eP014: Establishment of a clinically validated in vitro functional assay to score pathogenicity of novel GAA variants in Pompe patients


      Pompe disease was added to the Recommended Uniform Screening Panel in the United States in 2015. Currently, 28 states screen for Pompe disease at birth, and patients with late-onset Pompe disease (LOPD) represent about 80% of all confirmed cases. While infantile-onset Pompe disease (IOPD) can often be rapidly diagnosed based on absent or severely deficient GAA enzyme activity and notable clinical features such as cardiomegaly and muscle weakness, diagnostic confirmation of LOPD is often more challenging due to the lack of obvious clinical features in the newborn period. Once a diagnosis of Pompe disease is suspected based on low GAA enzyme activity, GAA sequencing is performed to confirm the diagnosis and better understand a patient’s expected disease course. After a positive newborn screen (NBS), sequencing can reduce the rate of false positives by identifying carriers or those with pseudo-deficiency alleles, and can predict cross reacting immunological material (CRIM) status and timing of disease onset (late vs. infantile). Many cases are identified in which the patient carries one pathogenic or likely pathogenic variant, and one variant of uncertain significance (VUS) which has never been reported.
      Of novel variants, missense variants are especially challenging to classify clinically and of the 626 missense GAA variants annotated in ClinVar, 448 (70%) of these are classified as VUS. Thus, with the increase of novel and unclassified GAA variants due to NBS, there is an urgent need for a reliable and reproducible in vitro functional analysis expression system that can be used to aid in classification of these variants which will enable appropriate clinical management. We established a HEK293 mammalian cell line system for transient in vitro expression of GAA VUS using site-directed mutagenesis in order to analyze the impact of a VUS on acid alpha glucosidase (GAA) enzyme activity. Using this assay, we analyzed eight VUS in six individuals with confirmed or suspected Pompe disease.


      pcDNA3.1 cloned GAA open reading frame (NM_0000152.5) was used as a reference sequence to generate targeted variants in the laboratory. Our expression system was validated using a set of twelve known GAA variants including seven known pathogenic variants (five missense and two nonsense) and five known benign variants (three missense and two synonymous change). Transfected cloned GAA (NM_000152.5) open reading frame was used as a positive (+ve) control, while un-transfected HEK293 cell line cultures were used as negative (–ve) controls. Intra- and inter-assay reproducibility was established.
      pcDNA3.1-GAA targeted variants (Clinical controls and VUS) were generated by QuickchangeII XL-site directed mutagenesis (Agilent Technologies). Cloned GAA variants were transiently expressed in HEK293 cell lines and cultured for forty-eight hours for expression studies. Sonicated cell lysates were used to assess the residual GAA activity for each VUS. Artificial substrate 4-Methylumbelliferone (4MU-G) (Sigma-Aldrich) was used to measure residual GAA enzyme activity indirectly using a fluorescence plate reader. Mutant protein was further characterized by western blot analysis using anti-GAA antibodies (Abcam).
      Results of these studies were used as evidence toward classification or reclassification of each selected VUS.


      Of the seven known pathogenic control variants, resulting enzyme activity level ranged from 0-11% as compared to the wild type (WT) +ve control. Of the five known benign variants, resulting enzyme activity ranged from 54-100% of the WT +ve control. Thus, our assay fulfills the criteria set forth by the Clinical Genomics Sequence Variant Interpretation Working Group for functional studies that can be used at the supporting level of evidence for variant classification (PS3/BS3).
      Three GAA missense VUS: c.1721T>C (p.Leu574Pro), c.1880C>T (p.Ser627Phe), and c.2450A>G (p.His817Arg) were expressed and assayed separately and were shown to have complete loss of GAA catalytic activity. Using Western blot analyses, these deleterious variants demonstrated only an inactive precursor protein band at 110kDa. Mature protein (76 and 70kDa) bands were absent, indicating a failure of protein processing attributable to these VUS. Four GAA VUS: c.266G>A (p.Arg89His), c.316C>T (p.Arg106Cys), c.664G>A (p.Val222Met), and c.1103G>A (p.Gly368Asp) were demonstrated to be damaging with reduced residual activity and reduced protein amounts. The final VUS, c.1642G>T (p.Val548Phe) demonstrated normal GAA expression and enzyme activity.
      See Table for complete results of each expression study and final evidence toward classification or reclassification.


      We have established a HEK293 mammalian cell line system, which is robust and reproducible, for the in vitro expression of novel GAA variants using a site-directed mutagenesis system. This system has allowed us to study the impact of VUS on GAA enzyme activity and protein processing. Our work has allowed reclassification of 6 of the 8 variants studied as either ‘likely pathogenic’ or ‘likely benign’ following ACMG-AMP classification guidelines and other published standards, and thus has helped improve the management of these patients.
      Tabled 1
      Patient/Case ReportGAA missense variant of unknown significance (VUS)Pathogenicity calculation as per LSDVCEP guidelines following ACMG-AMP standards
      Functional study results [PS3_supporting/BS3_supporting]Population databases Absent /Allele frequency <0.001 in all continental gnomAD populations

      In trans with P/LP variant [PM3/BP2]Computational prediction SIFT/Polyphen [PP3/BP4]Clinical phenotype for Pompe disease:

      DBS glucosidase activity below cut off and absence of pseudodeficiency allele

      Erasmus Mc(# patients/Phenotype with null alleleEvidence based Pathogenicity calculation


      10.7% of normal PS3_supportingallele frequency <0.001 in all continental gnomAD populations PM2_supportingc.-32-13T>G,


      PP3DBS glucosidase activity- 1.59μmol/lt, CK upper end of limit PP4_Moderate-Likely pathogenic (No conflict)
      2c.1103G>A (p.Gly368Asp) [VCV001037598]2.8% of normal PS3_supportingNot present in gnomAD PM2_supportingc.-32-13T>G,


      PP3DBS glucosidase activity- 0.42μmol/lt CK,AST,ALT elevated PP4_Moderate-Likely pathogenic (No conflict)
      3c.1721T>C (p.Leu574Pro) —No detectable activity PS3_supportingNot present in gnomAD PM2_supportingc.-32-13T>G,


      PP3DBS glucosidase activity- 0.61μmol/lt CK,AST,ALT elevated PP4_Moderate1, unknownLikely pathogenic (No conflict)
      4c.266G>A (p.Arg89His) [VCV000283219]21.3% of normal PS3_supportingallele frequency <0.001 in all continental population gnomAD PM2_supportingc.1478C>T (p.Pro493Leu) VCV000379593 PM3PP3DBS glucosidase activity- 4.10pmol/punch/hr. Low activity in fibroblast. Biomarkers WNL PP4_Moderate2/unknownVUS Lowered but significant activity with normal processing


      11.7% of normal PS3_supportingSouth Asian allele freq >0.005 (gnomAD) BS1c.2450A>G (p.His817Arg) VUS ---PP3DBS glucosidase activity- 8.2% of normal. CK slightly elevated16, unknownVUS (Evidences in conflict)
      c.2450A>G (p.His817Arg) ---No detectable activity PS3_supportingNot present in gnomAD PM2_supportingc.664G>A

      (p.Val222Met) VCV000290220 VUS ---
      PP3PP4 Moderate-Likely Pathogenic (No conflict)


      100% of normal, BS3_supportingNot present in gnomAD PM2_supportingc.2560C>T; (p.Arg854*) VCV000004034BP4DBS glucosidase activity- 1.52pmol/punch/hr CK, AST,ALT, Hex4 elevated3, unknownLikely Benign (No conflict)


      No detectable activity PS3_supportingNot present in gnomAD PM2_supportingPM3PP3PP4 Moderate5, classic infantileLikely Pathogenic (No conflict)