Late-onset hearing loss case associated with a heterozygous truncating variant of DIAPH1

Abstract

DIAPH1 is a formin homology F-actin elongating protein encoded by DIAPH1. Homozygous recessive variants resulting in the loss of DIAPH1 function cause seizures, cortical blindness and microcephaly syndrome (SCBMS), but hearing loss has not been reported. In contrast, dominant variants of human DIAPH1 are associated with DFNA1 nonsyndromic sensorineural hearing loss. The deafness phenotype is due partly to abnormal F-actin elongation activity caused by disruption of the DIAPH1 autoinhibitory mechanism. We report an elderly female heterozygous for the c.3145C>T: p.R1049X variant who showed late-onset sensorineural hearing loss in her fifth decade. p.R1049X lacks F-actin elongation activity because this variant truncates one third of the FH2 domain, which is vital for DIAPH1 dimerization and processive F-actin elongation activity. Concordantly, no increase of F-actin nor processive F-actin elongation activity was observed after overexpression of p.R1049X DIAPH1 in HeLa cells or by single-molecule microscopy using Xenopus XTC cells. However, overexpression of the p.R1049X variant impairs formation of cell-cell junctions and mitosis. We speculate that late-onset hearing loss is a long-term consequence of heterozygosity for the recessive p.R1049X variant, a phenotype that may have been overlooked among carriers of other recessive alleles of DIAPH1.

Graphical Abstract

Schematic DIAPH1 gene structure and DIAPH1 domains, including c.3145C>T: p.R1049X mutation. Cells transiently expressing AcGFP-DIAPH1 (p.R1049X) mutant shows impaired cell division (arrows, lower left) and spaces at cell-cell junctions (circles, lower right). Lower middle panels show single-molecule fluorescent microscopy expressing AcGFP-fused DIAPH1 (p.R1049X or wild-type). Fluorescent puncta indicating p.R1049X mutant show no directional movement.

Introduction

Diaphanous-related formin 1 (DIAPH1), encoded by DIAPH1, is a formin homology protein that can processively elongate F-actin¹. In human, DIAPH1 variants are associated with autosomal dominant hearing loss (DFNA1) resulting from constitutive activation of F-actin elongation by disrupting the wild type autoinhibition between the DIAPH1 inhibitory domain (DID) and autoregulatory domain (DAD)². Heterozygosity for the missense variant p.A265S in the DID, and five other variants in or near the DAD (p.A1221VfsX22, p.E1184AfsX11, p.E1192_Q1220del, p.R1213X and p.A1210SfsX31) have been reported in families segregating hearing loss^2–9. In contrast, three biallelic recessive truncating variants (p.F923fs, p.Q778X p.R1049X) and one splice variant (c.684+1G>A) that cause a complete loss of DIAPH1 protein were associated with seizures, cortical blindness and microcephaly syndrome (SCBMS; OMIM: #616632)^10–12. DIAPH1 is involved in a variety of cell functions including cytokinesis, cell migration, stress fibers, filopodia and cell-cell junctions although its physiological functions are still unclear^13–15.

Here, we report a person with late-onset hearing loss who is heterozygous for the DIAPH1 truncation variant c.3145C>T: p.R1049X. This variant is missing F-actin elongation activity and impairs formation of cell-cell junctions and cell proliferation. We speculate that late-onset hearing loss is a long-term consequence of heterozygosity for the p.R1049X variant.

Materials and Methods

Patient and Sanger sequencing

Genomic DNA was sampled after obtaining written informed consent. Exome sequencing (ES) was performed followed by filtering steps (Table S1). Exon 23 of DIAPH1 (NM_005219.4) was amplified by PCR using primers tagcctctatgagttcagtgttgg and tgaactgaaaatgctctcacatct, which was Sanger-sequenced to confirm the variant. Details of other methods are in the Supplementary Information.

Results and Discussion

A 72-year-old woman (SB70–122) with a family history of hearing loss has bilateral, moderate hearing loss across all frequencies (Figure 1, A and B), while other family members declined to participate. No other abnormalities were found upon a physical examination and blood tests did not reveal a meaningful abnormality either in the platelet count nor its morphology, excluding macrothrombocytopenia. Variants from ES were filtered using bioinformatics tools, allele frequencies and relevance to hearing loss (Table S1). We finally identified two candidates, c.3145C>T:p.R1049X in DIAPH1 and c.1703_1712dupTTGCCCTCCC:p.Ile572fs in WFS1. WFS1 was excluded since variants associated with autosomal dominant nonsyndromic hearing loss are almost always missense alleles via ‘gain-of-function’, while ‘loss-of-function’ variants including nonsense or frameshift variants are associated with Wolfram syndrome¹⁶. Approximately 200 cases associated with autosomal dominant non-syndromic hearing loss (DFNA 6/14/38) found in ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) are missense mutations mainly in WFS1 exon 8¹⁷. A final candidate nonsense variant, c.3145C>T, in exon 23 of DIAPH1 (NM_005219) replaced an arginine codon with a stop codon truncating one third of the FH2 domain (Figure 1, C and D). Detailed genotypic characteristics of seventeen candidate variants before the final filtering step are summarized (Table S2).

Figure 1: Clinical data of the proband and identified DIAPH1 variant.

(A) Audiogram of the proband (72-year-old female) showing bilateral, moderate hearing loss. Thresholds are shown in red and blue for her right and left ears, respectively.

(B) Pedigree of the family. Auditory functions of family members (asterisks) were interviewed since other family members declined to participate in this study. Filled symbols indicate individuals with hearing loss. Diagonal lines represent deceased individuals.

(C) Candidate variant validated by Sanger sequencing.

(D) Schematic DIAPH1 gene structure and DIAPH1 domains. Previously-reported variants affect either the autoinhibitory domains DID or DAD. The p.R1049X variant causes truncation within the FH2 domain.

Almost all reported variants associated with DFNA1 deafness affect residues in or near the autoinhibitory DID and DAD^2–9 with one exception¹⁸. Thus, we considered whether or not p.R1049X can be the cause of progressive, late onset hearing loss via a novel mechanism and not constitutive activation of F-actin elongation activity. We evaluated the F-actin elongation activity of p.R1049X after overexpression in HeLa cells. Wild-type and a constitutively active variant associated with hearing loss, p.R1213X², were used as controls. Abnormal, microvilli-like structures indicating upregulated actin polymerization were observed only in cells expressing AcGFP-DIAPH1 (p.R1213X), but not in cells expressing AcGFP-fused DIAPH1 (p.R1049X) or wild-type DIAPH1 (Figure 2A), suggesting that F-actin elongation activity of the p.R1049X variant is lost or significantly disrupted. Both AcGFP-fused DIAPH1 (p.R1049X) and DIAPH1 (p.R1213X) localized at the plasma membrane (Figure 2B) probably via exposed DIDs due to the lack of functional DADs¹⁹.

Figure 2: Disrupted F-actin elongation activity and autoinhibition in p.R1049X.

(A) Actin structures of HeLa cells expressing AcGFP-fused wild-type, (p.R1049X), or (p.R1213X) DIAPH1. Cells expressing DIAPH1 are indicated by asterisks. Abnormal, microvilli-like structures indicating upregulated actin polymerization are observed in cells expressing AcGFP-DIAPH1 (p.R1213X) (magenta arrows) while microvilli in cells expressing AcGFP-DIAPH1 (p.R1049X) appear normal (arrows). Stress fibers appear normal in all cells. Bars, 10 μm.

(B) Localization of AcGFP-fused DIAPH1 (p.R1049X) and DIAPH1 (p.R1213X) at the plasma membrane (arrows). Bars, 10 μm.

(C) Single-molecule fluorescent microscopy of XTC cells expressing AcGFP-fused DIAPH1 (p.R1049X, wild-type or p.R1213X). Fluorescent puncta appear in cells expressing p.R1049X without showing directional movements (indicated by circles). Directional movements observed in cells expressing wild-type DIAPH1 and the p.R1213X variant indicating intact F-actin elongation activities. Directionally-moving molecules are more frequent in cells expressing constitutively-active p.R1213X variant compared with cells expressing wild-type DIAPH1 (circles and trajectories indicate molecules showing directional movements for more than two frames; crosses indicate disappearance). Time-lapse, every 200 ms. Bars, 5 μm.

Corroborating data were obtained using single-molecule fluorescence microscopy of Xenopus XTC cells expressing AcGFP-fused wild-type and variant DIAPH1 (Figure 2C). The number of fluorescent puncta appeared in the cytoplasm was larger in cells expressing AcGFP-DIAPH1 (p.R1049X) than those expressing wild-type AcGFP-DIAPH1, indicating that p.R1049X protein is likely trapped by structures limiting diffusion. AcGFP-DIAPH1 (p.R1049X) molecules did not show directional movements and appear to lack processive F-actin elongation activity. Immunoprecipitation assays showed that truncated FH1-FH2 of p.R1049X DIAPH1 cannot dimerize with intact FH1-FH2 domains (Figure S1).

Overexpression of the p.R1049X variant impairs formation of cell-cell junctions and mitosis. MDCK cells overexpressing AcGFP-DIAPH1 (p.R1049X) have heterogenous shapes (particularly in size) and form a layer significantly thicker than cells expressing AcGFP-fused wild-type DIAPH1 (Figure 3, A and B). HeLa cells overexpressing AcGFP-DIAPH1 (p.R1049X) die more frequently and undergo fewer mitoses compared to cells overexpressing wild-type AcGFP-DIAPH1 (Figure 3C). Immunoblotting of HeLa cell lysates showed that expression levels of AcGFP-DIAPH1 (p.R1049X) were lower than AcGFP-fused wild-type and the p.R1213X variant (Figure 3D). The p.R1049X protein may be translated only at a low efficacy, unstable in the cytoplasm or lethal for cells at high expression. DIAPH1 is involved in the maintenance and stabilization of adherens and tight junctions¹³ and in the cortical tension and spindle assembly during mitotic metaphase and anaphase^14,15. Overexpressed p.R1049X DIAPH1 may interfere or compete with the functions of wild-type DIAPH1, perhaps explaining how one copy of the p.R1049X DIAPH1 allele causes late-onset hearing loss.

Figure 3: Impaired formation of cell-cell junctions and mitosis caused by p.R1049X.

(A) Increased height of MDCK cells expressing AcGFP-DIAPH1 (p.R1049X). Average height of cells expressing AcGFP-DIAPH1 (p.R1049X) were significantly higher than those expressing wild-type AcGFP-DIAPH1. ***P = 0.0002 by Student’s t-test, n = 5 each. Bars, 10 μm.

(B) Impaired formation of cell-cell junctions (Stacked images of the same samples shown in A). MDCK cells expressing AcGFP-DIAPH1 (p.R1049X) have heterogenous shapes compared with cells expressing wild-type AcGFP-DIAPH1. Cell-cell junctions are visualized using anti-ZO1 antibodies. Bars, 20 μm.

(C) Impaired mitosis caused by DIAPH1 (p.R1049X). HeLa cells died during mitosis and floating are indicated by arrows. The number of undergoing mitosis was significantly fewer among cells overexpressing AcGFP-DIAPH1 (p.R1049X) than cells overexpressing wild-type AcGFP-DIAPH1 (***P = 0.0006) while the number of cells that died during mitosis was significantly larger among cells overexpressing AcGFP-DIAPH1 (p.R1049X) than cells overexpressing wild-type AcGFP-DIAPH1 (**P = 0.0019). Student’s t-test, n = 5 each.

(D) Immunoblotting showing low expression level of AcGFP-DIAPH1 (p.R1049X). The amount of GFP was compared between lysates of HeLa cells expressing AcGFP-fused wild-type DIAPH1, DIAPH1 (p.R1213X) and DIAPH1 (p.R1049X) using an anti-GFP antibody.

p.R1049X DIAPH1 and other large truncating variants (p.F923fs, p.Q778X, c.684+1G>A) has been considered to be non-functional. Homozygosity for these variants have similar phenotypes characterized by seizures, cortical blindness and microcephaly^10–12. In these studies, auditory functions were reported to be normal via interviews with some patients and by screening at birth in one patient using ABR and ASSR. However, the long-term consequence of heterozygosity for the p.R1049X variant, especially a mild to moderate progressive hearing loss, may have been overlooked or could not be evaluated since sensitive auditory tests, such as pure-tone audiogram, would be difficult for patients due to the significantly delayed psychomotor development with severe intellectual disability, poor speech and lethality in the first or second decade. Hearing loss caused by the long-term consequence of p.R1049X DIAPH1 is hardly distinguishable in an outpatient clinic from age-related hearing loss. Here, we propose that one copy of the p.R1049X allele, previously thought to be benign in carriers, is associated with a late-onset hearing loss.