Dnase1l3^-/- mice (22) were generated as described. Siglecg ^-/- mice were generated as described (35) and were backcrossed to the C57BL/6 background by marker-assisted speed congenics (31). Dnase1 mutant mice were generated by CRISPR/Cas9 mediated mutagenesis in JM8A3 embryonic stem (ES) cells from C57BL/6N origin (36). In brief, ES cells were transfected with pX458 (obtained from Addgene, Watertown, MA, (37)) in which the Dnase1 specific gRNA-sequence 5´ TGACATCGCTGTTATCCAAG 3´ was inserted. GFP-expressing ES cells were sorted and mutations in individual ES cell clones were analyzed by sequencing the amplicon generated by primers flanking the target sequence in exon 3. One clone showing a 65 bp deletion from intron 2 into exon 3 was selected for blastocyst injections, generation of chimeric mice and further breeding with C57BL/6N mice. The deletion does not allow splicing into exon 3, and potential alternative splicing into exons 4, 5 or 6 containing the active sites for Dnase1 enzymatic activity lead to frameshift mutations and premature stop codons. For genotyping of mice these primers were used: DNAase1forw: 5´GGGAGGGACAAAGTCTGAGGTC3´ and DNAase1rev 5´AACAATAGAGCACAGAGGGCGT 3´ creating a 230 bp wildtype product and a 165 bp band for the mutant allele.

Anti-nuclear antibodies in the sera of aging mice were detected using Hep-2 cell slides (Bio-Rad). For this purpose, serum was diluted 1:250 in 1x PBS (Gibco) containing 0.1% BSA (Roth) and 0.05% sodium azide (Sigma-Aldrich). Buffer without serum was used as a negative control, and sera from sick MRL/lpr mice were carried as positive controls at the same dilution. 50 µl of the dilutions were pipetted as drops onto the wells of the slides and incubated for 30 minutes at room temperature in a humid chamber. The drops were then discarded and the slides were washed with 1x PBS for 10 min. Detection of anti-nuclear antibodies was performed with an Fc-specific goat-anti-mouse IgG antibody labeled in AF488 (Jackson ImmunoResearch). 50 µl of the diluted goat-anti-mouse antibody (1:500 in 1x PBS containing 0.1% BSA and 0.05% sodium azide) was pipetted onto the wells and incubated for 30 minutes at room temperature in a dark and humid chamber. Again, the slides were washed for 10 minutes in 1x PBS. After that, the slides were covered with cover glasses (Paul Marienfeld) using DePeX mounting medium (Serva). Data were acquired on an Axio Scope.A1 fluorescence microscope (Zeiss) with identical exposure times for all samples. The images were analyzed using ImageJ software by determining the average pixel intensity of ten nuclei per sample.

Anti-double stranded DNA antibodies in the serum of aging mice were detected using an ELISA-based method. For this purpose, the wells of a maxisorp plate (96-Well plate; NuncImmuno) were coated with 20 µg/ml Poly-L-Lysine (Sigma-Aldrich) in TE buffer (10 mM Tris and 1 mM EDTA in H₂O, pH 8). 50 µl per well was used and incubation was performed for two hours at room temperature or overnight at 4°C. The plates were then washed three times in TE buffer before 100 µl of 20 µg/ml calf thymus DNA (Sigma-Aldrich) in TE buffer was added to the wells. This second coating step was carried out overnight at 4°C and afterwards the plates were washed three times with 1xPBS/0.05%Tween20. Wells were then blocked with 150 µl of ELISA buffer (1x PBS containing 0,05% Tween20, 2% FCS, pH 7,4) for one hour at room temperature. The blocking solution was discarded and 50 µl of the serum dilutions (1:100 in ELISA buffer) were pipetted into the wells in duplicates. Pooled serum from NZB/NZW F1 mice was used as standard and ELISA buffer without serum was used as blank. The serum dilutions were incubated for one hour at room temperature. Afterwards, the plates were again washed three times with 1x PBS/0.05%Tween20. Detection of anti-dsDNA-antibodies was performed by incubation with an Fc-specific goat anti-mouse IgG-HRP (Jackson ImmunoResearch) secondary antibody (50 µl, 1:5000 diluted in ELISA buffer) for one hour at room temperature. The plates were washed three times before 50 µl Substrate (o-Phenylenediamine dihydrochloride tablet (Sigma-Aldrich) to a final concentration of 1 mg/ml in 20 mM Na₂HPO₄ and 10 mM C₆H₈O₇ and 3% H₂O₂) was added into the wells. Before saturation was reached, 50 µl of Stop solution was added (4M H₂SO₄) and the optical density was measured at 492-620nm wavelength using the Plate reader and SoftMax Pro software. The amount of anti-dsDNA antibody was determined as a relative unit (arbitrary unit) compared with the pooled serum of NZB/NZW F1 mice.

For the measurements of the blood urea nitrogen (BUN) content in the sera of aging mice the TECO-Diagnostics BUN kit was used. According to manufacturer`s protocol, 150 µl of enzyme reagent was pipetted in the wells of a 96 well MaxiSorp plate (Nunc-Immuno) and 1 µl Serum was added. 1 µl of pooled serum from sick MRL/Lpr mice was used as a positive control and for negative reference no serum was added. The standard included in the kit had a concentration of 20 mg/dl and was also diluted 1:150 in enzyme reagent and was used in triplicates. The samples were incubated for 10 minutes at room temperature. Following this, 100 µl color reagent were added and followed by an incubation step of 10 minutes at room temperature (dark). Optical density was measured in the ELISA reader at 600 nm wavelength using SoftMax Pro software. The following formula was used to determine the urea nitrogen content in the blood: B U N   ( m g d l ) = o D   s a m p l e o D   s t a n d a r d × 20 m g / d l .

Mice at 42-48 weeks of age or 60 weeks of age were sacrificed and the analyzed organs were collected. For the analysis of bone marrow cells, cells were isolated from one hind leg (femur and tibia). Single cell suspensions were prepared in 1x PBS (Gibco) using a 70 µm cell strainer (Greiner) and cell suspensions were kept at 4°C. Then the cells were pelleted by centrifugation (1300 rpm, 4°C, 10 minutes). After discarding the supernatant, the cells were taken up in ACK buffer (1,5 M NH₄Cl, 100 mM NaHCO₃, 10 mM EDTA in H₂O) (2 ml for bone marrow cells, 3 ml for splenic cells) for erythrocyte lysis. Red cell lysis was performed precisely for three minutes at room temperature before adding twice the volume of 1x PBS and centrifuging the cells. Cells were then taken up in 1x PBS and were counted using trypan blue staining (Gibco) and the Neubauer counting chamber. 4x10⁶ cells were used per staining. Therefore, the cells were transferred into 5 ml FACS tubes and after centrifugation the cells were taken up in 50 µl of the corresponding antibody mix in FACS buffer (1x PBS containing 0,1% BSA, 2mM EDTA, 2mM sodium azide). Staining was carried out for 30 minutes at 4°C protected from light. Cells were then washed with 1 ml FACS buffer and centrifuged. If the staining mix contained biotinylated antibodies, the cells were taken up in 50 µl FACS buffer containing fluorescent labeled streptavidin and incubated again for 30 minutes at 4°C. After a washing step the cells were then taken up in FACS buffer and were measured on flow cytometer CytoflexS (Beckman Coulter). The raw data were analysed using the FlowJo software.

The following antibodies were used for flow cytometer analyses: anti-PD1 PE (clone RMP1-30, BioLegend), anti-CD8-PE/Cy7 (clone 53-6.7, BioLegend), anti-CD44-APC (clone IM7, eBioscience), anti-CD62L-AF700 (clone MEL-14, BioLegend), anti-CXCR5-FITC (clone L138D7, BioLegend), anti-CD4-BV421 (clone GK1.5, BioLegend), GL7-PE (clone GL7, BioLegend), anti-Fas/CD95-PE/Cy7 (clone Jo2, BD Pharmingen), anti-B220-PerCPCy5.5 (clone RA3-6B2, BioLegend), anti-CD138-bio (clone 281-2, BioLegend), anti-TACI/CD267-APC (clone ebio8F10-3, eBioscience), anti-B220-AF700 (clone RA3-6B2, BioLegend) anti-CD21-bio (clone 7E9, BioLegend), anti-CD23-PE (clone B3B4, eBioscience), anti-CD5-PE (clone 53-7.3, BD Pharmingen).

For histological analysis, kidneys were fixed with 4% paraformaldehyde in phosphate buffered saline (PBS), dehydrated in an ascending series of ethanol and xylol followed by embedding in paraffin. Paraffin-embedded tissues were cut into sections of 2 µm thickness and stained with periodic acid Schiff (PAS) using routine protocol. Renal morphology was investigated by light microscopy as described below with the investigator being blinded to the genotype of the mice. Glomerulosclerosis index (GSI) was assessed on PAS-stained paraffin sections at a magnification of 200x. GSI of each animal was derived as the mean of at least 45 randomly sampled glomeruli using a semiquantitative scoring system expressing the severity of glomerulosclerosis on an arbitrary scale from 0 to 4. The glomerular score for individual glomeruli was: grade 0, normal glomerulus; grade 1, presence of mesangial expansion/thickening of the basement membrane; grade 2, mild/moderate segmental hyalinosis/sclerosis involving less than 50% of the glomerular tuft; grade 3, diffuse glomerular hyalinosis/sclerosis involving more than 50% of the tuft; grade 4, diffuse glomerulosclerosis with total tuft obliteration and collapse.

Spleens were harvested and frozen in Tissue-Tek OCT Cryomold (Sakura). At a cryotome, 8 µm slices were produced and fixed on Superfrost Plus adhesion Microscopy slides with -20°C acetone for 10 minutes. After rehydration with 1x PBS for 5 minutes, the spleen cryosections were saturated with 10% FCS and 1% Fc-block (2.4G2, hybridoma generated in house) for 30 minutes. The spleen cryosections were stained with Peanut Agglutinin (PNA)-FITC (Vector Laboratories) and anti-IgD-bio (clone 11-26c, eBioscience) for 45 minutes. The slices were washed with 0,05% Tween20 in 1x PBS and subsequently stained with streptavidin-Cy3 (Jackson ImmunoResearch) for 15 minutes. The spleen cryosections were washed with 1x PBS, mounted with Aqua-Poly/Mount (Polysciences) and covered with cover glasses. All spleen cryosections were analyzed the next day with a Axio Scope.A1 fluorescence microscope (Zeiss) while maintaining the same exposure time. Merge of images and analysis were performed with ImageJ.

The spleens of mice at the age of 15-18 weeks were harvested. Generation of single cell suspensions and erythrocyte lysis were performed as described above. 2.5x10⁵ spleen cells were seeded in 1 ml medium (1% heat-inactivated FCS, 1000 U/ml Penicillin-Streptomycin, 1% Non-Essential Amino Acids, 1 mM sodium pyruvate, 200 nM L-glutamine in RPMI-1640 medium) onto 24-well cell culture plates with or without the addition of either 10 µg/ml LPS (Sigma-Aldrich), 10 µg/ml LPS + 10 ng/ml IL-4 (BioLegend), 10 µg/ml LPS + 0,2 µg/ml anti-CD40 antibody (BioLegend) or 10 µg/ml anti-IgM F(ab’)₂ (Jackson ImmunoResearch). Activation markers were analyzed by flow cytometry after 24 hours by harvesting the cells and staining with anti-CD19-BV421 (clone 1D3, BD Horizon), anti-CD86-FITC (clone GL1, eBioscience) and anti-MHCII-APC (clone AF6-120.1, BioLegend). Antibody concentration in the supernatant was analyzed by ELISA after 6 days of stimulation. Maxisorp plates (96-Well plate; NuncImmuno) were coated with either goat anti-mouse IgM (SouthernBiotech) or goat anti-mouse IgG1 (SouthernBiotech) and saturated with 1% BSA in 1x PBS. The harvested supernatants and IgM or IgG1 standards (both Southern Biotech) were pre-diluted with 0,1% BSA in 1x PBS and added to 96-well plates to produce a 1:3 dilution series. The diluted samples were transferred to the coated ELISA plates. After incubation for 2 hours at 37°C, the ELISA plates were washed and goat anti-mouse IgM-AP (SouthernBiotech) or goat anti-mouse IgG1-AP (SouthernBiotech) were added and incubated for 2 hours at 37°C. After washing, the substrate (4-Nitrophenyl phosphate disodium salt hexahydrate in diethanolamine buffer) was added. The optical density was measured at 405 nm wavelength using a VersaMax microplate reader and analyzed by the Softmax Pro software. Antibody concentration was determined by using the standard curve.

Statistical analyses were performed using GraphPad PRISM software. Outliers were excluded and then the results were tested for normal distribution. For statistical analysis, ANOVA test with Šidák post-hoc test was used in case of normal distribution, whereas the Kruskal-Wallis test with Dunn’s post-hoc test was performed for non-normally distributed samples. Only significant differences in samples are indicated in the figures, non-significant differences are not shown.

In order to study the epistatic effects of genes involved in inhibitory signaling and immune complex clearance on autoimmunity, Siglecg deficient mice (31) were crossed either with Dnase1l3 deficient mice (22) or with a newly generated Dnase1 deficient mouse line. The Dnase1 deficient mouse line was generated by deleting parts of the Dnase1 gene by CRISPR/Cas. This led to a 65 bp deletion from intron 2 into exon 3, thereby deleting the splice acceptor site from exon 3 and to predicted frameshift mutations and premature stop codons by potential alternative splicing ( Supplementary Figure 1A ). The 65 bp deletion in genomic DNA was detected by flanking primers and confirmed by sequencing ( Supplementary Figure 1B ). DNase1 activity was tested by incubation of plasmid DNA with urine of WT mice, which led to DNA digestion due to abundant levels of DNase1 in the urine (38). The same assay did not lead to plasmid DNA digestion when performed with the urine of Dnase1 ^-/- mice ( Supplementary Figure 1C ). Since the Dnase1 ^-/- mice were newly generated, we checked their composition of lymphocytes and activation capacity. We noted no changes in splenocyte numbers or in lymphocyte populations. B cells were analyzed in detail. Dnase1 ^-/- mice had normal numbers of total B cells, follicular B cells, marginal zone B cells in the spleen and B1a cells in the peritoneum ( Supplementary Figure 2A ). When stimulated with various stimuli, B cells of Dnase1 ^-/- mice had a normal upregulation of activation markers ( Supplementary Figure 2B ) and normal antibody production ( Supplementary Figure 2C ). Dnase1 ^-/- mice and all other mouse lines used in this study were on a C57BL/6 background. All mice were aged and analyzed at 24 weeks, 36 weeks, 48 weeks and 60 weeks of age. At 24 and 36 weeks of age no clear differences in autoimmune parameters were observed between the various genotypes, therefore the results from 48 week and 60 week-old animals are presented here.

Spontaneous development of germinal centers is a hallmark in SLE formation in humans and mice (39). Therefore, GC B cell percentages and numbers were determined for Dnase1 ^-/- x Siglecg ^-/-, Dnase1l3 ^-/- x Siglecg ^-/- mice, the corresponding Siglecg ^-/-, Dnase1 ^-/- and Dnase1l3 ^-/- mice as well as wt mice. Siglecg deficiency alone tended to have an enhancing influence on GC B cell numbers in aging mice since both percentage and total GC B cell numbers were increased in Siglecg ^-/- mice compared to the wt mice ( Figures 1A–C ). In contrast, single-deficient Dnase1 ^-/- or Dnase1l3 ^-/- mice did not show this tendency ( Figures 1A–C ). However, both Dnase1 ^-/- x Siglecg ^-/- and Dnase1l3 ^-/- x Siglecg ^-/- double-deficient mice show significantly increased GC B cell numbers at most time points, when compared to single-deficient Dnase1 ^-/- or Dnase1l3 ^-/-mice. However, only Dnase1 ^-/- x Siglecg ^-/- mice showed a relative increase of GC B cells at 60 weeks, when compared to Siglecg ^-/- mice ( Figure 1C ). Similar results were also found in lymph nodes ( Supplementary Figures 3A–C ). Analysis of the GC structure by histology revealed a significantly higher number of GCs only in Dnase1 ^-/- x Siglecg ^-/- mice compared to wt mice, but it is more difficult to quantify small differences in GC numbers by histology ( Figures 1D, E ). We did not detect differences in the sizes of individual GCs in the different mutant mice. Thus, the Siglecg deficiency seems to be mainly responsible for the increase of GC B cell numbers and the Dnase1 deficiency enhances this phenotype.

Beside increased numbers of GC B cells due to spontaneous germinal center formation, plasma cells producing pathogenic autoantibodies contribute to SLE formation and are therefore an established marker (40). Therefore, a flow cytometry analysis of plasma cells in the spleen ( Figures 2A–C ) or in the bone marrow ( Supplementary Figures 3D–F ) was performed. Similar to GC B cells, plasma cell percentages and numbers tended to be increased in Siglecg ^-/- mice at the age of both 48 weeks and 60 weeks in the spleen ( Figures 2B, C ). In contrast, single-deficient Dnase1 ^-/- or Dnase1l3 ^-/- mice did not show this general tendency ( Figures 2B, C ). Again, both Dnase1 ^-/- x Siglecg ^-/- and Dnase1l3 ^-/- x Siglecg ^-/- double-deficient mice showed significantly increased plasma cell numbers at most time points in the spleen, when compared to single-deficient Dnase1 ^-/- or Dnase1l3 ^-/- mice, but not in comparison to Siglecg ^-/- mice ( Figures 2B, C ). Similar tendencies of increased plasma cell numbers in double-deficient mice were also found in the bone marrow, albeit with no significant differences ( Supplementary Figures 3E, F ). Again, the Siglecg deficiency seems to be mainly responsible for the increase of plasma cell numbers in aging mice.

Alongside germinal center B cells, numbers of follicular helper T cells (TFH) were analyzed in this study to determine spontaneous germinal center formation. For this purpose, TFH cell numbers in the spleen of Dnase1 ^-/- x Siglecg ^-/- and Dnase1l3 ^-/- x Siglecg ^-/- as well as the corresponding single knock out mice and wt mice were investigated ( Figures 3A–C ). While at all time points no clear difference in TFH cell numbers of single-KO mice to wt controls could be observed, Dnase1 ^-/- x Siglecg ^-/- or Dnase1l3 ^-/- x Siglecg ^-/- mice showed significantly increased TFH cell numbers at most time points, when compared to single-deficient Dnase1 ^-/- or Dnase1l3 ^-/- mice. Dnase1 ^-/- x Siglecg ^-/- mice also showed a relative increase of TFH cell numbers, when compared with Siglecg ^-/- mice at 60 weeks of age ( Figures 3B, C ). In contrast to these changes of TFH cells, numbers of total CD4 T cells were not changed in Dnase x Siglecg ^-/- mice ( Figure 3D ).

A further phenotype of SLE is a higher transition of naïve T cells to effector memory (EM) T cells and central memory (CM) T cells. To examine whether the ratio of naïve and activated memory T cells is altered in Dnase1 ^-/-, Dnase1 ^-/- x Siglecg ^-/-, Dnase1l3 ^-/- and Dnase1l3 ^-/- x Siglecg ^-/- mice, Siglecg ^-/- and wt mice, splenic T cells were analyzed via flow cytometry ( Figure 3E ). Indeed, the percentages of CD4⁺ naïve T cells in 48 week-old Dnase1 ^-/- x Siglecg ^-/- and Dnase1l3 ^-/- x Siglecg ^-/- mice were significantly reduced compared to wt and Dnase single-KO mice ( Figure 3F , left). However, this is neither visible in total cell numbers nor at 60 weeks of age ( Figures 3F, G ). With respect to the CD4⁺ memory T cells, the CM T cell numbers were comparable between the different genotypes, whereas EM T cell numbers were significantly increased in Dnase1l3 ^-/- x Siglecg ^-/- mice compared to wt and Dnase1l3 single-KO mice ( Figures 3F, G ). Similar tendencies are detectable for CD8⁺ T cell subsets ( Supplementary Figure 4 ).

Next, autoantibodies directed against nuclear components (ANAs) and double stranded DNA (anti-dsDNA) were tested in aging mice. First, sera of naïve aging mice were tested for the presence of ANAs. Both exemplary images and quantification showed that a deficiency in Siglecg leads to more ANAs compared to wt mice at 48 weeks and 60 weeks, although this increase was not significant ( Figures 4A, B ). There was no increase of ANAs detected in Dnase1 ^-/- mice and a trend towards higher ANA levels in Dnase1l3 ^-/- mice at 60 weeks. Dnase1l3 ^-/- x Siglecg ^-/- mice showed a significant increase of ANAs, when compared to Dnase1l3 ^-/- or wt mice ( Figures 4A, B ). As a positive control, sera of sick, 4-5 months old MRL/lpr mice are shown. At 60 weeks, Dnase1l3 ^-/- x Siglecg ^-/- mice reached ANA levels which were comparable to sick MRL/lpr mice.

Additionally, anti-dsDNA antibodies in the serum of the aging mice were quantified via ELISA. No increases of anti-dsDNA antibodies were detected in any single-KO mice, when compared to wt controls ( Figure 4C ). A substantial increase in anti-dsDNA IgG autoantibodies was detected in the serum of naive Dnase1l3 ^-/- x Siglecg ^-/- compared to wt mice and to Siglecg ^-/- and Dnase1l3 ^-/- single knockout mice, however ( Figure 4C ). This increase was already present at the age of 48 weeks and demonstrates epistatic effects of both deleted genes on the development of anti-dsDNA antibodies. Interestingly, in double-deficient Dnase1 ^-/- x Siglecg ^-/- mice no such epistatic effect is observable ( Figure 4C ).

Because deposition of immune complexes in the kidney frequently occurs during the course of lupus disease and can lead to severe renal damage, we examined the kidneys histopathologically and evaluated them for glomerular changes including mesangial matrix expansion, glomerular size and cell counts using periodic acid Schiff (PAS) stain. Examples of PAS-stained glomeruli of the various mice genotypes are shown in Figure 5A . While the glomeruli of Dnase1 ^-/- and Dnase1l3 ^-/- exhibited a normal PAS-positive mesangium, the mesangium of Siglecg ^-/- mice appeared slightly widened. This change was more pronounced in the two double-knockout mice, with the Dnase1l3 ^-/- x Siglecg ^-/- animals exhibiting additional mesangial proliferation, particularly at 60 weeks of age, which also resulted in enlarged glomeruli and narrowing of capillary volume ( Figure 5A ), indicating progressed glomerular injury.

When quantified with a mean glomerulosclerosis index (GSI), single-KO mice did not show significant alterations. However, both Dnase1 ^-/- x Siglecg ^-/- and Dnase1l3^-/- x Siglecg^-/- mice revealed a significantly increased mean glomerulosclerosis index compared to Dnase1 ^-/- and Dnase1l3 ^-/- single knockout mice ( Figure 5B ). The GSI of Dnase1l3^-/- x Siglecg^-/- mice was higher than the GSI of Dnase1 ^-/- x Siglecg ^-/- mice, at both time points. Histopathological changes in the kidney were confined to the glomerulus in investigated mice, with only Dnase1l3 ^-/- x Siglecg ^-/- showing mild tubulointerstitial changes with tubule dilatations and an increased inflammatory response. The renal vessels showed no pathological changes in all mice.

In addition, the blood urea nitrogen (BUN) content was examined in aging Dnase1 ^-/-, Dnase1 ^-/- x Siglecg ^-/-, Dnase1l3 ^-/- and Dnase1l3 ^-/- x Siglecg ^-/- mice, Siglecg ^-/- mice and wt mice to examine the consequences of kidney damage. As depicted in Figure 5C , all tested genetically-modified mice showed no signs of elevated blood urea nitrogen levels, compared to wt controls, neither at 48 weeks of age nor at 60 weeks. The BUN levels found were between 20 to 30 mg/dl. This was in contrast to sick MRL/lpr animals, analyzed as a positive control, with high BUN values around 100 mg/dl. In conclusion, both Dnase1 ^-/- x Siglecg ^-/-, as well as Dnase1l3 ^-/- x Siglecg ^-/- aging mice developed glomerulonephritis, with a more severe disease score detected in the latter mouse line.