Report

Cryo-EM structures of the human cation-chloride cotransporter KCC1

See allHide authors and affiliations

Science  25 Oct 2019:
Vol. 366, Issue 6464, pp. 505-508
DOI: 10.1126/science.aay3129

Coupled transport

Cation-chloride cotransporters move chloride and cations across the cell membrane and are important in regulating cell volume and setting the chloride concentration inside the cell. Mutations lead to serious diseases, such as epilepsy. Liu et al. present the structure of the human potassium-chloride cotransporter KCC1, as determined by cryo–electron microscopy. Based on the structure, functional studies, and molecular dynamics simulations, they propose an ion transport model. The structure provides a framework for interpreting disease-related mutations in potassium-chloride cotransporters.

Science, this issue p. 505

Abstract

Cation-chloride cotransporters (CCCs) mediate the coupled, electroneutral symport of cations with chloride across the plasma membrane and are vital for cell volume regulation, salt reabsorption in the kidney, and γ-aminobutyric acid (GABA)–mediated modulation in neurons. Here we present cryo–electron microscopy (cryo-EM) structures of human potassium-chloride cotransporter KCC1 in potassium chloride or sodium chloride at 2.9- to 3.5-angstrom resolution. KCC1 exists as a dimer, with both extracellular and transmembrane domains involved in dimerization. The structural and functional analyses, along with computational studies, reveal one potassium site and two chloride sites in KCC1, which are all required for the ion transport activity. KCC1 adopts an inward-facing conformation, with the extracellular gate occluded. The KCC1 structures allow us to model a potential ion transport mechanism in KCCs and provide a blueprint for drug design.

The human solute carrier 12 (SLC12) gene family encodes the cation-chloride cotransporters (CCCs) that mediate the coupled, electroneutral symport of Na+ and/or K+ with Cl across the plasma membrane (1, 2). Defined by their transport properties and amino acid sequences, CCCs can be divided into several branches, including two Na-K-2Cl cotransporters (NKCC1 and NKCC2), one Na-Cl cotransporter (NCC), and four K-Cl cotransporters (KCC1 to KCC4) (fig. S1). CCCs are expressed in many tissues and organs, especially in kidney and the nervous system, and play important roles in a variety of physiological processes. NKCCs and KCCs participate in cell volume regulation by transporting ions across the plasma membrane to reduce the osmotic difference when cells are exposed to hypertonic or hypotonic environments (3). In kidney, NCC and NKCC2 are major carriers for the reabsorption of NaCl in the distal convoluted tubule and the thick ascending loop of Henle, respectively (4). In central neurons, NKCC1 and KCC2 collaboratively set the intracellular Cl concentration and thereby regulate γ-aminobutyric acid (GABA)– and glycine-mediated synaptic inhibition (5).

Considering their important functions in ion transport, it is not surprising that mutations in CCCs lead to a wide spectrum of human diseases (610). For example, five single amino acid substitutions in KCC2 can cause human epilepsy. R952H (Arg952→His) and R1049C (Arg1049→Cys) mutations give rise to idiopathic generalized epilepsy (9), whereas L311H (Leu311→His), L426P (Leu426→Pro), and G551D (Gly551→Asp) substitutions induce epilepsy of infancy with migrating focal seizures (10). Therefore, CCCs are important drug targets. The diuretics furosemide and thiazide are used as antihypertensive drugs by inhibition of NKCC2 and NCC, respectively (11). Recently, efforts have been made to develop KCC2-specific agonists and NKCC1-specific antagonists to treat epilepsy diseases (12, 13). Drugs targeting KCC2 or NKCC1 may also be effective in treating other diseases by regulating GABAergic inhibition. For example, a KCC2 agonist, CLP290, promoted functional recovery after spinal cord injury (14).

Although the physiology, biochemistry, and pharmacology of different CCC members have been extensively studied, our knowledge of their ion cotransport mechanism is hindered by the lack of high-resolution structures. Structures are available for the Methanosarcina acetivorans CCC C-terminal domain (15) and Danio rerio NKCC1 (DrNKCC1) (16). Although the DrNKCC1 structure reveals the architecture and ion binding sites of the sodium-driven CCCs, the structures and ion cotransport mechanisms of the potassium-driven KCCs remain unknown. Here, we present structural studies of human KCC1.

We determined three cryo–electron microscopy (cryo-EM) structures of human KCC1: one in 150 mM KCl and detergent glyco-diosgenin (GDN) at 2.9-Å resolution, one reconstituted into lipid nanodisc in 150 mM KCl at 2.9-Å resolution, and one in 150 mM NaCl and GDN at 3.5-Å resolution (figs. S2 to S7 and table S1). Because the overall architectures of the three structures are very similar (fig. S8), we mainly focus on the structure in KCl and GDN in this report. The cryo-EM map in the transmembrane domain (TMD) and extracellular domain (ECD) is of high quality, enabling us to build an accurate model containing residues 116 to 652 (Fig. 1, A and B, and fig. S3). Owing to the flexibility of the linker that connects the TMD and C-terminal domain (CTD), the CTD is resolved at a lower resolution and is not modeled.

Fig. 1 Overall structure of human KCC1 determined in KCl and GDN.

(A) Side view of a three-dimensional reconstruction of KCC1 with each subunit individually colored. The gray bars on either side of the structure define the position (top and bottom) of the cell membrane. (B) Cartoon representation of the KCC1 dimer in the same orientation as the electron microscopy maps in (A). N-Acetylglucosamine (NAG) and putative GDN molecules are shown as green and cyan sticks, respectively. The three ions are shown as purple (K+) and orange (Cl) spheres. N and N′ indicate the N termini; C and C′ indicate the C termini. (C) Topology and domain arrangement of the KCC1 subunit. The three ions are shown as purple (K+) and orange (Cl) circles. (D) Structure of one KCC1 subunit in intracellular (left) and side (right) views. Each domain is colored the same as in (C). Side chains of two pairs of Cys residues that form disulfide bonds are shown as yellow sticks. Numbers indicate transmembrane helices. TM1, TM2, TM6, and TM7 form the core domain; TM3 to TM5 and TM8 to TM10 form the scaffold domain.

The TMD of KCC1 contains 12 transmembrane helices (TMs), with both the amino and carboxyl termini on the intracellular side (Fig. 1, C and D). TM1 to TM5 and TM6 to TM10 are inverted structural repeats related by a pseudo–twofold symmetry axis parallel to the membrane plane, with a root-mean-square deviation of 4.6 Å over 136 Cα atoms (fig. S9). Whereas TM1, TM2, TM6, and TM7 form the “core domain,” TM3 to TM5 and TM8 to TM10 form the “scaffold domain,” providing ion transport cavities between the core domain and the scaffold domain. TM11 and TM12 wrap around TM10 at one side of the transporter (Fig. 1D). The KCC1 TMD adopts a LeuT-like fold, which is shared among a superfamily of secondary transport proteins termed amino acid, polyamine, and organocation (APC) transporters (17).

The ECD, formed by a 120-residue-long linker between TM5 and TM6, is a characteristic and conserved feature of KCCs, whereas NKCCs and NCC have much shorter linkers (fig. S1). The ECD consists of two pairs of anti-parallel β strands (β1-β4 and β2-β3) and four short α helices (EL2 to EL5). The β2-β3 hairpin extends like an antenna from the ECD (Fig. 1D). The relatively rigid architecture of the ECD is stabilized by two disulfide bonds, C308–C323 and C343–C353 (Fig. 1D). Mutations of each of the four highly conserved cysteine residues abolish KCC2 transport activity, suggesting that the ECD folded structure is important for function (18). Two asparagine residues, Asn312 and Asn361, are glycosylated with visible density for the covalently linked N-acetylglucosamine moiety of the sugar (Fig. 1B and fig. S3B).

The cryo-EM structure of KCC1 reveals that both the TMD and ECD participate in dimeric assembly. At the dimer interface of the ECD, there are a pair of hydrogen bonds between the carbonyl oxygen of Ser403 and the amine group of Lys405 from neighboring subunits (fig. S10A). In addition, hydrophobic interactions between the side chains of Pro355, Pro402, and Leu404 from both ECDs strengthen this dimerization.

In the TMD of KCC1, the dimeric interaction involves TM12 and TM9 from the adjacent molecules in a dimer. On the periphery, hydrophobic amino acids from TM12 of one KCC1 molecule interdigitate with nonpolar residues from TM9 of the adjacent molecule, sealing the dimer interface at the edges but leaving an interior hydrophobic cavity between the monomers (fig. S10B). Within the cavity, two tube-like cryo-EM densities with the shape and size of two GDN molecules were observed (fig. S3C). The tail of the putative GDN inserts in the hydrophobic cavity and mediates hydrophobic interactions (fig. S10B). In comparison with other APC transporters such as LeuT (17) and the Vibrio parahaemolyticus sodium-galactose symporter (vSGLT) (19), KCC1 has relatively loose dimeric packing in the TMD (fig. S11).

Multiple biochemical studies have shown that CCCs form dimers in vivo (2024). Consistently, three lines of evidence support that this dimeric assembly of KCC1 involving the ECD and the TMD is stable and therefore probably physiologically relevant. First, this dimeric state is detergent-independent, because KCC1 has similar retention volumes in gel filtration in various detergents such as n-dodecyl-β-d-maltopyranoside (DDM), n-decyl-β-d-maltopyranoside (DM), GDN, and lauryl maltose neopentyl glycol (LMNG) (fig. S12A). Second, single or multiple mutations at the dimer interface in the TMD were unable to disrupt the dimeric assembly, confirming the strong dimeric interactions (fig. S12, B and C). Third, the cryo-EM structure of KCC1 reconstituted in lipid nanodisc has a very similar overall architecture to that in GDN (figs. S8 and S12D).

In the TMD core domain, TM1 and TM6 are disrupted in the middle by short Gly-containing loops, providing potential ion binding sites approximately halfway across the membrane. The two 2.9-Å resolution structures of KCC1 in KCl reveal three ion binding sites, designated as SK, SCl1, and SCl2 (Fig. 2, A and B, and fig. S13). SK is surrounded by four main-chain carbonyls from both Gly-containing loops (Asn131, Ile132, Thr432, and Pro429) and two hydroxyl groups of Tyr216 and Thr432, with coordination distances of around 2.9 to 3.8 Å (Fig. 2B). Because such a local chemical environment clearly favors a cation, we modeled a K+ at this site, the only cation used in the buffer. The SCl1 site is 3.8 Å above SK; the ion at this site is mainly coordinated by three main-chain amide groups (Gly134, Val135, and Ile136) and the K+ at the SK site. The hydroxyl group of Ser430 is about 4 Å from the ion at the SCl1 site, potentially allowing weak coordination. The ion at the SCl2 site is also coordinated mainly by three main-chain amide groups (Gly433, Ile434, and Met435) as well as the hydroxyl group of Tyr589 from TM10 (Fig. 2B). Because the chemical environments of SCl1 and SCl2 are suitable for anions, we modeled one Cl at each site. A similar Cl binding site has been observed in the structure of the prokaryotic Cl/H+ transporter StClC (fig. S14) (25).

Fig. 2 The ion binding sites in KCC1.

(A) The potassium and chloride ions bind at the breakage points of TM1 and TM6 of KCC1 in KCl and GDN. Two tyrosine residues involved in the ion coordination are shown as sticks. (B and C) Magnified views of the potassium and chloride binding sites in the KCC1 structure in KCl (B) or NaCl (C) from the same orientation. In both (B) and (C), all electron densities are shown at the same contour level of 4.5σ. Numbers show distances in angstroms between the ions and coordinating atoms. (D) K+ influx in Xenopus laevis oocytes injected with wild-type (WT) and mutant mouse KCC3 complementary RNA and measured under isosmotic (basal) and hypoosmotic (stimulated) conditions. The Western blot shows that all mutants are expressed at similar levels as the wild type. The table shows equivalent residues in human KCC1 and mouse KCC3. For more sequence alignments, see fig. S1. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

To test the functions of the ion binding sites, we measured the ion transport activity using a standard K+ (86Rb) influx assay in Xenopus laevis oocytes. Although human KCC1 has low activity (fig. S15A), mutations of the K+ coordination residue Tyr216 or the Cl coordination residue Ser430 at the SCl1 site did reduce KCC1 activity (fig. S15B). To further confirm the mutants’ activity, we studied mouse KCC3, which is a close relative to human KCC1 (78% sequence identity) but with much higher K-Cl cotransport activity (fig. S15C). Mutations of either of the K+ coordination residues Tyr283 (Tyr216 in KCC1) or Thr497 (Thr432 in KCC1), or the Cl coordination residue Tyr654 (Tyr589 in KCC1) at the SCl2 site, abolished the ion transport activities of KCC3 (Fig. 2D). Mutation of the weak Cl coordination residue Ser495 (Ser430 in KCC1) at the SC11 site also substantially reduced KCC3 activity. This suggests that all three sites are required for K-Cl cotransport activity.

KCCs cotransport K-Cl at a 1:1 ratio (26). This raises the questions of why there are two Cl sites and which one is the transport site. Three structural features suggest that SCl1 is likely a transport site. First, SCl1 sits deeply in the cytosolic-facing cavity, whereas SCl2 is more exposed to cytosolic solvent. The SCl1 and SK sites align well with substrates from other APC transporters such as LeuT and GkApcT, a prokaryotic homolog of the SLC7 transporter (fig. S16). Second, the 3.8-Å distance between SCl1 and SK allows the two ions to coordinate each other and thus to achieve stoichiometric cotransport; by contrast, SCl2 is 7.8 Å below SK, too far to allow direct interaction with the K+. Third, Cl binding at SCl2 is independent of K+ binding at SK, whereas Cl binding at SCl1 is coupled to K+ binding at SK, as revealed by the KCC1 structure determined in the presence of NaCl. In the KCC1 structure in NaCl, SCl2 is occupied by Cl, but no Cl binds at the SCl1 site; instead, a fourth site (S4) was observed between SK and SCl1 (Fig. 2C and supplementary text). Binding of Cl at the SCl2 site stabilizes the architecture of the K+ binding site by direct coordination with the main-chain amide group of Gly433 (Fig. 2B). In addition, it strengthens interactions between TM6a, TM6b, and TM10 and thus could couple their conformational changes during the ion transport process. Tyr589 has a coordination distance of 2.5 Å to Cl at SCl2; mutation of Tyr589 disrupts the SCl2 site, which would likely affect the structural integrity of the K+ binding site or the coupling of TM6a, TM6b, and TM10 during the ion transport process, resulting in loss of activity.

To confirm the three ion binding sites and to probe the function of the SCl2 site, we performed molecular dynamic simulations (fig. S17). In an equilibrium state without an electrochemical gradient, one K+ and two Cl ions remained stably bound at the SK, SCl1, and SCl2 sites throughout a 200-ns simulation (fig. S17A). By contrast, water molecules bound at the SCl1 and SCl2 sites are easy to dissociate (fig. S17, B and C). In the absence of Cl binding at the SCl2 site, the glycine-containing loop between TM6a and TM6b relaxes (fig. S17, D and E), resulting in the disturbance of K+ at the SK site. The movement of K+ further affects the binding of Cl at SCl1 and even drags it away from its original binding site (fig. S17, D and E). Thus, the removal of Cl at SCl2 triggers the correlated dissociation of K+ from SK and Cl from SCl1 with a correlation coefficient of 0.93. In addition, we performed steered molecular dynamics to pull the K+ ion from the SK site. The corresponding potential of mean force profile was used to estimate the binding free energy of K+, which reduces from −10.28 ± 1.00 kcal/mol to −3.34 ± 1.22 kcal/mol with the removal of Cl at the SCl2 site. Thus, Cl at SCl2 can facilitate K+ and Cl binding at SK and SCl1, respectively.

The KCC1 structure adopts an inward-facing conformation that is characterized by a cytosolic-facing cavity and solvent accessibility from the intracellular side to the substrate binding sites (Fig. 3A). Multiple polar residues on the surface of this cavity favor the recruitment of potassium and chloride ions (Fig. 3B). Above this cavity, several hydrophobic residues form a constriction that blocks extracellular solvent access (Fig. 3C). The extracellular gate is sealed by salt bridges and hydrogen bonds contributed by residues Arg140, Glu222, Asp575, Lys485, and Tyr486 (Fig. 3D). These interactions are expected to be disrupted when the extracellular gate opens, as observed in LeuT structures, where a salt bridge between Arg and Asp residues also participates in closing the extracellular gate (27).

Fig. 3 An inward-facing conformation of KCC1 in KCl and GDN.

(A) The surface-rendered KCC1 model shows a cytosolic-facing cavity. The yellow dashed lines indicate the opening of the intracellular gate. (B) Polar residues on the surface of the cytosolic-facing cavity. (C) Hydrophobic residues form a constriction above ion binding sites. (D) The extracellular gate is occluded by extracellular linker EL6-EL7 (pink) and pre-6a loop and post-EL5 loop (orange). (E) ECD dimerization stabilizes the closed state of the extracellular gates. For clarity, TM4, TM5, TM11, and TM12 are omitted in (B) and (C), as is TM4 in (E).

On the surface of the TMD, two regions cover the extracellular gate. First, a long extracellular linker between TM7 and TM8 folds into two short helices (EL6 and EL7) and occludes the extracellular pathway by packing tightly against TM1b and TM7 on one side and TM3 and TM8 on the other side. Second, two antiparallel loops (pre-6a loop and post-EL5 loop) from the ECD form a lid-like structural unit on top of TM6a and TM1b (Fig. 3D). EL6, EL7, the pre-6a loop, and the post-EL5 loop are tightly sandwiched by the TMD and the ECD (Fig. 3E). The packing interactions imposed on TM1b and TM6a by EL6, EL7, the pre-6a loop, and the post-EL5 loop are further stabilized by dimerization of the ECD in a relatively rigid architecture. Therefore, the ECD probably modulates KCC1 activity by constraining conformational changes on TM1b and TM6a when the extracellular gate opens.

Structural comparisons of KCC1 and DrNKCC1 reveal the conservation and divergence among different CCC members. First, both KCC1 and DrNKCC1 adopt a LeuT-like fold in an inward-facing conformation, with a root-mean-square deviation of 1.5 Å over 389 Cα atoms between two structures in the TMD (Fig. 4A). Second, the identified SK, SCl1, and SCl2 sites in KCC1 align well with those in DrNKCC1 (Fig. 4B). In DrNKCC1, the Na+ sits in the Na2 site that is conserved across many APC transporters (16, 17); by contrast, in KCC1, the two Na+ coordination Ser residues are replaced with Gly and Ala, accompanying a 2- to 3-Å shift of the TM8 backbone (Fig. 4C). With the loss of the Na+ site in KCC1, the energy barrier for transporting the Cl at the SCl2 site is unlikely to be crossed. Third, KCC1 and DrNKCC1 share a similar dimeric architecture with relatively loose packing, mediated by either lipid or detergent molecules in the TMD (fig. S18). The long ECD in KCC1 participates in the dimeric assembly, whereas the shorter CAP domain in DrNKCC1 does not.

Fig. 4 Structural comparisons of human KCC1 and DrNKCC1.

(A) The structural superimposition of KCC1 (blue) and DrNKCC1 (green) [Protein Data Bank (PDB) 6NPL] in the TMD. For clarity, large extracellular or intracellular loops are not shown. (B) The K+ and Cl bind at similar positions in KCC1 (blue) and DrNKCC1 (green). In the KCC1 structure, K+ and Cl are shown as purple and orange spheres, respectively, whereas in DrNKCC1, all three ions are shown as green spheres. (C) Compared with DrNKCC1 (green), KCC1 (blue) loses the Na+ binding site. The Na+ in the DrNKCC1 structure is shown as a cyan sphere. Red numbers show the distances in angstroms between Cα atoms of equivalent residues in KCC1 and DrNKCC1.

The structural and functional characterization of KCC1 allows us to generate a K-Cl cotransport model (fig. S19). The KCC1 structure provides a framework for interpretation of disease-related mutations of other KCCs such as KCC2 and KCC3, whose mutations cause genetic disorders of the human nervous system (fig. S20) (810). The structural studies of human KCC1, together with zebrafish NKCC1, will promote drug development and disease treatment targeting human CCCs.

Supplementary Materials

science.sciencemag.org/content/366/6464/505/suppl/DC1

Materials and Methods

Supplementary Text

Figs. S1 to S20

Table S1

References (2847)

References and Notes

Acknowledgments: We thank Y. Jiang’s support for the initiation of this project. We are grateful for X. Zhang’s help with data processing, B. Liu’s assistance with protein expression, and F. Yang’s support with the Tl+ uptake assay. Single-particle cryo-EM data were collected at the Center of Cryo-Electron Microscopy at Zhejiang University, UT Southwestern Medical Center Cryo-Electron Microscopy Facility, and Cryo-EM Platform of Peking University. We thank X. Zhang, N. Gao, and N. Li for support with facility access and data acquisition. Funding: This work was supported in part by the Ministry of Science and Technology (2018YFA0508100 to J.G. and 2016YFA0500404 to S.Y.), the National Natural Science Foundation of China (31870724 to J.G.; 31525001 and 31430019 to S.Y.; and 11722434 and 11874319 to J.L.), Zhejiang Provincial Natural Science Foundation (LR19C050002 to J.G. and LR16H0900001 to W.Y.), and the Fundamental Research Funds for the Central Universities (to J.G. and S.Y.). S.C. is supported by grants from laboratory and equipment management, Zhejiang University (SJS201814). X.B. is supported by Welch Foundation grant I-1944, the Cancer Prevention and Research Initiative of Texas and the Virginia Murchison Linthicum Scholar in Medical Research fund. E.D. is supported by NIH grants DK093501 and DK110375 and by Leducq Foundation grant 17CVD05. Author contributions: J.G., X.B., and S.Y. conceived and designed this project; S.L., J.G., C.Z., and F.W. prepared the samples; S.C., X.B., J.G., S.L., L.X., and S.Y. performed data acquisition, image processing, and structure determination; S.L., M.Z., and W.Y. performed the cell imaging assay; B.H. and J.L. performed modeling and simulation; and E.D. performed the K+ influx assay. All authors participated in the data analysis and manuscript preparation. Competing interests: The authors declare no competing financial interests. Data and materials availability: Structure coordinates and cryo-EM density maps have been deposited in the Protein Data Bank under accession numbers 6KKR and EMD-0701 for KCC1 in 150 mM KCl and GDN; 6KKT and EMD-0702 for KCC1 in 150 mM KCl and nanodisc; and 6KKU and EMD-0703 for KCC1 in 150 mM NaCl and GDN.
View Abstract

Stay Connected to Science

Navigate This Article