Deep Intronic GBE1 Mutation in Manifesting Heterozygous Patients With Adult Polyglucosan Body Disease

Importance  We describe a deep intronic mutation in adult polyglucosan body disease. Similar mechanisms can also explain manifesting heterozygous cases in other inborn metabolic diseases.

Objective  To explain the genetic change consistently associated with manifesting heterozygous patients with adult polyglucosan body disease.

Design, Setting, and Participants  This retrospective study took place from November 8, 2012, to November 7, 2014. We studied 35 typical patients with adult polyglucosan body disease, of whom 16 were heterozygous for the well-known c.986A>C mutation in the glycogen branching enzyme gene (GBE1) but harbored no other known mutation in 16 exons.

Main Outcomes and Measures  All 16 manifesting heterozygous patients had lower glycogen branching activity compared with homozygous patients, which showed inactivation of the apparently normal allele. We studied the messenger ribonucleic acid (mRNA) structure and the genetic change due to the elusive second mutation.

Results  When we reverse transcribed and sequenced the mRNA of GBE1, we found that all manifesting heterozygous patients had the c.986A>C mutant mRNA and complete lack of mRNA encoded by the second allele. We identified a deep intronic mutation in this allele, GBE1-IVS15 + 5289_5297delGTGTGGTGGinsTGTTTTTTACATGACAGGT, which acts as a gene trap, creating an ectopic last exon. The mRNA transcript from this allele missed the exon 16 and 3′UTR and encoded abnormal GBE causing further decrease of enzyme activity from 18% to 8%.

Conclusions and Relevance  We identified the deep intronic mutation, which acts as a gene trap. This second-most common adult polyglucosan body disease mutation explains another founder effect in all Ashkenazi-Jewish cases.

Adult polyglucosan body disease (APBD) is characterized after 50 years of age by the onset of progressive pyramidal paraparesis, distal sensory deficits, neurogenic bladder, ambulation loss, and premature death owing to complications of myelopathy and peripheral neuropathy.^1,2 The disease, which is often included in the differential diagnoses of multiple sclerosis and amyotrophic lateral sclerosis, is distinct from multiple sclerosis by late-onset progressive symmetric course and peripheral neuropathy; from amyotrophic lateral sclerosis by sensory deficits, incontinence, and florid subcortical and spinal cord changes on magnetic resonance imaging; and from both by autosomal recessive inheritance.^2-6 The neuropathological hallmarks of APBD are polyglucosan bodies (PBs), which are accumulations of aggregated, poorly branched, and insoluble glycogen both in the central nervous system and in the peripheral nervous system. In neurons, PBs are principally in axons, often appearing to clog the axonal flow. Other features include central nervous system demyelination and gliosis and loss of peripheral nervous system myelinated fibers.^6-12

Adult polyglucosan body disease is allelic to glycogenosis IV (glycogen storage disease IV [GSD-IV]; OMIM 232500). Patients with classic GSD-IV have profound glycogen branching enzyme (GBE) deficiency and die in childhood of liver failure with massive hepatic and extrahepatic polyglucosan accumulations. Cases with complete deficiency usually die prenatally. Patients with APBD have approximately 18% GBE activity and smaller PBs. Their livers and other nonnervous tissues, although containing PBs, remain clinically spared.^13-16

Most patients with APBD are Ashkenazi-Jewish and have a homozygous GBE1 mutation, p.Y329S. Few are compound heterozygotes of p.Y329S with p.L224P. As expected, carrier parents are unaffected.^1,14,16 A large fraction of patients (30%) carry the p.Y329S mutation in the heterozygous state with no other mutation in the 16 exons of the gene.² There are no known Ashkenazi patients who do not carry at least 1 copy of the p.Y329S mutation. In the current study, we resolved this genetic conundrum and identifed a deep intronic second mutation in the manifesting heterozygotes, which in the homozygous state, is likely lethal. To our knowledge, all known Ashkenazi patients with APBD are now explained.

Glycogen branching enzyme activity was measured in leukocytes, as previously described.^12,13 The study was approved by the Columbia University Medical Center and Hadassah-Hebrew University Medical Center institutional review boards. Written informed consent was obtained from participants.

Two microsatellite markers flanking GBE1 were genotyped at chr3:81312404-81312683 (UCSC hg-19 build; http://genome.ucsc.edu/), with primer sequences F-GTGATGCTCTGGTGCACAT and R-TACCCCCTGATAACCATCCT and product sizes in our patients of 270 base pairs (bp), 278 bp, 280 bp, and 282 bp, as well as at chr3:81751119-81751368, with primers F-CTCAGGTCAGTAGGCACAAAA and R-GTGATGACAACCTACGTGAAAA, sizes 242 bp, 246 bp, and 250 bp.

Messenger ribonucleic acid (mRNA) interruption was determined by a nested polymerase chain reaction (PCR) amplifying complementary DNA (cDNA) in 2 rounds: first, with GBE1-Ex13F primer (5′-ATATGGTTGGCTTGCAGCTC) and oligo dT(T)₁₈ and second, the first product was amplified with GBE1-Ex14 primer (5′-ACTTCCATCCAAGCAAGAGC) and oligo dT(T)₁₈. Because of the nonspecificity of oligo-dT(T)₁₈, the resulting product was sequenced with a third primer, GBE1-Ex15F (5′-GGAATATGGAGGGCATCAGA).

The IVS15 + 5289_5297delGTGTGGTGGinsTGTTTTTTACATGACAGGT mutation was screened on genomic DNA (gDNA) by PCR performed with primers GBE1-Int15F 5′-TGAGTGTTCTTGGAGCATTTTT and GBE1-Int15R7 5′-CCTCCTGAGTAGCTGGGATT, which generate 290 bp and 300 bp normal and mutant allele products, respectively. The sequence was confirmed by Sanger sequencing.

In a cohort of 35 patients with APBD, 16 were homozygous and 19 heterozygous for the c.986A>C (p.Y329S) mutation. This mutation segregated with 2 haplotypes; 3 heterozygous patients harbored a third haplotype associated with a second missense exonic mutation, c.671T>C (p.L224P). However, in the remaining 16 heterozygous patients for the c.986A>C mutation, extensive searches, including exonic and cDNA sequencing, multiplex ligation-dependent probe amplification, and whole-genome sequencing (Complete Genomics; see further on), failed to reveal a second mutation. The possibility arose that in these patients with c.986A>C (p.Y329S), 1 allele alone was somehow sufficient to cause the disease, ie, that these cases were manifesting heterozygotes. However, haplotype analysis showed that all these patients shared a common haplotype separate from their c.986A>C-associated haplotypes, suggesting that they actually did harbor a second mutation, which we were simply failing to detect despite the whole-genome sequencing.

By definition, all manifesting heterozygote patients were heterozygous for the exon 7 c.986A>C mutation in gDNA (Figure 1A). We synthesized cDNA from leukocyte mRNA from these patients using oligo-dT priming (oligo-dT is complementary to the poly-A tail at ends of mRNA). We PCR amplified the cDNA using various primer pairs flanking GBE1 exon 7 and, with most pairs, obtained the heterozygous c.986A>C sequence as in the gDNA. However, when we placed the reverse primer in exon 16, the PCR product was no longer heterozygous but homozygous for c.986A>C (Figure 1B). This suggested that the manifesting heterozygotes generate 1 transcript containing the exon 7 c.986A>C mutation and another transcript that lacks exon 16 or part thereof. We set out to identify and characterize the latter. This putative abnormal transcript had a poly-A tail because our starting cDNA was generated with an oligo-dT primer. We PCR amplified with forward primer in exon 13 and used oligo-dT as the reverse primer. The products were a mix with non-GBE1 sequences, which was unsurprising given the nonspecific nature of oligo-dT priming. To enrich for GBE1, we performed nested PCR, amplifying the exon 13–oligo-dT products with forward primer in exon 14 and oligo-dT reverse and then sequencing the PCR products with an exon 15 forward primer. The sequence obtained was the normal GBE1 coding sequence until the point where exon 16 should have started. At that point, the sequence became double: the normal exon 16 sequence and a second abnormal nonexon 16 sequence. We blasted this unknown sequence against the human genome and found that it was, in part, from GBE1 intron 15. We PCR amplified this intron 15 region in gDNA from manifesting heterozygotes and found that a 9-bp 5′-GTGTGGTGG sequence from intron 15 was replaced by a 20-bp 5′-TGTTTTTTTACATTACAGGT new sequence (Figure 2A). This abnormal sequence contained a highly potent mRNA splice acceptor site (http://www.fruitfly.org/seq_tools/splice.html).¹⁷ Exon 15 spliced into this ectopic splice site rather than into exon 16. Reverse transcription PCR using forward primer from exon 7 and reverse primer from the new abnormal putative exon generated precisely this product, ie, exon 15 spliced into the ectopic splice acceptor, creating a new abnormal exon 16 with a new stop codon (Figure 2B). Therefore, the missing mutation in the manifesting heterozygotes was IVS15 + 5289_5297delGTGTGGTGGinsTGTTTTTTACATGACAGGT, generating an mRNA that encoded a truncated protein (Figure 2B).

The truncated protein was unstable. Western blotting of leukocyte extracts from the manifesting heterozygote patients using an antibody against the middle unaffected portion of the GBE protein revealed more reduced total protein compared with p.Y329S homozygote patients (Figure 2C). The mean (SD) GBE activity in the manifesting heterozygote patients was 8% (6%) of normal (n = 3) compared with 18% (4%) (n = 6) in simultaneously measured p.Y329S homozygotes. The new mutation was present in all manifesting heterozygotes and was not found in 120 anonymous Ashkenazi individuals.

In light of having identified the missing mutation, we reanalyzed our whole–genome sequence data. The mutation was in a region of repetitive sequences. The computerized genome reassembly failed to resolve the repetitive sequences in this region and failed to detect the mutation.

We resolved the genotype of manifesting heterozygous Ashkenazi patients with APBD. They were compound heterozygotes for 2 mutations: c.986A>C (p.Y329S) and IVS15 + 5289_5297delGTGTGGTGGinsTGTTTTTTACATGACAGGT. Both are founder mutations. No patient was homozygous for the new mutation. This fact, combined with the greater severity of the new mutation (as determined by Western blot analysis) and with the absence of any patients with classic GSD-IV within our families with APBD, indicated that homozygosity for the new mutation is likely prenatal lethal. Compound heterozygosity for the 2 mutations did not appear to be more severe than homozygosity for p.Y329S, suggesting that 8% residual GBE activity was not significantly worse than 18%. However, careful clinical reassessment is warranted to confirm this point.

While the mutation identified here is not the first intronic mutation causing neurological disease, it is unusual in its last intron location, which both complicated and simplified its discovery. It was complicated because there was no exon beyond the mutation on which to place a primer and amplify the aberrant transcript. It was simplified because premature termination involving last exons did not cause nonsense-mediated mRNA decay¹⁸ and the aberrant transcript was therefore present and eventually specifically amplifiable.

Whole-genome sequencing shears gDNA, sequences billions of fragments, and bioinformatically reassembles their sequences into a genome. Output files catalog the millions of variants as well as innumerable regions where the program fails, usually in noncoding regions where DNA is repetitive, to generate correct reassembly.¹⁹ Our mutation was in such a region and highlighted present-day shortfalls of whole-genome sequencing in that a mutation in a known locus, ie, in a known single gene, could not be detected. How many such mutations are missed in projects where the locus is unknown?

Our study predicted that therapy might be harder to achieve in compound heterozygous cases (8% residual GBE activity) compared with p.Y329S homozygotes (18% activity). However, new therapies exploiting antisense oligonucleotides may be used to neutralize the cryptic splice site in the former and convert these patients into true unaffected heterozygotes, as are being tested in Duchenne dystrophy.²⁰ To our knowledge, our study is the first that allows specific diagnosis in most Ashkenazi patients with APBD, as well as precise genetic and prenatal diagnosis and counseling.

Corresponding Author: H. Orhan Akman, PhD, Department of Neurology, Columbia University Medical Center, 630 W 168th St, Room 4-431, New York, NY 10032 (hoa2101@cumc.columbia.edu).

Published Online: February 9, 2015. doi:10.1001/jamaneurol.2014.4496.

Author Contributions: Dr Akman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Akman, Kakhlon, Rosenmann, Turnbull, Meiner, Shpitzen, Leitersdorf.

Acquisition, analysis, or interpretation of data: Akman, Kakhlon, Coku, Peverelli, Rosenmann, Rozenstein-Tsalkovich, Turnbull, Meiner, Chamma, Lerer, Shpitzen, Paradas, Wallace, Schiffmann, DiMauro, Lossos, Minassian.

Drafting of the manuscript: Akman, Kakhlon, Turnbull, Meiner, Schiffmann, Lossos.

Critical revision of the manuscript for important intellectual content: Akman, Kakhlon, Coku, Peverelli, Rosenmann, Rozenstein-Tsalkovich, Turnbull, Meiner, Chamma, Lerer, Shpitzen, Leitersdorf, Paradas, Wallace, DiMauro, Minassian.

Statistical analysis: Minassian.

Obtained funding: Akman, Kakhlon, Meiner, Lossos.

Administrative, technical, or material support: Akman, Kakhlon, Coku, Rosenmann, Turnbull, Meiner, Chamma, Shpitzen, Leitersdorf, Lossos.

Study supervision: Akman, Kakhlon, Rosenmann, Turnbull, Meiner, DiMauro.

Conflict of Interest Disclosures: Columbia University has filed for a patent on the mutation in the GBE1 gene. Dr Minassian holds the University of Toronto Michael Bahen Chair in Epilepsy Research. No other disclosures were reported.

Funding/Support: Funding was obtained from the Adult Polyglucosan Body Disease Research Foundation (APBDRF), Keith B. Hayes Foundation, and the Yahel Foundation.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Correction: This article was corrected on March 4, 2015, to fix an error in the byline.