Animal experiments were approved by the Netherlands Central Committee for Animal Experiments and the University of Groningen. Mice were housed specific-pathogen free in macrolon cages, a 12 h light–dark cycle, and with ad libitum access to food and water unless specified otherwise. Conditional VISTA KO mice (Cx3cr1creERT2/WT VISTAloxP/loxP) were generated by crossbreeding VISTAloxP/loxP  and B6.129P2(Cg)-Cx3cr1tm2.1(cre/ERT2)Litt/WganJ mice (Jax, 021,160). Genotype was confirmed (see below) before depletion of VISTA in conditional VISTA KO mice by administration of 20 mg tamoxifen (Sigma-Aldrich, T5648) via oral gavage and subsequent 5-week rest to allow turnover of peripheral CX3CR1pos cells (= VISTA KO). Littermate controls (Cx3cr1creERT2/WT VISTAloxP/loxP) treated with corn oil served as control (= VISTA WT). For LPS experiments, 10-week-old conditional VISTA KO mice were injected with 1 mg/kg LPS (E. coli, O111:B4, L4391) intraperitoneally and terminated 3 h later. PBS-injected mice served as control. For EAE experiments, 11-week-old female conditional VISTA KO mice were immunized with MOG35-55 in complete Freund’s adjuvant in the neck and lower back according to manufacturer’s instructions (Hooke, EK-2110). At the time of immunization and after 24 h, mice were injected intraperitoneal (i.p.) with 150 ng pertussis toxin intraperitoneally. Animals were scored daily for disease symptoms and terminated at score 1 (limp tail; early disease), score 4 (hind leg paralysis; peak disease), and chronic (3 weeks after onset of symptoms). Unimmunized mice served as control.
Genotype for experimental animals (Cx3cr1creERT2/WT VISTAloxP/loxP) was confirmed by genomic PCR for Cx3cr1 WT and creERT2 alleles, and for VISTA. Genomic DNA was extracted from earcuts of mice using MyTaq Extract-PCR kit (Bioline, BIO-21127) according to manufacturer’s instructions. Genes and alleles were amplified using MyTaq HS Red Mix (Bioline, BIO-25047) and 10 μM primer and analyzed on 2% agarose gels. For Cx3cr1 WT and creERT2 alleles, two primer-pairs were used for WT allele (fwd: 5′-CTCAC GTGGA CCTGC TTACTG; rev: 5′-GTACC GGTCG ATGCT GATGA) and creERT2 allele (fwd: 5′-AAGAC TCACG TGGAC CTGCT; rev: 5′-CGGTT ATTCA ACTTG CACCA) with the following PCR program: 1) 95 °C 3 min, 2) 95 °C 15 s, 3) 58 °C 15 s, 4) 72 °C 20 s, repeated 30 times. Cx3cr1 WT alleles present with a band at 482 bp, whereas Cx3cr1 creERT2 alleles present with a band at 260 bp. For VISTA, one primer pair was used (fwd: 5′-CTAAT GGCAC AGCAG GGACT; rev: 5′-CAACA AATCA CGGTG GAGTG) with the following PCR program: 1) 95 °C 3 min, 2) 95 °C 15 s, 3) 51.7 °C 30 s, 4) 72 °C 30 s, repeated 35 times. VISTA WT alleles present with a band at 468 bp, whereas VISTA loxP alleles present with a band at 651 bp.
Microglia were isolated from whole brain (LPS) or spinal cord (EAE) as previously described in detail . The whole isolation procedure was performed on ice. Mice were perfused with PBS (Lonza, BE17-512F) and CNS tissue was mechanically dissociated in HBSS (Gibco, 14,170–088) containing 0.6% glucose (Sigma-Aldrich, G8769) and 15 mM HEPES (Lonza, BE17-737E). Myelin was removed by 24.4% percoll (GE Healthcare, 17-0891-01) density gradient centrifugation at 950 g for 20 min at 4 °C. Fc receptors were blocked with anti-CD16/32 (5 μg/ml, clone 93, eBioscience, 14-0161-85), and cells were stained with anti-CD11b-APC-Cy7 (1 μg/ml, clone M1/70, eBioscience, A15390), anti-CD45-PE-Cy7 (1 μg/ml, clone 30-F11, eBioscience, 25-0451-82), anti-Ly6C-APC (1.5 μg/ml, clone HK1.4, Biolegend, 128,016), and anti-VISTA-PE (20 μg/ml, clone MIH63, Biolegend, 150,204) antibodies 30 min at 4 °C in HBSS without phenol red (Gibco, 14,175-053) containing 0.6% glucose, 15 mM HEPES, and 1 mM EDTA (Invitrogen, 15,575-020). Microglia were sorted on a MoFlo Astrios (Beckman Coulter) in siliconized tubes containing RNAlater (Qiagen, 76,104), centrifuged at 5000 g, and lysed in RLT + lysis buffer (Qiagen, 74,034).
RNA extraction and sequencing
RNA was extracted using AllPrep DNA/RNA Micro Kit (Qiagen, 80,284) according to manufacturer’s instructions. Total RNA concentration was measured on TapeStation 4200 and samples with a RIN value below 7 were excluded. Sequencing libraries were generated using a NEBNext single cell/low input RNA library prep kit for Illumina (NEBNext, E6429), which was modified to include unique molecular identifiers for removal of PCR duplicates. Libraries were sequenced at 20 million read depth on an Illumina NovaSeq6000.
After removal of low-quality reads and adapter sequences, reads were mapped to the mouse GRCm38.p6 genome using Tophat (v2.0.14)  with default settings. Feature counting to quantify gene expression was done using HTSeq (v0.11.0) .
Differential gene expression analysis
Low expressed genes with no expression in at least two samples were excluded before DESeq2 R-package (v1.30.0)  was used for transformation, normalization, and differential gene expression analysis. Two samples were excluded based on low library size and complexity. Variance stabilizing transformation of counts was done for principal component analysis (PCA). Genes were regarded differentially expressed with a log2 Fold change > 1 and a Benjamini–Hochberg adjusted p-value of < 0.05. Additional file 1: Tables S4 and S5 provide normalized expression of genes for each sample and results of differential gene expression analysis for LPS and EAE experiments.
Weighted gene co-expression network analysis (WGCNA)
After filtering of low expressed genes as above, variance stabilizing transformed counts were used as input for WGCNA (v1.69) . Genes with zero variance and missing values were excluded before constructing a signed network using dissimilarities of topological overlap matrix. Modules with a minimum size of 30 genes were constructed and merged with a threshold of 0.25. Module trait correlation was regarded significant with a p-value < 0.05.
Enrichment for biological processes and transcription factors
To determine potential biological processes and transcription factors associated with genes in WGCNA modules and differentially expressed genes, enrichr (v3.0)  was used to perform an enrichment analysis for gene ontology biological processes, molecular signatures database hallmarks, and ENCODE/ChEA transcription factor targets.
Primary neonatal mouse microglia culture
Primary neonatal mouse microglia cultures were generated from P0-3 pups as previously described [6, 40]. After removal of meninges and cerebellum, the cerebrum was minced and incubated in HBSS containing 0.6% glucose, 1 × DNase (Sigma-Aldrich, DN25), 0.25% trypsin (Lonza, BE02007E), 1% penicillin–streptomycin (Gibco, 15,140,122), and 15 mM HEPES (Lonza, BE17-737E) for 20 min. The solution was mechanically dissociated further and centrifuged at 230 g for 10 min. Cells were plated in flasks in DMEM (Gibco, 1,500,416) containing 1 mM sodium pyruvate (Lonza, BE13-115E), 1 × GlutaMAX (Gibco, 35,050,038), 1% penicillin–streptomycin, and 10% FCS (Gibco, 10,500,064) (microglia medium). Medium was replaced after 1, 4, and 6 days. After 9 days, medium was replaced with fresh medium containing 33% L929 cell-conditioned medium that contains M-CSF to stimulate microglia proliferation. Microglia were harvested after 2 days by mitotic shake-off and seeded at 30,000 cells/cm2 in experimental plates. For deletion of VISTA, primary microglia cultures were prepared from pups of Cx3cr1creERT2/WT VISTAloxP/loxP mice. After mitotic shake-off, microglia were treated with 0.1 μM (Z)-4-hydroxytamoxifen (Sigma-Aldrich, H7904) or 0.01% ethanol for 24 h and left to rest for 48 h in microglia medium. Subsequently, microglia were used for phagocytosis assay (see below) or stimulated for 5 h with 100 ng/ml Pam3CSK4 (Invivogen, #tlrl‐pms), 100 ng/ml LPS (E. coli 0111:B4, Sigma‐Aldrich, L4391), 50 μg/ml polyI:C (Invivogen, #tlrl‐pic), 10 μg/ml beta-glucan (Sigma-Aldrich, G5011), 10 ng/ml IL1b (Peprotech, 400-18), 10 ng/ml TNFalpha (Peprotech, 400-14), 10 ng/ml IL10 (Peprotech, 400-19), 20 mg/ml IL4 (Peprotech, 400-04), 10 ng/ml TGFbeta (Peprotech, 100-16), or mouse myelin (see below).
Induction of early apoptotic Jurkat cells
Jurkat human T cell line was kindly provided by Prof. J. Smit (UMCG) and cultured in RPMI (Gibco, 21,875,034) supplemented with 10% FBS, 1 × penicillin–streptomycin, and 1 mM sodium pyruvate. Apoptosis induction for phagocytosis assay (described below) was done on the day of performing the assay as previously described . Jurkat cells were collected at 1*106 cells/mL in medium and treated with 1 μM staurosporine (Sigma-Aldrich, S5921) for 4 h. To verify induction of apoptosis, Jurkat cells were labelled with Annexin V/propidium iodide (Biolegend, 640,914) according to manufacturer’s instructions and analyzed on a MacsQuant (Miltenyi Biotec). After 4 h of staurosporine exposure, > 90% of cells were early apoptotic (Additional file 7: Fig. S6D).
Myelin from 10 week old C57BL/6 mouse whole brains was isolated as described previously  with minor adjustments. Whole brains were mechanically dissociated in HBSS containing 0.6% glucose and 15 mM HEPES and centrifuged in 24.4% percoll at 950 g for 20 min. The upper layer containing myelin was collected, diluted 1:3 in HBSS, and centrifuged at 950 g for 15 min. The pellet was resuspended in 0.32 M sucrose and a layer of 0.85 M sucrose was carefully added on top. The solution was centrifuged at 75,000 g for 30 min at 4 °C and the cloudy interphase between the two sucrose layers was collected and resuspended in Milli-Q water. After centrifugation at 75,000 g for 30 min at 4 °C, the pellet was washed twice by resuspending in Milli-Q water and centrifuging at 13,500 g for 15 min at 4 °C. As before, the pellet was resuspended in 0.32 M sucrose and 0.85 M sucrose was added on top before centrifuging at 75,000 g for 30 min at 4 °C and collecting the cloudy interphase. The interphase was diluted in Milli-Q and centrifuged at 75,000 g for 15 min at 4 °C and the remaining pellet containing pure myelin was resuspended in 1 mL sterile PBS. The Pierce BCA protein assay kit (Thermo Scientific, 23,225) was used to determine the myelin concentration.
Fresh early apoptotic Jurkat cells (EAJ) and mouse myelin were labelled with pHrodo Red succinimidyl ester (Invitrogen, P36600) according to manufacturer’s instructions. Briefly, 10*106 EAJ cells/mL or 1 mg/mL mouse myelin was mixed with 0.1 mg/mL pHrodo Red succinimidyl ester and incubated for 1 h at RT with intermittent mixing. Labelled compounds were diluted in cold PBS and centrifuged before resuspending the pellet in DMEM containing 1 mM sodium pyruvate, 1 × GlutaMAX and 1% penicillin–streptomycin (quiescence medium).
Primary neonatal mouse microglia were stimulated with LPS or PBS 5 h prior to the addition of phagocytosis compounds. Microglia were pretreated with 10 μM cytochalasin D (Sigma-Aldrich, C8273) to inhibit phagocytosis or with 0.1% DMSO before adding 20 μg/mL pHrodo-labelled myelin, 20 μg/mL pHrodo red E. coli BioParticles (Invitrogen, P35361), and pHrodo-labelled EAJ at equal numbers as seeded microglia. PHrodo signal was detected every 15 min over 12 h using an IncuCyte Live Cell Analysis System (Sartorius).
Total RNA from primary neonatal mouse microglia was isolated using TRIzol (Invitrogen, 15,596,018) and cDNA was generated using RevertAid First Strand cDNA Synthesis Kit (ThermoFisher, K1622) according to manufacturer’s instructions. Quantitative PCR was done using iTag Universal SYBR Green Super-Mix (Bio-Rad, 1,725,125) and exon-exon spanning primers on a QuantStudio 7 Flex (ThermoFisher). Primer pairs used were for VISTA (fwd: 5′- AACAA CGGTT CTACG GGTCC; rev: 5′-CGTGA TGCTG TCACT GTCCT), Tnf (fwd: 5′- TCTTC TGTCT ACTGAA CTTCGG; rev: 5′- AAGAT GATCT GAGTGT GAGGG), Ccl2 (fwd: 5′-TCAGC CAGAT GCAGT TAACG; rev: 5′-CTGGT GATCC TCTTG TAGCTC), and Hprt1 (fwd: 5′-ATACA GGCCA GACTTT GTTGGA; rev: 5′-TGCGC TCATCT TAGGC TTTGTA).
Primary neonatal mouse microglia were detached using accutase (Sigma-Aldrich, A6964) for 10 min at 37 °C and resuspended in HBSS without phenol red containing 15 mM HEPES, 0.6% glucose, and 1 mM EDTA. Microglia were blocked using anti-CD16/32 (5 μg/ml) for 15 min at 4 °C and stained with anti-VISTA-PE (20 μg/ml) for 30 min at room temperature (RT). Microglia were analyzed on a MacsQuant (Miltenyi Biotec).
Immunohistochemical staining was performed on formalin-fixed paraffin-embedded (FFPE) (human) or paraformaldehyde-fixed frozen (FF) (mouse) tissue. FFPE tissue was deparaffinized in xylene (J.T. Baker, 9490) and rehydrated. FF tissue was dried in an exsiccator. Heat-induced epitope retrieval was performed in a microwave in sodium citrate (pH = 6.0) using a pressure cooker. Endogenous peroxidase activity was blocked for enzymatic immunohistochemistry using 0.3% hydrogen peroxide for 30 min. Mouse tissue was additionally blocked 1 h in 5% normal serum. Antibodies were diluted in PBS containing 0.1% Triton-X (mouse) and 1% normal serum, or in Normal Antibody Green Bright Diluent (human) (ImmunoLogic, BD09-500). Antibodies used were anti-VISTA (1:200, clone D1L2G, Cell Signaling, #64,953), anti-IBA1 (1:1000, polyclonal, Wako, 019-19,741), anti-HLA-DR (1:750, clone LN3, eBioscience, 14-9956-82), anti-PLP (1:500, clone plpc1, Bio-Rad, MCA839G), anti-CD68 (1:500, clone PG-M1, Dako, M0876), and anti-TMEM119 (1:500, polyclonal, Atlas Antibodies, HPA051870). Biotinylated secondary antibodies (1:400, Vector, BA-1000 and BA-2001) were applied for 1 h at RT. Vectastain Elite ABC-HRP (Vector, PK-6100) was applied and immunoreactivity was revealed using 3,3′-diaminobenzidine (DAB).
Microglia morphological analysis
Mouse tissue was stained with anti-IBA1 and slides were scanned on a NanoZoomer Digital Pathology System (Hamamatsu Photonics) with a 40 × objective. An analysis tool was developed [27, 51] to evaluate 27 microglia morphological features. Briefly, images of at least 20 randomly selected individual microglia in cortex per mouse were processed to obtain cell silhouette images using semi-automated thresholding. Following thinning and pruning of branch areas to obtain cell skeletons, branch endings (end nodes), branch crossings (junctions) and all branchpoints going out from the cell soma (start nodes) were marked. Cell silhouettes and skeletons were used for fully automated morphological analysis including Sholl analysis and 23 other morphometric features. A non-supervised clustering algorithm was used to identify microglia subpopulations based on morphology. To this end, morphometric features were normalized and scaled before reducing the dimensionality using principal component analysis (PCA). Hierarchical clustering using Ward’s method was used to determine the top-contributing PC with an eigenvalue > 1. Morphometric features of individual cells are provided in Additional file 1: Table S2.
Statistical analyses were performed using GraphPad Prism (v8.4.0). The type of statistical test used is indicated in the figure legends. Values of p < 0.05 were considered significant: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.