Craniosynostosis syndromes encompass a large group of heterogeneous conditions characterized by the premature closure of one or more cranial sutures, which leads to abnormal skull shape and distinctive craniofacial deformities. Their combined estimated incidence is approximately one in 2100 live births (Wilkie et al., 2017). Six genes, FGFR2, FGFR3, TWIST1, EFBN1, TCF12 and ERF, account for approximately 84% of all genetically diagnosed cases (Paumard-Hernández et al., 2014; Wilkie et al., 2017; Hyder et al., 2021). Recently, a seventh gene has been added to this list, SMAD6 (Calpena et al., 2020).

Heterozygous loss-of-function variants in TWIST1 cause Saethre-Chotzen syndrome (SCS, MIM 101400), characterized by coronal suture synostosis, dysmorphic facial features, ptosis, hypertelorism and distal limb abnormalities such as brachydactyly or cutaneous syndactyly (El Ghouzzi et al., 1997; Howard et al., 1997).

TWIST1, located on chromosome 7p21.1 (Bourgeois et al., 1996), encodes a 202 amino acid class II basic helix-loop-helix (bHLH) transcription factor involved in calvarial development. TWIST1 plays an important role in the formation of the fronto-parietal boundary within the developing coronal suture (Merrill et al., 2006; Yen et al., 2010), and exerts an anti-osteogenic function preventing its premature closure (Yousfi et al., 2001; Bialek et al., 2004; Ting et al., 2009; Teng et al., 2018).

The major underlying pathogenic mechanism responsible for the craniofacial and limb anomalies observed in SCS patients is haploinsufficiency, usually caused by heterozygous variants affecting the coding region of the gene or structural variants, such as inversions or translocations encompassing TWIST1 regulatory elements, leading to functional loss of one gene copy (Hyder et al., 2021; Hirsch et al., 2022).

However, two pathogenic variants, in the 5′ untranslated region (5′ UTR) of TWIST1 have been recently reported in two patients with SCS (Zhou et al., 2018). Both were shown to create novel upstream initiation sites and a reduction in translation of TWIST1. Here, we report the clinical, genetics and functional characterization of a third variant in the 5′ UTR of TWIST1, in a patient with bicoronal craniosynostosis.

SCS is one of the most common craniosynostosis syndromes. Although the vast majority of SCS causing variants are located within the coding region of TWIST1 (Zhou et al., 2018), two pathogenic variants, c.-263C>A and c.-255G>A, have been identified in the non-coding regulatory region in the 5′ UTR of TWIST1, located in exon 1. Both variants create an alternative upstream ATG that affects translation of the TWIST1 mORF (Zhou et al., 2018).

Here, we report a third pathogenic variant in the 5′ UTR of TWIST1 (c.-18C>T), very close to the main start site, which is predicted to create an out-of-frame upstream ATG codon. The de novo variant was identified in a proband with SCS, who presented with bilateral coronal synostosis with large fontanels, intrasutural bones, and extensive parietal and occipital foramina (Figure 1A).

The context and the position of the ATG play a key role in the initiation of translation (Kozak, 1981). In eukaryotes, the flanking sequence of the ATG initiation codons has a conserved pattern of nucleotides known as the Kozak consensus (A/GCCATGG) (Kozak, 1986). Thus, the efficiency of an alternative upstream ATG to serve as translation initiation codon is influenced by the strength of the Kozak consensus. The optimal context requires a purine (A/G) at position -3 and a guanine at +4 is also preferable (Kozak, 1986) (Figure 2B). All three variants have a guanine at -3 of the Kozak sequence, whilst only one of the three, c.-263C>A has a guanine at +4. In silico analysis suggested that the c.-18C>T variant could act as potential competitor with the main ATG start site similarly to the c.-263C>A control mutant (Figure 2B).

The presence of an upstream ATG codon might reduce or abolish translation from the downstream main initiation site (Lomedico and McAndrew, 1982) so we tested whether the variant could reduce TWIST1 translation by comparing the luciferase reporter activity in HEK293 cells transfected with either the c.-18C>T variant, WT or the previously reported variant c.-263C>A as a positive control (Zhou et al., 2018). We observed that Renilla luciferase activity was significantly reduced for the c-18C>T mutant compared to WT, similar to that observed for the positive control (>80% and >85%, respectively) (Figure 2C).

Previously, it was shown that the length of the upstream ORF and the distance between its stop codon and the wild type start initiation site are important for the repressive effect on translation of TWIST1 (Zhou et al., 2018). Translation of the majority of eukaryotic mRNAs occurs by binding of the 43S preinitiation complex to the 5′ cap, and then scanning along the leader sequence in search for the first AUG codon available. Upstream ATGs and ORFs can also influence ribosome scanning efficiency, and thus modulate the translation levels of the main ORF (Iacono et al., 2005).

If translated, the c.-18C>T variant is predicted to result in an aberrant protein of 242 amino acids, while the c.-255G>A variant produces an 85 amino acid longer TWIST1 protein through the generation of an in-frame ATG. In contrast, the c.-263C>A variant creates an out-of-frame ORF which terminates in a stop codon located 59 bases upstream of the mORF. An in vitro transcription-translation assay and subsequent visualization of the protein products would determine if an aberrant protein is produced. If so, this aberrant mutant protein may be degraded or lack the dimerization and DNA binding capability.

With the three variants now identified to date in TWIST1, it is important that the entire 5′ UTR is screened for variants in patients with SCS, especially in genetically underdiagnosed SCS cases negative for coding or copy-number/structural variants. Although this region in TWIST1 is exonic, it is a region that may be excluded from analysis for various reasons: it is not necessarily included in customized panel designs, bioinformatic analysis often excludes these regions and finally it is poorly covered in exomes. Pathogenic variants in the 5′ UTR of two other craniosynostosis genes also create upstream ORF-creating variants, EFBN1 (Twigg et al., 2013; Romanelli Tavares et al., 2019) and more recently, SMAD6 (Calpena et al., 2020). Thus, variants within the 5′ UTR of genes, and especially in haploinsufficient genes, should be carefully assessed in patients in whom the molecular defect has not been identified. Similar cases may be increasingly identified with the implementation of whole genome sequencing in healthcare but functional studies will be necessary to confirm the pathogenicity of these upstream ATG generating variants.

We have identified a third upstream ORF-generating variant in the 5′ UTR of TWIST1 in a child with SCS, which reduces the translation efficiency of the TWIST1 mORF. The child now has a definitive clinical diagnosis which enables both appropriate clinical monitorization and genetic counselling. Thus, this case report shows the necessity for the creation of platforms/services for the functional characterization of variants of unknown significance within national health systems.