Rare PSAP Variants and Possible Interaction with GBA in REM Sleep Behavior Disorder

INTRODUCTION

Rapid eye movement (REM) sleep behavior disorder (RBD) is characterized by the enactment of dreams during the REM phase of sleep [1]. In its idiopathic/isolated form (iRBD, presenting before the clinical diagnosis of a neurodegenerative disease), it represents a common prodromal stage of synucleinopathies, including Parkinson’s disease (PD), dementia with Lewy bodies (DLB), and multiple system atrophy (MSA) [1, 2]. Notably, over 80% of iRBD cases convert to a synucleinopathy within 10–15 years [2, 3]. In line with their clinical overlap, iRBD and overt synucleinopathies also share some of their genetic risk factors. For example, iRBD and PD are both associated with GBA variants, which represent one of the most common genetic risk factors for both diseases [4, 5]. GBA variants display an incomplete penetrance in iRBD as well as in PD [4, 5], suggesting that other factors, genetic and/or environmental, contribute to the development of these disorders among GBA carriers.

GBA encodes glucocerebrosidase (GCase), a lysosomal hydrolase whose main function is the degradation of glucocerebrosides into ceramide and glucose, although it has additional substrates [5]. To properly function, GCase requires a co-activator, saposin C (sapC) [6]. This protein is one of the four active domains of a protein precursor, prosaposin, encoded by the PSAP gene. After its synthesis, prosaposin is cleaved by cathepsin D (CTSD) into its functional proteins: saposins A, B, C and D [7, 8]. Saposins are lysosomal cofactors that activate enzymes degrading sphingolipids. Mutations in PSAP have been associated with the accumulation of sphingolipids and with different lysosomal storage disorders (LSD). For example, Gaucher’s disease, an LSD that is typically caused by biallelic mutations in GBA, is also rarely caused by biallelic mutations in the sapC domain of PSAP [7, 9].

Whereas the association of GBA variants with PD is widely accepted, the role played by PSAP in general and sapC specifically in PD remains controversial. Studies in Asian populations suggested an association between PSAP variants and PD [10–14], yet these results did not replicate in Europeans [15–17]. These conflicting results may suggest a possible role played by ethnic differences and/or by the extreme rarity of deleterious PSAP variants, reducing their detection in PD. Despite the clinical, biological, and, possibly, genetic links of PSAP with GBA and PD, the role of PSAP in iRBD has not been investigated. Herein, we analyzed a multi-center cohort of 1,113 iRBD patients and 2,324 healthy controls to evaluate a possible association between rare PSAP variants and iRBD.

METHODS

RESULTS

We identified 59 rare variants within the PSAP region, of which 15 were nonsynonymous and 3 were loss of function (LoF) variants (Supplementary Table 1). The mean coverage was 568X, and a minimum threshold of 30X was applied for variant quality control. To evaluate if rare PSAP variants are associated with iRBD, we performed SKAT-O comparing iRBD patients and healthy controls. There was a nominally significant enrichment of rare PSAP LoF variants (p = 0.018, Q = 1255) in iRBD patients. However, after FDR correction, the results lost statistical significance (p = 0.1, Table 1). Three out of 1,113 iRBD patients (0.3%) carried a rare PSAP LoF variant, while no carriers of LoF variants were found among the controls (0/2324, Supplementary Table 1). In particular, p.Gln260Ter and p.Glu166Ter are both stop variants located, respectively, within the sapB and between sapA and sapB domains, therefore the sapC domain is not translated. The p.332_333del mutation is an in-frame deletion located within the sapC domain.

Given the interplay between sapC and GCase, we examined whether any of these three iRBD patients with PSAP LoF mutations also carry a GBA variant. Furthermore, we tested the presence of GBA copy number variants (CNVs), as was done previously [29]. We found that two of the patients, carrying the p.Glu166Ter and p.332_333del variants, also carried a GBA variant: p.Arg349Ter and p.Glu326Lys, respectively. None carried GBA CNVs. All PSAP and GBA variants were confirmed by Sanger sequencing.

We further examined the frequency of PSAP LoF variants on gnomAD database v2.11 (https://gnomad.broadinstitute.org). None of the LoF variants found in this study have been reported in gnomAD, and the overall frequency of PSAP LoF variants was extremely low, with a total allele count of high-quality LoF variants of 10 in 141,456 individuals (∼0.007%, compared to ∼0.3% in the iRBD cohort). With a frequency of ∼5% in the general European population for GBA variants (based on gnomAD data), the estimated combined carrier frequency of both LoF PSAP variants and GBA variants is 0.00035%, compared to 0.2% observed in the iRBD cohort, more than a 500-fold difference.

In silico structural analyses

To evaluate the impact of the three iRBD-associated variants on the structure and function of the saposin chains we performed in silico analyses. The p.Glu166Ter variant, located between saposin chains A and B, would result in the termination of expression for chains B-D. The p.Gln260Ter variant is located towards the C-terminus of the sapB domain and would result in the deletion of its C-terminal helix (Fig. 1A), as well as in the termination of sapC and sapD translation. This deletion would also unfold sapB and prevent its dimerization, which is critical for binding lipids [30]. Finally, the variant p.332_333del is located in a linker between helices 1 and 2 of sapC (Fig. 1B) [31]. This shortened linker would prevent the formation of stabilizing contacts between these helices and thus interfere with its ability to bind membranes and GCase. Therefore, all three variants result in a loss of function of the sapC chain.

OPEN IN VIEWER

Fig. 1

Structural analysis of the saposin B and C domains. A) Crystal structure of the saposin B dimer (green and cyan, pdb 1n69). A bound phosphatidylethanolamine (PE) is shown in violet. The C-terminal helix (a.a. 260-273) deleted in the p.Gln260Ter variant is shown in magenta. B) Solution NMR structure of the human saposin-C domain (cyan, pdb 1m12). In-frame deletion of amino acids Asn332 and Lys333 is shown in magenta.

DISCUSSION

In this study, we found three iRBD patients with extremely rare PSAP LoF variants, not reported on gnomAD, while no controls were found with LoF variants. Interestingly, two of the three PSAP LoF variant carriers also carried a GBA variant. While the enrichment of rare PSAP LoF variants in iRBD was only nominally significant, given their rarity it is plausible that this reflects a real association. Furthermore, assuming that in the general European population the carrier frequency of GBA variants is about 5%, and the carrier frequency of LoF variants (based on gnomAD) is about 0.007%, the probability to carry both a GBA variant and a PSAP LoF variant is 0.00035%. In the iRBD cohort, the carrier frequency of both was ∼0.2%, suggesting that this is likely not due to chance alone. The deleteriousness of the three PSAP LoF variants was further exemplified by structural analyses (Fig. 1A, B). All iRBD patients met the MDS criteria for probable prodromal PD (Table 2).

Although the role of PSAP in iRBD and in synucleinopathies in general is still controversial, this study provides the first evidence for a possible role of PSAP variants in iRBD. The lack of a statistically significant enrichment in iRBD patients after correction for multiple comparisons can be explained by the extreme rarity of PSAP variants, resulting in insufficient power. The Residual Variation Intolerance Score (RVIS) of PSAP is –1, putting it in the top 8.47% of genes in the human genome which are intolerant to genetic variance, especially for LoF variance (FDR corrected p = 0.00037 for the observed versus expected number of LoF variants - http://genic-intolerance.org/Search?query=psap).

Two iRBD carriers of PSAP LoF variants were also carriers of a GBA variant. Given the incomplete penetrance of GBA in iRBD, the presence of potentially pathogenic variants in PSAP among GBA carriers may suggest oligogenic inheritance and that PSAP variants might act as genetic modifiers of risk in GBA-iRBD. This is in line with the biological link between sapC and GCase [6, 9]. In particular, it is possible that an impairment of the sapC-mediated activation of GCase contributes to an increased risk to develop iRBD in GBA variant carriers. These hypotheses require additional genetic and functional studies. We cannot rule out that the co-occurrence of GBA and PSAP variants is a coincidence, due to chance alone. However, the fact that two out of three extremely rare PSAP LoF variant carriers also carried a GBA mutation makes a coincidental association less likely.

It is still unclear whether PSAP mutations alone can increase the risk of iRBD or PD. It is possible that LoF of sapC, as seen in our patient with the p.332_333del mutation, will result in reduced activation of GCase and be an independent risk factor. On the other hand, it is also possible that PSAP variants might lead to iRBD through mechanisms independent of GBA. A possible mechanism can be due to an impairment of CTSD and progranulin (PGRN) activity, as previously hypothesized in PD [8]. PSAP, CTSD and PGRN interact in a network involved in lysosomal homeostasis and clearance of alpha-synuclein. PSAP dysfunction might lead to decreased transport of PGRN into the lysosome, reduction of the pro-CTSD conversion into active CTSD, and consequently to impaired lysosomal trafficking and degradation of deleterious or overrepresented proteins, such as alpha-synuclein [8].

This study has several limitations. Age and sex differed between patients and controls. However, this difference would generally lead to false negative results (as young mutation carriers still would not develop the disease), and is, therefore, less likely to affect our results, as no carriers were found in the controls. Although this study was performed in the largest genetic cohort of iRBD patients worldwide, the sample size may still be insufficient to detect extremely rare variants in PSAP. Finally, we were able to find PSAP LoF variants, different from each other, in only 3 iRBD patients. However, the absence of such variants in the ∼twofold larger control group and in the ∼140-fold larger gnomAD control population suggests that this finding might not be random.

Further studies in larger cohorts and functional analyses will be required to clarify the role of PSAP variants in iRBD and alpha-synuclein physiopathology. In addition, studies in other populations, such as East Asians, where PSAP variants have already been proposed as PD risk factors [10–14], will be necessary to further explore differences in the genetic underpinnings of synucleinopathies between different ethnic groups.