To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.

 
 

Publication Date

2-19-2025

Description

Purpose

KCC2 is a potassium-chloride cotransporter that plays a critical role in neuronal function by regulating GABAergic signaling via chloride gradients. Maintaining this concentration gradient is crucial for balancing excitation and inhibition in the brain. While KCC2 dysregulation has been implicated in epilepsy and seizures, recent studies suggest that KCC2 inhibition results in phenotypes mirroring those seen in chronic opioid dependence. As such, increasing evidence support KCC2 being a potential therapeutic target for modulating behaviors associated with substances of abuse. Currently, only a few direct small molecule agonists against KCC2 have been reported, and no definitive active site on the protein has been thoroughly documented. Here, we report on findings resulting from the utilization of a machine-learning computational approach for modeling KCC2 protein interactions with a direct antagonist to find nine potential binding sites.

Methods

An in-silico model of the KCC2 protein was prepared and generated using published cryo-EM structural data from the RCSB Protein Data Bank. The protein surface was extrapolated from the atomic structure model and screened for potential active sites which were scored based on a variety of parameters including size, depth, hydrophobicity, and ability to facilitate intermolecular interactions including hydrogen bonds and Van der Waals forces. In-silico ligand models were also prepared and generated for each drug candidate. These models were collated into a small molecule library containing structures for the experimental agonists as well as a number of FDA-approved negative controls. The resulting library was screened against each identified binding site using a rigid docking protocol and each ligand interaction was scored on a variety of parameters including Van der Waals energy, Coulomb energy, hydrophobicity, and capacity for hydrogen bonding.

Results

Surface analysis of the KCC2 protein identified nine potential active sites with site scores ranging from 0.895 to 0.979 (0.80 indicates ability to distinguish between drug-binding and non-drug-binding sites). The highest scoring potential active site was identified in the transmembrane domain of KCC2 associated with residues SER896-PHE1090. Common residue interactions occurred at HIE1051 (Pi-Pi), ASN1086 (H-bond), and PHE1090 (Pi-Pi) with all potential agonist candidates. The average docking score for ligand candidates was -4.5228 while negative control candidates scored -3.6087 on average (a more negative score is indicative of a more energetically favorable conformation). These results are consistent with our previous findings.

Conclusion

Our machine-learning computational approach successfully identified nine potential active binding sites on the KCC2 protein, with the highest-scoring site located within the transmembrane domain (SER896–PHE1090). Key residue interactions including Pi-Pi stacking and hydrogen bonding were consistently observed across all agonist candidates, highlighting possible regions for drug binding. The difference in docking scores between agonist candidates and negative controls support the specificity of these identified sites. While these findings are promising, it is noted that the docking scores for our library of ligand candidates are not as highly scored as other well-documented protein:ligand interactions (ibuprofen covalent binding to cyclooxygenase-2 results in a docking score of -10.3 using the same experimental parameters). These docking scores could be improved using induced fit modeling which would account for protein conformation changes resulting from ligand interactions. Finally, we acknowledge the need for binding and functional assays need to be performed to validate on-target protein:ligand activitiy. Overall, these findings provide insights for the rational design of novel KCC2-targeted therapeutics, offering avenues for developing treatments aimed at modulating behaviors associated with substance abuse.

Disciplines

Biochemistry | Medical Biochemistry | Medical Cell Biology | Medical Molecular Biology | Medical Neurobiology | Medical Pharmacology | Medicinal and Pharmaceutical Chemistry | Molecular and Cellular Neuroscience | Molecular Biology | Nervous System Diseases | Neurosciences | Pharmaceutics and Drug Design | Structural Biology | Substance Abuse and Addiction

Keywords

KCC2; Substance Abuse, Neuroscience, Structural Biology, Medicinal Chemistry, Drug Design

Document Type

Poster

Event Website

https://www.roseman.edu/research/research-symposium/

Download

Share

COinS
 

In-silico modeling and characterization of KCC2 protein interaction with a small molecule direct agonist identifies potential binding sites for modulating behaviors associated with substances of abuse

Purpose

KCC2 is a potassium-chloride cotransporter that plays a critical role in neuronal function by regulating GABAergic signaling via chloride gradients. Maintaining this concentration gradient is crucial for balancing excitation and inhibition in the brain. While KCC2 dysregulation has been implicated in epilepsy and seizures, recent studies suggest that KCC2 inhibition results in phenotypes mirroring those seen in chronic opioid dependence. As such, increasing evidence support KCC2 being a potential therapeutic target for modulating behaviors associated with substances of abuse. Currently, only a few direct small molecule agonists against KCC2 have been reported, and no definitive active site on the protein has been thoroughly documented. Here, we report on findings resulting from the utilization of a machine-learning computational approach for modeling KCC2 protein interactions with a direct antagonist to find nine potential binding sites.

Methods

An in-silico model of the KCC2 protein was prepared and generated using published cryo-EM structural data from the RCSB Protein Data Bank. The protein surface was extrapolated from the atomic structure model and screened for potential active sites which were scored based on a variety of parameters including size, depth, hydrophobicity, and ability to facilitate intermolecular interactions including hydrogen bonds and Van der Waals forces. In-silico ligand models were also prepared and generated for each drug candidate. These models were collated into a small molecule library containing structures for the experimental agonists as well as a number of FDA-approved negative controls. The resulting library was screened against each identified binding site using a rigid docking protocol and each ligand interaction was scored on a variety of parameters including Van der Waals energy, Coulomb energy, hydrophobicity, and capacity for hydrogen bonding.

Results

Surface analysis of the KCC2 protein identified nine potential active sites with site scores ranging from 0.895 to 0.979 (0.80 indicates ability to distinguish between drug-binding and non-drug-binding sites). The highest scoring potential active site was identified in the transmembrane domain of KCC2 associated with residues SER896-PHE1090. Common residue interactions occurred at HIE1051 (Pi-Pi), ASN1086 (H-bond), and PHE1090 (Pi-Pi) with all potential agonist candidates. The average docking score for ligand candidates was -4.5228 while negative control candidates scored -3.6087 on average (a more negative score is indicative of a more energetically favorable conformation). These results are consistent with our previous findings.

Conclusion

Our machine-learning computational approach successfully identified nine potential active binding sites on the KCC2 protein, with the highest-scoring site located within the transmembrane domain (SER896–PHE1090). Key residue interactions including Pi-Pi stacking and hydrogen bonding were consistently observed across all agonist candidates, highlighting possible regions for drug binding. The difference in docking scores between agonist candidates and negative controls support the specificity of these identified sites. While these findings are promising, it is noted that the docking scores for our library of ligand candidates are not as highly scored as other well-documented protein:ligand interactions (ibuprofen covalent binding to cyclooxygenase-2 results in a docking score of -10.3 using the same experimental parameters). These docking scores could be improved using induced fit modeling which would account for protein conformation changes resulting from ligand interactions. Finally, we acknowledge the need for binding and functional assays need to be performed to validate on-target protein:ligand activitiy. Overall, these findings provide insights for the rational design of novel KCC2-targeted therapeutics, offering avenues for developing treatments aimed at modulating behaviors associated with substance abuse.

https://ecommons.roseman.edu/researchsymposium/2025/other/11