smpd1-/- zebrafish display abolished acid sphingomyelinase activity and marked sphingolipid accumulation
We identified a single smpd1 orthologue in zebrafish (ENSDARG00000076121) with 59% shared identity to the human SMPD1 gene at both the DNA and the protein level. CRISPR/Cas9 technology was used to generate a smpd1 stable mutant line (smpd1-/-). The selected mutant allele contained a 5bp deletion and 136bp insertion within exon 3, resulting in a frame-shift and the generation of a premature stop codon (Figure 1A and Supplemental Figure S1). Enzymatic activity of ASM in smpd1-/- at 5 dpf was reduced by 93% (p=0.006, Figure 1B). The large reduction in ASM enzymatic activity resulted in a significant increase of key glycolipid substrates in the smpd1-/- larvae already at 5 dpf (Figure 1C).
Combined ASM and GCase deficiency synergistically increases sphingolipid metabolites
We had previously reported marked sphingolipid accumulation in gba1-/-zebrafish [8]. We hypothesised that combined (enzymatic) GCase and ASM deficiency would synergistically increase distinct sphingolipid subtypes. Using mass spectrometry, a comprehensive panel of glycolipid substrates was analysed in the brains of gba-/- and smpd1-/- single mutant as well as in gba-/-;smpd1-/-double mutant zebrafish and WT controls at 12 weeks of age. A marked additive effect of combined GCase and ASM deficiency was observed for Glucosylceramide levels (the direct substrate of GCase) (Figure 2A). Combined GCase and ASM deficiency also resulted in an additive effect on lactosylceramide, ceramide and sphinganine levels (Figure 2B-D). Sphingosine levels were increased in gba-/-;smpd1-/- compared to WT, reflecting an increase compared to gba1-/- but not compared to smpd1-/- (Figure 2E). Unexpectedly, there was no synergistic effect in sphingomyelin levels in the gba-/-;smpd1-/- double mutants (Figure 2F).
The inflammation markers chitotriosidase and β-Hexosaminidase are markedly increased in the serum of GD patients and used as biomarkers to monitor disease activity [16]. We previously observed a marked increase in chitotriosidase and β-hexosaminidase activity in gba1-/-zebrafish brain tissue at 12 weeks [8]. As key GCase substrates were synergistically increased in gba-/-;smpd1-/- double mutant zebrafish, we investigated whether combined GCase and ASM inactivation may also result in a further increase chitotriosidase and β-hexosaminidase activity. Unexpectedly, gba-/-;smpd1-/- double mutant zebrafish displayed a similar increase in chitotriosidase and β-hexosaminidase activity compared to gba1-/- (Supplemental Figure 2), suggesting persistent yet unaltered neuroinflammatory states in the double mutants despite a marked synergistic increase in sphingolipid metabolites.
ASM deficiency unexpectedly prolongs survival in GCase deficiency
The marked additive effect of combined GCase and ASM deficiency on sphingolipid levels led us to hypothesize that ASM deficiency would further worsen the motor phenotype and shorten survival in gba-/-;smpd1-/- double mutant zebrafish. Unexpectedly, genetic inactivation of ASM led to a complete rescue of this behaviour in the gba-/-;smpd1-/- double mutant zebrafish (Supplemental Video S1 (WT), S2 (smpd1-/-), S3 (gba1-/-) and S4 (gba-/-;smpd1-/-)). Importantly, lifespan was also markedly increased by 22% in gba1-/-;smpd1-/- double mutant zebrafish compared to gba1-/- (median survival of 102 dpf in gba1-/- and 125dpf in gba1-/-;smpd1-/-, p=0.0055, Figure 3A).
RNAseq base pathway analysis confirms restored neuronal health in gba1-/-;smpd1-/-
We next applied RNAseq-based pathway analysis to further elucidate the underlying mechanisms of the observed rescue effect. The differential gene expression analysis all four genotypes (wild type, gba1-/-and smpd1-/- single mutants, gba1-/-;smpd1-/-double mutants) identified a total of 512 genes which were dysregulated in gba1-/- but rescued in gba1-/-;smpd1-/- . Amongst these, there are a downregulation of 219 genes and an upregulation of 293 genes in gba-/- compared to wild-type and gba1-/-;smpd1-/- (adjusted p-value ≤0.05, |log2Fold change| ≥1). The five most up- or down-regulated genes are listed in Supplementary Table 1 and Table 2. We next employed ClusterProfiler analysis on gene ontology (GO) categories to identify functionally relevant pathways within the rescued gene sets. Key neuronal pathways including the GO terms for synaptic signalling, chemical synaptic transmission and calcium ion regulated exocytosis were markedly downregulated in in gba-/- but normalized in gba1-/-;smpd1-/- (Figure 3B). This suggests that key aspects of neuronal function were restored in the gba1-/-;smpd1-/- double mutants.
We also observed an enrichment of upregulated genes in gba1-/-compared to gba-/-;smpd1-/- in a broad range of GO terms, the top 5 of which thought to regulate muscle function. However, since our RNA-seq analysis was carried out on brain tissue, we consider these changes to be of limited relevance only (Suppl Figure 3). Upregulation of the inflammatory signature in gba1-/- was retained in the gba-/-;smpd1-/- but not further enhanced (data not shown).
As both GCase and ASM are lysosomal hydrolases, we specifically focused on the effect of isolated GCase deficiency in gba1-/-compared to combined GCase and ASM deficiency in gba1-/-;smpd1-/- on lysosome transcriptomic pathways. Gene set enrichment analysis led to the identification of 27 leading-edge, dysregulated lysosomal genes, which account for the pathway’s enrichment signal. The expression of these 27 lysosomal genes was increased in gba1-/-compared to wild-type and gba1-/-;smpd1-/- (Fig3C, Supplementary Table 3). Amongst these 27 genes, acid hydrolases contributed the most. Cathepsin L, involved in the initiation of protein degradation, ranked as the top rescued gene. The apparent normalisation of lysosomal gene expression profiles in gba-/-;smpd1-/- was in contrast to the observed marked increase in a wide range of sphingolipid levels in gba-/-;smpd1-/- compared to gba-/- or smpd1-/- single mutants (see above).
We had previously observed marked mitochondrial dysfunction in gba1-/-. We therefore also focussed on the analysis of mitochondrial genes involved in the oxidative phosphorylation pathway. This leading-edge mitochondrial gene subset included 16 genes encoding the subunits of the Complex I, II, IV and V in mitochondrial electron transport chain (Supplementary Table 4). Interestingly, gene set enrichment analysis showed an upregulation of this mitochondrial gene subset in gba1-/-, presumably as a compensatory mechanism to the impaired function of the mitochondrial respiratory chain, but similar mitochondrial gene expression levels in wild-type and gba1-/-;smpd1-/-(Figure 6B).
Reciprocal restoration of mitochondrial function in gba1-/-;smpd1-/-
We next compared the mitochondrial respiratory chain function across all four genotypes to further determine whether the normalised gene expression levels for oxidative phosphorylation-related genes would be reflected in normalised mitochondrial function. Complex I activity was reduced by 65% in smpd1-/- compared to WT levels (p=0.0198, Figure 4A) but restored to 92% of WT levels in gba1-/-;smpd1-/- (p=0.0445, Figure 3A). Complex II was not significantly altered in any of the genotypes (Fig. 4B). Complex III activity in gba1-/- was reduced by 45% compared to WT levels (p=0.0091, Figure 4C) as previously observed [8]. In contrast, Complex III activity in the gba1-/-;smpd1-/- double mutant zebrafish was reduced by only 9% compared to WT levels and thus considerably less pronounced than the reduction observed in the gba1-/-, suggesting a rescue effect in gba1-/-;smpd1-/-. However, this 36% increase in complex III activity in gba1-/-;smpd1-/- compared to gba1-/- did not reach statistical significance. Complex IV activity was unchanged in smpd1-/- compared to WT, but reduced by 40% in gba1-/- compared to WT as previously reported (p=0.0491, Figure 4D). Remarkably, there was a marked improvement of complex IV activity in gba1-/-;smpd1-/- with an increase in activity of 69% compared to gba1-/- (p= 0.0005, Figure 4D). Thus, there is reciprocal rescue of mitochondrial respiratory chain function – ASM deficiency normalizes mitochondrial respiratory chain dysfunction in gba1-/- and GCase deficiency normalizes mitochondrial respiratory chain function in smpd1-/-. Malfunction of the mitochondrial respiratory chain can result in oxidative stress and subsequent lipid peroxidation. We therefore investigated next whether the observed rescue in mitochondrial function results in reduced oxidative stress-related damage. Lipid peroxidation was increased in gba1-/- brains by 63% above WT levels (p= 0.0214, Figure 4E). As predicted, lipid peroxidation levels were reduced by 70% in gba1-/-;smpd1-/- double mutants compared to gba1-/- and thus effectively normalized (p=0.0094, Figure 4E).
The observed mitochondrial rescue effect is autophagy independent
Inhibition of ASM is known to promote autophagy [17]. We (MG, AS) and others had previously observed impaired autophagy/mitophagy in GCase deficiency [18,19]. We therefore investigated in our previously established human dopaminergic SH-SY5Y GCase-deficient in vitro model whether ASM deficiency may exert its observed rescue effect on mitochondrial dysfunction in GCase deficiency by enhancing autophagy/mitophagy. A siRNA-approach was applied to inactivate SMPD1/ASM, whilst GCase was inhibited with conduritol B-epoxide (CBE) for 10 days [19]. SMPD1 mRNA levels in the absence or presence of CBE were significantly decreased relative to scrambled (scram) control treated cells (p = 0.0003 and 0.0007, Figure 5A). Treatment with CBE alone or in the presence of SMPD1 siRNA significantly decreased GCase activity (79% and 76% reduction, respectively; p < 0.01) relative to scram treated cells. Notably GCase activity was significantly increased in SMPD1 knock down cells, compared to scram (p < 0.05, Figure 5B). Macroautophagy flux was measured by quantifying the levels of LC3-II by western blot, a marker for autophagosome number. Under basal conditions, LC3-II levels were similar in all groups. As expected treatment with bafilomycin A1, which prevents the fusion of lysosomes with autophagosomes, caused an accumulation of autophagosomes, and was similar in all groups (Figure 5C-D). These data suggest that there is neither an impairment nor increase in macroautophagy flux under any conditions. α-synuclein is a key driver of PD pathology.
Combined GCase and ASM inhibition results in lowered α-synuclein levels
Given the effect of (isolated) GCase or ASM deficiency on α-synuclein in PD [7,20], we also investigated the effect of combined GCase and ASM deficiency on α-synuclein homeostasis. Unexpectedly, the levels of α-synuclein were significantly decreased (rather than increased as predicted) in SMPD1-KD + CBE-treated cells (p=0.0139, Figures 5E and F), when compared to CBE-treated cells. The decrease in intracellular α-synuclein levels in SMPD1-KD+CBE was not due to increased release of α-synuclein in to the media (Figure 5G) or becoming insoluble (data not shown).