Both homo- (causing autosomal-recessive Gaucherʼs disease; GD) and heterozygous mutations in the glucocerebrosidase gene (GBA) are associated with Parkinson’s disease (PD), and represent the most robust known genetic susceptibility factor identified in PD [1, 2]. Neudorfer et al. [3] in 1996 reported six patients with GD type 1 who developed progressive parkinsonism refractory to conventional antiparkinsonian therapy, although GD patients with some features of parkinsonism were sporadically reported as early as 1939 [4]. In 2004 it has been shown that mutations in the GBA gene were more frequent in PD patients than in controls, and that heterozygous carriers of the GBA mutation might have an increased risk to develop PD [5, 6]. The main confirmation of such association was the extensive multicentric study (5691 PD patients and 4898 controls) that found the odds ratio of 5.43 for detecting any GBA mutation in PD patients compared to the controls [7].

Proposed mechanism for GBA-associated parkinsonism

Glucocerebrosidase (GCase) is a lysosomal enzyme that hydrolyzes β-glycosyl linkage of glucosylceramide (GlcCer) to glucose and ceramide [8, 9]. Different mechanisms have been proposed in attempt to elucidate common pathological links between GBA mutations and PD, including glia-mediated inflammatory response and α-synuclein accumulation. Astrocytosis and microglial activation were described both in human GD brain samples, as well as in Gba-deficient mouse model of neuronopathic GD [10–12]. The latter model demonstrated that glial cells’ activation preceded neuronal loss in affected brain regions and that the level of inflammatory mediators correlated with disease progression [11]. Futhermore, recent reviews provided extensive analysis of the cellular relationship between GCase and α-synuclein (αS) [13–15], but the mechanisms underlying the relation between GBA mutations and the development of PD is still not clear. There is clinical, genetic and experimental evidence for both mutant GBA-mediated loss-of-function or toxic gain-of-function hypotheses, which are not mutually exclusive [1, 16].

The loss-of-function theories mainly focus on altered lipid metabolism and speculate that GBA mutations result in unstable or deficient protein, that, due to a lack of enzymatic activity, contributes to accumulation of GCase substrate, GluCer, within the lysosome, altering the cell membrane sphingolipid composition. Membrane binding is a key biological feature of αS and many functions of αS occur in association with lipid membranes. For example, in Lewy bodies (LB) αS layers are formed around central lipid core [17]. Hence, accumulation of GluCer can change lipid homeostasis, with subsequent alterations in αS processing. In support to this theory, null GBA alleles have been reported in PD patients and chemical inhibition of GBA could lead to accumulation of αS [18], a finding that was later replicated in different models with GBA mutations [16, 19–21]. Even in brains of PD patients without GBA mutations (PD-non-GBA), the reduction of GCase enzymatic activity was associated with increased αS levels [22].

The gain-of-function theories elaborate the effect of misfolded mutant GCase. Most GBA mutations are missense mutations, resulting in a misfolded protein. The presence of mutant GCase was confirmed in a significant proportion of αS inclusions in brain samples of GD and PD patients [1, 21]. The abnormal conformation can affect lysosomal function and cellular proteostasis systems and lead to αS accumulation, either due to increased formation of αS aggregates, or decreased clearance. Studies on fibroblasts derived from patients with GD showed that mutant GCase failed to fold correctly and was therefore retained in the endoplasmic reticulum (ER), overwhelming ubiquitin–proteasome system [23–25]. As a result, lysosomal GCase levels were significantly decreased, with subsequent accumulation of GluCer. Indeed, most mutant alleles identified in patients with GBA-associated PD (PD-GBA) were missense mutations that resulted in a misfolded protein. Moreover, mutant GCase was found in LB, suggesting its role in αS oligomerisation or impaired degradation [1].

Glucocerebrosidase, α-synuclein and lysosomes: a vicious circle

Deficient lysosomal GCase activity causes accumulation of GlcCer that in turn, accelerates formation and stabilizes soluble αS oligomers. Increased αS oligomers inhibit ER–Golgi trafficking of GCase and its translocation to the lysosome, resulting in further decrease in GCase lysosomal activity [20, 26, 27]. This amplifies GlcCer accumulation and stabilization of soluble αS oligomers, and results in a stronger inhibition of GCase ER–Golgi trafficking with each pathogenic cycle, resulting in more αS aggregates [20]. This way, bidirectional interaction between αS and GCase creates a positive feedback loop that, after a certain threshold, leads to self-propagating disease. Such reciprocal interaction between GCase and αS may underlie the aggregation of αS in the brain of PD-GBA patients. Data from autopsies of PD-GBA patients revealed elevated levels of oligomeric αS and αS-immunoreactive LB in the cortex and hippocampus [28]. However, some studies failed to demonstrate the effect of GCase pharmacological inhibition on αS accumulation/aggregation via lysosomal dysfunction [29].

In vitro experiments showed that in the lysosome-enriched fractions isolated from brain tissues and cultured neuronal cells, αS directly inhibited the lysosomal GCase activity (oligomers > monomers) [30]. It was shown that lysosomal GCase interacts with the C terminus of αS under acidic conditions, which mimicked the lysosomal lumen [31], and that membrane-bound α-helical form of αS inhibited GCase hydrolytic activity [32]. It is not known why αS oligomers inhibit GCase activity more strongly than the monomers. One possibility is that αS oligomers have a higher binding affinity with GCase and can therefore modulate its enzymatic activity to a greater extent. The colocalization of αS and GCase in LB indicates that aggregated αS can tightly bind to GCase [33]. In cultured neuronal cells, in addition to directly inhibiting GCase, αS oligomers may indirectly reduce GCase activity by blocking its transport from the ER to the lysosomes [20, 27].

GCase is located on the surface of the inner membrane of the lysosome. Under normal conditions, newly synthesized GCase is correctly folded in the ER and then translocated to the lysosomes by the trafficking receptor, lysosomal integral membrane protein-2 (LIMP-2) [34]. Cells overexpressing αS show less GCase delivered to the lysosome, as a result of less binding to LIMP-2 [35]. Since LIMP-2 is crucial for the correct trafficking of GCase and that its malfunction may lead to a reduction in GCase levels and activity, it is interesting to consider possible role of LIMP-2 in the development of PD. In LIMP-2 knock-out mice, lysosomal GCase activity was reduced, resulting in αS accumulation and disturbed lysosomal function, leading to neurotoxicity in dopaminergic neurons [36].

The main lysosomal degradation pathway for wild-type alpha αS is chaperone-mediated autophagy (CMA) [37]. Proteins destined for CMA form a complex with cytosolic chaperone, heat shock cognate protein of 70 kDa (Hsc70), which is targeted to the lysosomal membrane where it interacts with lysosomal associated membrane protein 2A (LAMP-2A), the receptor for CMA of αS, and undergoes translocation to the lysosome, followed by degradation [38]. To transport αS into the lysosome, the LAMP-2A protein must form a multimeric complex on the lysosomal membrane [39].

Namely, LAMP-2A exists as a monomer at the lysosomal membrane, in specific membrane lipid microdomains and substrate proteins only bind to monomeric LAMP-2A. To translocate substrate across the lysosomal membrane, it forms a multimeric complex with other proteins, enabling translocation of the substrates into the lysosomal lumen, which only relates to LAMP-2A molecules outside these microdomains [40]. Retention of LAMP-2A within the lipid microdomains may not allow protein-complex formation, which in turn will prevent the translocation of αS, resulting in its accumulation in the cytosol (Fig. 1).

Fig. 1
figure 1

Interaction between glucocerebrosidase, α synuclein, and lysosomes (adapted from Siebert et al. Ref. [13]). 1 After selective recognition by the lysosomal integral membrane protein-2 (LIMP-2), glucocerebrosidase (GCase) is sorted via ER and Golgi to the lysosomes. 2 Inside the lysosomes, GCase is associated with inner face of the lysosomal membrane, and hydrolyzes its substrate, glucocerebroside (GC), thus enabling the maintenance of the lysosomal membrane composition. In addition, α synuclein (αS) monomers interact with lysosomal associated membrane protein 2A (LAMP 2A), and upon formation of multimeric complex, enter the lysosome, where GCase facilitates their breakdown. 3 In case of LIMP-2 deficiency or absence, GCase is not sorted to the lysosomes, and is therefore secreted into the extracellular environment. 4 Impaired GCase activity may affect lysosomal membrane composition, leading to increased density of lipid rafts, affecting the formation LAMP 2A-protein-complexes required for αS translocation across lysosomal membrane, resulting in its accumulation in the cytosol. The increased cytosolic level of soluble monomers facilitates formation of oligomers and fibrils, affecting regular traffic of GCase from ER to Golgi. LIMP-2 lysosomal integral membrane protein-2, GCase glucocerebrosidase, ER endoplasmic reticulum, GC glucocerebroside, αS α synuclein, LAMP 2A lysosomal associated membrane protein 2A, PM plasma membrane, ECE extracellular environment

Full size image

In a cellular and mouse models of GD, GBA inhibition results in changes of lysosomal membrane composition, including increased concentration of GCase in lipid rafts [41, 42]. It was hypothesized that this process may prevent or reduce the formation of LAMP-2A protein-complexes, which will then reduce the removal of αS by CMA and increase its accumulation in cytosol [43]. In another study, LAMP-2A protein levels were found to be selectively decreased, which directly correlated with increased levels of αS and decreased levels of Hsc70 in the same PD samples, as well as with accumulation of cytosolic CMA substrate proteins, MEF2D and IkΒα [44]. It is possible that alterations in the composition of the lysosomal membrane may affect autophagy-lysosomal pathways in general, hypothetically contributing to PD development.

The role of aging

Aging, an important risk factor for PD, is associated not only with decline in efficiency of the mechanisms responsible for disposal of misfolded proteins in post-mitotic neurons, but also with decline of GCase activity [45]. Age-related increase of oligomeric αS levels in the brains of cynomolgus monkeys was accompanied by a decrease in the expression and activity of GCase [30]. Besides, levels of αS phosphorylated at serine 129, a modification that promotes its oligomerization, also increase with age in the brain and are associated with a reduction in the activity of protein phosphatase 2A (PP2A), an enzyme that facilitates αS dephosphorylation [30]. In the light of described deleterious αS-GCase feedback loop, aging-related αS accumulation may affect GCase activity even in PD-non-GBA patients. Indeed, decreased GCase protein levels and enzymatic activity were recently identified in the substantia nigra (SN) of PD-non-GBA patients [22, 35, 46], that might, unlike general lysosomal inhibition, increase the levels of intracellular αS [18, 47]. Interestingly enough, reduced GCase occured when cellular αS levels were increasing, but before αS deposition in LB, and correlated with reduced LAMP-2A and increased αS levels [22].

Glucocerebrosidase affects α-synuclein cell-to-cell transfer

Although initially thought to be a purely intracellular protein, it gradually became clear that αS could be detected in the conditioned medium of cells and in extracellular fluids, such as plasma and CSF [48, 49]. Recent evidence has pointed towards the significance of cell-to-cell transfer of αS aggregates in PD and other synucleinopathies, as a mechanism of disease propagation [50, 51]. The existence of extracellular αS in rodent and human brain interstitial fluid has been confirmed by microdialysis [52]. Importantly, secreted αS can impact neuronal homeostasis and lead to neuronal death, even at concentrations close to those identified in body fluids [51, 53]. It is conceivable that GCase depletion may increase αS release, while the endolysosomal dysfunction increases its uptake [54, 55]. A recent report described increased neuronal cell-to-cell transmission of endogenous αS in grafted cells lacking GCase [47]. Furthermore, the ectopic expression of wild-type GCase (but not of an activity-deficient GBA mutant) reversed the effects of GBA deletion on the propagation of αS aggregates, indicating that the enzyme hydrolytic activity had a role in cell-to-cell αS transfer [47].

How frequent are GBA mutations among PD patients?

More than 300 different mutations of the GBA have been reported, but the N370S (the most frequent among European populations and Ashkenazi Jews) and L444P (dominated in Asian populations, where N370S was rare) account for the majority of those found in both GD and PD. Both GD patients and asymptomatic heterozygous gene carriers were recognized to be at risk of PD [15]. The proportion of PD patients that carry GBA mutations is estimated to be approximately between 5 and 10 % [7, 56–59] (range 3.2–21 %) [5, 6, 60]. The highest frequency of GBA mutations in PD patients (31.1 %) was reported in Ashkenazi Jews [5], while studies in Norwegian [61] and North African Arab-Berber [62] populations failed to demonstrate different frequencies among PD patients and controls (Table 1).

GBA mutations are important risk factor for developing PD, but most people with mutations will never develop PD [63]. Patients with GD have an almost 20-fold increased life-time risk of developing PD [64], irrespective of either GD severity or enzyme replacement therapy [65]. Carriers of severe (for example L444P) when compared to those with mild GBA mutations (for example N370S) have three to four times higher risk to develop PD, characterized by an earlier age at onset (on average 5 years) and more common cognitive impairment [66, 67].

Risks for PD at the age of 60 and 80 years was 4.7 and 9.1 % for GD patients and 1.5 and 7.7 % for GBA mutations carriers, respectively [68]. Cumulative risk for PD for GBA mutations carriers was 5 % at the age of 60 rising to 15 % at the age of 80 years [58], while Rana et al. [69] reported risks of 2.2 and 10.9 % at the age of 65 and 85 years, respectively. Due to the high cumulative age-specific risk (i.e. penetrance) found in the study by Anhelm et al. [70] (29.7 % at the age of 80 years), it has been suggested that GBA might be considered as an autosomal-dominant PD causing gene with reduced penetrance, although, these results should be interpreted with caution.

It is not surprising that the first-degree relatives of GD patients more frequently have PD, since they are obligate heterozygous GBA carriers [71–73]. PD patients with affected relatives are more frequently carriers of GBA mutations than those with negative family history for PD [57].

Clinical characteristics of Parkinsonʼs disease associated with GBA mutations

Pathological examinations of PD-GBA patients have revealed morphological abnormalities within the spectrum of classical (idiopathic) PD; therefore, this condition is not considered to represent an atypical form of the disease [33, 59]. Autopsy study also found that cortical LB are more frequently seen in PD-GBA patients than in non-carriers [59]. Clinically, GBA mutation carriers have an earlier age at PD onset (between 1.7 and 6.0 years earlier than PD-non-GBA [1, 5, 74–77]; observation not confirmed in several studies) [56, 78, 79], more likely have a positive family history for PD, but also more prevalent non-motor symptoms (NMS) when compared to PD-non-GBA [80].

Both GD patients and asymptomatic GBA mutation carriers may exhibit mild parkinsonian motor signs insufficient to reach the diagnosis of PD that may remain stable after two years of follow-up [81, 82]. In addition, hyposmia, cognitive decline, REM-sleep behavior disorder and depressive symptoms are more frequent among them in comparison to subjects who are not carriers of GBA mutations [81, 83]. We identified a family where both parents of a GD patient (obligate GBA carriers) have stable mild parkinsonian signs during the follow-up of 3 years (Fig. 2).

Fig. 2
figure 2

DaT-SPECT findings in GBA mutation carriers in premotor phase of PD. a Normal DaT-SPECT in a person without parkinsonism and without GBA mutations; b and c show DaT-SPECT findings for parents of GD patient. GD patient doesn’t have any signs of parkinsonism, hence, both parents (obligate carriers of GBA mutations) are in premotor phase of PD. b father of our GD patient (55 years) had no criteria for PD, but expressed hyposmia, REM-sleep behavior disorder, hypomimia, and mild bradykinesia of the left leg, that were stable for a 3 year follow-up and bilateral, but predominantly right-side decrease in DaT-SPECT binding; c mother of the same GD patient (51 years) had only bradykinesia of the right extremities that was stable for 3 year follow-up and reduced striatal DaT binding bilaterally, predominantly on the left side

Full size image

PD-GBA has typical clinical manifestations: asymmetric parkinsonism with a good response to levodopa [84]. PD phenotype in GD patients and heterozygous GBA mutations carriers is similar [1]. Rarely, PD may precede clinical manifestations of GD for years [56, 85]. Initial symptoms of PD seem to be similar between GBA mutations carriers and non-carriers [5, 86–88]. However, although not confirmed in all studies, there are reports of more common bradykinesia, tremor, weakness and shoulder pain, as well as a lower frequency of rigidity as initial symptoms in PD-GBA [76, 86, 89].

Motor phenotype

When diagnosed, PD-GBA is clinically indistinguishable from PD-non-GBA [57, 76, 77, 79, 88, 90, 91]. However, some studies suggested less prevalent asymmetric onset, more postural instability [7, 56, 87], and more frequent levodopa-induced motor fluctuations and dyskinesia (LID), in PD-GBA patients [57, 92] (conversely, other studies observed that LID occurred with a lower frequency or with a similar prevalence when compared to PD-non-GBA patients) [93–96]. In conclusion, individual PD-GBA patients cannot be discriminated from PD-non-GBA patients on clinical or pathological grounds.

GBA mutations and polymorphisms influence the course of PD [79]. Progression of motor deterioration is more rapid in PD-GBA than in PD-non-GBA patients: GBA mutations carriers have fourfold increased risk of progression to Hoehn and Yahr stage 3 (HY3) [79, 90, 97–99]. Winder-Rhodes et al. [79] estimated that the time of progression to HY3 was 23.5 months for PD-GBA patients, 32 months for PD patients with GBA polymorphisms, and 49 months for non-carriers.

Non-motor features

Prominent NMS appeared to be the key clinical aspect of PD-GBA patients: they suffered more commonly and severely from a variety of NMS than those without GBA mutations [58, 80, 87]. PD-GBA patients show more rapid progression of cognitive decline and have >5fold increased risk of progression to dementia when compared PD-non-GBA patients [78, 79, 96, 98]. Winder-Rhodes et al. calculated projected median time to dementia to be 46 months for PD-GBA patients, 96 months for PD patients with GBA polymorphisms, while <50 % of those without GBA mutations developed dementia over the median follow-up period of 82 months [79].

Visual hallucinations, delusions and psychosis occured in PD-GBA, probably due to an extension of the LB pathology to the temporal lobe [55, 94, 95]. Depression, apathy, indifference, and anxiety disorder have been, although not uniformly, [89, 98, 101–103] reported to have higher prevalence in PD-GBA [58, 80, 87].

Autonomic dysfunction (e.g. orthostatic hypotension, constipation, sweating, urinary, bowel and sexual dysfunction), as well as sleep disturbances, fatigue, and unexplained pain, were also found to be more frequent in PD patients with than those without GBA mutations [58, 80, 87, 104, 105].

Therapeutic considerations

While several reports suggested poor response to levodopa, majority of the studies found an excellent therapeutic effect of levodopa and dopamine agonists in PD-GBA [5, 56, 78, 84, 90]. DBS also had favorable outcome in PD-GBA patients, although these patients might require application of the procedure earlier in the course of the disease and might have faster cognitive decline and axial impairment after DBS [65, 99, 106, 107]. The enzyme replacement therapy for GD neither prevented development of PD, nor modified severity of symptoms and PD progression, probably due to a lack of passage of the enzyme across the blood–brain barrier [108–110]. Alternative therapy for GD is with miglustat, an iminosugar which inhibits the biosynthesis of macromolecular substrates that accumulate pathologically in glycosphingolipidoses [111]. Miglustat is able to cross blood–brain barrier but no clear data are available on the impact of miglustat on parkinsonism in GD patients [110, 112, 113].

Table 1 Frequency of GBA mutations in patients with Parkinson’s disease (updated table from Ref. [1]; only studies with more than 100 patients with Parkinson’s disease and included controls were analyzed)
Full size table

Establishing the concept of GBA-related PD promoted a search for the pathogenic mechanisms through which GCase deficiency may influence pathogenesis of PD, suggesting that targeting the GCase–lysosomal pathway might be a rational approach for the “development of neuroprotective drugs in PD” [15].