SCARF1 deficiency exacerbates gut inflammation and autoimmune pathology

Abstract

Systemic lupus erythematosus (SLE) is a complex autoimmune disease known for its heterogeneity in both manifestation and presentation. Recent evidence has increasingly implicated the gut microbiome within immunomodulation and autoimmunity. This study aims to characterize the intestinal inflammation and microbial profile associated with autoimmune diseases, particularly SLE, and to identify unique biomarkers and shared microbial signatures for potential therapeutic measures. Our lab identified scavenger receptor class F, member 1 (SCARF1, SREC-1) as an efferocytosis receptor essential for the clearance of apoptotic debris, and its deficiency results in the development of lupus-like disease. SCARF1 is crucial in immune homeostasis, and defects in efferocytosis lead to inflammation. However, the role of SCARF1 in gut homeostasis remains to be elucidated. To answer our question, we analyzed and compared the metagenomic datasets generated through whole genome shotgun sequencing between our Scarf1^−/− lupus-prone mouse model and healthy counterparts. We found that Scarf1^−/− mice had significantly lengthened intestines, elevated immune cell infiltration, and structural changes in the colon. Microbiome analysis revealed gut dysbiosis, including reduced alpha diversity and increased Firmicute/Bacteroidetes ratio. Notably, beneficial taxa such as Akkermansia muciniphila was absent in Scarf1^−/− mice. Linear regression analysis identified positive associations between lupus disease severity and increased abundances of Alistipes, Lachnospiraceae, and Clostridium. Function analysis of the gut microbiome in Scarf1^−/− mice indicated downregulation of multiple pathways related to cell proliferation. These findings highlight the role of SCARF1 involvement in the gut microbiome and immune regulation in the context of inflammation and SLE.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-39378-7.

Introduction

The innate immune system is the first line of defense following a pathological insult¹. During infection, pattern recognition receptors (PRRs) identify pathogen associated molecular patterns (PAMPs) to initiate an immune response². This leads to an intracellular signaling cascade with the end results of eliminating the pathogen. The gut contains 70–80% of the body’s immune cells due to its susceptibility towards antigen exposure through pathogens, diet, and the gut microbiome³. Nonetheless, humans have a variety of microbes that reside in our bodies that contribute to homeostasis⁴.

The mammalian immune system co-evolved with commensal microbes. Comprising the gut microbiota, these commensal microbes not only contribute towards shaping the immune system during neonatal development, but also directly participate to protect us from infection³. Recent research increasingly highlights the gut microbiome’s complex and multifaceted role in the immune system⁵. This includes, but is not limited to, (1) providing colonization resistance through competition between commensal and pathogenic bacteria; (2) interactions with the intestinal epithelium and mucosal membranes to regulate mucus production, immune signaling, and epithelial barrier integrity; and (3) producing microbial metabolites that act as immune-protective signals, influencing T-cell differentiation in the gut and activating innate immune cells, such as neutrophils and macrophages, beyond the gut⁶.

Dysbiosis, or an imbalance of the microbiota, disrupts immune homeostasis and can induce an inflammatory response⁷. Gut dysbiosis has been linked to multiple diseases, including inflammatory bowel disease (IBD), cancer and diabetes^8,9. Recent evidence increasingly suggests that gut microbiota dysbiosis contributes to the development and pathogenesis of systemic lupus erythematosus (SLE)¹⁰. SLE is a complex and heterogeneous disease with a wide range of manifestations known to impact multiple organ systems, such as the gastrointestinal (GI) tract¹¹. Side effects of medications and increased susceptibility to infection may disrupt the balance of immune activation, ultimately impacting immune homeostasis¹². Nonetheless, host genetics also play a significant role in shaping the composition and function of the microbiome¹³.

Over the last decade the impact on the gut microbiota and the development of lupus has attracted the attention of researchers. However, the link between the gut microbiota and the development of lupus remains unclear. Due to the heterogeneity of SLE, each mouse model offers a unique set of advantages and limitations¹⁴. One of the most prominent mouse model strains, NZB/NZW F1 (NZB), have been shown to have a significantly increased gut microbiota correlating with disease progression^15,16. Specifically, it was reported that the gut microbiota changed significantly after the onset of Lupus in NZB mice¹⁷. Contrary, MRL/lpr shows an inverse correlation between the relative abundance of the beneficial bacteria Lactobacillus and inflammation¹⁸. Supplementation with Lactobacillus on MRL/lpr mice resulted in lower inflammatory cytokines and an increase in the anti-inflammatory response, as measured by the presence of autoantibody levels¹⁹. The study of a Toll Like Receptor (TLR)−7, also known as TLR-7.Tg, dependent lupus mouse model, identifies the presence of Lactobacillus reuteri in these mice enhancing autoimmune manifestations and the increase of inflammation²⁰. Further, the gut microbiota from the lupus-prone mice B6.sle1sle2sle3, also known as triple congenic (TC), described a change in the microbiota after the onset of autoimmune disease²¹. The group found that in TC mice, the increased inflammation was a result of altered microbiota resulting in metabolic changes²². Although multiple lupus mouse models show gut microbiome changes, it is unclear what molecular drivers influence these variations.

Scavenger receptors (SR) are considered a subset of PRRs, as they can identify a broad spectrum of ligands including apoptotic cells and microbial components^23,24. One member of the SR family, SCARF1 (scavenger receptor class F member 1, also known as SR-F1 or SREC1) was identified as a non-redundant efferocytosis receptor and as a lupus-prone mouse model²⁵. Scarf1-deficiency leads to the accumulation of dying cells in tissues, resulting in a lupus-like disease characterized by spontaneous autoantibody production against chromatin, immune cell activation, dermatitis, and nephritis. Our initial studies addressed several gaps in understanding SCARF1-mediated regulation of autoimmunity²⁵. In this study, we aim to expand our knowledge about the role of SCARF1 by investigating the gut microbiome and gut inflammation in Scarf1^−/− lupus-prone mice. Using shotgun sequencing, we compared fecal pellets from Scarf1^−/− and Scarf1^+/+ and identified changes in specific bacterial species. Together, our findings provide new insights into bacterial species associated with inflammation and SLE.

Results

Scarf1 deficiency results in enhanced gut inflammation and defects in efferocytosis

We previously showed that Scarf1 is responsible for the removal of apoptotic debris, and mice deficient in this receptor develop lupus-like disease²⁵. Our group observed that Scarf1^−/− mice exhibit a significantly longer gut when compared to wild-type mice, including colon length (Fig. 1A-B). Gut microbiome dysbiosis has often been linked to not only inflammation, but also anatomical and structural changes in the gastrointestinal organ system²⁶. We aimed to assess whether Scarf1^+/+ and Scarf1^−/− mice exhibited notable changes in gut inflammation and microbiome composition. We assessed the development of autoimmunity through the presence of antinuclear autoantibodies (Sup Fig. 1-A-B). Consistent with earlier work, Scarf1^−/− mice develop autoimmunity with multiple clinical manifestations, including alopecia, loss of whiskers, (Sup Fig. 1C-D) and nephritis (Sup Fig. 1E) by 20-weeks of age. Histological analysis of Scarf1-deficient murine colons revealed increased cell infiltration and structural alterations (Fig. 1C, bottom panels), while pathological analysis identified increased numbers of neutrophils and apoptotic cells (Fig. 1D). Increased gut inflammation has been linked to the development of obesity and related metabolic conditions²⁷. However, based on prior work and new data, we did not observed statically significant difference between Scarf1^+/+ and Scarf1^−/− mice weight and survival up to one year suggesting the presence of additional pathways that are able to partially compensate for Scarf1 deficiency in the development of autoimmunity²⁵ (Sup Fig. 5).

Fig. 1.

Scarf1^−/− mice exhibit enhanced gut inflammation. A) Scarf1^−/− gut is enlarged when compared to wild-type mice. Representative image of gut measurement. B) Quantification of gut length (cm) in wild-type vs. Scarf1^−/− mice. C) Increased inflammation in Scarf1^−/− gut tissue. Hematoxylin-and-eosin staining of fixed cryosections of duodenum, Ileum, and colon of 20-week-old female Scarf^+/+ and Scarf1^−/− mice (n = 3 per group). Original magnification, ×20. Scale bar 100 μm D) Pathology score. Slides were scored blinded, and the grade of inflammation as follows 1- minimal 2- mild, 3-moderate, 4-severe, 5-severe with ulceration. (n = 3 mice per group). Data was analyzed using 2-Way ANOVA. E) Fluorescence microscopy of Duodenum, Ileum and Colon sections from > 20-week-old Scarf1^+/+ and Scarf1^−/−mice stained by phalloidin (TRITC, red), EPCAM-1 (Alexa fluor 647, white) and DAPI (Alexa Fluor 421, blue) and apoptotic cells by TUNEL (FITC, green). Representative image. 3 independent experiments with n = 3 mice per group. Scale bar 170 μm at 20x magnification. F) Quantification of images at left by automated analysis of staining with the nuclear dye DAPI and by TUNEL. Statistical values ns = not significant, *p < 0.01, **p < 0.005 by Student’s t-test.

Fig. 5.

SLE-like disease score correlates with harmful relative microbiota abundance. A) Heat map of disease scores in 20-week-old female Scarf1^+/+ and Scarf1^−/− mice (n = 6 mice per group). The disease score represents the average of blinded assessments for ANA immunofluorescence on Hep-2 cells, nephritis, alopecia, and apoptotic cell counts in the duodenum, ileum, and colon. B-J) Multiple comparison of SLE-like disease and relative abundance of bacteria taxonomic levels. Statistical values are measured using simple linear regression analyses between individual bacterial species and disease score. N = 6 mice per group. Data are from two independent experiments. B) Akkremencia municiphila, C) Bacteroidales bacterium, D) Dubosiella, E) Lactobacillus johnsonii, F) Clostridium difficile, G) Porphyromonadaceae bacterium, H) Lepagella muris, I) Allistipes, J) Lachnospiraceae bacterium. Statistical values by simple linear regression.

Our previous work has shown that scavenger receptor, SCARF1, mediates the efferocytosis of apoptotic cells (ACs) in mice in a non-redundant manner and dysregulation of SCARF1 leads to the accumulation of ACs^25,28. We next aimed to investigate whether Scarf1-deficiency leads to an accumulation of apoptotic debris in the gut. Using fluorescent microscopy, we detected a significant increase in cellular debris in the Scarf1^−/− mice (Fig. 1E-F). Our data confirms that Scarf1 is essential for the removal of apoptotic debris and suggests a potential role for Scarf1 in the mediation of tissue homeostasis.

The gut Microbiome in Scarf1^−/− mice exhibits reduced alpha diversity and an increased Firmicutes/Bacteroidetes (F/B) ratio

The gut microbiota is sensitive to changes in the tissue homeostasis²⁹. We compared the fecal microbiome of 20-week-old Scarf1^+/+ and Scarf1^−/− mice (Fig. 2). Our data shows reduced alpha diversity in the Scarf1^−/− mice in both female and male animals when compared to Scarf1^+/+ mice (Fig. 2A-B). Changes in the Firmicutes/Bacteroidetes (F/B) ratio are a biomarker for gut dysbiosis^30–32. Assessing F/B ratio across strains and sexes, we uncovered a significant dysbiosis in the Scarf1^−/− female mice, through a notably marked increase in Firmicutes and decrease in Bacteroidetes (Fig. 2C). Interestingly, we noted that Scarf1-deficient males reveal a F/B ratio similar to wild-type mice, although still trending higher (Fig. 2C). Our data corroborate the large sex difference observed in the development of autoimmune disease²⁵. In SLE, there is a predisposition for the female population of 9:1 incidence ratio to develop lupus in comparison to males³³. Nonetheless, a significant difference between wild-type and Scarf1-deficient mice was present. Our data shows statistically significant clusters (p = 0.002) within our Principal Coordinates Analysis (PCoA) of bacterial beta-diversity as analyzed by Bray-Curtis dissimilarity between Scarf1^+/+ and Scarf1^−/−. This dissimilarity is independent of sex (Fig. 2D).

Fig. 2.

Lower alpha diversity and increased Firmicute/Bacteroidetes (F/B) ratio in female Scarf1^−/− mice. A-B) Alpha diversity of fecal microbiota from wild-type and Scarf1^−/− mice analyzed by A) Shannon and B) Simpson diversity indices. N = 32 mice, 8 mice per group. C) F/B ratio calculated from normalized read counts. N = 54 mice (Female mice n = 16 mice per group, Male mice n = 11 mice per group) D) Principal Coordinates Analysis as shown by Principal Components (PC) plot of bacterial beta-diversity based on Bray–Curtis dissimilarities. Data represent two combined independent experiments. Statistical analysis: ** p < 0.03 **** p < 0.001 by One-way ANOVA. Bray-Curtis p = 0.002 comparing Scarf1^+/+ vs. Scarf1^−/− by PERMANOVA (Permutational Multivariate Analysis of Variance).

Due to the coprophagic behaviors of mice, we next investigated whether co-housed wild-type and Scarf1^−/− mice would lead to fecal microbial self-reinoculation and subsequent alterations in the microbiome³⁴ (Supp Fig. 2 and Supp Fig. 3). To determine whether co-housing affects the development of autoimmune, we collected serum from female mice after 14-days of co-housing. As expected, we observed the presence of autoantibodies, as examined by antinuclear antibody (ANA) immunofluorescence study, only in the Scarf1^−/− mice (Supp Fig. 2A). However, a significant difference in the alpha diversity between wild-type and Scarf1-deficient mice within control and co-housed groups was found (Supp Fig. 2B-C), suggesting the presence of fecal microbiota transplantation. In addition, Bray-Curtis dissimilarity analysis indicated that co-housed mice clustered together, suggesting that co-housing impacts microbiome composition (Supp Fig. 2D). Taxonomic analysis revealed a reduction in Akkermansia and Porphyromenadaceae, and an increase in Alistipes in wild-type mice (Supp Fig. 2E). No significant differences were observed in Firmicutes (also known as Bacillota) or F/B ratio between co-housed and control mice (Supp Fig. 2F-G). Because of the length of the co-housing, we did not observe transfer of autoimmunity from Scarf1^−/− to Scarf1^+/+. As expected, co-housing male mice (Supp Fig. 3) does not exhibit gut microbiome changes between Scarf1^+/+ and Scarf1^−/−. Comparing alpha diversity between control and co-housed mice (Supp Fig. 3A-B) shows no difference between conditions. Further, in the Scarf1^−/− mice, a shift towards a decreased Firmicutes/Bacteroidetes (F/B) ratio was observed, but the data is not significant (Supp Fig. 3C). Bray-Curtis dissimilarity analysis indicated the co-housed mice do not cluster together (p value 0.166), implying that co-housing male mice have no effect on microbiome composition. This can be observed by the taxonomic analysis (Supp Fig. 3E), where the only change we see is the presence of Alistipes MGBC116833 in the WT co-housed mice while the rest of the species remain the same across the different conditions. Deficiency in SCARF1 will affect immune homeostasis and mutations of the receptor will affect ligand-receptor interactions³⁵. Therefore, based on our data we can conclude that the observed dysbiosis in Scarf1-deficient female mice is likely attributed to the absence of the SCARF1 receptor itself and the associated inflammation.

Fig. 3.

Female wild-type mice have a higher alpha diversity and reduced F/B ratio compared to littermates. Female F2 littermates (Scarf1^+/+, Scarf1^−/+ and Scarf1^−/−) were co-housed for 20 weeks. A) Scarf1^−/+ develop autoimmunity. A) Antinuclear antibody (ANA) immunofluorescence of HEp-2 cells using serum from 20-week-old female Scarf1^+/+, Scarf1^−/+ and Scarf1^−/− mice. Total mice n = 24 for two independent experiments (n = 3 mice per group/experiment). Quantification by frequency of serum ANA. Microscopy is a representative image. B) Scarf1 is necessary for the clearance of apoptotic debris. Fluorescence microscopy of Duodenum, Ileum and Colon sections from > 20-week-old Scarf1^+/+, Scarf1^−/+and Scarf1^−/−mice, stained with phalloidin (TRITC, red), EPCAM-1 (Alexa fluor 647, white) and DAPI (Alexa Fluor 421, blue) and apoptotic cells by TUNEL (FITC, green). Representative image. Representative image of 3 independent experiments. 10x magnification, Scale bar 170 mm. Quantification of images at left by automated analysis of staining with the nuclear dye DAPI and by TUNEL. C-D) Alpha diversity of fecal microbiota is lower in Scarf1-/+ and Scarf1^−/−. Alpha diversity was assessed by C) Shannon and D) Simpson diversity indices Scarf1^+/+and Scarf1^−/− n = 8 female mice per group; Scarf1^−/+ n = 15 mice per group. E) Wild-type mice have a lower F/B ratio. F/B ratio calculated from normalized read counts. F) Principal Coordinates Analysis as shown by Principal Components (PC) plot of bacterial beta-diversity based on Bray–Curtis dissimilarities. Data represent two independent experiments combined. G-H) Comparison of gut microbiota profiles between 20-week-old female wild-type, Scarf1^−/+Scarf1^−/− mice. Whole genome shotgun sequencing was performed by Transnetyx and analyzed using OneCodex. Scarf1^+/+, and Scarf1^−/− n = 8 female mice per group; Scarf1^−/+ n = 15 mice. G) Comparison of species, H) Comparison of Phylum. Statistical analysis NS = not significant, *p < 0.05, *** p > 0.001, **** p < 0.0001 by One-way ANOVA. Bray-Curtis statistical analysis p = 0.35 as measured by PERMANOVA.

Wild-type littermates exhibit greater relative bacterial taxa and higher alpha diversity

Since co-housing female mice for 2 weeks did indeed show an impact on the microbiome, we next asked whether a longer timeline would have a clear impact on the clinical development of autoimmune disease. To address this question, F2 littermate mice were co-housed for a minimum of 20-weeks. 50% of heterozygous mice (Scarf1^−/+) developed low levels of autoimmunity, as examined by ANA immunofluorescence study (Fig. 3A). Like our SCARF1-deficient mice, Scarf1^−/+ exhibit an accumulation of apoptotic debris (Fig. 3B). The accumulation of debris is lower than when compared to the full knock-out strain; however, we notice the presence of a cluster of apoptotic debris in the colon of the Scarf1^−/+ mice (Fig. 3B). Scarf1^−/+ mice had a higher F/B ratio (Fig. 3E); however, no difference in diversity between Scarf1^−/− and Scarf1^−/+ was identified through Shannon nor Simpson index measurements (Fig. 3C-D). The presence of AC clusters in Scarf1-/+ mice was only observed in two out of five mice analyzed for the presence of apoptotic cells; however, we noticed that they correlated with the higher number of the F/B ratio (Fig. 3E). Scarf1^+/+ mice display a more diverse microbiota compared to Scarf1^−/− or Scarf1^−/+ (Fig. 3C-D), suggesting even a partial deficiency of Scarf1 could alter gut microbial homeostasis. Nonetheless, Bray-Curtis dissimilarity analysis of Scarf1^−/−, Scarf1^−/+ and Scarf1^+/+ mice showed some clustering, suggesting diverse microbial taxa across the Scarf1^+/+ and Scarf1^−/− mice. However, Scarf1^−/+ did not cluster together. Instead, we observed that Scarf1^−/+ mice are distributed among both Scar1^+/+ and Scarf1^−/− (Fig. 3F). Using PERMANOVA to measure the dissimilarities between groups, our data shows non-significant differences between the different genotypes (p value 0.35), suggesting the presence of a shared microbial taxa with Scarf1^+/+ and Scarf1^−/− (Fig. 3F). Relative abundance data comparing Scarf1^−/−, Scarf1^−/+ and Scarf1^+/+ mice indicates a higher relative taxa for wild-type mice when compared to Scarf1^−/− and Scarf1^−/+ (Fig. 3G). We observed that all littermates faced a reduction in Akkermasia and an increase in Closterium (Fig. 3G-H); however, wild-type mice experienced an increase in other beneficial bacteria such as Bacteroidota, a phylum known for maintaining gut health.

Healthy mice express higher levels of Akkermansia

Our data indicate that beneficial bacteria are reduced or absent in Scarf1-deficient mice, prompting us to investigate which other beneficial bacterial species may also be reduced or absent in Scarf1^−/− mice. To answer this question, we compared the top 20 microbiota species between Scarf1^+/+ and Scarf1^−/−. Overall, bacterial species abundance was lower in the Scarf1^−/− mice when compared to wild-type controls (Fig. 4A-B). Examining the top 20 species, we observed higher abundance of Akkermansia muciniciphila, Dubosiella sp004793885 and Bacteriodales bacterium M2 in wild type animals compared to Scarf1^−/− mice (Fig. 4). As expected, we observed an increase in Firmicutes (Bacillota) in Scarf1^−/− mice (Fig. 4C). Our data also show decreased levels of Akkermansia (Fig. 4D), Dubosiella (Fig. 4E), and Bacteroidales (Fig. 4F), along with increased abundances of Alistipes (Fig. 4E) and Duncaniella (Fig. 4F) in Scarf1^−/− mice. Akkermansia muciniciphila is associated with a healthy gut by stimulating metabolic and immune responses³⁶. Interestingly, wild-type mice co-housed with Scarf1^−/− mice for 2 weeks showed decreased levels of Akkermansia (Supp Fig. 2E). However, the co-housed wild-type mice did not develop autoimmune disease, as assessed by ANA analysis (Supp Fig. 2A).

Fig. 4.

Healthy bacterial species are present in wild-type mice but absent in Scarf1^-/- mice. A-C) Scarf1^-/- mice express gut microbiota dysbiosis at 20-weeks of age. Comparison of gut microbiota profiles between 20-week-old female wild-type and Scarf1^-/- mice. Whole genome shotgun sequencing was performed by Transnetyx, and data was analyzed using OneCodex. N = 21 female mice per group; n = 9 wild-type and n = 12 Scarf1^-/- mice total. A) Relative abundance of the top 15 species. B) Heat map of the top 15 genera. C) Mean relative abundance of the top 10 phyla. D-H) Beneficial bacteria are absent is Scarf1^-/- mice. Levels of specific microbial communities measured by shotgun sequencing as described above. D) Akkermansia muciniphila, E) Dubosiella sp004793885, F) Bacteroidales bacterium M2, G) Alistipes MGBC116833, H) Duncaniella MGBC142302. Data represent two independent experiments, N = 13 mice per group. Statistical significance: **** p < 0.0001,*** p < 0.001, ** p < 0.01, by One-way ANOVA and Mann-Whitney test.

Loss of specific bacterial species is associated with autoimmunity

An increasing number of studies suggest that the gut microbiota is involved in the initiation and progression of inflammatory and autoimmune diseases³⁷. To assess this, we investigated whether we could identify specific species potentially associated with disease development. We developed a disease score (Fig. 5A and Supp Fig. 1) based on the levels of ANA staining, alopecia, nephritis (as measured by glomerular inflammation), and the average number of apoptotic cells as shown by a heatmap. Using a simple linear regression, we analyzed the relationship between bacterial species abundance of species and disease score (Fig. 5B-J). Healthy mice exhibit significantly higher levels of Akkermansia (Fig. 5B), Bacteroides (Fig. 5C), and Dubosiella (Fig. 5D) in contrast to Scarf1^−/− mice which exhibit significantly increased levels of Lepagella muris (Fig. 5H), Alistipes (Fig. 5I), Porphyromonadaceae (Fig. 5G), Lachnospiraceae (Fig. 5E) and Hominisplanchenecus (not shown). While not statistically significant, we noticed a trend toward increased abundance of Lactobacillus (Fig. 5H), Porphyromonadaceae (not shown), Clostridium (Fig. 5F) and Heminphilus (not shown) in Scarf1^−/− mice. Together, these findings suggest that SCARF1 deficiency decreases beneficial bacterial populations, leading to an altered gut microbiome and potentially contributing to the gut inflammation and autoimmune pathology.

Gut functional analysis suggests a decrease in host “internal component of membrane pathway” and an increase in “regulation of cell proliferation pathway” in Scarf1-/- mice.

Microbiome changes affect both the metabolic and immune systems³⁸. To gain more insight into the microbiome-host interaction, we performed a functional gene pathway analysis on the bacterial species (Fig. 6). In this functional analysis we can identify gene data present in the metagenomic data. Using GeneOntogoly (GO) Consortium and KEGG database^39–41, we noticed a decrease in the host “integral component of membrane” in Scarf1^−/− mice (Fig. 6A, Supp Fig. 3A). The integral membrane is composed by proteins to compose an impermeable membrane that only allows the transport of metabolites and communication of information to pass the membrane⁴². This data suggests that in Scarf1^−/− mice, the membrane lacks components necessary for maintaining its permeability. In turn, we also observed an increase in the “positive regulation of cell proliferation” (Fig. 6A, Supp Fig. 3B). Further GO gene analysis identified the gene as “upregulator of cell proliferation” (urgpc) in mice, and this gene is responsible for nucleotide binding⁴³.The function of this gene is to stimulate cell growth⁴⁴. However, our data shows that accumulation of apoptotic debris may be a contributor to the intestinal changes observed in Scarf1^−/− when compared to wild-type controls. Further pathway analysis showed that Scarf1^−/− mice have decreased levels of L-valine biosynthesis, L-isoleucine biosynthesis and adenosine ribonucleotide de novo biosynthesis compared to wild-type (Fig. 6B). These compounds are essential in the ability to produce branched-chain fatty acids (BCFA), which are essential lipid membrane components of the gut bacteria⁴⁵. Furthermore, these compounds have been shown to protect against intestinal damage⁴⁶. Alterations in the biosynthesis of BCFA have been linked to dysbiosis⁴⁷, further suggesting that changes in lipids can affect homeostasis. As described above, we found that Akkermansia is present only in the microbiota of Scarf1^+/+ mice (Fig. 6C-D, refer Figs. 4D and 5B). Studies have shown that A. muciniphila is involved in regulating lipid metabolism and modulating the immune response by reducing inflammation⁴⁸. Taken together, our data demonstrate distinct microbiome and inflammation profiles between Scarf1^+/+ and Scarf1^−/− mice.

Fig. 6.

Gut functional analysis shows a decrease in genes of host internal component of the membrane in Scarf1^-/- mice. A) Top 20 GO Term analysis (https://geneontology.org) in gut samples as copies per million abundances of gene families comparing Scarf1^+/+ and Scarf1^-/- mice. B) Pathway analysis of gut samples as copies per million comparing Scarf1^+/+ and Scarf1^-/- mice. Data shows biochemical pathways represented by a probability that an entire pathway’s enzymes are present in the sample. C-D) Flow diagram of functional group annotated ad the gene groups and the relationship to taxonomy. GO aspect is organized in: Molecular Function (MF), Cellular Component (CC), and Biological Process (BP). C) Scarf1^+/+ and D) Scarf1^-/-. Representative analysis of n = 6 Scarf1^+/+ and n = 10 Scarf1^-/- female mice.

Discussion

Dysbiosis of the gut microbiome has been linked to the development of SLE^49,50. Numerous factors have been studied associating dysbiosis with the development of SLE. Side effects of medications, increased susceptibility to infection, and tissue damage are a few examples that lead to the impairment of gut homeostasis^12,51. However, the impact of genetic SLE-associated risk factors on gut dysbiosis is less understood.

Here, we characterize the gut microbiome in a mouse model of spontaneous lupus. Using the lupus-prone Scarf1^−/− mouse, we observed a significant size increase in the colon size when compared to wild-type mice. This observation was striking, as typically an inflamed gut is shorter in size^52,53. Given SCARF1’s role as an efferocytosis receptor^25,54, the accumulation of uncleared apoptotic cells and subsequent inflammatory immune and epithelial stimulation in diseased mice may suggest a possible explanation for the increase in colon size.

The absence of SCARF1 disrupts immune homeostasis²⁵, leading us to assess inherent changes in composition and diversity of gut microbiota between control and diseased mice. Wild-type mice express a higher diversity and a richer bacterial composition when compared to Scarf1^−/− mice and Scarf1^−/+ mice. Data also shows differences that are sex specific, with diseased female mice exhibiting low diversity and an increased F/B ratio. Similar observations indicating significant sex-specific differences in gut mucosa have been shown using SWR × NZB F1 (SNF1) mice⁵⁵. Sex hormones might be influencing the microbiome, as this was indicated in a type 1 diabetes study in which male mice were castrated resulting in disease progression^56,57. In female MRL/lpr mice, there is deficiency in the beneficial bacterial Lactobacillus, Mu el at supplemented the mice with Lactobacillus where they observed alleviated lupus nephritis and increased IL-10¹⁹. We observed differences in the gut microbiome that were not only strain specific but sex specific as Scarf1^−/− mice showed lower microbiota only for female mice, and co-housing the mice from 2 to 20 weeks with Scarf1^+/+ did not have an effect. However, we do observe the development of disease and changes in the microbiota in Scarf1^−/+, albeit not as strong when compared to Scarf1^−/− mice. Although additional work is needed, we hypothesize that changes in the microbiome take place after autoimmunity develops. This was shown on an MRL/lpr mouse model, where microbial shift correlated with elevated levels of IgG2, IgG2a and urine albumin to creatine ratio⁵⁸, suggesting that changes in the microbiome could be an immunological marker for disease development and severity.

SCARF1 plays a significant role in maintaining lipid homeostasis. Initially, SCARF1 on endothelial cells as a receptor for modified lipoproteins^54,59. De novo biosynthesis of L-valine, L-isoleucine and adenosine ribonucleotide is essential to produce BCFA and dampen inflammation^60,61. Consistent with maintaining lipid homeostasis, Scarf1^−/− exhibit decreased beneficial bacteria that synthesize BCFA and short-chain fatty acids (SCFA). Cells can alter their metabolism to rely more on mitochondrial oxidation, resulting in apoptosis in the absence of growth factor signaling⁶². Although additional work is needed, we can propose that the development of autoimmunity in Scarf1-deficient mice is driven in part by the dysbiosis and defects in efferocytosis. Efferocytosis is essential for resolving inflammation⁶³. However, when efferocytosis fails, it leads to the release of noxious molecules and inflammatory mediators⁶⁴. SCARF1 is a receptor for efferocytosis²⁵, which we hypothesize is the first insult to the immune system and initiating inflammation. Additional work is needed to understand whether the accumulation of apoptotic debris is necessary to cause the change in the microbiota. However, this link between efferocytosis and dysbiosis have been previously described in other models, such as IBD, Type I diabetes, multiple sclerosis, and systemic lupus erythematosus⁶⁵.

In Scarf1^−/− we noticed changes in the microbiome in addition to the development of autoimmunity. Our observation is the first step to understand the mechanism of SCARF1 in autoimmunity and how it’s affecting individual organs. We noticed that the gut had an increased cellular debris and deficiency in A. muciniphila. Multiple questions arise based on these findings that required a mechanistic analysis of our observations. Akkermansia muciniphila is a gram-negative anaerobic bacterium that colonizes the intestinal tract early in life⁶⁶. This beneficial intestinal commensal is known for colonizing the mucosal layer, where it plays an essential role in host metabolism and immune response⁴⁸. Control mice express high levels of A. muciniphila, however this bacterium is completely absent in Scarf1^−/− mice. A. muciniphila is associated with health and maintaining the mucosal barrier^67,68. Furthermore, disease mice express high levels of Alistipes, Bacteroides, Lachnospiraceae which are associated with dysbiosis and metabolic diseases⁶⁹.

Altogether, our data shows various lupus-associated changes in the gut microbiome. Intestinal colonization of A. muciniphila was found to negatively correlate with disease development. A. muciniphila probiotics and derived postbiotics have already been identified as promising therapeutics within multiple inflammatory diseases, including SLE⁷⁰. Our data corroborate the potential use of A. muciniphila as a probiotic to decrease lupus-like symptoms and lower inflammation. However, further research is required to understand the impact of A. municiphila postbiotics, or metabolites, directly. Assessing metabolic changes in control and disease murine models would be an interesting avenue given previous studies implicating SCARF1 in lipid metabolism and homeostasis²⁵. The work presented here represents the first steps towards understanding the role of SCARF1 in gut inflammation and homeostasis. Our initial findings raise multiple questions, including how metabolic pathways are altered and whether changes in gut permeability contribute to inflammation. Future work will focus on elucidating the mechanistic interactions between apoptotic debris and gut dysbiosis.

Materials and methods

Mice

All mice were maintained under micro isolation in specific pathogen–free conditions at the animal facility of UMass Chan Medical School under a protocol approved by the Institutional Animal Care and Use Committee. In addition, experimental design shows a comparative study between Scarf1^+/+ (wild-type) or Scarf1^−/− following ARRIVE guidelines (https://arriveguidelines.org/). This includes the use of control animals, inclusion criteria for male and female mice, blinding of samples for data analysis and the use of statistical methods are described below. In accordance with ALAAS learning library and IACUC, for end-point studies mice were euthanize using a two-step euthanasia protocol. Mice were anesthetized with isoflurane using saturated vapor, then we performed cardiac puncture for blood collection. To ensure death, cervical dislocation was performed before harvesting organs for analysis.

Wild-type (WT, Scarf1^+/+) C57BL/6 mice were obtained from Jackson Laboratories and bred in-house. Scarf1^−/− mice were transferred from Massachusetts General Hospital and bred in-house for at least 10 generations. All mice were used after 20-weeks of age to allow for disease development, as previously described²⁵. Offspring of Scarf1^−/− and WT mice were produced at normal Mendelian ratios. Mice were not randomized or placed in specific groups for these studies.

Wild-type and Scarf1^−/− mice were kept in individual colonies in the same room and rack. To generate the co-house mice, 20-week old Scarf1^+/+ and Scarf1^−/− mice were placed in the same cage for 14-days. At the end of the 14-days, feces from each mouse were collected for further analysis. To test for the presence of autoantibodies, serum was collected by cheek-bleed.

To generate littermates, breading pairs were made using WT x Scarf1^−/− mice to generate F1 generation. F2 breeding, Scarf1^−/+ x Scarf1^−/+, was generated to obtain Scarf1^+/+. Scarf1^−/+ and Scarf1^−/−. Mice were separated by sex at the time of weaning and same sex mice were co-housed for a minimum of 20-weeks. At 20-weeks of age, feces were collected and mice were euthanized to collect gut, spleen, kidney for further experimentation. Tissue was collected for genotyping at this time. Tissue and feces were sent to Transgenyx (Cordova, TN) for genotype and microbiome analysis, respectively.

Fecal collection for Microbiome Preparation

Mice were selected for analysis from a cohort of Scarf1^−/− mice and their healthy littermate controls. Following euthanasia, fecal samples were collected and immediately frozen at −80 C until processing. Samples were then placed in individual tubes containing DNA stabilization buffers (Transnetyx Microbiome Kits, Cordova, TN) to preserve sample integrity and stability during shipment. All DNA extraction, library preparation, and sequencing were performed by Transnetyx (Cordova, TN).

DNA extraction and metagenomic sequencing

DNA extraction and metagenomic sequencing was performed by Transnetyx Inc. Briefly, stool DNA was extracted using a robust method that ensures reproducible extraction that captures the accurate microbial diversity. DNA quality control was performed to confirm sample integrity. Genomic DNA was converted into sequencing libraries and sequenced using shotgun metagenomic sequencing, generating approximately 2 million 2 × 150 bp read pairs to obtain microbial species and strain-level taxonomic resolution.

Raw sequencing data were uploaded to the OneCodex platform for analysis and aligned against a database of ~ 148 K complete microbial genomes, including 71 K bacterial, 72 K viral, and thousands of archaeal and eukaryotic genomes. To reduce false positives, classification results underwent further analysis to group the samples in their required experimental approach.

Microbial taxonomy and diversity metrics were computed within OneCodex to compare Scarf1^−/− mice and healthy littermate controls. Samples were also compared with Transnetyx’s global diversity averages derived from historic datasets for quality control. Low read counts were normalized relative to the total number of identifiable reads within each host sample. Finally, sequencing data were aligned against the Gene Ontology (GO) and KEGG Orthology databases via the OneCodex platform for downstream functional analysis of microbial communities.

Autoantibody profiles

Antinuclear autoantibodies (ANA) were measured as previously described⁷¹. ANA assays from mouse serum were performed using immunofluorescence assays according to the manufacturer’s instructions (Bio-Rad). Mouse serum was diluted at 1:200 and incubated with Hep-2 cells, followed by Alexa Fluor 488 secondary to detect bound ANAs (Cat #A11001, Invitrogen). Staining was scored by three independent observers ‘blinded’ to the genotypes of the mice. ANA severity was scored on a scale from 0 to 3, where grade of inflammation 1- minimal, 2- moderate, 3-high based on fluorescence levels.

Histology

The small and large intestines were dissected from 20-week-old Scarf1^−/− and C57BL/6J (B6) wild-type mice. To assess histological signs of gut inflammation, the intestines were fixed in 10% phosphate-formalin and embedded in paraffin. Sections were prepared and stained in hematoxylin-eosin. For all other studies, the intestines were flash frozen on Optimal Cutting Temperature (OCT) Compound (Sciegen Scientific, Gardena CA) embedding media. Sections from frozen samples were prepared (7 μm) and stained as described below. All slides were imaged using ECHO fluorescent microscope equipped with a high-resolution (Discover Echo, San Diego, CA, USA) and analyzed using Discover ECHO App (iOS 16+) (Discover Echo, San Diego, CA, USA) and Adobe Photoshop (Adobe, San Jose CA).

TUNEL assay

Frozen sections (7 μm) were allowed to warm to 25 °C. A TUNEL assay (terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling) for the detection of apoptotic cells was done according to manufacturer’s instructions (Invitrogen). Paraformaldehyde-fixed tissue was made permeable with 60 min 0.25% Triton X-100 in PBS at room temperature and was washed twice with PBS. Slides were incubated with reaction buffer (25 mM Tris-HCl, pH 6.6, 200 mM sodium cacodylate, 0.25 mg/ml BSA and 1 mM cobalt) containing terminal deoxynucleotidyl transferase, then were washed twice and were incubated for 45 min at 25 °C in the dark (protected from light) with reaction buffer containing the fluorescent label Alexa Fluor 488. To identify tissue structures, sections we stained with Phalloidin-Rhodamin for 25 min at RT and EPCAM1 cells were stained for 30 min at 25 °C with Alexa Fluor 647–anti-EPCAM1 (1:50 dilution; Invitrogen A22283) in PBS. Samples were washed twice with PBS. Finally, DNA was stained for 10 min at 25 °C with Hoechst 33,342 (1:1,000 dilution; Molecular Probes, Invitrogen). Slides were mounted with Prolong Gold antifade reagent (Invitrogen P36935) and were visualized with ECHO fluorescent microscope equipped with a high-resolution (Discover Echo, San Diego, CA, USA). Data were analyzed with Discover ECHO App (iOS 16+) (Discover Echo, San Diego, CA, USA) and Adobe Photoshop (Adobe, San Jose CA). The total number of apoptotic cells was determined by counting 100 Hoechst (DAPI) cells as a baseline per individual slide. Three independent sections were quantified per slide. FITC+ cells were quantified among the Hoechst cells. The average of the FITC+ cells was determined and graphed as a percentage of apoptotic cells.

Histological assessment and disease scoring

Histological evaluation of the intestinal H&E sections was performed in a blinded manner by a pathologist (Dr. Nupur Jadhav). Inflammation and cellular infiltration were scored, and cell types were identified. Tissue inflammation severity was scored on a scale from 0 to 5, where grade of inflammation 1- minimal, 2- mild, 3-moderate, 4-severe, 5-severe with ulceration. For presence of apoptosis in the tissue, 1- present and 2- absent. A composite disease score was calculated by averaging the individual scores for antinuclear autoantibodies (ANA), alopecia, nephritis, and the number of apoptotic cells.

Statistical analysis

Statistical calculations were done with a statistical software package GraphPad Prism, version 10.4.2 (GraphPad Software, San Diego, CA). For comparisons between two or more groups, the mean ± s.e.m was analyzed by unpaired two-tailed Student’s t test or ANOVA, respectively. Statistical analysis of the microbiota profiling data was performed on the proportional representation of the taxa using Shapiro-Wilk normality test. Parametric test with Welch’s corrections or nonparametric test with Mann-Whitney corrections were used depending on if passed the normality test. Multivariate analyses of disease score versus relative abundance of bacteria were done. The investigators were not blinded to the genotype of the mice except where indicated. Values of P < 0.05 were considered statistically significant.

In addition, OneCodex python package and Jupityr notebook was used for Bray-Curtis Dissimilarity statistical analysis. PERMANOVA and post-test were used to determine the values between groups. Data is shown as a principal component analysis. Network analyses are constructed by OneCodex package on Jupytr.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

DMS and ZGRO designed and performed the experimental, analyzed data and wrote the manuscript. ZGRO developed the mouse model. SH, MC, JMR, HN provided advice in the experimental design and edited the manuscript. JMR, DVW and NF validated the results. NJ analyzed pathology samples. JMR, DVW, HN provided advice on data analysis.

Funding

This work was funded by the Department of Defense LRP-Impact Award (W81XWH-21-1-0803) (ZGRO), Lupus Research Alliance Innovation Award (ZGRO) and UMass Chan Medical School Start-up funds (ZGRO).

Data availability

The datasets for the current study are available through OneCodex upon request to the corresponding author. All analyzed data were presented in the manuscript and in the supplement. The datasets generated and/or analyzed during the current study are available in the Microbiome_manuscript repository, https://app.onecodex.com/projects/ee4b590b724e4def.

Declarations

Competing interests

The authors declare no competing interests.