Fig 2: The soluble GPNMB extracellular domain (ECD) enhances fibrillar aSyn uptake and aggregation in iNeurons.(A). Proposed mechanism of GPNMB ECD action. The GPNMB ECD is cleaved from the membrane by metalloproteases, where it can act non-cell-autonomously to trigger downstream molecular events, including enhanced fibrillar aSyn uptake.(B). Experimental scheme encompassing differentiation of iNeurons, addition of GPNMB ECD and aSyn fibrils, and timing of aSyn fibril uptake and phosphorylated aSyn pathology assessments.(C). Representative images of WT iNeurons treated with labeled aSyn PFFs alone (WT) or PFFs and GPNMB ECD (WT + ECD); or KO iNeurons treated with labeled PFFs alone (KO) or PFFs and GPNMB ECD (KO + ECD). Nuclei were stained with DRAQ5. Representative image (scale bar = 50um) with inset highlighting region of interest (scale bar = 20um). PFFs internalized by cells are shown in red, while tubulin staining (Tub) is shown in grey.(D). Quantification of aSyn puncta in internalization experiments. KO neurons exhibit a significant reduction in internalization compared to WT neurons (*p=0.04). This is rescued by treating KO neurons with GPNMB ECD (**p=0.007). Treatment of WT neurons with GPNMB ECD also results in significantly increased internalization of labeled aSyn fibrils (*p=0.017). Bar depicts mean. Each dot represents one field, with 4-10 fields quantified per replicate (well, n = 6) across 3 differentiations. Statistics were calculated using nested t-test (one-tailed, given expected direction) to account for non-independent fields (see Methods).(E). Representative images of WT iNeurons treated with untagged aSyn PFFs alone (WT) or PFFs in combination with GPNMB ECD (WT + ECD); and KO iNeurons treated with untagged PFFs alone (KO) or in combination with GPNMB ECD (KO + ECD). aSyn aggregates phosphorylated at serine 129 (pSyn, green) were detected 14 days after PFF treatment, following extraction of soluble proteins with 1% Triton-X, using an antibody against phosphorylated synuclein (pS129, EP1536Y). Nuclei were stained with DRAQ5. Representative image (scale bar = 50um) with inset highlighting region of interest (scale bar = 20um).(F). Quantification of insoluble, phospho-Syn (pSyn) aggregates, normalized to WT. GPNMB KO neurons exhibit a significant reduction in aggregate number compared to WT neurons (**p=0.007) and to KO neurons treated with GPNMB ECD (*p=0.038). In contrast, GPNMB ECD treatment did not significantly alter pSyn aggregates in WT neurons compared to baseline. Bar depicts mean. Each dot represents one field, with 3-7 fields quantified per replicate (well, n = 6) across 4 differentiations. Statistics were calculated using nested t-test (one-tailed, given expected direction) to account for non-independent fields (see Methods).

Fig 3: GPNMB is predominantly expressed in microglia, and PD brain has more GPNMB-expressing microglia.(A). UMAP projections of four single-cell RNA sequencing (scRNAseq) datasets. Cell types are assigned to clusters according to canonical markers (see Methods).(B). Corresponding UMAP projections showing expression of GPNMB transcript, visualized as normalized log-transformed counts (red). GPNMB expression is enriched in the microglia cell cluster across all datasets examined.(C). Percentage of cells expressing detectable levels of GPNMB transcript, subdivided by cell type (neurons, astrocytes, oligodendrocytes, endothelial cells, microglia – see Methods for assignment), from each scRNAseq study.(D). Percentage of microglia expressing detectable levels of GPNMB transcript, stratified by neuropathological diagnosis (NC or PD) from each scRNAseq study, as well as the integrated data from all four studies, showing increased numbers of GPNMB+ microglia in PD (**p=0.0067). Statistics were calculated using an unpaired t-test (two-tailed).(E). Volcano plot comparing microglia from PD vs. NC samples in the integrated dataset. GPNMB is one of the most upregulated genes in PD microglia.

Fig 4: Anti-GPNMB mAb treatment rescues iNeurons from developing aSyn pathology.(A). Control conditions showing formation of hyperphosphorylated insoluble aSyn aggregates in WT iNeurons, but not GPNMB KO iNeurons, 14 days after seeding with aSyn PFFs. Phosphoserine 129 aSyn aggregates (green) were stained with EP1536Y, and nuclei with DRAQ5. Scale bar = 50 μm.(B) Representative images of WT iNeurons seeded with aSyn fibrils and also treated with low (top row, 75 ng/mL) or high (bottom row, 750 ng/mL) dose of negative control (isotype-matched IgG) antibody, anti-GPNMB mAb-26, or anti-GPNMB mAb-1. Cells were stained for hyperphosphorylated, 1% Triton-X-insoluble aSyn aggregates (green, EP1536Y) and nuclei (DRAQ5, blue) 14 days after seeding with aSyn PFFs. Scale bar = 50 μm.(B). Quantification of phosphorylated αSyn aggregates in WT iNeurons seeded with PFF and treated with negative control antibody (IgG, grey), anti-GPNMB mAb-1 (green), or anti-GPNMB mAb-26 (blue). Untreated WT iNeurons (purple) and GPNMB KO iNeurons (pink) seeded with PFF are also shown for reference. Each dot represents one field, with 3-6 fields per replicate (n=4 replicate wells), across 2 differentiations. Statistics were calculated using nested one-way ANOVA, followed by pairwise comparisons of each antibody and dose with PFF-only condition, adjusted for testing 3 antibodies (see Methods). One-tailed p-values are presented, given expected direction of effect. Treatment with mAb-1 at low (adjusted p=0.004), and high (adjusted p=0.001) doses, as well as mAb-26 at high dose (adjusted p=0.003) significantly rescued pSyn pathology.

Fig 5: Expression of GPNMB in the human brain, across brain regions and neuropathological diagnoses(A). Representative immunohistochemistry images of GPNMB stained in the temporal cortex, cerebellum, cingulate cortex, and hippocampus across neurologically normal control (NC), and Lewy body disease (LBD) cases. GPNMB is detected (AF2550 antibody) in cells with the morphology of neurons (blue arrowheads) and microglia (red arrowheads) across all brain regions. Representative image (scale bar = 2mm) with insets highlighting region of interest (scale bar = 100um), indicated on low-power image with triangle corresponding to color of inset frame.(B). Quantification of GPNMB protein expression (percentage area with GPNMB staining) in the temporal cortex, cerebellum, hippocampus, and cingulate cortex across NC, LBD cases. Within the LBD group, the presence (LBD/AD+, pink dots) or absence (LBD/AD−, teal dots) of concomitant AD pathology is indicated. Bars depict mean +/− SEM, with each dot representing one case. Statistics were calculated using an unpaired t-test (two-tailed).