Therapeutic potential of human microglial transplantation in a chimeric model of CSF1R-related leukoencephalopathy

Summary

Microglia replacement strategies are increasingly being considered for the treatment of primary microgliopathies like Adult-onset leukoencephalopathy with axonal spheroids and pigmented glia (ALSP). However, available mouse models fail to recapitulate the diverse neuropathologies and reduced microglia numbers observed in patients. In this study, we generated a xenotolerant mouse model lacking the fms-intronic regulatory element (FIRE) enhancer within Csf1r that develops nearly all the hallmark pathologies associated with ALSP. Remarkably, transplantation of human iPSC-derived microglia (iMG) progenitors restores a homeostatic microglial signature and prevents the development of axonal spheroids, white matter abnormalities, reactive astrocytosis, and brain calcifications. Furthermore, transplantation of CRISPR-corrected ALSP-patient-derived iMG reverses pre-existing spheroids, astrogliosis, and calcification pathologies. Together with the accompanying study by Munro and colleagues, our results demonstrate the utility of FIRE mice to model ALSP and provide compelling evidence that iMG-transplantation could offer a promising new therapeutic strategy for ALSP and perhaps other microglia-associated neurological disorders.

eTOC

Chadarevian et al. demonstrate that transplantation of iPSC-derived microglia can prevent and reverse diverse neuropathologies in a mouse model of Adult-onset leukoencephalopathy with axonal spheroids and pigmented glia (ALSP). Collectively, these findings suggest that iPSC-microglial transplantation could offer a promising new therapeutic strategy for ALSP and other primary microgliopathies.

Graphical Abstract

Introduction

CSF1R-related leukoencephalopathy, also known as adult-onset leukoencephalopathy with axonal spheroids and pigmented glia (ALSP), is a rare, autosomal dominant leukodystrophy caused by mutations in colony stimulating factor 1 receptor (CSF1R)^1,2. CSF1R signaling is necessary for microglial differentiation and survival, and ALSP-associated mutations have been shown to impair CSF1R signaling. As a result, ALSP patient brains have fewer microglia, and those microglia that remain exhibit a chronically activated proinflammatory phenotype^3,4. ALSP patients typically present during their third or fourth decade with various cognitive and motor symptoms, including executive function impairments, personality disruptions, memory decline, sensory deficits, bradykinesia, rigidity, and tremors^{5 6}. The overlap in clinical symptoms with more common neurological disorders including multiple sclerosis, atypical parkinsonism, and frontotemporal dementia often leads to misdiagnosis. However, increased genetic testing and awareness is beginning to improve this problem while also revealing an increased prevalence of ALSP¹.

The great majority of ALSP-associated mutations involve amino acid substitutions within the intracellular domain of CSF1R that ablate tyrosine kinase activity. Stables et al. recently produced a knock-in mouse model that harbors the mouse homologue of the ALSP-associated CSF1R-E633K mutation. Analysis of these mice confirmed a dominant-negative impact of the heterozygous mutation on responsiveness to CSF1⁷. The dominant inhibitory effect of kinase-dead ALSP mutations was further confirmed in zebrafish models⁸. In a heterozygous state, Csf1r-E631K mice exhibited reduced numbers of microglia and diminished dendritic arborization, consistent with ALSP patients⁷. However, there was no evidence of neuropathology, suggesting that mice are more resilient to impaired CSF1R signaling within the brain, perhaps related to differing levels of CSF1 and CSF1R expression between mice and humans⁹. In contrast, null mutations in the Csf1r gene in mice and rats lead to a complete absence of microglia and many other tissue macrophage populations but also cause early mortality^9,10. A similar exacerbated condition has recently been described in human patients that harbor homozygous or compound heterozygous loss of function mutations in CSF1R. This disease, termed BANDDOS (Brain abnormalities, neurodegeneration, and dysosteosclerosis), exhibits a much earlier pediatric or perinatal age of onset and as the name implies neurological symptoms are often accompanied by skeletal abnormalities^11,12. In addition to reduced numbers of microglia, ALSP and BANDDOS patients develop an array of neuropathologies, including axonal spheroids, reactive astrocytosis, lipid accumulation, white matter atrophy, and myelin disruption^3,5,8,13,14. Between 40–83% of CSF1R-related leukoencephalopathy patients also exhibit brain calcifications and varying degrees of blood-brain barrier (BBB) dysfunction, ranging from more subtle alterations in endothelial tight junctions, to cerebral amyloid angiopathy (CAA), to microinfarcts^{2,11–13,15,16}. Interestingly, a recent analysis of two severe homozygous mutant CSF1R pediatric cases revealed a complete absence of microglia, agenesis of the corpus callosum, and more extensive calcification, accompanied by extreme intellectual disability and perinatal mortality^17,18.

There are currently no FDA-approved treatments for ALSP. However, hematopoietic stem cell transplantation (HSCT) has been explored with variable results and at least one documented HSCT case resulted in death^19–22. Potentially prophylactic effects of anti-inflammatory steroids are also being examined²³. An alternative therapeutic approach that likely warrants further examination is to investigate whether replacing unhealthy mutant microglia with healthy stem cell-derived microglia could offer a safer, more effective therapy.

The mammalian CSF1R locus contains a highly conserved super-enhancer fms-intronic regulatory element (FIRE). Deletion of this element produced a novel mouse model of microglial deficiency. In the brain, the complete absence of microglia is not associated with any detectable impacts on early postnatal gene expression other than the loss of a microglial gene signature²⁴. Unlike CSF1R knockout rodents, FIRE-knockout mice (FIRE mice) are not osteoporotic, somatic growth is unaffected, and macrophage populations in the liver, spleen, gut and lung still express CSF1R. However, some other yolk-sac derived macrophage populations (ie. Langerhans cells) are also absent. McNamara et al. recently demonstrated that the absence of microglia in FIRE mice does not affect white matter development. However, with age, mutant mice develop histological features of white matter disruption observed in ALSP, including disruptions in myelin integrity and lipid metabolism²⁵. In our own recent studies, we have also observed additional intriguing evidence of various ALSP-associated pathologies in aging FIRE mice. Thus, we speculated that these mice could be further developed to provide a promising preclinical model of ALSP and to explore the potential therapeutic application of human microglial transplantation.

In the current study, FIRE mice were crossed onto a hCSF1/Rag2^−/−/il2rg^−/y background to enable human microglial engraftment, and the resulting ‘hFIRE’ mice were examined at 2 and 8.5 months of age. As hFIRE mice age, they develop several key pathological hallmarks observed in ALSP patients including axonal spheroids, reactive astrocytosis, lipid accumulation, BBB dysfunction, brain calcifications, and myelin disruption. Importantly, in the accompanying study Munro and colleagues independently replicate many of these findings within immune-intact FIRE mice and further reveal significant age-related changes in glial gene expression. To determine whether healthy human microglia can prevent these ALSP-associated pathologies, we transplanted 2-month-old hFIRE mice with human microglial progenitors derived from a healthy control patient induced pluripotent stem cell (iPSC) line. 6.5 months later, brains were examined revealing a near complete prevention of ALSP-related neuropathologies. To further determine whether this approach could be adapted toward an autologous cell therapy, iPSCs were generated from an ALSP patient carrying CSF1R-L786S heterozygous mutation and genetically corrected via CRISPR gene editing. In vitro analysis of these isogenic mutant and wildtype-corrected iPSC-microglia revealed impaired proliferation consistent with the dominant-negative impact of disease-associated mutations on CSF1R signaling. Furthermore, when CRISPR-corrected microglia were transplanted into 4-month-old hFIRE mice, we observed a significant reduction of pre-existing spheroid, astrogliosis, and calcification pathologies. In contrast, uncorrected ALSP microglia exhibited greatly diminished engraftment and failed to reduce pathology. Together, these experiments provide preclinical evidence to support the further therapeutic development of iPSC-microglia transplantation for a human disease.

Results

Microglia maintain their numbers throughout life via self-renewal/proliferation, a process that is highly dependent on CSF1R signaling^26,27. In ALSP, dominant mutations in CSF1R impair this process, leading to age-related reductions in microglial numbers⁸. To model this pathognomonic feature, we took advantage of ‘FIRE’ mice, a recently developed model that harbors homozygous deletions in the fms-intronic regulatory element (FIRE) of Csf1r. Unlike heterozygous Csf1r^+/− knockout mice that exhibit increased numbers of microglia, FIRE mice lack all microglia^24,28–30. While ALSP patients typically exhibit a partial reduction in microglial numbers, homozygous and compound heterozygous mutations in CSF1R have recently been shown to cause a more severe earlier onset disease, with at least two pediatric cases exhibiting a complete absence of microglia^17,18. We, therefore, speculated that FIRE mice, in contrast to heterozygous Csf1r-E631K mutant mice, may develop more of the pathological features observed in ALSP patients.

hFIRE mice recapitulate many of the pathological features of ALSP

Xenotolerant mice were generated by backcrossing FIRE mice onto a hCSF1/Rag2^−/−/il2rg^−/y background (Figure 1A). The resulting ‘hFIRE’ mice lack B-, T-, and NK-cells thereby preventing xenotransplant rejection and harbor humanized CSF1 which is necessary for survival of human microglia within mice^26,27. WT (hCSF1 littermates) and hFIRE mice were compared at 2- and 8.5-months of age for endogenous microglia numbers, levels of glial fibrillary acid protein (GFAP) immunoreactivity, and the presence of calcification using a fluorescently modified derivative of bisphosphonate that binds hydroxyapatite calcium crystals (Risedronate-647, RIS)³¹. Consistent with prior reports in non-humanized immune-intact FIRE mice²⁴, we found hFIRE mice developed with complete loss of PU.1+/IBA1+ microglia (Figure 1B-F) without the skeletal abnormalities or viability concerns seen in constitutive CSF1R knockout mice³². At 2-months of age, hFIRE mice exhibited normal gross brain anatomy and little evidence of neuropathology. However, by 8.5 months, hFIRE mice develop widespread astrogliosis (Figure 1G-I) and robust calcification (Figure 1J-L). Immunostaining with SMI312, a marker of hyperphosphorylated axonal neurofilaments, also identified progressive development of neurofilament-positive axonal spheroids from 2- to 8.5-months-old hFIRE mice (Figure 1M,N). Taken together, these findings recapitulate not only the diverse neuropathologies observed in patients with rare homozygous CSF1R mutations, but also the canonical ALSP pathologies observed in patients carrying dominant heterozygous CSF1R mutations.

Figure 1: hFIRE mice progressively develop ALSP-related neuropathologies.

(A) hFIRE mice were generated by crossing immune-intact FIRE mice onto the immune-deficient hCSF1^+/+/Rag2^−/−/il2rg^−/y knock-in background and restoring homozygosity for all four alleles. (B-C) Representative confocal imaging of 2-month-old female hCSF1 (B) and hFIRE (C) coronal brain sections for murine microglia (IBA1, green; PU.1, orange); scale bar, 50 μm. (D-F) Quantification of PU.1+/IBA1+ murine microglia within the cortex (D), hippocampus (E), and thalamus (F) of hCSF1 wildtype female littermates and hFIRE mice. Data represented as average number IBA1+/PU1+ microglia per field of view (FOV); P values from unpaired, two-sided t-test. (G-H), Representative immunofluorescence staining of 2-month-old (G) and 8.5-month-old (H) hCSF1 and hFIRE brain sections for astrocytes (GFAP, yellow); scale bar, 50 μm. (I) Quantification of astrogliosis in the thalamus of 2-month-old and 8.5-month-old hCSF1 and hFIRE mice; P values from one-way ANOVA (F_3,25=17.55; ****P<0.0001) with Tukey multiple comparisons tests. (J-K) Representative immunofluorescence staining of calcification (Risedronate-647) in 2-month-old (J) and 8.5-month-old (K) hCSF1 and hFIRE brain sections (RIS-647, gray); scale bar, 50 μm. (L) Quantification of RIS+ calcification in the thalamus of 2-month-old and 8.5-month-old hCSF1 and hFIRE mice; P values from one-way ANOVA (F_3,25=71.20; ****P<0.0001) with Tukey multiple comparisons tests. (M) Representative confocal imaging of 8.5-month-old hFIRE brain sections stained for neurofilament-positive axonal spheroids (SMI312, green); scale bar, 20 μm. (N) Quantification of SMI312 positive spheroids per hemibrain of 2-month-old and 8.5-month-old hCSF1 and hFIRE mice; P values from one-way ANOVA (F_3,25=52.85; ****P<0.0001) with Tukey multiple comparisons tests. For all quantification data is represented as average mean value ±SEM (2-month-old hCSF1, n = 6; 2-month-old hFIRE, n=7; 8.5-month-old hCSF1, n=8; 8.5-month-old hFIRE, n=8 biological replicates). *P<0.05; ****P<0.0001. Comparisons not shown are not significant.

Human microglia fully repopulate the brains of hFIRE mice

To determine whether transplantation of CSF1R-wildtype human microglia can prophylactically prevent the development of ALSP-like pathologies, 2-month-old female hFIRE mice were transplanted with human iPSC-derived microglia progenitors (hematopoietic progenitor cells; HPCs) or saline control (PBS) (Figure 2A). Within 7 days, human microglia proliferated and migrated out from the injection site into the surrounding parenchyma, exhibiting a mitotic Ki67+ ‘wavefront’ of proliferative migratory cells (Figure S1A). By 2–4 weeks post-transplantation, human microglia, co-labeled with IBA1 and the human-specific nuclear marker Ku80, occupied a majority of the brain but continued to proliferate at sites directly adjacent to the few areas that remained devoid of microglia (Figure 2B-D; Figure S1B). After 6.5 months, mouse brains were fully engrafted with IBA1+ human microglia (Figure 2E; Figure S1C,D).

Figure 2: Human microglial progenitors efficiently engraft within adult hFIRE brains, restoring canonical microglial gene expression.

(A) Schematic representing adult female hCSF1 and hFIRE mice transplantation paradigm. (B-D) Representative hemibrain confocal stitches of hFIRE mouse showing proliferative expansion and migration of human microglia (IBA1, green; Ki67, red/white) 7 days (B), 14 days (C), and 30 days (D) after stereotactic hippocampal transplantation of HPC microglial progenitors, scale bar, 500μm. (E) Representative 20x confocal imaging of 8.5-month-old hFIRE engrafted with human microglia (IBA1, green); scale bar, 50 μm. (F) Heat map comparing bioinformatically classified mouse or human transcripts from hemibrain samples of PBS-injected littermates (hCSF1, blue; n=6), hFIRE-PBS (light turquoise; n=5), and hFIRE-HPC aligned to human (hFIRE-HPC Human genes, green; n=5) and mouse transcriptomes (hFIRE-HPC Mouse genes, brown; n=5). (G) Examples of fully recovered canonical microglial genes and non-recovered transcripts (also see Figure S2). (H) Representative confocal imaging of hFIRE-PBS and hFIRE-HPC for engrafted microglia (P2RY12, magenta; IBA1, green), scale bar, 50 μm. (I) Representative high power confocal imaging demonstrating highly ramified homeostatic (P2ry12, magenta) human microglia (IBA1, green) in hFIRE-HPC mice 6.5 months after transplantation; scale bar, 20 μm.

Previous studies have shown reduced expression of several canonical microglial genes in ALSP-patient brains including CSF1R, CX3CR1, P2RY12, AIF1 and TREM2^3,8. To test whether transplantation of wildtype human microglial progenitors can restore these transcripts, bulk RNA sequencing was performed on whole brain lysates isolated from PBS-transplanted hCSF1 littermates, PBS-transplanted hFIRE (hFIRE-PBS), and HPC-transplanted hFIRE mice (hFIRE-HPC) (Figure 2F,G; Supplementary Table S1). As expected, hFIRE-PBS mice exhibited similar microglia-related transcriptional deficiencies to those previously reported in ALSP-patients (Figure 2F). By aligning the hFIRE-HPC data to human (Figure 2F; green) versus mouse (Figure 2F; brown) transcriptomes, we find that many of the murine microglia transcripts that are lost in hFIRE-PBS mice are now expressed by engrafted human microglia in hFIRE-HPC mice (Figure 2F; green), including expression of the homeostatic microglial genes CSF1R, P2RY12, CX3CR1, AIF1, GPR34, and OLFML3 (Figure 2G; Figure S2A). Immunohistochemical (IHC) analysis further confirmed that transplanted hFIRE-HPC mice express the homeostatic microglial marker P2ry12 (Figure S2B-D) and exhibit highly ramified microglial morphology within the cortex, hippocampus, and thalamus (Figure 2H,I). However, we also observed several transcripts that were not highly expressed by human microglia in hFIRE-HPC mice including Fcrls, Ccr6, Il7r and Ly9. Importantly, these genes have previously been shown to be undetectable or expressed at 20–40-fold lower levels in human microglia than mouse microglia^33,34. In contrast, many of these transcripts are known to be enriched in border associated macrophages that are largely preserved in FIRE mice³⁵. Thus, these murine transcripts remain expressed, albeit at lower levels, in mouse-aligned hFIRE-PBS reads (Figure 2F; light turquoise).

Transplantation of microglial progenitors prevents the formation of axonal spheroids

Loss of ramified microglia has been shown to precede the development of axonal spheroids in ALSP patient brains^4,36. hFIRE mice lack microglia and similarly develop increasing numbers of phosphorylated neurofilament accumulating axonal spheroids (SMI312+) with age (Figure 1M). Co-staining of hFIRE-PBS brain sections with SMI312 and MAP2, a neuronal dendritic marker, or DegenoTag, an antibody designed against degenerating axons³⁷, demonstrate that SMI312+ spheroids arise form axonal projections, but not dendrites (Figure 3A,B). In addition, imaging of DAPI+ nuclei confirm SMI312+ spheroids do not represent neuronal cell bodies (Figure S3A). Axonal spheroids have also been shown to accumulate a variety of cellular components including cytoskeletal proteins (e.g., neurofilaments and kinesins), organelles (e.g., lysosomes and mitochondria), and pathological proteins including amyloid precursor protein (APP), phosphorylated Tau (Tau231), and ubiquitin^38–40. Immunostaining further revealed axonal spheroids that accumulate LAMP1 immunoreactivity occur throughout the 8.5-month-old hFIRE-PBS brain (Figure 3C). To quantify spheroid numbers, confocal imaging of LAMP1 and SMI312 immunostained brain sections (Figure S3B-D) was performed, and the number of single and double positive axonal spheroids was calculated by a blinded observer using IMARIS 3D imaging software. We found that 96% of axonal spheroids in the hippocampus and cortex were LAMP1+ of which 27% were LAMP1/SMI312 double-positive (Figure 3D). A recent study has further shown that SMI312+ spheroids can be categorized by diameter as being associated with milder (1.5 to 4 μm) or more severe (>4 μm) axonal damage⁴¹. A similar categorization in hFIRE-PBS mice revealed 95% of SMI312+ spheroids within the hippocampus and 85% in the fornix exhibited diameters >4μm, indicative of severe axonal damage. Immunostaining further demonstrated that a subset of axonal spheroids in hFIRE-PBS mice can accumulate APP and phosphorylated Tau (Figure 3E,F). Additionally, axonal spheroids have been reported to occur in either myelinated or unmyelinated axons in response to injury, oxidative stress, Ca²⁺ imbalance, and inflammation^38,42. Therefore, we performed co-staining with myelin basic protein (MBP) and confirmed the presence of spheroids in both myelinated and unmyelinated axons (Figure 3G).

Figure 3: Microglia progenitor transplantation prevents axonal spheroid formation in hFIRE mice.

(A-B) Representative immunostaining of MAP2- (blue), Degenotag+ (magenta), and SMI312+ (green) spheroids in 8.5-month-old hFIRE-PBS mice; scale bar, 20 μm. (C) Representative confocal microscopy of SMI312+ axonal spheroids exhibiting LAMP1 (red) immunoreactivity in 8.5-month-old hFIRE-PBS mice, scale bar 20 μm. (D) Quantitative Venn diagram of LAMP+, SMI312+, and LAMP1+/SMI312+ spheroids in the hippocampus and fornix of hFIRE-PBS mice (n=8). Representative images of littermate, hFIRE-PBS, and hFIRE-HPC are provided in Figure S3B-D. (E) Representative confocal imaging of axonal spheroids in hFIRE-PBS brain sections exhibiting accumulation of LAMP1 immunoreactivity, amyloid precursor protein (E) (APP; blue), or phosphorylated-Tau (F) (Tau231; yellow). Examples of both myelinated and unmyelinated axonal spheroids are observed (G) (MBP, cyan); scale bar, 20 μm. (H) Confocal imaging of wildtype littermate, hFIRE-PBS, and hFIRE-HPC brain sections immunostained for LAMP1+ and APP+ axonal spheroids; scale bar 100 μm. (I,J) Quantitative Venn diagrams of LAMP1+, APP+, and LAMP1+/APP+ double-labeled axonal spheroids in the hippocampus (I) and fornix (J) of 8.5-month-old hFIRE-PBS mice (n=8). (K,L) Quantification of LAMP1+ (K) and APP+ (L) spheroid numbers in the hippocampus of littermate, hFIRE-PBS, and hFIRE-HPC mice; P values from one-way ANOVA (LAMP1+: F2,23=76.17; ****P<0.0001) (APP+: F2,23=78.51; ****P<0.0001) with Tukey multiple comparisons tests. (M,N) Quantification of number of LAMP1+ (M) and APP+ (N) spheroids in the fimbria fornix of littermate, hFIRE-PBS, and hFIRE-HPC mice; P values from one-way ANOVA (LAMP1+: F2,23=67.60; ****P<0.0001) (APP+: F2,23=85.76; ****P<0.0001) with Tukey multiple comparisons tests. Data represented as average mean value ±SEM (Littermates, n=9; hFIRE-PBS, n=8; hFIRE-HPC, n=9). ns=not significant; **P<0.01; ***P<0.001; ****P<0.0001.

To further assess the heterogeneity of axonal spheroids, and the therapeutic potential for microglial transplantation to prevent axonal spheroid formation, low-power confocal imaging of LAMP1 and APP immunostained hemibrain sections was performed. Analysis of wildtype-hCSF1 brains revealed very few enlarged LAMP1+ or APP+ aggregates (Figure 3H). In contrast, both single and double positive spheroids were observed throughout the brain of hFIRE-PBS mice (Figure 3I,J). An average of 54±7.9 (SEM) LAMP1+ and 34±3.4 (SEM) APP+ spheroids per field of view (FOV) were detected adjacent to CA2/CA3 of the hippocampus in hFIRE-PBS mice (Figure 3K,L), and an average of 101±12.6 (SEM) LAMP1+ and 27±2.7 (SEM) APP+ spheroids per FOV were observed within the fimbria fornix white matter tract of PBS-injected hFIRE mice (Figure 3M,N). Remarkably, engraftment of human microglia led to a complete reversal of spheroid pathology in both the hippocampus and fornix of hFIRE-HPC mice, significantly reducing spheroid numbers to the levels of age-matched wildtype-hCSF1 littermates (Figure 3K-N).

Microglia engraftment prevents the development of additional ALSP-related pathologies

Brain calcifications have been reported in up to 83% of ALSP patients and are further increased in homozygous CSF1R mutation carriers^12,17. Staining with the fluorescent bisphosphonate analog Risedronate-647 (RIS) revealed significant calcifications within the thalamus of hFIRE-PBS mice (Figure 4A). In striking contrast, transplantation of microglia progenitors fully prevented the accumulation of RIS+ calcium aggregates in hFIRE mice (Figure 4A,E; Figure S4A). ALSP patients also consistently exhibit signs of reactive astrocytosis as evidenced by increased glial fibrillary acidic protein (GFAP) mRNA and protein and elevation of other proinflammatory astrocyte-associated transcripts such as complement C4b^3,8. As expected, based on our initial assessment of 8.5-month-old hFIRE mice (Figure 1), we detected a significant increase in GFAP+ immunoreactive astrocytes within the thalamus of hFIRE-PBS mice (Figure 4B) that was prevented in hFIRE-HPC mice (Figure 4F; Figure S4B). Prior studies have shown that the pro-inflammatory cytokine osteopontin (OPN) is necessary for astrocyte activation following traumatic brain injury⁴³ and has been observed adjacent to calcium aggregates in mouse models of primary familial brain calcification. Staining for OPN revealed significant accumulation within the thalamus of hFIRE-PBS mice that was also prevented by microglia transplantation (Figure 4C,G; Figure S4C). Interestingly, we observed a positive correlation between OPN and GFAP immunoreactivity within the thalamus (R²=0.5725; *p<0.0183). However, the correlation between OPN and RIS+ calcification was even greater (R²=0.9578; ****p<0.0001) (Figure 4D,H,I), confirming the previously described tight association between OPN protein and calcium deposition in mouse models of primary familial brain calcifications⁴⁴. Importantly, these findings were further validated in male hFIRE-HPC mice following transplantation of microglial progenitors derived from three additional independent control iPSC lines, demonstrating the highly consistent capacity of healthy human microglia to reduce these ALSP-associated pathologies (Figure S5).

Figure 4: Microglia engraftment prevents ALSP-related neuropathologies in hFIRE mice.

(A-C) Representative immunostaining for Risedronate+ calcifications (A) (RIS, gray), astrogliosis (B) (GFAP, yellow), and osteopontin (C) accumulation (OPN, red) in hFIRE-PBS and hFIRE-HPC mice, scale bar 50 μm. (D) Confocal imaging of calcification (white), osteopontin (red), and astrogliosis (yellow) pathologies within the thalamus of hFIRE-PBS mice, scale bar 50 μm. (E-G) Quantification of RIS+ calcification (E), astrogliosis (F), and OPN accumulation (G) in the thalamus of 8.5-month-old littermate, hFIRE-PBS, and hFIRE-HPC mice. Representative images of littermate controls provided in Figure S4; P values from one-way ANOVA (calcification: F2,24=7.805; **P=0.0024) (astrogliosis: F2,24=13.91; ****P<0.0001) (OPN accumulation: F2,24=18.65; ****P<0.0001) with Tukey multiple comparisons tests. (H) Simple linear regression plotted between OPN accumulation and astrogliosis levels (R²=0.57256, *P=0.0183) in thalamus of hFIRE-PBS mice. (I) Simple linear regression between OPN accumulation and RIS+ calcification (R²=0.9578, ****P<0.0001) in the thalamus of hFIRE-PBS mice. (J-K) ELISA quantification of GFAP levels in soluble brain samples (J) and plasma (K); P values from one-way ANOVA (brain: F2,24=9.053; **P=0.0012) (plasma: F2,24=11.98; ***P=0.0002) with Tukey multiple comparisons tests. (L) Simple linear regression plotted between levels of GFAP in soluble brain extracts and plasma (R²=0.1884, *P=0.0183) of hCSF1-PBS littermates (blue), hFIRE-PBS (light turquoise), hFIRE-GFP (green) mice. (M) Biochemical quantification of MCP-1 levels in soluble brain extracts; P values from one-way ANOVA (F2,24=11.67; ***P=0.0003) with Tukey multiple comparisons tests. (N-O) Simple linear regression between levels of MCP-1 in soluble brain extracts and GFAP levels in the brain (N) (R²=0.1656, *P=0.0352) and plasma (O) (R²=0.6257, ****P<0.0001) of hCSF1-PBS littermates (blue), hFIRE-PBS (light turquoise), hFIRE-GFP mice (green). (P) ELISA quantification of plasma GFAP levels in hCSF1 and hFIRE mice at 1–2 months of age (n=10–11), 4–5 months (n=20), and 7–8 months (n=9); P values from one-way ANOVA (F5,73=35.96; ****P<0.0001) with Tukey multiple comparisons tests. Comparisons not shown are not significant. Data represented as average mean value ±SEM (Littermates, n = 9; hFIRE-PBS, n=9; hFIRE-HPC, n=9). ns=not significant; **P<0.01, ***P<0.001, and ****P<0.0001.

The impact of FIRE deletion and microglial transplantation on GFAP was further analyzed using an enzyme-linked immunosorbent assay (ELISA). Consistent with IHC analysis, we found significantly elevated levels of GFAP within the brain of hFIRE-PBS mice that were in turn prevented by microglia transplantation (Figure 4J). GFAP levels in the periphery have also been shown to correlate with neurodegeneration and increasingly considered as a potential biomarker for several neurodegenerative disorders^45–47. Analysis of blood plasma revealed elevated levels of GFAP within the periphery could also be readily detected and significantly reduced in microglial engrafted hFIRE-HPC mice (Figure 4J). Brain and plasma levels of GFAP also correlated across all three groups (R²=0.1884; *p=0.0237) (Figure 4K,L). These findings replicate similar transcriptomic and proteomic observations of elevated levels of GFAP in ALSP patient brains, a finding that was accompanied by increased levels of monocyte chemoattractant protein-1 (MCP-1/CCL2), a proinflammatory chemoattractant implicated in neurodegenerative disease^8,48. We therefore examined CCL2 levels by ELISA and found a significant increase within the brains of hFIRE-PBS mice and a corresponding reduction of CCL2 in hFIRE-HPC mice (Figure 4M). GFAP and CCL2 levels were also correlated within the brain (R²=0.1656, *p=0.0352), but this association was surprisingly far more robust when comparing brain CCL2 levels with plasma GFAP (R²=0.6257, ****p<0.0001) (Figure 4N,O). CCL2 is predominantly expressed by microglia within the brain^33,34. However, similar to OPN, prior studies have shown that activated astrocytes can also produce CCL2⁴⁹.

To determine whether hFIRE mice chronically exhibit elevated levels of plasma GFAP, we collected plasma from hCSF1 littermates and hFIRE mice across 1–8 months of age. Biochemical analysis reveals no significant differences between these groups at 1–2 months of age (P>0.9999) or 4–5 months of age (P=0.9967). However, 7–8-month-old hFIRE mice exhibited significantly elevated levels of plasma GFAP in comparison to both PBS-injected littermates and younger hFIRE mice (Figure 4P). Taken together, these results indicate blood plasma GFAP levels correlate with neuropathology in hFIRE mice, but significant changes are only detectable at later ages. While plasma GFAP does not provide a disease-specific biomarker, when paired with a genetic diagnosis of ALSP, this measure may serve as a potential peripheral biomarker for the onset, progression or treatment of ALSP.

To determine whether hFIRE mice exhibit altered levels of other peripheral immune-related markers that could indirectly influence brain health, analysis of blood sera was performed. Multiplex ELISA analysis of 19 murine cytokines and chemokines revealed no significant differences in IL-27, IL-2, IL-10, IL-6, TNF-α, KC/GRO, IL-12, IL-1B, IL-15, IL-33, IL-9, IL-17, MIP-2, IP-10, MIP-1α, or MCP-1 levels between 8.5-month-old wildtype littermates, hFIRE PBS, and hFIRE-HPC transplanted mice, although levels of many of these cytokines correlated significantly with each other (Figure S6A-P,R). Interestingly, a small but significant increase in IL-5 levels in hFIRE-PBS sera that was prevented by microglial transplantation was detected (Figure S6Q). IL-5 is involved in the maturation of B- and T-cells, which are absent in our Rag2- and Il2rg-deficient hFIRE mice⁵⁰. However, IL-5 has also been shown to regulate eosinophil maturation^51,52. Although the potential interactions between microglia and eosinophils remains largely unknown, a recent study detected increased eosinophil numbers and IL-5 levels following sustained pharmacological depletion of murine microglia⁵³. Importantly, cessation of treatment and subsequent repopulation of murine microglia returned these levels to baseline. Taken together, and with previous characterizations of FIRE mice²⁴ and the accompanying findings from Munro and colleagues, these results indicate that genetic deletion of the Csf1r FIRE enhancer has minimal impact on peripheral macrophage populations.

Microglia engraftment prevents thalamic microhemorrhages and neuronal and synaptic loss

Amyloid precursor protein (APP) plays a vital role in regulating calcium homeostasis within the brain^54,55. Having observed significant calcifications in hFIRE-PBS mice, we next examined the expression of APP. We detected strong APP immunoreactivity surrounding RIS+ calcifications within the thalamus (Figure 5A). Microglia transplantation not only prevented calcification but also the observed APP accumulation (Figure 5B). Previous studies have also reported amyloid β-related cerebrovascular lesions and BBB disruption in patients with ALSP ^2,5,15. Interestingly, we and others have further shown that pharmacological or genetic depletion of microglia in AD transgenic mice leads to a dramatic increase in cerebral amyloid angiopathy (CAA)^28,56. We therefore sought to determine whether 8.5-month-old hFIRE mice exhibit any evidence of BBB disruption. Prussian blue staining was performed, revealing significant levels of ferric iron deposition within the thalamus of hFIRE-PBS mice that were completely prevented by microglia transplantation (Figure 5C,D). Immunostaining to detect parenchymal localization of the plasma protein albumin, a commonly used additional marker of BBB disruption⁵⁷, also revealed diffuse extravascular localization of albumin surrounding a subset of CD31 labeled blood vessels within the thalamus of hFIRE-PBS mice (Figure 5E). In contrast, albumin labeling within hFIRE-HPC exhibited a typical blood vessel restricted distribution. Taken together, these data support a growing body of evidence that microglia play an important role in the blood-brain barrier^28,44,58,59.

Figure 5: Microglia engraftment prevents thalamic microhemorrhages and protects against neuronal and synaptic loss in hFIRE mice.

(A) Representative immunostaining for calcification (RIS, gray) and amyloid precursor protein (APP, blue), scale bar 50 μm. (B) Quantification of amyloid precursor protein in the thalamus of 8.5-month-old littermates, hFIRE-PBS, and hFIRE-HPC. P values from one-way ANOVA with Tukey multiple comparisons tests. (C) Representative imaging of Prussian blue positive microbleeds, scale bar 50 μm. (D) Quantification of Prussian blue in the thalamus of 8.5-month-old littermates, hFIRE-PBS, and hFIRE-HPC; P values from one-way ANOVA (F2,24=44.52; ****P<0001) with Tukey multiple comparisons tests. (E) Thalamic microhemorrhages in hFIRE-PBS mice exhibit diffuse parenchymal accumulation of the plasma protein albumin (purple) outside of the vasculature (CD31, green) that is prevented by microglia (IBA1, white) transplantation in hFIRE-HPC mice, scale bar 50 μm. (F) Representative confocal microscopy of NeuN+ neurons (cyan) in 8.5-month hFIRE-PBS and hFIRE-HPC transplanted hemibrains, scale bar 500μm. (G-I) Quantification of NeuN+ neurons in the visual cortex (G), hippocampus (H), and thalamus (I) of 8.5-month littermates, hFIRE-PBS, and hFIRE-HPC mice; P values from one-way ANOVA (cortex: F2,24=2.164; ns=0.1368) (hippocampus: F2,24=0.7322; ns=0.4913) (thalamus: F2,24=29.77; ****P<0.0001) with Tukey multiple comparisons tests. (J-L) Quantification of neuron specific enolase (J), post-synaptic density protein 95 (K), and synaptic vesicle glycoprotein 2A (L) levels in soluble brain extracts; P values from one-way ANOVA (NSE: F2,24=53944; **P=0.0080) (PSD 95: F2,24=8.072; **P=0.0021) (SV2A: F2,24=2.614; ns) with Tukey multiple comparisons tests. Data represented as average mean value ±SEM (Littermates, n = 9; hFIRE-PBS, n=9; hFIRE-HPC, n=9). ns=not significant; *P<0.05, **P<0.01, ***P<0.001, and ****P<0.0001.

To better understand the potential impact of ALSP-associated pathologies on neurons, we next quantified the number of NeuN+ neurons within the visual cortex, CA2/CA3 of the hippocampus, and thalamus. We detected no significant difference in the number of NeuN+ neurons in the cortex or hippocampus, but a highly significant and consistent loss of NeuN+ neurons within the thalamus of 8.5-month-old hFIRE-PBS mice (Figure 5F-I). Using an orthogonal approach, we quantified levels of neuron-specific enolase (NSE) by ELISA revealing a small, but significant reduction in brain NSE levels in hFIRE-PBS mice that was entirely prevented by microglia transplantation in hFIRE-HPC mice (Figure 5J). Biochemical analysis of postsynaptic density protein 95 (PSD-95) levels similarly revealed a significant but subtle decrease in hFIRE-PBS mice that was also fully prevented by microglia transplantation (Figure 5K). Additionally, analysis of synaptic vesicle glycoprotein 2a (SV2A), a presynaptic marker, showed a similar but non-significant pattern (Figure 5L). While quantitative histological or biochemical assessments of neuronal and synaptic density in ALSP patients have yet to be reported, several studies have shown marked reduction in brain volume by MRI^60,61. These results show that hFIRE mice exhibit a significant, but localized loss of neuronal and synaptic markers that can be largely prevented by microglia transplantation.

Microglia engraftment prevents myelin-related abnormalities in hFIRE mice

We next utilized a combination of biochemical and histological approaches to examine myelin integrity in hFIRE mice. Staining with FluoroMyelin Red, a lipid stain with high affinity for myelin, revealed a significant increase in lipid-enriched myelin in hFIRE-PBS mice that was prevented by microglia transplantation in hFIRE-HPC mice (Figure 6A,B). Interestingly, we also observed a significant increase in white matter autofluorescence under UV illumination within the fimbria fornix of 8.5-month-old hFIRE-PBS mice (Figure 6A,C). Lipid accumulation is a common source of autofluorescence, and autofluorescence spectroscopy has previously been shown to measure white matter injury in both animal models and multiple sclerosis patients⁶². To test if the observed autofluorescence was correlated with myelin levels in hFIRE mice we next performed a biochemical analysis of MBP. Interestingly, we not only found significantly elevated levels of MBP in hFIRE-PBS, but also detected a significant but partial reduction in MBP levels within microglial engrafted hFIRE-HPC mice (Figure 6D). Furthermore, linear regression analysis revealed a significant correlation between MBP lysate levels and autofluorescence (R²=0.7508; ****p<0.0001) (Figure 5E), consistent with the recent reported findings of altered lipid metabolism and impaired myelin compaction in both FIRE mice and ALSP patient brains^3,14,25.

Figure 6: Microglia engraftment prevents myelin-related abnormalities in hFIRE mice.

(A) Representative confocal imaging for 405-autofluorescence (blue) and lipid-enriched myelin (FluoroMyelin Red) of 8.5-month-old Littermates, hFIRE-PBS, and hFIRE-HPC mice; hemibrain scale bar 500μm, fornix scale bar 50μm. (B-C) Quantification of FluoroMyelin (B) and autofluorescence (C) in the fimbria fornix of 8.5-month-old; P values from one-way ANOVA (FluoroMyelin F2,24=15.93; ****P<0.0001) (Autofluorescence: F2,24=47.98; ****P<0.0001) with Tukey multiple comparisons tests. (D) Biochemical quantification of myelin basic protein levels in soluble brain samples of 8.5-month-old mice; P values from one-way ANOVA (F2,24=70.42; ****P<0.0001) with Tukey multiple comparisons tests. (E) Simple linear regression model plotted between quantified autofluorescence mean intensity of fimbria fornix and soluble MBP levels of brain samples (R²=0.7508, ****P<0.0001) of 8.5-month-old littermates, hFIRE-PBS, and hFIRE-HPC mice. (F) Representative immunostaining of SERPINA3N (teal) within the fimbria fornix; scale bar, 100μm. (G-H) Quantification of SERPINA3N count (G) and expression (H) in fimbria fornix of 8.5-month-old mice; P values from one-way ANOVA (count: F2,24=32.02; ****P<0.0001) (sum mean intensity: F2,24=25.39; ****P<0.0001) with Tukey multiple comparisons tests. (I) Representative immunostaining of oligodendrocytes (OLIG2, green) of the fimbria fornix; scale bar, 50μm. (J-K) Quantification of OLIG2 count (J) and expression (K) in fimbria fornix of 8.5-month-old mice; P values from one-way ANOVA (count: F2,24=0.4403; ns=0.6489) (sum mean intensity: F2,24=1.310; ns=0.2885) with Tukey multiple comparisons tests. Data represented as average mean value ±SEM (Littermates, n = 9; hFIRE-PBS, n=9; hFIRE-HPC, n=9). ns=not significant; **P<0.01, ***P<0.001, and ****P<0.0001.

Age- and disease-associated changes in oligodendrocyte gene expression are increasingly recognized as a standard component of many neurodegenerative conditions^63,64. Among these changes, increased oligodendroglial expression of SerpinA3N/SERPINA3 is highly conserved, is inversely correlated with cognitive function⁶⁵, and has also been shown to be elevated in FIRE mice²⁵. We therefore examined SerpinA3N by IHC and confocal microscopy. Whereas SerpinA3N immunoreactivity was virtually undetectable in littermate (hCSF1-PBS) control mice (Figure 6F) hFIRE-PBS mice exhibit a substantial elevation of this oligodendroglial protein within the fimbria fornix which was prevented in hFIRE-HPC mice (Figure 6G,H). To determine whether changes in SerpinA3N are indicative of alterations in oligodendrocyte numbers we further examined levels of OLIG2 (Figure 6I), a commonly used marker of oligodendrocytes. This analysis revealed no differences between groups demonstrating that the absence of microglia neither increases nor decreases oligodendrocyte numbers (Figure 6J) or OLIG2 expression (Figure 6K) in 8.5-month-old hFIRE mice. Interestingly, SerpinA3N has been implicated in both neuroinflammation, neuroprotection, and senescence^66–68. Thus, it remains unclear whether increased SerpinA3N directly contributes to altered myelination or represents a compensatory response to white matter pathology. Nevertheless, our data further supports the notion that oligodendroglial SerpinA3N is commonly upregulated in neurodegenerative disease⁶³.

CRISPR correction restores proliferative deficiency in ALSP patient-derived iPSC-microglia

To increase translational relevance, iPSCs were generated from fibroblasts derived from a female ALSP patient carrying a L786S mutation in CSF1R (L786S-Het). CRISPR-Cas9 gene editing was performed using a gRNA designed to specifically target the mutant CSF1R-L786S allele and a template sequence designed to correct the mutation (Figure 7A,B). Isogenic L786S-Het and L786L-corrected iPSCs were then differentiated into iPSC-microglia (iMG) following a widely used and highly validated approach^69,70. We immediately observed that L786S-Het iMG exhibited altered proliferation during the initial 7 days of iMG differentiation. We therefore plated an equivalent density of day 7 iMG into a 96-well plate and employed time lapse imaging to quantify microglial proliferation as previously described⁷¹. Interestingly, we found that L786S-Het iMG were unable to proliferate and exhibited a decrease in confluency over 48 hours (Figure 7C).

Figure 7: CRISPR correction rescues proliferative deficiencies in ALSP patient-derived microglia.

(A) Diagram of CRISPR-targeted mutant sequence within CSF1R. (B) Chromatograms of L786S-Het and L786L-corrected CSF1R iPSC clones following CRISPR/Cas-9 editing. (C) Quantification of L786S-Het (purple) and L786L-corrected (green) iPSC-microglia (iMG) confluency across 48 h of culture in complete iMG media; P values from repeated two-way ANOVA (F1,392=2000; ****P<0.0001) with Sidak’s multiple comparisons tests. Data represented as average normalized confluency. Error bars, SEM. (D) Schematic representing transplantation paradigm of L786S-Het and L786L-corrected iMG transplantation into 4-month-old hFIRE mice. (E) Representative confocal imaging of engrafted hemibrains with mutant L786S and corrected L786L human microglia (xMG; IBA1, green; Ku80, magenta). (F) 6-weeks post-transplantation the number of Ku80/IBA1 double positive human microglia within half brain sections was quantified; P value from unpaired, two-sided t-test (t(4)=16.40; ****P<0.0001). (G-H), Area of microglia engraftment within representative cortical (G) and hippocampal (H) fields; P values from unpaired, two-sided t-test (cortex: t(4)=34.24; ****P<0.0001) (hippocampus: t(4)=5.188; **P=0.0066). (I-J) Higher power view of mutant (I) and corrected (J) xMG within the hippocampus co-labeled with IBA1 (green) and the homeostatic microglia marker P2RY12 (orange). (K-L) Quantification of hippocampal IBA1 (K) and P2RY12 (L) average mean intensity within mutant and corrected microglia; P values from unpaired, two-sided t-test (IBA1: t(4)=0.9525; ns=0.9525) (P2RY12: t(4)=3.634; *P=0.0221). Data represented as average mean value ±SEM (L786S-Het xMG, n = 3; L786L-corrected xMG, n=3). ns=not significant; *P<0.05, **P<0.01, and ****P<0.0001.

To further determine whether the observed L786S-associated iMG proliferation deficit was simply an in vitro phenomenon, equivalent numbers of L786S-Het and L786L-corrected day 7 iMG were transplanted via stereotactic bilateral injections into the hippocampus of 4-month-old female hFIRE mice (Figure 7D). 6-weeks after transplantation, mice were sacrificed, and immunohistological analysis was performed to examine xenotransplanted microglia (xMG). Interestingly, Ku80+/IBA1+ human L786S-Het xMG remained primarily adjacent to the initial hippocampal injection site, whereas L786L-corrected xMG proliferated and migrated extensively to occupy the entire forebrain (Figure 7E-H). Immunostaining with Ki67 and cleaved Caspase-3 to detect proliferative and apoptotic cells, respectively, showed that 6 weeks after transplantation, L786S-Het xMG still exhibited Ki67+ proliferating microglia near the hippocampal injection site (Figure S8A) at significantly higher proportions than L786L-corrected xMG (Figure S8B,C). In contrast, no evidence of microglial apoptosis was detected (Figure S8D,E) although cleaved Caspase-3 positive spheroids were seen in L786S-Het xMG mice consistent with prior reports of caspase activation within axonal spheroids⁴⁰. We previously showed that healthy iPSC-microglia rapidly proliferate along an expanding ‘wavefront’ as they migrate from their injection site to fully occupy an empty microglia niche⁷¹. By one-month healthy microglia have almost entirely populated the brain, ceased proliferating, and exhibit homeostatic morphology and markers (Figures 2D, S1). However, 6 weeks after transplantation, L786S-Het microglia have barely moved beyond their hippocampal injection site, but nevertheless exhibit greater levels of Ki67 expression than L786L-corrected xMG, which had fully engrafted and lost the need for continued proliferation. These results therefore suggest that L786S-Het xMG engraftment deficiency results from impaired or delayed proliferation, but not increased microglial apoptosis.

Further immunostaining revealed L786S-Het and L786L-corrected xMG exhibited similar levels of the microglial/myeloid marker IBA1 (Figure 7I,J,K). However, CRISPR correction of L786S-Het led to a significant increase in expression of the purinergic receptor P2RY12 (Figure 7I,J,L), a homeostatic microglia marker previously shown to be reduced in ALSP-patient brains³. These results underscore the important role CSF1R signaling plays in microglia viability, proliferation, and migration and demonstrate that CRISPR-correction of a patient-derived iPSC line can alleviate ALSP-mutant microglial deficiencies, restoring a homeostatic microglial state.

Transplantation of patient-corrected microglia reverses ALSP-associated neuropathologies

To determine whether microglia transplantation could reverse ALSP-related neuropathologies, IHC was performed on 4-month-old non-transplanted wildtype-hCSF1 littermates and hFIRE mice as well as 5.5-month-old hFIRE mice that received bilateral hippocampal injections of L786S-Het or L786L-corrected iMG at 4-months of age. In comparison to age-matched littermates, 4-month-old hFIRE mice exhibited increased levels of LAMP1+ axonal spheroids within the hippocampus and fornix (Figure S9A,C). Interestingly, by 5.5 months, we observed a significant increase in spheroid volume (Figure S9B,D), but not number, of L786S-Het transplanted mice compared to 4-month-old transplanted mice (Figure 8A,D,C,F). Previous reports have shown axonal spheroids can continue to swell until they rupture³⁸. Thus, it appears that transplantation of mutant iMG fails to slow or may even promote the continued enlargement of axonal spheroids in hFIRE mice. In stark contrast, engraftment of L786L-corrected iMG eliminated virtually all observable axonal swellings within the hippocampus and fornix within just 6 weeks of transplantation (Figure 8B,C,E,F). Additional analysis of RIS+ calcifications and GFAP immunoreactivity revealed very similar differential effects of mutant versus corrected xMG (Figure 8G,H; Figure S9E). Within 6-weeks of transplantation, L786L-corrected xMG significantly reduced both calcifications and astrogliosis within the thalamus of transplanted hFIRE mice, decreasing these pathologies to levels that were not significantly different from 4-month old wildtype hCSF1 littermates (Figure 8I,J; Figure S9E). We similarly found that L786L-corrected xMG significantly reduced OPN accumulation within the thalamus (Figure 8K-M; Figure S9F). Any remaining OPN was found to be tightly associated with residual RIS+ calcifications which were in turn surrounded by IBA1+ microglial processes. This colocalization of microglial processes and calcium deposits, along with the observed reduction in RIS levels, and a prior report that iMG can phagocytose calcium hydroxyapatite crystals²⁸, suggest that L786L-corrected iMG can partially clear existing calcium deposits via phagocytosis.

Figure 8: Transplantation of patient-corrected microglia reverses pre-existing ALSP-associated neuropathologies.

(A-B) hFIRE mice were transplanted with L786S mutant or L786L CRISPR-corrected human microglia at 4-months of age and examined six weeks later (5.5-months). Representative immunostaining of axonal spheroids (LAMP1, red) and microglia (IBA1, green) within the CA2/CA3 of the hippocampus 5.5-month-old hFIRE mice transplanted with L786S-Het (A) or L786L-corrected (B) human microglia, scale bar 100 μm. (C) Quantification of LAMP1+ spheroid numbers within the hippocampus of non-transplanted 4-month-old Littermates (blue) and non-transplanted hFIRE mice (light green) in comparison to 5.5-month-old hFIRE mice that were transplanted at 4 months of age with L786S-Het (purple) or L786L-corrected (dark green) microglia. Representative images of littermate controls provided in Figure S9; P values from one-way ANOVA (F3,10=39.83; ****P<0.0001) with Tukey multiple comparisons tests. (D-E) Representative immunostaining of LAMP1+ axonal spheroids and microglia within the fimbria fornix of 5.5-month-old hFIRE mice transplanted with L786S-Het (D) or L786L-corrected (E) human microglia, scale bar 100 μm. (F) Quantification of LAMP1+ spheroid numbers within the fimbria fornix of non-transplanted 4-month-old Littermates and non-transplanted hFIRE mice in comparison to 5.5-month-old hFIRE mice that were transplanted at 4 months of age with L786S-Het or L786L-corrected microglia. Representative images of littermate controls provided in Figure S9; P values from one-way ANOVA (F3,10=47.92; ****P<0.0001) with Tukey multiple comparisons tests. (G-H) Representative immunostaining of astrogliosis (GFAP, yellow) and calcification (RIS, gray) in the thalamus of L786S-Het (G) and L786L-corrected (H) transplanted 5.5-month-old hFIRE mice, scale bar 100μm. (I-J) Quantification of RIS+ calcifications (I) and GFAP+ astrogliosis (J) of non-transplanted 4-month-old Littermates and hFIRE mice and 5.5-month-old L786S-Het and L786L-corrected transplanted hFIRE mice. Representative images of littermate controls provided in Figure S9; P values from one-way ANOVA (GFAP: F3,10=14.44; ***P=0.0006) (RIS: F3,10=22.81; ****P<0.0001) with Tukey multiple comparisons tests. (K-L) Representative immunostaining of calcification (RIS, gray) and osteopontin (OPN, red) accumulation in the thalamus of L786S-Het (K) and L786L-corrected (L) human microglia (Ku80, magenta; IBA1, green) transplanted 5.5-month-old hFIRE mice, scale bar 100 μm. High power scale bar 20 μm. (M) Quantification of OPN accumulation in 4-month-old Littermates and hFIRE mice and 5.5-month-old L786S-Het and L786L-corrected transplanted hFIRE mice; P values from one-way ANOVA (fornix: F3,10=14.36; ***P=0.0006) with Tukey multiple comparisons tests. Data represented as mean value ±SEM (4-month-old Littermates, n = 4; 4-month-old hFIRE, n=4; 5.5-month-old hFIRE-L786S-Het, n=3, 5.5-month-old hFIRE-L786L-corrected, n=3). *P<0.05, **P<0.01, ***P<0.001, and ****P<0.0001. Comparisons not shown are not significant.

To determine whether microglial transplantation can similarly reverse white matter abnormalities, immunostaining with FluoroMyelin Red and SerpinA3N was performed. Although 4-month-old hFIRE mice exhibited a slight but insignificant accumulation of myelin-associated lipids, 5.5-month-old L786S-Het transplanted hFIRE mice exhibited significantly higher levels while L786L-corrected microglial transplantation returned levels to baseline (Figure S9G-H). However, 4-month-old hFIRE mice exhibited significantly higher expression of SerpinA3N within the fimbria fornix compared with 4-month-old littermates. Within 6-weeks of transplantation, L786S-Het xMG maintained these elevated levels while transplantation of L786L-corrected microglia resulted in a reduction of SerpinA3N expression. Taken together, these results provide the first preclinical evidence that transplantation of either allogeneic CSF1R-wildtype iPSC-microglia or autologous CSF1R-corrected microglia could provide significant therapeutic potential by reducing the diverse neuropathologies found in ALSP.

Discussion

ALSP is a devastating and currently untreatable neurodegenerative disease driven by dominant mutations in CSF1R, a gene critical for microglial development, proliferation, and survival^1,72. As a primary microgliopathy, the pathological and clinical features of ALSP arise from a progressive loss of homeostatic microglia. Whether ALSP-associated CSF1R mutations drive disease via haploinsufficiency or dominant-negative mechanisms continues to be debated^7,29. However, published studies of kinase-dead mutations in mice and zebrafish strongly support a dominant negative model^7,8. Arguably, the potent effect of the heterozygous L786S mutations on human iPSC microglia observed in the current study also favors a dominant inhibitory effect of the mutant allele. However, it is also the case that loss-of-function CSF1R mutations are not dosage compensated⁹ so a heterozygous mutant does produce at least 50% loss of expression which may be sufficient to produce microglial loss in some individuals. In any case, reduced microglial density is consistently observed in ALSP patient brains; thus, models that mimic this key feature are likely to be most informative^3,4. A handful of prior studies have sought to model ALSP via heterozygous deletion of CSF1R^29,30,73,74. However, these mice exhibit either increased microglial density or no changes in microglia numbers, thus failing to recapitulate the underlying cause of human ALSP. Likewise, heterozygous deletion of CSF1R in rats did not lead to any detectable brain phenotypes¹⁰.

In the current and accompanying study by Munro and colleagues, we demonstrate that as FIRE-knockout mice age, they develop axonal spheroids, brain calcifications, alterations in myelination and oligodendrocytes, astrogliosis, and transcriptional and biochemical changes that resemble some of the neuropathological features of human ALSP (Figures 1,3-6). Importantly, we also show that transplanted wildtype human microglial progenitors can rapidly expand to fill the microglial niche restoring canonical microglial transcripts (Figure 2), and prophylactically preventing many of these ALSP-associated pathologies (Figure 3–6). In addition, we find that ALSP patient-derived microglia exhibit deficits in microglia proliferation that can be rescued via CRISPR-correction of mutant CSF1R, enabling robust brain-wide engraftment (Figure 7) and reversing pre-existing pathologies (Figure 8). Taken together, our study demonstrates that FIRE mice provide a valuable new model of ALSP, and that transplantation of iPSC-microglia warrants further examination as a promising new therapeutic approach. However, it is also important to note that the impacts of Csf1r mutations in mice depend on genetic background. Interestingly, clinical assessments of patients have shown uneven penetrance and disease progression among family members with CSF1R mutations^75–78. Therefore, the severe pathology associated with human disease may also be associated with an environmental trigger and/or epistatic interactions with other gene variants.

Current therapies for ALSP are limited and focus primarily on symptom management. Although a handful of ALSP patients have received HSCT, at least one death has been described in the literature as having directly resulted from preconditioning²². A second major challenge to HSCT is that currently approved paradigms lead to minimal chimerism within the brain parenchyma, with only 1–2% of the microglial population being replaced by infiltrating macrophages⁷⁹. Recent preclinical mouse studies have developed novel and promising approaches to rectify this specific problem by combining HSCT with microglial depletion via CSF1R antagonists, genetic approaches, or whole-body irradiation,^80–85 treatments that would likely exacerbate ALSP by promoting additional microglial loss. Importantly, these same studies have also provided a wealth of consistent new data that support a third major problem with HSCT: when bone marrow-derived monocytes, macrophages, or circulation-derived myeloid cells (CDMC) are successfully recruited to the brain to “replace microglia” they remain transcriptionally and functionally distinct from microglia, regardless of long-term brain residence^80–84. Whether these differences could impact brain health currently remain unclear, but bone marrow derived macrophages typically exhibit far greater phagocytic and antigen presentation activity than microglia, which could in turn elicit detrimental effects within the nervous system. These three important HSCT-associated problems likely explain the deterioration of cognition in ALSP patients following HSCT^20,22.

An alternative approach that our data suggests would avoid the challenges of HSCT, involves the replacement of missing and defective microglia with healthy stem cell derived-microglia, instead of peripheral HSCT-derived myeloid cells. The current study provides promising initial evidence that human iMG can prevent and even reverse many neuropathologies associated with ALSP, including axonal spheroids, astrogliosis, brain calcifications, and myelination defects. In addition, we demonstrate the feasibility of either an allogeneic or autologous approach by demonstrating that ALSP-associated pathologies can be treated by transplanting healthy donor-derived or CRISPR-corrected ALSP patient-derived iPSC-microglia.

Limitations of the study

The current study provides initial preclinical evidence that transplantation of human iPSC-derived microglia can reduce many of the pathologies associated with ALSP. However, the potential translation of these findings toward the clinic will of course require significant further study. As is the case for all mouse models of human neurological disease, the hFIRE model mimics many, but not all of the features of human ALSP. In particular, several groups have examined but been unable to detect significant cognitive or motor impairments in FIRE mice using traditional behavioral paradigms⁸⁶ (Munro et al.). It also remains to be seen whether transplantation of CSF1R-corrected microglia into brains with reduced numbers but not absence of microglia, will require the elimination of endogenous CSF1R-mutant microglia. Future studies should therefore determine whether healthy CRISPR-corrected microglia can outcompete endogenous CSF1R mutant microglia within the brain. Based on the considerable differences in microglial proliferation rates observed in the current study coupled with prior estimates that human microglia turnover at a rate of ~28% per year⁸⁷, it is quite plausible that healthy iPSC-microglia will carry a selective advantage over mutant ALSP microglia, leading to a gradual replacement of mutant cells. However, experiments designed to specifically test this hypothesis are needed. Lastly, it will be critical to carefully assess the long-term safety of iPSC-microglia transplantation. Fortunately, our fully defined differentiation methods have been shown by many independent groups to yield a highly pure population of microglia^88–91. Additionally, there are no known examples of human microglial-derived tumors^92–94. Thus, microglia may offer an intrinsically safe cell therapy that can be developed as a treatment for ALSP and perhaps many other neurological diseases.

STAR Methods

Resource Availability

Lead Contact

Further information and requests for resources and reagents Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Mathew Blurton-Jones (mblurton@uci.edu).

Materials availability

All unique/stable reagents and human iPSC lines generated in this study are available from the Lead Contact, Mathew Blurton-Jones (mblurton@uci.edu), with a completed Materials Transfer Agreement.

Data and Code Availability

• Data availability: All data in this paper will be shared by the Lead Contact upon request.

  • RAW files of bulk RNA seq are available through the GEO SuperSeries accession number GSE253623.

• Code: This paper reports original code for FASTQ read classification, species-specific FASTQ sorting, and differential expression analysis using the DESeq2 R package available at https://github.com/jhasselmann808/Chadarevian_2024_Leukoencephalopathy, https://doi.org/10.5281/zenodo.11211703.

• Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact upon request.

Experimental Models and Subject Details

Mouse Model

All animal procedures were conducted in accordance with guidelines set forth by the National Institutes of Health and the University of California, Irvine Institutional Animal Care and Use Committee. CSF1R^{ΔFIRE/ΔFIRE} mice were generated and previously characterized by Clare Pridans and David Hume²⁴. FIRE mice were generated on a B6CBAF1/J background. Heterozygous FIRE mice were then crossed with wild-type C57BL/6J mice for four generations before backcrossing for four generations with xenotransplantation-compatible hCSF1 mice (Jackson Laboratories # 017708) which harbor deletions in Rag2 and iL2rg and knockin of humanized CSF-1 on a BALB/c background. The resulting hFIRE mice were maintained via breeding of heterozygous FIRE females (FIRE^−/+, RAG2^−/−, iL2rg^−/−, hCSF1^+/+) with homozygous (FIRE^−/−, RAG2^−/−, iL2rg^−/y, hCSF1^+/+) or heterozygous males (FIRE^−/+, RAG2^−/−, iL2rg^−/y, hCSF1^+/+). Genotyping of FIRE mice was performed using the following primers: mCSF1R_FIRE_F (5’-GCGGTTGTAGGAAACCCTGA-3’), mCSF1R_WT_F (5’-GGTGCCAGCAATGTGTTTCC-3’), and mCSF1R_R: (5’-CACTCCTACCACTGGGCATC-3’). All female mice were age-matched and group-housed on a 12-h/12-h light/dark cycle with food and water ad libitum. Male mice (Figure S5) were age-matched but individually housed following intracranial transplantation on a 12-h/12-h light/dark cycle with food and water ad libitum. All mice were housed with ambient temperature and humidity. Cages and bedding were changed every 1–2 w.

Induced Pluripotent Stem Cell (iPSC) generation and culture

An ALSP patient carrying a heterozygous L786S mutation in CSF1R was clinically assessed by Dr. Wszolek at the Mayo Clinic Jacksonville. Following IRB-approved informed consent (Mayo Clinic IRB #09–003803.), a skin biopsy was collected, and primary fibroblasts were expanded and stored in liquid nitrogen. Frozen fibroblast cultures were then transferred to the University of California Irvine. The parental ALSP L786S-Het patient-derived iPSC line was generated using non-integrating Sendai virus (CytoTune-iPS 2.0 Sendai Reprogramming Kit; A16517; Thermo Fisher Scientific) thereby avoiding any integration-induced effects. Resulting iPSCs were confirmed to be karyotypically normal by Array Comparative Genomic Hybridization (Cell Line Genetics), sterile (mycoalert, Lonza), and pluripotent via trilineage in vitro differentiation (STEMdiff Trilineage Differentiation Kit; #05230; STEMCELL Technologies) (Figure S7).

The parental ADRC75, ADRC76 and ADRC77 iPSC lines were similarly generated by the University of California Alzheimer’s Disease Research Center (UCI-ADRC) from subject fibroblasts under approved Institutional Review Board and human Stem Cell Research Oversight protocols and provided by the Institute for Memory Impairments and Neurological Disorders. The parental WTC11-mEGFP (AICS-0036–006; Coriell) human iPSC line was acquired through the Allen Cell Collection, available from Coriell Institute for Medical Research.

All iPSC lines were maintained on matrigel-coated plates (CB356238; Corning) according to the manufacturer’s specifications in TeSR-E8 (05990; STEMCELL Technologies). Cultured iPSC media was replenished every day with fresh medium. Cells were passaged every 5–7 d using ReLeSR (100–0484; STEMCELL Technologies) according to the manufacturer’s specifications. For all in vitro experiments, iPSCs were cultured in 5% O₂, 5% CO₂ at 37°C.

Method Details

CRISPR/Cas9 correction of patient-derived iPSC line

Generation of L786L-corrected iPSC line was generated following a previously published protocol by Chadarevian et al¹⁰⁰. L786S-Het iPSCs were collected following Accutase (07920; STEMCELL Technologies) enzymatic digestion for 3 min at 37°C. 250,000 cells were resuspended in a 100 μL nucleofection buffer from Human Stem Cell Nucleofector Kit 2 (VPH-5022; Lonza). Ribonucleoprotein (RNP) complex was formed by incubating HiFi Cas9 Nuclease V3 (50 μg; IDTDNA, 1081061) with CRISPR RNA (5′-CGUAACGUGCUGUCGACCAA-3′): trans-activating CRISPR RNA (IDTDNA, 1072534) duplex. L786S-targeting RNP complex was then combined with L786L-corrected single-stranded oligodeoxynucleotide template (5′-GTGCTTTCCCTCAGTGCATCCACCGGGACGTGGCAGCGCGTAACGTGCTGTTGACCAATG GTCATGTGGCCAAGATTGGGGACTTCGGGCTGGCTAGGGACATCATGAAT-3′) (2–4 μM; IDTDNA) and the cellular suspension for nucleofection using the Amaxa Nucleofector program B-016. Cells were plated on a matrigel-coated plate in TeSR-E8 media with 0.25 μM Thiazovivin (72254; STEMCELL Technologies) and CloneR (05889; STEMCELL Technologies) overnight to recover. The following day, cells were collected following Accutase enzymatic digestion for 1 min at 37°C then manually single-cell plated into 96-well (3595; Corning) matrigel-coated plates in TeSR-E8 media with 0.25 μM Thiazovivin and CloneR supplement for clonal isolation and expansion. Culture media was replenished everyday with fresh medium. Plates were visually screened to identify single-clone wells after 5 d. Visually clonal wells were passaged with ReLeSR after 10 d. A cell pellet was collected from each well from which genomic DNA was extracted using Extracta DNA prep for PCR (95091; Quantabio) amplification with Taq PCR Master Mix (K0172; Thermo Fisher Scientific). To confirm CRISPR-repair of the CSF1R-L786S mutation, clones were screened via PCR followed by Sanger sequencing using the following primers: CSF1R_L786S_F (5’-GAAGGCCCAAGACTAACCCT-3’) and CSF1R_L786S_R (5’-GAGGATGCCATAGGACCAGAC-3’).

Differentiation of Hematopoietic progenitor cells (HPCs) and microglia (iMG) from iPSCs

HPCs and iPSC-derived microglia (iMG) were differentiated following the highly replicated protocol published by McQuade et al⁷⁰. To begin HPC differentiation, iPSCs were passaged onto 6-well matrigel-coated plates in TeSR-E8 at a density of 80 colonies of 100 cells each per 35 mm well. On day 0, cells were transferred to Medium A from the STEMdiff Hematopoietic Kit (05310; STEMCELL Technologies). On day 3, cells were exposed to Medium B, and remained in Medium B for 7 additional days while small round HPCs began to lift off the colonies. On day 10, non-adherent CD43+ HPCs were collected by carefully removing medium and non-adherent cells with a serological pipette. At this point, HPCs were frozen in BamBanker freezing solution (Wako, NC9582225) for long-term storage. Cells used for HPC transplantation were later thawed in complete iMG medium (DMEM/F12 [11039–021; Thermo Fisher Scientific], 2× insulin-transferrin-selenite [41400045; Thermo Fisher Scientific], 2× B27 [17504044; Thermo Fisher Scientific], 0.5× N2 [17502048; Thermo Fisher Scientific], 1× glutamax [35050–061; Thermo Fisher Scientific], 1× non-essential amino acids [11140–050; Thermo Fisher Scientific], 400 mM monothioglycerol, and 5 mg/mL human insulin [I2643; Sigma-Aldrich] freshly supplemented with 100 ng/mL IL-34 [Cat #200–34; Peprotech], 50 ng/mL TGFb1 [Cat #100–21; Peprotech], and 25 ng/mL M-CSF [Cat #300–25; Peprotech]) for 18–24 h to recover before being resuspended at 62,500 cells/μl in 1× DPBS (low Ca2+, low Mg2+) for transplantation. For iMG differentiation, HPCs were cultured on 6-well matrigel-coated plates in complete iMG medium for 7 days. Fresh complete iMG media was added every 24 h.

iMG Confluency

ALSP L786S-Het and L786L-corrected iMG were plated at 90,000 cells per 96-well matrigel-coated plate (six wells per line per condition). At time 0 h, all microglia were cultured with iMG basal media supplemented with 100 ng/mL IL-34, 50 ng/mL TGFb1, and 25 ng/mL M-CSF. Four 20x images per well were collected every hour for 48 h. Using IncuCyte 2020B software, image masks for phase confluence were generated. Graphs display confluency across 48 h normalized to 0 h time point.

Adult Intracranial Transplantation

All mouse surgeries and use were performed in strict accordance with approved National Institutes of Health (NIH) and institutional guidelines. Direct bilateral intracranial injections were performed as detailed in Hasselmann et al²⁶. Briefly, adult mice (2 m old) were anesthetized under continuous isoflurane and secured to a stereotaxic frame (Kopf), and local anesthetic (Lidocaine 2%; Medline, 17478–711-31) was applied to the head before exposing the skull. Using a 30-gauge needle affixed to a 10-mL Hamilton syringe, mice received 2 μL of cell suspension in sterile 1×DPBS (14190144; Thermo Fisher Scientific) at 62,500 cells/μL at each injection site. HPC transplantation was conducted bilaterally into the lateral parietal association cortex and dorsal hippocampus at the following coordinates relative to bregma: anteroposterior, 2.06 mm; mediolateral, ± 1.75 mm; dorsoventral, 1.75 mm (hippocampus) and 0.95 mm (cortex). Cells were injected at a rate of 62,500 cells/30 s with 4 min diffusion time in between injections. iMG transplantation was conducted bilaterally into the dorsal hippocampus only with 2 μL volume of cells at 62,500 cells/μL concentration per side. The needle was cleaned with consecutive washes of DPBS, 70% (vol/vol) ethanol, and DPBS in between hemispheres and animals. Animals were allowed to recover on heating pads before being placed in their home cages and received 2 mg/mL Acetaminophen (Mapap; Major, 0904–7014-16) diluted in water for 10 d.

Collection and Isolation of Peripheral Blood Plasma/Sera

Blood was collected into an EDTA-anticoagulant collection tube and immediately placed on ice. Tubes were then transferred to a pre-chilled centrifuge and spun at 1,600 rcf for 15 min at 4°C. Following centrifugation, plasma supernatant was gently collected and transferred to a 1.5 mL Eppendorf tube on ice. Eppendorf tubes were then transferred to a pre-chilled centrifuge and spun at 1,600 rcf for 15 min at 4°C. Following the second centrifugation, plasma supernatant was collected and aliquoted into 30 μL sub-aliquots and kept at −80°C for long-term storage. To isolate sera, blood was collected into a 1.5 mL Eppendorf tube and kept at 23°C for 1 hr before transferring to 4°C overnight. The following day, samples were centrifuged at 1500 rcf for 10 min. The translucent supernatant was then removed and transferred to a new tube, and samples were centrifuged at 4,500 rcf for 10 min. Following the second centrifugation, sera supernatant was collected and transferred to a fresh 1.5 mL Eppendorf tube and kept at −80°C for long-term storage.

Isolation of Soluble Protein Homogenates

Animals were perfused (20 mL per min) with 1X DPBS (4°C) for 3 min. Brains were surgically excised from each mouse and cut in half along the mid-sagittal plane. The left hemisphere was placed in 4% (wt/vol) paraformaldehyde (PFA) for 36 h at 4°C for subsequent immunohistochemistry (IHC) and the right hemisphere fresh-frozen on dry ice and stored at −80°C for biochemical analysis. Fresh frozen hemispheres were crushed on dry ice using mortar and pestle, then homogenized in solution of T-PER (Thermo Scientific, 78510) and Halt protease and phosphatase inhibitor cocktail (Thermo Scientific, P178446). Homogenates were centrifuged at 16,000 rcf for 30 min at 4°C and supernatants were stored at −80°C for analysis. Protein concentration of soluble brain samples was determined using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23227) and 1 mg/mL of total protein per sample was run for ELISAs.

Biochemical Analysis

Analysis of soluble brain samples were conducted using commercially available ELISAs following manufacturer guidelines to measure Mouse MCP1/CCL2 (ab208979; Abcam), Glial Fibrillary Acidic Protein (ABIN6574131; antibodies-online), Synaptic Vesicle Glycoprotein 2A (abx548000; Abbexa), Neuron Specific Enolase (ab233626; Abcam), Postsynaptic density protein 95 (K15OQND; MSD), and Myelin Basic Protein (CSB-E08285m; CUSABIO). Analysis of peripheral blood plasma was conducted using commercially available Glial Fibrillary Acidic Protein (ABIN6574131; antibodies-online) ELISA. Analysis of blood sera was conducted using the V-PLEX Plus Mouse Cytokine 19-Plex Kit (MSD, K15255G-1). Protein levels for each sample were then calculated via comparison to each assay-specific standard curve.

Immunohistochemistry

Animals were perfused with DPBS and isolated brain hemispheres were drop fixed in 4% (wt/vol) PFA for 36 h then cryoprotected in a 30% (wt/vol) sucrose at 4°C. Brains were sectioned coronally into 30-μm-thick slices on a freezing microtome (Leica, SM 2010R) and stored in a solution of 0.05% NaN3 (S2002; Sigma-Aldrich) in 1× PBS (P44017–100TAB; Sigma-Aldrich) as free-floating slices. For staining, tissue was blocked for 1 h in 1× PBS, 0.2% Triton X-100 (9002–93-1; Thermo Fisher Scientific), and 10% donkey serum (NC9624464; Thermo Fisher Scientific). Immediately following blocking, brain sections were placed in primary antibodies diluted in 1× PBS and 1% donkey serum and incubated ON whilst shaking at 4°C. Samples were then incubated in conjugated secondary antibodies for 1 h followed by mounting on microscope slides. Sections were labeled with combinations of goat anti-IBA1 (1:300; ab5076; Abcam), rabbit anti-IBA1 (1:300; 019–19741; Wako), rabbit anti-Ku80 (1:200; ab80592; Abcam), mouse anti-Ku80 (1:250; MAS-12933; Invitrogen), rabbit anti-Ki67 (1:200; ab16667; Abcam), rabbit anti-P2RY12 (1:500; HPA014518; Millipore Sigma), rabbit anti-NeuN (1:1000; abn78; Abcam), chicken anti-GFAP (1:2000; ab4674; Abcam), goat anti-OPN (1:500; AF808; R&D biosystems), mouse anti-SMI312 (1:1000; 837904; biolegend), Rat anti-Lamp1 (1:300; ab25245; Abcam), rat anti-MBP (1:400; ab7349; Abcam), rabbit anti-MBP-Degraded (1:2000; M9758–04; US Biological), rabbit anti-NF-L-Degenotag (1:5,000; RPCA-NF-L-Degen; EnCor Biotechnology), chicken anti-MAP2 (1:500, AB5392; Abcam), rabbit anti-cleaved Caspase-3 (1:1000; 9664; Cell Signaling), rat anti-CD31 (1:100; BD Biosciences; 553370), goat anti-mouse albumin (1:50; Bethyl Laboratories; A90–234A), RIS-647 (1:1000; BV500101; Biovnc), Fluoromyelin Red (1:300; F34652; Invitrogen), goat anti-Olig2 (1:500; AF2418; R&D biosystems), rabbit anti-pT231 (1:1000; 70105G; Invitrogen), mouse anti-APP (1:300; 14974982; clone 22C11; Invitrogen), or goat anti-SerpinA3N (1:150; AF4709; R&D biosystems) overnight at 4°C followed by 1 h incubation at 23°C whilst shaking. After washing with 1× PBS for 5 m three times, sections were incubated with highly cross-adsorbed AlexaFluor-conjugated secondary antibodies (1:400; ThermoFischer) for 1 h in the dark, then washed three times with 1× PBS before mounting with Fluoromount-G (0100–01; SouthernBiotech) DAPI Fluoromount-G (0100–20; SouthernBiotech). Amplification of cleaved Caspase3 signal was performed using biotinylated anti-rabbit igG (H+L) (1:200; BA-1000; Vector Laboratories) and rabbit anti-Streptavidin 555 (1:500; S32355; Life technologies).

Immunocytochemistry

Following trilineage differentiation (STEMdiff Trilineage Differentiation Kit; #05230; STEMCELL Technologies) in 4-well chamber slide, media was aspirated from each well. Cells were then washed with 1× PBS twice then fixed with 4% (wt/vol) PFA for 15 m at 23°C. Fixative was removed and cells were washed with 1× PBS twice then blocked with 10% donkey serum. Immediately following blocking, wells were individually labeled with goat anti-OTX2 (1:40; AF1979; R&D biosystems), goat anti-Brachyury (1:40; AF2085; R&D biosystems), or goat anti-SOX17 (1:50; AF1924; R&D biosystems) for 1 h at 23°C. Cells were then washed with 1× PBS for 5 m three times then incubated with highly cross-absorbed AlexaFluor-conjugated anti-goat 555 antibody (1:400; ThermoFischer) for 1 h in the dark. Following secondary incubation, cells were then washed three times with 1× PBS before mounting with DAPI Fluoromount G (00100–20; SouthernBiotech).

Imaging acquisition and processing

Immunofluorescent sections were visualized and captured using an Olympus FV3000 confocal microscope using identical confocal and Z-stack settings for a given quantitative comparison. Half brain stitches to capture half-brains with high resolution were performed using Fluoview FV31S-DT software. For each quantitative analysis, three sections were imaged per region per animal using the “Surfaces” function in IMARIS software (Bitplane). All images were examined blinded to genotype and treatment using batch processing. Numbers, area, volume, or intensity were calculated for each section and these values were averaged to provide a single value per animal for subsequent statistical comparisons. For some representative images, brightness and contrast settings were slightly and equally adjusted across groups to reveal fine structures and morphology. Importantly, no such changes were made to any images used for quantification.

RNA analysis, Library Construction and Bulk RNA-seq

Following RNA isolation from crushed fresh frozen hemispheres using commercially available kit (Qiagen, 74106) RNA integrity (RIN) values were determined using an Agilent Bioanalyzer 2100 series and RNA concentrations were assayed by Qubit and all samples had RIN values ranging from 9.5–10. 350ng of total RNA was used as input for library construction using the Illumina TruSeq Stranded mRNA kit which utilizes oligo-dT beads to select and amplify mRNA. cDNA was synthesized using nine rounds of PCR and quality and concentrations of the DNA libraries were assayed using the Agilent 2100 Bioanalyzer high sensitivity DNA assay and the DNA high sensitivity Qubit. The libraries were quantified by Kapa qPCR, normalized to 1.5nM and then multiplexed at equimolar concentrations for sequencing on the Illumina NovaSeq 6000 platform with paired-end read 100 base chemistry on an S4 flow cell.

RNA-Seq FASTQ Preprocessing and Mixed Species Alignment

FASTQ files were preprocessed using BBDuk to filter out ribosomal RNA and PhiX reads, trim Illumina adapters, and to quality trim any base pairs below a PHRED score of 10. FASTQC was then performed to verify the quality of the sequencing files and all files were determined to be of sufficient quality for downstream processing. Reads were then aligned to a mixed species genome consisting of the unmasked top level FASTA files for the human GRCh38 (Ensembl release 99) and mouse GRCm39 (Ensembl release 107) DNA assemblies and aligned using HISAT2 using the -q, −1, and −2 options to align paired-end FASTQ files. The output SAM files were then parsed using a modified version of the “byPrim.py” Python script from Song et al. (2022)⁹⁵ which assigned each read to either mouse or human based on the species for which the primary alignment was specified by HISAT2.

Separation of Human and Mouse Bulk RNA-seq Reads and Pseudoalignment

A Python script was then written to split the preprocessed FASTQ files into mouse and human specific read files. Using the species assignments stored in the output of the “byPrim_MBJ.py” script, the “MBJ_FASTQSorter.py” script generated two sets of paired end reads for each original file, one pair consisting of the forward and reverse human reads and one pair consisting of the forward and reverse mouse reads. As microglia only make up a fraction of the overall brain, the human files averaged approximately 1.07×10⁶ paired reads per sample and the mouse files averaged 39.75×10⁶ paired reads per sample. Species-specific FASTQ files were then pseudoaligned to the human GRCh38 transcriptome (Ensembl release 107) or the mouse GRCm39 transcriptome (Ensembl release 107) using Kallisto v0.46.1⁹⁷.

Differential Gene Expression Analysis

Mouse transcripts were then summarized to the gene level via tximport⁹⁸ and the dataset was filtered to retain only genes with 1:1 mouse to human gene homologues and then subsequently filtered to only retain genes with greater than 130 cumulative reads (average of 10 reads per sample). Human transcripts were then similarly summarized to the gene level, filtered to only retain genes with 1:1 homologue, and then both the mouse and human datasets were intersected to only retain genes present in both datasets. This resulted in a total of 13,947 genes included for downstream analyses (Table S1). As the mouse libraries were ~40 times deeper than the human libraries on average, VST normalization of the merged object failed to fit a regression model to the data. Therefore, differential gene expression analysis was performed on the DESeq2⁹⁶ object containing only the mouse files using the formula “~Mouse_Cell” to distinguish between the different mouse strains and injection type (see Table S1 for metadata table). The human and mouse samples were then independently log normalized using the DESeq2 varianceStabilizingTransformation (VST) function and the independently normalized values were then merged into a single data table for generation of mixed species heatmaps and boxplots.

Bulk RNA-seq Data Visualization

Heatmaps were generated using the R “Pheatmap” package while box plots were generated using the “ggplot2” package. Plots for Figure 2 were generated using the mouse-only VST file and differential gene expression analysis values, while the heatmap and box plots in Figure 2 were generated using the mixed species VST normalized data and filtered according to significant genes identified by the mouse-only differential gene expression analysis.

Quantification and statistical analysis

GraphPad Prism 9 was used to perform statistical tests and generate P values. We used standard designation of P values throughout the Figures (ns = not significant or p≥0.05; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001). Values are represented as average mean ± SEM. Details of number of replicates and the specific statistical test used are provided in the individual figure legends. Data represented for all antibodies calculated from an average of three matched coronal sections per animal.

Supplementary Material

1

Document S1. Figures S1-S9

2

Table S1. Bulk RNA seq of 8.5-month-old hCSF1 littermates, hFIRE-PBS, and hFIRE-HPC brains, related to Figure 2.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Goat anti-IBA	Abcam	Cat #ab5076
Rabbit anti-Ku80	Abcam	Cat #80592
Rabbit anti-IBA	Wako	Cat #019–19741
Mouse anti-Ku80	Invitrogen	Cat #MAS-12933
Rabbitt anti-P2RY12	Millipore Sigma	Cat #HPA014518
Rabbit anti-NeuN	Abcam	Cat #abn78
Chicken anti-GFAP	Abcam	Cat# ab4674
Goat anti-OPN	R&D biosystems	Cat #AF808
Mouse anti-SMI312	Biolegend	Cat #837904
Rat anti-Lamp1	Abcam	Cat# ab25245
Rat anti-MBP	Abcam	Cat# ab7349
Goat anti-Olig2	R&D biosystems	Cat# AF2418
Rabbit anti-pT231	Invitrogen	Cat# 70105G
Mouse anti-APP	Invitrogen	Cat# 14974982
Goat anti-SerpinA3N	R&D biosystems	Cat #AF4707
Goat anti-OTX2	R&D biosystems	Cat #AF1979
Goat anti-Brachyury	R&D biosystems	Cat #AF2085
Goat anti-SOX17	R&D biosystems	Cat #AF1924
Rabbit anti-MBP-Degraded	US Biological	Cat #M9758–04
Rabbit anti-NF-L-Degenotag	EnCor Biotechnology	Cat #RPCA-NF-L-Degen
Chicken anti-MAP2	Abcam	Cat #AB5392
Rabbit anti-cleaved Caspase-3	Cell Signaling	Cat #9664
Rat anti-CD31	BD Biosciences	Cat #553370
Goat anti-mouse albumin	Bethyl Laboratories	Cat #A90–234A
Biotinylated anti-rabbit igG (H+L)	Vector Laboratories	Cat # BA-1000
Rabbit anti-Streptavidin 555	Life technologies	Cat #S32355
Alexa Fluor 405 donkey anti-rabbit	Life Technologies	Cat# A48258
Alexa Fluor 488 donkey anti-goat	Life Technologies	Cat #A11055
Alexa Fluor 488 donkey anti-rabbit	Life Technologies	Cat #A21206
Alexa Fluor 488 donkey anti-mouse	Life Technologies	Cat #A21202
Alexa Fluor 555 donkey anti-rabbit	Life Technologies	Cat #A31572
Alexa Fluor 555 donkey anti-mouse	Life Technologies	Cat #A31570
Alexa Fluor 555 donkey anti-rat	Life Technologies	Cat #A78945
Alexa Fluor 555 donkey anti-goat	Life Technologies	Cat #A21432
Alexa Fluor 647 donkey anti-chicken	Life Technologies	Cat #A78952
Biological Samples
hCSF1 isolated brain RNA	This paper.	N/A
hFIRE-PBS isolated brain RNA	This paper.	N/A
hFIRE-GFP isolated brain RNA	This paper.	N/A
Chemicals, Peptides, and Recombinant Proteins
mTESR-E8	StemCell Technologies	Cat #05990
DMEM/F-12, HEPES, no phenol red	Thermo Fisher Scientific	Cat #11039021
1-Thioglycerol (Monothioglycerol)	Sigma Aldrich	Cat #M1753
B27 Supplement	Gibco	Cat #17504044
N2 Supplement	Gibco	Cat #17502048
Non-essential Amino Acids (NEAA)	Thermo Fisher Scientific	Cat #11140050
GlutaMax Supplement	Thermo Fisher Scientific	Cat #35050061
Insulin-Transferrin-Selenium (ITS-G) (100X)	Thermo Fisher Scientific	Cat #41400045
Human Insulin (25mg)	Sigma Aldrich	Cat #I2643
Recombinant Human IL-34	Peprotech	Cat #200–34
Recombinant Human M-CSF	Peprotech	Cat #300–25
Recombinant Human TGF-β1	Peprotech	Cat #100–21
T-PER	Thermo Fisher Scientific	Cat# 78510
Protease and Phosphatase Inhibitor Cocktail (100X)	Thermo Fischer Scientific	Cat #P178446
ReLeSR	StemCell Technologies	Cat #05872
CloneR	StemCell Technologies	Cat #05889
Matrigel, Basement Membrane Matrix Growth Factor Reduced, Phenol Red-Free	Corning	Cat #356231
Thiazovivin	StemCell Technologies	Cat #72252
BamBanker	Wako	Cat #NC9582225
DPBS, no Ca2+, no MG2+	Thermo Fisher Scientific	Cat #14190144
Paraformaldehyde	Sigma-Aldrich	Cat #P6148–500G
Alt-R® S.p. HiFi Cas9 Nuclease V3	IDTDNA	Cat #1081061
Trans-activating CRISPR RNA (tracrRNA)	IDTDNA	Cat #1072534
Nuclease Free Duplex Buffer	IDTDNA	Cat #11010301
Taq PCR Master Mix	Thermo Fisher Scientific	Cat #K0172
Sodium Azide	Sigma Aldrich	Cat #S2002
Phosphate Buffer Saline (PBS)	Sigma Aldrich	Cat #P44017
Triton X-100	Thermo Fisher Scientific	Cat #9002–93-1
Donkey Serum	Thermo Fisher Scientific	Cat #NC9624464
Isoflurane	Patterson Veterinary	Cat #14043070406
Puralube Vet ointment	Dechra	Cat #17033–211-38
Lidocaine 2%	Medline	Cat #17478–711-31
Acetaminophen (Mapap)	Major	Cat #0904–7014-16
Vetbond	3M	Cat #084–1469SB
DAPI Fluoromount-G	SouthernBiotech	Cat #0100–20
Fluoromount-G	SouthernBiotech	Cat #0100–01
RIS-647	Biovnc	Cat #BV500101
Fluoromyelin Red	Invitrogen	Cat #F34652
Critical Commercial Assays
StemDiff Hematopoietic Kit (includes Basal Media, Supplement A, and Supplement B)	Stem Cell Technologies	Cat #05310
Human Stem Cell Nucleofector™ Kit 2	Lonza	Cat #VPH-5022
Extracta DNA prep for PCR	QuantBio	Cat #95091
MycoAlert PLUS Mycoplasma Detection Kit	Lonza	Cat #LT07–710
Pierce BCA Protein Assay Kit	Thermo Fisher Scientific	Cat #23227
Mouse MCP1/CCL2 ELISA Kit	Abcam	Cat #Ab208979
Glia Fibrillary Acidic Protein ELISA Kit	Antibodies-online	Cat #ABIN6574131
Synaptic Vesicle Glycoprotein 2A ELISA Kit	Abbexa	Cat #abx548000
Neuron Specific Enolase ELISA Kit	Abcam	Cat #ab233626
Postsynaptic density protein 95 ELISA Kit	MSD	Cat #K15OQND
Myelin Basic Protein ELISA Kit	CUSABIO	Cat #CSB-E08285m
V-PLEX Plus Mouse Cytokine 19-Plex Kit	MSD	Cat #K15255G-1
CytoTune-iPS 2.0 Sendai Reprogramming Kit	Thermo Fisher Scientific	Cat #A16517
StemDiff Trilineage Differentiation Kit	Stem Cell Technologies	Cat #05230
RNeasy Mini Kit	Qiagen	Cat #74106
Deposited Data
hCSF1 brain bulk RNA-seq	This paper.	GEO: GSE253623
hFIRE-PBS brain bulk RNA-seq	This paper.	GEO: GSE253623
hFIRE-GFP brain bulk RNA-seq	This paper.	GEO: GSE253623
Experimental Models: Cell Lines
Human: WTC11-mEGFP iPSC line	Coriell Institute	Cat #AICS-0036–006
Human: ADRC75 iPSC line	UCI-ADRC Stem Cell Core	https://stemcells.minduci.edu/
Human: ADRC76 iPSC line	UCI-ADRC Stem Cell Core	https://stemcells.minduci.edu/
Human: ADRC77 iPSC line	UCI-ADRC Stem Cell Core	https://stemcells.minduci.edu/
Human: L786S-Het fibroblast line	Mayo Clinic Jacksonville	N/A
Human: L786S-Het iPSC line	This paper.	N/A
Human: L786L-corrected iPSC line	This paper.	N/A
Experimental Models: Organisms/Strains
Mouse: human M-CSF knockin mouse (Rag2−/−, il2rγ−/−, CSF1h/h)	The Jackson Laboratory	JAX #017708
Mouse: FIRE-knockout mouse (Csf1rΔFIRE/ΔFIRE)	Rojo et al., 201924	N/A
Mouse: Humanized FIRE (hFIRE) mouse (Rag2−/−, il2rγ−/−, CSF1h/h, Csf1rΔFIRE/ΔFIRE)	This paper.	N/A
Oligonucleotides
Primer: mCSF1R-WT_F; 5’-GGTGCCAGCAATGTGTTTCC-3’	This paper.	N/A
Primer: mCSF1R-FIRE_F; 5’-GCGGTTGTAGGAAACCCTGA-3’	This paper.	N/A
Primer: mCSF1R_R; 5’-CACTCCTACCACTGGGCATC-3’	This paper.	N/A
crRNA: CSF1R-L786S5′-CGUAACGUGCUGUCGACCAA-3′	This paper.	N/A
ssODN: CSF1R-L786L-correction; 5′-GTGCTTTCCCTCAGTGCATCCACCGGGAC	This paper.	N/A
GTGGCAGCGCGTAACGTGCTGTTGACCAA TGGTCATGTGGCCAAGATTGGGGACTTCG GGCTGGCTAGGGACATCATGAAT-3′
Software and algorithms
SnapGene (Version 4.3.11)	Dotmatics	http://www.snapgene.com/
Incucyte 2020B	Incucyte	https://www.sartorius.com/
Imaris	Bitplane	https://imaris.oxinst.com/products/imaris-for-core-facilities
Fluoview FV31S-DT software (Version 2.6)	Olympus	https://www.olympus-lifescience.com/en/downloads/detailiframe/?0[downloads][id]=847252002
Graphpad Prism (Version 9)	Graphpad	https://www.graphpad.com/
byPrim.py	Song et al., 202295	https://github.com/rhart604/optimized
DESeq2 R package (Version 3.18)	Love et al., 201496	https://bioconductor.org/packages/release/bioc/html/DESeq2.html
Kallisto (Version 0.46.1)	Bray et al., 201697	https://pachterlab.github.io/kallisto/
tximport	Soneson et al., 201598	http://bioconductor.org/packages/tximport
BBMAP (Version 39.04)	Brian Bushnell/Joint Genome Institute	https://sourceforge.net/projects/bbmap/
R (Version 4.3.2)	The R Foundation	https://www.r-project.org
FastQC (Version 0.12.0)	Babraham Bioinformatics	https://www.bioinformatics.babraham.ac.uk/projects/fastqc/
GSEA Desktop Application (Version 4.3.2)	Subramanian et al., 200599	https://software.broadinstitute.org/gsea/index.jsp
Original Code Chadarevian_2024_Leukoencephalopathy	This Paper.	Github https://doi.org/10.5281/zenodo.11211703
Other
Olympus FV3000 confocal microscope	Olympus	https://www.olympus-lifescience.com/en/laser-scanning/fv3000/
Agilent Bioanalyzer 2100	Agilent	Cat #G2939BA
10uL Gastight syringe, Model 1701 RN	Hamilton	Cat #7653–01
30-gauge, Small hub RN needle, 12mm, Pt:4, 45° tip	Hamilton	Cat #7803–07
0.2mL PCR 8-tube Strips	USA Scientific	Cat #1402–3900
Nuclease-Free Water	Ambion	Cat #AM9937
6-well plates	Corning	Cat #3516
48-well plates	Corning	Cat #3548
12-well plates	Corning	Cat #3512
96-well plates	Corning	Cat #3596
15 mL Centrifuge Tube	Corning	Cat #430791
50 mL Centrifuge Tube	Corning	Cat #430829
Cryotube vial	Thermo Fisher Scientific	Cat #374081
Parafilm	Sigma Aldrich	Cat #HS234526B
Sliding Microtome	Leica	Cat #SM2010 R
Microscope slides	VWR	Cat #16004–392
Coverslips	Thermo Fisher Scientific	Cat #12–548-5M
100 mL Reservoir	Thermo Fisher Scientific	Cat #07–200-130
Lonza Amaxa Nucelofector II	Thermo Fisher Scientific	Cat #13458999

Highlights.

Adult hFIRE mice progressively develop diverse ALSP-related neuropathologies

Human microglial transplantation prevents development of neuropathologies in hFIRE mice

CRISPR-correction of ALSP-patient-derived iPSCs rescues microglial deficiencies

Transplantation of CRISPR-corrected patient iMG can reverse pre-existing neuropathology