SGK1-mediated deficits in microglial phagocytosis drive pathological progression in amyotrophic lateral sclerosis

Abstract

Alterations in microglial function and transcriptomic profiles are major pathological hallmarks of amyotrophic lateral sclerosis (ALS). However, the dynamics and regulatory mechanisms underlying microglial phagocytic activity during disease progression remain unclear. In this study, we observed stage-dependent alterations in microglial phagocytic activity during disease progression in SOD1^G93A mice. Single-cell RNA sequencing suggested that this change was associated with a reduced abundance of microglial subpopulations enriched for phagocytosis-related pathways. Transcriptomic analysis identified serum- and glucocorticoid-regulated kinase 1 (SGK1) as a potential mediator of this process. Notably, SGK1 knockout in SOD1^G93A mice was associated with improved microglial clearance of myelin debris and reduced aberrant engulfment of neuronal material after disease onset. Our results further showed that, after disease onset, the accumulation of myelin debris and apoptotic neurons induced SGK1 upregulation in microglia from SOD1^G93A mice. Mechanistically, SGK1 appeared to promote lipid accumulation in microglia by suppressing lipophagy, thereby impairing the ability of microglia to clear cellular debris. Moreover, pharmacological inhibition of SGK1 with GSK650394 attenuated motor deficits and prolonged survival in SOD1^G93A mice. Together, our findings provide evidence for a previously unrecognized role of SGK1 in regulating microglial phagocytosis in ALS models and support SGK1 as a potential therapeutic target in SOD1 mutation-associated ALS models.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12974-026-03925-w.

Background

Amyotrophic lateral sclerosis (ALS) is a relentlessly progressive neurodegenerative disorder affecting the motor system. It is characterized by degeneration of spinal motor neurons, resulting in progressive muscle weakness that ultimately culminates in paralysis, and most patients succumb to respiratory failure within 2–5 years after diagnosis [1]. Approximately 90–95% of ALS cases are sporadic, while 5–10% are inherited, resulting from familial genetic factors [2]. Approximately 20% of familial and 3% of sporadic cases of ALS are associated with mutations in the gene encoding superoxide dismutase-1 (SOD1) [3], but the molecular mechanisms linking this mutation to motor neuron degeneration remain unclear.

Microglia are the resident immune cells of the brain that contribute to synaptogenesis, synaptic maintenance and immune surveillance [4, 5]. Since their initial characterization, multiple distinct microglial states have been described. Accumulating evidence positions microglia at the center of central nervous system (CNS) pathology: they shape the inflammatory milieu through secretion of modulatory cytokines and remove apoptotic neurons via phagocytic and restorative mechanisms, thereby both responding to and actively influencing disease course [6–8]. The notion of a binary, dynamic polarization between pro‑inflammatory and anti‑inflammatory phenotypes is now regarded as overly simplistic and insufficient to capture the multifaceted roles of microglia in CNS disorders, including ALS [9]. Analyses of post‑mortem tissue from ALS patients and studies in animal models indicate that microglial dysregulation in ALS is a complex, multi‑stage process involving numerous signaling pathways, during which microglia display pronounced functional plasticity and dynamic phenotypic transitions [10, 11]. Early studies in the SOD1^G93A mouse model of ALS suggested that microglia undergo a phenotypic shift from an initially neuroprotective state to a late‑stage pro‑inflammatory, neurotoxic phenotype [12]. However, the simplistic M1/M2 polarization paradigm is now considered insufficient to capture the dynamic and heterogeneous nature of microglial responses. Advances in epigenomics, transcriptomics, proteomics, and metabolomics have substantially deepened our understanding of microglial function, morphology, and ultrastructure. Recent spatial transcriptomics analyses in SOD1^G93A mice implicate altered phagocytic programs as a potential contributor to microglial functional transitions: phagocytosis‑related genes are significantly upregulated prior to symptom onset [13]. However, the phagocytic ability of microglia declines markedly at the advanced stage of disease [14]. The drivers of these temporally distinct changes in phagocytic function remain poorly understood, and the precise molecular mechanisms underlying this dynamic regulation have yet to be elucidated.

Serum- and glucocorticoid-inducible kinase 1 (SGK1) is a serine/threonine kinase belonging to the AGC family of kinases. It is involved in the regulation of various physiological and pathological processes, including cell survival, ion transport, and signal transduction [15, 16]. Previous studies have mainly focused on the intimate relationship between SGK1 and microglia-mediated neuroinflammation or cell death. For example, in neurodegenerative diseases like Alzheimer's disease, SGK1 can modulate microglial inflammatory responses, thereby promoting the deposition of Aβ and tau and exacerbating brain injury [17]. Similarly, in other CNS diseases, such as multiple sclerosis [18, 19], stroke [20, 21], and Parkinson’s disease [22, 23], the regulatory role of SGK1 in microglia-released inflammatory mediators has been thoroughly discussed. Although studies have demonstrated that SGK1 expression is upregulated in ALS mouse models [24], the mechanisms underlying this change and its functional consequences remain unclear, and its potential role in regulating microglial function has received little attention.

In this study, we investigated the mechanisms underlying the functional transformation of microglia in SOD1^G93A mice and the role of SGK1 in this process. First, we examined microglial state and phagocytic capacity across different stages of disease progression in SOD1^G93A mice. Single-cell RNA sequencing analysis revealed that the proportion of cell clusters enriched for phagocytosis-related pathways decreased markedly after disease onset. Subsequent transcriptomic analysis implicated SGK1 as a key regulator of this change. We then knocked out SGK1 in SOD1^G93A mice and observed an enhancement in the phagocytic ability of microglia. In BV2 cells transfected with SOD1^G93A, SGK1 appeared to impair phagocytic capacity by suppressing lipophagy. Finally, pharmacological inhibition of SGK1 with GSK650394 improved motor function and survival in SOD1^G93A mice. Together, these findings support SGK1 as a potential therapeutic target in SOD1-associated ALS and suggest that SGK1 is an important regulator of microglial phagocytosis under certain pathological conditions.

Results

Microglial state changed during the symptomatic stage of SOD1^G93A mice

ALS is a progressive neurodegenerative disorder characterized by degeneration of motor neurons and functional impairment [25]. Characterizing the dynamic changes of microglia during disease progression may provide insight into the regulatory mechanisms governing microglial function and may help identify potential therapeutic targets. First, we assessed motor neuron numbers in the spinal cord of SOD1^G93A mice along the course of disease. Beginning at P90 (early symptomatic stage), the number of ChAT-positive motor neurons in SOD1^G93A mice was significantly reduced and continued to decline until P150 (end stage), compared with age-matched wild type (WT) mice (Fig. 1a, b). Next, to investigate the dynamics of microglia during neuronal loss, we applied qRT-PCR to assess the expression levels of homeostatic microglia (HOM) markers and neurodegenerative microglia (MGnD) markers [26, 27]. Consistent with prior research [27], the heatmap showed that, compared to WT mice, the expression of selected HOM markers was suppressed from the pre-symptomatic to the end stage. In contrast, the mRNA levels of MGnD markers such as Clec7a and Spp1 began to increase at the early stage. At the advanced stage, four MGnD markers (Apoe, Clec7a, Itgax, and Spp1) were significantly upregulated and remained elevated until the end stage of disease progression (Fig. 1c).

Fig. 1.

Dynamic changes in microglial state during disease progression in SOD1^G93A mice. a-b Representative immunofluorescent micrographs (a) and quantification (b) of ChAT-positive motor neurons in WT and SOD1^G93A mice at different ages. Motor neurons were labeled with ChAT. n = 4 mice per condition. c Heat map showing the relative mRNA expression levels of HOM and MGnD markers. Darker colors indicate higher fold changes. *P < 0.05, **P < 0.01, ***P < 0.001 versus the WT group. n = 4 mice per condition. d Representative immunofluorescent micrographs and 3D reconstructions of IBA1 and CD68 staining in WT and SOD1^G93A mice at different ages. e–h Quantification derived from panel (d), including microglial volume measured by IBA1 staining (e), endpoints (f) and maximum process length (g) determined by Sholl analysis, and CD68-positive volume within IBA1-positive microglia (h). CD68 was used as a lysosomal marker. In (e-h), n = 4 mice per condition. Scale bar = 10 μm. Data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant

We subsequently assessed microglial morphology and function to further characterize reactive adaptations to the evolving microenvironment during disease progression in SOD1^G93A mice. Previous studies have shown that changes in CD68 expression are closely associated with dynamic alterations of microglia, including migration, morphological remodeling, and functional conversion, during the progression of certain neurodegenerative conditions such as Alzheimer's disease and depression [28, 29]. Therefore, we performed immunofluorescence staining for CD68 and the microglial marker IBA1 to analyze the adaptive changes of microglia during disease progression in SOD1^G93A mice (Fig. 1d). Immunofluorescence analysis showed that the volume of microglia in SOD1^G93A mice was significantly increased at the symptomatic stages (P90–P150) compared with WT mice (Fig. 1e). To further characterize microglial morphology, we performed Sholl analysis and found that, beginning at the pre-symptomatic stage (P60), microglia in SOD1^G93A mice already exhibited reduced process length and a decreased number of endpoints, and these alterations persisted throughout disease progression (Fig. 1f, g). In parallel, although no significant enlargement of total microglial volume was observed at P60, the volume of CD68 within microglia began to increase significantly at this stage and remained elevated during the symptomatic stages (Fig. 1h). These findings suggest that microglia in SOD1^G93A mice undergo early morphological remodeling and increased phagocytic activation prior to overt motor neuron loss.

Along with neuronal loss and changes in microglial state, SOD1^G93A mice exhibited an inflammatory phenotype during the symptomatic phase. qRT-PCR results showed that Cox2 expression increased at the early stage compared with WT mice. At the advanced stage, the mRNA levels of Cox2, Il1β, Tnfα and iNOS increased and persisted until the end stage of ALS (Figure S1). Altogether, these results showed that during disease progression in SOD1^G93A mice, microglia progressively shift from a homeostatic to a reactive phenotype, characterized by altered phagocytic capacity and enhanced neuroinflammation.

Microglial phagocytic ability declined in the symptomatic stages of SOD1^G93A mice

Previous studies have demonstrated a state shift in microglial phenotypes during ALS progression [30]. Our results further demonstrated that, with disease progression, microglia showed a progressive increase in CD68 volume, indicating dynamic alterations of microglial phagocytic capacity in SOD1^G93A mice. Consistent with prior research, ALS is a neurodegenerative disorder characterized in part by pathological changes associated with demyelination [31]. In the spinal cord of SOD1^G93A mice, the expression of myelin basic protein (MBP) began to decrease at the pre-symptomatic stage (Fig. 2c). Microglia, the primary phagocytic cells of the CNS, are essential for clearing myelin debris, which is a prerequisite for effective remyelination in neurodegenerative diseases [32]. We employed co-staining with MBP, CD68, and IBA1 to assess the dynamics of myelin debris processing within the degradative compartment of microglia during the disease progression of SOD1^G93A mice (Fig. 2a, b). Immunostaining analysis revealed that the volume of CD68-positive MBP within microglia in SOD1^G93A mice was significantly increased at the pre-symptomatic stage compared with WT controls, indicating enhanced phagocytic engagement at early stages. Notably, CD68-positive MBP volume peaked at P60 and then progressively declined with disease progression, although remaining higher than that in age-matched WT mice (Fig. 2b). These findings suggested that myelin debris accumulation and active phagocytic processing occurred at early stages, whereas the degradative capacity of microglia declined during disease progression. Krasemann et al. [27] reported that, in neurodegenerative disease, phagocytosis of apoptotic neurons can drive phenotypic transitions in microglia. Accordingly, we assessed the capacity of microglia in SOD1^G93A mice to phagocytose apoptotic neurons during disease progression using NeuN and IBA1 co‑localization analysis (Fig. 2d, e). Importantly, at the pre-symptomatic stage (P60), microglia in SOD1^G93A mice did not exhibit a significant change in neuronal engulfment compared to control mice, indicating that neuronal material is not a major phagocytic substrate at this early stage. Immunofluorescence analysis revealed that the phagocytic activity of microglia toward neurons peaked at P90 and then declined markedly between P120 and P150. Notably, in contrast to the increased myelin content within microglia at P60, the absence of enhanced neuronal engulfment at this stage suggested that early microglial phagocytic activity was preferentially directed toward myelin rather than neuronal material. These findings showed that, although lysosomal volume within microglia increased after disease onset, the intracellular content of myelin and neuronal material was reduced. Together, these observations suggested that microglial phagocytic capacity might be compromised during disease progression in SOD1^G93A mice.

Fig. 2.

Microglial phagocytic ability declined during the symptomatic stages of SOD1^G93A mice. a Representative micrographs and 3D reconstructions showing microglial phagocytic processing of myelin debris, assessed by IBA1, CD68, and MBP staining in WT and SOD1^G93A mice at different ages. Scale bars: 10 μm in representative micrographs and 5 μm in 3D reconstructions. b Quantification of MBP volume within CD68-positive compartments of IBA1-positive microglia in WT and SOD1^G93A mice at different ages. Phagocytic processing was quantified by normalizing the volume of MBP within CD68-positive structures to the total volume of IBA1-positive microglia. n = 4 mice per condition. c Representative immunoblots and quantification of MBP and β-ACTIN in spinal cord tissue. Protein expression levels are shown relative to the age-matched WT group. n = 7 mice per condition. d Representative micrographs and 3D reconstructions of IBA1 and NeuN staining in WT and SOD1^G93A mice at different ages. Scale bar = 10 μm. e Quantification of NeuN volume within IBA1-positive microglia. n = 4 mice per condition. f Unsupervised clustering of cell populations from the published single-cell RNA-seq dataset. g Subclustering of microglia isolated from WT and SOD1^G93A mice. Cells are colored according to their microglial subcluster identity. h Relative abundance of microglial subclusters under different conditions: SOD1^G93A-P60 (left), SOD1^G93A-P90 (right). i Gene set enrichment analysis (GSEA) scores of the indicated pathways in MG clusters. Data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant

To gain deeper insights into the mechanisms behind the increased lysosomal volume and the diminished phagocytic function in microglia, we conducted an analysis of the published single-cell RNA sequencing data of SOD1^G93A mice from the GEO database GSE206330 (Fig. 2f-i). The reactive microglia were divided into five clusters (Fig. 2g). Subclustering of microglial subpopulations together with pathway enrichment analysis revealed that, compared with the pre-symptomatic stage, the MG3 subcluster, which was enriched for phagocytosis-related pathways, showed a marked decrease in relative abundance after disease onset. Meanwhile, MG2, characterized by low expression of phagocytosis-associated genes, became the dominant microglial subpopulation post-onset (Fig. 2h, i). These findings suggested that alterations in phagocytic capacity in SOD1^G93A mice before and after disease onset led to impaired clearance of cellular debris in the surrounding environment, which might represent a critical factor influencing the progression of the disease.

SGK1 expression was upregulated in spinal cord microglia during the symptomatic stages of SOD1^G93A mice

Previous studies have indicated that microglia undergo functional reprogramming during ALS progression. To explore the molecular events driving early disease pathology, we reanalyzed the transcriptomic data of microglia from early-stage SOD1^G93A mice in the public dataset GSE43366. In addition to MGnD markers such as Apoe, Clec7a, Spp1, bulk RNA-seq analysis revealed that Sgk1 was upregulated at the early stage of disease, which drew our attention (Fig. 3a). SGK1, a serine/threonine kinase, is implicated in various pathological conditions, including Alzheimer's disease, ischemic stroke, and multiple sclerosis [33]. However, its role in ALS remains largely unexplored. To investigate this, we examined the temporal expression pattern of SGK1 in the spinal cord of SOD1^G93A mice. The protein expression of SGK1 was significantly upregulated from P90 to P150 compared to age-matched WT mice, with peak expression observed at P150 (Fig. 3b, c). Immunofluorescence staining revealed significant colocalization of SGK1 with IBA1 in the symptomatic stages of the disease (Fig. 3d, e). While SGK1 was expressed in a subset of microglia, the Pearson's correlation coefficient between SGK1 and IBA1 in SOD1^G93A mice showed a significant increase starting from P90 compared with age-matched WT controls, and this elevated correlation persisted through the end stage of disease progression (P150). Together, these results showed that microglia in the spinal cord of SOD1^G93A mice exhibited elevated expression of SGK1 at the early symptomatic stage, and this elevation persisted until the end stage of the disease. This sustained elevation suggested a potential role for SGK1 in the pathophysiological processes of ALS.

Fig. 3.

SGK1 expression was upregulated in spinal cord microglia during the symptomatic stages of SOD1^G93A mice. a Volcano plot of differentially expressed genes (DEGs) in microglia from SOD1^G93A and WT mice at the early stage of disease progression. Thresholds: fold change > 1.5 and FDR < 0.05. Blue dots indicate downregulated genes, red dots indicate upregulated genes, and grey dots indicate genes without significant differential expression. b Representative immunoblots of SGK1 and β-ACTIN in spinal cord tissue. c Quantification of SGK1 protein expression levels normalized to β-ACTIN. Expression levels are shown relative to the age-matched WT group. n = 7 mice per condition. d Representative immunofluorescent micrographs of SGK1 and IBA1 staining of WT and SOD1^G93A mice at different time points. e Pearson’s correlation coefficient between SGK1 and IBA1. n = 3 mice per condition. Scale bar = 10 μm. Data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant

Sgk1 knockout ameliorated impaired phagocytosis in SOD1^G93A microglia

To further determine whether SGK1 influences the phagocytic capacity of microglia in SOD1^G93A mice, we generated SOD1^G93A mice with SGK1 knockout. We crossed SGK1^KO mice with SOD1^G93A mice to generate SOD1^G93A; SGK1^KO mice (Fig. 4a). As shown in Figs. 4b and 4c, in the spinal cord of SGK1^KO mice, the protein and mRNA expression levels of SGK1 were significantly reduced compared with those in WT mice. Furthermore, to evaluate the knockout efficiency of SGK1 specifically within microglia, we analyzed its colocalization with the microglial marker IBA1. We found that SGK1 deletion resulted in a significant decrease in the Pearson's correlation coefficient between SGK1 and IBA1 in the spinal cords of SOD1^G93A mice (Figure S2a). To further define the stage-specific role of SGK1, we analyzed the effects of SGK1 deletion at pre-symptomatic (P60) and early symptomatic (P90) stages. At P60, SGK1 deletion did not significantly affect synaptic density (PSD95 and SYP) (Figure S2b, c), motor neuron number (Figure S2d), or microglial morphology (Figure S2f, h) in SOD1^G93A mice. In addition, neuronal engulfment by microglia (Figure S2h), as well as CD68-positive lysosomal volume and myelin debris clearance (Figure S2j, l), remained unchanged following SGK1 deletion. These results suggested that SGK1 does not play a major role in regulating microglial phagocytic function before its pathological upregulation.

Fig. 4.

Sgk1 knockout ameliorated impaired phagocytosis in SOD1^G93A microglia. a An experimental schematic showing the generation of SOD1^G93A; SGK1^KO mice by crossing SOD1^G93A and SGK1^KO lines. b Representative immunoblots and quantification of SGK1 and β-ACTIN in WT mice and SGK1^KO mice. n = 4 mice for the WT group and n = 3 mice for the SGK1^KO group. c Relative mRNA expression of Sgk1 in WT and SGK1^KO mice measured by qRT-PCR. n = 5 mice for the WT group and n = 4 mice for the SGK1^KO group. d Representative micrographs and Sholl analysis of IBA1-labeled microglia in SOD1^G93A mice and SOD1^G93A; SGK1^KO mice at P120. Sholl curves were generated from 30 microglia sampled from 4 mice per condition to illustrate the distribution of microglial morphological complexity. Quantification of endpoints and process length was performed at the animal level after averaging analyzed cells within each mouse, with each data point representing one mouse. n = 4 mice per condition for endpoint and process length quantification. e Representative micrographs and 3D reconstructions showing microglial phagocytic processing of myelin debris, assessed by IBA1, CD68, and MBP staining in WT, SOD1^G93A, SGK1^KO, and SOD1^G93A; SGK1^KO mice at P120. f Representative micrographs and 3D reconstructions showing microglial engulfment of neuronal material, assessed by IBA1 and NeuN staining. g Representative micrographs and 3D reconstructions showing lysosomal damage in microglia, assessed by IBA1, CD68, and GAL-3 staining. h-j Quantification of microglial volume (h), lysosomal content (i), and MBP volume within microglia (j) from panel (e). k Quantification of neuronal material within microglia from panel (f). l Quantification of GAL-3 volume within CD68-positive compartments from panel (g). m Quantification of motor neuron number from the corresponding ChAT staining. In (h-m), n = 4 mice per condition, and each data point represents the mean value from one mouse after averaging technical measurements from multiple cells or fields of view. Data are presented as mean ± SEM. Scale bar = 10 μm. *P < 0.05, **P < 0.01, ***P < 0.001

In contrast, at P90, when SGK1 expression was significantly increased and motor neuron degeneration became evident, SGK1 deletion markedly altered microglial function. Crucially, SGK1 deficiency significantly attenuated motor neuron loss (Figure S2e). This neuroprotection was accompanied by a marked phenotypic shift in microglia; specifically, SGK1 knockout reduced microglial cell volume and substantially diminished aberrant neuronal engulfment (Figure S2g, i). Concurrently, SGK1-deficient microglia exhibited a decreased lysosomal burden and enhanced efficacy in myelin debris clearance (Figure S2k, m). These findings indicated that the effects of SGK1 on microglial phagocytic activity are stage-dependent and become prominent during the early symptomatic phase.

We further analyzed the effects of SGK1 on microglia at the advanced stage (P120) of SOD1^G93A mice. Morphological analysis of microglia revealed that knockout of SGK1 increased Sholl intersections, endpoints, and process length of microglia in SOD1^G93A mice (Fig. 4d). We next investigated whether SGK1 influenced the state of microglia in the spinal cord of SOD1^G93A mice (Fig. 4e). We found that knocking out SGK1 in SOD1^G93A mice significantly reduced the volume of microglia (Fig. 4h) and intracellular volume of CD68 (Fig. 4i), while enhancing microglial phagocytic capacity for myelin debris (Fig. 4j). Nevertheless, genetic ablation of SGK1 markedly attenuated microglial engulfment of neurons (Fig. 4k), likely as a consequence of the reduced neuronal loss associated with SGK1 deletion (Fig. 4m). As previously reported by Monsorno et al. [34], alterations in the volume of CD68‑positive structures may indicate impaired lysosomal degradation. Recent studies have highlighted lysosomal membrane permeabilization (LMP) as a key mechanism driving lysosomal dysfunction. Galectin‑3 (GAL‑3), a member of the galectin family that binds β‑galactoside-containing carbohydrates, has been identified as a robust marker of LMP [35, 36]. Immunofluorescence triple staining for IBA1, CD68, and GAL-3 revealed that the volume of GAL-3 within CD68-positive compartments increased in microglia of SOD1^G93A mice during the symptomatic stage (Figure S3), while knockout of SGK1 in SOD1^G93A mice significantly ameliorated lysosomal damage (Fig. 4g, l). These results suggested that SGK1 could impair lysosomal function in microglia of SOD1^G93A mice, thereby affecting the clearance of myelin debris and apoptotic neurons.

SOD1^G93A microglia exhibited increased lipid droplet accumulation and transition toward an MGnD phenotype

To further elucidate the mechanism underlying the accumulation of cellular debris in microglia of SOD1^G93A mice, subsequent studies were performed in BV2 cells following transfection with the SOD1^G93A plasmid. To simulate phagocytic environments relevant to disease progression, treatments included either myelin debris alone or a combination with apoptotic N2a cells. BV2 cells were co-cultured with these fluorescently labeled debris to assess their phagocytic activity (Fig. 5a). Although BV2 cells cannot fully recapitulate the features of primary microglia, they exhibit similar phagocytic machinery [37]. Microglial phagocytosis is generally described as a three-step process: "find-me", "eat-me", and "digest-me" [38]. In the present study, we initially employed an in vitro engulfment assay to assess the phagocytic capacity of BV2 microglia. We first examined the phagocytic response of BV2 cells expressing SOD1^G93A (G93A-BV2) after exposure to myelin debris alone, apoptotic N2a cells alone, or a combination of both. The results showed that, compared with SOD1^WT-BV2 (WT-BV2) cells, exposure of SOD1^G93A-BV2 (G93A-BV2) cells to myelin debris alone increased their engulfment capacity, whereas no significant difference was observed under stimulation with apoptotic N2a cells alone. However, co-stimulation with myelin debris and apoptotic neurons markedly impaired the engulfment capacity of G93A-BV2 cells (Fig. 5b-d). Similar effects were observed in primary microglia isolated from SOD1^G93A mice: treatment with myelin debris alone enhanced their phagocytic capacity, whereas apoptotic N2a cells alone had no significant effect. However, co-stimulation with both substrates markedly diminished their phagocytic efficiency (Figure S5).

Fig. 5.

SOD1^G93A microglia exhibited increased lipid droplet accumulation and MGnD-associated changes. a Experimental schematic showing the evaluation of the effects of myelin debris and apoptotic neurons on BV2 microglia. Created with BioRender.com. b Representative micrographs showing the phagocytic activity of BV2 cells following 3 h incubation with CFSE-labeled myelin debris alone, Red CMTPX-labeled apoptotic N2a cells alone, or a combination of both. c–d Quantification of BV2 microglial phagocytosis under the indicated treatment conditions. c Phagocytosis of myelin debris. d Phagocytosis of apoptotic N2a cells. e Representative micrographs showing lipid droplets in BV2 cells treated with myelin debris alone, apoptotic N2a cells alone, or a combination of both. f Representative micrographs showing lipid droplets in BV2 cells following co-treatment with myelin debris and apoptotic N2a cells and subsequent EBSS treatment. g Representative micrographs showing the phagocytic activity of BV2 cells following 3 h incubation with CFSE-labeled myelin debris and Red CMTPX-labeled apoptotic N2a cells after EBSS treatment. h Percentage of BODIPY-positive cells following treatment with myelin debris alone, apoptotic N2a cells alone, or their combination (quantified from panel e). i Percentage of BODIPY-positive cells in BV2 cells following co-treatment with myelin debris and apoptotic N2a cells and subsequent EBSS treatment (quantified from panel f). j Quantification of phagocytosis of myelin debris and apoptotic N2a cells by BV2 cells after EBSS treatment. k Relative mRNA levels of Apoe, Itgax, Spp1, and Clec7a in BV2 cells measured by qRT-PCR. For the quantitative cell-based imaging assays in panels (c), (d), and (h–j), 3 independent biological replicates were analyzed for panels (c), (d), (h), and (i), and 4 independent biological replicates were analyzed for panel (j). For each replicate, 3–5 fields of view (FOVs) were analyzed per condition, and the averaged value was used as a single data point for statistical analysis. Each dot represents one independent biological replicate. For qRT-PCR analysis in panel (k), n = 3 independent experiments. Scale bar = 40 μm. Data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant

We next investigated whether the impaired phagocytic function of SOD1^G93A microglia was associated with defective intracellular degradation pathways. Previous studies have shown that when microglia engulf large amounts of lipid-rich myelin debris or apoptotic cells, lipid overload occurs, leading to the accumulation of intracellular lipid droplets. Impaired lipid degradation has been identified as a key mechanism contributing to lipid droplet accumulation [39–41]. Gouna et al. reported that lipid clearance is primarily mediated by lipophagy, a selective autophagy pathway that degrades lipid droplets via lysosomal hydrolysis [41]. Based on this mechanism, we utilized BODIPY staining to assess lipid accumulation in G93A-BV2 cells following phagocytosis of myelin debris or apoptotic neurons. These results revealed no significant differences in lipid droplet accumulation among the indicated treatment conditions in WT-BV2 cells. In G93A-BV2 cells, myelin debris alone or apoptotic N2a cells alone did not induce lipid droplet formation within 3 h, whereas combined treatment with myelin debris and apoptotic neurons significantly increased lipid droplet content (Fig. 5e, h). To further explore the relationship between lipid accumulation and phagocytosis, G93A-BV2 cells were subjected to starvation with Earle’s Balanced Salt Solution (EBSS) to promote lipid droplet degradation (Fig. 5f, i), followed by phagocytosis assays (Fig. 5g). These results showed that EBSS treatment significantly improved the phagocytic or processing capacity of G93A-BV2 cells, resulting in increased intracellular signals of myelin debris and apoptotic neurons at the assay endpoint (Fig. 5j). These findings suggest that the SOD1 mutation primarily impairs post-engulfment processing and degradative capacity, thereby compromising overall phagocytic efficiency under co-stimulation. To rule out technical artifacts, we first performed reciprocal labeling experiments, swapping CMTPX and CFSE between myelin debris and apoptotic N2a cells. Phagocytic outcomes remained consistent, confirming no fluorophore-dependent bias (Figure S4a-d). Time-course analysis further showed that at 1 h, no significant differences in the uptake of myelin debris or apoptotic neurons were observed between WT and G93A-BV2 cells, suggesting that initial engulfment is largely preserved (Figure S4e-h). To further assess cargo processing within acidic degradative compartments, we performed pHrodo-based assays. These results revealed that while pHrodo signals increased over time in WT cells, this increase was markedly blunted in G93A microglia under co-stimulation with myelin debris and apoptotic neurons, particularly at 3 h. A similar trend was observed for apoptotic N2a cargo (Figure S4i-l). These findings indicated that the phagocytic defect in G93A microglia is not due to reduced uptake, but rather reflects impaired post-engulfment trafficking and/or maturation of cargo into acidic degradative compartments. qRT-PCR results showed that lipid stress induced by myelin debris and apoptotic neurons also upregulated the mRNA expression of MGnD markers (Apoe, Itgax, Spp1, and Clec7a) in G93A-BV2 cells. Notably, Spp1 was also modestly induced by myelin debris alone, suggesting that it may represent an early or lipid-sensitive MGnD-associated gene (Fig. 5k). One possible explanation is that distinct MGnD-associated genes have different activation thresholds, with Spp1 being relatively more sensitive to lipid-associated stress. This interpretation is consistent with previous studies linking lipid metabolism and myelin-associated pathology to disease-related microglial state transitions, including Spp1-positive microglial subsets [27, 40, 42]. Accordingly, broader induction of MGnD-associated markers appears to require combined pathological stimuli. Furthermore, lipid-accumulating G93A-BV2 cells exhibited increased ROS levels and elevated inflammatory cytokine expression (Figure S6a-e). Together, co-stimulation with myelin debris and apoptotic neurons provoked lipid metabolic dysregulation in SOD1-mutant microglia, exacerbated neuroinflammation, and impaired microglial phagocytic function.

SGK1 promoted lipid accumulation in SOD1^G93A microglia

As described above, microglial phagocytic capacity was reduced during the symptomatic stage in SOD1^G93A mice, concomitant with an increase in SGK1 expression. Notably, deletion of SGK1 significantly enhanced the ability of microglia to clear cellular debris. These findings led us to further examine whether the impaired phagocytic capacity of microglia is directly associated with SGK1, and to investigate the underlying mechanisms by which SGK1 regulates phagocytosis under pathogenic conditions. First, we assessed the expression of SGK1 by Western blotting. Our results demonstrated that SGK1 protein expression in G93A-BV2 cells was significantly upregulated only upon co-stimulation with both myelin debris and apoptotic neurons, whereas treatment with either stimulus alone failed to induce any significant changes in SGK1 levels (Fig. 6a). We further explored the role of SGK1 in microglial dysfunction. We employed the inhibitor of SGK1, GSK650394, to treat WT-BV2 and G93A-BV2 cells for 48 h, followed by engulfment assays (Fig. 6b). These results showed that GSK650394 significantly enhanced the phagocytic capacity of G93A-BV2 cells against myelin debris and apoptotic neurons (Fig. 6c). BODIPY staining showed that GSK650394 treatment significantly reduced lipid droplet accumulation in G93A-BV2 cells, although not to basal levels (Fig. 6d, e). This partial reduction suggested that alleviation of lipid accumulation contributes to, but might not fully account for, the improvement in phagocytic capacity. Additionally, GSK650394 significantly attenuated the expression of MGnD markers, such as Apoe, Itgax and Spp1, which were upregulated by the stimulation of apoptotic cells and debris. However, it did not significantly reduce the expression of Clec7a (Fig. 6f). As lipid accumulation declined, GSK650394 attenuated cell-debris induced ROS levels and the expression of inflammatory factors (Tnfα, Ifnγ, and Cox2) in G93A-BV2 cells (Figure S6f-h).

Fig. 6.

SGK1 promoted lipid accumulation and phagocytic dysfunction in SOD1^G93A microglia. a Representative immunoblots and quantification of SGK1 and β-ACTIN in BV2 cells treated with myelin debris alone, apoptotic N2a cells alone, or a combination of both. Expression levels were normalized to β-ACTIN and are shown relative to the WT group. n = 3 independent experiments. b Representative micrographs showing phagocytosis of myelin debris and apoptotic N2a cells by WT-BV2 and G93A-BV2 cells with or without GSK650394 treatment. c Quantification of phagocytosis of myelin debris and apoptotic N2a cells by BV2 cells after GSK650394 treatment. Phagocytic activity was quantified as the engulfed signal normalized to total IBA1 fluorescence intensity to account for potential variations in cell density. n = 3 independent experiments. d Representative micrographs showing BODIPY staining of lipid droplets in BV2 cells treated with myelin debris and apoptotic N2a cells in the presence or absence of GSK650394 treatment. e Quantification of the percentage of BODIPY-positive cells from panel (d). Lipid accumulation was quantified as the percentage of BODIPY-positive cells. n = 3 independent experiments. f Relative mRNA expression levels of Apoe, Clec7a, Itgax, and Spp1 in BV2 cells measured by qRT-PCR. n = 3 independent experiments per group. g Representative micrographs showing TFEB and Hoechst staining in BV2 cells. h Quantification of TFEB nuclear localization. n = 4 independent experiments. i-j Representative immunoblots (i) and quantification (j) of lipophagy- and lysosome-related proteins in BV2 cells treated with myelin debris and apoptotic N2a cells in the presence or absence of GSK650394 treatment. β-ACTIN was used as the internal reference. Expression levels are shown relative to the WT group. n = 3–4 independent experiments per group. For cell-based imaging assays in (b–h), experiments were performed with 3–4 independent biological replicates. For each replicate, 3–5 fields of view (FOVs) were analyzed per condition, and the averaged value from these FOVs was used as a single data point for statistical analysis. Thus, each dot represents one independent biological replicate. Scale bar = 40 μm. Data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ns: not significant

Next, we examined changes in the expression of lipophagy-related molecules to investigate how SGK1 influenced this phagocytic process. In addition to its involvement in lysosome biogenesis and autophagy, TFEB has recently been reported to be closely associated with the process of lipophagy [43, 44]. Dephosphorylation of TFEB promotes its nuclear translocation, where it activates transcription of target genes involved in lipid metabolism and lysosomal function [45]. For assessing the nuclear distribution of TFEB, we performed immunostaining against TFEB and Hoechst. We found that stimulation with myelin debris and apoptotic N2a cells significantly reduced nuclear translocation of TFEB in G93A-BV2 cells, which could be significantly enhanced by GSK650394 treatment (Fig. 6g, h). Western blotting was employed to investigate the impact of GSK650394 on molecules involved in lipophagy (Fig. 6i, j). Compared with WT-BV2 cells, co-treatment with myelin debris and apoptotic neurons increased the expression of P62, LAMP1, and GAL-3 in G93A-BV2 cells, while also increasing the ratios of P-mTOR/mTOR and P-TFEB/TFEB. After GSK650394 treatment, the phosphorylation levels of mTOR and TFEB were markedly reduced, accompanied by an increase in LC3 II protein levels and a decrease in the expression of P62, LAMP1, and GAL-3. These results suggest that SGK1 may suppress TFEB activation, possibly through enhanced mTOR signaling, thereby disrupting lipophagy, blocking the degradation of intracellular myelin debris and apoptotic neurons, and consequently leading to lipid accumulation in G93A-BV2 cells.

To further investigate whether SGK1 expression is associated with intracellular lipid metabolic status, we examined the effect of enhancing lipophagy on SGK1 levels. G93A-BV2 cells were subjected to EBSS treatment to promote lipid droplet degradation. Notably, EBSS treatment significantly reduced SGK1 expression in G93A-BV2 cells (Figure S7). These results suggested that SGK1 expression is responsive to intracellular lipid metabolism and can be modulated by promoting lipid clearance.

SGK1 inhibitor GSK650394 ameliorated disease-related phenotypes in SOD1^G93A mice

Based on the above findings, we next examined whether pharmacological inhibition of SGK1 could ameliorate disease-related phenotypes in SOD1 mutation-associated ALS models. Behavioral tests were applied to verify whether the SGK1 inhibitor slowed disease progression (Fig. 7a-d). As anticipated, SOD1^G93A mice performed comparably to healthy controls in Grip Strength (Fig. 7a), Wire Suspension (Fig. 7b), and Rotarod test (Fig. 7c) up to around P90, after which their performance deteriorated sharply. Interestingly, SOD1^G93A mice treated with GSK650394 performed significantly better than SOD1^G93A mice in all tasks. The Kaplan–Meier survival curve analysis showed that GSK650394 administration extended the lifespan of SOD1^G93A mice by 14.5 days (Fig. 7d). In addition to enhancing motor performance in mice, GSK650394 increased the number of motor neurons in the spinal cord of SOD1^G93A mice (Fig. 7e, f) and elevated the protein level of MBP (Fig. 7g). qRT-PCR analysis revealed that the mRNA levels of MGnD-associated markers (Fig. 7h-k) and inflammatory factors (Figure S8) were significantly reduced in the SOD1^G93A mice treated with GSK650394. Together, these experiments demonstrated that the SGK1 inhibitor GSK650394 improved long-term neurological function and extended survival in SOD1^G93A mice.

Fig. 7.

The SGK1 inhibitor GSK650394 attenuated motor deficits and pathological changes in SOD1^G93A mice. a-d Quantification of Grip Strength (a), Wire Suspension (b), Rotarod test (c), and Kaplan–Meier survival analysis (d) in WT and SOD1^G93A mice treated with or without GSK650394 at P120. n = 8 mice per group for behavioral assessment. *P < 0.05, **P < 0.01, ***P < 0.001, SOD1^G93A + Vehicle versus the WT + Vehicle group; #P < 0.05, ##P < 0.01, SOD1^G93A + GSK650394 versus the SOD1^G93A + Vehicle group. e–f Representative immunofluorescent micrographs (e) and quantification (f) of ChAT-positive motor neurons in WT and SOD1^G93A mice treated with or without GSK650394 at P120. n = 4 mice per condition. Each data point represents one mouse; values from multiple sections or fields of view within the same mouse were averaged before statistical analysis. Scale bar = 10 μm. g Representative immunoblots and quantification of MBP and β-ACTIN in spinal cord tissue. n = 4 mice per group. h–k Relative mRNA expression levels of Apoe (h), Itgax (i), Clec7a (j), and Spp1 (k) measured by qRT-PCR. n = 4 mice per group. Data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001

Discussion

In this study, we provide in vitro and in vivo evidence for temporally distinct alterations in microglial state and function during ALS progression in the SOD1^G93A model. Our results suggest that the accumulation of apoptotic neurons and myelin debris imposes a substantial phagocytic burden on microglia, leading to disrupted intracellular lipid metabolism and impaired phagocytic function. We further found that targeting SGK1 by genetic ablation or pharmacological inhibition promotes lipophagy and restores microglial phagocytic function, thereby improving debris clearance and attenuating disease progression.

Here, we analyzed the spatiotemporal evolution of microglial states during disease progression in SOD1^G93A mice on the basis of transcriptional, morphological, and functional changes. Transcriptional analysis indicated that microglial homeostatic genes (e.g., Egr1 and Mertk) were downregulated before disease onset. Following motor neuron loss after symptom onset, MGnD-associated markers were markedly upregulated in microglia. MGnD is primarily localized to regions of motor neuron degeneration, suggesting that its activation may be driven by danger-associated signals released from degenerating neurons [46]. Krasemann et al. further showed that, across several neurodegenerative diseases, including multiple sclerosis and Alzheimer’s disease, microglia similarly transition from a homeostatic state to an MGnD phenotype in response to neurodegeneration [27]. These observations support a conserved role for the transition to the MGnD state in neurodegenerative pathology. Notably, the temporal dissociation between homeostatic collapse and MGnD emergence suggests that these phenotypic transitions are governed by distinct regulatory mechanisms. While post-symptomatic MGnD activation is likely influenced by neuronal degeneration, the mechanisms that initiate suppression of homeostatic genes prior to disease onset remain incompletely understood. Recent studies have shown that myelin debris can activate a pro-phagocytic transcriptional program in microglia through lipid-sensing pathways, thereby suppressing homeostatic signatures [32, 47, 48]. This suggests that early disruption of microglial homeostasis in SOD1^G93A mice may be associated with local accumulation of myelin debris, although the precise spatiotemporal dynamics and causal contribution of this interaction require further investigation.

Previous studies have suggested that microglial phagocytic programs are dynamically altered during ALS progression. Spatial transcriptomic analysis by Maniatis et al. [13] showed early activation of phagocytosis-associated molecules in SOD1^G93A mice, whereas Barreto-Núñez et al. [14] reported impaired microglial phagocytic function at later disease stages. These observations suggest a close relationship between microglial phagocytosis and ALS progression, but the temporal evolution of this process remains insufficiently characterized. In the present study, we further found that microglial phagocytic function in SOD1^G93A mice followed a biphasic pattern, with early phagocytic engagement followed by progressive functional impairment at later stages. At P60, we observed significant myelin abnormalities accompanied by increased microglia–myelin colocalization, suggesting enhanced phagocytic engagement with myelin debris before overt symptom onset. Notably, this increase occurred before pronounced transcriptional reprogramming, motor neuron loss, and overt neuroinflammatory responses, indicating that it might represent an early event rather than a downstream consequence of advanced pathology. Consistently, Sholl analysis showed that microglia in SOD1^G93A mice had already undergone morphological remodeling at P60, as reflected by reduced process length and decreased branching complexity. Because myelin disruption has been reported before symptom onset in SOD1^G93A mice [49], and myelin debris has been shown to influence microglial transcriptional and functional states [50], these early changes may reflect an adaptive response to alterations in the local microenvironment. Moreover, it has been reported that ALS-associated mutations may intrinsically prime microglia toward functional dysregulation [11]. Together, these observations suggest that early microglial alterations in SOD1^G93A mice may be associated with both myelin debris accumulation and intrinsic effects of ALS-associated mutations. During disease progression, apoptotic neuronal material and myelin debris progressively accumulated in the spinal cord of SOD1^G93A mice. Although microglial CD68-positive lysosomal volume continued to increase, the amount of myelin and neuronal material detected within microglial degradative compartments gradually declined. These results suggest that the increased lysosomal volume did not reflect more efficient debris clearance, but rather may indicate an impaired ability of microglia to process accumulated substrates. Thus, microglial phagocytic function in SOD1^G93A mice appears to undergo a biphasic change, characterized by early phagocytic engagement followed by reduced debris-processing capacity during disease progression. This later-stage impairment occurred in parallel with increased inflammatory gene expression, suggesting that phagocytic dysfunction and neuroinflammatory activation may coexist during symptomatic disease progression.

Reanalysis of published single-cell RNA-seq data revealed a striking shift in microglial subclusters during disease progression in SOD1^G93A mice. The MG3 cluster, which was enriched for phagocytosis-associated pathways, decreased after symptom onset, whereas the MG2 cluster, characterized by relatively lower expression of phagocytosis-associated genes, became predominant. These transcriptomic changes indicate a reduction in the relative abundance of microglial populations associated with phagocytic programs. Together with our observations of increased debris burden, enlarged CD68-positive lysosomal compartments, lipid accumulation, and elevated inflammatory gene expression, this shift is consistent with the possibility that multiple pathological stresses collectively contribute to the progressive impairment of microglial debris clearance. Prior studies have described enhanced microglial phagocytic responses in debris-rich neurodegenerative environments [51]. Our data further highlight a progressive disruption of intracellular lipid homeostasis following sustained pathological stimulation. Notably, the lipid droplet accumulation observed in SOD1^G93A microglia mirrors pathological features reported in demyelinating diseases [39, 41]. However, in our in vitro model, marked lipid droplet accumulation required co-stimulation with myelin debris and apoptotic neurons. Treatment with myelin debris alone failed to induce marked lipid droplet accumulation in G93A-BV2 cells, whereas co-stimulation with apoptotic neurons induced a pronounced lipid-rich transformation, accompanied by inflammatory cytokine upregulation and impaired debris clearance. Through engulfment assays, time-course analyses, and pHrodo-based experiments, we found that the phagocytic defect in SOD1^G93A microglia was not primarily due to reduced initial uptake, but rather to impaired post-engulfment cargo processing under conditions of increased degradative burden. In addition, primary microglia derived from SOD1^G93A mice accumulated excessive lipid droplets after internalizing lipid-rich cellular fragments, likely as a result of dysregulated lipophagy, which in turn was associated with reduced phagocytic activity. Notably, starvation-induced enhancement of lipophagy substantially restored microglial phagocytic function and promoted intracellular lipid droplet clearance. Together, these findings indicate that the accumulation of apoptotic neurons and myelin debris disrupts intracellular lipid homeostasis in SOD1^G93A microglia, resulting in excessive lipid accumulation and impaired lipophagy. This metabolic disturbance compromises the clearance of cellular debris and diminishes the pool of phagocytically competent microglia, thereby potentially contributing to disease progression. Importantly, targeting SGK1 to modulate lipophagy restored microglial lipid homeostasis and rescued phagocytic function, highlighting SGK1-regulated lipophagy as a potential therapeutic target in ALS.

Transcriptomic analysis of microglia at early disease stages suggested that SGK1 may contribute to microglial dysfunction in SOD1^G93A mice. Previous studies on SGK1 have primarily focused on central nervous system disorders, including stroke, multiple sclerosis, Parkinson’s disease, and Alzheimer’s disease [52], whereas its role in ALS remains largely unexplored. Our data provide evidence that SGK1 contributes to microglial dysfunction in ALS and further suggest a stage-dependent role for SGK1 in regulating microglial phagocytic function during disease progression. Importantly, our temporal analysis provides insight into the cellular context in which SGK1 exerts its effects. At the pre-symptomatic stage (P60), SGK1 deletion did not significantly affect motor neuron number, synaptic density, or the examined microglial morphological and engulfment-related parameters in SOD1^G93A mice. In contrast, at the early symptomatic stage (P90), when SGK1 expression was elevated and motor neuron degeneration had become detectable, SGK1 deletion was associated with improved motor neuron preservation, reduced microglial activation, decreased neuronal material within microglia, and enhanced myelin debris clearance. These findings suggest that the association between SGK1 deletion and improved microglial debris-handling capacity becomes more evident after disease onset. However, because constitutive SGK1 knockout was used in this study, contributions from neuronal SGK1 or SGK1 in other cell types cannot be excluded.

In G93A-BV2 cells, EBSS-induced enhancement of lipophagy was associated with reduced SGK1 expression, suggesting that SGK1 levels may be responsive to intracellular lipid metabolic status. Our data suggest that SGK1 influences lipophagy-related pathways, thereby contributing to impaired phagocytic function in SOD1^G93A microglia. Previous studies have suggested that SGK1 stabilizes lipid droplet formation through its α-helical hydrophobic side chain [53]. In the present study, SGK1 inhibition was associated with reduced phosphorylation of mTOR and TFEB, increased LC3II levels, and decreased expression of P62, LAMP1, and GAL-3. These changes support a link between SGK1 and pathways related to mTOR–TFEB–LC3 signaling, lipophagy, and lysosomal homeostasis. In parallel, knockout of SGK1 in SOD1^G93A mice alleviated lysosomal damage in microglia and enhanced the clearance of myelin debris and apoptotic neurons, consistent with our in vitro findings. Finally, treatment of SOD1^G93A mice with the SGK1 inhibitor GSK650394 significantly improved motor performance and prolonged survival. Together, these findings support SGK1 as a potential therapeutic target in SOD1-associated ALS pathology.

Limitations

These findings should be interpreted in light of several limitations. First, the BV2 cell line cannot fully recapitulate the dynamic responses of primary microglia in vivo, and our use of the SOD1^G93A model requires further validation in other ALS subtypes. Second, although targeting SGK1 alleviated lipid accumulation and microglial phagocytic deficits, the rescue was partial, suggesting that additional pathological processes, such as persistent lysosomal dysfunction or broader metabolic dysregulation, also contribute to disease progression. Third, because our interventions involved constitutive SGK1 knockout and systemic GSK650394 administration, off-target effects and contributions from other cell types cannot be excluded; definitive evidence for a cell-autonomous role will require microglia-specific genetic approaches. Finally, reanalysis of publicly available ALS patient transcriptomic datasets did not reveal consistent SGK1 upregulation (data not shown) [54], suggesting that its dysregulation may be context-dependent rather than universal across ALS. Because most available datasets lack genetic stratification, particularly with respect to SOD1 mutation status, future studies in genetically defined patient cohorts will be needed to clarify the broader clinical relevance of SGK1.

Conclusion

In conclusion, our data indicate that, during disease progression in the SOD1^G93A ALS model, the accumulation of apoptotic neurons and myelin debris may impose an excessive degradative burden on microglia, thereby perturbing intracellular lipid metabolism and impairing microglial phagocytic function. Targeting SGK1 by genetic ablation or pharmacological inhibition enhanced lipophagy-associated pathways and improved microglial phagocytic capacity, leading to more efficient debris clearance and attenuation of disease progression in SOD1^G93A mice. At the mechanistic level, SGK1 inhibition was associated with activation of the mTOR–TFEB–LC3 signaling axis, enhanced lipophagy, reduced inflammation, and restored the ability of microglia to clear myelin and apoptotic neuronal debris. As a result, SGK1 inhibition improved motor performance and prolonged survival in SOD1^G93A mice. These findings suggest that targeting SGK1 to improve microglial lipophagy may represent a potential therapeutic strategy in SOD1 mutation-associated ALS pathology. Overall, our study identifies a close association between SGK1 activation and impaired microglial phagocytic function in ALS models, providing new insight into the role of SGK1 in SOD1-associated ALS pathology.

Methods

Animals

All experimental protocols and animal handling procedures were approved by the Animal Research Ethics Committee of China Pharmaceutical University. All the animals were raised and bred under pathogen-free conditions in the Animal Centre of China Pharmaceutical University. The 6- to 8-week-old male C57BL/6J mice and neonatal C57BL/6J mice (within 3 days after birth) were purchased from Vital River Laboratory Animal Technology Co., Ltd (Beijing, China). SOD1^G93A mice (B6.Cg-Tg (SOD1^G93A)1Gur/J) were purchased from The Jackson Laboratory (stock no: #004435). SOD1^G93A mice were bred on a C57BL/6J background to produce SOD1^G93A and wild-type (WT) mice. Progeny was genotyped by standard PCR following the Jackson Lab protocol (www.jax.org). SGK1^KO mice (KOCMP-20393-Sgk1-B6J-VA) were purchased from Cyagen Bioscience (Suzhou) Inc. SOD1^G93A; SGK1^KO mice were generated by crossing SOD1^G93A mice with SGK1^KO mice. To verify the knockout of SGK1, genotyping was performed on SGK1^KO and SOD1^G93A; SGK1^KO mice using the following genotyping primers (forward: 5’-GAAACAGAGAAGGATGGGCCTG-3’ and reverse: 5’-TGGTTTCAACTGGAGAGGCTTG-3’). The absence of SGK1 product at 593 bp confirmed knockout of SGK1 in these animals. Mice were bred and kept at 22–25 ℃ with a 12 h light/dark cycle, and provided adequate food and water. All animal experiments were conducted in accordance with the guidelines of the Animal Care and Use Committee of the pharmaceutical animal experimental center of China Pharmaceutical University (2019-06-006).

Drug administration

For SGK1 inhibition, SOD1^G93A and WT mice were intraperitoneally injected with 1 mg/kg GSK650394 (MedChemExpress, USA) every three days, starting from postnatal day 60 (P60) until the disease endpoint. For immunohistochemistry and Western blotting analysis, animals were treated with GSK650394 at P60 and sacrificed at P120.

Behavior tests

Disease progression and survival analysis

Survival analysis of SOD1^G93A mice was determined by the inability of the animal to right itself within 15–30 s if laid on either side. This is a widely accepted and published endpoint for life span studies in SOD1^G93A mice and guarantees that euthanasia occurs prior to the mice being unable to reach food or water [55].

Grip strength

Grip Strength Meter (Yiyan, YLS-13A) was used to measure maximal limb muscle grip strength [56]. In this test, grip strength was measured in mice starting from P60. Weekly measurements were conducted using a grip strength testing system. Mice were trained to grasp horizontal rods, which were then pulled until the mice released their grip. The maximum force exerted was recorded. Each mouse was tested three times, with a 10-min interval between trials, and the highest value was taken. The onset of disease was defined as a reduction in grip strength to 75% (early stage), 50% (advanced stage), and 25% (end stage) of the data collected at P60.

Rotarod test

Rotarod performance was measured with a Rotamex (Zhongshidichuang, ZS-RDM-DS) using an accelerating rod protocol [57]. The initial speed of the rod was set to 2 revolutions per minute (rpm) with a linear acceleration to 20 rpm over 300 s. The latency to fall was measured in three trials. The mean latency was registered and used as an indicator of rotarod performance.

Wire suspension test

The wire suspension test was used to evaluate the muscle strength and coordination of mice [58]. In this experiment, mice were suspended upside down on a 1-m-long wire rope with a diameter of 0.5 cm, suspended at a height of 30 cm above the ground. A foam pad was placed below to prevent injury. The time until the mouse falls was recorded, with a maximum duration of 240 s. Each mouse was tested three times, with a 10 min interval between trials, and the average time was recorded.

Immunofluorescence

Mice were anesthetized with isoflurane and perfused with 0.9% saline, followed by cold 4% paraformaldehyde (PFA). The spinal cord was removed from the vertebral column and postfixed in 4% PFA overnight at 4 ℃. After being dehydrated in 20% and 30% sucrose, the spinal cord was sliced into 20 μm thick sections using a Leica-1950 cryostat (Leica Instruments, Germany). The sections were blocked by 10% goat serum and 0.1% Triton at room temperature for 1 h and incubated in primary antibody at 4 ℃ overnight. The primary antibodies used were as follows: guinea pig anti-IBA1 (1:500, Oasis, OB-PGP049), rat anti-CD68 (1:50, Abcam, ab53444), rat anti-MBP (1:100, Millipore, mab386), rabbit anti-NeuN (1:200, Abcam, ab177487), rabbit anti-TFEB (1:200, HUABIO, HA723419), rabbit anti-SGK1 (1:50, Abcepta, AP3924a), rat anti-ChAT (1:100, Oasis, OB-PRT115), and mouse anti-Galectin-3 (1:200, Proteintech, 60207–1-Ig). After rinsing three times with phosphate buffer saline (PBS), the sections were incubated in Alexa Fluor 488 conjugated goat anti-rabbit IgG (1:500, Invitrogen, A-11008), Alexa Fluor 405 conjugated goat anti-mouse IgG (1:500, Invitrogen, A-31553), YSFluor™ 488 Goat Anti-Guinea Pig IgG (H + L) (1:500, Yeasen, 33406ES60), YSFluor™ 488 F(ab')2 Fragment Goat Anti-Rat IgG (H + L) (1:500, Yeasen, 33316ES60), or Alexa Fluor 647 conjugated goat anti-mouse IgG (1:500, Invitrogen, A-21235) at room temperature for 1h. The sections were rinsed and mounted on microscope slides with mounting media (Beyotime, China). All fluorescent images were acquired using Olympus FV3000 confocal microscope (Olympus, Japan) and processed by ImageJ software (USA). Control and experimental samples were imaged under identical settings. The mean fluorescence intensity (MFI) and the area of positive signals were analyzed by ImageJ software (USA).

For microglia morphologic analysis, the sections were immunostained with IBA1 and fluorescent secondary antibody. The microglial traces were generated and analyzed using the Sholl analysis plugin in ImageJ. For in vivo microglial morphology analyses, multiple microglia were sampled from each animal, and cell-level measurements were averaged within each mouse before group-level statistical analysis. Thus, the animal, rather than the individual microglial cell, was used as the biological replicate for statistical testing.

For phagocytosis analysis, the sections were stained with IBA1, MBP, NeuN, or CD68 depending on the purpose of the experiment. The z-stacks were acquired using a 100 × oil objective with a step size of 0.75 μm. The acquired images were 3D reconstituted by Imaris software (UK). Engulfed material, CD68-positive compartments, and microglial signals were detected and quantified using the colocalization, surface, and spots functions of Imaris as appropriate. For in vivo engulfment-related analyses, values from multiple cells or fields of view within the same animal were averaged before statistical analysis, and each mouse was treated as one biological replicate.

Western blotting

The spinal cord tissues or collected cells were homogenized on ice in RIPA buffer (Beyotime, P0013B) supplemented with protease inhibitor (Beyotime, P0013B) and 1 mM phenylmethanesulfonyl fluoride (Beyotime, P1081). Extracts were cleared by centrifugation (14,000 × g, 20 min, 4 ℃). Samples (20–30 μg) were analyzed by SDS-PAGE (sodium dodecyl sulfate–polyacrylamide gel electrophoresis) and immunoblotted. PVDF (polyvinylidene difluoride) membranes were incubated (overnight, 4 ℃) with primary antibodies as follows: rabbit anti-SGK1 (1:50, Abcepta, AP3924a), rabbit anti-MBP (1:1000, Oasis, OB-PRB130), rabbit anti-P62 (1:1000, Abcam, ab109012), mouse anti-LC3 (1:1000, CST, 83506), rabbit anti-LAMP1 (1:1000, Proteintech, 33243–1-AP), mouse anti-Galectin-3 (1:200, Proteintech, 60,207–1-Ig), rabbit anti-TFEB (1:5000, HUABIO, HA723419), rabbit anti-Phospho-TFEB (Ser142) (1:1000, Abmart, PC22337), rabbit anti-mTOR (1:5000, HUABIO, ET1608-5), and mouse anti-Phospho-mTOR (S2448) (1:1000, HUABIO, HA600094). The corresponding secondary antibodies used were HRP (horseradish peroxidase)-conjugated donkey anti-rabbit IgG (Jackson, 711–035–152) and HRP-conjugated donkey anti-mouse IgG (Abclonal, AS033), followed by detection with ECL Select (Vazyme, China). Band intensities were quantified with ImageLab 6.1 software (USA).

qRT‐PCR

The total RNA of the spinal cord tissues or collected cells was extracted using TRIzol reagent (Vazyme, China), and quantified by NanoDrop 2000 (Thermo Fisher Scientific, USA). RNA was then reversely transcribed by HiScript 1st Strand cDNA Synthesis Kit (Vazyme, China) according to the manufacturer’s protocol. Quantitative PCR was performed using SYBR Green Master Mix (Vazyme, China) on an ABI7000 real-time PCR system (Applied Biosystems, USA). Ct values were normalized to β-actin. The primer sequences used for qRT-PCR are provided in Table S1, titled “Primer sequences used for qRT-PCR”. The revised table includes only the primers used for the results presented in the current manuscript.

Microglial cell line (BV2 cells) culture and plasmid transfection

The BV2 murine microglial cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (KeyGEN Biotech, KGL1206) supplemented with 10% fetal bovine serum (FBS) (Royacel, RY-F32-01), 1% penicillin–streptomycin (Beyotime, C0222). The medium was changed every day. BV2 cells (1 × 10⁵ cells per well) were transfected with 1 μg plasmid DNA of SOD1^WT (ADDGENE, #26397) or SOD1^G93A (ADDGENE, #26401) using the jetPRIME transfection reagent (Polyplus, 101000046). BV2 cells were initially seeded at equal densities. Differences in apparent cell density at the time of imaging may reflect variation in cell survival and/or growth after transfection. To minimize potential bias arising from these differences, phagocytosis-related fluorescence signals were normalized to total IBA1 fluorescence intensity within each field of view.

Isolation and culture of primary mouse microglia cells

Spinal cords from P0-2 SOD1^G93A mice and age-matched non-transgenic controls were dissected and enzymatically dissociated with trypsin (Beyotime, C0201) in a 37 ℃ incubator for 10 min. Trypsinization was stopped by adding DMEM/F12 supplemented with 10% FBS and mechanical disaggregation by repeated pipetting. The resulting extract was passed through a 40 μm cell-strainer to eliminate tissue debris and then centrifuged. The pellet was resuspended in culture medium (DMEM/F12 (KeyGEN, KGM12400) supplemented with 10% FBS, 1% penicillin/streptomycin). Cells were plated, and the culture medium was removed after 48 h and then replaced every 72 h.

Measurement of intracellular ROS

The level of ROS was measured by confocal microscopy. For the fluorescence microscopy assay, BV2 cells were incubated with DCFH-DA (10 μM, Beyotime, S0033S) for 30 min at 37 ℃. The stained cells were then washed with PBS and observed using a fluorescence microscope (Olympus, Japan).

Myelin isolation and CFSE labeling

Myelin isolation and CFSE labeling were performed as previously described [59]. Myelin was isolated from brains of C57BL/6J mice. Brains were dissected and homogenized in 0.32 M sucrose of Tris–Cl solution using a Dounce homogenizer. The homogenate was layered onto a 0.83 M sucrose of Tris–Cl solution in a 50 ml polypropylene centrifuge tube. The mixture was centrifuged at 100,000 g at 4 ℃ for 45 min. The crude myelin was collected at the interface of the sucrose gradient. The pellet was washed in Tris–Cl buffer and resuspended in sterile PBS to a concentration of 100 mg/mL. The myelin was aliquoted into 1.5 mL centrifugation tubes and centrifuged at 22,000 g at 4 °C for 10 min. The supernatant was removed, and the pellet was weighed and stored at −80 ℃ in sterile PBS at 100 mg/mL. Crude myelin was resuspended in 50 µM CFSE in PBS and incubated at room temperature for 30 min. The CFSE-labeled myelin was washed and resuspended in PBS to 100 mg/mL for storage.

Apoptosis induction

Neuro-2a (N2a) cells were cultured in DMEM supplemented with 10% FBS and 1% penicillin–streptomycin at 37 °C in a humidified atmosphere containing 5% CO₂. For experimental procedures, cells were harvested by trypsinization and centrifuged at a density of 10⁶ cells per tube. The cell pellet was resuspended in pre-warmed CellTrace™ Red CMTPX staining solution (Yeasen, 40717ES50) and incubated at 37 °C for 20 min in the dark. After staining, cells were washed twice with pre-warmed PBS and resuspended in 1 μM staurosporine at 10⁶ cells/mL for 3 h. Subsequently, cells were centrifuged at 400 g for 10 min, washed once with PBS, and resuspended in microglial cell culture medium for downstream experiments [60].

Phagocytosis and degradation assays

Microglial phagocytic uptake and intracellular degradation were assessed using complementary engulfment assays and pHrodo-based assays. BV2 cells or primary microglia were seeded onto coverslips placed in 24-well plates at a density of 1 × 10⁵ cells per well and cultured overnight to allow cell attachment. Cells were then incubated with CFSE-labeled myelin debris (5 μg/mL) and/or Red-CMTPX-labeled apoptotic N2a cells at a BV2:N2a ratio of 1:2 for 3 h or 1 h at 37 °C in a humidified incubator containing 5% CO₂. For reciprocal labeling control experiments, myelin debris and apoptotic N2a cells were alternatively labeled with Red-CMTPX and CFSE, respectively. After incubation, cells were washed three times with pre-warmed PBS to remove non-engulfed extracellular materials and then processed for immunofluorescence staining or immediate confocal imaging.

For degradation assays, myelin debris or apoptotic N2a cells were labeled with pHrodo Green dye (Thermo Fisher Scientific) according to the manufacturer’s instructions. Briefly, substrates were incubated with pHrodo Green dye at room temperature for 30 min in the dark, followed by washing to remove excess dye. Microglia were then incubated with pHrodo-labeled substrates under the same conditions as described above. Because pHrodo fluorescence increases in acidic environments, fluorescence intensity reflects successful internalization and lysosomal processing of the engulfed material.

Images were acquired using an Olympus FV3000 confocal microscope under identical acquisition settings across experimental groups. Each condition included 3 or 4 independent biological replicates, depending on the assay. Within each biological replicate, 3–5 fields of view (FOVs) were randomly selected for image quantification, resulting in approximately 10–20 FOVs per group in total. FOVs were used as technical sampling units, and values from multiple FOVs within the same biological replicate were averaged before statistical analysis. Thus, statistical analyses were performed using biological replicates rather than individual FOVs.

For engulfment assays, phagocytosis was quantified as the total intracellular fluorescence intensity of CFSE or Red-CMTPX normalized to total IBA1 fluorescence intensity within each FOV, reflecting the average phagocytic capacity of microglia in that field. For pHrodo-based assays, phagocytic degradation was quantified as the mean fluorescence intensity (MFI) of pHrodo signal normalized to total IBA1 fluorescence intensity within each FOV. For lipid droplet analysis, the percentage of BODIPY-positive cells was calculated for each FOV. For TFEB localization analysis, nuclear translocation was assessed by calculating the Pearson’s correlation coefficient between TFEB and nuclear staining (Hoechst) within each FOV. All image analyses were performed using ImageJ and Imaris software.

Single-cell RNA sequencing (scRNA-seq) data analysis

Single-cell RNA-seq data were obtained from the published GEO database (GSE206330, normalized mouse transcriptome matrix) and analyzed in R (v4.4.0) using Seurat (v5). Doublets were removed with DoubletFinder, and cells with > 10% mitochondrial genes or 200–3000 detected genes were excluded. Normalization and variable gene selection (n = 2000) were performed with FindVariableFeatures. PCA and UMAP were used for dimensionality reduction and visualization. Cell types were annotated with CellMarker 2.0 (e.g., Tmem119, Cx3cr1 for microglia). Following the code provided by Kloosterman et al., functional enrichment analysis of KEGG pathways was performed using clusterProfiler (v4.6.2), and significance was evaluated by hypergeometric tests [61].

Statistical analysis

Data were analyzed by GraphPad Prism software, version 10.0 (USA) and presented as mean ± SEM. Individual comparisons were assessed using an unpaired Student's t-test. One-way or two-way ANOVA tests followed by Tukey’s or Bonferroni’s post hoc test were used for multiple comparisons. Values with P < 0.05 were considered statistically significant.

Abbreviations

SGK1

Serum- and glucocorticoid-regulated kinase 1

ALS

Amyotrophic Lateral Sclerosis

SOD1

Superoxide dismutase-1

WT

Wild type

CNS

Central nervous system

HOM

Homeostatic microglia

MGnD

Neurodegenerative microglia

FOV

Field of view

mTOR

Mammalian target of rapamycin

TFEB

Transcription factor EB

LC3

Microtubule-associated protein 1 light chain 3

Ifnγ

Interferon-γ

Apoe

Apolipoprotein E

Cox2

Cyclooxygenase-2

Tnfα

Tumor necrosis factor-α

Il1β

Interleukin-1 beta

iNOS

Inducible nitric oxide synthase

IBA1

Ionized calcium-binding adapter molecule 1

Authors’ contributions

Hong Liao and Meijun He conceived this project and designed the experiments. Meijun He performed most experiments and interpreted the data. Mengqiu Hu, Ruiqi Liu and Haoran Hu performed some of the behavioral experiments. Xiayi Shi, Heng Wang and Yujie Tong aided in the data analysis. Meijun He and Chaoran Wu wrote the manuscript. Hong Liao revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by grants of National Natural Science Foundation of China, 82372577 and 82073831.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

All animal experiments were conducted in accordance with the guidelines of the Animal Care and Use Committee of the Pharmaceutical Animal Experimental center of China Pharmaceutical University (2019–06-006).

Competing interests

The authors declare no competing interests.