C57BL/6, prdm1eGFP, Tcrβδ ^-/-, Sle123 (all C57BL/6 background) mice were bred under specific pathogen-free conditions at the DRFZ and Charité Universitätsmedizin, Berlin, Germany. C57BL/6 mice were also bred under GF or SPF conditions at the German Institute of Human Nutrition, Potsdam, Germany. Sle123 mice were evaluated for proteinuria twice weekly (Multistick, Siemens) starting from week 24 onwards. Sle123 mice were considered to be sick and sacrificed for analysis when proteinuria >100mg/dL. Age-matched WT, healthy age-matched or young Sle123 mice (9-12 weeks old) without proteinuria were included as healthy control mice. Mice were between 10-20 weeks old, unless otherwise stated. Gender-matched mice were used.

Human peripheral blood was collected from 12 HDs and six SLE patients. Human bone marrow (BM) was obtained from nine patients undergoing total hip arthroplasty.

Murine specific monoclonal antibodies: CD3 (clone 17A2), CD11a (clone M17/4), CD18 (clone M18/2), CD16/CD32 (Fc Block, clone 2.4G2), CD19 (clone 6D5), CD39 (clone Duha59), CD44 (clone IM7), CD47 (clone miap301), CD49d (clone R1-2), CD54 (clone YN1/1.7.4), CD81 (clone Eat-2), CD98 (clone RL388), CD130 (clone 4H1B35), CD138 (clone 281-2), CD155 (clone TX56), CD172a (clone P84), CD205 (clone NLDC- 145), CD229 (clone Ly9ab3), CD267 (clone 8F10), CD274 (clone 10F.9G2), CD319 (clone 4G2), CD326 (clone G8.8), IgA (clone mA-6E1), IgM (clone RMM-1), LAG-3 (Clone eBioC9B7W), SCA-1 (clone D7) and human specific monoclonal antibodies: CD3 (clone UCHT1), CD14 (clone M5E2), CD19 (clone SJ25C1), CD20 (clone 2H7), CD27 (clone L128), CD39 (clone A1), CD81 (clone 5A6), CD130 (clone 2E1B02), CD138 (clone MI15), and CD326 (clone 9C4) were purchased from BioLegend, Miltenyi Biotec, BD Biosciences, Thermo Fisher or produced in house. LEGENDScreen™ Mouse PE Kit was purchased from BioLegend.

C57BL/6 mice were immunized intraperitoneally with 200µg of NP-KLH precipitated in alum and analyzed 7 days after immunization. NP-reactive ASCs were detected using fluorochrome-conjugated NP.

Cell isolation and flow cytometry were performed as previously reported (1, 23). Briefly, BM cells were flushed from femur and tibia using a 1-ml syringe with a 26G needle in PBS supplemented with 0.5% (w/v) bovine serum albumin (BSA) (PBS/BSA). Spleen and mesenteric lymph node (mLN) were mashed in a Petri dish with PBS/BSA, passed through a 70-µm cell strainer (BD Biosciences), and then transferred into a 15-ml tube using a 25G needle to obtain single cell suspension. For isolation of kidney cells, the kidneys were minced using a scalpel and manually crushed on a 70-µm cell strainer using a syringe plunger, and then transferred into a 15-ml tube using a 25G needle. Collagenase D (Sigma) was used to isolate splenocytes in some experiments, in these cases each spleen was cut into two equal pieces, each piece of spleen was then minced separately using a scalpel and transferred into 4ml of serum-free RPMI (Life Technologies) supplemented or not with 1mg/ml collagenase D and incubated at 37°C for 30 minutes. Non-digested tissue pieces were also crushed on 70-µm cell strainer using a syringe plunger, and subsequently transferred into a 15-ml tube using a 25G needle to obtain single cell suspension. Bone marrow, spleen, and kidney red blood cells were lysed using red blood cell lysis buffer (Sigma). After lysis, the cells were resuspended in PBS/BSA and counted using MACSQuant ^® Flow Cytometer (Miltenyi Biotec). Prior to staining, the cells were blocked with anti-Fc receptor antibody (clone 2.4G2) for 15 minutes and then incubated with fluorochrome- and/or biotin-conjugated antibodies against surface molecules for 20 minutes. Biotinylated antibodies were subsequently stained with fluorochrome-conjugated Streptavidin for 15 minutes. For the screening experiments, surface-stained cells were then incubated another 15 minutes with PE-conjugated screening antibody (LEGENDScreen™ Mouse PE Kit, BioLegends). For the kinetic experiment, BM cells were stained at 30 minutes, 2, 4 and 6 hours after red blood cell lysis. Dead cells were excluded using propidium iodide. Stained cells were measured by BD FACSymphony or BD LSRFortessa.

BM mononuclear cells were isolated as previously reported (24). Briefly, fresh BM samples were physically minced in PBS/BSA/EDTA (Miltenyi Biotec) and subsequently filtered with 70-µm cell strainer. BM mononuclear cells were obtained by density gradient centrifugation using Ficoll-Paque PLUS. BM mononuclear cells were collected and washed twice with PBS/BSA/EDTA.

Human peripheral blood mononuclear cells were isolated from EDTA-anticoagulated whole blood by Ficoll density gradient centrifugation. BM (30x10⁶) and peripheral blood (PB) mononuclear (20x10⁶) cells were initially labelled with live/dead dye (Blue fluorescent reactive dye, Molecular Probes Invitrogen) in PBS according to the manufacturer’s instructions and washed with PBS. Labelled cells were blocked with FcR blocking reagent (Miltenyi Biotec) for 5 min. Cells were then stained for surface molecules with fluorochrome-conjugated antibodies for 20 minutes. Stained cells were measured using a BD LSRFortessa.

Mice blood was collected postmortem by cardiac puncture, and stored at 4°C for 8–12 hours in 1.5ml microtubes. The tubes were centrifuged at 4°C for 10 min at 13000rpm and the sera were transferred into new microtubes and stored at –20°C until analysis.

Anti-dsDNA IgG and IgM antibodies were measured by ELISA as previously described (25). Briefly, 96-well polystyrene plates were pre-coated with 10µg/ml methylated BSA (Sigma-Aldrich) in PBS for 3 hours at 37°C. The plates were washed with PBS and coated with calf thymus dsDNA (10µg/ml in PBS; Sigma Aldrich) for 3 hours at 37°C. The plates were washed with PBS and blocked with PBS/BSA (3%, w/v) (Sigma-Aldrich) overnight at 4°C, and washed with deionized water. One hundred microliters of 5-fold serial dilutions of each serum were prepared in PBS/BSA (1%, w/v) to obtain 1:20 to 1:62500 dilutions, and were deposited on the plate. After 2 hour incubation at 37°C, the plates were washed with PBS 0.05% (v/v) Tween-20 (Sigma). Next, 100µl of either biotin-conjugated anti-mouse IgG or IgM (0.5µg/ml; Southern Biotech) were added to the wells and incubated for 1 hour at 37°C. The plates were washed with deionized water and incubated with 100µl of ExtrAvidin^®-Peroxidase (Sigma Aldrich) at a 1:5000 dilution in PBS/BSA (1%, w/v) for 30 min at 37°C. After washing with deionized water, 50µl of Tetramethylbenzidine (TMB) substrate solution (Thermo Scientific) were added to the wells. The reaction was stopped with 50µl of 2M sulfuric acid and read at 450nm on a spectrophotometer (SpectraMax Plus 384, Molecular Devices, USA).

Relative serum anti-dsDNA antibody titer was calculated by plotting the dilution factors of each serum sample in logarithm scale against the absorbance. All curves were drawn on the same graph, and an absorbance cut off value was decided, which crossed the curves in the linear regions of the majority of the curve. The dilutions at cut off values were read on the x axis and gave the relative dsDNA antibody titers of the samples.

Flow cytometry data were analyzed with FlowJo™ software (version 10.7.1, TreeStar). Quantification and statistical analysis were performed using GraphPad Prism 9. Groups were compared using two-way ANOVA, unpaired t test or multiple unpaired t test, as indicated in figure legends. Considered p-values were: ^nsp > 0.05; ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; ^∗∗∗∗p < 0.0001. Correlation matrices were generated using R corrplot package.

To identify new makers expressed by ASCs, we used splenocytes and BM cells isolated from naive prdm1eGFP reporter mice to screen 255 surface molecules (LEGENDScreen™, BioLegend, Inc.) by flow cytometry. ASCs identified as BLIMP-1eGFP⁺CD138⁺ were compared with B and non-B cells, identified as CD19⁺BLIMP-1eGFP^-CD138^- and CD19^-BLIMP-1eGFP^-CD138^- cells, respectively ( Supplementary Figure 1A ). This strategy yielded 20 surface molecules, of which the high expression by ASCs from BM, spleen, and mLN was confirmed by further flow cytometric analysis ( Figure 1A ; Supplementary Figures 1B, C ). Among these surface molecules, CD138, CD98, CD267, CD319, SCA-1, CD44, CD18 (Integrin β2), and CD11a (Integrin α-L) were found consistent with earlier reports for ASCs (6, 26). Most importantly, certain surface molecules, such as CD39, CD229 (Ly-9), CD81, CD326, CD54 (ICAM-1), CD130, CD49d (Integrin α4), CD172a (SIRPα), CD274 (PD-L1), CD155, CD205 (DEC-205), and CD47 were identified and have not been clearly reported as ASCs markers before. Importantly, flow cytometry clearly confirmed the co-expression with BLIMP-1eGFP^hi cells in BM ( Figure 1B ), spleen ( Supplementary Figure 1D ) and mLN ( Supplementary Figure 1E ). To evaluate the co-expression of BLIMP-1eGFP and the candidate markers with the BLIMP-1eGFP⁺CD138⁺ population, we performed a Spearman correlation analysis. The majority of the populations defined by the new markers showed positive correlation (r > 0.94) with BLIMP-1eGFP⁺CD138⁺, confirming the co-expression of these molecules by the same ASCs ( Figure 1C ). Thus, we concluded that differential expression of CD39, CD229, CD81, CD326, CD54, CD130, CD49d, CD172a, CD274, CD155, CD205, or CD47 together with BLIMP-1eGFP can be used to identify ASCs independently of CD138 in lymphoid organs of naive mice.

Next, we addressed whether the above identified markers can be used to identify ASCs independently of BLIMP-1. Thus, we first co-stained each marker with CD138, a bona fide ASC identifier. ASC populations co-expressing CD138 and each of the markers could be clearly identified in BM, spleen, and mLN. CD39, CD81, CD326, and CD130 provided an improved resolution compared other markers ( Figure 2A ; Supplementary Figure 2 ). Then, we asked whether double positive populations co-expressing CD138 and each marker would include only ASCs in lymphoid organs. To address this question, we quantified the frequencies of BLIMP-1-expressing cells within these populations. Most of these cells expressed BLIMP-1eGFP⁺ in all tested organs, except for a lower frequency in CD138⁺CD18⁺ cells, which contained higher proportions of non-BLIMP-1eGFP⁺ ASCs in both BM and mLN, as did CD138⁺CD319⁺ and CD138⁺CD49d⁺ cells in BM ( Figure 2B ). Therefore, these markers identified almost exclusively ASCs. Finally, we used a Spearman correlation matrix to evaluate which surface molecules would perform better than the others as ASCs identifiers in BM, spleen, and mLN, respectively. Most notably, only CD138⁺CD39⁺, CD138⁺CD81⁺, CD138⁺CD130⁺ populations exhibited a significant correlation with CD138⁺BLIMP-1eGFP⁺ population in BM when a p-value cutoff of 0.01 was applied ( Figure 2C ). CD326 was also significantly co-expressed when a p-value of 0.05 was considered. Of note, combinations of CD138 with CD98, CD267, CD319, CD39, CD229, CD81, CD326, CD54, CD130, CD49d, CD155, CD205, or CD47 allowed the identification of ASCs in spleen, whereas CD98, CD267, CD39, CD326, CD54, CD130, CD49d, and CD205 identified ASC in mLN ( Figure 2C ). Thus, CD39, CD81, and CD130 followed by CD326 are the best molecular ASC identifiers.

ASCs infiltrating in the kidney medulla of patients with SLE (systemic lupus erythematosus) correlate with kidney inflammation and disease severity (27). In animal models of lupus such as NZB/W F1 and MRL/lpr mice, ASCs infiltrate the kidneys and influence disease severity (27–29). Thus, we tested whether the herein identified markers can serve to quantify kidney-resident ASCs in mice with lupus. To do so, we analyzed mice of the B6.Sle123 triple congenic strain, in which Sle1 locus mediates the loss of immune tolerance; Sle2 locus results in hyperactivity of B cells, and Sle3 locus results in reduced CD4⁺ T cell apoptosis. B6.Sle123 mice develop highly penetrant severe systemic autoimmunity and fatal glomerulonephritis, beginning at 6 months of age (30–32). Here, we analyzed 8 months old mice with established proteinuria (> 300 mg/dL, sick) and age-matched mice of the same strain in which proteinuria was absent (healthy). We found that co-expression of CD138 and CD39, CD81, CD326, and CD130 identified kidney-resident ASCs in sick Sle123 mice while this ASC population was absent in proteinuria negative age-matched Sle123 controls ( Figure 2D ). We conclude that CD39, CD81, CD326, and CD130 mark ASCs in kidneys of murine lupus nephritis.

In mice, BM ASCs can be identified as BLIMP-1eGFP⁺CD138⁺ or CD19/B220^low/-CD138^hi ( Supplementary Figure 3A ). Recently, it was shown by flow cytometry that ASCs can be identified, in the absence of BLIMP-1, as CD138⁺CD267⁺ (33) ( Supplementary Figure 3A ). While evaluating the stability of the newly identified markers, we observed that CD138 and CD267 expression were down-regulated shortly after ASCs are released from the hypoxic BM environment ( Figure 3A ; Supplementary Figure 3B ). In contrast, newly identified ASC markers CD39, CD81, CD326, and CD130 showed a stable expression at all tested time points ( Figure 3A and Supplementary Figure 3B ). Next, we evaluated optimal marker combinations to mark BM ASCs. For BM ASCs, CD39⁺CD130⁺, CD39⁺CD81⁺, CD39⁺CD326⁺, or CD81⁺CD326⁺ exhibited stable staining with more than 95% of the BLIMP-1eGFP⁺ cells ( Figure 3B and Supplementary Figure 3C ).

Isolation of ASCs from non-lymphoid organs is a bottle neck for studying tissue-resident cells. That is, the immune cells need to be released from the tissues by mechanical or enzymatic methods or combinations thereof. In particular, enzymatic digestion affects the detection of plasma cell markers, such as CD138 and CD267 (34). Therefore, we tested whether the newly identified ASC markers would resist to enzymatic digestion. We compared the expression of the markers—CD138, CD267, CD39, CD81, CD326 and CD130—before and after digestion of prdm1eGFP spleens with collagenase. This lymphoid organ does not need digestion for ASC retrieval, and therefore, allows for the quantitative comparison of markers stability with or without enzymatic digestion. While CD138 and CD267 disappeared from the cell surface upon collagenase treatment, CD39, CD81, CD326 and CD130 remained clearly detectable ( Figure 3C and Supplementary Figures 3D, E ). Thus, CD39 combined with CD81 or CD130 identified tissue resident ASCs with more than 95% of the cells being BLIMP-1eGFP⁺ cells ( Figure 3D ). Of note, collagenase treatments increased ASC retrieval, which was visible with both BLIMP-1eGFP reporter and CD39, CD81, and CD130.

Our initial findings that CD39, CD81, CD326, and CD130 identify ASCs in mouse led us to examine whether these molecules also identify human ASCs. We isolated mononuclear cells from peripheral blood and BM from heathy donors (HDs) and analyzed the expression of selected molecules by ASCs and other B lineage cells. Particularly, CD27⁺⁺CD20^low plasmablasts (PBs), memory CD27⁺CD20⁺, naïve CD27^-CD20⁺ B cells in peripheral blood, and CD38⁺⁺CD27⁺⁺ PCs, CD27⁺CD38^- memory, CD27^-CD38⁺ and CD27^-CD38^- immature B cells in the BM were analyzed ( Supplementary Figures 4A, B ) (35). CD27⁺⁺CD20^low PBs from peripheral blood expressed higher level of CD39, CD81, and CD130 compared with other B cell subsets; and a higher level of CD326 than naïve B cells did ( Figure 4A ). CD38⁺⁺CD27⁺⁺ PCs from BM expressed higher level of CD39, CD130, and CD326 compared with other B cell subsets but lower level of CD81 than CD27^-CD38⁺ immature B cells did ( Figure 4B ).

We then asked whether the expression of these molecules was sufficient to identify human ASCs. We found similar frequencies of peripheral blood PBs ( Figure 4C ; Supplementary Figure 4C ) and BM PCs ( Figure 4D ; Supplementary Figure 4D ) using the newly identify markers or with conventional CD20^lowCD27⁺⁺ and CD38⁺⁺CD27⁺⁺ gating, respectively. Analysis with Spearman correlation matrices showed that the ASC populations identified with new markers correlated with CD27⁺⁺CD20^low PBs from peripheral blood (r > 0.97; Figure 4C ) or with CD38⁺⁺CD27⁺⁺ PCs from BM (r = 1, Figure 4D ). Most importantly, we verified the expression of these surface molecules in patients with SLE, which have been known to have a higher number of PBs in peripheral blood. CD39, CD81, and CD130 were found on ASCs from peripheral blood of these patients ( Figure 4C ). Next, we evaluated which combination of these markers would be useful to stain PCs in human BM ( Figure 4E ; Supplementary Figure 4E ). We were able to gate human BM PCs as CD81⁺CD130⁺ cells ( Figure 4E ). We conclude that CD39, CD81, and CD130 expression marks human ASCs from peripheral blood and BM. These surface markers expressed by mouse and human ASCs overcome prior limitations of detection and might candidate as treatment targets of ASCs.

The identification of these new and overarching ASC markers may allow new insights into ASC induction and lifestyle. We conducted additional studies to address which signals could promote the expression of the identified markers in ASCs. The level of expression of the several surface markers is organ specific ( Supplementary Figure 5A ). For instance, the transcription factor BLIMP-1 and surface markers, such as CD39, CD81, CD326, or SCA-1 are higher expressed in BM than in spleen and mLN. In contrast, CD138 is higher expressed in spleen and mLN. In addition, splenic ASCs express higher levels of CD98 than those residing in BM and mLN while mLN ASCs express higher levels of CD44 ( Supplementary Figure 5A ). Some ASCs subsets are enriched in certain organs, such as long-lived plasma cells in the BM and microbiota driven ASCs found in the gut-associated tissue, such as the mLN. However, which signals drive the expression of certain molecules on these ASCs is unknown. To delineate a critical pathway of ASC induction and to examine whether CD39, CD81, CD130, and CD326 are expressed by ASCs generated in microbiota-free conditions, we compared the expression of these markers in mice raised in specific pathogen-free (SPF) or in germ-free (GF) conditions. ASCs from GF mice and controls were identified as CD138⁺CD39⁺ cells ( Supplementary Figure 5B ). The frequency of ASCs in BM and spleen were similar in GF and SPF mice ( Supplementary Figure 5B ). As previously described (36), we found a decreased frequency of ASCs in mLN of GF compared to SPF mice, highlighting the importance of microbial signals on ASC generation in mLN ( Supplementary Figure 5B ). We then compared the expression levels of ASCs markers in the two cohorts of mice. Most markers, including CD39, CD130, and CD326, showed similar expression in both conditions ( Figures 5A, B ). In contrast, CD81 was found to be up-regulated on ASCs from GF mice in all analyzed organs ( Figure 5B ; Supplementary Figure 5C ). We conclude that CD81 expression is downregulated by microbiota-derived signals.

As ASCs can be generated through T-dependent or T-independent pathways, next we examined whether some of the new markers could identify ASCs generated upon canonical T:B cell interaction. To this aim, we analyzed ASCs from T cell-deficient mice (Tcrβδ ^-/-). These mice accumulated less ASCs in BM and more in mLNs than WT mice, while similar frequencies of these cells were found in the spleen ( Supplementary Figure 5D ). The expression of CD39, CD81, CD326, and CD130 by ASCs from T cell-deficient mice was similar or higher than those from WT mice in BM, spleen, and mLN, suggesting that these markers identify ASCs independently of their T-dependent or independent origin ( Figure 5C ). We found similar expression of the newly identified markers in splenic ASCs from WT and T cell-deficient mice, while there were some differences in BM and mLN. CD11a was decreased and CD54 was increased on ASCs from Tcrβδ ^-/- mice in both BM and mLN compared to those from WT mice ( Figure 5D ; Supplementary Figure 5E ). Furthermore, we found an increased expression of CD98, SCA-1, CD39, CD326, CD54, CD172a, and CD274 in mLN ASCs from T cell deficient mice compared to WT ( Figure 5D ; Supplementary Figure 5E ). To further investigate the role of T cell help in the regulation of expression of these molecules on ASCs, we immunized WT mice with 4-hydroxy-3-nitrophenylacetyl keyhole limpet hemocyanin (NP-KLH), a well-established T cell-dependent antigen, and analyzed the expression of ASC markers by non NP- and NP-specific ASCs ( Supplementary Figure 5F ). In agreement with the findings in T cell-deficient mice, non-NP-specific ASCs expressed higher amounts of CD98, SCA-1, CD54, CD172a, and CD274 than NP-specific ASCs ( Figure 5E ; Supplementary Figure 5G ). Altogether, these data suggest that ASCs generated in the context of T-dependent antigens express lower level of CD98, SCA-1, CD54, CD172a, and CD274.

We and others showed that ASCs are accumulated in BM and spleen with age (1, 37). We analyzed the expression of the new markers by ASCs from aged mice, hereafter named, old ASCs in order to identify their origins (37). Flow cytometry analyses confirmed the accumulation of ASCs in BM, spleen, and mLN with aged when CD138 and CD39 were applied to identify ASCs. The increased frequency of ASCs was associated with increase of LAG-3⁺ regulatory ASCs ( Supplementary Figure 5H ) (1). These ASCs were generated independently of T cells as they were also present in T-cell deficient mice ( Supplementary Figure 5I ). Consistent with the data from Tcrβδ ^-/- and the analysis of non-NP-specific ASCs ( Figures 5D, E ), we found that old ASCs expressed higher levels of a set of four markers, including CD98, CD54, CD172a, and CD274 than ASCs obtained from young animals ( Figures 5F, G ). Collectively, the biased expression of the T independent markers CD98, CD54, CD172a, and CD274 by old ASCs support that ASCs accumulated in old mice share phenotypic characteristics with T-independent ASCs.

In autoimmunity, pathogenic ASCs produce autoantibodies and pro-inflammatory cytokines (3, 38–41). Thus, is of interest to find molecules that are specifically expressed by pathogenic ASCs, as this would allow their selective targeting. We analyzed the expression of the newly identified ASC markers in the context of lupus using the B6.Sle123 triple congenic mouse strain with established proteinuria (>100mg/dL, sick). We used B6.Sle123 mice in which proteinuria was not detectable as controls (healthy). We found that CD39, CD81, and CD130 were highly expressed by ASCs from sick Sle123 mice ( Supplementary Figure 6A ) and sick Sle123 mice had an increased frequency of CD138⁺CD39⁺ ASCs in BM, spleen, and mLN compared with the control group ( Figure 6A ). In the BM, this increase was partly associated with enhanced LAG-3⁺ regulatory ASCs ( Figure 6B ).

Next, we evaluated which of the ASC markers could identify disease-associated ASCs in sick Sle123 mice. CD18, CD326, and CD172a were over-expressed in sick Sle123 ASCs compared to control ASCs in all analyzed organs ( Figure 6C ; Supplementary Figure 6B ). In order to fine tune our analysis to pathogenic ASCs, we analyzed the expression of molecules of interest on regulatory LAG-3⁺ ASCs and their counterparts. We found that CD39 and CD326 were over-expressed in BM LAG-3^- ASCs compared with LAG-3⁺ ASCs in sick Sle123 mice, LAG-3^- and LAG-3⁺ ASCs in control SLE123 mice and age-matched WT ( Figure 6D ). We did not find similar differences in spleen or mLN ( Supplementary Figure 6C ). Given that CD39 and CD326 were over-expressed only by LAG-3^- ASCs in sick Sle123, we examined whether their co-expression could identify a new subpopulation of PCs associated with lupus. We found that co-expression of CD39 and CD326 marked an ASC subpopulation (CD39⁺⁺CD326⁺⁺) that accounted for half of total ASCs and the majority of LAG-3^- ASCs in sick Sle123 mice while this ASC population represented less than 20% of total or LAG-3^- ASCs in controls. The increased total number of CD39⁺⁺CD326⁺⁺ ASCs in sick Sle123 mice was mainly associated with increased LAG-3^-CD39⁺⁺CD326⁺⁺ ASCs ( Figure 6E ). Moreover, the increased frequency of CD39⁺⁺CD326⁺⁺ ASCs in sick Sle123 was associated with an elevated titers of anti-dsDNA autoantibodies in SLE ( Figure 6F ). We found that CD39⁺⁺CD326⁺⁺LAG-3^- were IgM⁺ cells ( Figure 6G ) and the frequency of CD39⁺⁺CD326⁺⁺LAG-3^- ASCs correlated with IgM anti-ds-DNA autoantibodies ( Figure 6H ). These cells were not found in spleen, mLN, and kidneys, suggesting that they preferentially reside in the BM and did not correlate with IgG anti-dsDNA autoantibodies in sera of sick Sle123 mice ( Supplementary Figures 6D, E ).