Graphical abstract

   graphic file with name fx1.jpg
   [53]Open in a new tab

Highlights

     * •
       RSAD2 as a pathogenic interferon-stimulated gene at the
       maternal-fetal interface
     * •
       RSAD2 excess increases lipid load in structural cell populations
     * •
       RSAD2-induced lipid accumulation impairs placental vasculogenesis
     * •
       LCA targeting of RSAD2 alleviates lipid-induced vascular
       inflammation
     __________________________________________________________________

   The role of type I interferon response at the maternal-fetal interface
   is complex. Ding et al. identify RSAD2 as a pathogenic ISG in SLE
   pregnancies, where it drives placental lipid accumulation and impairs
   vasculogenesis, with LCA targeting RSAD2 to improve pregnancy outcomes.

Introduction

   The prolonged and excessive induction of type I interferon (IFN-I)
   signaling impairs embryonic development,[54]^1^,[55]^2^,[56]^3 which is
   observed in some autoimmune diseases such as systemic lupus
   erythematosus (SLE) and Aicardi–Goutières
   syndrome.[57]^4^,[58]^5^,[59]^6 Upon binding to their receptor, IFN-I,
   mainly the multiple subtypes of IFN-α and IFN-β, induces hundreds of
   interferon-stimulated genes (ISGs), which function to restrict
   pathogenic infections and regulate the IFN
   response.[60]^7^,[61]^8^,[62]^9 However, the functions of ISGs in
   different diseases could be complex. Therefore, identifying the
   protective and pathogenic ISGs in these processes is essential for the
   accurate diagnosis and treatment of aforementioned diseases.

   Several protective ISGs are required for a successful pregnancy,
   whereby it plays a role in pregnancy recognition and uterine artery
   remodeling during pregnancy.[63]^10^,[64]^11 Previous studies have
   shown impaired decidual spiral artery remodeling in mice lacking the
   IFN-I receptor (Ifnar^−/− mice), indicating that activation of the
   IFN-I signaling pathway is crucial for early decidualization.[65]^12
   Unlike the deciduomata of pseudopregnant mice, the presence of an
   embryo in the decidua of pregnant mice upregulated the expression of
   many ISGs, including Irf8, Iarp, Isg15, and Ifi27.[66]^13 Recently, a
   spatial multiomics map of trophoblast development in human early
   pregnancy showed that the transformation of interstitial extravillous
   trophoblast cells into placental bed giant cells, the receptors of the
   IFN-I signaling pathway, and the corresponding target, IFI27, are
   upregulated.[67]^14 This also suggests that a series of ISGs play a
   protective role in promoting embryo implantation, trophoblast invasion,
   and vascular remodeling. By contrast, excessive IFN-I signaling
   increases the activation of IFN-induced transmembrane proteins
   (IFITMs), which leads to the inhibition of placental
   syncytiotrophoblast formation and ultimately, fetal demise.[68]^2
   Moreover, the inactivation of G protein-coupled estrogen receptor 1
   contributes to the upregulation of canonical ISGs such as Ifit2, Isg15,
   and Ccl5, halting fetal development and promoting fetal demise during
   maternal inflammation.[69]^15 Considering that some ISGs exert
   protective effects during pregnancy, the identification of pathogenic
   ISGs could improve the treatment accuracy of IFN-I-related
   complications during pregnancy.

   SLE is an autoimmune disease characterized by the excessive production
   of IFN-I, especially IFN-α.[70]^16^,[71]^17^,[72]^18^,[73]^19 An ISG
   expression analysis of peripheral blood mononuclear cells (PBMCs)
   isolated from patients with SLE revealed that the expression of many
   key ISGs, including ISG15, ISG20, MX1, MX2, IFIT1, IFIT2, IFIT3,
   IFITM1, IFITM2, and RSAD2 (also called cytomegalovirus-inducible gene 5
   [Cig5] and Cig33), was significantly
   activated.[74]^20^,[75]^21^,[76]^22 It is ∼7–9 times more prevalent in
   women than in men, especially in women of childbearing
   age.[77]^23^,[78]^24 Furthermore, elevated IFN-α activity early in
   gestation is associated with the onset of preeclampsia in patients with
   SLE, which occurs via the inhibition of angiogenesis.[79]^25^,[80]^26
   In summary, the excessive activation of IFN-α signaling pathway is a
   high-risk factor for adverse pregnancy outcomes in SLE. Although
   specific guidelines for the management of SLE in pregnancy have been
   implemented, with hydroxychloroquine and aspirin treatment being
   recommended, recent data show that pregnancy outcomes in SLE have not
   improved over the past three decades.[81]^27^,[82]^28 Nevertheless,
   studies of the mechanisms by which IFN-α and its downstream pathogenic
   ISGs mediate adverse pregnancy outcomes in patients with SLE and those
   with other IFN-α-related pregnancy diseases are still very limited.

   Here, we demonstrated that RSAD2 is a pathogenic ISG at
   the maternal-fetal interface in patients with SLE, which causes lipid
   accumulation in the placenta, inhibited angiogenesis, and impaired
   fetal development. The small molecular compounds targeting RSAD2
   alleviate abnormal pregnancy outcomes (APOs) in the SLE mouse model.
   Our work will contribute to the development of new treatment strategies
   for pregnant patients with a dysregulated IFN-I response.

Results

RSAD2 is highly expressed at the maternal-fetal interface in patients with
SLE

   To identify potentially pathogenic ISGs in SLE pregnancy, we monitored
   ISGs expression in the PBMCs, decidua, and chorionic villi from
   pregnant patients with SLE using the quantitative real-time PCR array
   platform. We found that the expression of RSAD2 in PBMCs of pregnant
   patients with SLE was higher than in those from the group of healthy
   pregnant individuals ([83]Figures 1A and [84]S1A). The results
   indicated that the IFN-I response in the peripheral blood of pregnant
   patients with SLE was more activated than that of the control group.
   Next, we detected the expression of the indicated ISGs in situ in the
   decidua and chorionic villi and found that the expression of RSAD2 was
   also higher in these tissues than in those of the normal pregnant group
   ([85]Figure 1B). To investigate the response of the maternal-fetal
   microenvironment to IFN-α, we tested the induction of ISG expression in
   decidual and trophoblast organoids by treating them with IFN-α or
   phosphate-buffered saline (PBS, as a negative control). We found that
   the expression of RSAD2 was strongly induced by IFN-α in trophoblast
   organoids and decidual organoids ([86]Figures 1C and [87]S1B). We found
   that IFNAR1 was expressed in situ in the decidua and chorionic villi,
   as well as in the decidual and placental trophoblast-like organoids
   ([88]Figures 1D and 1E). To explore how various cell types responded to
   IFN-α, we sorted decidua-derived immune cells, including CD56^+ natural
   killer (NK) cells, CD3^+ T cells, and CD14^+ macrophages, as well as
   structural cells such as decidual stromal cells (DSCs) and
   trophoblasts. Gating strategy for immune cells was showed in
   [89]Figure S1C. We then detected ISG expression changes in these cells
   in response to IFN-α. We found that the structural cell populations at
   the maternal-fetal interface (i.e., DSCs and trophoblasts) mounted a
   strong IFN-α response ([90]Figures 1F and [91]S1D). Among the immune
   cells, macrophages were the most responsive to IFN-α, followed by
   T cells and NK cells ([92]Figure 1F). The high expression of RSAD2 in
   DSCs and trophoblast cells has also been confirmed at the protein level
   ([93]Figures 1G and 1H). The aforementioned results indicate that the
   structural cell population of the maternal-fetal interface is more
   capable of upregulating the expression of RSAD2 in response to IFN-I.

Figure 1.

   [94]Figure 1
   [95]Open in a new tab

   RSAD2 is highly expressed in pregnant patients with SLE

   (A) Heatmap showing the expression of interferon-stimulated genes
   (ISGs) in the peripheral blood mononuclear cells (PBMCs) of healthy
   women (Healthy PBMCs), female patients with SLE (SLE PBMCs), healthy
   pregnant women (Healthy pregnancy PBMCs), and pregnant patients with
   SLE (SLE pregnancy PBMCs).

   (B) Heatmap showing the expression of ISGs in the decidua and villi of
   healthy early pregnant women and pregnant patients with SLE.

   (C) Heatmap showing the expression of ISGs in the decidual and
   trophoblast organoids from the first trimester decidual gland or
   trophoblast cells (passage 1, day 5), treated with 1,000 U/mL IFN-α or
   PBS for 12 h. (A, B, and C). The 2^−ΔCt values of ISGs were calculated
   from the ISG qPCR array data.

   (D) Representative IHC image of IFNAR1 and CD55 (expressed in DSCs)
   co-staining of decidual tissue sections (top), and representative IHC
   image of IFNAR1 and EPCAM (expressed in VCTs, the proliferative villous
   cytotrophoblast cells) co-staining of chorionic villi sections
   (bottom); scale bar, 50 μm.

   (E) Representative IHC image of decidual organoids (top) from the first
   trimester decidual glands (passage 1, day 5) and of trophoblast
   organoids (bottom) from the first trimester trophoblast cells (passage
   1, day 5); scale bar, 50 μm.

   (F) Expression of ISGs in decidual CD56^+ NK cells, CD3^+ T cells,
   CD14^+ macrophages, decidual stromal cells (DSCs), and trophoblasts
   (n = 3); the 2ˆ^−ΔΔCt values of ISGs were calculated from the qPCR
   data.

   (G and H) Western blot showing RSAD2 and ISG15 expression in DSCs and
   trophoblasts, treated with 1,000 U/mL IFN-α or PBS for 12 h.

   Data were analyzed using the unpaired t test and are presented as the
   mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns,
   not significant.

RSAD2 upregulation is accompanied by accumulation of lipid droplets and
elevated levels of proinflammatory prostaglandin

   To explore the role of RSAD2 in a high IFN-α environment at the
   maternal-fetal interface, we performed immunoprecipitation (IP)
   experiments targeting RSAD2 protein in DSCs and identified the IP
   products by liquid chromatography-tandem mass spectrometry (LC-MS/MS)
   ([96]Figure 2A). RSAD2 and some binding partners of it are typically
   associated with the mitochondria and lipid droplets (LDs)
   ([97]Figure 2B), which suggested that it could potentially alter the
   metabolism of the maternal-fetal interface. To test the ability of
   RSAD2 to regulate lipid metabolism in DSCs and trophoblasts, we
   analyzed the relationship between the RSAD2 expression and LD
   accumulation. Lipid metabolism was assessed in 293T and JEG3 (human
   placental choriocarcinoma) cells following the overexpression of RSAD2
   through transduction with an RSAD2-overexpressing lentiviral vector or
   a negative control vector. RSAD2 overexpression efficiency was
   quantified by qPCR ([98]Figures 2D and 2F). LD staining showed that
   high RSAD2 expression in both 293T and JEG3 cells was associated with
   increased LD accumulation ([99]Figures 2C and 2E). In addition, we
   found that IFN-α treatment induced high RSAD2 expression in DSCs, which
   was accompanied by the marked accumulation of LDs; by contrast, with
   PBS treatment, DSCs lacking RSAD2 expression had only basal levels of
   LDs ([100]Figures S2A and S2B). We confirmed that RSAD2 expression and
   the accumulation of LDs were also observed in trophoblastic organoid
   models treated with IFN-α ([101]Figures S2D and S2E). To further
   characterize the effect of RSAD2 on the formation of LDs in the decidua
   and chorionic villi, we performed immunohistochemistry (IHC) on frozen
   sections from PBS- or IFN-α-treated tissue samples, and LDs were
   detected using oil red O staining. We observed an increase of LD level
   in the decidua ([102]Figure S2C), although it was only weakly increased
   in the chorionic villi under the same conditions ([103]Figure S2F). The
   aforementioned results show that RSAD2 upregulation is associated with
   increased LD concentration in DSCs and trophoblasts. To further test
   whether the RSAD2 overexpression is associated with lipid metabolism,
   we measured the expression levels of the major transcriptional
   regulators sterol regulatory element-binding protein and the
   endoplasmic reticulum-resident diacylglycerol acyltransferase 1; the
   genes encoding the fatty acid synthase, FASN, which catalyzes de novo
   lipid synthesis, and CD36, which assists in fatty acid uptake from the
   exogenous environment; and the genes encoding lipid transporter
   proteins, including apolipoprotein (apo) A1, APOA4, APOB, and APOD. The
   expression levels of the genes encoding these proteins were increased
   in RSAD2-overexpressing cell types ([104]Figures 2D and 2F) and
   IFN-α-treated trophoblasts than in those treated with PBS
   ([105]Figure S2G). Moreover, cells stably expressing RSAD2 displayed an
   elevated ability to take up fatty acids ([106]Figures S2H and S2I). To
   further confirm the downstream genes of RSAD2, the expression levels of
   RSAD2 were knocked down in 293T cells that had been overexpressed using
   small interfering RNA (siRNA) interference assays, which resulted in
   the downregulation of lipid metabolism genes ([107]Figure S2L). The
   effective siRNAs targeting RSAD2 were screened by qPCR and western blot
   ([108]Figures S2J and S2K). LD development has marked effects on the
   production of inflammatory mediators, including prostaglandin E2
   (PGE2), interleukin (IL)-1β, and IL-6, in foamy macrophages.[109]^29 We
   found that the level of proinflammatory PGE2, PGI2, and PGD2
   was significantly increased in DSCs in response to IFN-α
   ([110]Figures 2G and 2H). Taken together, our data indicate that RSAD2
   is associated with lipid metabolic reprogramming at the maternal-fetal
   interface accompanied by LD accumulation and elevated levels of
   production of proinflammatory mediators.

Figure 2.

   [111]Figure 2
   [112]Open in a new tab

   Upregulation of RSAD2 is associated with inflammatory lipid
   accumulation at the maternal-fetal interface

   (A) Decidual stromal cells (DSCs) were treated with 1,000 U/mL IFN-α
   (+) or PBS (−) for 12 h. Cell lysates were subjected to
   immunoprecipitation (IP) with a rabbit anti-RSAD2 monoclonal antibody
   (mAb) or a rabbit IgG control. RSAD2 IP was determined by western
   blotting (upper). Heatmap of proteins interacting with RSAD2,
   identified using LC-MS/MS (lower).

   (B) Sankey map of cell localizations and functions of RSAD2-interacting
   proteins.

   (C) Representative IF image of LDs in 293T cells, which has been
   overexpressed RSAD2; scale bar, 10 μm.

   (D) qPCR analysis of lipid metabolism-related genes in
   RSAD2-overexpressing 293T cells.

   (E) Representative IF image of LDs in JEG3 cells, which has been
   overexpressed in RSAD2; scale bar, 10 μm.

   (F) qPCR analysis of lipid metabolism-related genes in
   RSAD2-overexpressing JEG3 cells. Data are representative of three
   independent experiments with similar results (C, D, E, and F).

   (G) Schematic diagram of the prostaglandin synthesis pathway.

   (H) DSCs from 12 donors were isolated and either treated with PBS or
   IFN-α for 12 h. To obtain enough cells for LC-MS/MS assay, the cells
   from every two donors were mixed as one sample. The threshold for
   identifying significant differences was VIP ≥ 1 and t test p < 0.05 in
   the orthogonal partial least squares discriminant analysis (OPLS-DA)
   model.

   ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not
   significant.

   To confirm that the high expression of RSAD2 is associated with the
   accumulation of LDs at the maternal-fetal interface in vivo, we
   employed a mouse pregnancy model in which high IFN-I levels were
   induced by poly(I:C) treatment at E10.5 ([113]Figure S3A). We found
   that poly(I:C) treatment induced a severe reduction in dam weight
   ([114]Figure S3B) and APOs ([115]Figures S3C–S3G). Intrauterine growth
   restriction in pups of the poly(I:C)-treated group was observed at
   E11.5 ([116]Figure S3C). The body weights, lengths, and widths of the
   live fetuses and the diameters of their placentas were also
   significantly smaller compared with the controls
   ([117]Figures S3D–S3G). We found that the murine decidua and placenta
   expressed significantly higher levels of RSAD2 and ISG15 following
   poly(I:C) treatment compared with those of control mice
   ([118]Figures 3A–3C and [119]S3H). Moreover, BODIPY 493/503 staining
   showed that LDs accumulated in the decidua and placenta after the
   intraperitoneal injection of poly(I:C), which was not observed in the
   control animals ([120]Figures 3D, 3E, and 3F). Oil red O staining
   confirmed that an increase in RSAD2 levels associated with excessive
   lipid accumulation ([121]Figures 3G, 3H, and 3I). Meanwhile, LC-MS/MS
   revealed that the triacylglycerol (TAG)56:6(16:0) and diacylglycerol
   (DAG) (16:0/18:1) content of the decidua and placenta, respectively,
   were increased in the poly(I:C)-treated mice compared with the controls
   ([122]Figures 3J and 3K). We performed an RNA sequencing (RNA-seq)
   analysis of saline- or poly(I:C)-treated decidua and placenta collected
   at E12.5 and found that the expression of the apolipoprotein (Apo)
   D-encoding gene, Apod, was significantly increase in the decidua
   ([123]Figure S3I), while that of Apoa1, Apoa4, and Apob was increased
   in the placenta ([124]Figure S3J). The significantly increased
   expression of a series of Apo genes at the maternal-fetal interface
   suggested that lipid metabolism was markedly perturbed. Collectively,
   our results showed that the upregulation of Rsad2 is associated with
   lipid accumulation at the maternal-fetal interface in the placenta.

Figure 3.

   [125]Figure 3
   [126]Open in a new tab

   Elevated RSAD2 expression is accompanied by lipid accumulation in the
   placenta

   (A) Rsad2 and Isg15 mRNA from the decidual and placental samples were
   analyzed with quantitative real-time PCR (n = 3).

   (B) Western blot analysis of RSAD2 and ISG15 protein expression in
   decidual and placental samples. The experiment has been independently
   repeated at least five times.

   (C) Data statistics of RSAD2 relative to tubulin expression were
   obtained by Image Lab.

   (D and E) LD staining of the decidua (D) and placenta (E) of pups from
   three independent pregnant dams at gd12.5. Images are representative of
   three independent experiments with similar results. Scale bar, 50 μm.

   (F) Quantification of LD staining by ZEN 2.6.

   (G and H) Oil red O staining of the decidua (G) and placenta (H) of
   pups from independent pregnant dams at gd12.5. Scale bar, 50 μm. Images
   are representative of three independent experiments.

   (I) Quantification of oil red O staining by Image-Pro Plus 6.0.

   (J) Heatmap representing the relative abundances of lipid subspecies.
   Each column represents a single decidual or placental sample from
   independent pregnant dams, and each row corresponds to a single
   subspecies of lipid.

   (K) Quantification of TAG56:6 (16:0) and DAG (16:0/18:1) subspecies.
   Poly(I:C)-treated decidual and placental samples are shown as red bars,
   and saline-treated samples are depicted as blue bars.

   All the data are presented as mean values ± SEM. p values were computed
   using the multiple t tests. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001,
   ∗∗∗∗p < 0.0001; ns, not significant.

RSAD2 depletion alleviates DAG-lipid accumulation in the placenta and adverse
pregnancy outcomes

   To demonstrate that RSAD2 is a pathogenic ISG at the maternal-fetal
   interface, we next used Rsad2^−/−mice, which were generated using
   CRISPR-Cas9 technology, to determine the effect of RSAD2 on pregnancy
   outcomes ([127]Figure S4A). There were no obvious histopathological
   changes in the liver, spleen, and kidneys of Rsad2^−/−mice, compared
   with wild-type (WT) animals ([128]Figure S4B). We found that poly(I:C)
   treatment significantly reversed the phenotype of embryo absorption
   (the loss of embryos or fetuses, typically characterized by the absence
   of visible viable embryos or the presence of only residual blood clots
   in place of what would normally be a complete embryo) in
   Rsad2^−/−pregnant mice compared with WT pregnant mice at E12.5
   ([129]Figures 4A and 4B). Specifically, compared with poly(I:C)-treated
   WT mice, embryonic developmental abnormalities and the embryo
   absorption rate significantly decreased in Rsad2^−/−pregnant mice
   ([130]Figures 4B and 4C). Moreover, the body lengths and the body
   widths of live fetuses were significantly smaller in poly(I:C)-treated
   WT mice compared with saline-treated controls. By contrast, the fetuses
   of the poly(I:C)-treated Rsad2^−/−mice were similar in size to those of
   the saline-treated control group ([131]Figure 4D). Moreover, we
   observed the following abnormalities in the poly(I:C)-treated WT mice
   but not in the Rsad2^−/−group: decidual vessel wall thickening, a
   reduction in the size of placental vascular spaces, and a decrease in
   the fetal blood cell content in the labyrinthine zone
   ([132]Figure S4C). The aforementioned results suggest that Rsad2, as a
   pathogenic ISG, induces adverse pregnancy outcomes.

Figure 4.

   [133]Figure 4
   [134]Open in a new tab

   RSAD2 depletion alleviates poor pregnancy outcomes and DAG-lipid
   accumulation

   (A) Representative image of pups from poly(I:C)- or saline-treated
   pregnant WT or Rsad2^−/−mice at gd12.5. Scale bar, 1 cm.

   (B) One representative litter from each of the indicated treatment
   groups collected at gd12.5 is shown. Scale bar, 1 mm.

   (C) Pie charts indicating the total number of fetuses analyzed and
   their developmental phenotypes (i.e., resorbed or halted development
   versus phenotypically normal), related to (A). WT-saline (n = 5),
   Rsad2^−/−-saline (n = 5), WT-poly(I:C) (n = 11), and
   Rsad2^−/−-poly(I:C) (n = 8) were enrolled.

   (D) The lengths (left) and widths (right) of live fetuses were measured
   with vernier calipers. WT-saline (n = 4), Rsad2^−/−-saline (n = 4),
   WT-poly(I:C) (n = 5), and Rsad2^−/−-poly(I:C) (n = 4) were measured. p
   values were computed using unpaired t test. (A), (B), (C), and (D) are
   representative of at least two independent experiments.

   (E and G) (E) LD staining and (G) oil red O staining of the deciduas
   and placentas from poly(I:C)-treated pregnant WT or Rsad2^−/−mice and
   saline-treated pregnant WT mice, collected at gd12.5. Images are
   representative of three independent experiments. Scale bar, 50 μm. DB,
   decidual basalis; LZ, labyrinth zone.

   (F) Quantification of mean fluorescence intensity (MFI) of LD staining
   in the labyrinth zone by ZEN 2.6.

   (H) Quantification of oil red O staining in the labyrinth zone by
   Image-Pro Plus 6.0. (F and H) Data are presented as the mean ± SEM and
   were obtained from the placentas of independent pregnant dams. p values
   were computed using the one-way analysis of variance (ANOVA).

   (I) Heatmap showing the relative abundances of lipid subspecies. Each
   column represents a single decidual or placental sample from
   independent WT or Rsad2^−/−pregnant dams (poly[I:C]-treated), and each
   row corresponds to a single subspecies of lipid.

   (J) Gene set enrichment analysis (GSEA) of genes associated with the
   triacylglyceride synthesis pathways expressed in the decidual samples
   of Rsad2^−/−mice compared to those expressed in the decidual samples of
   WT mice (poly[I:C]-treated). (A–J) Rsad2^−/−and WT mice were injected
   intraperitoneally with poly(I:C) or saline at gd10.5, and the tissues
   were harvested at gd12.5.

   All the data are presented as mean values ± SEM. (D) p values were
   computed using the multiple t tests. ∗p < 0.05, ∗∗p < 0.01,
   ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

   To explore whether high Rsad2 expression caused fetal demise in mice by
   promoting LD accumulation in the placenta, whole placenta frozen
   sections were stained with BODIPY 493/503. We found that, following
   poly(I:C) treatment, unlike that mass LD accumulation was observed in
   the WT animals, the LD content of the decidua and labyrinthine zone in
   Rsad2^−/−mice was similar with saline-treated WT mice ([135]Figures 4E
   and 4F). In addition, oil red O staining showed that large LDs were
   present in the decidua and placenta of WT mice following poly(I:C)
   treatment, while the lipid content of Rsad2^−/−mice was similar to that
   of saline-treated WT controls ([136]Figures 4G and 4H). The widely
   targeted quantitative lipidomics results revealed that the levels of
   DAG (16:0/18:1) and other sn-1,2-DAG species were significantly lower
   in the Rsad2^−/−placenta than those in WT placenta under the same
   poly(I:C) treatment ([137]Figure 4I). Although no significant changes
   in the decidual lipidome were observed in the poly(I:C)-treated
   Rsad2^−/−mice and WT mice ([138]Figure 4I), transcriptome data revealed
   that the TAG synthesis pathway was inhibited in Rsad2^−/− animals
   compared with poly(I:C)-treated WT mice ([139]Figure 4J). It has been
   reported that the sn-1,2-DAG-mediated activation of protein kinase Cε
   is implicated in obesity-related lipid metabolism disorders, where it
   induces hepatic insulin resistance and type Ⅱ diabetes.[140]^30
   Moreover, the excessive accumulation of lipids leads to macrophage
   activation and the production of proinflammatory PGE2. In accordance,
   we found COX-2^+ macrophages in the placenta of poly(I:C)-treated WT
   mice but not in those of poly(I:C)-treated Rsad2^−/−mice or
   saline-treated WT animals ([141]Figure S4D). Taken together, our
   results show that Rsad2 is a pathogenic ISG, which causes adverse
   pregnancy outcomes by inducing DAG-lipid accumulation in the placenta.

RSAD2 depletion alleviates placental vascular injury caused by DAG-lipid
accumulation

   The fetus-derived placenta is composed of a layer of trophoblast giant
   cells, spongiotrophoblast layer, and a labyrinthine zone, the site of
   metabolic exchange between the maternal and fetal blood. The
   labyrinthine zone contains nonnucleated maternal red blood cells (RBCs)
   and fetal blood spaces, which are populated by nucleated fetal RBCs
   ([142]Figure 5A). Our earlier findings led us to hypothesize that the
   high expression of RSAD2 attenuated blood vessel development in the
   placenta by inducing excessive lipid accumulation. To test this
   hypothesis, we stained frozen placenta sections of saline- or
   poly(I:C)-treated WT mice with cytokeratin to label trophoblasts and
   BODIPY 493/503 probes to label LDs. We found that LDs mainly
   accumulated in the placental labyrinth ([143]Figure 5B) and partially
   colocalized with trophoblasts ([144]Figure 5C). This result suggests
   that lipid accumulation in the placenta damaged cells other than
   trophoblasts.

Figure 5.

   [145]Figure 5
   [146]Open in a new tab

   RSAD2 depletion alleviates placental vascular injury caused by lipid
   accumulation

   (A) Schematic diagram showing the decidual and placental architecture
   and the localization of different cell types within these structures.

   (B and C) OCT-embedded frozen placental sections from poly(I:C)- or
   saline-treated pregnant WT mice were stained for BODIPY 493/503 (used
   to stain LDs, green). In (B), the whole placental images were obtained
   by a Pannoramic MIDI scanner (3DHISTECH; Hungary) and analyzed using
   Case Viewer software. Scale bar, 500 μm.

   (C) Local enlargement of the placental labyrinth, obtained by a
   confocal LSM980 microscope and analyzed using ZEN 2.6 software. CK17/19
   is used to stain trophoblast cells. Scale bar, 20 μm. Representative
   images from at least three placentas per genotype from at least two
   litters are shown.

   (D) GSEA of genes expressed in the placentas of poly(I:C)-treated mice
   compared to those expressed in the placentas of WT mice
   (saline-treated), showing inhibition of vasculogenesis pathway.

   (E) Placentas from poly(I:C)- or saline-treated pregnant WT or
   Rsad2^−/−mice were fixed in PFA. Paraffin-embedded sections were then
   stained for CD31 (used to stain blood vessel) with DAB. Representative
   blood vessel images in labyrinth zone from three placentas of three
   pregnant mice per genotype and per treatment are shown. Scale bar,
   50 μm.

   (F) Quantification of the labyrinth zone area occupied by blood
   vessels, visualized using CD31 staining as described in (E) and
   analyzed using Case Viewer software, n = 240. Error bars indicate SEM,
   and statistical analyses were performed using the unpaired t test. For
   measuring vascular space, a researcher was blinded to the samples and
   asked to quantify the area within ∼10 blood vessels from each image
   irrespective of the orientation of the vessel cross-section. Blood
   vessels were defined using CD31 staining and were quantified from at
   least eight different fields of view from three different placentas for
   each group.

   (G and H) GSEA of genes expressed in the placentas of Rsad2^−/−mice
   compared to those expressed in the placentas of WT mice
   (poly[I:C]-treated), showing the enrichment of genes associated with
   the vasculogenesis (G) and erythrocyte development (H) pathways.

   (I) Comparison of FPKM values of Aplnr (Apelin receptor), Kdr (VEGF
   receptor 2), Tie1 (Tie1, angiopoietin receptor 1), Tek (Tie2,
   angiopoietin receptor 2), Apln (Aplin), and Apela (Elabela) in mouse
   placentas, quantified using RNA-seq; the unpaired t test was used for
   statistical analyses.

   ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not
   significant.

   To verify this notion, IHC staining of placenta tissue sections with an
   anti-CD31 antibody was conducted. With poly(I:C) treatment, the severe
   vascular injury was observed in WT mice, but not in Rsad2^−/−mice
   ([147]Figure 5E). Moreover, the size of the area occupied by blood
   vessels in the placenta was significantly decreased in
   poly(I:C)-treated WT mice compared with that in Rsad2^−/−mice,
   indicating that RSAD2 depletion alleviated placental vascular injury
   ([148]Figures 5E and 5F). We next analyzed transcriptional data from
   the placentas of WT and Rsad2^−/−mice following poly(I:C) treatment and
   found that vasculogenesis and the erythrocyte development pathway were
   significantly restored in Rsad2^−/−mice compared with the WT animals
   ([149]Figures 5G and 5H). Moreover, the genes encoding the apelin
   receptor (Aplnr), the VEGF receptor 2 (Kdr), and the angiopoietin
   receptors (Tie1 and Tie2), which mediate proangiogenic signaling in the
   placenta, were significantly downregulated in poly(I:C)-treated WT
   mice, but not in Rsad2^−/−mice ([150]Figure 5I). The adipokine apelin
   (encoded by Apln), a ligand of Aplnr, had an increased trend in the
   placenta of poly(I:C)-treated WT mice (relative to the PBS-treated WT
   controls), a result of a feedback effect. By contrast, the expression
   of Apln had a tendency to decrease in poly(I:C)-treated Rsad2^−/−mice
   (relative to poly(I:C)-treated WT mice) ([151]Figure 5I). The
   Apln-Aplnr axis is implicated in cardiovascular function and
   vasoconstriction.[152]^31 Apela, which encodes the peptide Elabela (an
   endogenous ligand of the Aplnr in placenta), is a circulating hormone
   secreted by the placenta to promote placental development and prevent
   preeclampsia.[153]^32 We found that the expression of Apela was higher
   in Rsad2^−/−mice ([154]Figure 5I). Our findings collectively showed
   that the high expression of Rsad2 in WT mice is associated with severe
   vessel injury in the placenta and the excessive accumulation of neutral
   lipids.

RSAD2 depletion ameliorates APOs and lipid accumulation in pristane-induced
pregnant mice

   To further demonstrate that Rsad2 is a pathogenic ISG in SLE pregnant
   model, we next used an improved SLE induction model. Pristane-induced
   lupus is a classical model of SLE with overactivation of IFN-I
   signaling.[155]^33 We found that the expression of Rsad2, Isg15, and
   Ifit1 in the decidua and placenta was upregulated in pregnant mice
   ([156]Figures S5A and S5B). To examine the links between the
   upregulation of Rsad2 and the pathogenesis of SLE pregnancy, Rsad2^−/−
   and WT pregnant mice were treated with pristane. Pristane challenge
   causes embryonic developmental abnormalities, especially the fetal
   brain and liver, and medium absorption (relative to the degree of
   embryo resorption in the poly[I:C] model) ([157]Figure S5C). We found
   that pristane treatment significantly reversed the phenotype of
   embryonic developmental abnormalities in Rsad2^−/−pregnant mice
   compared with WT pregnant mice at E12.5 ([158]Figures S5C and S5D).
   Compared with pristane-treated WT mice, embryonic developmental
   abnormalities and the embryo absorption rate significantly decreased in
   Rsad2^−/−pregnant mice ([159]Figure S5E). Moreover, the body length,
   head widths, and weight of live fetuses were significantly smaller in
   pristane-treated WT mice compared with saline-treated WT control mice
   ([160]Figure S5F). By contrast, the fetuses of the pristane-treated
   Rsad2^−/−mice were alleviated in size to those of the pristane-treated
   WT group ([161]Figure S5F). In conclusion, unlike the apparent APO
   phenotype in pristane-treated WT mice, the Rsad2^−/−mice with
   pristane-treatment significantly alleviated embryo resorption or
   abnormal development. The aforementioned results suggest that Rsad2, as
   a pathogenic ISG, induces adverse pregnancy outcomes in a
   pristane-induced SLE-like model. Consistent with the observations made
   in the poly(I:C) model, lipid accumulation-related genes were
   upregulated ([162]Figure S5G). Furthermore, following pristane
   treatment, unlike that LD accumulation was observed in the WT animals,
   the LD content of the decidua and labyrinthine zone in Rsad2^−/−mice
   was similar with saline-treated WT mice ([163]Figures S5H and S5I). In
   addition, vasculogenesis was significantly inhibited in the
   pristane-induced SLE model ([164]Figure S5J). Compared with that in WT
   mice with pristane inducement, the size of the area occupied by
   placental blood vessels was significantly increased in the Rsad2^−/−
   mice, indicating that RSAD2 depletion alleviated placental vascular
   injury ([165]Figures S5J and S5K). Taken together, our results show
   that Rsad2 is a pathogenic ISG, which causes adverse pregnancy outcomes
   by inducing placental lipid accumulation and vascular injury in a
   pristane-induced SLE-like model.

RSAD2 triggers lipid accumulation in the spontaneous SLE pregnancy model and
SLE patients

   To further explore whether the upregulation of Rsad2 caused fetal
   demise by promoting LD accumulation in the placenta, we used New
   Zealand White (NZW) and New Zealand Black (NZB) mice as spontaneous
   murine models of SLE. Among the NZW, NZB, C57BL/6J, and BALB/c pregnant
   mice, the worst pregnancy outcomes were observed in the NZW mice
   ([166]Figures 6A and [167]S6A). The lengths and widths of NZW live
   fetuses were significantly lower, and the fetal resorption rates were
   significantly higher than those of the NZB, C57BL/6J, and BALB/c mice
   ([168]Figures 6B and 6C). We found that the expression of Rsad2 in the
   uteruses of NZW mice was also higher than in those of the NZB,
   C57BL/6J, and BALB/c mice ([169]Figure 6D). To explore whether high
   Rsad2 expression promoted the accumulation of placental LDs in NZW
   mice, whole frozen placenta sections were stained with BODIPY 493/503.
   Indeed, the LD content of the decidua and the labyrinthine zone was
   significantly higher in the NZW mice compared with C57BL/6J animals
   ([170]Figures 6E and 6F). Oil red O staining was used to further
   characterize the effect of Rsad2 expression on placental LDs in NZW
   mice ([171]Figures 6G and 6H). Hematoxylin and eosin (H&E) and
   anti-CD31 antibody IHC staining of whole placenta tissue sections
   showed evidence of vessel wall thickening and a reduction in the number
   of blood vessels in the decidua and placenta of NZW mice
   ([172]Figures S6B and S6C). We next examined whether this phenomenon
   also occurred in pregnant patients with SLE. Indeed, we found that the
   LD content was also increased in the placental villi of patients with
   SLE using LD staining ([173]Figures 6I and 6J) and oil red O staining
   ([174]Figures 6K and 6L). Thus, the aforementioned results indicate
   that the upregulation of RSAD2 induced lipid accumulation in the
   placenta, leading to blood vessel injury and ultimately causing adverse
   pregnancy outcomes in spontaneous SLE murine models and patients with
   SLE.

Figure 6.

   [175]Figure 6
   [176]Open in a new tab

   Lipid accumulation at the maternal-fetal interface in a mouse model of
   spontaneous SLE and pregnant patients with SLE

   (A) One representative litter from each of the NZW, NZB, C57BL/6J, and
   BALB/c groups of mice, collected at gd12.5. Scale bar, 1 cm. Sires and
   dams of all groups for mating have the same genetic backgrounds.

   (B) Lengths (left) and widths (right) of live fetuses NZW (n = 4), NZB
   (n = 3), C57BL/6J (n = 4), and BALB/c (n = 5) mice were recorded and
   counted.

   (C) Pie charts showing the total number of fetuses analyzed and their
   developmental phenotypes (i.e., resorbed or halted development versus
   phenotypically normal), related to (A). (A), (B), and (C) are
   representative of at least two independent experiments.

   (D) The expression of Rsad2 and Isg15 in the uteruses of NZW, NZB,
   C57BL/6J, and BALB/c mice was detected using quantitative real-time
   PCR. Each dot represents a sample from one mouse. In (B) and (D), error
   bars indicate SEM and statistical analyses were performed using the
   unpaired Student’s t test.

   (E and G) (E) LD staining and (G) oil red O staining of decidua basalis
   (DB) and labyrinth zone (LZ) of NZW and C57BL/6J mice, collected at
   gd12.5. Images are representative of three independent experiments.
   Scale bar, 50 μm.

   (F) Quantification of LD staining of placental labyrinth using ZEN 2.6.

   (H) Quantification of oil red O staining using Image-Pro Plus 6.0.

   (I and K) (I) LD staining and (K) oil red O staining of the first
   trimester deciduas and chorionic villi from each of the healthy donors
   and patients with SLE. Images are representative of three heathy donors
   and three patients with SLE. Scale bar, 50 μm.

   (J) Quantification of LD staining of placental labyrinth using ZEN 2.6.

   (L) Quantification of oil red O staining using Image-Pro Plus 6.0.

   ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not
   significant.

Virtual screening of a small-molecule compound targeting RSAD2 alleviates
APOs

   Given that RSAD2 acts as pathogenic ISGs in SLE pregnancy, virtual
   screening of two commercialized small-molecule compound libraries
   targeting the mouse RSAD2 protein structure was conducted. L-chicoric
   acid (LCA) has the highest docking score, whose structural formula is
   shown in [177]Figure 7A. LCA can form 7 hydrogen bonding interactions
   and 3 π-π interactions with mouse RSAD2 protein (As a hydrogen bond
   donor, a phenolic hydroxyl group forms 1 hydrogen bonding interaction
   with LEU252, and another phenolic hydroxyl group forms 1 hydrogen
   bonding interaction with LYS299, both at a distance of 2.0 Å; a
   carboxyl group acts as a hydrogen bond acceptor and forms 1 hydrogen
   bonding interaction with SER124 at a distance of 2.3 Å, and the oxygen
   atom on another carboxyl group acts as a hydrogen bond acceptor with
   LYS220 and ASN222, LYS247, with distances of 2.4, 1.7, and 2.4 Å,
   respectively, and ASN222 forms one hydrogen bond with the carboxyl
   oxygen atom on the main chain, with a distance of 2.8 Å)
   ([178]Figure 7B). To investigate whether LCA has a therapeutic effect
   by targeting pathogenic RSAD2, the compound was injected to pregnant
   mice after poly(I:C) treatment. The results showed that pregnancy
   outcome was significantly alleviated in the LCA-treated mice compared
   to the vehicle-treated group ([179]Figures 7C–7F), while the level of
   RSAD2 did not change in LCA-treated mice. ([180]Figures S7A and S7B).
   APOs were alleviated, and embryo resorption rate was significantly
   reduced after LCA treatment ([181]Figures 7C and 7E). Individual
   fetuses were evaluated morphologically for appearance and size by
   taking photographs with a stereoscope and measuring the crown-rump
   length, the width of the fetal head, and the body weight of live
   fetuses with a vernier caliper and an analytical balance. We found that
   the size of fetuses (especially fetal brain and liver) was more closely
   resembling normal healthy embryonic size in LCA treatment mice compared
   to the vehicle group ([182]Figures 7D and 7F). LCA treatment
   significantly restored the postnatal survival rate and reversed the
   weight loss in offspring mice from poly(I:C)-treated dams, compared to
   the vehicle control ([183]Figures S7C and S7D). The aforementioned
   results demonstrate that LCA treatment targeting RSAD2 can alleviate
   adverse pregnancy outcomes in poly(I:C)-challenged mice. While LCA
   alleviates APOs, lipid accumulation in maternal decidual and placenta
   labyrinthine zone also reduced, especially the large size of LDs
   ([184]Figure 7G). Both H&E staining and immunohistochemistry of CD31
   labeling of vascular endothelial cells of the placenta showed a
   significant improvement of vascular injury after LCA treatment compared
   with vehicle-treated mice ([185]Figures S7E and S7F). To examine
   whether LCA also has a therapeutic effect by targeting pathogenic RSAD2
   in pristane-induced SLE-like model, the compound was injected to
   pregnant mice after pristane challenge. Consistent with previous
   results, LCA did not directly affect the expression level of RSAD2
   ([186]Figures S7G and S7H), while the pregnancy outcome was alleviated
   in the LCA-treated mice compared to the vehicle-treated group
   ([187]Figures 7H–7K). Compared with the small size of individual
   fetuses observed in the vehicle group, the live fetuses had healthier
   morphology in LCA treatment mice ([188]Figure 7I). Measurement of the
   length, width, and body weight of live fetuses ([189]Figure 7K), and
   the weight of offspring mice ([190]Figure S7J), also illustrates the
   same conclusion. More to the point, LCA treatment significantly
   attenuated the rate of embryo absorption and abnormal development
   during pregnancy ([191]Figure 7J) and the survival rate of offspring
   ([192]Figure S7I). Moreover, the accumulation of LDs of placenta
   sections from LCA-treated dams was lower than that from the vehicle
   group ([193]Figure 7L). Both H&E staining and IHC of placental blood
   vessel suggested a significant improvement of vascular damage after LCA
   treatment ([194]Figures S7K and S7L). Surprisingly, LCA treatment
   reduced the proportion of embryonic resorption in NZW mice
   ([195]Figure S7M), a model of spontaneous SLE that has previously been
   shown to have serious fertility problems ([196]Figure 6C), although no
   significant improvement in embryonic size ([197]Figure S7N). In
   conclusion, a small-molecule compound LCA targeting pathogenic RSAD2
   alleviates APOs in poly(I:C)-challenged mice, pristane-induced SLE-like
   model, and NZW mice, which provides new target and therapeutic ideas
   for the clinical improvement of adverse pregnancy outcomes in women
   with SLE and other pregnancies with abnormal activated IFN-I signal.

Figure 7.

   [198]Figure 7
   [199]Open in a new tab

   Virtual screening of a small-molecule compound targeting RSAD2
   alleviates APOs

   (A) Structural formulae of LCA compound.

   (B) 3D diagram (left) of the binding pattern of LCA and RSAD2 and the
   detailed binding patterns were further analyzed (right). LCA was
   colored with rose stick. C skeleton of mouse RSAD2 protein is shown in
   green, N atoms are shown in blue, O atoms are shown in bright red, and
   H atoms are shown in white.

   (C–F) Poly(I:C) was injected at E10.5. 10 mg/kg LCA and vehicle were
   intraperitoneally injected at E10.5 and E11.5. (C) Representative
   litter from vehicle-treated and LCA-treated groups of mice, collected
   at gd12.5. Scale bar, 1 cm. (D) One representative litter from each of
   the indicated treatment groups collected at gd12.5 is shown. Scale bar,
   1 mm. (E) Pie charts showing the total number of fetuses analyzed and
   their developmental phenotypes (i.e., resorbed or halted development
   versus phenotypically normal), related to (C). Saline-vehicle (n = 4),
   saline-LCA (n = 4), poly(I:C)-vehicle (n = 4), and poly(I:C)-LCA (n =
   4) were recorded. (F) Lengths (left), widths (middle), and weights
   (right) of live fetuses from indicated each group were recorded and
   measured. (C)–(F) are representative of at least two independent
   experiments.

   (G) Oil red O staining of decidua basalis (DB) and labyrinth zone (LZ)
   of LCA or vehicle-treated mice with same poly(I:C) treatment, which is
   harvested at E12.5. Representative image of placentas from at least
   three dams. Scale bar, 50 μm.

   (H–K) 10 mg/kg LCA was injected at E10.5 after induced with 500 μL
   pristane or same volume saline, and vehicle was injected to control
   mice at the same time.

   (H) Representative litter from vehicle-treated and LCA-treated groups
   of mice, collected at gd12.5. Scale bar, 1 cm.

   (I) One representative litter from each of the indicated treatment
   groups collected at gd12.5 is shown. Scale bar, 1 mm.

   (J) Pie charts showing the total number of fetuses analyzed and their
   developmental phenotypes (i.e., resorbed or halted development versus
   phenotypically normal), related to (H). 6–7 dams per group enrolled.

   (K) Lengths (left), widths (middle), and weights (right) of live
   fetuses from indicated each group were recorded and measured.
   Saline-vehicle (n = 3), saline-LCA (n = 3), pristane-vehicle (n = 5),
   and pristane-LCA (n = 7) were recorded. (H)–(K) are representative of
   two independent experiments.

   (L) Oil red O staining of DB and LZ of LCA or vehicle-treated mice with
   pristane inducement, collected at gd12.5. Images are representative of
   three independent experiments. Scale bar, 50 μm.

   In (F) and (K), error bars indicate SEM, and statistical analyses were
   performed using the unpaired Student’s t test. ∗p < 0.05, ∗∗p < 0.01,
   ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Discussion

   Our work examined SLE pregnancy as a model and revealed that RSAD2 acts
   as a pathogenic ISG. Unlike the protective ISGs, RSAD2 promotes
   excessive DAG-lipid accumulation at the maternal-fetal interface,
   causing placental vascular damage and poor pregnancy outcomes. In
   addition, LCA, as a new inhibitor targeting RSAD2 by virtual screening,
   significantly alleviated APOs and lipid accumulation in three pregnant
   mice model with dysregulated IFN-I signaling. Thus, RSAD2 may represent
   a promising target in the treatment of adverse pregnancy outcomes
   caused by an abnormally activated IFN-I response.

   Anifrolumab, a human monoclonal antibody to IFN-I receptor subunit 1,
   has been investigated for the treatment of SLE.[200]^34 The clinical
   trial showed that patients who received anifrolumab experienced a
   positive outcome. However, there was a higher incidence of herpes
   zoster and bronchitis.[201]^35 The trial assessing the long-term safety
   of anifrolumab revealed higher exposure-adjusted event rates of
   COVID-related adverse events, including asymptomatic infections,
   compared to the placebo.[202]^36 It is suggested that distinguishing
   between pathogenic and protective ISGs offers the potential for the
   precise treatment of SLE or SLE-related pregnant disorders.

   Several studies have demonstrated that the abnormal activation of
   placental IFN-I signaling leads to adverse pregnancy outcomes during
   viral infection.[203]^2^,[204]^3^,[205]^15 A previous study found that
   IFITMs inhibited placental syncytial trophoblast fusion, leading to
   embryonic loss.[206]^2 This provided a possible explanation for how
   IFN-I and its downstream pathogenic ISGs mediated placental dysfunction
   in a mouse pregnancy model stimulated by the IFN-I inducer poly(I:C).
   In our study, we found that RSAD2 caused placental damage by perturbing
   lipid metabolism at the maternal-fetal interface and led to placental
   vascular injury in SLE. In addition, poly(I:C) injections at different
   gestational stages resulted in different changes in pregnancy outcomes.
   Existing literature shows that starting poly(I:C) injections at
   E7.5—before placental formation—leads to nearly complete embryo
   absorption.[207]^2 Injection of poly(I:C) typically results in a low
   probability of embryonic loss after E12.5.[208]^37 Based on the adverse
   pregnancy outcomes observed in clinical cases of SLE and SLE mouse
   models, we opted to administer poly(I:C) at E10.5 to better simulate
   adverse outcomes such as stillbirth and intrauterine growth
   restriction.

   RSAD2 encodes an IFN-inducible protein, RSAD2 (also named viperin,
   CIG5, CIG33), which is a member of the radical S-adenosyl-L-methionine
   superfamily of enzymes. RSAD2 inhibits the replication of multiple
   viruses by producing the ribonucleotide
   3′-deoxy-3′,4′-didehydro-cytidine
   triphosphate.[209]^38^,[210]^39^,[211]^40 However, RSAD2 facilitates
   the infectious process by disrupting cellular lipid metabolism during
   infection with human cytomegalovirus. Meanwhile, RSAD2 was relocalized
   from the endoplasmic reticulum to the mitochondria.[212]^41^,[213]^42
   RSAD2 is considered to be a key diagnostic ISG in SLE and may
   contribute to the personalized targeted treatment of this
   disease.[214]^43 Moreover, RSAD2 is significantly upregulated in
   atherosclerosis, where its expression is regulated by proinflammatory
   agents (i.e., by lipopolysaccharide, cytomegalovirus, and IFN-γ but not
   by tumor necrosis factor alpha or IL-1β).[215]^44 RSAD2 is expressed in
   the atherosclerotic lesions of humans and Apoe^−/−mice but not in
   normal arteries.[216]^44 RSAD2 expression is also observed in various
   types of cancer, where RSAD2-mediated metabolic alterations drive the
   progression of tumors.[217]^45 The aforementioned studies suggest that
   RSAD2 is a pathogenic ISG in pregnant patients with SLE and others with
   an overactive IFN-I response. In this study, we reported a novel role
   for RSAD2 in the regulation of cellular metabolism at the
   maternal-fetal interface in individuals with a disease that affected
   pregnancy outcomes. We found that RSAD2 may lead to abnormal embryonic
   dysplasia by inducing the accumulation of sn-1,2-DAGs in the placenta.

   Through virtual screening, we found one small-molecule compound
   targeting RSAD2. As a dicaffeoyltartaric acid, LCA is a potent,
   selective, and reversible HIV-1 integrase inhibitor.[218]^46 Chicoric
   acid is isolated and purified from plant and vegetables, which is
   reported to possess antioxidant and anti-inflammatory
   activities.[219]^47 Given that LCA has effects on alleviating adverse
   pregnancy outcomes in treating the SLE mice models and its natural
   product properties, small-molecule compound therapy targeting RSAD2 has
   potential clinical applications. In addition, elevated RSAD2 has been
   observed in atherosclerotic plaques, where it is associated with lipid
   accumulation and foam cell formation, exacerbating the disease.[220]^48
   This is similar to the phenomenon observed in the placental labyrinth
   zone of our SLE pregnancy model, expanding the potential role of RSAD2
   as a target in vascular inflammatory diseases and the therapeutic value
   of LCA.

   Interestingly, we found that IFN-I stimulation upregulated APOD in the
   decidua ([221]Figure S4H), while its receptor-encoding genes LRP1 and
   SR-B1 were highly expressed in the placental region (data not showed).
   The APOD-LRP1/SR-B1 signaling axis may mediate lipid transfer from the
   decidua to the placenta. Our previous research found that decidual NK
   cells used the APOD-LRP1 signaling axis to deliver lipids to
   trophoblasts, which helped trophoblasts resist apoptosis (induced by
   intracellular bacterial infection) and promoted their survival.[222]^49
   In an IFN-I-rich environment, the upregulation of APOD expression in
   DSCs could further transport lipids from the decidual tissue to the
   placental trophoblast layer, leading to the abnormally excessive
   accumulation of LDs in the placenta. In patients with atherosclerosis,
   macrophages induced vascular injury by phagocytosing lipids.[223]^50
   Whether placental macrophages are responsible for the induction of
   placental injury via disrupting lipid metabolism deserves further
   investigation, and whether RSAD2 contributes to the proatherogenic
   effects of IFN-α in SLE also deserves further investigation.
   Ultimately, our work provides new insights into how targeting
   pathogenic ISGs may benefit pregnant patients with diseases associated
   with dysregulated IFN-I signaling.

Limitations of the study

   Although we have shown that RSAD2 is mainly expressed in structural
   cell subpopulations using purified cells, we used full rather than
   conditional knockout mice based on the fact that RSAD2 is broadly
   expressed at the maternal-fetal interface. Our results cannot rule out
   the possibility that RSAD2 plays a role at other sites in SLE, such as
   the spleen and kidney. In addition, RSAD2 plays an important antiviral
   infection role by inhibiting viral replication.[224]^39 Thus, RSAD2
   deficiency may not be beneficial in some mouse models of viral (i.e.,
   ZIKA) infection-induced APOs. The factors contributing to adverse
   pregnancy outcomes in patients with SLE are complex, including
   antinuclear antibodies, complement proteins, inflammatory cytokines,
   and vascular factors.[225]^51 Our study explains the molecular basis of
   risky pregnancies associated with an overactive IFN-I response in the
   SLE model.

Resource availability

Lead contact

   Further information and requests for resources and reagents should be
   directed to and will be fulfilled by the lead contact, Haiming Wei
   (ustcwhm@ustc.edu.cn).

Materials availability

   Plasmids generated in this study are available from the [226]lead
   contact without restriction.

Data and code availability

     * •
       RNA-seq data files, quasi-targeted metabolomics, and widely
       targeted quantitative lipidomics are available in the GSA or OMIX
       database repository of the National Genomics Data Center, China
       National Center for Bioinformation. These data are publicly
       available as of the date of publication. Accession numbers are
       listed in the [227]key resources table.
     * •
       This study did not report new original code.
     * •
       Any additional information required to reanalyze the data reported
       in this paper is available from the [228]lead contact upon request.

Acknowledgments