Systematic Assessment of GATA2 Genetic Variation Reveals the Presence of Novel Disease-Causing Synonymous Exonic Mutations

Among its different clinical presentations, GATA2 deficiency had been recognized as the most frequent hereditary predisposition to pediatric myelodysplastic syndromes (MDS), accounting for 15% of MDS with excess of blasts (MDS-EB), with a particularly high prevalence among children with monosomy 7 (37%). A pattern of three major classes of germline mutations can be recognized in the roughly 350 reported cases with GATA2 deficiency: I) protein truncating (stop-gain, frameshift, splice-site), II) missense within or adjacent to zinc finger 2 (ZnF2), and, III) noncoding substitutions in the conserved EBOX-GATA-ETS regulatory region in intron 4, (NM_032638.4). The rare aberrations (class IV) include in frame indels and whole gene deletions. For the purpose of this study we introduced the above classification to systematically categorize various types of GATA2 variations and to assess the prevalence and pathogenic significance of silent (synonymous) substitutions, referred to as class V.

Our study cohort consisted of 723 children and adolescents with primary MDS enrolled in prospective registry of the European Working Group of childhood MDS (EWOG-MDS). Typical GATA2 germline mutations were found in 69/723 cases. Additional 28 GATA2-deficient individuals with heterogeneous phenotypes diagnosed at our institutions were added to the analysis. We excluded patients with secondary MDS, in which GATA2 is not etiologically involved as previously reported by our group. Results were compared to those of 138632 individuals in the Genome Aggregation Database (GnomAD). Various in-silico prediction tools to evaluate deleteriousness were employed, and selected mutations were investigated in a GATA reporter system.

First, we categorized the "typical" germline GATA2 mutations identified in 96 patients. Class I = truncating mutations were present in 51.5% (49/96), class II = missense in ZnF2 in 37.5% (36/96), class III = deep intron 4 in 8.3% (8/96) and class IV = other in 3.15% (3/96) of patients. Genotype-phenotype analysis failed to establish a significant correlation. We next systematically assessed the prevalence of other potentially pathogenic coding variants. Missense alterations p.Ala164Thr, p.Pro161Ala, and p.Pro250Ala (present in 34.8%, 1.6% and 0.9% GnomAD individuals) were neither significantly enriched, nor they displayed "synergistic" co-occurrence in pediatric MDS. This essentially depicts an unlikely role in MDS predisposition. We further performed transactivation activity in a GATA reporter system for p.Pro161Ala and observed no significant alteration.

Finally, we focused on the evaluation of synonymous exonic substitutions resulting in nucleotide exchange without alteration of protein sequence. Recent cancer studies proposed that synonymous mutations might alter protein folding and stability, also such variants can introduce novel splice sites leading to premature RNA degradation. The synonymous substitutions that are common according to GnomAD: p.Pro5Pro (88.3% GnomAD), p.Thr188Thr (8.0%) and p.Ala411Ala (4.3%) showed similar frequencies in our patient cohort. Ultimately, we identified 6 patients with 5 novel or very rare synonymous exonic substitutions, referred to as "class V" variants (Fig. 1): p.Thr117Thr found in 2 unrelated pedigrees, p.Gly327Gly, p.Leu217Leu, p.Ala341Ala, and p.Pro472Pro (detected in 0, 0, 2, 6, and 70 out of >130 Tsd. GnomAD individuals, respectively). Median age at MDS diagnosis was 12 (4-15) years and 4/6 patients have acquired monosomy 7.

In silico prediction assigned a high chance of splice defects to p.Thr117Thr, p.Leu217Leu, and p.Ala341Ala variants. Sequencing from blood of 2 unrelated patients with p.Thr117Thr confirmed complete loss of mutated allele in mRNA as compared to 50:50% distribution in DNA; while the degradation of p.Leu217Leu mutant mRNA was less prominent. Expectedly, GATA reporter assays performed for synonymous variants did not reveal altered transactivation. Further ongoing studies aim at detailed examination of altered splicing and characterization of protein stability.

In sum, we establish that noncoding substitutions in GATA2, believed thus far to be silent, constitute a novel disease-causing mutational class. In contrast, rare SNPs (p.Pro161Ala and p.Pro250Ala) have a likely neutral effect towards MDS predisposition.