GAA compound heterozygous mutations associated with autophagic impairment cause cerebral infarction in Pompe disease

Clinical characterization

The proband was hospitalized for persistent dizziness and unsteady gait. At the age of 14, the patient started to develop weakness in both legs and complained constant fatigue. Examination revealed that the patient has normal intelligence, no facial palsy, scoliosis, instability in the finger-nose test, and muscle atrophy throughout the body. Craniocerebral CT showed left cerebellar cerebral infarction (Figure 2A); cerebral MRI SWI shows multiple microhemorrhage loci in both hemispheres (Figure 2C); MRI T₂FLAIR showed multiple ischemic lesions in both lateral ventricles and deep white matter (Figure 2D). Computed tomography angiography (CTA) showed bilateral acute curvature of the internal carotid artery (C2 segment); the left side was significant. Calcified plaques were formed in bilateral vertebral artery and the lumen exhibited segmental stenosis. Multiple bilateral local stenoses occurred in the P1-P2 segments of the posterior cerebral arteries. Basilar artery aneurysm was observed (Figure 2B). Electromyography revealed myogenic damage. The patient’s hematological examination readings were: alanine aminotransferase (ALT) 39 IU/L, aspartate aminotransferase (AST) 132 IU/L, creatine kinase (CK) 325 IU/L, and lactate dehydrogenase (LDH) 180 IU/L.

H&E staining showed that the patient’s muscle fibers were slightly different in size, polygonal in shape, and slightly increased in kernel fibers. A large number of muscle fibers were seen in the submucosal and intramuscular area with irregular vacuoles, and some muscle fibers were almost completely vacuolated. Occasionally, basophilic granules deposition were seen between the muscle fibers. No necrosis accompanied by phagocytosis and regenerated fibers were observed. No focal infiltration of inflammatory cells in the muscle interstitium, and there was a slight hyperplasia of the muscle inner membrane (Figure 2I, 2J). Lysosomal Associated Membrane Protein 2 (LAMP2) immunohistochemical staining was enhanced in the vacuolar muscle fiber, and the distribution was enriched at the margin of vacuolar muscle fiber (Figure 2K). Coenzyme I (NADH) staining showed the interphase distribution of two types of fibers, the mesh-like structure in vacuolar fibers was disordered, and the activity of NADH in vacuolar region was absent (Figure 2L). No increase of lipid droplets was observed by OR0 staining. Electron microscopy was used to examine muscle fibril structure; myofibril and muscle glycogen particles increased significantly under the plasma membrane. We also observed focal myofibril destruction, glycogen granules, cystic structure, and lysosomes in addition to the phenomenon of many membranous vacuoles full of glycogen particles. A large number of mitochondria accumulated near the nuclear membrane but we did not see lattice-form inclusion bodies in the mitochondria. Another phenomenon was that mesenchymal cells were visible within the glycogen accumulation.

The female patient, the younger sister of the proband, died in December 2017 from respiratory failure caused by a lung infection. Her intelligence was normal, and her left nasolabial groove was shallow; she exhibited whole-body muscle atrophy and left finger-nose test instability. Electrocardiogram showed a decreased ST segment. She had ruptured mitral tendinous cord and valve prolapse. Her mitral valve was thickened and with neoplasm. Her liver and spleen were large. MRI susceptibility-weighted imaging (SWI) showed recurrent cerebellopontine old hemorrhagic focus and multiple micro-hemorrhagic foci in cerebellum and brainstem (Figure 2F, 2G). Multiple patchy abnormal signals were detected in the brain stem, bilateral brain semi-oval center, radiation crown area, and left sub frontal cortex. Craniocerebral MRI T₂/FLAIR images shows multiple ischemic lesions in para lateral ventricles and white matter (Figure 2H). CTA of the brain showed stenosis at the left anterior cerebral artery and the right posterior cerebral artery (Figure 2E). Electromyography revealed myogenic damage. The patient’s hematological examination readings were: ALT 71 IU/L, AST 280 IU/L, CK 484 IU/L, and LDH 375 IU/L.

The child of the female patient appeared normal. However, some abnormal blood biochemical indexes were observed during the physical examination, including: ALT 14 IU/L, AST 48 IU/L, CK 556 IU/L, LDH 246 IU/L, which led to the suspicion that the child may be a Pompe disease patient. A year after the mother's death, the child's biochemical index became normal, the hematological examination readings were: ALT 14 IU/L, AST 31 IU/L, CK 97 IU/L, and LDH 225 IU/L.

Compound heterozygous mutations of GAA gene

WGS data from all subjects were thoroughly analyzed (Figure 1B). In comparing with the normal family members, both patients beard two GAA compound heterozygous mutations, one was a missense mutation c.2238G>C (p.Trp746Cys) from the father and the other was a frameshift mutation c.1388_1406del19 (p. Arg463fs) from the mother. The first mutation c.2238G>C (p.Trp746Cys) in exon 16 causes a change from nonpolar aromatic tryptophan to polar aliphatic cysteine at codon 746 which is in a highly conserved region (Figure 3C) and has been reported to affect the enzymatic function of acid α-glucosidase [5, 17]. The second mutation c.1388_1406del19 (p. Arg463fs) in exon 9 was a frameshift mutation caused by a deletion of 19 bases. This mutation leads to premature termination of protein translation; a frameshift mutation occurs once the deletion takes place, which results in an unexpected stop to translation at amino acid 462 (Figure 3A, 3B). The truncated polypeptide chain cannot maintain its normal conformation, so the protein is unable to function normally. In addition to the above single nucleotide polymorphisms (SNP) and indel, c.1726G>A and c.2065G>A mutation sites from the father GAA genome and c.2446G>A from the mother GAA genome were also found. These three mutation sites have been reported and have racial heterogeneity, c.1726G>A and c.2065G>A are mainly distributed in Asian populations, with higher frequency in China, Taiwan, and Japan, c.1726G>A mutation can lead to a decreased enzyme activity, but this mutation does not cause disease; mutation of c.2446G>A also has a higher allele mutation frequency in the population [4, 18, 19]. So, in this study, the compound heterozygous mutations c.2238G>C and c.1388_1406del19 of GAA gene lead to familial glycogen storage disease type II. For the child of the female patient, we only found the c.2238G>C mutation. The other chromosome was normal; therefore, we can rule out the possibility of the child to develop the disease. We performed Sanger sequencing on these two pathogenic sites in all the members, and the results of Sanger validation were consistent with the whole-genome analysis (Figure 4).

Analysis of genetic risk locus and autophagy-related gene SNPs associated with ischemia/hemorrhagic stroke and cardiovascular diseases

In addition to GAA gene, we examined variations of two panels of genes to explore the mechanism behind the cerebral infarction of the patients. One is 15 genome-wide risk loci associated with ischemia and hemorrhagic stroke [20]; the other is 7 autophagy genes with the most significant association with cardiovascular disease [21]. The risk allele frequencies for these loci are highly identified in big cohort studies. In our studies, 7 out of 15 risk alleles of hemorrhagic stroke-associated genes were examined in all family members including the 2 Pompe patients and 6 normal members (Supplementary Table 1). Considering each individual risk allele has 2-66% frequency in the general population [20], all 8 family members showed more than 50% of the risk alleles may indicate high genetic risk with the disease. For autophagy genes associated with cardiovascular disease, 2 out of 7 risk alleles were examined in the family, both appear in the 2 Pompe patients, and 6/6 or 3/6 in the normal members (Supplementary Table 2).

GAA activity in patient samples

GAA activity of the proband and the child measured by peripheral blood lymphocytes enzyme activity method were 3.77 nmol/h/mg and 9.55 nmol/h/mg (reference value 14 nmol/h/mg), respectively. In general, the residual GAA activity in patients was 1%~30%. Although the child's enzyme activity was slightly below normal, it was still in the normal range. There was no GAA activity result for the female patient because she passed away.

Functional characterization of GAA mutations

HEK-293 cells constitutively express human GAA which is considered as a background signal (Figure 5A). Cells transfected with the GAA wild type construct (GAA-WT) contained 3 forms of GAA molecular species, i.e., the 110 kD precursor, the 95 kD partially processed intermediate and the 76 kD mature GAA. The medium contained only the 110 kD GAA precursor (Figure 5A). The quality and quantity of different molecular species of GAA protein were evaluated according to the previous reported criteria [22]. Although all 3 forms of GAA were detected in either of the GAA-Trp746Cys or GAA-Arg463fs transfected cells, the quantity of each decreased moderately with GAA-Trp746Cys mutation and severely with GAA-Arg463fs mutation (Table 1). In the cells transfected with both constructs, all 3 forms of GAA were barely detected (Table 1). Mutation severity was evaluated using the scoring system based on the GAA activity in the medium and in the cells [22]. Results in Table 1 indicated that the combination of GAA-Trp746Cys and GAA-Arg463fs mutations led to a “class B—potentially severe” deficiency of GAA activity.

Autophagy impairment was examined in the GAA constructs-transfected HEK293 cells after culturing with glucose-free medium for 24-hour. Morphological analyses showed the presence of ≥5 LC3 puncta in the vast majority of cells with the double GAA-Trp746Cys and GAA-Arg463fs transfection (Figure 5B, 5C). Protein expressions of LC3 and p62, two markers of autophagosome formation and clearance, respectively, were quantified by western blot analysis (Figure 5D). GAA-Trp746Cys and Arg463fs transfected cells had a massive increase in lipidated LC3 and p62 protein, indicating a autophagy failure.

Analysis of gut microbiome metagenomics

We tried to explore why the child had abnormal blood biochemical index when he lived with his Pompe mom. Because intestinal flora has been shown to be associated with a variety of chronic diseases [23–25] and cohabiting family members normally share microbiota with one another [26, 27], we extracted DNAs from the fecal samples of the family members and performed microbiome metagenomics analysis. We generated 111.15Gb raw intestinal metagenomic data and 102.33 Gb high quality data after removing adaptor sequences, low quality data, and host data (Table 2). An average of 12.79Gb high quality data was obtained for each sample. Subsequent analysis found that II-3 (the female patient) has the lowest alpha diversity among the three diversity indices, compared to other individuals of the family (Table 3), indicating her intestinal flora was disordered. Her child’s (III-2) gut flora composition is similar to hers (Table 3), with the second lowest alpha diversity. Specifically, the correlation analysis shows that II-3 and III-2 has the highest correlation coefficient (Figure 6) according to their respective gene abundance, suggesting that the mother may have a certain influence on the child's intestinal flora [28]. Furthermore, we identified several differential species such as Actinobacillus, Histophilus, Mobiluncus, and Thermoanaerobacterium between patients and normal family members, however, due to the limited number of samples, we could not conclude any practical significance.

Patients and controls

Two sibling patients, clinically diagnosed with Pompe disease, were recruited from Liaocheng People's Hospital Department of Neurology. Additionally, a 5-year-old boy, the female patient’s child, had suspected symptoms and had been mis-diagnosed with Pompe disease. Six healthy subjects from this family were selected as controls, including the parents of the patients, children of the patients, sister of the patients, and the spouse of the female patient (Figure 1).

The study conformed to the tenets of the Declaration of Helsinki and was approved by the ethics committee of Liaocheng People's Hospital, Shandong province. Informed consent was obtained from all patients and their families.

Whole-genome sequencing and data analysis

Peripheral blood samples were obtained from the sibling patients and their family members. Genomic DNA was extracted from peripheral blood using standard protocols. The DNA samples were sequenced by WGS on a BGISEQ-500 platform for paired-end 100 bp reads. The average sequencing depth ranged from 40.54 to 44.12X.

Low quality reads were filtered by SOAPnuke, then mapped against UCSC hg19 (http://genome.ucsc.edu/) by BWA (http://bio-bwa.sourceforge.net/). The SNPs and indels were detected by GATK. Single nucleotide variants and indels were filtered using the following criteria: (i) homozygous or compound heterozygous; (ii) absent or with a minor allele frequency value <0.01 in public databases [including dbSNP, 1000 Genomes Project, the NHLBI Exome Sequencing Project (ESP), the Exome Aggregation Consortium (ExAC) and gnomAD]; (iii) exclude non-coding regions variant, intergenic region variant, downstream variant, upstream variant, intron variant, synonymous variant, 3-prime UTR variant, 5-prime UTR variant; and (iv) possible pathogenic effects of the identified variant were predicted by SIFT (https://sift.bii.a-star.edu.sg/) and PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/). Sanger sequencing was used to validate the novel compound heterozygous mutations in GAA gene identified by the WGS.

GAA activity assays

GAA activity in peripheral white blood cells was measured using a standard fluorescence assay. A synthetic substrate 4-mug was used, which when hydrolyzed by GAA releases a fluorophore. Acarbose was used to inhibit its isoenzyme, and the peripheral white blood cells were detected by the Molecular Devices SpectraMax Gemini XPS. White blood cell homogenate was prepared by ultrasound and protein concentration was determined by the BCA method. Enzyme activity measurement mixture included: 50mL cell homogenate, 10μL 100μM acarbose, and 50μL 6mM 4-methyl ketone of umbrella-α-D-pyran glycosidase fluid. After a 1 hr incubation in a 37°C water bath, 1.25 mL glycine carbonate buffer (pH=10.3, 0.17 M) was added to terminate the reaction. Gemini XPS was used to measure the fluorescence intensity; each group required a blank control.

Functional analysis of the GAA mutations

A pSHAG2 expression vector containing the wild-type GAA open reading frame (GAA-WT) was used to generate site-directed mutagenesis. The constructs of missense mutation c.2238G>C (GAA-Trp746Cys) and frameshift mutation c.1388_1406del19 (GAA-Arg463fs) were introduced in the plasmid individually or simultaneously by using the QuickChange® Site-Directed Mutagenesis Kit (Agilent Technologies Inc.). The integrities of the mutant constructs were confirmed by direct sequencing. The constructs were transduced into HEK293 cells when 80–90% of confluence at 1.5 mg of GAA-WT or mutant constructs using transient transfection reagent Effectene (Qiagen). Mock transfected cells served as negative controls. Cells were harvested at 48-hour with lysis buffer (50 mM Tris–HCl pH 7.0, 150 mM NaCl, 50 mM NaF, and 1% TritonX-100), centrifuged at 10,000 g for 10 min, then the supernatant fraction was recovered. GAA activity was measured in both medium and cell homogenates. Mutation severity was evaluated based on the GAA activity in the medium and in the cells, and on the quality and quantity of the different molecular species of GAA proteins [22]. Cell lysate and immunoprecipitated GAA from the medium were subjected to SDS-PAGE and Western-blotting analysis using GAA specific polyclonal rabbit antibody (SAB2100872, Sigma Aldrich) and goat anti-rabbit secondary antibody.

Examination of autophagy deficiency in vitro

HEK293 cells with different GAA constructs were cultured in glucose-free medium for 24-hour before analyzing autophagy. LC3 Antibody Kit for Autophagy (L10382, Thermo Fisher Scientific Inc) was used to stain HEK293 cells and fluorescent images were taken by Olympus FV1000 confocal microscope. Cells with ≥5 LC3 puncta were quantified manually by two independent research staff in a single-blind way. Protein expressions of LC3 (LC3 I and II) and p62 were quantified by western blot analysis using rabbit anti-LC3 and anti-p62 primary antibodies (#2775 and #5114, Cell Signaling Technology Inc).

Gut microbiome metagenomic sequencing and data processing

Fecal samples were collected from all 8 family individuals and then frozen immediately at -80°C before transportation to the laboratory for DNA extraction. Subjects did not receive any antibiotic treatment and did not take any probiotics for at least one month before sample collection. DNA was extracted using the standard protocol. All samples were sequenced on a BGISEQ-500 platform with paired-end 100 bp reads, and filtered to remove adaptor contamination, low quality reads, and host data (hg19) by SOAPnuke. The remaining high quality data was aligned to IGC to profile gene abundance by bowtie2, and the species abundance and the functional abundance were summarized from their respective genes [41]. Differential genera were determined using the function of t-test in R. The alpha diversity was analyzed with the QIIME program. The Pearson coefficient was used to calculate the relationship between family individuals according to their gene abundance.

Accession numbers

The sequence data has been deposited in the CNSA databases under the following accession numbers: CNP0000237.