Positive and negative design in stability and thermal adaptation of natural proteins

The aim of this work is to elucidate how physical principles of protein design are reflected in natural sequences that evolved in response to the thermal conditions of the environment. Using an exactly solvable lattice model, we design sequences with selected thermal properties. Compositional analysis of designed model sequences and natural proteomes reveals a specific trend in amino acid compositions in response to the requirement of stability at elevated environmental temperature: the increase of fractions of hydrophobic and charged amino acid residues at the expense of polar ones. We show that this “from both ends of the hydrophobicity scale” trend is due to positive (to stabilize the native state) and negative (to destabilize misfolded states) components of protein design. Negative design strengthens specific repulsive non-native interactions that appear in misfolded structures. A pressure to preserve specific repulsive interactions in non-native conformations may result in correlated mutations between amino acids that are far apart in the native state but may be in contact in misfolded conformations. Such correlated mutations are indeed found in TIM barrel and other proteins.     

Coupling Protein Side-Chain and Backbone Flexibility Improves the Re-design of Protein-Ligand Specificity

Interactions between small molecules and proteins play critical roles in regulating and facilitating diverse biological functions, yet our ability to accurately re-engineer the specificity of these interactions using computational approaches has been limited. One main difficulty, in addition to inaccuracies in energy functions, is the exquisite sensitivity of protein–ligand interactions to subtle conformational changes, coupled with the computational problem of sampling the large conformational search space of degrees of freedom of ligands, amino acid side chains, and the protein backbone. Here, we describe two benchmarks for evaluating the accuracy of computational approaches for re-engineering protein-ligand interactions: (i) prediction of enzyme specificity altering mutations and (ii) prediction of sequence tolerance in ligand binding sites. After finding that current state-of-the-art “fixed backbone” design methods perform poorly on these tests, we develop a new “coupled moves” design method in the program Rosetta that couples changes to protein sequence with alterations in both protein side-chain and protein backbone conformations, and allows for changes in ligand rigid-body and torsion degrees of freedom. We show significantly increased accuracy in both predicting ligand specificity altering mutations and binding site sequences. These methodological improvements should be useful for many applications of protein – ligand design. The approach also provides insights into the role of subtle conformational adjustments that enable functional changes not only in engineering applications but also in natural protein evolution.     

Mechanistic Explanations for Restricted Evolutionary Paths That Emerge from Gene Regulatory Networks

The extent and the nature of the constraints to evolutionary trajectories are central issues in biology. Constraints can be the result of systems dynamics causing a non-linear mapping between genotype and phenotype. How prevalent are these developmental constraints and what is their mechanistic basis? Although this has been extensively explored at the level of epistatic interactions between nucleotides within a gene, or amino acids within a protein, selection acts at the level of the whole organism, and therefore epistasis between disparate genes in the genome is expected due to their functional interactions within gene regulatory networks (GRNs) which are responsible for many aspects of organismal phenotype. Here we explore epistasis within GRNs capable of performing a common developmental function – converting a continuous morphogen input into discrete spatial domains. By exploring the full complement of GRN wiring designs that are able to perform this function, we analyzed all possible mutational routes between functional GRNs. Through this study we demonstrate that mechanistic constraints are common for GRNs that perform even a simple function. We demonstrate a common mechanistic cause for such a constraint involving complementation between counter-balanced gene-gene interactions. Furthermore we show how such constraints can be bypassed by means of “permissive” mutations that buffer changes in a direct route between two GRN topologies that would normally be unviable. We show that such bypasses are common and thus we suggest that unlike what was observed in protein sequence-function relationships, the “tape of life” is less reproducible when one considers higher levels of biological organization.     

Design of multi-specificity in protein interfaces

Interactions in protein networks may place constraints on protein interface sequences to maintain correct and avoid unwanted interactions. Here we describe a “multi-constraint” protein design protocol to predict sequences optimized for multiple criteria, such as maintaining sets of interactions, and apply it to characterize the mechanism and extent to which 20 multi-specific proteins are constrained by binding to multiple partners. We find that multi-specific binding is accommodated by at least two distinct patterns. In the simplest case, all partners share key interactions, and sequences optimized for binding to either single or multiple partners recover only a subset of native amino acid residues as optimal. More interestingly, for signaling interfaces functioning as network “hubs,” we identify a different, “multi-faceted” mode, where each binding partner prefers its own subset of wild-type residues within the promiscuous binding site. Here, integration of preferences across all partners results in sequences much more “native-like” than seen in optimization for any single binding partner alone, suggesting these interfaces are substantially optimized for multi-specificity. The two strategies make distinct predictions for interface evolution and design. Shared interfaces may be better small molecule targets, whereas multi-faceted interactions may be more “designable” for altered specificity patterns. The computational methodology presented here is generalizable for examining how naturally occurring protein sequences have been selected to satisfy a variety of positive and negative constraints, as well as for rationally designing proteins to have desired patterns of altered specificity.     

Composites of local structure propensities: Evidence for local encoding of long-range structure

To estimate how extensively the ensemble of denatured-state conformations is constrained by local side-chain–backbone interactions, propensities of each of the 20 amino acids to occur in mono- and dipeptides mapped to discrete regions of the Ramachandran map are computed from proteins of known structure. In addition, propensities are computed for the trans, gauche−, and gauche+ rotamers, with or without consideration of the values of phi and psi. These propensities are used in scoring functions for fragment threading, which estimates the energetic favorability of fragments of protein sequence to adopt the native conformation as opposed to hundreds of thousands of incorrect conformations. As finer subdivisions of the Ramachandran plot, neighboring residue phi/psi angles, and rotamers are incorporated, scoring functions become better at ranking the native conformation as the most favorable. With the best composite propensity function, the native structure can be distinguished from 300,000 incorrect structures for 71% of the 2130 arbitrary protein segments of length 40, 48% of 2247 segments of length 30, and 20% of 2368 segments of length 20. A majority of fragments of length 30–40 are estimated to be folded into the native conformation a substantial fraction of the time. These data suggest that the variations observed in amino acid frequencies in different phi/psi/chi1 environments in folded proteins reflect energetically important local side-chain–backbone interactions, interactions that may severely restrict the ensemble of conformations populated in the denatured state to a relatively small subset with nativelike structure.     

A new hydrogen-bonding potential for the design of protein-RNA interactions predicts specific contacts and discriminates decoys

RNA-binding proteins play many essential roles in the regulation of gene expression in the cell. Despite the significant increase in the number of structures for RNA–protein complexes in the last few years, the molecular basis of specificity remains unclear even for the best-studied protein families. We have developed a distance and orientation-dependent hydrogen-bonding potential based on the statistical analysis of hydrogen-bonding geometries that are observed in high-resolution crystal structures of protein–DNA and protein–RNA complexes. We observe very strong geometrical preferences that reflect significant energetic constraints on the relative placement of hydrogen-bonding atom pairs at protein–nucleic acid interfaces. A scoring function based on the hydrogen-bonding potential discriminates native protein–RNA structures from incorrectly docked decoys with remarkable predictive power. By incorporating the new hydrogen-bonding potential into a physical model of protein–RNA interfaces with full atom representation, we were able to recover native amino acids at protein–RNA interfaces.     

The interplay of SARS-CoV-2 evolution and constraints imposed by the structure and functionality of its proteins

The unprecedented pace of the sequencing of the SARS-CoV-2 virus genomes provides us with unique information about the genetic changes in a single pathogen during ongoing pandemic. By the analysis of close to 200,000 genomes we show that the patterns of the SARS-CoV-2 virus mutations along its genome are closely correlated with the structural and functional features of the encoded proteins. Requirements of foldability of proteins’ 3D structures and the conservation of their key functional regions, such as protein-protein interaction interfaces, are the dominant factors driving evolutionary selection in protein-coding genes. At the same time, avoidance of the host immunity leads to the abundance of mutations in other regions, resulting in high variability of the missense mutation rate along the genome. “Unexplained” peaks and valleys in the mutation rate provide hints on function for yet uncharacterized genomic regions and specific protein structural and functional features they code for. Some of these observations have immediate practical implications for the selection of target regions for PCR-based COVID-19 tests and for evaluating the risk of mutations in epitopes targeted by specific antibodies and vaccine design strategies.     

Coupling backbone flexibility and amino acid sequence selection in protein design.

Using a protein design algorithm that considers side-chain packing quantitatively, the effect of explicit backbone motion on the selection of amino acids in protein design was assessed in the core of the streptococcal protein G beta 1 domain (G beta 1). Concerted backbone motion was introduced by varying G beta 1's supersecondary structure parameter values. The stability and structural flexibility of seven of the redesigned proteins were determined experimentally and showed that core variants containing as many as 6 of 10 possible mutations retain native-like properties. This result demonstrates that backbone flexibility can be combined explicitly with amino acid side-chain selection and that the selection algorithm is sufficiently robust to tolerate perturbations as large as 15% of G beta 1's native supersecondary structure parameter values.     

Plasticity and steric strain in a parallel beta-helix: rational mutations in the P22 tailspike protein

By means of genetic screens, a great number of mutations that affect the folding and stability of the tailspike protein from Salmonella phage P22 have been identified. Temperature-sensitive folding (tsf) mutations decrease folding yields at high temperature, but hardly affect thermal stability of the native trimeric structure when assembled at low temperature. Global suppressor (su) mutations mitigate this phenotype. Virtually all of these mutations are located in the central domain of tailspike, a large parallel beta-helix. We modified tailspike by rational single amino acid replacements at three sites in order to investigate the influence of mutations of two types: (1) mutations expected to cause a tsf phenotype by increasing the side-chain volume of a core residue, and (2) mutations in a similar structural context as two of the four known su mutations, which have been suggested to stabilize folding intermediates and the native structure by the release of backbone strain, an effect well known for residues that are primarily evolved for function and not for stability or folding of the protein. Analysis of folding yields, refolding kinetics and thermal denaturation kinetics in vitro show that the tsf phenotype can indeed be produced rationally by increasing the volume of side chains in the beta-helix core. The high-resolution crystal structure of mutant T326F proves that structural rearrangements only take place in the remarkably plastic lumen of the beta-helix, leaving the arrangement of the hydrogen-bonded backbone and thus the surface of the protein unaffected. This supports the notion that changes in the stability of an intermediate, in which the beta-helix domain is largely formed, are the essential mechanism by which tsf mutations affect tailspike folding. A rational design of su mutants, on the other hand, appears to be more difficult. The exchange of two residues in the active site expected to lead to a drastic release of steric strain neither enhanced the folding properties nor the stability of tailspike. Apparently, side-chain interactions in these cases overcompensate for backbone strain, illustrating the extreme optimization of the tailspike protein for conformational stability. The result exemplifies the view arising from the statistical analysis of the distribution of backbone dihedral angles in known three-dimensional protein structures that the adoption of straight phi/psi angles other than the most favorable ones is often caused by side-chain interactions. Proteins 2000;39:89-101.     

Evolutionary variation and adaptation in a conserved protein kinase allosteric network: Implications for inhibitor design

The activation of protein kinases involves conformational changes in key functional regions of the kinase domain, a detailed understanding of which is essential for the design of selective protein kinase inhibitors. Through statistical analysis of protein kinase sequences and crystal structures from diverse organisms, we recently proposed that the activation of protein kinases involves a hidden strain switch in the catalytic loop. Specifically, we demonstrated that the backbone torsion-angles of residues in the catalytic loop switch from a “relaxed” to “strained” conformation upon kinase activation and the strained geometry results in a network of hydrogen bonds involving conserved non-catalytic residues in the ATP and substrate binding lobes. Here, we further explore this activation mechanism by analyzing families that lack the canonical hydrogen bonding interactions with the strained backbone. We find that alternative mechanisms have evolved to maintain catalytic loop strain. In PIM kinase, for example, two water molecules account for the lack of a conserved aspartate in the substrate binding by hydrogen bonds to the strained backbone. We discuss the relevance of these findings in the design of family-specific allosteric inhibitors, and in predicting the structural and functional impact of cancer mutations that alter the strain associated hydrogen bonding network. This article is part of a Special Issue entitled: Inhibitors of Protein Kinases (2012).     

Aromatic cluster mutations produce focal modulations of β-sheet structure

Site-directed mutagenesis is a powerful tool for altering the structure and function of proteins in a focused manner. Here, we examined how a model β-sheet protein could be tuned by mutation of numerous surface-exposed residues to aromatic amino acids. We designed these aromatic side chain “clusters” at highly solvent-exposed positions in the flat, single-layer β-sheet of Borrelia outer surface protein A (OspA). This unusual β-sheet scaffold allows us to interrogate the effects of these mutations in the context of well-defined structure but in the absence of the strong scaffolding effects of globular protein architecture. We anticipated that the introduction of a cluster of aromatic amino acid residues on the β-sheet surface would result in large conformational changes and/or stabilization and thereby provide new means of controlling the properties of β-sheets. Surprisingly, X-ray crystal structures revealed that the introduction of aromatic clusters produced only subtle conformational changes in the OspA β-sheet. Additionally, despite burying a large degree of hydrophobic surface area, the aromatic cluster mutants were slightly less stable than the wild-type scaffold. These results thereby demonstrate that the introduction of aromatic cluster mutations can serve as a means for subtly modulating β-sheet conformation in protein design.     

Intrinsic α‐helical and β‐sheet conformational preferences: A computational case study of alanine

A fundamental question in protein science is what is the intrinsic propensity for an amino acid to be in an α-helix, β-sheet, or other backbone dihedral angle (-ψ) conformation. This question has been hotly debated for many years because including all protein crystal structures from the protein database, increases the probabilities for α-helical structures, while experiments on small peptides observe that β-sheet-like conformations predominate. We perform molecular dynamics (MD) simulations of a hard-sphere model for Ala dipeptide mimetics that includes steric interactions between nonbonded atoms and bond length and angle constraints with the goal of evaluating the role of steric interactions in determining protein backbone conformational preferences. We find four key results. For the hard-sphere MD simulations, we show that (1) β-sheet structures are roughly three and half times more probable than α-helical structures, (2) transitions between α-helix and β-sheet structures only occur when the backbone bond angle τ (N–C_α–C) is greater than 110°, and (3) the probability distribution of τ for Ala conformations in the “bridge” region of-ψ space is shifted to larger angles compared to other regions. In contrast, (4) the distributions obtained from Amber and CHARMM MD simulations in the bridge regions are broader and have increased τ compared to those for hard sphere simulations and from high-resolution protein crystal structures. Our results emphasize the importance of hard-sphere interactions and local stereochemical constraints that yield strong correlations between -ψ conformations and τ.     

Functional analysis of predicted coiled-coil regions in the Escherichia coli K-12 O-antigen polysaccharide chain length determinant Wzz

Wzz is a membrane protein that determines the chain length distribution of the O-antigen lipopolysaccharide by an unknown mechanism. Wzz proteins consist of two transmembrane helices separated by a large periplasmic loop. The periplasmic loop of Escherichia coli K-12 Wzz (244 amino acids from K65 to A308) was purified and found to be a monomer with an extended conformation, as determined by gel filtration chromatography and analytical ultracentrifugation. Circular dichroism showed that the loop has a 60% helical content. The Wzz periplasmic loop also contains three regions with predicted coiled coils. To probe the function of the predicted coiled coils, we constructed amino acid replacement mutants of the E. coli K-12 Wzz protein, which were designed so that the coiled coils could be separate without compromising the helicity of the individual molecules. Mutations in one of the regions, spanning amino acids 108 to 130 (region I), were associated with a partial defect in O-antigen chain length distribution, while mutants with mutations in the region spanning amino acids 209 to 223 (region III) did not have an apparent functional defect. In contrast, mutations in the region spanning amino acids 153 to 173 (region II) eliminated the Wzz function. This phenotype was associated with protein instability, most likely due to conformational changes caused by the amino acid replacements, which was confirmed by limited trypsin proteolysis. Additional mutagenesis based on a three-dimensional model of region I demonstrated that the amino acids implicated in function are all located at the same face of a predicted α-helix, suggesting that a coiled coil actually does not exist in this region. Together, our results suggest that the regions predicted to be coiled coils are important for Wzz function because they maintain the native conformation of the protein, although the existence of coiled coils could not be demonstrated experimentally.     

Carbonyl-carbonyl interactions stabilize the partially allowed Ramachandran conformations of asparagine and aspartic acid

Asparagine and aspartate are known to adopt conformations in the left-handed alpha-helical region and other partially allowed regions of the Ramachandran plot more readily than any other non-glycyl amino acids. The reason for this preference has not been established. An examination of the local environments of asparagine and aspartic acid in protein structures with a resolution better than 1.5 A revealed that their side-chain carbonyls are frequently within 4 A of their own backbone carbonyl or the backbone carbonyl of the previous residue. Calculations using protein structures with a resolution better than 1.8 A reveal that this close contact occurs in more than 80% of cases. This carbonyl-carbonyl interaction offers an energetic sabilization for the partially allowed conformations of asparagine and aspartic acid with respect to all other non-glycyl amino acids. The non-covalent attractive interactions between the dipoles of two carbonyls has recently been calculated to have an energy comparable to that of a hydrogen bond. The preponderance of asparagine in the left-handed alpha-helical region, and in general of aspartic acid and asparagine in the partially allowed regions of the Ramachandran plot, may be a consequence of this carbonyl-carbonyl stacking interaction.     

A backbone‐dependent rotamer library with high (ϕ, ψ) coverage using metadynamics simulations

Backbone‐dependent rotamer libraries are commonly used to assign the side chain dihedral angles of amino acids when modeling protein structures. Most rotamer libraries are created by curating protein crystal structure data and using various methods to extrapolate the existing data to cover all possible backbone conformations. However, these rotamer libraries may not be suitable for modeling the structures of cyclic peptides and other constrained peptides because these molecules frequently sample backbone conformations rarely seen in the crystal structures of linear proteins. To provide backbone‐dependent side chain information beyond the α‐helix, β‐sheet, and PPII regions, we used explicit‐solvent metadynamics simulations of model dipeptides to create a new rotamer library that has high coverage in the (ϕ, ψ) space. Furthermore, this approach can be applied to build high‐coverage rotamer libraries for noncanonical amino acids. The resulting Metadynamics of Dipeptides for Rotamer Distribution (MEDFORD) rotamer library predicts the side chain conformations of high‐resolution protein crystal structures with similar accuracy (~80%) to a state‐of‐the‐art rotamer library. Our ability to test the accuracy of MEDFORD at predicting the side chain dihedral angles of amino acids in noncanonical backbone conformation is restricted by the limited structural data available for cyclic peptides. For the cyclic peptide data that are currently available, MEDFORD and the state‐of‐the‐art rotamer library perform comparably. However, the two rotamer libraries indeed make different rotamer predictions in noncanonical (ϕ, ψ) regions. For noncanonical amino acids, the MEDFORD rotamer library predicts the χ ₁ values with approximately 75% accuracy.     

Substitutions of proline 76 in yeast iso-1-cytochrome c. Analysis of residues compatible and incompatible with folding requirements

Fine-structure genetic mapping previously revealed numerous nonfunctional cyc1 mutations having alterations at or near the site corresponding to amino acid position 76 of iso-1-cytochrome c from the yeast Saccharomyces cerevisiae. DNA sequencing of the alterations in four of these cyc1 mutations indicated that the normal Pro-76 was replaced by Leu-76. Revertants containing at least partially functional iso-1-cytochromes c were isolated, and the alterations were analyzed by DNA sequencing and protein analysis. Specific activities of the altered iso-1-cytochromes c were estimated in vivo by growth of the strains in lactate medium; compared to normal iso-1-cytochrome c with Pro-76, the following activities were associated with the following replacements: approximately 90% for Val-76, approximately 60% for Thr-76, approximately 30% for Ser-76, approximately 20% for Ile-76, and 0% for Leu-76. In order to develop an understanding of the factors that determine whether or not an altered iso-1-cytochrome c will function, we undertook a theoretical analysis which led to the conclusion that the activity of the proteins was dependent on both short- and long-range interactions. Short-range interactions were estimated from studies on known protein structures which gave the likelihood that various amino acids would be found in a local backbone configuration similar to the native protein; long-range interactions with the rest of the molecule were analyzed by considering the size of the side chain. We believe this approach can be used to analyze a wide variety of mutant proteins.     

Word Decoding of Protein Amino Acid Sequences with Availability Analysis: A Linguistic Approach

The amino acid sequences of proteins determine their three-dimensional structures and functions. However, how sequence information is related to structures and functions is still enigmatic. In this study, we show that at least a part of the sequence information can be extracted by treating amino acid sequences of proteins as a collection of English words, based on a working hypothesis that amino acid sequences of proteins are composed of short constituent amino acid sequences (SCSs) or “words”. We first confirmed that the English language highly likely follows Zipf''s law, a special case of power law. We found that the rank-frequency plot of SCSs in proteins exhibits a similar distribution when low-rank tails are excluded. In comparison with natural English and “compressed” English without spaces between words, amino acid sequences of proteins show larger linear ranges and smaller exponents with heavier low-rank tails, demonstrating that the SCS distribution in proteins is largely scale-free. A distribution pattern of SCSs in proteins is similar among species, but species-specific features are also present. Based on the availability scores of SCSs, we found that sequence motifs are enriched in high-availability sites (i.e., “key words”) and vice versa. In fact, the highest availability peak within a given protein sequence often directly corresponds to a sequence motif. The amino acid composition of high-availability sites within motifs is different from that of entire motifs and all protein sequences, suggesting the possible functional importance of specific SCSs and their compositional amino acids within motifs. We anticipate that our availability-based word decoding approach is complementary to sequence alignment approaches in predicting functionally important sites of unknown proteins from their amino acid sequences.     

IIAGlc inhibition of glycerol kinase: a communications network tunes protein motions at the allosteric site

Steady-state and time-resolved fluorescence anisotropy methods applied to an extrinsic fluorophore that is conjugated to non-native cysteine residues demonstrate that amino acids in an allosteric communication network within a protein subunit tune protein backbone motions at a distal site to enable allosteric binding and inhibition. The unphosphorylated form of the phosphocarrier protein IIAGlc is an allosteric inhibitor of Escherichia coli glycerol kinase, binding more than 25 A from the kinase active site. Crystal structures that showed a ligand-dependent conformational change and large temperature factors for the IIAGlc-binding site on E. coli glycerol kinase suggest that motions of the allosteric site have an important role in the inhibition. Three E. coli glycerol kinase amino acids that are located at least 15 A from the active site and the allosteric site were shown previously to be necessary for transplanting IIAGlc inhibition into the nonallosteric glycerol kinase from Haemophilus influenzae. These three amino acids are termed the coupling locus. The apparent allosteric site motions and the requirement for the distant coupling locus to transplant allosteric inhibition suggest that the coupling locus modulates the motions of the IIAGlc-binding site. To evaluate this possibility, variants of E. coli glycerol kinase and the chimeric, allosteric H. influenzae glycerol kinase were constructed with a non-native cysteine residue replacing one of the native residues in the IIAGlc-binding site. The extrinsic fluorophore Oregon Green 488 (2',7'-difluorofluorescein) was conjugated specifically to the non-native cysteine residue. Steady-state and time-resolved fluorescence anisotropy measurements show that the motions of the fluorophore reflect backbone motions of the IIAGlc-binding site and these motions are modulated by the amino acids at the coupling locus.     

Automated scaffold selection for enzyme design

A major goal of computational protein design is the construction of novel functions on existing protein scaffolds. There the first question is which scaffold is suitable for a specific reaction. Given a set of catalytic residues and their spatial arrangement, one wants to identify a protein scaffold that can host this active site. Here, we present an algorithm called ScaffoldSelection that is able to rapidly search large sets of protein structures for potential attachment sites of an enzymatic motif. The method consists of two steps; it first identifies pairs of backbone positions in pocket‐like regions. Then, it combines these to complete attachment sites using a graph theoretical approach. Identified matches are assessed for their ability to accommodate the substrate or transition state. A representative set of structures from the Protein Data Bank (～3500) was searched for backbone geometries that support the catalytic residues for 12 chemical reactions. Recapitulation of native active site geometries is used as a benchmark for the performance of the program. The native motif is identified in all 12 test cases, ranking it in the top percentile in 5 out of 12. The algorithm is fast and efficient, although dependent on the complexity of the motif. Comparisons to other methods show that ScaffoldSelection performs equally well in terms of accuracy and far better in terms of speed. Thus, ScaffoldSelection will aid future computational protein design experiments by preselecting protein scaffolds that are suitable for a specific reaction type and the introduction of a predefined amino acid motif. Proteins 2009. © 2009 Wiley‐Liss, Inc.     

Methodology for protein-ligand binding studies: application to a model for drug resistance, the HIV/FIV protease system.

A protocol for the rapid energetic analysis of protein-ligand complexes has been developed. This protocol involves the generation of protein-ligand complex ensembles followed by an analysis of the binding free energy components. We apply this methodology toward understanding the origin of binding specificity within the human immunodeficiency virus/feline immunodeficiency virus (HIV/FIV) protease system, a model system for drug resistance studies. A distinct difference in the internal strain of an inhibitor within each protein environment clearly favors the HIV protease complex, as observed experimentally. Our analysis also predicts that residues within the S2-S3 pockets of the FIV protease active site are responsible for this strain. Close examination of the active site residue contributions to interaction energy and desolvation energy identifies specific amino acids that may also play a role in determining the binding preferences of these two enzymes. Proteins 1999;36:318-331.