Transport unplugged: KCCs are regulated through an N‐terminal plug of the ion pathway

The ability to regulate transmembrane ion transport in response to various cues is vital to any living cell. In neurons, one key example of critical ion control relates to the extrusion of chloride mediated by the potassium‐chloride‐cotransporters (KCC1‐4). In a recent hallmark study, Chi et␣al (2021) report cryo‐EM structures of human KCC1 and KCC3b, delineating in detail how regulation by phosphorylation inhibits the transport activity. The authors also identify a stabilizing binding site for nucleotides and speculate on its functional role.

Recent structural work uncovers control mechanisms of human potassium‐chloride‐cotransporters by phosphorylation and nucleotides.

The cation chloride cotransporters (CCCs) constitute a subgroup of␣membrane proteins of the solute carrier (SLC) family referred to as the SLC12 transporters. The SLC12 family members can be roughly divided into two branches: the sodium transporting chloride importers (NCC, NKCC1‐2) and the potassium‐only transporting chloride exporters, KCC1‐4. In correspondence with their opposite directions of chloride transport, the activity regulation is inversed with importers being activated by phosphorylation while exporters are inhibited by phosphorylation (Fig 1A; Rinehart et␣al, 2009; Gagnon & Delpire, 2010).

Figure 1. Ion transport by SLC12 proteins and regulatory mechanisms.

(A) KCC1‐4 transporters are responsible for extrusion of chloride ions utilizing the cellular potassium electrochemical gradient (top panel). The activity of KCCs is inhibited by phosphorylation. Uptake of chloride ions utilizing the sodium electrochemical gradient is mediated by NKCC1‐2 and NCC (bottom panel). The activity of NKCC1‐2/NCC is stimulated by phosphorylation. (B) In response to phosphorylation, KCCs adopt an auto‐inhibited state where part of the N‐terminal domain (yellow ribbon) binds to the transmembrane domain (gray surface) preventing ion access to the binding sites (top panel). Upon removal of phosphoryl modifications, the N‐terminal domain dislodges allowing potassium ions (purple) and chloride ions (green) to access the ion‐binding sites in the transmembrane domains (bottom panel). (C) Phosphorylated residues in the C‐terminal domains (orange dots) might perturb the local spatial arrangement of the nucleotide binding sites preventing ATP/ADP from associating with the protein (top panel). When phosphoryl modifications are not present, as in the constitutive active KCC1 mutant, the nucleotide binding sites can stably engage with ATP/ADP.

Whereas many SLC transporter proteins have been structurally characterized by protein crystallography over the years, the same level of detailed insights to the SLC12 proteins has remained elusive. Fortunately, tides are changing and the ever‐improving capabilities of electron cryo‐microscopy (cryo‐EM) have enabled high‐resolution structural studies of, in particular, difficult‐to‐crystallize membrane proteins. Thus, in recent years a number of SLC12 protein cryo‐EM structures have been reported, e.g., the zebrafish NKCC1 showing the overall dimeric architecture and the potassium binding site (Chew et␣al, 2019) together with a lower‐resolution structure of the transmembrane domain of the human ortholog NKCC1 (Yang et␣al, 2020). Likewise, a series of high‐resolution structures of the transmembrane domain of human KCC1 dimer in different ionic conditions beautifully showed the potassium and chloride binding sites of the transported ions along with a secondary chloride binding site facilitating the loading of the transported ions (Liu et␣al, 2019).

Whereas the transmembrane domain of SLC12 protein structures has been sufficiently well‐resolved to allow detailed analyses, the same cannot be said for the C‐terminal domain, with the exception of a sub‐4 Å map of the zebrafish NKCC1. Recently, cryo‐EM structures of full‐length human KCC2‐4 have been reported revealing the entirety of these transporters (Chi et␣al, 2020; Xie et␣al, 2020). The available structures reveal commonalities in coordination of potassium and chloride ions between KCC1, 3, and 4. Likewise, complementary observations were made of auto‐inhibited states of KCC2‐4 where a stretch of N‐terminal residues interacts with the cytosolic side of the inward‐open cavity formed by transmembrane helices TM1a, TM5, TM6b, and TM8. In their multidisciplinary study, Chi et␣al, (2021) make a similar observation of␣auto‐inhibition for a phospho‐mimetic version of human KCC3b where access to the ion‐binding cavity is obstructed by N‐terminal residues (Fig 1B, top panel). Interestingly, the authors observe that even in the presence of the substrate cation, auto‐inhibited phospho‐mimetic KCC3b does not bind potassium suggesting that the N‐terminal plug could regulate ion access and thereby transport activity. By comparing their structure of the auto‐inhibited KCC3b with another structure determined of a constitutively active KCC1 mutant, Chi et␣al, (2021) find that the non‐phosphorylated KCC1 does not have the N‐terminal plug obstructing the path to the ion‐binding sites (Fig 1B, bottom panel). These results allude to a regulatory mechanism where phosphorylation of KCCs induces conformational changes to the N‐terminal domains, causing their association with the transmembrane domains and thereby auto‐inhibition.

Further evidence from hydrogen–deuterium exchange mass spectrometry experiments highlights the profound impact of phosphorylation modifications on the protein dynamics of KCC transporters, where the so‐called scissor helix, linking the transmembrane‐ and C‐terminal domains, is observed as being more dynamic in a non‐phosphorylated KCC3b mutant as compared to the phospho‐mimetic KCC3b protein. In the non‐phosphorylated KCC1 structure, this dynamic behavior is relayed downwards to the C‐terminal domains, which adopts different conformations with respect to the transmembrane domains, but also locally where especially the outer lobe helices 6‐10 adopt strikingly different tertiary folds. These substantial conformational changes can be linked to the regulatory phosphorylation sites found in the C‐terminal domain. In their proteomics analyses, Chi et␣al, (2021) identified the scissor helix residue Ser685 as a target for phosphorylation, as well as the residues Thr727, Ser930, and Tyr1070, all embedded in the C‐terminal domain. These phosphorylation sites reside close to the flexible regions and thus might very well serve the function of regulating the conformation of the C‐terminal domains and the transport activity.

A meticulous inspection of their cryo‐EM map of the non‐phosphorylated KCC1 protein led Chi et␣al, (2021) to a remarkable observation of extra density in the inner lobe␣of the C‐terminal domains. The map features, as well as the local chemical environment, suggest that a mix of adenine nucleotides ATP or ADP is bound at a Rossmann‐like fold. This unique observation has not been reported before for other KCC protein structures (Chi et␣al, 2020; Xie et␣al, 2020), and close inspection of the cryo‐EM maps from these complementary studies fail to show similar extra densities for adenine nucleotides in equivalent positions within the C‐terminal domains. In line with this, the phospho‐mimetic KCC3b structure did not show any map features indicative of nucleotide binding (Chi et␣al, 2021). Obviously, this raises the question of the nature of the nucleotide observed in the non‐phosphorylated KCC1 structure—is it natively bound or exogenously associated from the purification protocol? The authors address this very question through molecular dynamics (MD) simulations together with biophysical experiments, where the latter show a significant and specific thermostabilizing effect of adenine nucleotides to KCC1. This is supported by MD simulations, observing a stable interaction between KCC1 C‐terminal domains and ATP. The results on KCC3b wild‐type protein and the respective mutants, however, are less clear as the thermostabilizing effect is observed indeed for the KCC3b mutants, but MD simulations suggest a much less stable interaction in line with the lack of nucleotide density in the cryo‐EM map. Thus, as interesting and truly fascinating the discovery of the bound nucleotide is, there are many outstanding questions to this observation that will be important to address for complete elucidation. However, Chi et␣al, (2021) make one key analysis as to whether the adenine nucleotide binding is a common feature to other SLC12‐family proteins and indeed by closely inspecting the published cryo‐EM map of zebrafish NKCC1 (Chew et␣al, 2019) they actually find unmodelled map density features in the C‐terminal domains corresponding to an adenine nucleotide. As already hypothesized by Chi et␣al, (2021), the nucleotide binding to the C‐terminal domain might be subjected to regulation by the phosphorylation state of the SLC12 protein in question (Fig 1C). Residue Thr727 in the phospho‐mimetic KCC3b was found to be phosphorylated, whereas the corresponding residue Thr713 was found non‐phosphorylated in the KCC1 mutant. Given the vicinity to the nucleotide binding site, the presence of a phosphoryl group might perturb the local environment enough that nucleotide binding is less favorable. These findings immediately spark the interest into the nucleotide binding as a general concept for SLC12‐family proteins, as also indicated from membrane proteomics studies (Jelcic et␣al, 2020). It prompts also new questions of a mechanistic coupling between chloride transport/signaling and ATP status, which is sensitive to, e.g., Na⁺, K⁺‐ATPase activity.

In conclusion, the study by Chi et␣al, (2021) opens an avenue of exciting research directions to further explore within the SLC12 family of membrane transporters and their roles in cellular or tissue physiology. Future work will be inspired to further investigate the regulatory functions of phosphorylation states, especially in the context of the transport cycle. Furthermore, elucidating the enigmatic association of nucleotides with these transporters will be of outstanding importance as will analyses to address whether there is a catalytic or scaffolding function.

The EMBO Journal (2021) 40: e108371.