DNMT3A is a highly conserved 130 kDa protein encoded by 23 exons[5]. DNMT3A is expressed both embryonically and postnatally and contributes to important developmental processes including genomic imprinting and maturation of the nervous system. Pathogenic variants in DNMT3A associated with variable overgrowth, intellectual disability, and autism spectrum disorder (ASD)[2, 3]. Overgrowth is mainly seen in Tatton-Brown-Rahman syndrome (TBRS), which is characterized by overgrowth, distinctive facial appearance, and intellectual disability[2, 6, 7]. Somatic variants in DNMT3A are also common in hematologic malignancies and are known to impact prognosis, with a hot spot mutation located at arginine 882 (R882) in the catalytic domain and additional pathogenic variants distributed throughout the gene. These variants are observed in 17–34% of cases with acute myeloid leukemia (AML), 3–13% of cases with myelodysplastic syndromes (MDS), 10% of cases with myeloproliferative neoplasms (MPN), and 4% of cases with MDS/MPN overlap neoplasm[8, 9].
However, growth retardation associated with DNMT3A is rare. We reported here a 2.8-year Chines boy presented with growth retardation who was associated with a de novo DNMT3A variation, c.911_913del (p.S304del). His motor and intellectual developmental milestone were slightly delayed. No abnormal dysmorphic features and signs of ASD were found. He did not fulfill the diagnostic criteria of Heyn-Sproul-Jackson syndrome (HSJS), which is also caused by DNMT3A variations mainly presented with mirocephaly, short stature and impaired intellectual development.
It is contradictory or mysterious that why DNMT3A variations could induce two opposite conditions, overgrowth vs growth retardation. There might be several reasons for this phenomenon as below.
First, the variation site and position might be different. As seen in patients with AML, there are two major classes of DNMT3A variation. One class is the highly recurrent set of variations at R882, while the other class is the major locations of variations[10]. It is known that there are three domains in DNMT3A, a proline-tryptophan-tryptophan-proline (PWWP) domain, an ATRX-DNMT3A-DNMT3L-type zinc finger (ADD) domain, and a C-terminal DNA methyltransferase (MTase) domain[5]. In hematologic malignancies, although somatic variants of DNTM3A cluster in all three functional domains, which are mainly seen in MTase domain[7]. The study of Sendzikaite[11] also proved this point. They established a mouse model carrying a D329A point variant in the DNMT3A PWWP domain, which caused dominant postnatal growth retardation, and resulted in progressive DNA hypermethylation and de-repression of developmental regulatory genes in adult hypothalamus at the molecular level[11]. The de novo DNMT3A variation, c.911_913del (p.S304del), of our boy was adjacent to the D329A, both located in the DNMT3A PWWP domain, and induced the same clinical feature of growth retardation. It was a pity that we could not measure the DNA methylation changes in our boy. However, overlap in the spectrum of DNMT3A variants is observed between TBRS and hematological malignancies[7], further studies are needed to elucidate the relationship between phenotype and genotype of DNMT3A.
Second, the variations in other associated genes might effected the phenotypes of DNMT3A. It is evident that there are some genes frequently co-occurrence with DNMT3A variants in haematopoietic malignancy including FLT3, TET2, NPM1, IDH1, IDH2, JAK2 and CEBPA[4, 12–16], while in neurodevelopmental disorders including MeCP2, NSD1[17–19]. Polonis[20] describe a 5-year-old female with a paternally inherited pathogenic variation in EZH2 (c.2050C > T, p.Arg684Cys) and a maternally inherited 505-kb duplication of uncertain significance at 2p23.3 (encompassing five genes, including DNMT3A) who presented with intrauterine growth restriction, slow postnatal growth, short stature, hypotonia, developmental delay, and neuroblastoma diagnosed at the age of 8 months. It has been previously shown that EZH2 directly controls DNA methylation through physical association with DNMTs, including DNMT3A. Our boy also carried an IGF1R (Insulin-like growth factor 1 receptor) heterozygous variant (c.1171C > T, p.R391C), whether which has some interaction between DNMT3A is not sure. However, it is proved that IGFBP3 (Insulin like growth factor binding protein 3) is an established target of DNMT3a, MTA1 (Metastasis-associated protein 1) can modify the expression of DNMT3a and IGFBP3 in breast cancer[21].
Third, despite a recognized role of DNMT3a in human diseases, it remains unclear to what extent these mutations result in differential phenotypes. The clinical phenotypes of DNMT3a are continuously expanding besides the known neurodevelopmental disorder and hematologic malignancies, such as Melanoma[22], breast cancer[23], congenital myopathy[24]. The growth retardation of our case might just due to the DNMT3A variation in the PWWP domain.
In conclusion, we reported an infantile case presented with growth retardation associated with a novel variation (c.911_913del, p.S304del) in the DNMT3A PWWP domain, and speculated the possible reasons. The shared genes, the allele-specific gene networks and cellular processes that may underlie the spectrum of phenotypes of DNMT3A disorders, further studies are needed.