Functional assessment of allelic variants in the SLC26A4 gene involved in Pendred syndrome and nonsyndromic EVA

Abstract

Pendred syndrome is an autosomal recessive disorder characterized by sensorineural hearing loss, with malformations of the inner ear, ranging from enlarged vestibular aqueduct (EVA) to Mondini malformation, and deficient iodide organification in the thyroid gland. Nonsyndromic EVA (ns-EVA) is a separate type of sensorineural hearing loss showing normal thyroid function. Both Pendred syndrome and ns-EVA seem to be linked to the malfunction of pendrin (SLC26A4), a membrane transporter able to exchange anions between the cytosol and extracellular fluid. In the past, the pathogenicity of SLC26A4 missense mutations were assumed if the mutations fulfilled two criteria: low incidence of the mutation in the control population and substitution of evolutionary conserved amino acids. Here we show that these criteria are insufficient to make meaningful predictions about the effect of these SLC26A4 variants on the pendrin-induced ion transport. Furthermore, we functionally characterized 10 missense mutations within the SLC26A4 ORF, and consistently found that on the protein level, an addition or omission of a proline or a charged amino acid in the SLC26A4 sequence is detrimental to its function. These types of changes may be adequate for predicting SLC26A4 functionality in the absence of direct functional tests.

Pendred syndrome (PS) (OMIM#274600) (1) is an autosomal recessive disorder characterized by sensorineural hearing loss (SNHL) and malformations of the inner ear, ranging from enlarged vestibular aqueduct (EVA) (2) to Mondini malformation (3), combined with deficient iodide organification in the thyroid gland, as demonstrated by the positive perchlorate discharge test in affected individuals (4–7). Another form of SNHL associated with EVA, however, showing normal thyroid function, is called nonsyndromic EVA (ns-EVA) (OMIM#600791). The clinical features of PS are the consequence of impaired pendrin function (1), a protein encoded by the SLC26A4 gene (NM_000441). It is a member of the multifunctional anion transporter family SLC26, which mediates the exchange of anions including Cl⁻, HCO₃⁻, OH⁻, I⁻, or formate (8). Pendrin seems to be responsible for the efflux of iodide in thyrocytes (9–11), and for mediating Cl⁻/HCO₃⁻ exchange in the kidney cortex (12) and inner ear. In the latter, pendrin is involved in the conditioning of endolymphatic fluid, presumably because of HCO₃₋ secretion (13), thereby modifying inner ear acid-base homeostasis. A variable feature of PS is the development of goitre (apparent in only about 50% of the affected individuals). At the thyroid level, the role of pendrin is not conclusive. The transporter could act as an iodide transporter at the apical membrane of thyroid cells and impaired function could therefore lead to the iodide organification defect observed in PS patients (10, 11, 14–16). PS seems to be linked to bi-allelic mutations of the SLC26A4 genes. ns-EVA is genetically more heterogeneous relative to PS, and although some cases show mono- or bi-allelic SLC26A4 mutations, other cases are not linked to the SLC26A4 locus. Since its characterization, (1) more than 160 allelic variants of the SLC26A4 gene have been described (www.healthcare.uiowa.edu/labs/pendredandbor), all of which are located along the coding sequence, the splice sites, and in the noncoding exon 1 within the FOXI1 binding transcriptional regulatory elements (17). Approximately 62% of all mutations are missense changes, most of which have been described in only one family.

Previously, we identified 26 SLC26A4 allelic variants, including 18 missense changes in a cohort of hearing-impaired Spanish families (18). In this study, we screened the SLC26A4 coding sequence in a cohort of normal hearing controls. Using the fast fluorometric method for measuring Cl⁻/I⁻ transport via pendrin as described earlier (19, 20), we performed functional analyses on one mutant in the SLC26A4 protein found only in our control cohort (L597S), as well as on nine mutants found in our cohort of controls or patients suffering from PS or ns-EVA (18). To further examine the SLC26A4 structure/function relationship, we also tested two in silico designed mutations. The summarized experiments allowed us to rate the mutations in accordance to the respective ion transport. In the future, this approach, combined with an evaluation of SLC26A4 genetics (mono- or bi-allelic, cis- or trans-position), could permit a more reliable approach for distinguishing PS and ns-EVA-associated mutations from polymorphisms.

Results

Mutations of the SLC26A4 Gene Found in the Normal-Hearing Spanish Population.

In our cohort of 214 normal-hearing Spanish subjects, we identified a total of 17 allelic variants (Table 1). The eight mutations, which did not lead to an amino acid sequence change, were not further investigated (see Table 1, Group A). They include the two silent mutations, L75L and D710D, the five intronic SLC26A4 polymorphisms, and the mutation in the noncoding region of exon 21. The other nine missense mutations located within the SLC26A4 open-reading frame (ORF) (see Table 1, Group B), all led to an amino acid sequence change in SLC26A4. Three of them (F354S, L597S, and F667C) were only found in the normal-hearing cohort and a functional implication of these mutations seemed unlikely. However, the L597S mutation was included in the present study because conflicting assumptions have been made regarding its transport function. The six remaining mutations found in the normal-hearing cohort were also identified in patients suffering from PS or ns-EVA (Table 2, group A). However, two of them (V609G and G740S) were previously described in the National Center for Biotechnology Information database as SNPs. In addition, it was previously shown that R776C did not compromise SLC26A4 transport function (21). Therefore, the transport ability of E29Q, D724G, V88I, and L597S were tested in the present study.

Table 1.

SLC26A4 allelic variants detected in Spanish normal-hearing controls and in a cohort of hearing impaired patients

Group	Nucleotide change	aa change	Exon/ Intron	Controls (n = 428)	* Patients (n = 409)	References
A	225C > G	L75L	Ex 3	1	0	Present study
	2130C > T	D710D	Ex 19	1	1	Present study
	1–32G > A	—	In 1	1	0	Present study
	165–32A > G	—	In 2	1	0	Present study
	601–17C > T	—	In 5	3	0	Present study
	1708–18T > A	—	In 15	16	>3	Present study
	2320–22T > C	—	In 20	5	0	Present study
	2498A > T	—	Ex 21	1	0	Present study
B	85G > C	E29Q	Ex 2	1	1	(4), (5), (17), (22), (36), (37)
	262G > A	V88I	Ex 3	1	1 (V88I/R409H)	†(18)
	1061T > C	F354S	Ex 9	3	0	(4), (22)
	1790T > C	L597S	Ex 16	1	0	(4), (5), (17), †(22), (34), (35)
	1826T > G	V609G	Ex 17	2	2	(17), (22), †(35)
	2000T > G	F667C	Ex 17	1	0	(1)
	2171A > G	D724G	Ex 19	2	2	(36), (38)
	2218G > A	G740S	Ex 19	1	1	(36)
	2326C > T	R776C	Ex 21	1	1	†(21), †(24), (35)

The mutants which did not (Group A) or did (Group B) lead to a change in amino acid sequence are presented. All mutations were identified in different alleles at the heterozygous state, except for V609G and G740S, which were identified on a single allele. Albert et al. (22) referred to the L597S allelic variant as an SNP, whereas others assumed it would have a negative functional effect, as subjects carrying the respective mutation showed PS/ns-EVA (4, 5, 34, 35). Our functional analyses showed that ion transport carried out by L597S mutants is not significantly different from WT SLC26A4. Therefore, our results support the findings by Albert et al. n, number of alleles tested. The numbers indicate how often the mutant allele was identified in a group of unrelated Spanish patients, including 144 cases with SNHL, 12 with PS, 35 with ns-EVA, and 20 cases with SNHL segregating with the DFNB4 locus (18).

*Data taken from ref. 18. References denote previous works that consider the variant as pathogenic.

^†Reports that questioned whether the variant is pathogenic.

Table 2.

Comparison of the genotype and phenotype of the patients with hearing loss mentioned in Table 1, and of the patients included in the study whose mutations were not found in the control population

Group	Patient	Genotype		Sex	Age	Phenotype				Origin
		Allele 1	Allele 2			Deafness	CT	Goitre	PT
A	S226.1	p.E29Q†	[p.V88I; p.R409H]†	F	44	R.E. moderate/L.E. mild	EVA	No	ND	Castilla
	S154.18	p.T410M	p.D724G*	M	14	R.E. Profound/L.E severe	Mondini	No (ET)	ND	ND
	E205.3	p.D724G*	—	F	2	Severe, bilateral	Normal	No	ND	Cataluña
	E206.3	[p.V609G; p.D710D;p.G740S]	—	F	5	Profound, bilateral	Normal	No	ND	Cataluña
	E277.3	p.V609G	—	F	5	Moderate, bilateral	Normal	No	ND	Cataluña
	E502.3	p.G740S	—	F	7	Moderate, bilateral	Normal	No	ND	Cataluña
	S877.1	p.R776C	—	M	14	Profound, bilateral	Normal	No	ND	Castilla/Andalucia
B	S9.3	p.P140H*	p.G497S	M	6	Profound, bilateral	Mondini	No	ND	ND
	S9.4	p.P140H*	p.G497S	F	4	Profound, bilateral	Mondini	No	ND	ND
	E343.1	p.Q413P*	p.L445W	M	7	Moderate, bilateral, progressive	Mondini	No (ET)	P	Cataluña
	S256.4	p.G424D†	p.T416P	F	22	Profound, bilateral	Mondini	Yes (HT)	ND	Aragón
	S256.5	p.G424D†	p.T416P	M	24	Profound, bilateral	Mondini	Yes (HT)	ND	Aragón
	S129.3	p.T485R†	p.G102DfsX4	F	22	Profound, bilateral	EVA	Yes (ET)	P	Pais Vasco
	S129.4	p.T485R†	p.G102DfsX4	M	17	R.E. mild/L.E. moderate	EVA	No (ET)	ND	Pais Vasco
	S145.3	p.C400FfsX67	p.Q514K*	M	21	R.E. moderate/L.E. severe	EVA	Yes (HT)	ND	Castilla
	S145.5	p.C400FfsX67	p.Q514K*	M	18	Profound, bilateral	EVA	Yes (HT)	P	Castilla
	S67.3	p.G209V	p.Q514K*	M	8	Severe, bilateral	EVA	No	ND	ND
	S67.4	p.G209V	p.Q514K*	M	4	Severe, bilateral	EVA	No	ND	ND
	E224.3	p.Q514K*	p.Q514K*	M	2	Profound, bilateral	EVA	No (ET)	ND	Cataluña
	S461.3	p.Q514K*	p.V138F	M	22	Profound, bilateral	EVA	No	ND	ND
	S461.4	p.Q514K*	p.V138F	M	19	Profound, bilateral	EVA	No	ND	ND
	S1206.1	p.Q514K*	—	M	1	Profound, bilateral	Normal	No	ND	Cataluña

Group A: Patients bearing mutations described also in controls. Group B: Patients bearing mutations described only in hearing impaired individuals. Abbreviations: CT; computed tomography; ET, euthyroid; F, female; HT, hypothyroid; L.E., left ear; M, male; ND, not determined; PT, perchlorate test; R.E., right ear. An iodide discharge higher than 10% is indicated as P. *, †, denote the transport ability of the mutated protein:

* no activity,

^† reduced. Age in years at the time of clinical examination.

Functional Assay of SLC26A4 Mutations Found in the Spanish Population.

For the functional assays, we first selected the four mutations mentioned above (L597S, E29Q, D724G, and V88I). The hearing impaired patient carrying V88I showed a combined mutation in cis (V88I/R409H); therefore, this double mutation was tested in addition to the V88I. Finally, we selected five new missense mutations that have only been described in Spanish patients suffering from PS/ns-EVA (P140H, Q413P, G424D, T485R and Q514K) (see Table 2, Group B) (18).

To assay the transport function of the SLC26A4 allelic variants, we used a recently described fast fluorometric method for measuring chloride/iodide (Cl⁻/I⁻) transport (19, 20). With this technique, SLC26A4-mediated anion flux is measured by monitoring I⁻- and Cl⁻-induced quenching of an enhanced-yellow fluorescent protein (EYFP) fluorescence signal (19). In the first set of experiments, the change in fluorescence in cells transfected with both EYFP and pendrin or EYFP alone (the empty vector) were measured after a Cl⁻ to I⁻ (or vice versa) substitution. As seen in Fig. 1, the expression of WT pendrin lead to a marked decrease (Cl⁻ to I⁻ substitution) (see Fig. 1) or increase (I⁻ to Cl⁻ substitution) [supporting information (SI) Fig. S1] in EYFP fluorescence intensity. These data are indicative of a facilitated iodide transport (a more detailed description of the technique is given in Methods and SI Materials and Methods). The transport values (ΔFmax%) of WT pendrin described in this study are similar to those described earlier (20).

Fig. 1.

Change of the fluorescence signal (ΔFmax%) after the Cl⁻ to I⁻ switch (for details see Methods) in HEK 293 Phoenix-cells expressing solely the EYFP protein (empty), the WT pendrin and EYFP (wt), or the different mutants together with the EYFP protein. The numbers of the experiments are given (n). (A) Summary of the experiments using the mutants, which are not reduced if compared to WT (gray), and whose activity is reduced (cross lines). (B) Summary of the experiments using the mutants whose activity is annihilated (crossed lines) compared to WT [the values from cells expressing only EYFP (empty), and WT pendrin (wt) are taken from (A)]. The numerical data to all histograms shown in this study are given in Table S1. Statistical analysis: ***, P < 0.001 vs. WT; **, P < 0.01 vs. WT; *, P < 0.05 vs. WT; ###, P < 0.001 vs. empty; ##, P < 0.01 vs. empty; #, P < 0.05 vs. empty.

The fluorescence values of L597S and V88I were not significantly reduced as compared to WT pendrin (see Fig. 1A and Fig. S1a). Four of the mutations tested (E29Q, V88I/R409H, G424D, and T485R) showed a statistically significant reduction in fluorescence values as compared to WT pendrin (Cl⁻ to I⁻ substitution or vice versa), indicating that anion transport carried by these mutants is reduced (see Fig. 1a and Fig. S1a). No iodide transport activity could be observed with P140H, Q413P, Q514K and D724G (see Fig. 1b and Fig. S1b), as evidenced by the fact that the changes in fluorescence were not statistically different from those carried out using the empty vector.

Functional Assay of in-Silico Designed SLC26A4 Mutations.

It became evident from the experiments mentioned above that a loss-of-function SLC26A4 protein can be observed if amino acids with structural impact (charged amino acids or proline) are deleted from or added to the sequence. To confirm this hypothesis, we used two mutations found in the control cohort that showed no reduction in transport activity. Accordingly, V88I and L597S were transformed into V88P and L597P, respectively. As expected, the insertion of the proline rendered SLC26A4 inactive (Fig. 2).

Fig. 2.

Change of the fluorescence signal (ΔFmax%) after the respective anion-exchange in HEK 293 Phoenix-cells expressing EYFP and WT pendrin (wt), or the two different in silico designed mutants: (A) Cl⁻ to I⁻ and (B) I⁻ to Cl⁻. To test the “proline/fixed-charge” role of our hypothesis we mutated V88I and L597S (see Fig. 1) into V88P and L597P, respectively. The insertion of a proline leads to an annihilation of the SLC26A4-induced anion transport. Statistical analysis: ***, P < 0.001 vs. WT; **, P < 0.01 vs. WT; *, P < 0.05 vs. WT.

Discussion

The aim of the present study was, (i) to validate previous criteria used to predict functional implications of SLC26A4 mutations; (ii) to functionally test 10 missense mutations identified in the healthy control population or in patients suffering from PS/ns-EVA; and (iii) to test if the insertion of an amino acid with structural impact alters SLC26A4 function. It is our hope that the results of the experiments summarized will provide an avenue for predicting SLC26A4 activity, especially in cases where functional assays cannot be performed.

The two criteria used so far for assuming pathogenicity of mutations in the SLC26A4 gene are low incidence of the mutation in the control population and substitution of evolutionary conserved amino acids. However, we found that five of the missense changes in healthy individuals (E29Q, F354S, L597S, F667C, and D724G) affect amino acids located in highly conserved regions of the SLC26A4 orthologs (Fig. 3). None of them, except L597S (22), were detected in other normal-hearing Caucasian controls. Two of these mutations—partially active E29Q and completely inactive D724G—have negative functional implications on SLC26A4 transport, and occurred one and two times out of 214 control subjects, respectively (see Tables 1 and 2). The latter two mutants were also identified in PS/ns-EVA patients (see Tables 1 and 2). Therefore, these observations demonstrate that the location of a mutation in a conserved string of amino acids is not a reliable parameter for predicting a functional impairment of SLC26A4, and a high incidence of a mutation in the control population is not a reliable parameter for predicting a lack of functional impairment. This conclusion is further substantiated when estimating the prevalence for hereditary hearing impairments using the aforementioned criteria and the mutations found in the healthy individuals described in this study. Assuming that E29Q, F354S, L597S (conflicting data), F667C, and D724G cause deafness because they occur in a highly conserved string of amino acids, the carrier frequency of SLC26A4 mutated alleles should be 1 in 27 (or 30, excluding L597S). The derived prevalence for SLC26A4 recessive hearing loss should then be 1 in 2,900, which comprises around 68% of hereditary hearing impairments. This figure is much too high and inconsistent with the reported prevalence derived from previous clinical and genetics studies (23, 24). The reason for the apparent high prevalence we obtained is obviously the result of an ill-defined link between hearing impairment and allelic variants found in the SLC26A4 gene, again underlining the fact that a mutation found in a string of highly conserved amino acids is not a reliable parameter for predicting SLC26A4 functionality.

Fig. 3.

Protein sequences of 10 SLC26A4 orthologs, including: (1) Homo sapiens (NP_000432), (2) Rattus norvegicus (NP_062087), (3) Mus musculus (NP_035997), (4) Pan troglodytes (XP_519308), (5) Macaca mulatta (XP_001094049), (6) Canis familiaris (XP_540382), (7) Bos taurus (XP_608706), (8) Monodelphis domestica (XP_001363598), (9) Gallus gallus (XP_425419), and (10) Danio rerio (XP_692273). The mutations found in the Spanish control population, which lead to a changed amino acid sequence as summarized in Table 1, are given with the position-number and arrows. Some of these mutations (i.e., at positions 29, 88, 609, 724, 740, and 776) are also found in hearing-impaired patients. ‘–’ indicates sequence interruptions. The single letter code for the amino acids is given in SI Materials and Methods.

We next tested six novel mutations described recently (18) in Spanish PS/ns-EVA patients (V88I/R409H, P140H, Q413P, G424D, T485R, and Q514K). The double mutation (V88I/R409H) showed a reduced transport activity, whereas the single V88I mutation did not lead to a functional impairment (see Fig. 1A and S1a). These data indicate that ns-EVA identified in the patient carrying the double mutation V88I/R409H is most likely not related to the V88I mutation, but instead to the partially functional R409H mutation (25). The mutations G424D and T485R resulted in reduced function, whereas the mutations P140H, Q413P, and Q514K showed no transport activity (see Fig. 1, and Fig. S1). It is difficult at the moment to unambiguously link the severity of hearing loss (moderate, severe, and profound) (see Table 2) to individual SLC26A4 mutations. Mutations with no transport activity (P140H, Q413P, Q514K, and D724G) and with reduced activity (E29Q, V88I/R409H, G424D, and T485R) were equally identified in patients with PS and ns-EVA suffering from moderate-to-profound hearing loss. The simplest explanation for the difficulty to correlate a specific SLC26A4 mutation to the clinical symptoms of a patient is that additional genetic, epigenetic, and environmental factors could substantially modify the observed disease phenotype. This explanation is substantiated by the fact that SLC26A4 zero-mutants can show EVA and severe hearing loss, and patients with mono-allelic SLC26A4 mutations (see Table 2) can show severe-to-profound hearing loss without EVA. Despite the fact that it is difficult at the moment to correlate genotype (SLC26A4) and phenotype (moderate-to-profound hearing loss), it is noteworthy mentioning that patient E224.3 (see Table 2), who is homozygous for the Q514K mutation (loss-of-function), had bilateral, sensorineural, profound deafness at the age of 2; whereas patient S226.1, who carries two mutations with only reduced transport activity (E29Q and V88I/R409H), had sensorineuronal, moderate-to-mild hearing loss at the age of 44.

From data presented here, it is apparent that certain biochemical parameters could act as guide-posts for possible functional impairments. In all mutations tested, an impairment of transport function was found if an amino acid bearing a fixed charge [aspartic-acid: (D⁻); single letter code and the sign indicates the fixed charge) or glutamic-acid (E⁻), lysine (K⁺) or arginine (R⁺), and histidine (H⁺); the positive charge of histidine depends on the pH] was missing or introduced. Furthermore, the loss or inclusion of a proline (P; proline acts as a structural disruptor of regular secondary structures such as α-helices or β-sheets) in the SLC26A4 sequence was detrimental for transport function. These findings are consistent with the functional data published so far, with one exception. Pfarr et al. (21) described an allelic variant (R776C) that, on a functional level, is indistinguishable from controls. However, this particular arginine is located on the extreme C terminus of SLC26A4 (a 780 aa protein). The N terminus and C terminus of a protein are usually not structurally defined. Therefore, it is plausible to assume that mutations occurring in these areas probably have little or no functional implication. Scott et al. (26) described two inserted prolines (L236P and T416P, two very common mutations in PS) associated with a loss of function. Furthermore, the same authors described two separate mutations occuring in PS/ns-EVA patients, in which an amino acid with a fixed charge was added or omitted (E384G and V480D). Both mutations showed a loss of or reduced function, respectively. In yet another study performed by Taylor et al. (27) on PS/ns-EVA patients, three mutations of the same type (G672E, G102R, and Q446R) were described, and all showed a loss of functionality. We recently described a mutation, S28R, which again, reduced functionality (14, 19). Another mutation, R409H, in which a fixed charge is altered (the pK_a of histidine is 6.0, and therefore only a minute amount of histidine will be charged at physiological pH of 7.4), also shows a reduced functionality, and therefore follows the proposed role (25). Moreover, the transformation of two allelic variants (V88I and L597S) that were not functionally different compared to WT SLC26A4, into mutations fitting the proline/fixed charge role (i.e., V88P and L597P), resulted in the annihilation of transport activity. Although, it is important to note that mutations that do not enter into this proline/fixed charge role can be functionally detrimental or without functional implication (15, 20, 21, 26, 27). In these cases, only functional tests can unambiguously distinguish between different SLC26A4 gene mutations in SNPs and those mutations that actually cause a reduced function and ultimately disease symptoms.

In conclusion, we show that the two parameters used so far—(i) low incidence of the mutation in the control population and, (ii) substitution of evolutionary conserved amino acids by the mutation—are not reliable for predicting SLC26A4 transport function. Our functional test using mutations found in healthy and hearing-impaired individuals further reveals that the proline/fixed charge role—that is, the addition or omission of proline, or the addition or omission of charged amino acids in the sequence of SLC26A4—might be a better option for predicting SLC26A4 function in the cases where direct functional tests cannot be performed.

Methods

Subjects.

A cohort of 214 normal-hearing individuals of Spanish–Caucasian population, 74 females and 140 males, aged 23 to 75 years, was collected. Auditory capacity was established by evaluating the pure-tone air- and bone-conduction threshold audiometry. Subjects gave written informed consent for genetic testing.

Mutational Analysis of SLC26A4 Gene.

DNA was obtained from blood samples following standard procedures. Mutation screening of the SLC26A4 gene was performed by PCR amplification of the coding 2–21 exons and the flanking intron sequences as described in detail in SI Materials and Methods.

SLC26A4 Site-Directed Mutagenesis.

The site-directed mutagenesis of SLC26A4 was done as described in detail in (14), and in SI Materials and Methods.

Cell Culture and Transient Transfection.

HEK 293 Phoenix cells (this is a second generation retrovirus producing cell line for the generation of helper free ectopic and amphotropic retroviruses) (28) were grown as previously described (14, 19). Briefly, cells were grown in Minimum Essential Medium Eagle (Sigma) supplemented with 10% FBS (Cambrex BioScience), 2-mM L-glutamine, 100-units/ml penicillin, 100-μg/ml streptomycin, and 1-mM pyruvic acid (sodium salt). The cells were maintained at 37 °C in a 5% CO₂, 95% air-humidified incubator; subcultures were routinely established every second to third day seeding cells in Petri dishes (Ø 10 cm) after trypsin/EDTA treatment. For in vivo functional fluorometric assays, HEK 293 Phoenix cells were transiently transfected with the plasmid expressing WT SLC26A4, the mutant or the empty plasmid (negative control) by calcium-phosphate precipitation. One day before transfection, cells were seeded on six multiwells (Ø 35 mm) and grown to 60% to 80% confluency. For each well, 3 μg of plasmid dissolved in 61.5-μl H₂O were mixed with 7.5 μl of buffer A (CaCl₂ 2.5 M), and 75 μl of buffer B (140-mM NaCl, 1.5-mM Na₂HPO₄, 50-mM Hepes, pH 7 adjusted with NaOH) and incubated 10 to 15 min at room temperature. The transfection mix was spread over the cells, adding it to the complete growing media; after incubation of 8 to 10 h, the cells were washed twice with fresh medium-containing serum. For the fluorescence measurements, 24 h after transfection cells were split on round, poly-L-lysine-coated coverslips (Ø 40 mm), kept in medium containing serum, and fluorescence measurements were performed 48 to 56 h after transfection, in the absence of serum, using the saline solutions described below.

Fluorometric Analyses.

The fluorometric method was described previously in detail (19, 20). In brief, the SLC26A4 protein is able to transport chloride (19, 29, 30) as well as iodide (11, 15, 27, 31, 32), and therefore, because the EYFP protein is a fluorescent dye sensitive to intracellular halides, this dye is particularly suitable for sensing changes in the intracellular chloride as well as iodide amount, and therefore to measure the respective ion fluxes across the cell membrane. By changing the extracellular chloride or iodide concentration in cells over-expressing the SLC26A4 protein and comparing the result with the result obtained in cells not over-expressing SLC26A4, the transport contribution of SLC26A4 can be easily determined. To evaluate the pendrin-induced halide transport, HEK 293 Phoenix cells were transfected with pIRES2-EYFP plasmid bearing the respective SLC26A4 or mutant SLC26A4 cDNA, or with the empty plasmid. The cells were continuously perfused in a laminar-flow chamber (FCS2 System, Bioptechs Inc.) with “isotonic high Cl⁻” (2-mM KCl, 135-mM NaCl, 1-mM MgCl₂, 1-mM CaCl₂, 10-mM D-glucose, 20-mM Hepes, pH 7.4, 308 mOsm with mannitol) or “isotonic high I⁻” (2-mM KCl, 135-mM NaI, 1-mM MgCl₂, 1-mM CaCl₂, 10-mM D-glucose, 20-mM Hepes, pH 7.4, 308 mOsm with mannitol) solutions. By changing the isotonic high Cl⁻ to isotonic high I⁻ solution, the iodide influx can be measured, whereas by switching back from isotonic high I⁻ to isotonic high Cl⁻ solution, the iodide efflux can be estimated. We define a reduced transport activity if the in- or efflux capacity of a mutant is reduced if compared to WT SLC26A4. The EYFP fluorescence measurements were performed using a Leica TCS SP2 AOBS confocal microscope (Leica Microsystem), using a 515-nm Ar/ArKr laser line for exciting the EYFP fluorescence. Measurements were started after steady-state conditions were reached. Maximal fluorescence variations (ΔF_max%) represent the maximal observed percentage difference with respect to the fluorescence intensity at the moment of the solution substitution (iodide/chloride). Because iodide is able to quench the EYFP signal more efficiently than chloride (33), the slight increase of fluorescence observed in control cells not over-expressing SLC26A4 (see Fig. 2, empty), is pointing to the fact that the chloride exit prevails the iodide influx. Therefore, the transporters or channels able to pass iodide from the extracellular space toward the cytosol are scarcely active in the native HEK 293 Phoenix cell line we use. This cell line is therefore particularly suited to test SLC26A4-induced halide transport activity after the over-expression of the transporter or its mutants. After the expression of WT SLC26A4, however, the exchange of extracellular chloride with iodide leads to a marked reduction of the EYFP fluorescence because of the SLC26A4-driven iodide influx (see Figs. 1 and 2, WT).

Statistical analysis was done by ANOVA–Bonferroni's multiple comparison test, and statistical significant differences were assumed at P-values ≤0.05.