Kathryn M. Hart

Discovery of multiple hidden allosteric sites by combining Markov state models and experiments

Significance Rational drug design efforts typically focus on identifying inhibitors that bind to protein active sites. Pockets that are not present in crystallographic structures yet can exert allosteric (i.e., long-range) control over distant active sites present an exciting alternative. However, identifying these hidden allosteric sites is extremely challenging because one usually has to simultaneously find a small molecule that binds to and stabilizes the open conformation of the pocket. Here, we present a means of combining advances in computer modeling—using Markov state models to capture long timescale dynamics—with biophysical experiments to identify hidden allosteric sites without requiring the simultaneous discovery of drug-like compounds that bind them. Using this technology, we discover multiple hidden allosteric sites in a single protein. The discovery of drug-like molecules that bind pockets in proteins that are not present in crystallographic structures yet exert allosteric control over activity has generated great interest in designing pharmaceuticals that exploit allosteric effects. However, there have only been a small number of successes, so the therapeutic potential of these pockets—called hidden allosteric sites—remains unclear. One challenge for assessing their utility is that rational drug design approaches require foreknowledge of the target site, but most hidden allosteric sites are only discovered when a small molecule is found to stabilize them. We present a means of decoupling the identification of hidden allosteric sites from the discovery of drugs that bind them by drawing on new developments in Markov state modeling that provide unprecedented access to microsecond- to millisecond-timescale fluctuations of a protein’s structure. Visualizing these fluctuations allows us to identify potential hidden allosteric sites, which we then test via thiol labeling experiments. Application of these methods reveals multiple hidden allosteric sites in an important antibiotic target—TEM-1 β-lactamase. This result supports the hypothesis that there are many as yet undiscovered hidden allosteric sites and suggests our methodology can identify such sites, providing a starting point for future drug design efforts. More generally, our results demonstrate the power of using Markov state models to guide experiments.

Thermodynamic System Drift in Protein Evolution

Tracking the evolution of thermostability in resurrected ancestors of a heat-tolerant extremophile protein and its less heat tolerant Escherichia coli homologue shows how thermostability has probably explored different mechanisms of protein stabilization over evolutionary time.

Modelling proteins’ hidden conformations to predict antibiotic resistance

TEM β-lactamase confers bacteria with resistance to many antibiotics and rapidly evolves activity against new drugs. However, functional changes are not easily explained by differences in crystal structures. We employ Markov state models to identify hidden conformations and explore their role in determining TEM’s specificity. We integrate these models with existing drug-design tools to create a new technique, called Boltzmann docking, which better predicts TEM specificity by accounting for conformational heterogeneity. Using our MSMs, we identify hidden states whose populations correlate with activity against cefotaxime. To experimentally detect our predicted hidden states, we use rapid mass spectrometric footprinting and confirm our models’ prediction that increased cefotaxime activity correlates with reduced Ω-loop flexibility. Finally, we design novel variants to stabilize the hidden cefotaximase states, and find their populations predict activity against cefotaxime in vitro and in vivo. Therefore, we expect this framework to have numerous applications in drug and protein design.

Designing small molecules to target cryptic pockets yields both positive and negative allosteric modulators

Allosteric drugs, which bind to proteins in regions other than their main ligand-binding or active sites, make it possible to target proteins considered “undruggable” and to develop new therapies that circumvent existing resistance. Despite growing interest in allosteric drug discovery, rational design is limited by a lack of sufficient structural information about alternative binding sites in proteins. Previously, we used Markov State Models (MSMs) to identify such “cryptic pockets,” and here we describe a method for identifying compounds that bind in these cryptic pockets and modulate enzyme activity. Experimental tests validate our approach by revealing both an inhibitor and two activators of TEM β-lactamase (TEM). To identify hits, a library of compounds is first virtually screened against either the crystal structure of a known cryptic pocket or an ensemble of structures containing the same cryptic pocket that is extracted from an MSM. Hit compounds are then screened experimentally and characterized kinetically in individual assays. We identify three hits, one inhibitor and two activators, demonstrating that screening for binding to allosteric sites can result in both positive and negative modulation. The hit compounds have modest effects on TEM activity, but all have higher affinities than previously identified inhibitors, which bind the same cryptic pocket but were found, by chance, via a computational screen targeting the active site. Site-directed mutagenesis of key contact residues predicted by the docking models is used to confirm that the compounds bind in the cryptic pocket as intended. Because hit compounds are identified from docking against both the crystal structure and structures from the MSM, this platform should prove suitable for many proteins, particularly targets whose crystal structures lack obvious druggable pockets, and for identifying both inhibitory and activating small-molecule modulators.

Evolutionary trend toward kinetic stability in the folding trajectory of RNases H

Significance Because protein folding is crucial to proper cellular function, there must be evolutionary pressures on how a protein achieves and maintains its folded structure. The outcome of these pressures on a folding pathway should be reflected in trends and patterns over a protein’s evolutionary history. To understand how folding pathways evolve, we characterized how reconstructed ancestral proteins of the ribonuclease H family fold. The deepest ancestors fold and unfold faster than their modern descendants, and kinetic stability evolved along both mesophilic and thermophilic lineages. This trend is possible because of a conserved partially folded intermediate state, which uncouples thermodynamic and kinetic stability to allow each parameter to evolve independently. Proper folding of proteins is critical to producing the biological machinery essential for cellular function. The rates and energetics of a protein’s folding process, which is described by its energy landscape, are encoded in the amino acid sequence. Over the course of evolution, this landscape must be maintained such that the protein folds and remains folded over a biologically relevant time scale. How exactly a protein’s energy landscape is maintained or altered throughout evolution is unclear. To study how a protein’s energy landscape changed over time, we characterized the folding trajectories of ancestral proteins of the ribonuclease H (RNase H) family using ancestral sequence reconstruction to access the evolutionary history between RNases H from mesophilic and thermophilic bacteria. We found that despite large sequence divergence, the overall folding pathway is conserved over billions of years of evolution. There are robust trends in the rates of protein folding and unfolding; both modern RNases H evolved to be more kinetically stable than their most recent common ancestor. Finally, our study demonstrates how a partially folded intermediate provides a readily adaptable folding landscape by allowing the independent tuning of kinetics and thermodynamics.

Tabtoxinine-β-lactam is a “stealth” β-lactam antibiotic that evades β-lactamase-mediated antibiotic resistance

Tabtoxinine-β-lactam (TβL) is a phytotoxin produced by plant pathogenic strains of Pseudomonas syringae. Unlike the majority of β-lactam antibiotics, TβL does not inhibit transpeptidase enzymes but instead is a potent, time-dependent inactivator of glutamine synthetase, an attractive and underexploited antibiotic target. TβL is produced by P. syringae in the form of a threonine dipeptide prodrug, tabtoxin (TβL-Thr), which enters plant and bacterial cells through dipeptide permeases. The role of β-lactamases in the self-protection of P. syringae from tabtoxin has been proposed, since this organism produces at least three β-lactamases. However, using in vitro and cellular assays and computational docking we have shown that TβL and TβL-Thr evade the action of all major classes of β-lactamase enzymes, thus overcoming the primary mechanism of resistance observed for traditional β-lactam antibiotics. TβL is a “stealth” β-lactam antibiotic and dipeptide prodrugs such as tabtoxin from P. syringae represent a novel antibiotic therapeutic strategy for treating multi-drug resistant Gram-negative pathogens expressing high levels of β-lactamase enzymes.

An Evolutionary Trend towards Kinetic Stability in the Folding Trajectory of RNases H

Proper folding of proteins is critical to producing the biological machinery essential for cellular function. Over the course of evolution, the rates and energetics of a proteins folding landscape must be maintained such that the protein folds and remains folded over its biological lifetime. Developing a comprehensive understanding of how a proteins folding process is modulated during evolution is critical to our understanding and engineering of protein biophysical properties. In this study, we characterized the folding trajectories of ancestral proteins of the ribonuclease H (RNase H) family by using ancestral sequence reconstruction to access the evolutionary history between RNases H from mesophilic and thermophilic bacteria. We find that the overall folding pathway of RNase H is preserved over billions of years of evolution. Although thermodynamic stabilities diverge between the mesophilic and thermophilic lineages, kinetic stability increases along both, with the last common ancestor folding and unfolding faster than the modern descendants. The conserved folding intermediate permits this paradoxical uncoupling of thermodynamics and kinetics, and allows for the folding landscape to independently respond to different selective pressures on global stability and kinetic barriers.