High-resolution views and transport mechanisms of the NKCC1 and KCC transporters

Abstract

Cation-chloride cotransporters (CCCs) are responsible for the coupled co-transport of Cl⁻ with K⁺ and/or Na⁺ in an electroneutral manner. They play important roles in myriad fundamental physiological processes––from cell volume regulation to transepithelial solute transport and intracellular ion homeostasis––and are targeted by medicines commonly prescribed to treat hypertension and edema. After several decades of studies into the functions and pharmacology of these transporters, there have been several breakthroughs in the structural determination of CCC transporters. The insights provided by these new structures for the Na⁺/K⁺/Cl⁻ cotransporter NKCC1 and the K⁺/Cl⁻ cotransporters KCC1, KCC2, KCC3 and KCC4 have deepened our understanding of their molecular basis and transport function. This focused review discusses recent advances in our structural and mechanistic understanding of CCC transporters, including architecture, dimerization, functional roles of regulatory domains, ion binding sites, and coupled ion transport.

Introduction

Cation-chloride cotransporters (CCCs), also known as the SLC12 transporter family, catalyze the electroneutral transport of K⁺, Na⁺, and Cl⁻ in various stoichiometric ratios [1–6]. CCC transporters are key regulators of osmolarity and water balance both on a cellular and organ-system level [5, 7–9]. Based on phylogenetic analysis and similarities in function, CCC family members can be divided into Na⁺-dependent and Na⁺-independent groups (Figure 1). The Na⁺-dependent group comprises NKCC1, NKCC2, and NCC. NKCC1 and NKCC2 transport Na⁺, K⁺, and Cl⁻ in a 1:1:2 ratio in the same direction, while NCC transports only Na⁺ and Cl⁻, unidirectionally in a 1:1 ratio. These transporters typically mediate cellular Cl⁻ uptake [10, 11]. The Na⁺-independent group includes the four closely related transporters KCC1, KCC2, KCC3, and KCC4, all of which transport K⁺ and Cl⁻ unidirectionally in a 1:1 ratio. They are often responsible for Cl⁻ export using the electrochemical gradient of K⁺ [12, 13]. Atypical transport stoichiometries that deviate from the above-mentioned canonical ratios also have been reported for CCC transporters from several species [6, 14, 15]. Finally, two additional members that are less well studied are CCC8 and CCC9; the identity of their substrates is unclear [14, 16–18].

Figure 1.

Topology and sequence/function relationship of CCC transporters. (a) Topology of NKCC and KCC transporters. Transmembrane domain (TMD) colored in green, cytosolic domain (CTD) in marine and scissor helix in yellow. For NKCCs, the most extensive extracellular loop is between TM7/TM8. (b) Phylogenetic analysis of selected CCC transporters.

Reflecting their specific physiological functions, CCC transporters exhibit distinct expression patterns and tissue distributions. NKCC1 is expressed ubiquitously across epithelial cells on the skin and gastrointestinal system, and also in the brain and nervous tissue [6]. NKCC2 and NCC are found exclusively in the kidney, where they regulate urinary production and water retention, playing important roles in modulating water balance and blood pressure [19–21]. Among the Na⁺-independent transporters, KCC1, KCC3, and KCC4 are widely expressed; KCC2 is mostly neuron-specific [22–25]. CCC transporters are regulated by cellular stimuli. More specifically, phosphorylation of the NKCC-KCC system controls the balance of cell volume and osmolarity [26]. During times of low cell volume, NKCC and NCC are phosphorylated, which activates these proteins to trigger an influx of ions [27–29]. Low intracellular volume also causes KCC phosphorylation, which leads to its inhibition and slows K⁺ and Cl⁻ efflux. In turn, this draws in water, causing cell swelling [30]. High intracellular volume causes the opposite effect – NKCC and NCC are dephosphorylated, leading to inactivation, while KCC is activated by its dephosphorylation. This leads to K⁺ and Cl⁻ efflux from the cell, causing water to exit the cell.

Indeed, the critical physiological roles played by CCC transporters translate to a variety of clinical applications. For more than half a century––and long before their molecular identities were revealed––CCC family members have been targeted pharmacologically for treating human disease. For example, the loop diuretics and thiazide diuretics, which target these transporters, have been used over several decades and are among the most commonly prescribed medications. The NCC-targeting thiazide diuretics have been a staple treatment for high blood pressure ever since their introduction in the 1950s [31]. Another example is the NKCC-targeting drug furosemide, a loop diuretic developed in the 1960s that was rapidly adopted to treat edema and fluid overload [32]. More recently, targeting CCC family members has garnered attention as a therapeutic strategy for some neurological disorders. NKCC1 and KCC2 are capable of modulating GABA signaling due to their ability to control intracellular chloride concentration [33, 34]. Thus, activating KCC2 or inhibiting NKCC1 represent potentially promising strategies for treating epilepsy [35–39].

Despite decades of use, these drugs’ molecular mechanisms remain unresolved. While mutagenesis and functional tests have given some clues as to drug binding and function [1, 40–44], key questions remain, including the drugs’ specific binding sites and the mechanism of specificity and inhibition. Sequence similarities between various CCC transporters raise the issue of cross-reactivity and, consequently, side effects of these drugs. For example, the loop diuretics can cause hearing loss because they cross-react with NKCC1 in the endolymph of the ear [45, 46]. Thus, the ability to target NKCC2 in the kidney without affecting NKCC1 may lead to better drugs with fewer side effects. Detailed structural information is needed to facilitate the design of such highly selective therapeutics.

This review focuses on recent breakthroughs in the structural and mechanistic understanding of the CCC family of transporters and explores their implications.

Overview of CCC transporter structures

NKCC and NCC were cloned around three decades ago, initially from the shark rectal gland and winter flounder urinary bladder [47–49]. Mechanistic studies quickly commenced given the important roles of CCCs in physiology and medicine. Early mutagenesis, biochemical, and functional studies provided important insights into these transporters and revealed their basic architectural features [50–55]. In particular, CCC transporters were predicted to have 12 transmembrane (TM) helices based on both topology prediction and in vitro translation studies [56]. Hydropathy analyses suggested large N- and C-terminal cytosolic domains [57].

High resolution views of CCC family members long remained elusive, however. A crystal structure of the C-terminal domain (CTD) of a prokaryotic CCC [58] (published in 2009) provided the first glimpse into these transporters’ conserved soluble regulatory domain. The CTD structure reveals a mixed α/β fold and has two structurally related subdomains that form an overall antiparallel architecture. Advances in single particle cryo-EM [59] and membrane protein biochemistry eventually enabled high-resolution structures of CCC transporters to be captured. The high-resolution structure of NKCC1 from zebrafish [60] (NKCC1_Dr) opened the flood gates for a number of CCC structures, including NKCC1 [60–62], KCC1 [63], KCC2 [62, 64, 65], KCC3 [64, 65] and KCC4 [65, 66].

These structures revealed the basic architecture of the CCC family, including both the TM domains and CTD (Figure 1 & 2). The transmembrane domain (TMD) adopts the amino acid-polyamine-organocation (APC) superfamily fold (also known as the LeuT fold), with TM1-5 and TM6-10 forming an inverted repeat architecture (Figure 1). These CCC transporters were almost all captured in an inward facing conformation with a substrate translocation cavity extending from the cytosolic surface to around the center of the TMD. The translocation pathway primarily involves residues from TM 1, 3, 6 and 8 – similar to pathways observed in the prokaryotic APC transporters AdiC and ApcT [67, 68]. Transporters with an APC fold are widespread, constituting many distinct transporter families with diverse functions. The structure of NKCC1_Dr was the first of any APC transporter that contained a soluble regulatory domain [60]. The CTD is thought to play an important role in activation and inactivation [53] as well as dimerization [69]. The core of the CTD in NKCC1 and KCCs superimposes well with that of the prokaryotic MaCCC [58], though they share only limited sequence homology. This reflects how strongly conserved the basic architecture of the CCC transporter family is.

Figure 2.

Overall structures of NKCC1 and KCCs. (a) Structures of representative CCC transporters. TMD colored in green, CTD in marine and scissor helix in yellow. From left to right: zebrafish NKCC1(PDB ID:6NPH), human KCC1 (PDB ID:6KKR), human KCC2 (PDB ID:7D8Z), human KCC3 (PDB ID: 6M22), mouse KCC4 (PDB ID:6UKN), human KCC4 (PDB ID:7D99). (b) Structural comparisons of individual TMD and CTD protomers. NKCC1 (PDB ID:6NPH) is colored in orange, and KCC2 (PDB ID:7D8Z) is colored in cyan.

The overall architecture of KCCs is similar to that of NKCC1. The individual TMDs of NKCC1 and KCCs all share the same APC fold, and they superimpose well (Figure 2). One unique aspect to the KCCs compared to their Na-dependent counterparts is the presence of a much longer extracellular loop between TM5 and TM6 [63], which forms an ordered extracellular domain (ECD). The ECD of KCCs is composed of several β strands and short α helices and is stabilized by disulfide bonds (Figure 1). Mutations in any of the four cysteine residues involved in disulfide bonding result in loss of function in KCC1, suggesting that stabilizing this loop region is critical to function [70]. In an interesting parallel, a conserved disulfide bond [71] stabilizes the long extracellular loop between TM7 and TM8 of NKCC1.

Dimeric architecture

The structures of NKCC1 and KCCs demonstrated a dimeric assembly; this is consistent with previous crosslinking experiments that showed dimer formation in NKCCs, NCC and KCCs [72–76]. The only exception is mouse KCC4 (expressed with a GFP fusion at the C-terminus), whose structure was determined in a monomeric form [66]. Given that structures of human KCC1-4 all showed a dimeric assembly, it is likely that the dimer is the predominant tertiary assembly form of the CCC transporters. Whether both monomer and dimer forms exist under physiological conditions remains to be determined, as does whether oligomerization state regulates transport.

These CCC structures revealed several intriguing features about the dimeric architecture. First, they highlighted the unexpected domain-swapped configuration of the TMD and CTD in the dimer. That is, the TMD and CTD from the same subunit do not directly interact. Instead, these two domains, connected by a scissor helix, are on two different sides of the dimer interface, making direct interaction with the CTD and TMD from the neighboring subunit. This domain swapped configuration has been observed in the structure of NKCC1_Dr [60] and was corroborated by structures of human KCCs that contained resolved CTDs [64, 65] (Figure 2). Thus, the domain-swapped dimer appears to represent the basic organization of the CCC transporters. Second, the TMD and CTD (scissor helix included) both contribute to dimer formation; compared with the TMD dimer interface, however, the CTD interface is more extensive and likely drives dimer formation. The two CTDs directly interact, and the scissor helix on one subunit interacts closely with the scissor helix and CTD from the neighboring subunit. Third, the predicted membrane boundaries of the two TMDs within the dimer tilt relative to each other so that the bilayer at the dimer interface is lower than the edge of the dimer. This is apparent from the bent micelle surface shown in the cryo-EM maps of CCCs, and the nanodiscs that contain reconstituted dimer likewise showed a bent membrane. The physiological implications of such local membrane deformation remain unclear. It is possible that the bending of the membrane might provide a mechanism to regulate dimer stability, formation, or localization. Alternatively, transport activity might be in part regulated by the change of membrane deformation as suggested in glutamate transporters [77, 78].

Comparing the dimers of various CCC transporters shows that the CTD dimers superimpose well among NKCC1 and KCCs. This suggests a conserved CTD dimer interface and organization, which is consistent with the relatively large protein-contacting interface of the CTD (with interactions of the scissor helices included). Although the isolated CTD from the prokaryotic MaCCC [58] was proposed to form a distinct dimer organization based on crystal packing, this result might have been due to the nature of that construct. In particular, it lacked the context of a full-length protein with a TMD and the scissor helix that is critical for dimer formation of the cytosolic portion in the full length NKCC1 and KCCs. Interestingly, the isolated soluble CTD from C. elegans KCC-1 forms a dimer similar to those observed in full length CCCs [79].

In contrast to the CTDs, various interaction modes were observed between the two TMDs within the dimer. In NKCC1, the dimer interface involves mainly TM11 and TM12 with some contribution from the C-terminal end of TM10 (Figure 3(a)), which shows a relatively poor overall shape complementarity. As a result, cavities and voids were formed at the interface, which are filled by three lipid molecules as shown in the NKCC1_Dr structure [60]. These lipid molecules help to bridge the dimer interaction and may thus function as “molecular glue”. It is conceivable that lipid composition in distinct cellular environments might regulate the strength of the association of TM domains in NKCC1. In KCC1_Hs, KCC2_Hs, KCC3_Hs and KCC4_Hs structures that contain clearly resolved CTDs, the TMD dimer interface similarly involves TM11 and TM12 (Figure 3(a)) [64, 65], which form a helix-turn-helix structure. Since this dimer mode has been observed in all CCC isoforms with available structures, it is designated as the canonical dimer for CCC transporters. It is worth noting that this TMD dimer interface does not involve extensive cross-protomer interactions. The relative tilt angle and distance between TM11-TM12 from two protomers (and thus the whole TMD) vary among CCC transporters. Intriguingly, different modes of dimerization between TMDs also have been observed in KCC transporters. In one KCC1_Hs structure where the CTD is invisible [63], the TMD dimer interface involves TM12 and TM9 (Figure 3(a)); this is distinct from TM11-TM12 in the canonical interface. A second mode of TMD dimerization was observed in lower-resolution structures (~5 – 7 Å) of KCC2 and KCC3 [65]: the two TMDs interact only tangentially, mediated by each protomer’s TM12 (Figure 3(a)) that is nearly parallel to each other. Despite significant diversity in TMD dimer architecture, a common feature is that TM12 is part of the dimer interface (Figure 3(a)). This is possibly due to the geometric constraints imposed by CTD dimerization, in which the scissor helix is at the dimer interface. Since TM12, the last transmembrane helix at the TMD, is directly linked to the scissor helix, the CTD dimer would thus dictate the close proximity of TM12 from both protomers. Together, structural observations indicate that the CTD drives dimer formation and that the TMD dimer is relatively loose.

Figure 3.

The domain organization of CCC transporters. (a) TMD dimer interface. The TM helices near the dimer interface are highlighted in different colors (TM12 in red, TM11 in marine, TM10 in yellow and TM9 in green). Three types of TMD dimer interfaces among CCC transporters are shown in each boxed area. (b) TMD-CTD interaction. Surface maps of NKCC1 (PDB ID:6NPL) and KCC2 (PDB ID:7D8Z) show direct interaction between the TMD and CTD (TMD in green, CTD in blue, and scissor helix in yellow).

In structures where the CTDs are not resolved, the question persists as to whether the CTD remains a dimer or becomes dissociated. In these cases, the CTD apparently does not form stable interactions with the TMD and is mobile relative to the TMD. As a result, the CTD can become fuzzy or invisible since the particle alignment in cryo-EM data processing is mainly driven by the larger TMD. Given that the CTD dimer appears more stable than that of the TMD, the CTD is probably a dimer in solution. This would help to stabilize the marginal contact between the two TMDs, as observed in several KCC structures [65]. In line with this, the isolated CTD (without the scissor helix) from C. elegans KCC-1 forms a dimer in solution that can largely survive size-exclusion chromatography [79]. Nonetheless, the CTD dimer interface shows no obvious conservation in sequence [79] and should be able to allow changes during the activation process, as suggested by FRET studies [53, 80].

Ion binding sites

Structures of NKCC1 and KCC1-4 demonstrate that the CCC transporters’ transmembrane domains share the same basic architecture. In an inward-facing conformation, the transmembrane helices of these transporters’ promoters superimpose well, with Cα RMSD less than 1.5 Å. Thus, a key question was how the same basic architecture can give rise to diverse stoichiometries of ions: Na⁺, K⁺, and Cl⁻ at a 1:1:2 ratio for NKCCs, Na⁺ and Cl⁻ at a 1:1 ratio for NCC, and K⁺ and Cl⁻ at a 1:1 ratio for KCC [1]. Structural observations, sequence-function correlation analysis, chemical environment of the binding sites, and molecular dynamic simulations together helped to pinpoint the ion binding sites.

Interpreting ion densities in a cryo-EM map is often challenging due to background non-proteinaceous densities. Hence, high-resolution maps are particularly useful for this purpose. For example, the focused refinement of NKCC1_Dr yielded a map at 2.9 Å, which revealed relatively strong density for three ions, well above the background noise level [60]. The density attributed to K⁺ is the strongest. K⁺ coordination is mediated by mainchain carbonyls of N220 (residue numbering corresponds to NKCC1_Dr) and I221 on TM1, P417 and T420 on TM6, and the sidechain oxygen of Y305 (TM3) (Figure 4) [60]; together these give rise to a coordination configuration compatible with K⁺ binding. The same ion binding was found at the equivalent position in the cryo-EM maps of KCC1-4 (which also transport K⁺) (Figure 4) [63–66]. Sequence comparison revealed that the ion-coordinating Y305 is conserved across K⁺-transporting NKCC1-2 and KCC1-4, but not NCC (Figure 4) [60]. For NCC, which does not transport K⁺, the position is occupied by histidine. This may reflect the functional requirement for a coordinating ion given that NKCC1-2 is closely related to NCC but not to KCC1-4 in terms of overall sequence conservation. It is conceivable that a positively charged element––either K⁺ in NKCCs and KCCs or a histidine side chain in NCCs––may be required for proper function. This proposed mechanism would allow NCCs to operate by a similar mechanism to NKCC and KCC even though K⁺ is not involved in its transport cycle. Further structural and functional studies on NCC are needed to probe this proposed mechanism.

Figure 4.

The ion binding sites in NKCCs and KCCs. (a) Schematic drawings of ion binding sites in NKCC (top) and KCC (bottom). Ions are shown as spheres (potassium in magenta, chloride in brown, and sodium in marine). The proposed coordination is shown as dashed lines. (b) Sequence alignment among NKCCs, NCCs and KCCs around the ions binding sites. Residues that are involved in coordinating ions are colored in the same way as ions. Other conserved residues are colored in grey.

The Na⁺ binding site of NKCC1 was identified based on structural comparisons [60]. A so-called Na2 Na⁺-binding site is conserved in a number of APC transporters [81–84]. The same site configuration was found at the equivalent position in NKCC1, where the backbone carbonyls of L219 and W222 (on TM1; NKCC1_Dr residue numbering) and A535 (on TM8) and the sidechain oxygens of S538 and S539 (on TM8) constitute the binding site (Figure 4). It superimposes well onto that from SiaT, where a Na⁺ is bound [84]. This indicates that NKCC1_Dr is in a conformation where the Na⁺ binding site has not been disrupted. On the other hand, non-proteinaceous density potentially for Na⁺ was only slightly above background, indicating a partial occupancy of Na⁺ [60]. A partly loosened Na⁺ binding site could be compatible with the observed partially inward open state [60], considering that a fully inward-open state typically is associated with the disruption of Na2 in other APC transporters [85–87]. This Na⁺ binding site is conserved among Na⁺-dependent CCCs, with identical residues involved in NKCC1, NKCC2 and NCC (Figure 4(b)). In Na⁺-independent CCCs, i.e. KCCs, the equivalent positions do not contain the two consecutive serine residues (Figure 4(b)). Instead, human KCC1, 2, 3, and 4 have glycine and alanine residues, whose sidechains do not directly coordinate Na⁺. Structures of KCCs indeed demonstrate that their equivalent positions do not constitute an ion binding site (Figure 4(a)), corroborating the assignment of the Na⁺ binding site in NKCC1.

In the cryo-EM map of NKCC1_Dr, two strong non-proteinaceous densities were attributed to Cl⁻ based on their chemical environments [60]. One Cl⁻ (site 1) is within interacting distance to K⁺ and is coordinated by the main-chain amides of three consecutive residues––G223, V224, and M225 (on TM1). The second Cl⁻ (site 2) is coordinated by mainchain amides of another three consecutive residues––G421, I422 and L423 (on TM6)––as well as the sidechain oxygen of Y611 (on TM10). These two chloride binding sites show some similarity to those observed in CLC transporters [88]. Cryo-EM maps of human KCCs also revealed the densities of both Cl⁻ ions at equivalent positions to those of NKCC1 [63–65]. The conservation of both Cl⁻ binding sites in KCCs is puzzling since KCCs transport K⁺:Cl⁻ in a 1:1 stoichiometry ratio and only one K⁺ binding site was identified. The discrepancy between the ratio of the bound ions and the ratio of the transported ions might be reconciled in two ways. Scenario 1: One of the two bound Cl⁻ ions is not transported. In this case, the question becomes which Cl⁻ is directly involved in the transport. Since Cl⁻ at site 1 directly interacts with K⁺, this appears to provide a relatively simple mechanism for direct coupling. On the other hand, Cl⁻ at site 2 is functionally critical as mutating its coordinating tyrosine residue abolishes transport activity. The site 2 Cl⁻ is also apparently linked to K⁺ through the helical break of TM6 (residues 421–423 in NKCC1_Dr) that coordinates both K⁺ and Cl⁻ (site2) on each side. Based on sequence alignment, Cl⁻ binding site 2 is conserved in NKCC1, KCCs, and NCC (the latter does not transport K⁺), while the likely absence of a bound K⁺ in NCC raises the question whether Cl⁻ binds at site1 in NCC. If a common Cl⁻ site were used across CCC transporters (which all transport Cl⁻), then Cl⁻ site 2 may potentially represent the transporting site. At this stage, the exact roles of site 1 and site 2 in KCCs still remain to be determined. Scenario 2: An alternative possibility is that KCC may transport K⁺ and Cl⁻ in a 2:2 stoichiometry. This would still give rise to electrically neutral co-transport of K⁺ and Cl⁻. Such a mechanism requires a second K⁺ binding site, however, which has not been observed in any of the NKCC1 or KCC structures. Hence, although this cannot be formally excluded yet, it is considered unlikely.

Molecular dynamics simulations provide further support for the assignment of ion binding sites. During simulations of NKCC1_Dr (time scale: ~ 2μs) [60], K⁺ and Na⁺ remain relatively stable in their binding sites. In contrast, when they are swapped, K⁺ and Na⁺ become less stable and in one case, Na⁺ spontaneously moves back to the Na⁺ binding site [60]. These studies demonstrate that the two cation binding sites can selectively bind to either K⁺ or Na⁺. Regarding the Cl⁻ binding sites, the strongest support for site assignment from MD simulations is that Cl⁻ ions placed in the solution spontaneously diffuse into the transporter and bind at the Cl⁻ binding sites [60]. In simulations, two stable Cl⁻ binding sites were identified that match well with the Cl⁻ densities revealed in the cryo-EM map. In addition, a transient site was found that is closer to the intracellular solution and may facilitate Cl⁻ movement en route to stable binding sites deep within the transporter. Finally, MD simulations showed that the binding of cations facilitates Cl⁻ binding and vice versa. It is likely that when one ion binds it helps to stabilize the loop conformation of the helical break, thereby facilitating binding of other ions. In APC-fold transporters, helical breaks allow the backbone amide or carboxyl groups to interact with substrate; in particular, helical breaks of TM1 and TM6 are frequently involved in substrate binding. In CCC transporters, the helical break of TM1 is involved in binding Cl⁻ (site2) and K⁺, and the helical break of TM6 mediates the binding of K⁺ and Cl⁻ (site1) and Na⁺. Thus, backbone interactions play major roles in coordinating Cl⁻, K⁺ and Na⁺ and the interconnected ion binding sites observed in the structure facilitate coupled ion binding. MD simulations in KCC1 (time scale ~0.2 μs) also suggested the stable binding of K⁺ and Cl⁻ [63], corroborating the assignment of K⁺ and Cl⁻ binding sites.

To achieve the coupled co-transport of Cl⁻ with K⁺ and/or Na⁺ that is the hallmark of CCC transporters, co-transported ions from one side of the membrane must all bind to the CCC transporter before translocation [1, 6, 89, 90]. With functional studies indicating the ions bind in an ordered sequence [91, 92], a key question is the sequence of ion binding/unbinding. Here, we will mainly focus on NKCC1. Previously, cell-based functional studies on NKCC1 led to well-documented kinetic models to account for coupled ion transport by NKCC1 [89, 93, 94]. In the “glide symmetry” model, the order of ion binding to an outward-open transporter was proposed to be Na⁺, followed by Cl⁻, and then K⁺, and then a second Cl⁻; then, when the transporter switches to an inward open conformation, the first ions bound are released first [93]. In an alternative “steady state” model [89], the order of external ion bindings is Cl⁻, then Na⁺, followed by a second Cl⁻, and then K⁺. After transporter isomerization, ions are then released intracellularly in the reverse order, with K⁺ first, followed by the second Cl⁻, then Na⁺ and lastly the initial Cl⁻. The structural observations in NKCC1 seem to be more directly compatible with the “glide symmetry” model. In the cryo-EM map for the NKCC1_Dr that was captured in a partial inward-open conformation, K⁺ and both Cl⁻ densities are fairly strong, indicating a high occupancy and stable binding. In contrast, the Na⁺ density is quite weak, suggesting partial occupancy or Na⁺ in a loosely bound state. These observations imply Na⁺ likely dissociates before K⁺ when NKCC1 is in an inward-open conformation, in line with the release of Na⁺ first in the “glide symmetry” model. In addition, in the structure, the Na⁺ is on the opposite side of TM1 from K⁺ and Cl⁻. This presumably can allow Na⁺ to enter (during ion binding from the outside) or leave the transporter (during ion release to the inside) first. On the other side, Cl⁻ (site 2) is closest to the intracellular side, with K⁺ in the middle and Cl⁻ (site1) on top. Assuming ion access in a single file, this arrangement of ion binding sites provides a relatively straightforward explanation to account for the proposed order of binding/unbinding of these ions in the “glide symmetry” model: Cl⁻, followed by K⁺, and then a second Cl⁻ during binding from the extracellular side and then their release to the intracellular side in reverse order. The caveats are that determining ion binding order from the extracellular side requires a structure of NKCC1 in an outward-open conformation, and the access route of each ion has also not yet been defined. Further studies are necessary to resolve the ion binding order, a critical component of cooperative binding and coupled transport.

Communication between the transmembrane and C-terminal domain

A common feature of all the CCC structures with visible CTDs is that the interaction surface between the CTD and TMD is relatively small (Figure 3(b)), allowing conformational flexibility between these two domains. Multi-body refinement during the cryo-EM reconstruction of NKCC1_Dr revealed various motions [60], such as rocking and swiveling, that the two domains can undertake relative to one another. These analyses showed transient interactions across multiple contact points between the two domains. For example, the intracellular loop region between TM2 and TM3 interacts with both the N-terminal region of the scissor helix as well as the C-terminal region of the CTD. Additionally, the loop between TM10 and TM11 interacts with the CTD loop following the α1 helix.

Structures of KCCs showed multiple modes of interaction between the TMD and CTD. The structures of human KCC1, KCC2, KCC3 and KCC4 are highly similar when their CTDs are engaged in substantial direct interactions with the TMD. In these KCCs, the scissor helices are wedged in between TM12 from both protomers rather than right below TM12, as in NKCC1. This is accompanied by a clockwise 70° rotation of the CTD with respect to the TMD dimer (compared with NKCC1, viewed from the intracellular side) [64]. KCC3 exhibited an additional mode of organization between the CTD and TMD as shown in a lower resolution structure of around 7 Å [65]. In this conformation, the CTD dimer moves down around 7 Å away from the TMDs, adopting a relative orientation to the TMD that is more similar to that of NKCC1. In human NKCC1, human KCC1, and mouse KCC4, cryo-EM reconstructions without discernable CTD densities have also been reported, pointing to the flexibility between the CTD and TMD. A challenge in the field is how to assign the functional states of the CCC transporters observed in the cryo-EM structures given that distinct modes of CTD-TMD interactions were observed in specific KCCs, sometimes under the same conditions.

What is the functional significance of the various degrees of engagement between TMD and CTD? Mutagenesis studies on NKCC1_Dr showed that mutating key contact residues, such as R630A, N682A, and S686A resulted in loss of transport activity, suggesting an important role for the CTD in regulating the transport activity of the TM domain [60]. In NKCC1, the CTD contacts the TMD through multiple points (Figure 3(b)), which might help to stabilize particular conformational states of the TMD and thus regulate its activity. In addition, the TMD-CTD interface can also influence the N-terminal domain’s impact on transport activity as discussed in the next section.

The TMD-CTD interaction is intrinsically linked to dimerization given the domain-swapped configuration of the CCC dimer. Mutations designed to disrupt the CTD dimer interface in KCC4 cause reduced transport activity, similar to the effect of truncating the CTD [66]. It is also worth noting that proteolysis of the CTD by calpain reduced the transport activity of KCC2 [95]. A plausible explanation is that disrupting the CCC dimer impairs the CTD-TMD interaction and thus reduces transport activity.

Role of the NTD in activation and regulation of transport activity

CCC transporters have an N-terminal domain (NTD) that sits on the intracellular side of the protein and regulates the TMD’s transport activity. Compared with the CTD, the NTD shows limited conservation. Its length varies, and it lacks an ordered domain structure.

In KCCs, the NTD can inhibit transport activity by binding to the entrance of the cytosol-facing vestibule, as has been observed in structures of KCC2, KCC3 and KCC4 [62, 64, 65]. These structures revealed an ~25 residue peptide within the NTD that consists of two short helices with a loop in between. This peptide seals the vestibule entrance and thus restricts solvent access to the substrate binding pockets in an inward-open conformation. Moreover, because the loop of the peptide inserts into the entrance of the vestibule and interacts with transmembrane helices that line the vestibule, it can stall conformational state transitions, where the movement of these helices collapses the vestibule. Deleting the peptide or mutating residues to disrupt its interactions with the TMD increased transport activity [62, 64, 65]. These results support a model where this N-terminal peptide functions as an autoinhibitory element to regulate the transporter’s activity. This likely reflects a conserved regulatory mechanism for KCCs, as this sequence is conserved across KCCs but not NKCCs, and similar binding modes of the peptide were observed in multiple KCC structures [62, 64, 65].

In contrast to KCCs, removing the NKCC NTD reduces transport activity [29]. Thus, the NTDs of NKCCs likely regulate transporter function through a distinct mechanism. Indeed, post-translational modifications such as phosphorylation are a physiologically important and well-established mode of regulating the NTD in Na⁺-dependent CCCs [27, 96, 97]. Notably, decreases in Cl⁻ concentration result in NKCC phosphorylation, activating the transporter [96]. Subsequently, three phosphorylation sites have been identified in the NTD that are conserved across Na⁺-dependent CCCs [27, 97]. Yet it remains unclear how NTD phosphorylation activates NKCC. Thus far, no CCC transporter structures have shown the full picture of the NTD, probably due to its conformational flexibility. The structure of the full length NKCC1_Dr failed to yield a clearly resolved NTD. Interestingly, this structure superimposes well onto that of the NTD-truncated NKCC1_Dr [60], which presumably should not be in an activated state given the NTD’s essential role in activating NKCC1 [27, 98]. Therefore, the structure of the full-length NKCC1_Dr likely captures a state that is not activated. In this state, the NTD apparently does not make any stable direct contacts with the TM or CTDs. Nonetheless, an important question––since the conformational states that are directly involved in NKCC1 activation remain unknown––is whether this represents an inactivated state. Despite the inability to resolve the NTD, the structure of full-length NKCC1 may still hint at how the transporter is activated: the NTD is connected to the TMD at the perimeter, which likely would be in proximity to the positively charged intracellular surface area of TMD [60]. It is conceivable that the NTD, when phosphorylated, might interact with the positively charged TMD intracellular surface to regulate transport activity.

Given that the NTD presumably exerts its influence on the TMD on the intracellular surface, which is right at the TMD-CTD interface, the CTD is poised to play important roles in modulating the NTD’s effect. The structure of KCCs showed that the N-terminal inhibitory peptide can interact with the CTD when the CTD is engaged in substantial interaction with the TMD [64, 65]. This raises the possibility that CTD interaction might regulate the autoinhibitory state posed by the NTD either directly [62, 64, 65] or by determining the accessibility of the TMD intracellular surface to the NTD. Overall, the interplay among the NTD, TMD, and CTD may provide multiple layers of regulation for CCC transport activity. Hence, these underlying mechanisms warrant further investigation.

Future directions and concluding remarks

Remarkable recent advances in the structural elucidation of CCC family members, together with functional characterizations and MD simulations, have provided significant insights into how CCC transporters work at the molecular level. Despite these impressive developments in the field, many fundamental questions remain.

Among the structures of NKCC1 and all KCC members, almost all were captured in an inward-facing state, with similar conformations in the transmembrane domain. This is remarkable not least because these structures were determined in a wide range of conditions: the detergents used ranged from GDN and digitonin to LMNG or DDM; the proteins were imaged in detergent micelles or nanodiscs (with scaffold proteins of different sizes and diverse lipid compositions); the substrate ion K⁺ was present or absent. This inward-facing conformation thus likely represents a relatively stable state in vitro. Whether this is the dominant conformation in the membrane awaits future investigation.

Understanding the transport cycle and its alternating access mechanism will require determining all the various conformational states of CCC transporters. Minimally, structures of outward-facing and occluded states are needed for each type of CCC transporter. In all conformations, structures with and without substrates will also provide important insights into the loaded and apo states. In addition, outward- or inward-facing conformations may have fully and partially open states. A recent preprint revealed the structure of KCC1 with an inhibitor (UV0463271) that traps the protein in an outward-facing conformation [99]. This provides the first glimpse into a conformation other than the inward-facing state and delivers important insights into KCC1 inhibition. This structure reveals that the inhibitor disrupts K⁺ binding and prevents the outer gate from closing through steric hindrance. How closely this resembles an outward-open conformation of KCC1 during the normal transport cycle will be a subject for future studies.

Beyond structural snapshots, it is important to understand how CCC transporters transition between states and what drives the conformational changes. Understanding the dynamics of these transporters will require studying CCC transporters using biophysical and computational approaches. Previous FRET studies suggest significant conformational changes during the activation of NKCC1 [53, 80]. Now, based on the near-atomic level resolution structures, specific reporter pairs can be chosen to probe the conformational changes and dynamics of CCC transporters using FRET, DEER (double electron-electron resonance), or other spectroscopic studies. MD simulations may also provide insights into the dynamics and sequence of molecular events during conformational transitions.

CCC transporters are tightly regulated to balance physiological needs. Phosphorylation of the NTD in NKCCs/NCC and the CTD in KCCs plays a critical role in regulating their transport activities [21, 27, 100–102]. To understand these important regulatory processes, it will be crucial to capture CCC structures when the phosphorylated NTD or CTD is engaged in regulating the TMD. Such studies, in combination with functional validation, will provide deep insights into transporter regulation.

Finally, structures of NKCC1 and KCCs provide a solid foundation for structure-guided approaches to develop small molecules that target these CCC transporters. The field still awaits the structures of NKCC2 and NCC, as well as NKCCs and NCC in complex with loop and thiazide diuretics to reveal the molecular basis of their pharmacology. Such structural information will be instrumental in developing improved CCC modulators for therapeutics. For instance, NKCC1 and NKCC2 offer how such understanding might be leveraged. Currently, loop diuretics that aim to target the primarily kidney-expressed NKCC2 also end up inhibiting the ubiquitously expressed NKCC1 due to sequence similarity between the two proteins. Directly comparing the structure of NKCC2 with NKCC1 should provide valuable information to optimize/develop loop diuretics with high NKCC2 selectivity and correspondingly fewer side effects. Thus, although no new drugs that target CCC family members have been approved for several decades, this new era of increased knowledge behind the mechanism and regulation of these transporters may bring improved therapeutic options for patients suffering from diverse diseases ranging from high blood pressure and kidney disease to epilepsy and other neurological disorders.

Highlights.

• CCC transporters form dimers that show commonalities and diversity in organization.

• The NTD and CTD regulate the transport activity of the transmembrane domain.

• K⁺, Na⁺ and Cl⁻ binding sites provide insights into ion selectivity and coupling.

Abbreviations

CCC

cation-chloride cotransporters

CTD

C-terminal domain

NTD

N-terminal domain

TMD

transmembrane domain

ECD

extracellular domain

APC

amino acid-polyamine-organocation