GBA1 L444P PD patient-derived cortical neurons have increased total C16:1 and C18:1 fatty acid associating with decreased αS T:M ratio that are rescued by pharmacological inhibition of SCD. We and others previously identified a primary α-synucleinopathy associated with increased cellular MUFAs (C16:1, C18:1) (22, 55, 56). Correcting this MUFA imbalance with SCD inhibitor treatment reversed PD-relevant phenotypes in vitro and in vivo, and an SCD inhibitor subsequently entered PD clinical trials (22, 44–46, 55, 57, 58). Here, we hypothesized that there may be increased MUFAs in a nonprimary α-synucleinopathy, resulting from mutations associated with GBA1-PD. To assess MUFAs in GBA1 mutation carriers, we performed a fatty acid profiling comparing patient-derived GBA1-PD (L444P) mutant iPSC neurons with those of an isogenic corrected control. Analysis focused on MUFAs C16:1 (n9, n7) and C18:1 (n9, n7), and their saturated chain length matched precursors C16:0 and C18:0. Both forms of C16:1 (n9, n7) were increased in the L444P neurons relative to the isogenic control neurons with a small decrease (as expected) for the saturated precursor C16:0 (Figure 1, A–C) (C16:0, P < 0.0001; C16:1n9, P = 0.0003; C16:1n7, P = 0.0081). Similarly, both C18:1n9 and C18:1n7 were increased in the mutant neurons relative to the isogenic control neurons with decreased C18:0 (Figure 1, D–F) (C18:0, P = 0.0028; C18:1n9, P < 0.0001; C18:1n7, P = 0.0002). This increase in MUFAs and decrease in saturated fatty acids C16:0 and C18:0 resulted in an overall increase in the desaturation index (Figure 1, G and H) (C16:1/C16:0, P = 0.0007; C18:1/C18:0, P = 0.0004). Based on these substantial increases in SCD-related C16 and C18 FADI, we hypothesized correcting MUFA dyshomeostasis via SCD inhibition as a candidate therapeutic approach for GBA1 mutation carriers.

Monounsaturated fatty acids (C16:1 and C18:1) are increased in GBA1 L444P mutant iPSC neurons and SCD inhibition increases αS T:M ratio of patient-derived GBA1 L444P and E326K mutant iPSC neurons. GBA L444P mutant and isogenic corrected neurons were differentiated (DIV 20) and harvested for FA analysis by gas chromatography. n = 6. (A–C) Total cellular C16:0, C16:1n9, and C16:1n7 of GBA1 L444P mutant neurons and isogenic corrected control neurons were measured by gas chromatography. Data are reported relative to the isogenic corrected line. GraphPad Prism 10, unpaired 2-tailed t test. (D–F) Total cellular C18:0, C18:1n9, C18:1n7 of GBA1 L444P mutant neurons and isogenic corrected control neurons were measured by gas chromatography. Data is reported relative to the isogenic corrected line. Statistical analysis: Graphpad Prism 10, unpaired 2-tailed t test. (G) Heatmap shows the calculated desaturation index for C16:1n9+C16:1n7/C16:0 for GBA L444P mutant neurons versus isogenic corrected control neurons. (H) Heatmap shows the calculated desaturation index for C18:1n9+C18:1n7/C18:0 for GBA L444P mutant neurons versus isogenic corrected control neurons. (G and H) n = 6. (I and L) L444P and E326K neurons were treated with 1 μM 5b or DMSO. Cells were crosslinked using DSG. Cell lysates were immunoblotted to detect and quantify αS14, αS60, and DJ-1 (crosslinking control). (J) L444P: Quantification of αS60:αS14 (T:M) ratio. Two-way ANOVA: statistically significant effects of condition (F1,46 = 18.75, P = 0.0011) and treatment (F1,46 = 21.69, P = 0.0005), Tukey’s’s HSD test, Corr without 5b (Corr–5b) versus L444P–5b, P = 0.0060, L444P–5b versus L444P with 5b (L444P+5b), P = 0.0028; N = 2, n = 12–19). (M) E326K: Quantification of αS60:αS14 (T:M) ratio. (K and N) No statistical differences in DJ-1 (P > 0.05). Data are shown as mean ± SD. Unpaired 2-tailed t test, P = 0.0036; N = 1, n = 8–9. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We previously showed that a brain-penetrant SCD inhibitor (5b) corrected the aberrant αS T:M ratio in 3K mice (46) and PD patient-derived αS triplication neurons (22). Here, we investigated 5b treatment of GBA1 L444P and E326K patient-derived cortical neurons. The cells were treated with 1 μM 5b and crosslinked using DSG. In line with the findings of Kim and colleagues (23), the αS T:M ratio was markedly reduced in L444P neurons relative to that of the isogenic corrected neurons (2-way ANOVA, F1,46 = 18.75, P = 0.0011). The L444P-associated αS T:M was normalized by 5b treatment (P = 0.0028; Figure 1, I and J). 5b treatment significantly increased αS T:M ratio to a similar degree in GBA1 E326K neurons (unpaired 2-tailed t test, P < 0.01; Figure 1, L and M). No differences were detected in DJ-1, which served as loading and crosslinking quality control (Figure 1, K and N). Quantifying αS tetramers plus probable higher molecular weight conformers of the tetramer (potential octamers and hexamers at ~80 and 100 kDa) (59) revealed a significant decrease in L444P versus isogenic corrected controls (P < 0.05), and 5b treatment raised the multimer/monomer ratio (P < 0.05) (Supplemental Figure 1, A and B; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.188413DS1). Previous studies show that GBA1 L444P–mediated GCase deficiency (50) negatively affects the αS T:M ratio in brain neurons (23). We next tested GCase activity in GBA1 E326K neurons and observed a significant reduction in activity at pH 4.5 (P < 0.0001; Supplemental Figure 1C).

These data suggest that GCase-deficient GBA-PD–derived neurons with increased SCD products have a decrease in the physiological αS tetramerization that can be increased by SCD inhibition.

Gba1 mutations cause a progressive motor syndrome associated with a decrease in mature GCase and the αS T:M ratio in mice. Previous studies have reported that homozygous E326K Gba1 KI mice develop progressive neuropathology, including a decrease in GCase activity, tyrosine hydroxylase+ (TH+) neuronal loss, and a higher susceptibility toward fibrillar αS, associating with an age-dependent motor deficit (60), while heterozygous E326K or L444P mice show only very subtle changes (13, 60). In order to promote phenotype development, we bred both L444P and E326K mouse lines to homozygosity. Previously, these L444P (The Jackson Laboratory, 024574) and E326K KI mice were reported to be viable and fertile and to live a normal life span (13, 60). Accordingly, we did not observe any changes in viability or fertility between the homozygous Gba1 KI mouse lines and control (Ctl) mice (data not shown).

Accelerated rotarod testing (4–40 rpm, average of 6 testing trials during 3 consecutive days [d]) of 12-month-old L444P and E326K mice showed a significant decrease versus age-matched Ctl mice (Ctl 262 ± 10 seconds; L444P 234 ± 7 seconds; E326K 135 ± 10 seconds; 1-way ANOVA; P < 0.0001). To more finely analyze the underlying gait changes at 12 months, Gba1 mutant and Ctl mice were additionally subjected to a gait scan analysis, using high-resolution photography of gait symmetry by placing mice on a transparent motoric belt. A one-way ANOVA showed an increase in stance (P < 0.0001), a shorter percentage of time in limb swing (P < 0.04) and an increase in the average rear track width (P = 0.01), but no changes were observed front track width (Figure 2B). Together, these data suggest motor deficits and an abnormal, insecure gait (paw-to-body support) in aged (12 months) Gba1 mutant mice.

Gba1 mutant mice develop a progressive motor decline and an insecure gait. (A) Graph quantifies balancing on a 4–40 rpm accelerating rotarod (avg. of 6 trials on 3 consecutive days) at 12 months. (B) Automated gait scans of mouse paw pattern on a horizontal treadmill (Cleversys) at 12 months. Schematic of stepping pattern derived from gait scans display insecure (bradykinetic) gait in Gba1 mutant mice. Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001; ****P < 0.0001. One-way ANOVA, Tukey’s post hoc test.

GCase activity is reduced in L444P (52) and in E326K (60) mice. We hypothesized that reduced GCase function contributes to a disequilibrium of saturated and MUFAs (Figure 1). These MUFAs may then be incorporated into lysosomal and other phospholipid membranes, leading to abnormal αS membrane interactions that decrease αS tetramerization and allowing the resultant free monomers to aggregate with lysosomes and lipid droplets (LDs), inducing neuropathology and phenotypes. An accumulation of αS aggregates can further trigger a decrease in mature/functional GCase, creating a bidirectional loop (3) (Figure 3A).

Gba1 mutant mice display a decrease in GCase maturation, a decrease in the αS T:M ratio, and more insoluble pS129+ mice. (A) Lysosomal dysfunction hypothesis. Reduced GCase associates with a disequilibrium of saturated and unsaturated FA in lysosomal and other phospholipid membranes leading to a shift in αS tetramers and the free monomers abnormally aggregate with lysosomes and lipid droplets, contributing to neuropathology and phenotypes. The αS+ aggregates can further decrease the amount of lysosomal active (mature) GCase. (B) WB of intact-cell crosslinked GCase and αS using cell-penetrant cross-linker DSG in cortical brain bits lysed with 1% Triton X-100/PBS. GCase signals reveal less mature (>62 kDa), more immature GCase (<62 kDa), and a reduced αS T:M signal in Gba1 mutant mice. DJ 1 signal serves as a control for crosslinking (dimer/monomer ratio) and loading. (C) Quantifies WBs in B. (D) Expression data from mouse SCD1 RNA in Ctl, L444P, and E326K (see also Supplemental Figure 2D). (E–F) WBs (noncrosslinked) of sequentially extracted TBS (soluble), RIPA-soluble (insoluble) extracts from cortex (left panels) and midbrain (right panels) (E), quantified in F and G, respectively. The TBS fraction was developed against (total) αS and the detergent-insoluble (RIPA) fraction against serine 129 phosphorylated αS and (total) αS of the corresponding blot. Actin serves as loading control. Data are shown as mean ± SEM. One-way ANOVA, Tukey’s post hoc. *P < 0.05, **P < 0.01.

The presence of lysosomally active GCase can also be inferred from the relative level of its biochemically higher molecular weight mature (glycosylated) form (>60 kDa, located at the lysosome) versus lower molecular weight immature forms (50–60 kDa, located at the endoplasmatic reticulum) (61). Quantitative Western blots (WBs) of the DSG-crosslinked cortical protein extracts showed a decrease of mature (>60 kDa) GCase and more immature (<60 kDa) GCase in both the L444P and the E326K mice (Figure 3, B and C). Reductions in GCase activity were also observed via a GCase activity assay, which is more sensitive than the relative GCase glycosylation (62), confirming reduced GCase activity in L444P and in E326K mouse cortex (Supplemental Figure 2A). Next, we evaluated the relative level of αS 60 tetramers to αS14 monomers (T:M) from the same protein extract, revealing a significant decrease in L444P and E326K mouse brain (Figure 3, B and C) (1-way ANOVA, P < 0.05) . Quantifying αS tetramers plus probable conformers (59) of the tetramer (potential octamers and hexamers) revealed a significant decrease in L444P (P < 0.05) and a trend toward a decrease in E326K (P = 0.08) (Supplemental Figure 2B). We next measured whether Gba1 mutations affect SCD expression, which may account for the detected increase in MUFAs in culture (Figure 1). The SCD1 mRNA level was increased in L444P and E326K mouse brain (Figure 3D and Supplemental Figure 2D).

Sequential extractions of brain cortices revealed an increase in (TBS) buffer-soluble and buffer-insoluble (RIPA) αS in Gba1 mutant mouse brain (Figure 3E). Buffer insolubility was confirmed in a second PD-vulnerable brain region, the midbrain (Figure 3E). We probed the insoluble lysates against phosphorylated serine (pS) 129 αS, an established marker for αS aggregation (63), and found that it was elevated in both the cortical and midbrain region of Gba1 mutant mice (Figure 3E). Given the accumulation in total insoluble αS, we quantified the pS129 αS/total αS ratio. Despite the overall accumulation of αS monomers, there was still a significant increase in pS129/total αS in brain cortex (1-way ANOVA, P < 0.03) (Figure 3F). There was no statistically evident change in the pS129/total αS ratio in the midbrain, which may be due to technical variability when dissecting this small brain region (Figure 3G).

Brain FADI correction induced by 5b improves motor phenotypes and DAergic fiber integrity in Gba1 mutant mice. Based on the in vitro efficacy of 5b in GBA1-PD patient-derived iPSC neurons, we tested in vivo whether long-term 5b treatment could ameliorate the observed motor phenotypes, GCase maturation, and αS dyshomeostasis in Gba1 mutant mice. We used 12-month-old symptomatic Gba1 mutant mice (Figure 2A) and fed these mice with 5b formulated in chow (0.15 g per kg/food) ad libitum from age 12–16 months (Figure 4A). A subset of L444P mice were fed with a low dose (0.075 g per kg/food) of 5b (termed “L444P-LD-5b” (Supplemental Figure 3). Baseline (BL), interim (30d) and final (90d) motor assessment were performed, and the study ended after 120d (~16 weeks). An average chow intake of ~3 g/day was maintained throughout the study, indicating a daily dose intake of ~15 mg/kg. During the 4 months of treatment, 30d and 90d were selected to compare the motor performance in 5b-treated Gba1 mice versus placebo-treated (Plb-treated; diet without 5b) mice, and the 5b treatment led to a consistent improvement in GBA mutant mice (Figure 4B). Two-way ANOVA revealed a significant treatment effect over time (F2,55 = 9.3; P < 0.001) and pairwise comparisons showed a statistical significance was achieved in both Gba1 mutant mouse lines at 90d (P < 0.05) (Figure 4B). While no improvement in the overall motor performance was detected in L444P-LD-5b (at 90d) (Supplemental Figure 3A), the mice displayed an improvement in balancing skills between trial 1 and trial 2 of the rotarod testing (Supplemental Figure 3B). This indicates enhanced motor skill learning in the low-dose–treated L444P versus Plb L444P similar to the improvements detected in the high-dose L444P-5b–treated mice (Supplemental Figure 3B).

5b SCD inhibition improves the motor performance, lowers the fatty acid desaturation index, and restores striatal DAergic fiber densities and DA level. (A) SCD-inhibitor 5b treatment study of symptomatic (12 months old) L444P and E326K Gba1 versus Ctl mice. All mice were treated either with 15 mg/kg 5b or Plb. Some additional L444P Gba1 mice were treated with 7.5 mg/kg 5b (low dose [LD]; see Supplemental Figure 3). (B) Graph quantifies balancing skill learning on a 4–40 rpm accelerating rotarod. (C) Fatty acid saturation indices in brain cortex of 5b versus Plb treated mice validating efficacy reducing the specific MUFAs (C16:1, C18:1 versus C16:0, C18:0) ratio by 5b treatment (see also Supplemental Figure 1, D and E). Heatmaps show the calculated desaturation index for GBA L444P and E326K Plb and 5b. Note: Planned pairwise comparisons showed a relative increase of FADI C16:1/C16:0 in Plb L444P and E326K versus Ctl. Quantifying FADI C16 and FADI C18 in Gba1 (E326K+L444P) showed a significant decrease in 5b versus Plb. (D) Representative images of TH+ nerve terminals and fibers of Ctl, L444P, and E326K Gba1 mice. Scale bar: 600 μm. (E) Relative TH optical density (total of 12 sections; n = 3–4 mice each cohort) was analyzed in the dorsal striatum. (F) HPLC assay of striatal dopamine measured by HPLC. Data are shown as mean ± SEM. Two-way ANOVA with Bonferroni (C) or Tukey’s (B, E, and F) post hoc tests. Two-tailed, unpaired 2-tailed t test comparing Gba1 (E326K+L444P) Plb versus 5b. *P < 0.05, **P <0.01, ***P < 0.001, ****P < 0.0001

Sixteen weeks after initiating 5b administration, mice were euthanized, and brains were harvested. The effect of SCD inhibition was confirmed by measuring the C16 and C18 FADI using liquid chromatography-mass spectrometry (LC-MS). Pairwise comparisons revealed a significant increase in the C16:1/C16:0 FADI in Plb Gba1 versus Ctl mouse brain, similar to changes detected in GBA1-PD patient-derived cortical neurons (Figure 1, A–C, and G) (Supplemental Figure 1, D and E). The 5b treatment normalized the C16 FADI in Gba1 mutant mice, consistently detected in all 5b-treated (Ctl, L444P and E326K) versus Plb-treated mouse cortices (2-way ANOVA, P < 0.0001) (Figure 4C). Testing for 5b treatment effects in Gba1 (L444P+E326K, combined) mice revealed a significant decrease in both the FADI C16 and FADI C18 by 5b (P < 0.0001; Figure 4C). Consistently, 5b treatment normalized the SCD1 RNA level between Ctl and Gba1 mutant mice and decreased the level comparing Gba1 (E326K+L444P, combined) Plb versus 5b (P < 0.0001) (Supplemental Figure 2D).

Given the improvement in motor phenotypes with 5b treatment in Gba1 mutant mice, we next determined the effects of 5b on PD-like brain DAergic pathologies, by assessing striatal TH immunoreactivity and striatal DA level. We quantified TH immunopositive nerve terminals in the dorsal striatum (caudate putamen), which is rich in projections from DAergic neurons in the substantia nigra pars compacta. Quantification in L444P and E326K showed a ~15% decrease in TH+ fibers versus Ctl at the analyzed age of 16 months (genotype × treatment, P = 0.001) and pairwise comparisons showed 5b-treated Gba1 (L444P and E326K) mice displayed no differences in TH+ neurites when comparing with 5b-treated Ctl mice (Figure 4, D and E). Testing for 5b treatment effects in Gba1 (L444P+E326K, combined) mice showed a significant increase in 5b versus Plb (P < 0.01) (Figure 4E). Accordingly, a lower striatal DA concentration was measured by HPLC in Gba1 mutant versus Ctl mice, reaching significance in L444P (P = 0.015), which was not detected when mice were treated with 5b (Figure 4F). The 5b treatment effects in Gba1 (L444P+E326K, combined) mice showed significantly increased DA level in Plb versus 5b (P < 0.05) (Figure 4F). The DA level was also similar between L444P-LD-5b, L444P-5b, and Plb control mice (Supplemental Figure 3C).

Having determined that 5b treatment effectively decreased brain SCD products and normalized striatal DAergic fiber integrity, we next investigated GCase maturation and αS homeostasis in the Plb- and 5b-treated Gba1 mice that had shown motor improvements over the Plb at 90d of treatment (Figure 4B).

First, we assessed the level of mature and immature GCase, which was lower in Plb Gba1 mutant mice, reaching significance in L444P versus Ctl (P = 0.03). However, these differences were not detected after 5b treatment in Gba1-5b versus Ctl-5b (Figure 5, A and D). Comparing Gba1 (L444P+E326K, combined) Plb versus 5b showed a nonsignificant trend in improved GCase maturation (P = 0.1) (data not shown). The αS T:M ratio revealed a significant decrease in Plb Gba1-mutant versus Plb-Ctl (2-way ANOVA; genotype, P = 0.02; treatment, P = 0.03), but no changes were detected between Ctl and Gba1 mutant mice after 5b treatment. The αS T:M level significantly raised when comparing Gba1 (L444P+E326K) Plb versus 5b mice (P < 0.01) (Figure 5, A and E; multimer-to-monomer quantification in Supplemental Figure 2C). No difference in the αS T:M ratio was detected between Plb Ctl and L444P-LD5b and L444P-5b (2-way ANOVA, P > 0.05; Supplemental Figure 3F). To assess whether the increase in the αS T:M ratio was associated with changes in αS buffer solubility and serine 129 phosphorylation, sequential extraction of cortical (Figure 5B) and midbrain (Figure 5C) homogenates was performed. While no differences in the relative TBS/RIPA-buffer solubility were detected between groups (Figure 5, F and H), both L444P and E326K Gba1 Plb mice showed a significant increase in buffer-insoluble pS129 in cortical and midbrain extracts (Figure 5, G and I). These increases in pS129 positivity were not observed after 5b treatment (Figure 5I). Testing for 5b treatment effects in Gba1 (L444P+E326K, combined) mice showed a significant decrease in pS129/total αS in brain cortex (P < 0.05, Figure 5G). In addition, dosing of 5b stepwise reduced the total pS129 level (Supplemental Figure 3D), and no differences in pS129/total αS were observed between Ctl, L444P-LD-5b, and L444P-5b mice (Supplemental Figure 3E). Together, the data suggest that 5b normalized GCase maturation and improved αS homeostasis in Gba1 mutant mice.

5b SCD inhibitor 5b normalizes GCase maturation, αS T:M ratio and insoluble pS129 positivity in Gba1 mutant mice. (A) Representative WB of GCase and αS using cell-penetrant cross-linker DSG in cortical brain bits lysed with 1% Triton X-100/PBS. 5b normalizes ratio of mature (>62 kDa) versus immature (<62 kDa) GCase (upper blot). 5b restores the αS T:M signal in GBA mutant mice (lower blot). DJ 1 signal serves as a control for crosslinking (dimer/monomer ratio) and loading. (B and C) Representative WBs (noncrosslinked) of sequentially extracted TBS (soluble), RIPA-soluble (insoluble) extracts from cortex and midbrain. (D and E) Quantified for mature/immature GCase and αS 60:14 kDa (T:M) ratio. (F–I) Quantified for TBS/RIPA-αS solubility and pS129/total αS in the insoluble (RIPA) extract in cortex and in midbrain. Data are shown as mean ± SEM. Two-way ANOVA, Tukey’s post hoc test (D and E, left panels). Two-tailed, unpaired 2-tailed t test comparing Gba1 (E326K+L444P) Plb versus 5b (E, G, and I, right panels). *P < 0.05, **P < 0.01.

SCD inhibition reduces pS129+ inclusions, normalizing lysosomal clustering and biogenesis in Gba1 mutant mice. Abnormal, membrane-lipid rich aggregates are found in human PD brain (26), in GBA1-PD patient-derived neurons (50), and in PD-type fly models (64). We consistently observed pS129+ multilaminar membranes and LB-type aggregates in 3K αS mutant mice with PD-like phenotypes (16). To address neuropathological characteristics of pS129+ deposits in Gba1 mutant mice, we searched for PK-resistant αS aggregates in cryostat sections. We applied a monoclonal antibody specific for pS129 and then digested the sections with PK (Figure 6A). Four months of 5b treatment efficiently decreased the build-up of larger-sized PK-resistant pS129 αS granules in Gba1 mutant mice (Figure 6A). The granular patterns of PK-resistant pS129 αS forms were relatively strong in neuronal somata of the cortical layers V and VI in L444P and E326K sections; therefore, these were used for quantification. Only very few and small pS129+ puncta were detected in Ctl mice, and these were similar to 5b-treated Gba1 mutant mice (2-way ANOVA, genotype × treatment; P = 0.0001) (Figure 6B). Testing for 5b effects in Gba1 (L444P+E326K, combined) mice showed a significant decrease in PK-resistant immunoreactive puncta in Plb versus 5b (P < 0.001; Figure 6B).

5b treatment reduces PK-resistant and vesicle/lipid-rich αS aggregates and normalizes lysosomal clustering and biogenesis. (A and B) Representative images of PK-resistant pS129+ aggregates in the cortex of Plb- and 5b-treated Gba1 mice and quantification (n = 2–3 sections, n = 3–4 mice per cohort). Note only background staining was seen in Ctl mice independent of treatment. (C and D) Confocal microscopy of cortical and midbrain (S. Nigra) region labeled with Plin2 (red) and quantification of puncta sizes in cortex and S. Nigra. (E and F) Midbrain sections triple labeled with pS129 (red), LAMP1 (green), and DAPI (blue) and quantification of LAMP1+pS129 clusters. (G and H) Adjacent sections were additionally stained for lysosomal biogenesis marker TFEB and tyrosine hydroxylase and graphs quantify the relative proportion of dopaminergic neurons displaying nuclear TFEB immunolabeling. Data are shown as mean ± SEM. Two-way ANOVA, Tukey’s post hoc test (B, D, F, and H, left panels). Two-tailed, unpaired 2-tailed t test comparing in Gba1 (E326K+L444P) Plb versus 5b (B, D, F, and H, right panels). *P < 0.05, **P < 0.01, ***P < 0.001; ****P < 0.0001. Scale bars: 25 μm.

Since αS also acquires PK resistance when aggregating into LB-type lesions that can include lysosomes, LDs, and other lipid-rich vesicle membranes (26, 65), we next stained sections with the LD membrane marker perilipin 2 (Plin2) (Figure 6C). Notably, Plins have high affinity for membrane lipid packing defects (66), and the Plin2 affinity toward LDs covered with MUFA C18:1 is higher than with saturated FA (67). The Plin+ LDs are shuttled to the lysosome, where Plins are extracted (e.g., by cathepsin B) from LD membranes, enabling lipophagy of the LD content (68, 69). When quantifying the relative Plin2 cluster sizes in cortical (frontal cortex, layers V–VI) and midbrain (DAergic) neurons, a significant increase was detected in midbrain of both Plb L444P and E326K Gba1 versus Ctl, and these Plin2+ puncta were normalized after 5b treatment (Figure 6D; 2-way ANOVA, genotype × treatment, P < 0.05; for full-sized images, see Supplemental Figure 4. When quantifying the 5b treatment effects in Gba1 (E326K+L444P, combined) mice, we detected a significant decrease in Plin2+ puncta between Plb and 5b in cortex (P < 0.05; Figure 6D). We previously observed enlarged LAMP1+pS129 clusters in the DAergic neurons of 3K αS mice that were reduced by 5b treatment (45). The LAMP1+pS129 clusters were increased by Gba1 mutation and normalized by 5b treatment in Gba1 mutant 5b versus Plb across genotypes (Figure 6, E and F). Ctl mice only showed background staining when using the pS129 antibody (Supplemental Figure 5) and, therefore, were excluded from this analysis.

Nuclear TFEB is considered the master regulator of autophagy and lysosomal function and augments the expression of autophagy-lysosomal genes, thereby contributing to the degradation of cytoplasmic lipidic material (70–72). Time-laps experiments have shown that, under normal (fed) conditions, TFEB continuously shuttles between the cytosol and the nucleus (73). We investigated whether lipidic aggregates in Gba1 mutant mice associate with a change in the cytosolic/nuclear TFEB immunoreactivity in midbrain DAergic neurons (Figure 6G). Two-way ANOVA revealed a significant interaction (genotype × treatment; P < 0.01) with similar nuclear reactivity in 5b-treated mice (Figure 6H). Analyzing the 5b-treatment effect on Gba1 (L444P+E326K) showed that 5b significantly raised the level of nuclear TFEB (P < 0.01) (Figure 6H). Together, these data suggest that 5b treatment normalized the lipid-rich, pS129+ lysosomal aggregates and promoted lysosomal biogenesis and distribution in Gba1 mutant mice.

5b treatment decreases lipid unsaturation, normalizing phospholipid membrane homeostasis in Gba1 L444P and E326K mouse brain. To determine the specific lipids in which the C16:1 and C18:1 species were reduced, we performed a focused species and subspecies analysis of the mouse brain lipidome of the 16-month-old Plb- and 5b-treated mice (Figure 7). We first focused on C16:1 and C18:1 containing phospholipids with an average increase of > 20% in Plb L444P mice relative to Ctl and further selected for those decreased by > 20% following 5b treatment. As expected, 5b treatment reduced several C18:1- and C16:1-containing lipids, predominantly PE species, as well as PC, PI and LPC species in L444P mice (Figure 7A). There was a striking change in these monounsaturated species in ether lipids including phosphatidylethanolamine-o (PE-O) as well as phosphatidylcholine-o (PC-O) and lysophosphatidylethanolamine-o (LPE-O) (Figure 7B). The same analysis was performed comparing E326K versus Ctl mice with or without 5b treatment, and this resulted in the identification of a larger set of phospholipids being identified as increased by the Gba1 mutation and reduced by 5b treatment. The phospholipid headgroups were PE, phosphatidic acid (PA), PC, and phosphatidylserine (PS), among others (Figure 7C). Notably, there was enrichment of these MUFAs in diacylglycerides, and this was also reduced by 5b treatment (Figure 7C). Similar to our observations in mice with the L444P mutation, mice with the E326K mutation also had an increase in PE ether lipid species, and a subset of these was reduced by 5b treatment including many in the PE-O class (Figure 7D).

SCD inhibition (5b) treatment decreases lipid unsaturation in Gba1 mutant mice. Lipidome data (pmol) were analyzed with a focus on lipids meeting the following criteria: inclusion of C16:1 or C18:1 fatty acids and a minimum of n = 3 replicates. Lipid species with an average increased of > 20% in mice with a GBA mutation and decreased by > 20% by 5b treatment were selected for heatmaps. (A) Heatmap shows C16:1- and C18:1-containing phospholipid species (pmol) increased in GBA L444P mice relative to control mice with > 20% decrease upon 5b treatment. Heatmaps depict median values with each plot representing a minimum n = 3 per condition. (B) Data per A but showing ether lipids. (C) Data per A for Gba1 E326K mice versus controls. Two DAG species are shown in addition to phospholipids. (D) Data per B showing ether lipids.