Impact of Gba2 on neuronopathic Gaucher’s disease and α-synuclein accumulation in medaka (Oryzias latipes)

Homozygous mutations in the lysosomal glucocerebrosidase gene, GBA1, cause Gaucher’s disease (GD), while heterozygous mutations in GBA1 are a strong risk factor for Parkinson’s disease (PD), whose pathological hallmark is intraneuronal α-synuclein (asyn) aggregates. We previously reported that gba1 knockout (KO) medaka exhibited glucosylceramide accumulation and neuronopathic GD phenotypes, including short lifespan, the dopaminergic and noradrenergic neuronal cell loss, microglial activation, and swimming abnormality, with asyn accumulation in the brains. A recent study reported that deletion of GBA2, non-lysosomal glucocerebrosidase, in a non-neuronopathic GD mouse model rescued its phenotypes. In the present study, we generated gba2 KO medaka and examined the effect of Gba2 deletion on the phenotypes of gba1 KO medaka. The Gba2 deletion in gba1 KO medaka resulted in the exacerbation of glucosylceramide accumulation and no improvement in neuronopathic GD pathological changes, asyn accumulation, or swimming abnormalities. Meanwhile, though gba2 KO medaka did not show any apparent phenotypes, biochemical analysis revealed asyn accumulation in the brains. gba2 KO medaka showed a trend towards an increase in sphingolipids in the brains, which is one of the possible causes of asyn accumulation. In conclusion, this study demonstrated that the deletion of Gba2 does not rescue the pathological changes or behavioral abnormalities of gba1 KO medaka, and GBA2 represents a novel factor affecting asyn accumulation in the brains. Supplementary Information The online version contains supplementary material available at 10.1186/s13041-021-00790-x.

Interestingly, a previous study reported that the deletion of GBA2 rescued the visceral manifestations, such as hepatosplenomegaly, cytopenia and osteopenia, in a GD1 mouse model through a reduction in sphingosine [30]. In addition, the deletion of GBA2 also rescues the CNS phenotypes, such as neuronal cell loss, motor coordination and lifespan, in the Niemann-Pick Type C (NPC) mouse model, which shows a decreased GBA1 activity in the brain [31]. These findings raise the possibility that GBA2 works as a novel factor affecting the CNS pathology in GBA1-related disorders, including nGD and PD.
GBA1 and GBA2 are widely conserved from Caenorhabditis elegans to humans, as well as in medaka, mice and rats. We previously reported that gba1 KO medaka survived long enough to perform pathological analysis of the disease progression, in contrast to the perinatal death observed in Gba1 KO mice [32,33]. gba1 KO medaka develop swimming abnormalities at 2 months post-fertilization (mpf ) and start to die from 3 mpf. Medaka have endogenous asyn, enabling the observation of the effects of genetic mutations or pharmacological intervention on the dynamics of asyn in the brains. We showed that gba1 KO medaka displayed not only the dopaminergic (DA), noradrenergic (NA) and serotonergic neuronal cell loss, neuroinflammation, and ALP dysfunction, but also asyn accumulation in the brains [33]. gba1 KO medaka are a useful model animal for investigating the pathological mechanisms underlying the CNS pathology in GBA1related disorders.
In the present study, we generated gba2 KO medaka and analyzed gba2 single-and gba1/gba2 double-KO (DKO) medaka to investigate the contributions of GBA2 to nGD. We found that the deletion of Gba2 did not rescue the CNS phenotypes of gba1 KO medaka but resulted in the perturbation of sphingolipid metabolism and an increase in the amount of asyn. The present study provides novel insights into the pathological role of GBA2 in nGD and asyn accumulation in the brains.

Generation of gba2 KO and gba1/gba2 DKO medaka
The ethics statement and maintenance of medaka were described previously [33]. Medaka experiments were approved by the Animal Experiments Committee of Kyoto University and conducted in accordance with national guidelines. Medaka were maintained in an aquaculture system with recirculating water at 27 °C in a 14-h light/10-h dark cycle.
Medaka of the Kyoto-cab strain, a substrain of Cab, were used in this study. The generation and characterization of GBA1-deficient medaka were reported previously [33]. GBA2-deficient medaka were generated using a clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated 9 (Cas9) system, as reported previously [34]. In brief, the cDNA sequence of medaka gba2 was determined by reverse transcriptionpolymerase chain reaction and rapid amplification of cDNA ends. The medaka gba2 gene consists of 18 exons encoding 858 amino acids. The crRNAs were designed using the CRISPR design tool (http:// viewer. shigen. info/ cgi-bin/ crispr/ crispr. cgi), and the following crRNA was used: 5′-GGA GGG CAA AGC ACT GTC GGGGG-3′. The crRNA and tracrRNA were constructed by Fasmac Co. (Kanagawa, Japan). The Cas9 RNA was synthesized from pCS2 + hSpCas9 vector (Addgene #51815) using mMessage mMachine SP6 Kit (Thermo Fisher Scientific, Waltham, MA, USA). The RNA mixture was injected into single-cell-stage embryos. The injected founders (F 0 ) were raised to sexual maturity and back-crossed with wild-type (WT) to generate F 1 s. The gba2 gene of F 1 s was sequenced, and novel heterozygous gba2 mutation (gba2 +/− ) medaka with 21 bases deleted and 2 bases inserted into exon 5 were obtained (Additional file 1: Fig. S1). These deletions and insertions resulted in a frame shift mutation, leading to the deficiency in protein expression and enzymatic activity of Gba2 in the brain. Off-target candidates were searched for using the Medaka pattern match tool (http:// viewer. shigen. info/ medak avw/ crisp rtool/). No alterations were found in three off-target candidates located on exons.
gba2 +/− medaka were back-crossed with WT medaka at least five times and then crossed with GBA1-deficient medaka to create gba1/gba2 DKO medaka. Medaka brains collected by surgery were directly snap-frozen in liquid nitrogen and then stored at − 80 °C until use. The gba2 +/− medaka and gba2 KO medaka were used in the previously reported study [35].

The GBA2 enzymatic activity assay
The measurement of GBA2 enzymatic activity was carried out as described previously [36,37]. Medaka brains were homogenized in 200 µl deionized water and centrifuged at 500×g at 4 °C for 10 min. The supernatant was collected and centrifuged at 20,000×g at 4 °C for 20 min. The pellet was rinsed with 200 µl of 50 mM potassium phosphate buffer, pH 5.8 and centrifuged at 20,000×g at 4 °C for 15 min. This step was repeated twice. Next, 30 µl of 50 mM potassium phosphate buffer, pH 5.8 was added to the pellet, and the pellet was resuspended. The resulted suspension was used for the enzyme assay.
The reaction mixture contained 10 µl of the suspension and 20 µl of 4.5 mM 4-methlumbeliferyl β-dglucopyranoside (Wako, #324-37411) in 100 mM citric acid and 200 mM disodium hydrogen phosphate buffer, pH 5.8. The reaction mixture was incubated at 37 °C for 60 min with or without 0.3 mM NB-DGJ (N-deoxygalactonojirimycin, #B690500; Toronto Research Chemicals Inc., Toronto, Canada). The reaction was terminated by adding 200 µl of 0.5 M sodium carbonate buffer at pH 10.7, and the fluorescence (excitation 55 nm and emission 460 nm) was measured by Fluoroskan Ascent FL (Thermo Fisher Scientific). The GBA2 enzymatic activity was measured as the GCase activity sensitive to NB-DGJ.

Locomotor function analysis
The medaka locomotor function analysis was performed as described previously [33]. Medaka were transferred to a 20-cm-diameter tank filled with water to a depth of 2 cm at room temperature. After 5 min of rest, the freeswimming distance for 5 min was analyzed using an ethovision XT 5 software (Noldus, Leesburg, VA).

Confirmation of the cross-species reactivity of antibodies
To confirm the cross-species reactivity of the antibodies used in present study against medaka, the homology search for β-actin, LC3, NSE and p62 genes was performed (https:// asia. ensem bl. org/ Oryzi as_ latip es/ Info/ Index), showing high homology of medaka β-actin, LC3, and NSE to the peptides used to raise each antibodies (93%, 93%, and 84%, respectively). Meanwhile, medaka p62 has only 45% homology to the peptide used to raise the p62 antibody. Therefore, we confirmed the crossspecies reactivity of the antibody by overexpression of medaka p62 in cultured cells. The medaka p62 cDNA was cloned into pcDNA3 and expressed in HEK293T cells. A strong signal was detected at the molecular weight close to that of medaka p62 in the lanes of medaka p62-overexpressing HEK293T cells and gba1 KO medaka brains, but not in the lane of mock transfected cells (Additional file 2: Fig. S2). The following is the brief description about this experiment. The sequence of medaka p62 cDNA was determined previously [33]. The full-length medaka p62 cDNA was inserted into pcDNA3 to generate a medaka p62-expressing vector. Transfection of HEK293T cells (RIKEN Cell Bank, Tsukuba, Japan) with vector alone (pcDNA3-mock) or vector containing full-length medaka p62 (pcDNA3p62) was performed using polyethyleneimine MAX (#24765, Polysciences, Warrington, PA, USA) according to the manufacturer's instructions. The cell lysates were obtained with RIPA buffer. The SDS-PAGE and chemiluminescent signal detection was performed as described above. The amount of loading protein was 5 μg for WT and gba1 KO medaka brains and 0.05 μg and 0.25 μg for cell lysates, respectively.

Immunohistochemical analysis
Paraffin sections were used for immunohistochemical analysis as reported previously [33]. Medaka brains were collected by surgery. The collected brains were fixed with 4% (w/v) paraformaldehyde in PBS at 4 °C for 1 day and stored in 70% ethanol until use. The fixed samples were dehydrated and embedded in paraffin using Surgipath FSC22 (Leica, Wetzlar, Germany), and sections were acquired using a Microm HM 325 (Thermo Fisher Scientific). The thickness of the sections was set at 20 µm for tyrosine hydroxylase (TH) -positive cell counting and 8 µm for other analysis. The following antibodies were used as the primary antibody: anti-TH (#MAB318, 1:1000; Merck Millipore, Burlington, MA) and antimedaka asyn (1:2000) [33]. The sections were incubated with the primary antibody at 4 °C for 1 day after blocking with 4% skim milk. Histofine (#414322; Nichirei Bioscience Inc., Tokyo, Japan) was used as the secondary antibody for diaminobenzidine staining.

Cell counting
The number of DA neurons in the middle diencephalon and NA neurons in the locus coeruleus were counted as previously described [33]. The numbers of TH-positive neurons with visible nuclei were counted under the microscope (Cx41; Olympus, Tokyo, Japan).

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)
RNA was isolated from medaka brains using Qiazol (QIAGEN). cDNA was synthesized using the Prime-Script RT reagent kit of Perfect Real Time (#RR037A; Takara, Kyoto, Japan). The quantification of cDNA was performed with the LightCycler 480 using LightCycler 480 SYBR GreenI Master (#04887352001; Roche Diagnostics, Mannheim, Germany). The following primer sets were used: TNF-α:
For the analysis of GlcCer, galactosylceramide (Gal-Cer), GlcSph and galactosylsphingosine (GalSph: psychosine), the neutralized reaction mixture was subjected to Folch's partition, and the lower phase was dried under a flow of N 2 gas. For the analysis of ceramide and sphingosine, the neutralized reaction mixture was dried under a flow of N 2 gas. The resulting lipid films were suspended in C:M (2:1, v/v) at a concentration of 100 µg frozen tissue/µl, and aliquots were subjected to LC-ESI-MS/MS.

LC-ESI-MS/MS for the sphingolipid profile analysis
LC-ESI-MS/MS was performed on an LC system Nexera X2 (SHIMADZU, Kyoto, Japan) attached to a triple-quadrupole linear ion trap mass spectrometer (QTRAP4500; SCIEX, Tokyo, Japan). The LC-ESI-MS/ MS datasets were analyzed with the MultiQuant ™ (ver. 2.1) and Analyst (SCIEX) software programs. Target lipids were monitored in multiple reaction monitoring (MRM) mode using specific precursor-product ion pairs, as detailed in Additional file 3: Table S1. Peak areas were integrated and quantified relative to the associated internal standard.

Statistical analysis
For comparison of two groups, at two-tailed unpaired Student's t-test was performed. An F test was performed to evaluate the differences in variances. For comparison of three or more groups, one-way ANOVA with Tukey's post-hoc test or Newman-Keuls multiple comparison test was performed. A Brown-Forsythe test was performed to evaluate the differences in variances. Differences with p values of less than 0.05 were considered significant. Statistical calculations were performed with GraphPad Prism Software (GraphPad) Version5.0 and 7.04.

gba2 KO medaka show no apparent phenotypes
First, we attempted to create gba2 KO medaka using a CRISPR/Cas9 system. We successfully generated gba2 nonsense mutant medaka via the deletion of 21 bases and insertion of 2 bases in exon 5, which resulted in a frame shift mutation (Additional file 1: Fig. S1).
There were 15 off-target candidates of crRNA used in this study, among which 3 were located on exons (Additional file 4: Table S2). We conducted direct sequencing and confirmed no mutations around these three off-target candidates in medaka carrying heterozygous mutations of the gba2 gene.
We then analyzed the phenotypes of gba2 KO medaka after heterozygous gba2 mutant medaka were backcrossed with WT medaka more than five times. gba2 KO medaka showed almost no enzymatic activity or protein expression of Gba2 in the brains, and gba2 +/− medaka showed ~ 58% of enzymatic activity and ~ 42% of protein expression level compared with WT medaka (Fig. 1a and Additional file 5: Fig. S3). The lifespan of both gba2 KO and gba2 +/− medaka were the same as that of WT medaka (Fig. 1b).
We next examined their locomotor function, as the loss of GBA2 causes locomotor dysfunction in mice and zebrafish [25,29]. In contrast to GBA2-deficient mice and zebrafish, gba2 KO medaka showed a normal locomotor function at 6 mpf (Fig. 1c). Such differential effects of GBA2 loss between medaka and mice are seen also in spermatogenesis, as Gba2 KO mice show infertility due to morphological abnormality of sperm, called globozoospermia, while gba2 KO medaka showed a normal fertilization rate (Fig. 1d). In summary, gba2 KO medaka did not show any apparently abnormal phenotypes.

Deletion of Gba2 in gba1 KO medaka does not rescue the CNS phenotype
The compensatory changes of GBA2 in GBA1 deficiency remain unclear given the previous studies showing conflicting data [18,28,[41][42][43]. Therefore, we examined Gba2 protein expression levels and enzymatic activity in Fig. 1 Phenotypes of gba2 KO medaka. a The Gba2 enzymatic activity in the brains of medaka. The Gba2 activity was almost eliminated in gba2 KO medaka and reduced by 42% in gba2 +/− medaka (n = 6 for each genotype). A one-way ANOVA with Tukey's multiple comparison test was performed. b A Kaplan-Meier plot showed no significant differences in the survival among genotypes (n = 24-33 for each genotype). A Log-rank (Mantel-Cox) test was performed. c Analysis of the spontaneous swimming movement. gba2 KO medaka showed a normal locomotor function compared with WT medaka at 6 mpf (n = 6 for each genotype). A two-tailed unpaired Student's t-test was performed. The bottom images are representative movement tracks for each genotype. d Fertilization rate analysis showed no marked differences among genotypes at 6 pmf (n = 3 for each group). A one-way ANOVA with Tukey's multiple comparison test was performed. n.s. not significant, ***p < 0.001. The bars indicate the mean ± SEM the brains of WT and GBA1 deficient medaka, showing no differences in protein expression levels or enzymatic activity among the gba1 genotypes (Additional file 6: Fig.  S4a and b). To examine whether or not GBA2 functioned as a novel factor affecting nGD, we crossed gba2 mutant medaka and gba1 mutant medaka to generate gba1/gba2 DKO medaka. Recent studies have reported that the deletion of GBA2 in a GD1 mouse model and a NPC mouse model improved the phenotypes [30,31]. However, gba1/gba2 DKO medaka showed a shorter lifespan than gba1 KO medaka (Fig. 2a). In addition, gba1/gba2 DKO medaka did not show improvement of abnormal swimming movement seen in gba1 KO medaka [33] (Additional files 7, 8, 9, 10: video S1-4).
We previously reported that gba1 KO medaka showed the DA, NA and serotonergic neuronal cell loss, increased inflammatory cytokine levels and microgliosis similar to nGD patients [33]. We then evaluated the neuronal cell Fig. 2 Deletion of Gba2 in gba1 KO medaka did not rescue their phenotypes. a A Kaplan-Meier plot showed that the deletion of Gba2 in gba1 KO medaka shortened the lifespan (n = 15-41 for each genotype). A Log-rank (Mantel-Cox) test was performed. b Representative TH immunohistochemical staining of the middle diencephalic dopaminergic neurons in WT, gba1 KO, and gba1/gba2 DKO medaka at 3 mpf. c The number of TH-positive (TH + ) neurons in the middle diencephalon at 3 mpf. The number of TH + neurons was decreased in gba1/gba2 DKO medaka compared with WT medaka but did not markedly differ between gba1/gba2 DKO and gba1 KO medaka (n = 6-10 for each genotype). A one-way ANOVA with Tukey's multiple comparison test was performed. d The number of TH + neurons in the locus coeruleus at 3 mpf. The number of TH + neurons was decreased in gba1/gba2 DKO medaka compared with WT medaka but did not differ markedly between gba1/gba2 DKO and gba1 KO medaka (n = 4-10 for each genotype). A one-way ANOVA with Tukey's multiple comparison test was performed. e qRT-PCR analysis of TNF-α mRNA in the brains of 3 mpf medaka. TNF-α mRNA levels were normalized to β-actin mRNA. The TNF-α mRNA levels did not differ markedly between gba1/gba2 DKO and gba1 KO medaka (n = 6 for each genotype). Dunn's multiple comparisons test was performed. f qRT-PCR analysis of ApoE-b mRNA in the brains of 3 mpf medaka. The ApoE-b mRNA levels were normalized to β-actin mRNA. The amount of ApoE-b mRNA was increased in gba1/gba2 DKO medaka compared with WT medaka but did not differ markedly between gba1/gba2 DKO and gba1 KO medaka (n = 6 for each genotype). A one-way ANOVA with Tukey's multiple comparison test was performed. n.s. not significant, ****p < 0.0001. The bars indicate mean ± SEM Nakanishi et al. Mol Brain (2021) 14:80 loss by counting the number of TH-positive neurons in the middle diencephalon (DA neurons) and locus coeruleus (NA neurons). The numbers of both DA and NA neurons were decreased in gba1/gba2 DKO medaka, similar to gba1 KO medaka at 3 months mpf (Fig. 2b-d).
The amount of DA in the brains showed trend towards decrease in gba1 KO and gba1/gba2 DKO medaka albeit not reaching statistical significance (Additional file 11: Fig. S5). We also examined the neuroinflammation in gba1/gba2 DKO medaka. The mRNA expression of tumor necrosis factor (TNF)-α was not markedly different between gba1/gba2 DKO and gba1 KO medaka at 3 mpf (Fig. 2e). We evaluated the expression of mRNA encoding ApoE-b, a marker for microglia in teleost fish [33,44], and confirmed a similar expression between gba1/gba2 DKO and gba1 KO medaka at 3 mpf (Fig. 2f ). gba1/gba2 DKO medaka also showed the Gaucher celllike cells which had been seen in gba1 KO medaka [33] (Additional file 12: Fig. S6). These results suggested that the deletion of Gba2 does not rescue any CNS phenotypes of gba1 KO medaka.

Perturbation of sphingolipid metabolism in the brains of gba1 and gba2 mutant medaka
To examine why Gba2 deletion did not rescue the CNS phenotypes of gba1 KO medaka, we analyzed the sphingolipid profiles in the medaka brains at 3 mpf using LC-ESI-MS/MS which can measure GlcCer and GalCer separately. We previously reported that gba1 KO medaka showed the accumulation of GlcCer, similar to GBA1deficient human and mice [33]. The sphingolipid analysis showed that the amounts of GlcCer and glucosylsphingosine (GlcSph) were further increased in gba1/gba2 DKO medaka compared with gba1 KO medaka ( Fig. 3a and e). The amount of GlcCer tended to be increased in gba2 KO medaka compared with WT medaka (Fig. 3a   Fig. 3 The amount of sphingolipids was altered in the brains of gba1 and gba2 mutant medaka. a The amount of GlcCer was increased in gba1/ gba2 DKO medaka compared with gba1 KO medaka (n = 3 for each genotype). b The amount of GalCer did not differ markedly between gba1 KO and gba1/gba2 DKO medaka (n = 3 for each genotype). c The amount of ceramide did not differ among genotypes (n = 3 for each genotype). d The amount of sphingosine did not differ markedly between gba1 KO and gba2 KO medaka (n = 3 for each genotype). e The amount of GlcSph was increased in gba1/gba2 DKO compared with gba1 KO medaka (n = 3 for each genotype). f The amount of GalSph was increased in gba1 KO and gba1/gba2 DKO medaka compared with WT medaka (n = 3 for each genotype). A one-way ANOVA with Tukey's multiple comparison test was performed. N.D.: not detected (below the detection threshold). n.s. not significant, *p < 0.05, **p < 0.01 and ***p < 0.001. The bars indicate the mean ± SEM and Additional file 13: Table S3). While the amount of ceramide did not differ among the genotypes (Fig. 3c), that of GalCer was decreased in the absence of Gba1, irrespective of the gba2 genotype (Fig. 3b). In contrast, GalSph was increased in gba1 KO and gba1/gba2 DKO medaka (Fig. 3f ). While the amount of GalSph was below the detection threshold in WT medaka, it was detected in gba2 KO and gba1 ± medaka ( Fig. 3f and Additional file 13: Table S3). GalSph is a substrate of galactocerebrosidase (GALC) and is believed to be the pathological sphingolipid involved in Krabbe's disease (KD) [45]. The amount of sphingosine, which was reported to have cellular toxicity in GD1 [30], was not decreased in gba1/gba2 DKO medaka compared with gba1 KO medaka (Fig. 3d).
In summary, compared with gba1 KO medaka, gba1/ gba2 DKO medaka showed a significant increase in Glc-Cer and GlcSph, and an increasing trend in sphingosine and GalSph in the brains. Although gba2 KO medaka did not show significant increase in the amount of Glc-Cer and GalSph when comparing all groups, it reached statistical significance when compared with WT medaka (Additional file 13: Table S3).

The impact of Gba2 on asyn accumulation
Recent studies have suggested that the perturbation of sphingolipid metabolism causes conformational changes in asyn, resulting in asyn accumulation [16,18,46]. gba1 KO medaka showed the perturbation of sphingolipid metabolism accompanied by asyn accumulation in axonal swellings containing autophagosomes, which we previously determined by electron microscopic analysis [33]. The increase in asyn was seen in the Triton X-soluble fractions, but not in SDS-soluble fractions (Additional file 14: Fig. S7). Compared with WT medaka, gba2 KO medaka showed an increased amount of asyn without any asyn-positive abnormal structures like axonal swellings seen in gba1 KO medaka (Fig. 4a-c). gba1/gba2 DKO medaka showed the similar phenotypes to those of gba1 KO medaka (Fig. 4a-c). The qRT-PCR for asyn mRNA showed no increase in asyn mRNA expression levels in gba2 KO medaka (Additional file 15: Fig. S8). This result suggests that asyn accumulation seen in gba2 KO medaka was the pathological process after the translation. Perturbation of the sphingolipid metabolism may be one of the causes of the asyn accumulation in the brains of gba2 KO medaka. Because both ALP dysfunction and the perturbation of the sphingolipid metabolism presumably leads to asyn accumulation in gba1 KO medaka [33], we also examined the autophagic function in gba2 KO medaka. p62 acts as a linker between microtubule-associated protein 1 light chain 3 (LC3) and poly-ubiquitinated proteins. The expression of p62 correlates with the autophagic function, while the LC3-II/I ratio correlates with the accumulation of autophagosomes [47]. The expression of p62 and the LC3-II/I ratio were not markedly different in the brains of gba2 KO medaka from WT ( Fig. 5a-c). We assumed that ALP was intact in gba2 KO medaka in contrast to gba1 KO medaka.

Discussion
This study revealed that the deletion of Gba2 in gba1 KO medaka did not rescue the phenotypes, including the altered sphingolipid metabolism and asyn accumulation in the brains. However, GBA2 represents a novel factor affecting asyn accumulation because gba2 KO medaka showed an increased amount of asyn in the brains, which was presumably caused by the perturbation of sphingolipid metabolism.
In GD, the accumulation of substrates of GBA1, GlcCer and GlcSph, is thought to play a central role in disease pathogenesis. At present, substrate reduction therapy to reduce these sphingolipids is available for GD1. Although the mechanisms by which GBA1 mutations cause PD are not fully understood yet, the following evidence suggests that a reduced GBA1 enzymatic activity is responsible for the increased risk of developing PD: (1) There is a negative correlation between GBA1 activity and asyn accumulation [17,48]; (2) heterozygous GBA1 mutations with Fig. 5 Autophagic dysfunction in gba1 KO and gba1/gba2 DKO medaka, but not in gba2 KO medaka. a, b Immunoblot analysis of p62 in the brains at 3 mpf. The expression of p62 was increased in the gba1/gba2 DKO and gba1 KO medaka but not in gba2 KO medaka (n = 6 for each genotype). a, c Immunoblot analysis of LC3-I and LC3-II in the brains at 3 mpf. The LC3-II/I ratio was increased in the gba1/gba2 DKO and gba1 KO medaka compared with the WT and gba2 KO medaka (n = 6 for each genotype). A one-way ANOVA with Tukey's multiple comparison test was performed. n.s. not significant, **p < 0.01, and ***p < 0.001. The bars indicate the mean ± SEM lower enzymatic activities confer a higher risk of developing PD than those with higher enzymatic activities [49]; (3) GBA1 overexpression ameliorates asyn accumulation in a PD mouse model [50]. Furthermore, previous studies have shown that a brain-penetrating GlcCer synthase inhibitor and a GBA1 chaperone alleviated pathological asyn accumulation in the brains of PD mouse models [51,52]. These findings suggest that decreased GBA1 enzymatic activity and the resultant substrate accumulation lead to increased risk of developing PD.
In the present study, we focused on another glucocerebrosidase, GBA2, as a novel factor affecting nGD, based on the findings of a previous study that the deletion of GBA2 ameliorated the visceral phenotypes of a GD1 mouse model [30]. However, contrary to our expectation, deletion of Gba2 did not rescue any CNS phenotypes of gba1 KO medaka. In the previous study using the GD1 mouse model, the deletion of GBA2 was presumed to reduce the amount of sphingosine, thereby ameliorating the phenotypes of GD1 mice [30]. However, in the present study, the analysis of the sphingolipids profile revealed that the deletion of Gba2 in gba1 KO medaka further tended to increase sphingosine in the brains, albeit without statistical significance. One possible explanation is that the gba1 KO medaka is a complete deletion of Gba1, whereas GBA1 is only deleted in hematopoietic and mesenchymal lineage cells in Mx1-Cre (+): GD1 mice [30]. Therefore the effect of the GBA2 deletion on sphingolipid metabolism may be different between Mx1-Cre (+): GD1 mice and gba1 KO medaka. Another possible explanation is the existence of an alternative pathway that produces sphingosine other than GBA1 and GBA2; human, rats and medaka, but not mice, possess GBA3. GBA3 hydrolyzes GlcCer and GlcSph to produce sphingosine in gba1/gba2 DKO medaka but not in gba1/gba2 DKO mice. This pathological effect of GBA3 is supported by a report that GBA3 does not affect the pathology of GD1, whereas GD1 patients with GBA3 deficiency showed a tendency to have a milder disease severity [53]. Thus, the failure to rescue the phenotypes of gba1 KO medaka by Gba2 deletion may be relevant to human patients having GBA1/GBA2 mutations. Moreover, it also needs to be considered that most patients with nGD have some residual GBA1 activity unlike gba1 KO medaka. The effect of GBA2 deletion in other animal models of nGD with residual GBA1 activity should be examined in future studies.
The present study reports, for the first time, the extensive sphingolipids profile in the brains of GBA2 KO animals. Compared with WT medaka, the amounts of GlcCer and GalSph were found to be significantly increased in gba2 KO medaka. Accumulating evidence has shown that metabolic abnormalities of sphingolipids are closely associated with the formation of asyn pathology [16,17,46,54]. Under physiological conditions, asyn maintains an equilibrium state between free monomeric and membrane-bound multimeric conformation, which can be disrupted by sphingolipid metabolic perturbation [54][55][56]. Previous studies demonstrated that GlcCer converts the physiological asyn conformers to assembly-state pathological asyn, leading to its accumulation in cultured cells [17,46]. In addition, a recent report indicated that GalSph also promotes asyn aggregation [57,58]. Based on these previous studies, an increased amount of Glc-Cer and/or GalSph presumably causes asyn accumulation in the brains of gba2 KO medaka. One of the biggest differences in asyn accumulation between gba1 KO medaka and gba2 KO medaka was observed by histological analyses. asyn accumulation was localized in axonal swellings in gba1 KO medaka [33], whereas subcellular localization of asyn accumulation in gba2 KO medaka could not be determined by immunohistochemistry (Fig. 4c). This is probably because asyn condensed in the axonal swellings in gba1 KO medaka with ALP dysfunction, whereas asyn diffusely accumulated in gba2 KO medaka. To our knowledge, the present study is also the first report showing the quantitative changes in GalSph in the brains of heterozygous GBA1 mutant animals. Although its synthetic pathway is unclear, GalSph might play some pathological role in GBA1-PD. Based on the findings of the present and previous studies, we have proposed pathological mechanisms of asyn accumulation under gba1 or gba2 KO backgrounds (Fig. 6). Interestingly, it was reported that the GBA2 activity declines with aging in the mouse brain [59]. Given the asyn accumulation in gba2 KO medaka, as shown in the present study, GBA2 may be a novel factor affecting asyn accumulation in the brains. Further studies are required to determine whether the perturbation of sphingolipid metabolism directly causes asyn accumulation in GBA2 KO animals and how the agedependent decline in GBA2 activity contributes to asyn accumulation. It should be noted there is no report that Miglustat (N-butyldeoxynojirimycin), an orphan drug for type 1 GD disease or NPC, which inhibits GBA2 activity as well as glucosylceramide synthase, induced parkinsonism [37,43].
One limitation of the present study is that we failed to detect asyn in the Triton X-insoluble, SDS-soluble fractions in any genotypes partly because of small brains of medaka. Another limitation is that we did not detect pathological asyn aggregates on brain sections due to the lack of antibodies that recognize pathological medaka asyn, such as misfolded or phosphorylated forms of asyn. The other limitation is that we saw a limited synergistic effect on asyn accumulation in gba1/gba2 DKO medaka compared with gba1 KO and gba2 KO medaka (Fig. 4b), which seems against the proposed mechanisms (Fig. 6). One possibility is a ceiling effect of Gba1 and Gba2 deficiency on asyn accumulation, but we don't have any evidence to support it.
In conclusion, the present study demonstrated that Gba2 deletion does not rescue the CNS phenotypes of gba1 KO medaka. gba2 KO presumably increases the amount of asyn in the medaka brain, through sphingolipid metabolic perturbation. Further studies are required to determine whether GBA2 could be a therapeutic target of α-synucleinopathy. Fig. 6 Proposed mechanisms of deletion of Gba1 and Gba2 leading to asyn accumulation. Deletion of Gba1 and Gba2 induces changes in the sphingolipid metabolism. These changes might induce conformational changes in asyn, resulting in asyn accumulation. Furthermore, the deletion of Gba1 causes autophagic dysfunction, resulting in the impairment of asyn degradation and leading to asyn accumulation. The solid and striped arrows indicate the downstream pathways of gba1 KO and gba2 KO, respectively