Gpr34 knockout in mice shifts microglial transcriptional profiles toward DAM-and inflammation-associated signatures, partially resembling 5xFAD Alzheimer’s model transcriptomic profile

A: Overview of the experimental design. B: UMAP of single-cell microglia data, colored by microglial states. Each condition (Gpr34 KO, WT, 5xFAD, and Gpr34 KO;5xFAD) is shown separately. C: Volcano plot of differentially expressed (DE) genes between Gpr34 KO and WT mice in the single-cell data, calculated with pseudobulk. Those colored red are significant. Labelled genes are known to be DAM specific based on previous publications. D: Heatmap of selected DE genes. The rows are mice, columns are genes, the color is the z score of the pseudobulked TPM per mouse. The bars on the left represent Gpr34 KO vs WT, the bars on the top represent which gene set each gene belongs to. All are significant at an FDR of 0.05. E: Enrichment analysis of chosen terms related to disease and the immune processes. The terms (see Methods) are on the x-axis, color is Normalized Enrichment Score (NES), size is – log10 pvalue, those outlined in black have FDR<0.1. The dataset of origin for each term is annotated on the top. F: Enrichment analysis of chosen terms related to microglia function. The terms are on the x-axis, color is NES, size is –log10 pvalue, those outlined in black have FDR<0.1. The dataset of origin for each term is annotated on the top. G: Percent of cells (y-axis) in each microglia cell type (x-axis) in each Gpr34 KO and WT mouse (color). There is a significant (FDR<0.05) reduction in homeostatic cells and increase in DAMs in the Gpr34 KO mice, represented by the asterisks. H: Percent of disease associated cells (y-axis) in astrocyte, microglia and oligodendrocyte (ODC) populations (x-axis) in each KO and WT mouse (color). We see an increase in all of them, though it only reaches significance in the astrocytes (FDR<0.05), represented by the asterisk. Sample sizes for sequencing analyses: For single-cell RNA-seq, WT, n = 6 mice; Gpr34 KO, n = 5 mice; and 5xFAD, n = 5 mice. For single-nucleus RNA-seq, n = 6 mice per condition.

Gpr34 deletion partially resembles 5xFAD Alzheimer’s model transcriptomic profile and potentiates DAM and inflammatory programs in the 5xFAD Alzheimer’s model.

A: Volcano plot of DE genes for 5xFAD vs WT microglia in the single-cell data; those colored red are significant. Those gene names colored blue are upregulated in Gpr34 KO vs WT mice and those in black are downregulated. B: Correlation between the logFC of genes in 5xFAD vs WT (x-axis) versus those in Gpr34 KO vs WT (y-axis) in the single-cell data. We only show those genes that are significant in at least one comparison, the color of each dot represents which comparison each gene is significant in—black dots are significant in KO vs WT, grey dots in 5xFAD vs WT, and red are significant in both. Those that are significant in both are labelled. The correlation and the associated permutation-based p-value (the p-value is calculated so as to overcome the issue of shared controls which can lead to inflated correlations) are shown at the top. C: Gpr34 expression for each mouse (y-axis, measured in TPM) in WT vs 5xFAD microglia (x-axis). D: Volcano plot of DE genes in KO;5xFAD vs 5xFAD mice in the single-cell data, calculated with a pseudobulk based method. Those colored red are significant. Those labelled genes are the top 20 most significant genes. E: Enrichment analysis of chosen terms related to disease, the immune system and microglia function in the single cell data. The terms are on the y-axis, color is NES, size is –log10 pvalue, those outlined in black have FDR<0.1. The comparison is on the x-axis, either 5xFAD vs WT or KO;5xFAD vs 5xFAD. The dataset of origin for each term is annotated on the right side. F: Percent of disease associated cells (y-axis) in each Gpr34 KO;5xFAD and 5xFAD mouse (color) for each cell type (x-axis) in the single-nucleus data. We see an increase in all of them, though it only reaches significance in the astrocytes and oligodendrocytes (FDR<.05), represented by the asterisks. Sample sizes for sequencing analyses: For single-cell RNA-seq, WT, n = 6 mice; Gpr34 KO, n = 5 mice; and 5xFAD, n = 5 mice. For single-nucleus RNA-seq, n = 6 mice per condition.

Gpr34 deletion exacerbates microglial reactivity and increases amyloid beta engulfment in 5xFAD mice

A: Representative image of Aβ (shown in green) immunofluorescence staining in the cortex of 5xFAD and Gpr34 KO;5xFAD mice. Scale bar indicates 10µm. B: Quantification of total Aβ signal in the region of interest (ROI) of the cortex of 5xFAD and Gpr34 KO;5xFAD mice using confocal z-stack imaging and Imaris 3D surface analysis. Values represent summed Aβ-positive signal within each ROI. P = 0.17. C: Quantification of total Iba1 immunoreactivity in the ROI of the cortex of 5xFAD and Gpr34 KO;5xFAD mice using confocal z-stack imaging and Imaris 3D surface analysis. P = 0.90. G: Representative image of Iba1 (shown in cyan) and CD68 (shown in magenta) immunofluorescence staining in the cortex of 5xFAD and Gpr34 KO;5xFAD mice. Scale bar indicates 10µm. D: Representative image of CD68 (shown in pink) and Iba1 (shown in blue) immunofluorescence staining in the cortex of 5xFAD and Gpr34 KO;5xFAD mice. Scale bar indicates 10µm. E. Quantification of CD68 signal within Iba1+ microglial compartments, normalized to total Iba1 signal, in cortical ROIs from 5xFAD and Gpr34 KO;5xFAD mice using Imaris 3D surface analysis. Values are normalized to total Iba1 signal. P = 0.0017. F: Quantification of Aβ signal within CD68+/Iba1+ compartments in cortical ROIs using Imaris 3D surface analysis. Values are normalized to total Aβ signal. P = 0.0040. G: Representative whole-slide immunofluorescence image of GFAP (shown in yellow) staining in the cortex of 5xFAD and Gpr34 KO;5xFAD mice. Scale bar indicates 20µm. H: Quantification of GFAP-positive immunoreactivity within cortical ROIs from 5xFAD and Gpr34 KO;5xFAD mice, using 2D whole-slide image analysis, measured as percent area. P = 0.084. Imaging quantification: 5xFAD, N=9 mice and Gpr34 KO;5xFAD, N=12 mice. 4-5 brain slices were analyzed for each animal for Aβ, Iba1 and CD68 measurements. 2-3 brain slices were analyzed for each animal for GFAP measurement. Data represent mean ±s.e.m. Welch’s t test. Sample sizes for sequencing analyses: For single-cell RNA-seq, 5xFAD, n = 5 mice, and Gpr34 KO;5xFAD, n = 6 mice. For single-nucleus RNA-seq, n = 6 mice per condition.

Generation and pharmacological validation of GPR34-deficient human iPSC-derived microglia (iMGLs)

A: Sequencing validation of two independent GPR34 iPSC knockout (KO) clones (KO-1 and KO-2) and their respective WT controls. Guide RNA sequences for KO-1 and KO-2 are indicated below the alignment, with predicted Cas9 cut sites marked. Red boxes indicate cut sites, confirming editing of GPR34 in the KO clones relative to WT and parental controls B: MS/MS analysis shows the loss of GPR34 protein in GPR34 KO-1 iMGLs C: Schematic illustrating GPR34 activation and downstream signaling pathways. D: Dose-dependent calcium response to the GPR34 agonist Compound 4B in WT and GPR34 KO iMGLs. E: Dose-dependent pERK activation in response to the GPR34 agonist Compound 4B in WT and GPR34 KO iMGLs. F: Dose-dependent calcium response to the GPR34 endogenous ligand lysoPS in WT and GPR34 KO iMGLs. The representative curve was shown from the average data from triplicates. G: Dose-dependent pERK activation in response to the GPR34 endogenous ligand lysoPS in WT and GPR34 KO iMGLs. H: Myelin-induced calcium signaling in WT and GPR34 KO iMGLs treated with (S)-E49. I: Dose-response of myelin-induced calcium flux in WT and GPR34 KO iMGLs. J-K: Dose-dependent calcium response to GPR34 unrelated agonists (P2Y12/2-MeSADP, C3aR/BR103) in WT and GPR34 KO iMGLs. Sample sizes and stats: GPR34 KO line 1 (KO-1) and its control (WT-1) were used for this study. The representative curve for D, F, G, H and I was shown from the average data are from triplicates and for E is from duplicates.

GPR34 deficiency selectively impairs myelin phagocytosis in iPSC-derived microglia

A: Experimental workflow for live-cell imaging and flow cytometry-based assessment of phagocytosis. iMGLs were incubated with pHrodo-labeled substrates and imaged following 5 and 24 hours of substrate incubation using the Opera Phenix live-imaging system. Cells were then detached from imaging plates onto PhenoPlates and processed by flow cytometry on the CytoFLEX LX for assessment of median fluorescence intensity (MFI). B-D: WT and GPR34 KO iMGLs were exposed to pHrodo-labeled myelin (B), amyloid fibrils (C), or E. coli bioparticles (D) and assessed by live cell imaging at 5 and 24 hours. Lines connect paired WT and KO biological replicates derived from the same experimental batch. E: WT and GPR34 KO iMGLs were exposed to GPR34 antagonist YL-365 (10 µM) and assessed for phagocytosis of pHrodo labeled myelin at 24 hours by live cell imaging. Sample sizes and stats: Data are represented as mean values with +/- SEM. Smaller circles represent technical replicates, and larger circles represent biological replicates. Statistical tests were undertaken at 5h, 24h, and 25h (n=2-4 independent biological replicates with 5-6 technical replicates per condition). Two-way ANOVA with Bonferroni correction, multiple comparisons test, paired t-test with Welch’s correction, as appropriate.

Multiomic analysis of GPR34-deficient iMGLs highlights dysregulation of immune, replicative, and lysosomal programs

A: Volcano plot of RNA-seq analysis in GPR34 KO vs WT iMGL samples. Red dots are significant, top genes (those with smallest p-values) are labelled. Based on the meta-analysis of two experimental batches. B: Plot of selected gene sets (y-axis) from gene set enrichment analysis for the GPR34 KO vs WT RNA-Seq comparison. Color is NES, size is -log10 p-value, those circled in black are significant (FDR<.1). C: Heatmap of selected genes from pathways altered in GPR34 KO vs WT iMGL samples. The columns are samples, rows are genes, the color is the z score of the TPM expression values. The bars on the top represent KO vs WT. Note the separation between genes from different gene sets (y-axis). D: Volcano plot of differentially expressed proteins (DEPs) in MS/MS quantitative proteomics analysis of GPR34 KO vs WT iMGLs. Top DEPs are labelled. E: Gene set enrichment analysis of quantitative proteomics data highlighting prominently changed molecular pathways in GPR34 KO vs WT iMGLs. Color indicates normalized enrichment scores (NES) and bubble size indicates FDR-adjusted P-value. All shown pathways are significantly altered (FDR-adjusted P < 0.05) in KO iMGLs (compared to WT). F: Volcano plot of quantitative proteomics analysis of GPR34 KO vs. WT iMGLs highlighting endosomal and lysosomal pathways. Proteins directly involved in endosome/lysosome function (“direct - endo/lyso”) and known regulators of these pathways (“regulators”) are indicated. GPNMB, a putative biomarker of lysosomal dysfunction, is also highlighted. G: Volcano plot of quantitative proteomics analysis of GPR34 KO vs WT iMGLs highlighting DNA replication and MHC class II pathways H: Comparison of transcriptomic (x-axis) and proteomic (y-axis) changes (log2FC) in GPR34 KO vs WT iMGLs. Color indicates DE in RNA, protein, or both. Only genes that are DE in either or both assays are shown.

GPR34 loss blunts microglial transcriptional responses to myelin challenge

A: PCA plot of iMGL bulk RNA-seq samples, colored by myelin treatment. The genotype is represented by the shape. B: Bar plots of the number of DE genes (x-axis) for each of the myelin-treatment comparisons (y-axis). C: Venn diagram of the number of DE genes after 24-hour myelin treatment versus no-myelin in WT and GPR34 KO iMGLs. Note overlaps are only counted if they have the same direction of effect. D: Volcano plots comparing GPR34 KO versus WT iMGLs after 24-hour or 2-hour myelin treatment. Significant hits are colored red (FDR<.05), genes with the smallest p-values are labelled. E: Expression of selected microglial response genes APOE, CCL3, CD68, CST7, GPNMB, TREM2 (TPM, y-axis) in WT and GPR34 KO iMGLs (x-axis) after 24-hour myelin treatment. In the box plots the center line is the median, the boxes span from 25% to 75%, the length of the whiskers is based off the smallest/largest value within 1.5 of the interquartile range from the median, and everything outside that range is plotted as an outlier. F: Heatmap of selected myelin-response genes across genotypes and myelin-treatment conditions. The columns are samples, rows are genes, the color is the z score of the TPM expression values. The bars on the top represent KO vs WT. Note the separation between samples with different treatments (x-axis) and genes from different gene sets (y-axis). G: Gene set enrichment analysis across genotypes and myelin-treatment comparisons. The terms are on the y-axis, color is NES, size is –log10 p-value, those outlined in black have FDR<0.1. The comparison is on the x-axis. The dataset of origin for each term is annotated on the right side.