The GBA p.G85E mutation in Korean patients with non-neuronopathic Gaucher disease: founder and neuroprotective effects

Background Gaucher disease (GD) is caused by a deficiency of β-glucocerebrosidase, encoded by GBA. Haplotype analyses previously demonstrated founder effects for particular GBA mutations in Ashkenazi Jewish and French-Canadian populations. This study aimed to investigate the clinical characteristics and mutation spectrum of GBA in Korean GD patients and to identify founder effect of GBA p.G85E in non-neuronopathic GD patients. Results The study cohort included 62 GD patients from 58 unrelated families. Among them, 18 patients from 17 families harbored the p.G85E mutation. Haplotype analysis was performed for 9 probands and their parents for whom DNA samples were available. In 58 unrelated probands, the GBA mutation p.L483P was the most common (30/116 alleles, 26%), followed by p.G85E (16%), p.F252I (13%), and p.R296Q (9%). The median age at diagnosis of the 18 patients harboring the p.G85E mutation was 3.8 (range 1.2–57) years. No patients developed neurological symptoms during follow-up periods of 2.2–20.3 (median 13.9) years. The size of the shared haplotype containing GBA p.G85E was 732 kbp, leading to an estimated age of 3075 years. Conclusion The GBA p.G85E mutation, which appears to be neuroprotective despite producing distinctive visceromegaly and skeletal symptoms, exhibited a potential founder effect in Korean GD patients.

The GBA p.G85E mutation has been identified in Asian populations, exclusively in Korea, China, and India [4,[9][10][11]16]. Furthermore, the p.G85E mutation was found exclusively in non-neuronopathic GD patients [4,9], suggesting a potential neuroprotective allele. This study aimed to investigate the clinical characteristics of GD patients carrying p.G85E mutation and mutation spectrum of GBA among Korean GD patients and to test for a founder effect for the GBA p.G85E mutation via haplotype analysis.

Three-dimensional (3D) structure of GBA and location of mutations
We determined the location of each mutation in the 3D structure of GBA (Additional file 1: Figure S1). All mutations were widely located in domains I, II, and III. The p.G85E, p.R87W, p.L483P, and R535H mutations were located in the immunoglobulin (Ig)-like domain (domain II), but they were distant from the active site, except for p.L483P, which was located in the hydrophobic core of the Ig-like domain, which may lead to disruption of the hydrophobic core and alteration of the domain folding. p.G85E, p.R87W, and R535H were identified only in type 1 GD patients. p.V54L, p.N425K, p.D438N, p.I442T, and p.D448H were located in domain I containing two disulphide bridges. p.V54, D438N, and p.I442T were identified in type 1 GD. The other 12 missense mutations were located in the catalytic domain, which is the triosephosphate isomerase (TIM)-barrel (domain III). Among mutations located in domain III, p.V211fs and p.W420* were identified in patients with the neuronopathic type. The splicing mutation (IVS2 + 1 G > A), which leads to disruption of domains I and II, was also found in type 2 GD patients.
Four patients (23%) underwent splenectomy during childhood, prior to the introduction of enzyme replacement therapy (ERT). Of the 4 splenectomized patients (Subjects 7, 12, 15, and 16), two (Subjects 7 and 16) manifested avascular necrosis (AVN) of the femur neck, while Subject 12 had osteoporosis, with a lumbar spine Z-score of − 5.4. Subject 15 had compression fractures in the thoracic spine at age 13 years. Four patients (Subjects 7, 9, 14, and 16) manifested AVN of the femur neck at ages 11, 12, 12, and 57 years, respectively. A 2-year-old female patient (Subject 11) suffered from severe bone pain and massive splenomegaly, with abnormally low T1 signal intensity in the long bone and femur neck on magnetic resonance imaging.
Following the introduction of ERT in Korea in 1994, all patients have been treated with it. Patient age at the last evaluation ranged from 3.7 to 67 years (mean 25.7 ± 17.0, median 19.0). No patients developed neurological symptoms, including abnormal eye movements, ataxia, seizures, and cognitive impairment during the follow-up period, which ranged from 2.3 to 21 years. After ERT treatment for 2.2-20.3 years (median 13.9), hematological findings improved among all patients, with mean hemoglobin and platelet levels of 13.6 ± 1.57 g/dL and 240 ± 81.2 × 10 3 /µL,    (Table 2).

Founder effect of the p.G85E mutation among Korean patients
On haplotype analysis, 9 patients were compound heterozygotes for the p.G85E allele with other mutant GBA alleles. The size of the shared haplotype was approximately 732 kbp (chromosome 1: 154,808,287-155,540,660) (Fig. 3) (Additional file 2: Table S1). All 9 alleles containing the p.G85E mutation shared a common haplotype surrounding it, suggesting a founder effect. The estimated age of the mutation was 123 generations (95% credible set; 69-248 generations); i.e., 3075 years (95% credible set: 1725-6200 years), based on an intergeneration time of 25 years [17].

Discussion
This study assessed the mutational spectrum and phenotypic characteristics of Korean GD patients, many of whom share a unique founder mutation p.G85E. In 1996, this mutation was first reported in two Korean patients: 2-and 18-year-old boys, who presented with massive hepatosplenomegaly and isolated splenomegaly, respectively [11]. Both were non-neuronopathic and their genotypes were compound heterozygous for p.L483P and homozygous for p.G85E, respectively [11]. GBA expression of the p.G85E mutant was as low as the activity of the p.F252I mutant in neuronopathic forms [4]. The 3D structure showed that p.G85E was located in the Ig-like domain II (Additional file 1: Figure S1) and distant from the active site residues E274-E379 [18]. In our study, patients with p.G85E presented with massive visceromegaly and bone involvement during early childhood. However, none had shown any neurologic symptoms during a relatively long follow-up period. Genotype-phenotype correlations have been previously reported, although they have not always been consistent. Patients with at least one p.N409S allele are non-neuronopathic in Caucasians [19][20][21][22][23]. Furthermore, homozygous p.K118N, p.N227S, and p.G416S mutations are also non-neuronopathic. However, compound heterozygosity with null mutant alleles results in the chronic neuronopathic form (type 3) [11,[24][25][26][27].
Half of the chronic non-neuronopathic type GD (type 3) patients displayed symptoms before 2 years of age, while others exhibited initial symptoms during late adolescence [22,[28][29][30]. In the present study, GD patients carrying the p.G85E mutation have yet to present with any neurologic symptoms at the latest evaluation. More than half of the patients (11/18, 61%) were adults. Furthermore, Subject 4, displaying compound heterozygosity for p.G85E and Rec8a, displayed no neurological symptoms until the latest follow-up at 15 years of age. However, many GD patients with the p.G85E mutation presented with more severe hematological, visceral, and skeletal manifestations, than did typical GD patients. These findings suggest that the p.G85E mutation may exert a neuroprotective effect, as does the p.N409S mutation in Caucasian GD patients. The p.G85E mutation was previously identified in two Chinese patients as compound heterozygotes, with one having p.N227S and the other having p.L483P, respectively [10]. The patients harboring the p.G85E and p.L483P mutations were diagnosed with type 1, whereas the other patient with p.G85E and p.N227S was diagnosed with type 3 GD; this latter patient presented with seizures and abnormal electroencephalography findings, but not oculomotor abnormalities. Recently, two Indian GD patients with p.G85E mutation were reported, but without phenotypic information [16].
In addition to having unique clinical characteristics, the p.G85E mutation is frequently observed among Korean GD patients, accounting for 16% of mutant alleles, second only to the p.L483P mutation (26%). We assessed the haplotypes of alleles with p.G85E and identified a haplotype 732 kbp in length, suggesting that a founder mutation occurred 3,075 years ago. To date, we have not encountered any case of Parkinson disease or Alzheimer disease in a GBA p.G85E carrier, and there have been no reports about the association between GBA p.G85E carriers and Parkinson disease [31]. However, we carefully counseled the patients, together with their parents, considering the late onset of Parkinson disease or Alzheimer disease [32]. Thus, we hypothesize that the p.G85E mutation may have passed through many generations within the Korean population, with positive selection for the neuroprotective effects conferred to carriers.

Conclusion
In conclusion, this study shows the mutation spectrum of GBA, as well as clinical features and a founder effect of GBA p.G85E in the Korean population. Our data and those from a previous study [4] strongly suggest that the p.G85E mutation may behave as a neuroprotective allele in GD patients, even though their visceral and skeletal phenotypes might be severe. Further long-term, meticulous clinical follow-up may be needed to understand the neuroprotective effect of p.G85E allele and to elucidate the mechanisms underlying this effect.

Patients
The study cohort included 62 patients with GD from 58 unrelated Korean families diagnosed through molecular analysis of GBA in Korea from 2000 to 2019. Among them, 18 patients from 17 families harbored the p.G85E mutation ( Table 2). Their clinical characteristics, including height, weight, visceromegaly, skeletal findings, hematological parameters, bone mineral density assessed through dual-energy X-ray absorptiometry, and neurologic symptoms, were retrospectively reviewed. This study was approved by the Institutional Research Board at each institute. Informed consent was obtained from all patients and their parents.

Molecular analysis of the GBA gene
Genomic DNA was extracted from peripheral blood leukocytes using the Puregene DNA isolation kit (Qiagen, Hilden, Germany). Polymerase chain reaction (PCR) amplification of all coding exons and exon-intron boundaries was performed using allele-specific primers. Direct bidirectional sequencing of amplification products was performed using an ABI3130xl Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). The longrange PCR approach was used for identification of GBA recombinant mutations between GBA and the non-functional GBAP pseudogene located 16 kb downstream of the functional GBA according to a previous report [38]. This report uses GBA nomenclature based on the reference sequence NP_000148.2. Traditional GBA nomenclature, which omits the first 39 amino-acid residues of the leader sequence of NP_000148.2, is also provided within parentheses without the p. prefix in Fig. 1.

In vitro expression of GBA mutant proteins
The β-glucocerebrosidase activity of the p.G85E, p.N227S, p.F252I, p.R296Q, and p.L483P mutant proteins was measured in COS-7 cells transfected with the pCMV6 vector. For site-directed mutagenesis, human GBA (NM_000157.4) cDNA (SC120080 OriGene, Rockville, MD, USA) was used as a template for the generation of mutants. GBA mutants, with a wild-type control, were transiently transfected into COS-7 cells using Effectene transfection reagent #301427 (Qiagen, Hilden, Germany). After transient transfection, all cells were further incubated at 37 °C in 5% CO 2 . Assays for β-glucocerebrosidase activity were performed using a standard fluorometric method. Fluorescence was detected using a fluorescence spectrophotometer (Molecular Devices, San Jose, CA, USA). Every experiment was performed in triplicate for each mutant.

Haplotype analysis
Analysis was performed for 9 unrelated probands and their 16 family members from whom DNA samples were available and who consented to participate in the study. Genotyping of short tandem repeat (STR) and singlenucleotide polymorphism (SNP) markers was performed within the 3.8-Mb genomic region encompassing at least three recombination hotspots on both sides of the muta-