Case Study: K-Cl co-transporter 2 (KCC2)

This is an example of functional and structural investigation of human co-transporter/ ion channel.

KCC2 is a K-Cl co-transporter. KCC2 plays multiple roles in the physiology of central neurons. KCC2 regulates intraneuronal chloride homeostasis. KCC2 strongly influences the efficacy and polarity of the chloride-permeable γ-aminobutyric acid (GABA) type A and glycine receptor (GlyR) mediated synaptic transmission. It is critically involved in many neurological diseases including brain trauma, epilepsies, neuropathic pain, autism and schizophrenia. From biochemical point of view KCC2 has 12 TM and exists at oligomer most probably dimer.

BAG3 is a soluble protein that is known to be present very often on the surface of the macrophage (fluorescence). The idea is to apply our solubilization methodologies to maintain protein-protein interactions and after co-purification, mass spectrometry could identify the protein partner.

The challenge was to be able to produce this co-transporter without having to mutate, truncate or add any fusion proteins.

We started by expressing the protein in HEK cells and then after lysis fractionated the membranes. Western Blot (A) showed that the protein was present in the cell lysis (P1), internal membranes (P2) and plasma membranes (P3). Since KCC2 was detected in internal and plasma membrane fractions, detergent screening using classical detergents (DDM, OG, Triton, FC12 and CHAPS), novel calixarenes based detergents (CALX reagents R1 to R6) or a combination of both was applied to both membrane fractions and results were assessed by Dot Blot using a KCC2 specific antibody. A striking difference in KCC2 solubilization between internal and plasma enriched membrane fractions was observed.

Indeed, KCC2 from internal membranes was difficult to extract. This may correspond to a misfolded fraction due to KCC2 overexpression. On the contrary, plasma membrane KCC2 showed good solubilization mainly when calixarene based detergents were used alone or in combination with classical detergents.

KCC2 was totally extracted and the purity obtained using Ni-NTA (A) was relatively good (~70 %). Flag- affinity was applied as a second affinity step to improve purity (B). Combining Ni-NTA and Flag-M2 purification, improved the purity to ~90% as shown in C. In summary, KCC2 was solubilized using CALX-R3 from enriched plasma membranes fractions and isolated to a high purity using a tandem affinity purification (Ni-NTA and Flag).

Gel filtration analysis demonstrated the non-aggregated state of purified human KCC2 since no KCC2 was noticed at the void volume fraction. The two observed peaks were shown to contain dimers and monomers by Native PAGE. The use of DDM to replace CALX-R3 reagent for gel filtration resulted in KCC2 aggregates (data not shown). Thus, we showed that KCC2 purified using CALX-R3 compound exists as monomer and dimer and not higher oligomers or aggregates. This is consistent with several members of the cation-chloride cotransporter (CCC) family including KCC2 reported to forms oligomers in vivo.

To investigate the structural organization of purified KCC2, we have used negative stain electron microscopy (EM) to analyze the gel filtration fractions of KCC2 monomers and dimers. When observed by EM, the KCC2 monomer fraction showed isolated particles about 90-120 Å in size.

A total of 2000 micrographs were recorded, and 114,370 particles were selected, aligned and clustered into classes. The resulting class averages showed different views of the KCC2 monomer which often appears to be formed by two domains. A 3-D model was calculated ab initio and the image dataset was clustered into 3 classes to reveal heterogeneities or conformational changes of the sample.

The three classes are slightly elongated and are organized into two major domains connected by a linker domain. The smallest “head” domain appears to be flexible and adopts different positions as compared to the largest “core” domain. Classes I, II and III correspond to 30, 32 and 38% of the monomeric particles, respectively, suggesting that the conformational changes are continuous. Class I in which the two domains were best resolved was further refined to reach a resolution of ~15 Å.

The EM observation of the KCC2 dimer fraction revealed particles about 150-180 Å in size, significantly larger than the monomers.

To gain more insights into the quaternary structure of dimeric KCC2, we recorded 2,000 micrographs and selected 78,781 dimeric particles, which were aligned and clustered into different classes. A 3-D model was calculated ab initio from the dimeric particles and the image dataset was clustered into 3 classes to identify distinct conformations. As the three classes did not reveal significant conformational changes, the whole dataset was refined to reach a 3-D model with a resolution of ~17 Å. The 3-D model of the dimeric fraction has the same overall architecture, consisting of 4 different structural modules connecting to each other to form an elongated bent “S” shape. A flexible “hinge” region observed in all classes may account for a very flexible anchoring dimerization interface between monomers.

Sequence analysis and topology predictions suggest that KCC2 possesses two different structural domains. The first could be composed of 12 transmembrane helices (TM) and the second corresponding to a carboxy-terminal cytoplasmic domain (C-ter). The first domain is homologous to the glutamate-GABA anti-porter structure (PDB: 4DJK), while the C-ter domain displays sequence/ secondary structure homology to the prokaryotic cation-chloride cotransporter (PDB: 3G4O).

Here we show a fit of both structures filtered to 12 Å. Additional density remains unfitted and may account for linkers and long loops. Two monomers were found to fit well to the dimer. Other fits are certainly possible and higher resolution Cryo-EM structures will be required to better map KCC2 domain organization (ongoing). Interestingly, the prokaryotic cation-chloride cotransporter was shown to dimerize in solution suggesting that the KCC2 C-ter domain may be at the interface of dimerization. This is consistent with the literature showing that there are key residues such as serine 940 or other phosphorylation sites absolutely required for KCC2 function.

Our study demonstrates that in contrast to KCC2 WT and the tagged in N-terminus KCC2, modification of KCC2 at the C-terminus by addition of a His-Avi-Flag-tag led to inactivation of the transporter in both thallium assays and in electrophysiology assays (not shown).

The 3-D model of the dimeric fraction had the same overall architecture, consisting of 4 different structural modules connecting to each other to form an elongated bent “S” shape. A flexible “hinge” region observed in all classes may account for a very flexible anchoring dimerization interface between monomers. To test the contribution of disulfide bridges to KCC2 dimerization, we incubated purified KCC2 with DTT and then subjected the sample to chemical crosslinking using glutaraldehyde.

A good correlation between crosslinker concentration and the appearance of the KCC2 dimer band was observed. Interestingly, the presence of DTT results in a significant reduction of dimerization suggesting that disulfide bridges are involved in KCC2 dimerization. Therefore, we propose that the dimerization interface is composed of at least two anchor points, one being mediated by disulfide bridges (perhaps between specific extracellular loops present on monomers) and the second maybe involving functional homodimerization of the C-ter domain.

Given the critical role that KCC2 plays in neuronal function and the growing interest in generating novel selective blockers or activators, we decided to investigate binding of small molecule compounds to purified KCC2. We took advantage of the N-terminal His-tag of KCC2 for immobilization on a Ni-NTA chip to evaluate binding of a selective inhibitor VU0463271. This compound has been previously shown to selectively inhibit KCC2 leading to hyperexcitability and epileptiform discharges in the hippocampal slices. Purified KCC2 could be effectively captured and covalently tethered to the biosensor allowing to generate a stable sensor surface for subsequent ligand binding experiments. Ligand-binding competence of the tethered material was assessed with the antagonist VU0463271 in a concentration-response experiment.

Sensorgrams for binding of VU0463271 demonstrated a clear saturation, and an affinity of 3-11 µM. We investigated binding of the antagonist compound VU0463271 to immobilized monomers or dimers separated by gel filtration (B). No significant difference in binding was observed for KCC2 monomer and dimer. The specificity of the binding was assessed using non-specific compound (C). Thus, KCC2 shows direct and indirect functional evidence by Biacore and electrophysiology/ thallium assays, respectively.

Other published work in collaboration with Dr. Igor Medina (INMED, Marseille) did help to establish that the N-terminus domain of KCC2 is involved in surface delivery.

We believe that we are getting a better picture of KCC2 function. The next step is to generate high-resolution structure of KCC2 (ongoing) to better understand the mechanism of transport and to enable drug discovery allowing to meet unmet medical needs related to neurological disorders and other human diseases.

Video:

Collaborative work with THERMO FISHER SCIENTIFIC presented by Dr. Anass JAWHARI (CSO) at PEGS US 2019

Off-the-shelf functional and pure KCC2 protein produced as described will be soon available for sale in our catalogue. For more informations, please contact us.