H-Fc). In addition, we demonstrate that PD-1H functions as a receptor on myeloid cells and that PD-1H agonism reduced proinflammatory cytokine production by neutrophils and pDCs, including IFN-α, a key pathogenic cytokine in lupus. Although neutrophils are thought to be critical players of lupus development through neutrophil extracellular trap (NET) formation (19), recent studies have suggested that neutrophils may actually be protective for murine lupus development (68). Human discoid lupus is infiltrated by more T cells than myeloid cells and rarely has significant numbers of neutrophils.

There are a few limitations of this study. Although PD-1H agonism in MRL/lpr mice reduces autoimmunity, the critical cells involved and the mechanism of immunosuppression remain to be examined. Furthermore, both murine models of cutaneous lupus (PD-1H KO and MRL/lpr) have a predominately myeloid infiltration within the skin. In contrast, human DLE is predominately infiltrated by CD3+ T cells. Therefore, the role of PD-1H in the pathogenesis of lupus in mice and humans may involve distinct immune cell subsets, tissue localization, and potential mechanisms. Despite demonstrating that PD-1H agonism attenuates proinflammatory function in stimulated neutrophils and pDCs, we do not understand how PD-1H restricts cytokine liberation or limits the function of activation by TLR ligands. Furthermore, the regulation of PD-1H expression remains unknown. The full characterization of immune subsets expressing PD-1H in human DLE, SLE, or other autoimmune tissues remains to be performed.

Our data support PD-1H agonist as a strategy to treat cutaneous and systemic lupus. This represents a significant shift from the currently available therapeutic mechanisms in modulating the immune system to overcome cutaneous lupus specifically and autoimmunity more generally. Human agonistic anti–PD-1H antibodies need to be developed so as to test PD-1H stimulation in clinical trials. Although more studies are needed to elucidate the mechanism of PD-1H immunomodulation, PD-1H remains an attractive target for treating autoimmune diseases with elevated levels of PD-1H at the site of organ-specific autoimmunity.

MATERIALS AND METHODS

Study design

The aim of this study was to identify the role of PD-1H in the development of cutaneous and systemic lupus as well as the use of an agonist PD-1H mAb for murine lupus therapy. We backcrossed the PD-1H KO mice onto a BALB/c background and observed spontaneous cutaneous and systemic lupus development. Using IHC, flow cytometry, mass cytometry, and quantitative real-time polymerase chain reaction (PCR), we characterized the spontaneous lupus phenotype of BALB/c PD-1H KO mice. Human archived tissue samples from patients with DLE, LP, and healthy controls were analyzed with IHC and mQIF for PD-1H expression. Using an agonist PD-1H mAb, we treated MRL/lpr mice and evaluated its effect on lupus development by IHC, flow cytometry, autoantibody detection by enzyme-linked immunosorbent assay (ELISA), and cytokine/chemokine arrays. Last, we tested the mechanism of PD-1H agonism from isolated immune cells in vitro by phospho-flow cytometry, flow cytometry, and cytokine arrays. All animal studies were performed in Yale University’s animal facility, and all were approved by Yale University’s Institutional Animal Care and Use Committee in accordance with the National Institutes of Health guidelines. No statistical methods were used to predetermine sample size. The investigators were not blinded to allocation during experiments or analysis with the exception of murine dermatitis clinical scoring, murine histopathological scoring of cutaneous lupus and lupus nephritis, and histopathological analysis of human DLE. All in vivo experiments were performed at least twice, and all in vitro experiments were performed at least three times. All data are included (no outlier values were excluded). Staining of archived human formalin-fixed paraffin-embedded (FFPE) tissues was approved by the Yale University Institutional Review Board (Human Investigative Committee no. 15010105235). Primary data are reported in data file S1.

Mice

C57BL/6 PD-1H-KO mice (23) were backcrossed with BALB/cAnNCr mice for 10 generations. Both BALB/c PD-1H KO homozygote (−/−) and WT homozygote (+/+) mice were generated from offspring littermates of F10 BALB/c PD-1H heterozygotes and maintained in identical conditions within our laboratory. Female BALB/c PD-1H KO mice or female BALB/c WT control mice derived from original littermates were used for all experiments. Primers and protocols used for mouse genotyping can be found at the Mutant Mouse Resource and Research Centers (MMRRC) site (www.mmrrc.org/catalog/sds.php?mmrrc_id=31656). Female MRL/MpJ-Faslpr/J mice (MRL/lpr) were purchased from the Jackson laboratory.

Histological studies and IHC

Tissues including skin, heart, and kidneys from PD-1H-KO, WT BALB/c, and MRL/lpr mice were formalin fixed, paraffin embedded, sectioned, and stained with hematoxylin and eosin (H&E). Heart tissues were stained with Von Kossa calcium deposit detection, and kidneys were stained with periodic-acid Schiff (PAS) for glomerular membrane visualization. IHC on mouse FFPE tissues was performed by the Yale Pathology Tissue Services using antibodies directed against murine CD3 (clone SP7, Abcam), CD4 (clone 4SM95), CD8 (clone 4SM15), CD11b (clone EPR1344, Biogenex), CD317/PDCA-1/BST2 (clone 120G8.04, Novus Biologicals), Ly6b (clone 7/4), and PD-1H/VISTA (clone 742002, R&D systems). Human control skin (n = 20), DLE (n = 21), and LP (n = 11) from archived FFPE tissue was performed by Yale Dermatopathology with antibodies against human CD3 (clone F7.2.28, Dako Agilent Technologies), CD4 (clone 4B12, Dako Agilent Technologies), CD8 (clone C8/144B, Dako Agilent Technologies), CD20 (clone L26, Dako Agilent Technologies), CD68 (clone PG-M1, Dako Agilent Technologies), CD123 (clone 7G3, BD Biosciences), MPO (polyclonal, Dako Agilent Technologies), and PD-1H/VISTA (clone D1L2G, Cell Signaling Technology). Quantification was determined by counting positive cells in 6 to 10 high-powered fields (magnification, ×400) in a blinded fashion.

Skin biopsy score

Histopathological score of inflammatory dermatitis was adapted from Yang et al. (51). Overall, score is 0 to 6 for each mouse using three features: (i) cellular infiltration, (ii) epidermal hyperplasia, and (iii) epidermal ulceration. For cellular infiltration: 0 to 2 points (0 = baseline/normal/none; 1 = mild, patchy inflammation; 2 = severe, dense, and diffuse inflammation). For epidermal hyperplasia (acanthosis, hyperkeratosis, and parakeratosis): 0 to 2 points (0 = baseline/normal/none; 1 = mild; 2 = severe). For epidermal ulceration: 0 to 2 points (0 = baseline/normal/none; 1 = mild erosion or small focus of ulceration; 2 = severe). Histological specimens were scored independently in a blinded fashion by a dermatopathologist (J.M.M.).

Dermatitis clinical score

Murine skin of the face, ears, and dorsal neck were evaluated in a blinded fashion for inflammatory dermatitis using a clinical score of 0 to 3 based on previous reports (51) (no visible skin changes = 0; minimal hair loss with redness and a few scattered lesions =1; redness, scabbing, and hair loss with a small area of involvement = 2; and ulcerations with an extensive area of involvement = 3).

Protein and monoclonal antibodies

Mouse PD-1H-Fc proteins (extracellular domain of PD-1H fused with mouse IgG2AFc tag) were generated in CHO cells in our laboratory or provided by Boehringer Ingelheim. Mouse PD-1H monoclonal antibody (MH5A) was previously generated in our laboratory (21) and was either purified by our laboratory or provided by Boehringer Ingelheim. Polyclonal hamster IgG or monoclonal mouse IgG controls were purchased from BioXcell.

Autoantibody detection

Serum samples were collected from mice at indicated age or time. Mouse anti-dsDNA, anti-ANA, anti-histone autoantibodies, anti-R052/SSA, and anti-Sm were detected by ELISA kits (Alpha Diagnostic International). Absorbance was measured by OD450 nm wavelength using a microplate reader (EL808, BioTek). Autoantibody levels were shown as either absorbance at OD 450 nm or concentrations (units per milliliter) calculated by using standards provided.

Cytokine analysis

Cytokine concentrations of mouse serum or cell culture supernatant were detected by using mouse inflammation or TH1/2/17 Cytometric Bead Array kits (BD Biosciences) or BioLegend Legendplex mouse antivirus response panel. Serum from MRL/lpr mice treated with either MH5A (generated in laboratory) or Hamster IgG control (BioXcell) was tested for 31 cytokines/chemokines using Mouse Cytokine/Chemokine Array 31-plex (Eve Technologies). IFN-α cytokine detection was performed using Invitrogen IFN-α mouse ELISA kit.

Urine protein

Urine protein concentrations were detected using 10SG Urine Reagent Strips (Fisherbrand) and analyzed by Germaine Laboratories AimStrip Urine Analyzer 2.

Lupus nephritis score

Histopathological evaluation of glomerulonephritis was determined by modifying criteria outlined in the 2003 International Society of Nephrology (ISN)/Renal Pathology Society (RPS) Classification of lupus nephritis. Both H&E- and PAS-stained kidneys from 20-week-old MRL/lpr mice treated with agonist PD-1H mAb (MH5A) or control Hamster IgG were evaluated in a blinded fashion by a kidney pathologist (G.W.M.). Ten glomeruli from each kidney were evaluated for mesangial proliferation, fibrinoid necrosis, cellular crescents, tubular atrophy, and fibrosis and scored from class I to VI (32).

Gene expression of type I ISGs

Normal skin from WT mice (n = 3) and PD-1H KO (n = 3) mice and cutaneous lupus tissue from PD-1H KO mice (n = 3) were isolated, and RNA was extracted from tissues using QIAGEN RNeasy Mini kit. All mice used were 4- to 5-month-old females. Complementary DNA was generated with the RT2 first strand kit (QIAGEN) and was then used to profile type I ISGs using Mouse Type I Interferon Response RT2 Profiler PCR Array kit (QIAGEN). Array was performed on a quantitative reverse transcription PCR machine ABI 7900HT (384-well fast block). Data were analyzed with SDS 2.4 software to obtain the Ct values. Ct values were uploaded to and further analyzed by the online QIAGEN data analysis service (www.qiagen.com/us/shop/genes-and-pathways/data-analysis-center-overview-page/). Average of arithmetic mean of Ct values of five housekeeping genes Actb, B2m, Gapdh, Gusb, and Hsp90ab1 was used for normalization. Average fold changes (KO compared with WT group) of expression and P values were used to make a volcano plot. Expression of up-regulated genes in each mouse was shown as fold change to the average of WT group.

Immunophenotyping studies by flow cytometry

BALB/c PD-1H KO and WT mice were euthanized between 5 and 10 months of age as indicated. Spleen or lymph nodes (pooled bilateral axillary, bronchial, and inguinal lymph nodes) were isolated and stained with fluorescently labeled antibodies against CD45, CD3, CD4, CD8, B220, PDCA-1, Ly6G, Ly6C, PD-1, CXCR5, CD44, CD62L, Bcl6, and FoxP3 purchased from BioLegend or BD Biosciences. Samples were analyzed by BD Caliber, LSRII, or Thermo Attune. Data were analyzed using FlowJo software version 10.

Mass Cytometry studies

BALB/c PD-1H KO and WT mice were euthanized at 4 or 6 months of age, and periorbital skin was harvested. Tissue from a single eyelid was used as one sample with the following treatment. Skin tissue was homogenized and digested with collagenase IV (200 μg/ml) and DNase (20 μg/ml) for 30 min before tissue dissociation using gentleMACS. Single-cell suspensions were stained with anti–CD45.2-PE (BD Biosciences), and CD45.2+ cells were sorted by BD FACSARIA cell sorter. Sorted cells were then incubated with the mAb against mouse CD16/CD32 for 10 min at room temperature to block Fc receptors and subsequently stained with the metal-labeled mAb cocktail against cell surface molecules. Then, cells were resuspended with RPMI 1640 and 10 μM cisplatin (Fluidigm) in a total volume of 400 μl for 60 s before quenching 1:1 with pure fetal bovine serum to determine viability. Cells were centrifuged at 600g for 7 min at 4°C and washed once with phosphate-buffered saline (PBS) with 0.5% bovine serum albumin (BSA) and 0.02% NaN3. Cells were then fixed using Fixation/Permeabilization Buffer (ebioscience) for 30 min at 4°C. After the treatment with the Fixation/Permeabilization Buffer (ebioscience), cells were further incubated with the mAb cocktails against intracellular proteins. Metal-conjugated antibodies were purchased from Fluidigm or from BioLegend and conjugated in the laboratory (see table S2 for complete list of antibodies used). Then, cells were washed twice in PBS with 0.5% BSA and 0.02% NaN3 and then stained with 1 ml of 1:4000 191/193Ir DNA intercalator (Fluidigm) diluted in PBS with 1.6% paraformaldehyde overnight. The day after, cell samples were then diluted in double-distilled water (ddH2O) containing bead standards to about 106 cells/ml and then acquired on a mass cytometer (Helios, CyTOF 3, Fluidigm) equilibrated with ddH2O. Data were analyzed using Cytobank.

Pristane-induced lupus arthritis model

Female BALB/c WT or PD-1H KO 8- to 10-week-old age-matched mice were given a single intraperitoneal injection of 500 μl of pristane (TMPD; Sigma-Aldrich). Mice were monitored weekly for arthritis. We used a scoring system on the scale of 0 to 16 (0 to 4 for each paw, adding the scores for all four paws), using the following criteria: 0, normal paw; 1, one toe inflamed and swollen; 2, more than one toe but not entire paw, inflamed and swollen, or mild swelling of entire paw; 3, entire paw inflamed and swollen; 4, very inflamed and swollen paw or ankylosed paw. Autoantibodies including ANA, anti-histone, anti-Sm, and anti-dsDNA were detected using ELISA kits from Alpha Diagnostics as described above. Single-cell suspensions of peritoneal cells were obtained on day 4 after pristane injection, and anti-CD11b, anti-Ly6G, and anti-Ly6C mAbs (BioLegend) were used for neutrophil gating in flow cytometry analysis. To stimulate PD-1H signaling, WT BALB/c mice were given 200 μg of anti–PD-1H MH5A or hamster IgG (BioXcell) on days 0 and 2 after pristane injection.

Multiplexed quantitative immunofluorescence

FFPE human DLE skin (n = 10) and control skin (n = 10) from Yale Dermatopathology (Human Investigative Committee no. 15010105235) were used for mQIF. We measured the expression of human CD3 (clone F7.2.28, Dako Agilent Technologies), CD8 (clone C8/144b, Dako Agilent Technologies), CD68 (clone PG-M1, Dako Agilent Technologies), CD11b (clone CL1719), and PD-1H/VISTA (clone D1L2G, Cell signaling technologies) as previously described (47). Briefly, freshly cut FFPE blocks were deparaffinized, and antigen retrieval was performed with 1 mM EDTA (pH 8) (Sigma-Aldrich) and boiled for 20 min at 97°C (PT module, Lab Vision, Thermo Scientific). Inactivation of endogenous peroxidase activity was performed using 0.3% hydrogen peroxide in methanol for 20 min, followed by incubating in blocking solution of 0.3% BSA in 0.05% Tween-20 and tris-buffered for 30 min. Primary antibody dilution and incubation were performed followed by isotype-specific horseradish peroxidase–conjugated antibodies, and tyramine-based amplification systems (PerkinElmer) were used for signal detection. Quantitative measurement of the fluorescence signal was performed using the AQUA method of QIF as previously described (47). Immunofluorescence was quantified using AQUA. Fluorescence images of DAPI, Cy3, Cy5, and Cy7 (as appropriate) for each field of view (FOV) were collected. Image analysis was carried out using the AQUAnalysis software (Navigate Biopharma Inc.), which generates an AQUA score for each compartment by dividing the sum of target pixel intensities by the area of the compartment in which the target was measured. AQUA scores were normalized to the exposure time and bit depth at which the images were captured, allowing scores collected at different times to be directly comparable. Specimens with less that 5% area per region of interest were not included in AQUA analysis for not being representative of the corresponding specimen. Each slide was visually examined to exclude tissue and staining artifacts. This allows PD-1H to be selectively measured in immune cell subpopulations defined by its colocalization with the immune phenotype markers CD3, CD4, CD8, CD11b, and CD68.

PD-1H mRNA (C10orf54) expression in SLE and healthy controls

To determine PD-1H expression in human lupus, we analyzed PD-1H mRNA (C10orf54) in patients with SLE and normal controls using database OmicSoft DiseaseLand release HumanDisease_B37_20180620_v10 (48). This dataset contains microarray and RNA-sequencing data from independent studies involving whole-blood samples from 111 patients with SLE and 25 healthy donors.

PD-1H agonism in the MRL/lpr mouse lupus model

Female MRL/lpr mice received weekly injections of 200 μg of anti–PD-1H MH5A or control hamster IgG (BioXcell) from 6 to 16 weeks of age. Mice were monitored closely for skin diseases, and dermatitis clinical score was performed (see above). Serum and urine were collected every month. Body weight was measured every 2 weeks. Mice were euthanized at 20 weeks for analysis.

Neutrophil activation assay

Neutrophils were isolated using EasySep mouse neutrophil enrichment kit (Stemcell Technologies) from female BALB/c WT or PD-1H KO bone marrow and stimulated with or without LPS (100 ng/ml) in a round-bottom plate. IL-6, TNFα (18 hours after stimulation), and MCP-1 (72 hours after stimulation) in culture media were measured with mouse inflammation Cytometric Bead Array (CBA) kit (BD Biosciences). For PD-1H agonist studies, isolated neutrophils from female BALB/c WT, PD-1H KO, or MRL/lpr mice bone marrow were stimulated with or without LPS (200 μg/ml) in round-bottom plates in the presence of plate-bound MH5A (10 μg/ml) or hamster IgG (10 μg/ml; BioXcell). Supernatants were collected 24, 48, and 72 hours after stimulation, and IL-1β, TNFα, and MCP-1 concentrations were measured using BioLegend LEGENDplex mouse antivirus response panel or mouse inflammation CBA kit (BD Biosciences).

T cell signaling studies

Mouse total lymph node cells were stimulated with immobilized anti-CD3 (2C11 clone, 3 or 10 μg/ml) in the presence of immobilized MH5A (10 μg/ml), hamster IgG (Rockland, 10 μg/ml), PD-1H-Fc fusion protein (5 μg/ml), or mouse IgG (Rockland, 5 μg/ml). Thirty minutes after stimulation, cells were treated by cold PBS, fixed by 2% paraformaldehyde, and stained with phosphoflow antibodies against phospho-Akt3+, phospho-p70S6+, or phospho-Erk1/2+ (Cell Signaling Technology) using intracellular staining protocol before flow cytometry acquisition. Percentages of phospho-Akt3+, phospho-p70S6+, and phospho-Erk1/2+ CD4+ T cells were detected.

pDC functional assays

pDCs were isolated using EasySep mouse pDC isolation kit (Stemcell Technologies) from spleens of WT, PD-1H KO, or MRL/lpr mice and incubated with immobilized agonist PD-1H mAb MH5A or control IgG (10 μg/ml) and stimulated with LPS (5 μg/ml) or R848 (50 μg/ml). Flow cytometry for activation markers CD86 and MHC-II were performed 48 hours after stimulation with LPS (5 μg/ml). Supernatants were collected 24, 48, and 72 hours after stimulation with R848 (50 μg/ml), and IFN-α levels were measured using BioLegend LEGENDplex mouse antivirus response panel or Invitrogen IFN-1α mouse ELISA kit.

Statistical analysis

Data were obtained from at least two independent experiments and were analyzed using GraphPad Prism software (version 7 or 8). Statistical significance was determined by an unpaired two-tailed Student’s t test, analysis of variance (ANOVA), log rank, or Mann-Whitney U test as indicated in the legend (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Flow cytometric data were analyzed using FlowJo v10. Mass cytometry data were analyzed using Cytobank. mRNA expression was analyzed using Qiagen GeneGlobe. The number of sampled units, n, is indicated in legend. A volcano plot was generated with mRNA expression data by performing multiple t tests in GraphPad Prism.

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/11/522/eaax1159/DC1

Fig. S1. PD-1H KO mice do not develop lupus nephritis.

Fig. S2. Serum cytokines from WT and PD-1H KO mice.

Fig. S3. PD-1H modulates type I IFNs, and type I IFN response genes are modestly up-regulated in cutaneous lupus lesions of PD-1H KO mice.

Fig. S4. Quantification of mass cytometry from WT and PD-1H KO skin.

Fig. S5. PD-1H KO mice and WT mice develop similar pristane-induced kidney disease.

Fig. S6. IHC of human control skin, discoid lupus, and LP.

Fig. S7. Multiplex immunofluorescence of human DLE and control skin.

Fig. S8. Histology and immunophenotyping of MRL/lpr mice treated with a PD-1H agonist.

Fig. S9. Proteinuria, kidney inflammation, and lupus nephritis in MRL/lpr mice treated with a PD-1H agonist.

Fig. S10. The PD-1H agonist MH5A does not affect T cell signaling or pDC and neutrophil function from PD-1H KO mice.

Table S1. Histological scoring of cutaneous lupus lesions.

Table S2. Murine mass cytometry antibody panel.

Data file S1. Primary data.

REFERENCES AND NOTES

Acknowledgments: We thank B. Cadugan for editing the manuscript and other members in the laboratory of L. Chen for the helpful discussions and technical assistance. Funding: This research was partially supported by sponsored research funding from Boehringer Ingelheim and the United Technologies Corporation Endowed Chair. M.D.V. is supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases Training in Investigative Dermatology grant (T32 AR007016), the Dermatologist Investigator Research Fellowship, and Physician-Scientist Career Development Award from the Dermatology Foundation. M.F.S. is supported by a Miguel Servet contract from Instituto de Salud Carlos III, Fondo de Investigacion Sanitaria (Spain), and, currently, at the Department of Oncology, Clinic University of Navarra, Pamplona, Spain. Author contributions: X.H., M.D.V., and L.C. designed the study. X.H., M.D.V., W.Y., M.F.S., T.B., J.A., S.W.L., J.-P.Z., X.N., A.N., A.B., D.B.F., L.Z., and T.K.K. performed experiments and data analysis. P.G. performed multiplex immunofluorescence and analysis of data. F.L.-G. performed microarray analysis of PD-1H expression in patients with SLE. J.M.M. evaluated histology of skin tissue. G.W.M. evaluated the histology of kidney tissue. X.H., M.D.V., and L.C. drafted the manuscript, and all authors critically revised the manuscript. Competing interests: L.C. is a consultant/board member for NextCure, Tayu, Junshi, Zai Lab, Vcanbio, and GenomiCare; is a scientific founder of NextCure and Tayu; and has sponsored research grants from NextCure, Pfizer, and Boehringer Ingelheim in the past 12 months. D.F.B. is a current employee of NextCure. The other authors declare that they have no competing interests. Several patents related to this study were submitted. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials. The BALB/c PD-1H KO mice are available from L.C. and will be transferred under a material transfer agreement (Yale UBMTA or MTA-TO); the receiving party will also need to sign the MMRRC MTA.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article