Spatial transcriptomics analysis uncovers ER stress in MANF-deficient Purkinje cells underlying alcohol-induced cerebellar neurodegeneration in mice

Abstract

Alcohol use disorders (AUD) is one of the most prevalent mental disorders in the United States affecting more than 10% of the adult population. Cerebellar atrophy and Purkinje cell (PC) degeneration are frequently observed in patients with AUD. Alcohol can cause endoplasmic reticulum (ER) stress in PCs and alter PC structure and function. Mesencephalic astrocyte-derived neurotrophic factor (MANF) is an ER stress inducible protein highly expressed in PCs. It is neuroprotective in various ER stress-related pathological conditions. However, it is unknown whether MANF plays a role in protecting PCs from alcohol-induced ER stress and neurodegeneration. In this study, we generated PC-specific MANF knockout (KO) mouse model to test the hypothesis that MANF-deficient PCs are more susceptible to alcohol-induced ER stress and neurodegeneration in the adult brain. We employed a binge alcohol exposure paradigm and tested the effect of alcohol and MANF deficiency on molecular, cellular, and behavioral outcomes in the adult animals. We also performed spatial transcriptomics and high throughput in situ analyses to profile gene expression changes in response to MANF deficiency. We found that alcohol exacerbated the motor function deficits in PC-specific MANF KO animals. Interestingly, female KOs were more sensitive to alcohol-induced motor function impairments than male KOs. In accordance with the behavior changes, alcohol exposure activated the unfolded protein response (UPR), increased intranuclear expression of calcium binding protein, and caused PC degeneration in female but not male MANF KO mice. Spatial transcriptomics and high throughput Xenium in situ analyses revealed that MANF deficiency altered the transcriptomic landscape in PCs in a sex-specific manner and triggered the expression of genes involved in protein folding and transportation, and response to ER stress. Our study reveals that MANF-deficient PCs are predisposed with a higher risk to UPR activation and disrupted calcium homeostasis in a sex-dependent manner, which may underline their sex-specific vulnerability to alcohol-induced neurodegeneration. These findings suggest that ER stress plays a significant role in alcohol-triggered neurodegenerative process in the cerebellum, and MANF may possess therapeutic potentials in AUD via its capacity in restoring ER and calcium homeostasis.

Keywords: Alcohol use disorder; Cerebellar dysfunction; Sex difference; Spatial transcriptomics; Unfolded protein response.

Fig. 1

Characterization of PC-specific MANF KO mouse model. A–C. Representative immunofluorescent images of MANF (red) and PC marker CALBINDIN (green) in the control and KO cerebellum at postnatal day (PD) 7 (A), PD 14 (B), and PD 21 (C). D–F. Morphology (D), body weight (E), and brain weight (F) of the adult control and MANF KO mice. The data was expressed as mean ± SEM. n = 8–16 per group for body weight; n = 6 per group for brain weight. Two-way ANOVA followed by Tukey’s post hoc test. ****p < 0.0001; ns not significant. G. Representative immunohistochemistry images of MANF in the adult control and KO cerebellum. Insets show the entire cerebellum lobule II. H. Representative immunofluorescent images of MANF (red) and PC marker CALBINDIN (green) in the adult control and KO cerebellum. Arrows indicated MANF expression in cells located in the granular layer. ML, molecular layer; PL, purkinje cell layer; GL, granule cell layer

Fig. 2

Schematic illustration of the experimental timeline. Both male and female adult mice at the age of 4 to 5 months old received equal volume of H₂O or ethanol (5 g/kg, 25% ethanol w/v) via intragastric gavage once daily for 10 days. Behavioral tests were performed 10 days after the last ethanol exposure. Cohort 1 animals were tested in open field, rotarod, and balance beam tests. Cohort 2 animals were tested in 3-chamber sociability task and Barnes maze test. After the tests, animals were euthanized, and brains were collected for further cellular and molecular analysis

Fig. 3

Effects of ethanol exposure on motor functions in PC-specific MANF KO mice. A–B. Total distance traveled in open field test for water- and ethanol-exposed control and KO female (A) and male (B) mice. C–D. The latency to fall from the accelerating rotarod for water- and ethanol-exposed control and KO female (C) and male (D) mice. E–F. The time to cross the balance beam for water- and ethanol-exposed control and KO female (E) and male (F) mice. All data was expressed as mean ± SEM. n = 8–11 per group. Two-way ANOVA followed by Tukey’s post hoc test. Significant main effect p values were highlighted in bold; *p < 0.05; **p < 0.01; ***p < 0.001; ns not significant

Fig. 4

Effects of ethanol exposure on social behaviors in PC-specific MANF KO mice. A–B. Time spent in the object and social chambers for water- and ethanol-exposed control and KO female (A) and male (B) mice. C–D. Time interacting with the object and social cylinders for water- and ethanol-exposed control and KO female (C) and male (D) mice. All data was expressed as mean ± SEM. n = 8–11 per group. Data was analyzed by three-way ANOVA followed by Tukey’s post hoc test. Significant main effect p values were highlighted in bold; *p < 0.05; ns not significant

Fig. 5

Ethanol induces neurodegeneration in female MANF deficient PCs. A–B. Representative immunohistochemistry images showing the expression of PC marker CALBINDIN in the adult control and KO cerebellum in female (A) and male (B) after water- or ethanol-exposure. Insets represent the enlarged view of the PCs framed with the black square. Note the diminished perikaryal staining and predominant intranuclear immunoreactivity of CALBINDIN in the ethanol-treated female KO PCs. C–F. Quantification of the number (C, D) and size (E, F) of PCs in female (C, E) and male (D, F) cerebellum lobule II. The data was expressed as mean ± SEM. n = 4 per group. Two-way ANOVA followed by Tukey’s post hoc test. Significant main effect p values were highlighted in bold; ****p < 0.0001; ns not significant

Fig. 6

Ethanol induces ER stress in female MANF KO PCs. A–F. Representative immunofluorescent images and quantifications for the expression of ER stress markers GRP78 (A), HYOU1 (B), XBP1s (C), p-PERK (D), p-eIF2α (E), and ATF6 (F) in water- and ethanol-exposed control and KO female mice cerebellum. PC marker CALBINDIN was co-labeled in green. Quantification of ER stress markers fluorescence intensity was expressed as mean ± SEM. n = 3–4 per group. Two-way ANOVA followed by Tukey’s post hoc test. Significant main effect p values were highlighted in bold; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

Fig. 7

Effects of ethanol exposure on ER stress in male MANF KO PCs. A–F. Representative immunofluorescent images and quantifications for the expression of ER stress markers GRP78 (A), HYOU1 (B), XBP1s (C), p-PERK (D), p-eIF2α (E), and ATF6 (F) in water- and ethanol-exposed control and KO male mice cerebellum. PC marker CALBINDIN was co-labeled in green. Quantification of ER stress markers fluorescence intensity was expressed as mean ± SEM. n = 3–4 per group. Two-way ANOVA followed by Tukey’s post hoc test. Significant main effect p values were highlighted in bold; *p < 0.05; ns not significant

Fig. 8

Ethanol induces apoptosis in female but not male MANF KO PCs. A–B. Representative immunofluorescent images of cleaved caspase-3 (red) and PC marker CALBINDIN (green) in water- and ethanol-exposed control and KO female (A) and male (B) cerebellum. The number of cleaved caspase-3 positive PCs in lobule II was quantified and expressed as mean ± SEM. n = 3–4 per group. Two-way ANOVA followed by Tukey’s post hoc test. Significant main effect p values were highlighted in bold; **p < 0.01; ns not significant. C–D. Representative images of TUNEL labeling (green) with immunofluorescent co-labeling of CALBINDIN (red) in water- and ethanol-exposed control and KO female (C) and male (D) cerebellum. The number of TUNEL positive apoptotic PCs in lobule II was quantified and expressed as mean ± SEM. n = 3–4 per group. Two-way ANOVA followed by Tukey’s post hoc test. Significant main effect p values were highlighted in bold; **p < 0.01; ns not significant. Arrow heads in A and C indicate cleaved caspase-3 and TUNEL positive PCs, respectively

Fig. 9

Visium spatial transcriptomics analysis identifies spatially defined tissue clusters in the mouse cerebellum. A. Layout of the 3 Visium slides with sagittal sections of the cerebellum vermis from age-matched male and female control and MANF KO adult mice. B. Mouse cerebellum vermis sagittal section (adapted screenshot from the Allen Reference Atlas, image 17 of 21, P56, Sagittal. Allen Mouse Brain Atlas, mouse.brain-map.org [1]). Cerebellar lobules were indicated by Roman numerals. C. Example of H&E staining image of Visium slide 1, sample 1. D. H&E staining image overlayed with unsupervised graph-based clustering of Visium spots in the cerebellum (spots outside of the cerebellum were not shown). E. UMAP plot of the Visium spots with annotated tissue clusters. F. Expression of selected marker genes for each tissue cluster in the cerebellum. G. PC cluster was selected for further pseudobulk analysis based on Calbindin (Calb1) transcripts distribution in the cerebellum. H. Venn diagram showing the male and female overlapping and distinct differentially expressed genes (DEGs) in the PC cluster

Fig. 10

Transcriptional impact of MANF deficiency in both female and male PCs. A. Heatmap of differentially expressed genes (DEGs) in control and MANF KO PCs shared in both female and male PC cluster identified by pseudobulk analysis. False discovery rate (FDR)-adjusted p value < 0.1, |log₂ FC|> 0.2, and baseMean value > 5. B. Common DEGs enrichment analysis using the Gene Ontology (GO) consortium database for molecular functions. The circular plot shows the top molecular functions and their most important genes in the PCs that were affected by MANF KO. Smallest common denominator-adjusted p value < 0.05, that is − log₁₀ (padj) > 1.3. C. Common DEGs enrichment analysis using the GO dataset for biological processes. The circular plot shows the top biological processes and their most important genes in the PCs that were affected by MANF KO. Smallest common denominator-adjusted p value < 0.05, that is -log₁₀ (padj) > 1.3. D. Common DEGs enrichment analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database for pathways. The circular plot shows the top pathways and their most important genes in the PCs that were affected by MANF KO. Bonferroni-adjusted p value < 0.05, that is − log₁₀ (padj) > 1.3. Upregulated genes were shown in red, downregulated genes were shown in blue

Fig. 11

Female-specific transcriptional changes in PCs in response to MANF deficiency. A. Heatmap of female specific DEGs in control and MANF KO PCs identified by PC cluster pseudobulk analysis. FDR-adjusted p value < 0.1 and |log₂ FC|> 0.2. B. Female-specific DEGs enrichment analysis using the Gene Ontology (GO) consortium database for molecular functions. The circular plot shows the female-specific top molecular functions and their most important genes in the PCs that were affected by MANF KO. Smallest common denominator-adjusted p value < 0.05, that is − log₁₀ (padj) > 1.3. C. Female-specific DEGs enrichment analysis using the GO consortium database for biological processes. The circular plot shows the female-specific top biological processes and their most important genes in the PCs that were affected by MANF KO. Smallest common denominator-adjusted p value < 0.05, that is -log₁₀ (padj) > 1.3. D. Female-specific DEGs enrichment analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database for pathways. The circular plot shows the female-specific top pathways and their most important genes in the PCs that were affected by MANF KO. Bonferroni-adjusted p value < 0.05, that is − log₁₀ (padj) > 1.3. Upregulated genes were shown in red, downregulated genes were shown in blue

Fig. 12

Male-specific transcriptional changes in PCs in response to MANF deficiency. A. Heatmap of male specific DEGs in control and MANF KO PCs identified by PC cluster pseudobulk analysis. FDR-adjusted p value < 0.1 and |log₂ FC|> 0.2. B. Male-specific DEGs enrichment analysis using the Gene Ontology (GO) consortium database for molecular functions. The circular plot shows the male-specific top molecular functions and their most important genes in the PCs that were affected by MANF KO. Smallest common denominator-adjusted p value < 0.05, that is − log₁₀ (padj) > 1.3. C. Male-specific DEGs enrichment analysis using the GO consortium database for biological processes. The circular plot shows the male-specific top biological processes and their most important genes in the PCs that were affected by MANF KO. No processes were identified with the significant smallest common denominator-adjusted p values < 0.05. The top 5 processes with the smallest adjusted p values were listed. D. Male-specific DEGs enrichment analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database for pathways. The circular plot shows the male-specific top pathways and their most important genes in the PCs that were affected by MANF KO. No pathways were identified with the significant Bonferroni-adjusted p value < 0.05. The top 3 pathways with the smallest adjusted p values were listed. Upregulated genes were shown in red, downregulated genes were shown in blue

Fig. 13

Xenium-based single cell in situ gene expression profile in control and MANF KO cerebellum. A. Example of Xenium slide 1, sample 1 cerebellum vermis section stained with DAPI and H&E. Transcripts for 300 genes and unsupervised graph-based clustering of segregated cells were overlayed with DAPI image. Zoom-in images demonstrated individual transcripts as colored dots and cells filled with colors representing different clusters. B. Transcript density maps of selected marker genes for distinct cell clusters in the cerebellum. C. Cells grouped as the PC cluster (purple) matched the expression of PC-specific gene Ppp1r17 (yellow dots) in the cerebellum. They were selected for further pseudobulk analysis.

Fig. 14

Xenium identifies DEGs in female and male PCs in response to MANF deficiency. A–B. Vulcano plot showing the -log₁₀ (padj) and log₂ FC for the 300 genes in the Xenium gene panel in female (A) and male (B). FC: fold change. Black dotted lines indicate the threshold of − log₁₀ (padj) > 1, that equaled to FDR-adjusted p value < 0.1, and |log₂ FC|> 0.2. Red dots: upregulated genes in MANF KO PCs; blue dots: downregulated genes in MANF KO PCs; grey dots: not significantly changed genes. C. Representative Xenium results showing the distribution of Hspa5 (green) transcripts in control and KO female and male PCs (purple). D–E. Frequency distribution of PCs containing Hspa5 transcripts in control and KO female (D) and male (E). F. Representative Xenium results showing the distribution of Sdf2l1 (orange) transcripts in control and KO female and male PCs (purple). G–H. Frequency distribution of PCs containing Sdf2l1 transcripts in control and KO female (G) and male (H). I. Representative Xenium results showing the distribution of Cfap100 (blue) transcripts in control and KO female and male PCs (purple). J–K. Frequency distribution of PCs containing Cfap100 transcripts in control and KO female (J) and male (K). The data was expressed as mean ± SEM. n = 3 animals per group. Multiple unpaired t tests. *p < 0.05. Red circles within PCs indicated the PC nuclei. PCs framed with the white square were enlarged on the right

Fig. 15

Enrichment analysis of DEGs identified in Xenium. A. Violin plots for the expression levels of common DEGs shared in both female and male control and MANF KO PCs. B. Enrichment analysis for the common DEGs. C. Violin plots for the expression levels of female-specific DEGs in control and MANF KO PCs. D. Enrichment analysis for the female-specific DEGs. E. Violin plots for the expression levels of male-specific DEGs in control and MANF KO PCs. F. Enrichment analysis for the male-specific DEGs. Data sources used in the analysis include Gene Ontology (GO) consortium database and Kyoto Encyclopedia of Genes and Genomes (KEGG) database. MF: molecular function; BP: biological process; CC: cellular component. Adjusted p value (padj) was calculated using g:SCS (Set Counts and Sizes) method developed by g:Profiler. Driver GO terms and KEGG terms with padj < 0.05 were listed. The table demonstrated the source of database, term names and IDs with genes involved in each term. Red square: the gene was involved in the term and was upregulated; blue square: the gene was involved in the term and was downregulated