Transcriptional control of SLC26A4 is involved in Pendred syndrome and nonsyndromic enlargement of vestibular aqueduct (DFNB4)

Abstract

Although recessive mutations in the anion transporter gene SLC26A4 are known to be responsible for Pendred syndrome (PS) and nonsyndromic hearing loss associated with enlarged vestibular aqueduct (EVA), also known as "DFNB4," a large percentage of patients with this phenotype lack mutations in the SLC26A4 coding region in one or both alleles. We have identified and characterized a key transcriptional regulatory element in the SLC26A4 promoter that binds FOXI1, a transcriptional activator of SLC26A4. In nine patients with PS or nonsyndromic EVA, a novel c.-103T-->C mutation in this regulatory element interferes with FOXI1 binding and completely abolishes FOXI1-mediated transcriptional activation. We have also identified six patients with mutations in FOXI1 that compromise its ability to activate SLC26A4 transcription. In one family, the EVA phenotype segregates in a double-heterozygous mode in the affected individual who carries single mutations in both SLC26A4 and FOXI1. This finding is consistent with our observation that EVA occurs in the Slc26a4(+/-); Foxi1(+/-) double-heterozygous mouse mutant. These results support a novel dosage-dependent model for the molecular pathogenesis of PS and nonsyndromic EVA that involves SLC26A4 and its transcriptional regulatory machinery.

Figure 1.

FOXI1-binding affinity to wild-type and mutant (c.−103T→C) FBS1. A, Conservation of FBS1 and FBS2. FBS1 and FBS2 are shown in bold, arrows indicate orientation, the c.−103T→C mutation is indicated by the vertical arrow, and the oligonucleotide EMSA probe is underlined. B, ³²P-labeled double-stranded oligonucleotide probes containing FBS1 and FBS2 bind to FOXI1 protein as shift bands. Protein binding to the mutant probe (lane 2) is significantly weaker than that to the wild-type probe (lane 3). Protein binding to the wild-type labeled probe was competed with either wild-type (lanes 4–7) or mutant (lanes 8–11) unlabeled competitor (10–100 molar fold excess). Mutant competitor competed significantly less efficiently than wild-type competitor. C, Quantification of EMSA competition result: intensity of the top shift band of lanes 3–11 relative to the intensity of the top shift band without competition. At 100× molar excess of competitor, the intensity of the top shift band drops to 0.12 fold for the wild-type competitor but to only 0.56 fold for the mutant competitor.

Figure 2.

Luciferase assay showing complete loss of FOXI1 transcriptional activation in mutant and modified promoter-reporter constructs. Top panel, Sequence of the FBS1-FBS2 promoter site in various promoter-reporter constructs (from top to bottom): wild type, c.−103T→C mutant, FBS1 deletion, FBS1 inversion (Flip FBS1), and FBS2 inversion (Flip FBS2). The −103C nucleotide is marked with lowercase letters. Bottom panel, COS-7 cells were transfected with 250 ng promoter-reporter construct and 750 ng empty vector (white bars) or FOXI1 construct (gray bars). Reporter activity is shown as “fold induction” compared with activity of the wild-type promoter-reporter construct without exogenous expression of FOXI1. Results from three independent experiments are shown (mean±SD).

Figure 3.

Conservation of human and mouse FOXI1 protein. The amino acid changes identified in patients with PS and nonsyndromic EVA are underlined and marked by arrows. The conserved forkhead DNA-binding domain is shaded in gray.

Figure  4.

The transcriptional activation ability of FOXI1 variants, as measured using a luciferase assay. Together with 750 ng of empty vector (EV), FOXI1 wild-type construct (WT), or one of the five FOXI1 mutant constructs, 250 ng of promoter-reporter construct was transfected into COS-7 cells. Reporter-gene activity was quantitated as “fold induction” relative to the empty vector control. Results from three independent experiments are shown (mean±SD); the activity of each mutant FOXI1 construct has been compared with that of the wild-type construct (P values calculated by two-tailed Student’s t test).

Figure 5.

Segregation of SLC26A4 and FOXI1 mutations in family 82230. Parents carry a single mutation in either SLC26A4 or FOXI1, the affected child carries both mutations, and her unaffected sister carries the SLC26A4 mutation only. Both siblings inherited the same wild-type SLC26A4 allele from their father.

Figure 6.

Temporal bone histology from Slc26a4^+/− embryos and Slc26a4^+/−; Foxi^+/− double-heterozygous embryos at the E16.5 stage. Red arrows mark selective enlargement of the endolymphatic duct in Slc26a4^+/−; Foxi^+/− double mutants, as compared with its normal morphology in Slc26a4^+/− mice. The endolymphatic duct is also normal in the Foxi1^+/− mice.

Figure 7.

Sequence alignment in multiple species shows the head-to-head FOXI1-binding motif in the promoter region of SLC26A4, SLC4A9 (GenBank accession number NC_000005.8), and ATP6V1B1 (GenBank accession number NC_000002.10) (UCSC Genome Browser). The two adjacent FOXI1-binding sites are shown in boxes; arrows indicate orientation. The binding sites experimentally confirmed to be required for gene transcriptional activation are in bold. Numbers indicate the upstream position of the nucleotides relative to the translational start site.