Expanding the genotypic spectrum of TXNL4A variants in Burn‐McKeown syndrome

The developmental disorder Burn‐McKeown Syndrome (BMKS) is characterised by choanal atresia and specific craniofacial features. BMKS is caused by biallelic variants in the pre‐messenger RNA splicing factor TXNL4A. Most patients have a loss‐of‐function variant in trans with a 34‐base pair (bp) deletion (type 1 Δ34) in the promoter region. Here, we identified two patients with BMKS. One individual has a TXNL4A c.93_94delCC, p.His32Argfs *21 variant combined with a type 1 Δ34 promoter deletion. The other has an intronic TXNL4A splice site variant (c.258‐3C>G) and a type 1 Δ34 promoter deletion. We show the c.258‐3C>G variant and a previously reported c.258‐2A>G variant, cause skipping of the final exon of TXNL4A in a minigene splicing assay. Furthermore, we identify putative transcription factor binding sites within the 56 bp of the TXNL4A promoter affected by the type 1 and type 2 Δ34 and use dual luciferase assays to identify a 22 bp repeated motif essential for TXNL4A expression within this promoter region. We propose that additional variants affecting critical transcription factor binding nucleotides within the 22 bp repeated motif could be relevant to BMKS aetiology. Finally, our data emphasises the need to analyse the non‐coding sequence in individuals where a single likely pathogenic coding variant is identified in an autosomal recessive disorder consistent with the clinical presentation.

Extra-craniofacial phenotypes of conductive and sensorineural hearing loss, congenital heart defects, inguinal hernias and short stature are observed in some patients. One BMKS individual has been reported with intellectual disability and developmental delay. 8 Wieczorek et al. identified biallelic variants in TXNL4A as causative in BMKS. 4 Most affected individuals carry a 34-base pair (bp) deletion (chr18: g.77748581_77,748614del [GRCh37, hg19]), known as the type 1 Δ34) in the TXNL4A promoter of one allele combined with a loss-of-function variant on the other allele. Loss-of-function variants include microdeletions, splice site, nonsense and frameshift variants. 4,6 Alternatively, some affected individuals are homozygous for a different 34 bp deletion, (chr18: g.77748604_77,748 637 [GRCh37, hg19], known as the type 2 Δ34) in the TXNL4A promoter. 4,6,7 It is proposed that reduced TXNL4A expression causes BMKS, with complete loss-offunction likely embryonically lethal.
TXNL4A/DIM1 is a spliceosomal U5 small nuclear ribonucleoprotein particle (snRNP) component, responsible for all precursor mRNA (pre-mRNA) splicing. [9][10][11] It is postulated that decreased TXNL4A expression reduces tri-snRNP assembly disrupting splicing of a specific subset of pre-mRNAs required for craniofacial development. 4,12,13 Mis-splicing of pre-mRNAs relevant to craniofacial development would result in the tissue-specific and restricted phenotype of BMKS patients.
A difficulty hindering the diagnosis of BMKS is the identification of the 34 bp TXNL4A promoter deletions from sequencing data. Promoter deletions may not be identified by whole-exome sequencing (WES), while bioinformatics pipelines for whole-genome sequencing (WGS) frequently do not cover non-coding sequences encompassing promoter and deep intronic regions. 14 Here, we identify two unreported individuals with BMKS with novel TXNL4A genotypes. We show that a novel TXNL4A c.258-3C>G splice acceptor variant in one patient, as well as a previously reported c.258-2A>G variant affecting the adjacent nucleotide, cause skipping of the final exon of TXNL4A.
Furthermore, we identify potential transcription factor binding sites within the TXNL4A type 1 and type 2 Δ34 promoter deletions and use a dual luciferase assay to identify a 22 bp repeated motif which is crucial for TXNL4A promoter activity. These findings expand the genetic spectrum of TXNL4A variants underlying BMKS and identify why TXNL4A Δ34 promoter deletions influence TXNL4A expression.

| MATERIALS AND METHODS
See Data S1. to have a heterozygous type 1 Δ34 promoter deletion, which was confirmed by Sanger sequencing (Figure 1).

| Proband phenotyping
The family 1 proband is a white British female only child born to unrelated parents who presented in the genetic clinic in adulthood with mixed conductive sensorineural hearing loss and jaw ankylosis ( Figure 2A). She had previous treatment for bilateral choanal atresia and displayed dysmorphic craniofacial features including lower eyelid coloboma, malar flattening, a high palate and micrognathia, right-sided microtia and protruding ears ( Figure 2B, Table S1). Sequencing revealed a heterozygous chr18:77748298TGG>T (GRCh37), TXNL4A c.93_94delCC (NM_006701), p.His32Argfs*21 (NP_006692) variant with a heterozygous type 1 Δ34 promoter deletion (Table S1). The frameshift variant is not present in gnomAD and has not been previously associated with BMKS. Parental genotyping revealed that the type 1 Δ34 promoter was maternally inherited, while the c.93_94delCC was paternally inherited (Figure 2A, Table S1). The mother is clinically unaffected. The father died at 44 years of oesophageal carcinoma. He possibly had choanal atresia as, at 11-12 years, he had an operation to drill one side out of his nose as his nasal passages had not fully developed. He also possibly had a flat malar region (Table S1). It is possible that the father may be mildly clinically affected based on his reported phenotype. Sanger sequencing of the whole TXNL4A coding and promoter sequence for the father did not reveal any additional variants which could account for his craniofacial features. As he was deceased, the father was not recruited to the 100 K Genomes Project. Therefore, WGS was not available. It is unlikely that the oesophageal carcinoma is related to his TXNL4A genotype as this association has not been described in other carriers of TNXL4A variants. While somatic mutations in some core spliceosome components have been associated with cancer, there are no reports to date of TXNL4A mutations in tumours. 15,16 The family 2 proband is a white British male and only child of  (Table S1). The splice acceptor variant is not observed in the gnomAD population database and has not been previously described in a BMKS patient. However, a variant in the adjacent nucleotide, TXNL4A c.258-2A>G (NM_006701) has been described in an individual with BMKS. 6 In silico prediction of variant pathogenicity suggested both splice site variants are disease-causing by disrupting the splice acceptor site (Table S1). We conducted minigene splicing assays for the c.258-2A>G and c.258-3C>G variants; both led to complete skipping of the TXNL4A final exon (Data S1; Figure S1). Deletion of TXNL4A exon 3 in trans to a type 1 Δ34 has been reported in another BMKS patient. 4 The heterozygous c.258-3C>G splice acceptor variant was maternally inherited while the heterozygous type 1 Δ34 was paternally inherited (Figure 2A, Table S1). Comparison of the clinical features observed in patients here and previously reported patients is provided in Table S1. 3.3 | Identifying putative transcription factor binding sites within the human TXNL4A promoter type 1 Δ34 Wieczorek et al. found that TXNL4A type 1 and type 2 Δ34 deletions reduced promoter activity by 59% and 72%, respectively. 4 This promoter region consists of two repeated 22 bp motifs separated by a 12 bp spacer, with each Δ34 deletion containing one of the 22 bp repeated motifs with the spacer region overlapping the type 1 and type 2 Δ34 ( Figure 3A). These 34 bp regions were proposed to contain binding sites for transcription factors which promote TXNL4A expression, the loss of which cause decreased promoter activity in patients and carriers of the deletions. 4 We predicted potential binding sites for four transcription factors (XBP-1, c-JUN, AhR/ARNT and ATF3) in the type 1 Δ34 ( Figure 3A).
All but three nucleotides in these binding sites were within the promoter were found in the gnomAD database, indicating an important and sequence-specific role in promoter activity (Data S1; Table S2).

| In vitro analysis of putative transcription factor binding sites on promoter function
To test whether the identified putative transcription factor binding sites are important in TXNL4A promoter function, we cloned a 601 bp TXNL4A promoter fragment into a luciferase reporter vector and performed dual luciferase assays. Constructs contained the wild type promoter region, the type 1 Δ34 or several smaller deletions ( Figure 3A). Similar to Wieczorek et al., we found type 1 Δ34 reduced promoter activity to 46% ( Figure 3B). 4 Smaller deletions (deletions 1 and 2) reduced promoter activity to 59% and 63%, respectively, while deletion 3 (12 bp spacer) only reduced promoter activity by 7% ( Figure 3B). Deletion 4 (spanning deletions 1 and 2) reduced activity to 47% ( Figure 3B). Scrambling deletion 4 reduced activity to 70%, suggesting sequence specificity of this region ( Figure 3B). We then deleted the 22 bp repeated motif within the type 1 Δ34 (repeated region 2, RR2) or type 2 Δ34 (repeated region 2, RR2) ( Figure 3A). RR1 reduced promoter activity to 54%, while RR2 reduced activity to 45%, the same as the full type 1 Δ34 ( Figure 3B). Deleting or scrambling both RR1 and RR2 together reduced promoter activity to 10% ( Figure 3B). These findings suggest F I G U R E 3 Analysis of the human TXNL4A promoter. (A) Structure of TXNL4A promoter region affected by type 1 (orange) and type 2 (red) 34 bp deletions in BMKS patients; 12 bp spacer region (yellow) and 22 bp repeated regions (pink). Putative transcription factor binding sites identified using ALGGEN PROMO indicated in grey. Hypothetical deletions 1-4 in luciferase reporter gene constructs are highlighted in purple. B) Effects of TXNL4A promoter deletions on luciferase expression. Relative firefly luciferase expression for each construct, normalised to renilla luciferase expression, is indicated as a percentage of the wild type promoter region expression. n = 4. **p-value <0.01, ****p-value <0.0001 [Colour figure can be viewed at wileyonlinelibrary.com] that RR1 and RR2 contain the critical nucleotides for TXNL4A promoter activity and act independently and cumulatively to promote TXNL4A expression.
This study has reiterated the power of WGS in diagnosing patients with rare disorders and emphasises the need to consider non-coding regions when analysing WGS data, especially when a single pathogenic coding variant is identified in a diseaseassociated gene known to cause a recessive condition consistent with the clinical presentation. We have also developed an analysis approach for screening existing and novel promoter variants in a gene of interest. This approach may prove useful for disorders associated with promoter variants where few patients have been identified and where it is unclear whether a single pathogenic variant or spectrum of different promoter variants underlie the phenotype.

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of the article at the publisher's website.