Glen E. Kellogg, Ph.D.
Focus areas: Computational medicinal chemistry, computer-aided drug discovery, structural biology
HINT (Hydropathic INTeractions)
One of Abraham’s ideas was that the hydrophobic effect had a vastly more important role in protein structure and in drugs binding to enzymes and receptors than was being currently exploited. In particular, the seemingly rather simple partition constants for small molecules partitioning between 1-octanol and water, LogP, was a very rich thermodynamic parameter with implications beyond its uses in describing drug absorption. The partition coefficient is in fact a free energy – for the transfer of a solute between two solvents (in this case 1-octanol and water). The hypothesis was: the two solvents provide microenvironments surrounding the molecules preferentially dissolved in each, with molecules found in the water phase favoring hydrogen bonding interactions and the molecules found in the 1-octanol phase favoring hydrophobic interactions. The binding sites in proteins and other macromolecules contain similar, albeit on a smaller scale, microenvironments. Thus, the binding preferences of a small molecule are driven by exactly the same effects, as it orients itself to maximize hydrophobic-hydrophobic and polar (hydrogen bonding, etc.) interactions with its site. Consequently, if the LogP information can be parsed to individual atoms, then an empirical free energy-based scoring function for all biomolecular interactions could be implemented. Dr. Kellogg coded this scoring function model into a computer program called HINT for Hydropathic INTeractions. This was a unique and unprecedented synthesis of medicinal chemistry principles (small molecule logP and related) and structural biology (protein and protein-ligand structures).
The large majority of his research since has featured the HINT scoring function. In addition to representing interactions as numerical scores, he adopted formalisms that allowed visualizations of: 1) the hydrophobic properties of a molecule or biomacromolecule, 2) the implied hydrophobic properties of what an interacting ligand or drug should possess for a binding site, and 3) the actual interactions between species in an association, representing their types and strengths.
Research Synergism
After establishing HINT as a foundational platform, Dr. Kellogg built an independent research program with two goals: 1) continue developing novel and transformative computational tools for understanding biological structure in the crease between medicinal chemistry and structural biology, and 2) exploit these tools (and any others available) in highly collaborative projects that move collaborator’s projects and ideas forward. These two goals are actually very synergistic! Unique problems from collaborators inspire novel solutions. Further, having trainees work and learn within this framework gave them very broad and invaluable experience for their careers, as they were both creators as well as users of computational methods for drug discovery and structural biology.
To expound on these two goals, he developed algorithms to represent logP as a 3D field property, particularly for 3D quantitative structure-activity relationship (3D-QSAR) studies, since LogP is quite often a key component of QSAR models. The paper describing this, in the Journal of Computer-Aided Molecular Design, has been referenced >350 times so far, and is among the top 20 all-time cited articles from that journal. Algorithms that were designed for characterizing and predicting water molecules in binding sites, Rank and Relevance, have been adopted by other researchers (and also used by others without credit). Another method called Computational Titration was reported for optimizing ionization states of the amino acid residues and ligands at binding sites and published in the Journal of the American Chemical Society. The most recent methods development effort has been in designing a protein structure prediction tool based on HINT and the collection of visualizable 3D maps that it can produce. This also includes the Computational Titration and water prediction algorithms, which should turn out to be very germane to accurate structure predictions. Eight articles this new method have been published so far, and the next steps involve machine learning/deep learning and or evolutionary algorithms.
The variety of collaborative projects Dr. Kellogg had undertaken is vast. His group was excited to share their expertise and facilities with anyone who can benefit from it. Dr. Kellogg’s publication list highlights the key role of collaboration in his career, although many other collaborations did not lead to funding and/or manuscripts. Two examples of collaborative research were: 1) Colchicine site inhibitors of tubulin – this work was initiated by John Gupton (University of Richmond) with a large series of pyrrole compounds that had anticancer activities. The team determined that they were likely microtubule disrupting agents, and after modeling, synthesis and pharmacology studies in seven publications, we designed a new compound with superior anticancer activity that we were able to patent. 2) Most recently, his group worked with Ronald Gartenhaus (VCU Massey, Veterans Administration Richmond) on identifying lead compounds as inhibitors of eIF4A1 and USP11 with antitumor activity in lymphoma and other cancers. At least two publications are emerging from this work. In addition, a patent has been filed with these results.
Publications, Presentations, Patents and Grants
Since joining VCU, Dr. Kellogg published >150 research manuscripts in sixty-six different journals and in twelve book chapters. Twelve of the articles are in the Journal of Medicinal Chemistry, the flagship journal for the field. Two of those are invited Perspectives, which have been cited >250 and >100 times so far. More than half of his publications were in traditional medicinal chemistry journals, such as the Journal of Medicinal Chemistry, Bioorganic and Medicinal Chemistry and Molecular Medicine, and in journals devoted to a specific discipline or therapeutic target class, e.g., the Journal of Virology, Free Radical Biology and Medicine and Nucleic Acids Research. The remaining articles have been in technique (structural biology and computational chemistry) journals, e.g., the Journal of Structural Biology and the Journal of Chemical Information and Modeling or in more general journals, e.g., the Journal of the American Chemical Society and PLoS ONE. Two book chapters were invited for the Seventh (2010) and Eighth (2021) Editions of Burger’s Medicinal Chemistry (Wiley), as lead author for “Docking, Scoring and Virtual Screening in Drug Discovery”. These multi-volume sets are considered the “bible” for medicinal chemistry.
Overall, Dr. Kellogg’s h-index is 43 (Web of Science) and his d-index is 48 (research.com). In a report prepared by research.com that ranked VCU’s best scientists in chemistry, his ranking was 8/11. His research support strategy was also driven by research synergism in that funding was pursued: 1) based on his innovative modeling and methods development, and 2) collaborative projects would help support involved the trainees. Collaborative projects successfully funded laboratory, due to the numerous leads for drug discovery produced by his team. He contributed in the preparation of >100 applications.