Genetic background variation impacts microglial heterogeneity and disease progression in amyotrophic lateral sclerosis model mice

Summary Recent single-cell analyses have revealed the complexity of microglial heterogeneity in brain development, aging, and neurodegenerative diseases such as amyotrophic lateral sclerosis (ALS). Disease-associated microglia (DAMs) have been identified in ALS mice model, but their role in ALS pathology remains unclear. The effect of genetic background variations on microglial heterogeneity and functions remains unknown. Herein, we established and analyzed two mice models of ALS with distinct genetic backgrounds of C57BL/6 and BALB/c. We observed that the change in genetic background from C57BL/6 to BALB/c affected microglial heterogeneity and ALS pathology and its progression, likely due to the defective induction of neurotrophic factor-secreting DAMs and impaired microglial survival. Single-cell analyses of ALS mice revealed new markers for each microglial subtype and a possible association between microglial heterogeneity and systemic immune environments. Thus, we highlighted the role of microglia in ALS pathology and importance of genetic background variations in modulating microglial functions.

The role of microglia in ALS has been extensively studied, but their function remains debatable due to their double-edged sword nature 15 and the complexity of heterogeneity.Infiltrating T-lymphocytes can regulate the balance between the beneficial and detrimental functions of microglia by inducing the secretion of insulin-like growth factor 1 (IGF-1), a neurotrophic factor. 16In addition, the upregulated expression levels of genes including Igf1 and DAM-related genes in microglia could correlate with the infiltration ratio of T-lymphocytes in mutant SOD1-expressing ALS model mice. 17Recent scRNA-seq analysis revealed the predominance of Igf1 expression in DAM. 3,18Furthermore, alterations in the peripheral immune system (in both immune cell populations and functions) have been reported in patients with ALS and in the ALS mouse model. 19Therefore, the peripheral immune environment may affect microglial heterogeneity and the disease course in ALS, but whether this environment affects microglia or immune cells in the CNS and ALS progression remains unclear.
This study explored the effect of genetic background variations on microglial heterogeneity and functions in ALS model mice.Following the generation of two kinds of SOD1 G93A ALS model mice with different genetic backgrounds, C57BL/6 and BALB/c strains, as well as wildtype littermates, single-cell RNA-seq analysis was performed in the spinal cords of the mice, followed by survival analysis.We established for the first time that the genetic change from C57BL/6 to BALB/c background affects microglial heterogeneity and ALS pathology, as evidenced by the reduced induction of DAM clusters, fewer surviving microglia, and accelerated disease progression in ALS mice.In addition, a neuroprotective microglial subtype was identified in these DAM clusters.Our findings suggest that environmental factors derived from peripheral immune cells may contribute to the phenotypic differences observed in microglia.

RESULTS
Single-cell RNA sequencing analysis revealed that genetic background variation affected microglial heterogeneity in amyotrophic lateral sclerosis model mice SOD1 G93A mice carrying a low copy number of mutant SOD1 in the C57BL/6J background (G93AL(B6)) were used, which spontaneously lost transgene copies ($À40%) during laboratory breeding.G93AL(B6) mice were backcrossed over 10 times with BALB/cA mice to generate congenic mice (G93AL(Balb)), and quantitative PCR (qPCR) was used to confirm the comparability of the transgene copy number for each mouse.
Low-copy SOD1 G93A mice with the BALB/c background displayed faster disease progression with a decreasing number of microglia To investigate the effect of altered microglial heterogeneity on ALS pathogenesis in mice, we performed survival analyses on G93AL(B6) and G93AL(Balb) mice having the same transgene copy numbers (Figure S2A).The G93AL(Balb) mice exhibited a significantly shorter survival time than the G93AL(B6) mice, despite having comparable onset times and time of 10% body weight loss (survival time: G93AL(B6), 225.2 G 21.7 days; G93AL(Balb), 203.2 G 22.1 d; onset time: G93AL(B6), 145.3 G 10.2 days; G93AL(Balb), 147.2 G 13.2 days; time of 10% body weight loss: G93AL(B6), 201.7 G 14.6 days; G93AL(Balb), 191.9 G 23.9 days; disease duration: G93AL(B6), 79.9 G 24.0 days; G93AL(Balb), 55.9 G 18.1 d, mean G SD) (Figures 2AÀ2F), indicating that G93AL(Balb) mice accelerated the late phase of disease progression.Although the sample number was very small, it was previously reported that another low-copy SOD1 G93A mice with the BALB/c background also had a shorter survival time than those with the C57BL/6 background. 23o investigate microglia and astrocyte phenotypes, we next performed triple-immunofluorescence staining with antibodies against Iba-1, Mac2, and glial fibrillary acidic protein (GFAP) in the G93AL(B6) and G93AL(Balb) spinal cords at the disease end-stage.Despite comparable microglial activation in both strains, the number of G93AL(B6) microglia exceeded that of G93AL(Balb) microglia (Figure 2G).However, astrocyte morphology, gene expression of astrocyte marker, Gfap, and matured oligodendrocyte numbers did not differ between these strains (Figures 2G and S2BÀS2D).
Flow cytometric analysis, which was performed to clarify the time course of microglial number during disease progression by staining the spinal cord microglia with cell surface marker CD45 and to distinguish them from lymphocytes and monocytes based on a lower level of CD45 expression (Figures S3A and S3B), revealed that the G93AL(B6) and G93AL(Balb) spinal cords displayed a gradual increase and decrease in microglial number after disease onset, respectively, compared to each WT (Figure 2H).
To clarify the cause of the decreased microglial number of in the G93AL(Balb) spinal cord, we analyzed microglial apoptosis.Doubleimmunofluorescence staining with antibodies against cleaved caspase 3 and Iba-1, which was performed to identify apoptotic microglia, showed that the G93AL(Balb) spinal cord displayed a higher number of apoptotic microglia than the G93AL(B6) spinal cord at the disease end-stage (Figures 2I and 2J), suggesting the defective proliferation and survival of G93AL(Balb) microglia.
Defective production of macrophage colony stimulating factor in microglia was observed in low-copy SOD1 G93A mice with the BALB/c background To determine whether the defective production of microglia growth factors affected microglia survival and proliferation in the G93AL(Balb) spinal cord, we examined the expression of microglia growth factors, including macrophage colony stimulating factor (M-CSF), IL-34, and granulocyte-macrophage-colony-stimulating factor (GM-CSF), in the lumbar spinal cords at the disease end-stage.While M-CSF expression was strongly induced in G93AL(B6) spinal cords, its expression was hardly induced in G93AL(Balb) spinal cords (Figure 3A).Meanwhile, decreased IL-34 expression levels were observed in both ALS models compared with each WT (Figure 3A).Reduced IL-34 expression levels may reflect the loss of motor neurons in both ALS models, as IL-34 is mainly produced by neurons. 24he scRNA-seq analysis revealed that microglia, especially the DAM cluster mainly express M-CSF in G93AL(B6), but the number of M-CSF-expressing microglia was lower in G93AL(Balb) (Figure 3B).To confirm the results, we next identified the M-CSF-expressing cells in the spinal cords by double-immunofluorescence staining combined with in situ hybridization.M-CSF-expressing cells were detected by in situ hybridization, after which microglia and astrocyte were stained with antibodies against Iba-1 and GFAP in the spinal cords at the disease end-stage.Similar to the scRNA-seq result, M-CSF was found to be mainly produced by microglia themselves in the G93AL(B6) spinal cord, while its number was lower in the G93AL(Balb) spinal cord (Figure 3C).
To confirm our findings, we isolated microglia from the spinal cords by magnetic-activated cell sorting (MACS) and quantified mRNA levels by RNA sequencing analysis.M-CSF expression was significantly induced in G93AL(B6) microglia but marginally induced in G93AL(Balb) microglia, while the M-CSF receptor (M-CSFR) expression did not differ between these strains (Figure 3D).The decreased expression of M-CSFR in both ALS model microglia may be attributed to microglial activation.These results suggest that these phenotypes are achieved through the cell-autonomous effect caused by different genetic backgrounds.
Therefore, we compared microglial phenotypes in vitro using primary cultured microglia derived from C57BL/6J and BALB/c mice.However, no differences were observed in the expression levels of M-CSF and M-CSFR in LPS-stimulated microglia (Figure 3E) or in the microglial proliferation of M-CSF-stimulated microglia (Figure 3F).These results indicate that extracellular environments in vivo may influence ALS pathogenesis by altering microglial growth and response.
Altered neuroinflammation-and neuroprotection-related gene expression and limited infiltration of peripheral immune cells were observed in low-copy SOD1 G93A mice with the BALB/c background To examine the changes in neuroinflammation-and neuroprotection-related gene expressions, we quantified the expression of activation markers of microglia, proinflammatory and phagocytic markers, and a neurotrophic factor by qRT-PCR.Although the expression of the typical activation marker of microglia, Lgals3 (Mac2), did not significantly differ between G93AL(B6) and G93AL(Balb) lumbar spinal cords, the induction of proinflammatory markers (Cd86, Ccl5, and Cxcl10), a phagocytotic marker (Cd68), and a neurotrophic factor (Igf1), in G93AL(Balb) lumbar spinal cords was weaker than in G93AL(B6) (Figure 4A).
A subpopulation of T-lymphocytes can infiltrate the spinal cord and exert a neuroprotective effect by inducing IGF-1 production from microglia in SOD1 G93A mice. 16As Ccl5 and Cxcl10 chemokines and Igf1 were weakly expressed in the G93AL(Balb) lumbar spinal cord, we further examined the infiltration of immune cells into the spinal cord during the disease course.While the infiltration of CD45 high /CD3 + cells (T-lymphocytes) and CD45 high /CD3 À cells (non-T-lymphocytes) incrementally increased after onset in the G93AL(B6) spinal cords, it was scarcely observed in G93AL(Balb) spinal cords (Figure 4B).The main populations of immune cells infiltrating the G93AL(B6) spinal cord were CD8 + T-lymphocytes (CD8 + -T), natural killer T-lymphocytes (NK-T), and natural killer (NK) cells (Figure 4C).These results corroborated our previous research on SOD1 G93A spinal cords carrying a high copy number of transgenes. 25NA sequencing analysis of MACS-isolated microglia at disease end-stage revealed a distinct gene expression profile in low-copy SOD1 G93A microglia with the BALB/c background compared to M1, M2, or DAM microglia.
To elucidate the differences in microglial responses and properties between G93AL(B6) and G93AL(Balb), we directly isolated microglia from the spinal cords by MACS and performed RNA sequencing analysis at the disease end-stage.Principal-component analysis (PCA) revealed that the gene expression profiles of microglia were largely separated between G93AL(B6) and G93AL(Balb) mice (Figure 5A).Gene enrichment analysis of upregulated genes (Fold change >1.5, q < 0.05) among all pairs of strains revealed that some gene sets in G93AL(B6) microglia were specifically enriched in gene ontology (GO) terms of ''response to interferon-gamma,'' ''regulation of leukocyte cell-cell adhesion,'' ''response to virus,'' ''adaptive immune response,'' and ''response to interferon-beta'' and Reactome pathway of ''immunoregulatory interactions between a lymphoid and a non-lymphoid cell'' (Figure 5B).Consistent with the gene enrichment analysis, TRRUST (transcriptional regulatory relationships unraveled by sentence-based text mining) 26 analysis also revealed that the identified gene sets were specifically regulated by transcription factors of RFX (regulatory factor binding to the X-box) complex that regulate MHC-class II (major histocompatibility complex class II) molecules and STAT1 (signal transducer and activator of transcription 1), a downstream of cytokine signaling pathways (Figure 5C).The comparison of the expression levels of the top 30 genes 3,27 predominantly expressed in M1, M2, DAM, and homeostatic microglia subtypes between G93AL(Balb) microglia and G93AL(B6) microglia, which was undertaken to examine phenotypic shifts, revealed that while some genes were upregulated, most of them were downregulated in G93AL(Balb) microglia (Figures 5D and S4AÀS4C).It seems that most of the G93AL(Balb) microglial genes are shifted from homeostatic microglia to DAM, but the gene profile was not consistent with typical DAM (Figures S4B and S4C), suggesting that G93AL(Balb) microglia exhibit a distinct gene expression profile.In addition, gene enrichment analysis revealed that a downregulated gene set in G93AL(Balb) microglia was enriched in GO terms of ''T cell activation'' and ''response to interferonbeta,'' while an upregulated one was enriched in terms of ''regulation of cell adhesion'' and ''actin filament-based process'' (Figures S5A  and S5B).

Part of disease-associated microglia cluster 5 in G93AL(B6) microglia exhibited neuroprotective properties
The comparison of the DAM marker gene expressions of each DAM cluster among all strains undertaken to thoroughly investigate the characteristics of DAM subtypes showed that the expression levels of DAM marker genes were particularly enriched in cluster 5 (Figure S7), with some DAM marker genes being oppositely regulated between G93AL(B6) and G93AL(Balb) microglia (Figures 7A and 7B).Specifically, Igf1 was highly expressed and distributed in G93AL(B6) DAM clusters 4, 5, 8, and 10, whereas Csf1 was highly expressed and distributed in G93AL(B6) DAM clusters 5, 6, and 10 (Figures 7A, S7, and S8A).In addition, Igf1-expressing cells barely co-expressed Csf1 (Figure S8B).
Because IGF-1 expression, especially in microglia, plays a neuroprotective role in ALS disease, 16,28,29 we next focused on Igf1-expressing microglia and examined their gene expression profiles.Igf1-positive cells were sorted based on their expression levels (read count >2) to obtain differentially expressed genes (DEGs) (log fold-change >0.25, adjusted p < 0.05) (data not shown).DEGs from each DAM cluster (data not shown) (log fold-change >0.25, adjusted p < 0.05) were also obtained, and all the obtained DEGs were then compared using Metascape's gene enrichment analysis.We found that DEGs of Igf1-positive cells mostly overlapped with those of DAM cluster 5 (Figure S8C) and were enriched in GO terms and Reactome pathways of ''response to wounding,'' ''inflammatory response,'' ''positive regulation of cell migration,'' ''Neutrophil degranulation,'' and ''regulation of apoptotic signaling pathway'' in common with DAM cluster 5 (Figure 7C).These results indicate that a part of Igf1-expressing DAM cluster 5, especially in G93AL(B6) mice, could exert a neuroprotective effect.
To extract genes that could exhibit neuroprotective functions, we merged the DEGs of Igf1-positive microglia with upregulated genes in cluster 5 of G93AL(B6) versus that of G93AL(Balb) (log fold-change >0.25, adjusted p < 0.05), from which one-third of each gene list was found to overlapped (Figure S8D).Similar to the Howell group report, 7 ribosomal genes were mostly enriched in the common 153 genes as ''SRPdependent co-translational protein targeting to membrane'' of the Reactome pathway, implying that some experimental conditions may have affected this result, as mentioned by them.Meanwhile, the common genes, with the exception of their ribosomal genes, were enriched in GO terms of ''negative regulation of cell activation'' (including Apoe, Cst7, Cd74, Lgals3, Tnfaip3, Tyrobp, Trem2, Gpnmb, Flt1, and Igf1), ''regulation of neuron death'' (including Apoe, Csf1, Igf1, Tyrobp, Trem2, and Gpnmb), ''response to type II interferon'' (including Cd74, Csf1, Flt1, Lgals3, Mif, and Trem2) and ''negative regulation of neuroinflammatory response'' (including Cst7, Igf1, Trem2, Tnfaip3, Gpnmb, Flt1, Tyrobp, Apoe, Csf1, Cd74, and Lgals3) (Figure 7D; Table S2).9][30][31][32][33] CD74 can function as an Mif receptor. 34Flt1 encodes VEGFR1 (vascular endothelial growth factor receptor 1), and the genetic association of VEGF with ALS and its neurotrophic function in ALS model mice have been previously reported. 35,36Based on our accumulating evidence, we infer that part of DAM cluster 5 in G93AL(B6) microglia exhibits neuroprotective properties.To determine whether a cell-autonomous (intracellular) or non-cell-autonomous (extracellular) factor significantly affects microglial phenotypes, we compared the gene expression profiles of microglia isolated from all strains and primary cultured microglia from WT mice (in vitro (B6) and in vitro (Balb)).PCA revealed that the contribution ratio of PC1 likely derived from multiple environmental effects was highest (57.6%), while those of PC2 likely derived from differences between healthy (normal) and ALS disease (with mutant SOD1) and PC3 likely derived from a cell-autonomous effect caused by the clear separation of each primary cultured microglial cell were 16.4% and 10.5%, respectively (Figure 7E).Overall, our in vitro results (Figures 3E and 3F) and these results indicate that cell-autonomous differences in gene expression alone cannot account for all microglial phenotype variations in vivo.Notably, inbred C57BL/6 and BALB/c strains usually exhibit biased peripheral helper T cell responses, Th1, and Th2, respectively.At the disease end-stage, we found altered ratios of peripheral immune cell populations and Th1-or Th2-biased peripheral immune responses between G93AL(B6) and G93AL(Balb) mice (Figures S9AÀS9F), indicating that the peripheral immune environment may affect disease progression by regulating microglial heterogeneity, survival, and DAM induction.

DISCUSSION
This study establishes for the first time that genetic background variations affect microglia heterogeneity, their responses, and disease progression in ALS model mice.The use of scRNA-seq analysis identified characteristic markers of microglia subtypes.Intriguingly, even in WT mice, a shift in genetic background from C57BL/6 to BALB/c altered the distribution of homeostatic microglia subtypes, significantly suppressing the induction of the neurotrophic factor-producing DAM subtype and the production of microglia growth factor in ALS model mice.As determined by PCA and detailed analyses of peripheral immune cells, microglia heterogeneity was found to correlate with the systemic immune environment in ALS model mice.
We found that G93AL(Balb) mice exhibited faster disease progression in the late phase than G93AL(B6) mice (Figures 2AÀ2F) and that the induction of DAM clusters 5 and 6 in G93AL(B6) microglia was stronger than that in G93AL(Balb) mice (Figures 6AÀ6D).Importantly, the expression levels of DAM marker genes were particularly enriched in cluster 5, and the DEGs of Igf1-positive cells mostly overlapped with those of cluster 5 (Figures 7C and S8C).The expression levels of neuroprotective genes Igf1, Gpnmb, Lgals3, and Mif, which have been previously reported in ALS model mice, [28][29][30][31][32][33] were found to be enriched in the overlapped 153 genes (Figure S8D) as well as Csf1 and Tnfaip3.We found that the induction of Csf1-encoded M-CSF was limited in G93AL(Balb) microglia (Figures 3AÀ3D).Notably, M-CSF can function as a microglial growth factor and neurotrophic factor. 37The anti-inflammatory activity of Tnfaip3-encoded A20 has been reported. 38Therefore, we concluded that a part of DAM cluster 5 in G93AL(B6) microglia could confer neuroprotection against ALS progression.
We also identified Trem2, Apoe, and Tyrobp in the overlapped 153 genes of cluster 5 and DEGs of Igf1-positive cells.While they were increasingly expressed in all DAM clusters of G93AL(B6), Apoe expression increased in those of G93AL(Balb) and Trem2 and Tyrobp expression only slightly increased in them (Figure S10).Intriguingly, Trem2 and Tyrobp expression levels in the homeostatic microglia clusters of WT(B6) exceeded those of WT(Balb), while Apoe expression was the opposite (Figure S10).4][5] DAP12 encoded by Tyrobp, a downstream molecule of TREM2, could also contribute to DAM induction.Therefore, our results suggest that elevated expression levels of Trem2/Tyrobp and Apoe may be required for the induction of DAM clusters 5 and 6, while the elevated expression of Apoe alone may be sufficient to induce other DAM clusters.Although Amit and colleagues revealed two stages of DAM transition that depend on whether or not Trem2 is expressed, 3 we could not clearly detect any difference in the expression level of Trem2 among DAM clusters, even in G93AL(B6) mice (Figure S10).However, our results corroborated those of Howell and colleagues who reported no evidence for TREM2-dependent DAM transition even in AD model mice. 7Overall, elevated Trem2 expression may be necessary for inducing neuroprotective DAM clusters, although the existence of two stages of DAM transition remains unclear, indicating the need for further study to clarify TREM2 involvement in neuroprotective DAM induction in ALS.0][41] Recently, the anti-inflammatory and phagocytic clearance roles of TREM2 have been identified as potential therapeutic targets for ALS. 42,43he scRNA-seq analysis of microglia revealed for the first time that, unlike WT(Balb) mice, the distribution of homeostatic MG clusters in WT(B6) mice was skewed toward clusters 0 and 1, which exhibited high Socs3 expression (Figures 6C, 6D, S6A, and S6B; Table S1).SOCS3 (suppressor of cytokine signaling 3) is a protein that plays a crucial role in negatively regulating cytokine signaling pathways by binding to Janus kinase 2 (JAK2).Several cytokines including interleukin 6 (IL-6), IL-10, and interferon gamma (IFN-g) induce this protein in macrophages and neutrophils. 44,45Hence, the observed bias in homeostatic microglial subpopulations in WT(B6) mice may be attributable to differences in spinal cord cytokine environments.Notably, Il6 expression was higher in WT(B6) microglia than in WT(Balb) microglia (Figure S4A, M1 marker  (B and C) The numbers of infiltrated CD45 high /CD3 + cells, CD45 high /CD3 À cells, and subsets of immune cells isolated from spinal cords were quantified by flow cytometric analysis at pre-onset (130 dA), onset (145 dA), 10% body weight loss (10%WL) (WT(B6) and G93AL(B6): 200 dA, WT(Balb) and G93AL(Balb): 190 dA), and end-stage of the disease (pre-onset, 10%WL, and end-stage: n = 3 each; onset: WT(B6), G93AL(B6), and WT(Balb), n = 3 each; G93AL(Balb), n = 5).Although CD4 + T, CD8 + T, NK-T, CD4 À /CD8 À double-negative (DN)-T and NK cells were mainly infiltrated in G93AL(B6) spinal cords time-dependently after onset, these infiltrations were limited in G93AL(Balb) mice.Data were represented as mean G SD. two-way ANOVA followed by Tukey-Kramer multiple comparison post hoc tests at each time point.*p < 0.05, **p < 0.01, ****p < 0.0001; G93AL(B6) vs. WT(B6) as indicated by blue color; G93AL(Balb) vs. WT(Balb) as indicated by red color.genes).Furthermore, because the C57BL/6 strain exhibits biased peripheral helper T cell responses toward IFN-g-producing Th1 (Figures S9CÀS9F), IFN-g derived from peripheral CD4 + helper T cells may also contribute to this difference.In addition, we newly identified distinct marker genes including Socs3 that separate each microglial cluster (Figures S6A and S6B; Table S1).Specifically, Klf2 (Kruppel-like factor 2), Ier2 (immediate-early response gene 2), Socs3, Fosb, and Crybb1 expressions separate homeostatic MG subtypes into four clusters (Table S1).Klf2, Ier2, and Fosb are immediate-early genes (IEGs) that encode transcriptional factors immediately induced by environmental stimuli.Both Barres and Kaminska groups independently identified IEGs expressing microglial subpopulations using single-cell analyses, but the former group suggested that the sorting procedure could induce IEG expressions such as Klf2, Ier2, and Fosb, as these expressions were hardly detected on microglia in vivo. 2,46,47Therefore, Socs3 and Crybb1 should be more appropriate markers to distinguish these . RNA sequencing analysis of MACS-isolated microglia at disease end-stage showed a distinct gene expression profile in low-copy SOD1 G93A microglia with the BALB/c background compared to M1, M2, or DAM microglia (A) PCA of gene expression profiles of MACS-isolated microglia at the disease end-stage of G93AL mice as well as age-matched WT mice with each strain (n = 4 each).(B and C) Gene enrichment analysis of upregulated genes among all pairs of strains by Metascape (B) or TRRUST (C).(D) Most of the M1, M2, and DAM marker genes were downregulated in G93AL(Balb) microglia relative to G93AL(B6) microglia.Relative fold changes in the indicated genes are plotted.
Finally, we compared gene expression profiles among isolated microglia from both WT and ALS model mice of each strain and cultured microglia of each strain to assess the effect of environmental factors derived from genetic background variations on microglial phenotypes.PCA revealed that the contribution ratio derived from the cell-autonomous effect (PC3) was only 10.5% (Figure 7E).No differences in the expression levels of Csf1 and Csf1r and the proliferation of cultured microglia were observed between C57BL/6 and BALB/c (Figures 3E  and 3F).Therefore, we inferred that environmental factors strongly affect microglia phenotypes.Amit and colleagues compared microglia between germ-free (GF) mice and control mice at postnatal and adult ages and revealed that microglia from GF mice exhibited the Reclustering analysis of single-cell microglia subsets showed that some parts of DAM subsets slightly increased in low-copy SOD1 G93A microglia with the BALB/c background (A-C) All single-cell microglia subsets (21,952 cells) of UMAP plots of WT(B6), WT(Balb), G93AL(B6), and G93AL(Balb) mice (4,742 cells from WT(B6), 7,765 cells from WT(Balb), 5,932 cells from G93AL(B6), and 3,513 cells from G93AL(Balb)).Each cluster was colored by microglia subtypes (A and C) or mouse strains (B).(D) Percentages of each microglia subtype in single-cell transcriptome plots from each strain were plotted.(E) Percentages of each homeostatic microglia cluster and each DAM cluster from each strain were plotted.

A B
C D E deregulation of many genes associated with microglial development and immune responses, 48 indicating that the immune system, such as intestinal immunity, strongly associates with microglial functions related to both brain development and neuroinflammation of neurodegenerative disease.Since the biased peripheral helper T cell responses between inbred C57BL/6 and BALB/c strains are traditionally known, these differences could also affect microglia development and their immune functions.Notably, Th1-or Th2-biased peripheral immune responses were observed in both WT and ALS model mice for each strain (Figures S9C-S9F).In addition, differences in the numbers of infiltrating immune cells and the ratios of peripheral immune cells were observed between G93AL(B6) and G93AL(Balb) mice at the disease end-stage (Figures 4B, 4C, S9A, and S9B).However, our previous research revealed that the depletion of CD8 + T, natural killer T (NK-T), and natural killer (NK) cells in ALS model mice with the C57BL/6 genetic background from the onset of age had no impact on survival times, 25 indicating that environmental factors derived from peripheral immune cells and/or infiltrating immune cells, with the exception of CD8 + T, NK-T, and NK cells, may affect the course of the disease by regulating microglial heterogeneity, survival, and DAM induction in ALS model mice.Intriguingly, some DAM marker genes were found to be oppositely regulated between G93AL(B6) and G93AL(Balb) microglia (Figures 7A and 7B).Further analyses will be required to reveal whether these differences could be linked to the disease mechanisms of ALS.In summary, our findings provide new evidence that genetic diversity may affect microglia heterogeneity, their responses, and disease progression in an ALS model.Identifying the therapeutic targets of ALS will require further study to elucidate the details of environmental factors regulating microglial heterogeneity and their neuroprotective functions.

Limitations of the study
This study suggests that genetic background variation affects microglial heterogeneity, their responses, and disease progression in ALS model mice.However, since this study conducted the experiments only between two strains, further analysis among ALS model mice with various genetic backgrounds will be required.In addition, although the comparative analyses of gene expression between in vivo and in vitro microglia suggest the contribution of immune system-derived environmental factors on microglial heterogeneity and functions, further validation using ALS model mice crossed with immunoregulatory gene knockout or transgenic mice will also be required.

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Koji Yamanaka (koji.yamanaka@riem.nagoya-u.ac.jp).

Materials availability
Plasmid and mouse lines generated in this study will be available from the lead contact upon request.

Data and code availability
RNA-seq and single-cell RNA-seq data have been deposited at NCBI GEO and NCBI SRA and are publicly available as of the date of publication.Accession numbers are listed in the key resources table.All data reported in this paper will be shared by the lead contact upon request.EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice
We used SOD1 G93A mice carrying a low copy number of mutant SOD1 G93A in the C57BL/6J background (G93AL(B6)), which were spontaneously lost transgene copies ($ À40%) during breeding of an original SOD1 G93A mouse (B6.Cg-Tg (SOD1*G93A)1Gur/J) obtained from Jackson Laboratory (Strain#:004435, RRID:IMSR_JAX:004435).BALB/c congenic SOD1 G93A mice (G93AL(Balb)) were generated by backcrossing G93AL(B6) with BALB/cA female mice over 10 times.Comparable transgene copy numbers in each mouse were confirmed by quantitative real-time PCR (qPCR).The genotyping method of mice was previously described elsewhere. 25qPCR was performed by using human SOD1-f primer (CAATGTGACTGCTGACAAAG), human SOD1-r primer (GTGCGGCCAATGATGCAAT), mouse b-actin-f primer (TTGGCCTCACTG TCCACCTT), and mouse b-actin-r primer (CGGACTCATCGTACTCCTGCTT).PCR was performed by using mouse SOD-A primer (GTTAC ATATAGGGGTTTACTTCATAATCTG), human/mouse SOD-C primer (CAGCAGTCACATTGCCCARGTCTCCAACATG), and human SOD-B primer (CCAAGATGCTTAACTCTTGTAATCAATGGC).Mice were maintained under the standard specific pathogen free environment (12 h light-dark cycle; 23 G 1 ºC; 50 G 5% humidity) and were treated in accordance with the guidelines established by the Institutional Animal Care and Use Committee of Nagoya University.The experiments using genetically modified animals and organisms were approved by the Animal Care and Use Committee and the recombinant DNA experiment committee of Nagoya University.We used the same numbers of male and female mice as possible for each experiment except for the scRNA-seq analysis.scRNA-seq analysis was performed with only female mice of each strain.

Primary microglia culture
Brains from male and female mice at postnatal days 1-2 were dissociated in 0.25% Trypsin and 10 mg/mL DNase I containing PBS at 37 C for 10 min.Dissociated cells were washed and plated on poly-L-Lysine coated flasks in 10% FBS DMEM (DMEM medium supplemented with 10% FBS, L-glutamine, 50 U/mL penicillin, and 50 mg /mL streptomycin) in the 5% CO 2 incubator.After 1-2 weeks, non-adherent microglial cells were collected by shaking flasks for 2 h at 150 rpm in the 5% CO 2 incubator.Microglia were seeded on culture dishes and stimulated with 1 mg /mL lipopolysaccharide (LPS) for 6 h or 50 ng/mL macrophage colony-stimulating factor (M-CSF) for 72 h.

Survival experiments of the mice
The times of disease onset and 10% weight loss were retrospectively determined as the times when mice reached a maximum body weight and 10% weight loss from the maximum body weight, respectively, and that of end stage was determined as the time when the mouse could not right itself within 20 seconds after being placed on its side.Statistical analyses of survival time were performed with a log-rank test and an unpaired t-test by using GraphPad Prism.

In situ hybridization of spinal cord sections
In situ hybridization was performed on 12 mm cryosections of lumbar spinal cords using a digoxigenin-labeled Csf1-specific cRNA probe.The cRNA probe was synthesized from pGEM-7Zf(-) vector inserted Csf1 cDNA by using DIG RNA Labeling Kit (SP6 / T7) according to the manufacturer's instruction.The sections were fixed with 4% PFA at RT for 15 min, washed with Diethyl pyrocarbonate (DEPC)-treated PBS, incubated with 1 mg/mL RNase-free Proteinase K Solution at 37 C for 15 min, and then post-fixed with 4% PFA at RT for 15 min.After washing with DEPC-treated PBS, the sections were incubated with hybridization buffer (50% deionized formamide / 2% Blocking Reagent / 0.1% N-Lauroylsarcosine / 0.1% SDS / 53 SSC (pH 4.5)) containing the cRNA probe at 70 C overnight in a slide mailer.Following hybridization with the cRNA probe, the sections were washed with 50% deionized formamide / 1% SDS / 23 SSC (pH 4.5) solution at 70 C for 30 min, three times, washed with 0.1 M TrisHCl (pH 7.5) / 150 mM NaCl / 0.05% Tween 20 solution at RT for 5 min, twice, and then incubated with 0.5% Blocking Reagent / 0.1 M TrisHCl (pH 7.5) / 150 mM NaCl / 0.05% Tween 20 solution for blocking at RT for 30 min.After blocking, the sections were incubated with an alkaline phosphatase-conjugated anti-digoxigenin antibody (1:1000) at RT for 2 h , washed with 0.1 M TrisHCl (pH 7.5) / 150 mM NaCl / 0.05% Tween 20 solution at RT for 10 min, twice, and then stained with Fast Red substrate according to the manufacturer's instruction.Immunofluorescence staining was conducted after in situ hybridization procedure.
Isolation of microglia from spinal cords by magnetic-activated cell sorting Spinal cord microglia were isolated by using magnetic-activated cell sorting (MACS) technique.Deeply anesthetized mice were transcardially perfused with PBS to remove blood from blood vessels.The spinal cords were then dissected and dissociated in Neural Tissue Dissociation Kit-Postnatal Neurons reagents by using gentleMACS Dissociator (Miltenyi Biotec, Germany) at 37 C for 15 min.For removal of myelin debris, the cells were incubated with Myelin Removal Beads II according to the manufacturer's instruction and purified by autoMACS Pro Separator (Miltenyi Biotec, Germany).The cells were then treated with anti-CD16/CD32 antibody (2 mg/mL) for blocking Fc receptors, followed by reacting with CD11b (Microglia) MicroBeads according to the manufacturer's instruction.CD11b-positive microglia were isolated by autoMACS Pro Separator (Miltenyi Biotec, Germany).

Quantitative PCR
Total RNAs from lumbar spinal cords, MACS-isolated microglia, or cultured microglia were purified using mirVana miRNA Isolation Kit or RNeasy Micro Kit according to the manufacturer's instructions.cDNAs were synthesized using PrimeScript RT reagent Kit (Perfect Real Time) and were subjected to quantitative PCR with the following protocol: 1 cycle at 95 C for 30 s, 40 cycles at 95 C for 5 s and 60 C for 30 s, with SYBR (TB Green) Premix Ex Taq II (Tli RNaseH Plus) by using the Thermal Cycler Dice Real Time System II (Takara Bio, Japan).The value of each gene was calculated in duplicate and averaged.The list of gene-specific primer pairs was shown in Table S3.Statistical analyses were performed using GraphPad Prism.

Flow cytometric analysis
Spinal cords, dissected from mice transcardially perfused with PBS for blood removal, were minced into 1mm 3 pieces in collagenase working solution (1 mg/mL Collagenase Type 4 / 0.4 mg/mL DNase I in HBSS) and incubated at 37 C for 15 min.For removal of myelin debris, the cells were resuspended in 37% Percoll in PBS, and then centrifuged at 780 3 g for 20 min.After the removal of myelin debris in the upper layer, a cell pellet containing microglia and immune cells was collected.Splenocytes were collected by mashing the spleen with slide glasses, followed by lysing red blood cells with an ammonium chloride solution.For the flow cytometry analysis, Fc receptors were blocked with 2 mg/mL anti-CD16/CD32 antibody on ice for 10 min, and then the cells were stained with combinations of the following antibodies, anti-CD45-APC-Cy7, anti-CD4-PE-Cy7, anti-CD3e-PerCP-Cy5.5, anti-CD8a-FITC, anti-NK1.1-PE,or anti-CD49b-PE at RT for 20 min.Data were obtained by using FACS Verse flow cytometer (BD Biosciences, USA) and further analyzed using Flowjo Software.Representative gating strategies for microglia and immune cells were shown in Figures S3A and S3B.Statistical analyses were performed using GraphPad Prism.

Intracellular cytokine staining
Peripheral bloods were collected from the central tail artery, and then peripheral blood mononuclear cells (PBMCs) were isolated by centrifugation (400 3 g for 30 min) with Ficoll-Paque PREMIUM 1.084.Splenocytes were obtained by the method mentioned above.Immune cells were stimulated with 10 ng/mL Phorbol 12-myristate 13-acetate (PMA) and 1mg/mL Ionomycin in the presence of BD GolgiPlug protein transport inhibitor (1:1000) in 10% FBS containing RPMI1640 medium for 4hrs at 37 C / 5% CO 2 , and then cell surface staining was performed with anti-CD45-APC-Cy7, anti-CD4-PE-Cy7, anti-CD8a-FITC, and anti-CD3e-PerCP-Cy5.5 by the method mentioned above.After cell surface staining, the cells were fixed and permeabilized with BD Cytofix/Cytoperm solution, and then intracellular cytokine staining was performed with anti-IFN-g-alexa fluor 488, anti-IL-17A-alexa fluor 647, and anti-IL-4-PE according to the manufacturer's instruction.Flow Cytometry analyses were performed by using FACS Verse flow cytometer (BD Biosciences, USA).The ratio of cytokine producing cells in the CD4-positive T-lymphocytes were analyzed by using Flowjo Software.Statistical analyses were performed using GraphPad Prism.

Proliferation analysis of microglia
Primary cultured microglia were seeded on culture dishes.On the next day, microglia were stained with CytoTell Red 650 fluorescent dye at 37 C for 30 min in the 5% CO 2 incubator, and then washed with PBS.Microglia labeled with CytoTell Red 650 were stimulated with 50 ng/mL macrophage colony-stimulating factor (M-CSF) in 10 % FBS DMEM medium for 72 h.The fluorescence changes in divided microglia were then monitored using FACS Verse flow cytometer (BD Biosciences, USA).The ratio of divided microglia was analyzed using Flowjo Software.Statistical analyses were performed using GraphPad Prism.

RNA sequencing
Total RNAs from MACS-isolated microglia and primary cultured microglia were purified using RNeasy Micro Kit (QIAGEN, Germany) and RNeasy Mini Kit (QIAGEN, Germany) according to the manufacturer's instructions, respectively.The RNA integrity was analyzed by 2100 Bioanalyzer (Agilent Technologies, USA) with RNA 6000 pico Kit.Libraries were prepared with TruSeq RNA Library Prep Kit v2 (Illumina, USA) or MGIEasy RNA Directional Library Prep Kit V2.0 (MGI Tech Co, China) and sequenced with 151-nt paired-end reads using HiSeq X (Illumina, USA) or DNBSEQ-G400 (MGI Tech Co, China).

Figure 1 .Figure 2 .
Figure1.Genetic background variation affects microglial heterogeneity in ALS model mice (A-C) All single-cell transcriptome data (31,216 cells) of UMAP plots for the spinal cords of ALS model mice at the disease end-stage and age-matched litter mate wild-type mice (6,473 cells from WT(B6), 9,762 cells from WT(Balb), 8,391 cells from G93AL(B6), and 6,590 cells from G93AL(Balb)).Each cluster was colored by cell type (A and C) or mouse strain (B).(D) Percentages of each cell type in the single-cell transcriptome plots from each strain were plotted.(E) Percentages of each homeostatic microglia cluster and DAM cluster from each strain were plotted.

Figure 3 .Figure 4 .
Figure3.Continued (E) The mRNA expression of Csf1 (M-CSF) and Csf1r (M-CSFR) in LPS-stimulated primary cultured microglia derived from WT C57BL/6 (B6) and BALB/c mice was quantified by quantitative RT-PCR.No differences were observed in Csf1 and Csf1r expression (n = 4 each).Data were represented as mean G SD. One-way ANOVA followed by Tukey-Kramer multiple comparison post hoc tests.n.s., not significant.(F) The proliferation of M-CSF-stimulated primary cultured microglia derived from WT C57BL/6 (B6) and BALB/c mice was quantified with Cytotell dye staining (n = 3 each).Distributed fluorescence dye intensities and the proportions of proliferating cells were quantified by flow cytometry.Data were represented as mean G SD. One-way ANOVA followed by Tukey-Kramer multiple comparison post hoc tests.n.s., not significant.