Asymmetric pendrin homodimer reveals its molecular mechanism as anion exchanger

Hearing loss affects over 5% of the world’s population, with 3–5% attributable to SLC26A4 mutations, which cause Pendred syndrome characterized by sensorineural hearing loss, goiter, and reduced blood pressure. Pendrin (SLC26A4) is a sodium-independent, electroneutral anion exchanger that mediates Cl−/HCO3− and Cl−/I− exchange in the apical membrane of epithelial cells in the inner ear, kidney, and thyroid. Genetic studies and mouse models implicate pendrin in regulating endolymph composition and pH in the inner ear, bicarbonate secretion and chloride reabsorption in renal intercalated cells, and iodide handling in the thyroid. Despite its physiological importance, the molecular structure and detailed mechanism of pendrin-mediated exchange have remained unknown. This study aims to determine high-resolution structures of mouse pendrin under different anion conditions and to elucidate the structural basis and mechanism of its electroneutral anion exchange, while mapping disease-associated variants onto the structure.

Prior work established pendrin as a sodium-independent anion exchanger for Cl−/HCO3− and Cl−/I− in epithelia and linked loss of function to Pendred syndrome and enlarged vestibular aqueduct in mice. Pendrin’s cellular localization and hormonal regulation (angiotensin II, aldosterone) in kidney intercalated cells have been described. Structural studies of other SLC26 family members (prestin/SLC26A5 and SLC26A9) revealed inward-open and intermediate states and a conserved UraA-like fold with core and gate regions, but pendrin’s structure was unknown. Disease-associated missense variants in SLC26A4 are abundant (>800 missense in databases), particularly near the anion-binding pocket, yet lacked structural rationalization. The study builds on these findings by providing pendrin cryo-EM structures and comparing conserved/anomalous features with prestin and SLC26A9 to infer determinants of anion selectivity and exchange.

• Constructs and expression: Full-length mouse pendrin (UniProt Q9R155-1) with an N-terminal 3xFLAG tag was cloned into a PEZTBacMam vector. HEK293E cells were infected with BacMam virus; expression was induced with sodium butyrate.
• Purification: Cells were solubilized in digitonin; pendrin was purified by anti-FLAG affinity chromatography, eluted with FLAG peptide, and further purified by SEC in GDN detergent. Distinct buffer conditions introduced specific anions: Cl− (150 mM NaCl), HCO3− exchange (e.g., 50 mM NaCl + 150 mM NaHCO3), or I− mixtures, and Cl−-free buffers using HEPES/Na2SO4 for HCO3− conditions.
• Cryo-EM sample preparation and data collection: Purified protein (~1.7 mg/mL) was vitrified on glow-discharged grids. Data were collected on a Titan Krios (300 kV) with K2 or K3 cameras, super-resolution, pixel size ~1.046–1.064 Å, defocus −1.2 to −2.2 μm, and total dose ~53–58 e−/Å2.
• Image processing: Motion correction (MotionCor2), CTF estimation (Gctf), particle picking and 2D classification (cryoSPARC), 3D classification/refinement (RELION), C1 or C2 symmetry as appropriate, CTF refinement and polishing. 3D variability analysis probed conformational heterogeneity. Maps were post-processed with DeepEMhancer and visualized in UCSF Chimera.
• Model building and refinement: Initial models were derived from AlphaFold predictions and manually adjusted in Coot; real-space refinement was performed with PHENIX. Inward-open C2 structures were refined to 3.3–3.5 Å; outward/asymmetric states to ~3.6–3.8 Å (gold-standard FSC 0.143).
• Functional assays: HEK293T cells expressing fluorescently tagged pendrin were used for anion exchange assays. Cl−/HCO3− exchange was monitored by intracellular pH-sensitive dye BCECF in CO2/HCO3− buffers. Cl−/I− exchange was measured via iodide-sensitive EYFP quenching with alternating Cl− and I− buffers. Site-directed mutants (e.g., Y105F, Y105A, P142A) were tested. Data were quantified in ImageJ and GraphPad Prism.

• Cryo-EM structures: Mouse pendrin forms a domain-swapped homodimer with NTD and STAS domains interchanged between protomers and a TMD adopting the UraA fold (core: TM1–4,8–11; gate: TM5–7,12–14). Symmetric inward-open dimers were obtained with Cl− (3.3 Å) and HCO3− (3.5 Å).
• Anion-binding site: A clear Cl− density occupies a pocket at the apex of the inward-open intracellular vestibule between short helices TM3 and TM10. Coordination involves TM3/TM10 dipoles with direct interactions from S408 and Y105; Q101 and Y105 help form a partially positive pocket; R409 stabilizes via interaction with Q101; hydrophobic A406, L407, P140, F141, P142 shape the pocket; N457 (gate) hydrogen bonds with Q101 (core) to stabilize core–gate packing. With HCO3−, density occupies the same pocket ~2 Å offset from Cl−, interacting with backbone nitrogens of S408 and L407 while Y105 is too distant to interact directly.
• Mixed-anion conditions reveal multiple states: In Cl−/HCO3− samples, three conformations were observed: ~15% symmetric inward-open, ~15% symmetric outward-open, and ~70% asymmetric dimers with one inward-open and one outward-open protomer. In the outward-open protomer, two anion densities were detected within the extracellular cavity above the binding pocket, ~4.2 Å apart, consistent with alternative binding modes during translocation; K237 and Q230 can interact with the upper and lower densities, respectively. Cl−/I− and HCO3−/I− pairs yielded predominantly symmetric inward-open (~75%) and asymmetric (~25%) states.
• Conformational change: Comparing inward vs outward protomers (gate-aligned), the core rotates ~15° and translates ~9 Å toward the extracellular side, elevating the binding pocket to release anion outward. The outward-open state shows an increased outer-leaflet cross-sectional area relative to the inward-open state.
• STAS domain and dimerization: STAS domains mediate dimerization and contact the partner TMD. Conserved interactions include D724 (STAS Cα4)–R24 (NTD Nβ1). Inter-protomer H-bonds (e.g., S552–S666) contribute to dimer stability. A hypothesized anion pre-binding site in STAS (between loops Cβ3–Cα1 and Cβ4–Cα2) interfaces with TM12–14.
• Comparative analysis: Pendrin’s TMD is similar to prestin and SLC26A9 in the inward state (RMSD < 1.3 Å), but pendrin displays the largest core rotation in the outward-open state among SLC26 structures; its outward binding pocket is positioned most apically, approaching AE1’s outward-open state.
• Functional validation and selectivity determinants: Mutations in key pocket residues modulate exchange. Y105F retains Cl−/I− exchange with reduced transport and shows significantly reduced Cl−/HCO3− exchange; Y105A further impairs I− transport and abolishes Cl−/HCO3− exchange. P142A maintains Cl−/I− exchange but loses Cl−/HCO3− exchange. These data support the role of pocket surface charge and hydrophobic triad (P140–F141–P142) in anion selectivity.
• Disease variant mapping: 761 missense variants mapped onto the resolved structure highlight hotspots around CL1–TM3 and EL5–TM10 near the binding pocket and along the anion pathway; many substitutions alter local charge or sterics in the core–gate interface or hydrophobic periphery, potentially disturbing alternate-access motions or membrane stability. Mutations in STAS dimer interfaces and the positively charged platform (Cal, Calb, Cx2) are enriched and likely disrupt dimerization/signaling.

The structures reveal that pendrin operates via an inverted alternate-access mechanism, wherein asymmetric homodimers permit simultaneous uptake and secretion: one protomer is inward-open while the partner is outward-open, enabling electroneutral exchange. Mixed-anion conditions favor this asymmetric state, suggesting that competitive binding by exchange partners (e.g., Cl− and HCO3−) drives allosteric transitions of the core relative to the gate. The ~15° rotation and ~9 Å elevator-like translation of the core move the anion binding site from the intracellular vestibule to the extracellular cavity for release. The outward-open cavity’s positive electrostatic potential and residues (K237, Q230) promote extracellular anion recruitment. STAS domains form the principal dimerization interface and are positioned beneath the TMD pathway; subtle rotation of helix Calb between states and the conserved positively charged STAS platform suggest a modulatory role in anion recruitment and/or coupling to intracellular partners. Comparative analysis with prestin and SLC26A9 identifies pocket-residue differences (e.g., Y105, P140–F141–P142, R409) that tune pocket charge and shape, rationalizing divergent anion selectivity and transport modes across the family. Mapping of clinical variants to structural elements explains how mutations near the pocket, core–gate interface, and STAS dimerization surfaces can impair anion binding, conformational cycling, or dimer stability, thus providing a structural framework for interpreting Pendred syndrome genetics.

This work determines high-resolution cryo-EM structures of mouse pendrin in symmetric inward-open, symmetric outward-open, and asymmetric inward–outward homodimer states. The data define a conserved anion-binding pocket between TM3 and TM10, show that Cl− and HCO3− bind the same site with slightly offset positions, and demonstrate that mixed-anion conditions stabilize an asymmetric dimer consistent with an inverted alternate-access elevator mechanism for electroneutral exchange. Structural comparisons across SLC26 members and functional assays of key pocket residues explain determinants of anion selectivity. Mapping hundreds of disease-associated variants onto the structure provides mechanistic insight into pathogenicity. Future work should capture additional intermediate states, resolve flexible regions (e.g., IVS and CTD), investigate lipid/nanodisc environments and membrane mechanics, and delineate how STAS-mediated protein–protein interactions regulate transport, thereby informing therapeutic strategies for Pendred syndrome.

• Structural models lack densities for flexible regions (NTD residues 1–17, the IVS 596–650, and CTD 738–780), limiting insight into their roles in regulation.
• Cryo-EM structures represent static snapshots from detergent-purified protein; native membrane context and potential lipid effects on conformational equilibria were not directly assessed.
• Asymmetric and outward-open states were enriched under mixed-anion conditions; their relative populations in vivo remain to be quantified.
• Functional validation focused on a subset of pocket residues; comprehensive mutational and kinetic analyses were not performed for all mapped variants.
• The proposed inverted alternate-access mechanism is inferred from structural comparisons and limited functional assays rather than direct real-time transport measurements of conformational changes.