A large scale test of computational protein design: folding and stability of nine completely redesigned globular proteins

A previously developed computer program for protein design, RosettaDesign, was used to predict low free energy sequences for nine naturally occurring protein backbones. RosettaDesign had no knowledge of the naturally occurring sequences and on average 65% of the residues in the designed sequences differ from wild-type. Synthetic genes for ten completely redesigned proteins were generated, and the proteins were expressed, purified, and then characterized using circular dichroism, chemical and temperature denaturation and NMR experiments. Although high-resolution structures have not yet been determined, eight of these proteins appear to be folded and their circular dichroism spectra are similar to those of their wild-type counterparts. Six of the proteins have stabilities equal to or up to 7kcal/mol greater than their wild-type counterparts, and four of the proteins have NMR spectra consistent with a well-packed, rigid structure. These encouraging results indicate that the computational protein design methods can, with significant reliability, identify amino acid sequences compatible with a target protein backbone.     

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.     

Contribution of the tyrosines to the structure and function of the human U1A N-terminal RNA binding domain.

RNA binding domains (RBDs) are members of a large family of proteins that share minimal sequence conservation but adopt an alpha beta sandwich global fold. Defining the contributions of specific amino acids to RBD structure and RNA binding is critical to understanding the functions of these proteins. In these experiments with the human U1A N-terminal RNA binding domain (RBD1), the contributions from each of its four tyrosines to protein structure, stability, and RNA binding were measured. Each tyrosine was substituted with phenylalanine and one other selected residue, and the resulting proteins were characterized by chemical denaturation to measure their unfolding free energy, by binding free energies to the wild-type RNA hairpin, and by 19F NMR to probe for structural changes. Features of the protein identified in these experiments include a possible tyrosine/lysine contact in an alpha-helix, which may be an example of an energetically favorable aromatic/amino side chain interaction. One long loop of the protein, which shows unusual 15N backbone and tyrosine side-chain dynamics, is implicated in protein:protein association. The diverse interactions of the four tyrosine residues in the organization of RBD1 illustrate how each member of this family of proteins will have unique molecular details that contribute to function.     

Relationships between protein structure and dynamics from a database of NMR-derived backbone order parameters

The amplitude of protein backbone NH group motions on a time-scale faster than molecular tumbling may be determined by analysis of (15)N NMR relaxation data according to the Lipari-Szabo model free formalism. An internet-accessible database has been compiled containing 1855 order parameters from 20 independent NMR relaxation studies on proteins whose three-dimensional structures are known. A series of statistical analyses has been performed to identify relationships between the structural features and backbone dynamics of these proteins. Comparison of average order parameters for different amino acid types indicates that amino acids with small side-chains tend to have greater backbone flexibility than those with large side-chains. In addition, the motions of a given NH group are also related to the sizes of the neighboring amino acids in the primary sequence. The secondary structural environment appears to influence backbone dynamics relatively weakly, with only subtle differences between the order parameter distributions of loop structures and regular hydrogen bonded secondary structure elements. However, NH groups near helix termini are more mobile on average than those in the central regions of helices. Tertiary structure influences are also relatively weak but in the expected direction, with more exposed residues being more flexible on average than residues that are relatively inaccessible to solvent.     

Accurate computer-based design of a new backbone conformation in the second turn of protein L.

The rational design of loops and turns is a key step towards creating proteins with new functions. We used a computational design procedure to create new backbone conformations in the second turn of protein L. The Protein Data Bank was searched for alternative turn conformations, and sequences optimal for these turns in the context of protein L were identified using a Monte Carlo search procedure and an energy function that favors close packing. Two variants containing 12 and 14 mutations were found to be as stable as wild-type protein L. The crystal structure of one of the variants has been solved at a resolution of 1.9 A, and the backbone conformation in the second turn is remarkably close to that of the in silico model (1.1 A RMSD) while it differs significantly from that of wild-type protein L (the turn residues are displaced by an average of 7.2 A). The folding rates of the redesigned proteins are greater than that of the wild-type protein and in contrast to wild-type protein L the second beta-turn appears to be formed at the rate limiting step in folding.     

Improving the accuracy of protein stability predictions with multistate design using a variety of backbone ensembles

Multistate computational protein design (MSD) with backbone ensembles approximating conformational flexibility can predict higher quality sequences than single‐state design with a single fixed backbone. However, it is currently unclear what characteristics of backbone ensembles are required for the accurate prediction of protein sequence stability. In this study, we aimed to improve the accuracy of protein stability predictions made with MSD by using a variety of backbone ensembles to recapitulate the experimentally measured stability of 85 Streptococcal protein G domain β1 sequences. Ensembles tested here include an NMR ensemble as well as those generated by molecular dynamics (MD) simulations, by Backrub motions, and by PertMin, a new method that we developed involving the perturbation of atomic coordinates followed by energy minimization. MSD with the PertMin ensembles resulted in the most accurate predictions by providing the highest number of stable sequences in the top 25, and by correctly binning sequences as stable or unstable with the highest success rate (≈90%) and the lowest number of false positives. The performance of PertMin ensembles is due to the fact that their members closely resemble the input crystal structure and have low potential energy. Conversely, the NMR ensemble as well as those generated by MD simulations at 500 or 1000 K reduced prediction accuracy due to their low structural similarity to the crystal structure. The ensembles tested herein thus represent on‐ or off‐target models of the native protein fold and could be used in future studies to design for desired properties other than stability. Proteins 2014; 82:771–784. © 2013 Wiley Periodicals, Inc.     

Backbone conformational flexibility of the lipid modified membrane anchor of the human N-Ras protein investigated by solid-state NMR and molecular dynamics simulation

The lipid modified human N-Ras protein, implicated in human cancer development, is of particular interest due to its membrane anchor that determines the activity and subcellular location of the protein. Previous solid-state NMR investigations indicated that this membrane anchor is highly dynamic, which may be indicative of backbone conformational flexibility. This article aims to address if a dynamic exchange between three structural models exist that had been determined previously. We applied a combination of solid-state nuclear magnetic resonance (NMR) methods and replica exchange molecular dynamics (MD) simulations using a Ras peptide that represents the terminal seven amino acids of the human N-Ras protein. Analysis of correlations between the conformations of individual amino acids revealed that Cys 181 and Met 182 undergo collective conformational exchange. Two major structures constituting about 60% of all conformations could be identified. The two conformations found in the simulation are in rapid exchange, which gives rise to low backbone order parameters and nuclear spin relaxation as measured by experimental NMR methods. These parameters were also determined from two 300 ns conventional MD simulations, providing very good agreement with the experimental data.     

High-resolution structural and thermodynamic analysis of extreme stabilization of human procarboxypeptidase by computational protein design

Recent efforts to design de novo or redesign the sequence and structure of proteins using computational techniques have met with significant success. Most, if not all, of these computational methodologies attempt to model atomic-level interactions, and hence high-resolution structural characterization of the designed proteins is critical for evaluating the atomic-level accuracy of the underlying design force-fields. We previously used our computational protein design protocol RosettaDesign to completely redesign the sequence of the activation domain of human procarboxypeptidase A2. With 68% of the wild-type sequence changed, the designed protein, AYEdesign, is over 10 kcal/mol more stable than the wild-type protein. Here, we describe the high-resolution crystal structure and solution NMR structure of AYEdesign, which show that the experimentally determined backbone and side-chains conformations are effectively superimposable with the computational model at atomic resolution. To isolate the origins of the remarkable stabilization, we have designed and characterized a new series of procarboxypeptidase mutants that gain significant thermodynamic stability with a minimal number of mutations; one mutant gains more than 5 kcal/mol of stability over the wild-type protein with only four amino acid changes. We explore the relationship between force-field smoothing and conformational sampling by comparing the experimentally determined free energies of the overall design and these focused subsets of mutations to those predicted using modified force-fields, and both fixed and flexible backbone sampling protocols.     

Solution structure and dynamics of melanoma inhibitory activity protein

Melanoma inhibitory activity (MIA) is a small secreted protein that is implicated in cartilage cell maintenance and melanoma metastasis. It is representative of a recently discovered family of proteins that contain a Src Homologous 3 (SH3) subdomain. While SH3 domains are normally found in intracellular proteins and mediate protein-protein interactions via recognition of polyproline helices, MIA is single-domain extracellular protein, and it probably binds to a different class of ligands.Here we report the assignments, solution structure, and dynamics of human MIA determined by heteronuclear NMR methods. The structures were calculated in a semi-automated manner without manual assignment of NOE crosspeaks, and have a backbone rmsd of 0.38 Å over the ordered regions of the protein. The structure consists of an SH3-like subdomain with N- and C-terminal extensions of approximately 20 amino acids each that together form a novel fold. The rmsd between the solution structure and our recently reported crystal structure is 0.86 Å over the ordered regions of the backbone, and the main differences are localized to the most dynamic regions of the protein. The similarity between the NMR and crystal structures supports the use of automated NOE assignments and ambiguous restraints to accelerate the calculation of NMR structures.     

Altering the RNA-binding mode of the U1A RBD1 protein

The N-terminal RNA-binding domain (RBD1) of the human U1A protein is evolutionarily designed to bind its RNA targets with great affinity and specificity. The physical mechanisms that modulate the coupling (local cooperativity) among amino acid residues on the extensive binding surface of RBD1 are investigated here, using mutants that replace a highly conserved glycine residue. This glycine residue, at the strand/loop junction of beta3/loop3, is found in U1A RBD1, and in most RBD domains, suggesting it has a specific role in modulation of RNA binding. Here, two RBD1 proteins are constructed in which that residue (Gly53) is replaced by either alanine or valine. These new proteins are shown by NMR methods and molecular dynamics simulations to be very similar to the wild-type RBD1, both in structure and in their backbone dynamics. However, RNA-binding assays show that affinity for the U1 snRNA stem-loop II RNA target is reduced by nearly 200-fold for the RBD1-G53A protein, and by 1.6 x 10(4)-fold for RBD1-G53V. The mode of RNA binding by RBD1-G53A is similar to that of RBD1-WT, displaying its characteristic non-additive free energies of base recognition and its salt-dependence. The binding mode of RBD1-G53V is altered, having lost its salt-dependence and displaying site-independence of base recognition. The molecular basis for this alteration in RNA-binding properties is proposed to result from the inability of the RNA to induce a change in the structure of the free protein to produce a high-affinity complex.     

Backbone dynamics of the N-terminal domain of the HIV-1 capsid protein and comparison with the G94D mutant conferring cyclosporin resistance/dependence.

Nuclear magnetic resonance (NMR) (15)N relaxation methods have been used to characterize the backbone dynamics of the N-terminal core domain of the HIV-1 capsid protein (CA(151)). The domain, which has an unusually flat, triangular shape, tumbles in solution at 28 degrees C with an effective rotational correlation time of 11.5 ns. Relaxation data for backbone amides in the domain's seven alpha-helices are indicative of fully anisotropic rotational diffusion. The principal axes of the rotational diffusion tensor calculated from the NMR data are aligned to within 12-23 degrees of the principal axes of the inertial tensor, with the axis of fastest rotational diffusion coincident with that of minimal inertia, and vice versa. Large variations in the (15)N-(1)H nuclear Overhauser effects for individual amino acids correlate with the degree of convergence in the previously calculated NMR structure. In particular, the partially disordered residues Val86-Arg97 that contain the human cyclophilin A (CypA) packaging signal have (15)N heteronuclear NOEs and transversal relaxation rates consistent with a high degree of dynamic conformational averaging. The N-terminal domain of a CA mutant (G94D) that confers both resistance to and dependence on cyclosporin A analogues was also analyzed. Our results indicate that this mutation does not influence the conformation or dynamics of CA(151), and therefore probably affects the function of the protein by modifying essential intermolecular CA-CA interactions.     

De novo backbone scaffolds for protein design

In recent years, there have been significant advances in the field of computational protein design including the successful computational design of enzymes based on backbone scaffolds from experimentally solved structures. It is likely that large‐scale sampling of protein backbone conformations will become necessary as further progress is made on more complicated systems. Removing the constraint of having to use scaffolds based on known protein backbones is a potential method of solving the problem. With this application in mind, we describe a method to systematically construct a large number of de novo backbone structures from idealized topological forms in a top–down hierarchical approach. The structural properties of these novel backbone scaffolds were analyzed and compared with a set of high‐resolution experimental structures from the protein data bank (PDB). It was found that the Ramachandran plot distribution and relative γ‐ and β‐turn frequencies were similar to those found in the PDB. The de novo scaffolds were sequence designed with RosettaDesign, and the energy distributions and amino acid compositions were comparable with the results for redesigned experimentally solved backbones. Proteins 2010. © 2009 Wiley‐Liss, Inc.     

Repacking the Core of T4 lysozyme by automated design

Automated protein redesign, as implemented in the program ORBIT, was used to redesign the core of phage T4 lysozyme. A total of 26 buried or partially buried sites in the C-terminal domain were allowed to vary both their sequence and side-chain conformation while the backbone and non-selected side-chains remained fixed. A variant with seven substitutions ("Core-7") was identified as having the most favorable energy. The redesign experiment was repeated with a penalty for the presence of methionine residues. In this case the redesigned protein ("Core-10") had ten amino acid changes. The two designed proteins, as well as the constituent single mutants, and several single-site revertants were over-expressed in Escherichia coli, purified, and subjected to crystallographic and thermal analyses. The thermodynamic and structural data show that some repacking was achieved although neither redesigned protein was more stable than the wild-type protein. The use of the methionine penalty was shown to be effective. Several of the side-chain rotamers in the predicted structure of Core-10 differ from those observed. Rather than changing to new rotamers predicted by the design process, side-chains tend to maintain conformations similar to those seen in the native molecule. In contrast, parts of the backbone change by up to 2.8A relative to both the designed structure and wild-type.Water molecules that are present within the lysozyme molecule were removed during the design process. In the redesigned protein the resultant cavities were, to some degree, re-occupied by side-chain atoms. In the observed structure, however, water molecules were still bound at or near their original sites. This suggests that it may be preferable to leave such water molecules in place during the design procedure. The results emphasize the specificity of the packing that occurs within the core of a typical protein. While point substitutions within the core are tolerated they almost always result in a loss of stability. Likewise, combinations of substitutions may also be tolerated but usually destabilize the protein. Experience with T4 lysozyme suggests that a general core repacking methodology with retention or enhancement of stability may be difficult to achieve without provision for shifts in the backbone.     

1H, 15N, and 13C NMR signal assignments of IIIGlc, a signal-transducing protein of Escherichia coli, using three-dimensional triple-resonance techniques

IIIGlc is an 18.1-kDa signal-transducing phosphocarrier protein of the phosphoenolpyruvate:glycose phosphotransferase system (PTS) of Escherichia coli. Virtually complete (98%) backbone 1H, 15N, and 13C nuclear magnetic resonance (NMR) signal assignments were determined by using a battery of triple-resonance three-dimensional (3D) NMR pulse sequences. In addition, nearly complete (1H, 95%; 13C, 85%) side-chain 1H and 13C signal assignments were obtained from an analysis of 3D 13C HCCH-COSY and HCCH-TOCSY spectra. These experiments rely almost exclusively upon one- and two-bond J couplings to transfer magnetization and to correlate proton and heteronuclear NMR signals. Hence, essentially complete signal assignments of this 168-residue protein were made without any assumptions regarding secondary structure and without the aid of a crystal structure, which is not yet available. Moreover, only three samples, one uniformly 15N-enriched, one uniformly 15N/13C-enriched, and one containing a few types of amino acids labeled with 15N and/or 13C, were needed to make the assignments. The backbone assignments together with the 3D 15N NOESY-HMQC and 13C NOESY-HMQC data have provided extensive information about the secondary structure of this protein [Pelton, J.G., Torchia, D.A., Meadow, N.D., Wong, C.-Y., & Roseman, S (1991) Proc. Natl. Acad. Sci. U.S.A. 88, 3479-3488]. The nearly complete set of backbone and side-chain atom assignments reported herein provide a basis for studies of the three-dimensional structure and dynamics of IIIGlc as well as its interactions with a variety of membrane and cytoplasmic proteins.     

Computational Design of a PAK1 Binding Protein

We describe a computational protocol, called DDMI, for redesigning scaffold proteins to bind to a specified region on a target protein. The DDMI protocol is implemented within the Rosetta molecular modeling program and uses rigid-body docking, sequence design, and gradient-based minimization of backbone and side-chain torsion angles to design low-energy interfaces between the scaffold and target protein. Iterative rounds of sequence design and conformational optimization were needed to produce models that have calculated binding energies that are similar to binding energies calculated for native complexes. We also show that additional conformation sampling with molecular dynamics can be iterated with sequence design to further lower the computed energy of the designed complexes. To experimentally test the DDMI protocol, we redesigned the human hyperplastic discs protein to bind to the kinase domain of p21-activated kinase 1 (PAK1). Six designs were experimentally characterized. Two of the designs aggregated and were not characterized further. Of the remaining four designs, three bound to the PAK1 with affinities tighter than 350 μM. The tightest binding design, named Spider Roll, bound with an affinity of 100 μM. NMR-based structure prediction of Spider Roll based on backbone and ¹³C^β chemical shifts using the program CS-ROSETTA indicated that the architecture of human hyperplastic discs protein is preserved. Mutagenesis studies confirmed that Spider Roll binds the target patch on PAK1. Additionally, Spider Roll binds to full-length PAK1 in its activated state but does not bind PAK1 when it forms an auto-inhibited conformation that blocks the Spider Roll target site. Subsequent NMR characterization of the binding of Spider Roll to PAK1 revealed a comparably small binding ‘on-rate’ constant (? 10⁵ M^− 1 s^− 1). The ability to rationally design the site of novel protein-protein interactions is an important step towards creating new proteins that are useful as therapeutics or molecular probes.     

Effects of the N-linked glycans on the 3D structure of the free alpha-subunit of human chorionic gonadotropin

To gain insight into intramolecular carbohydrate-protein interactions at the molecular level, the solution structure of differently deglycosylated variants of the alpha-subunit of human chorionic gonadotropin have been studied by NMR spectroscopy. Significant differences in chemical shifts and NOE intensities were observed for amino acid residues close to the carbohydrate chain at Asn78 upon deglycosylation beyond Asn78-bound GlcNAc. As no straightforward strategy is available for the calculation of the NMR structure of intact glycoproteins, a suitable computational protocol had to be developed. To this end, the X-PLOR carbohydrate force field designed for structure refinement was extended and modified. Furthermore, a computational strategy was devised to facilitate successful protein folding in the presence of extended glycans during the simulation. The values for phi and psi dihedral angles of the glycosidic linkages of the oligosaccharide core fragments GlcNAc2(beta1-4)GlcNAc1 and Man3(beta1-4)GlcNAc2 are restricted to a limited range of the broad conformational energy minima accessible for free glycans. This demonstrates that the protein core affects the dynamic behavior of the glycan at Asn78 by steric hindrance. Reciprocally, the NMR structures indicate that the glycan at Asn78 affects the stability of the protein core. The backbone angular order parameters and displacement data of the generated conformers display especially for the beta-turn 20-23 a decreased structural order upon splitting off the glycan beyond the Asn78-bound GlcNAc. In particular, the Asn-bound GlcNAc shields the protein surface from the hydrophilic environment through interaction with predominantly hydrophobic amino acid residues located in both twisted beta-hairpins consisting of residues 10-28 and 59-84.     

Design and folding of a multidomain protein

To test whether the folding process of a large protein can be understood on the basis of the folding behavior of the domains that constitute it, we coupled two well-studied small -helical proteins, the B-domain of protein A (60 amino acids) and Rd-apocytochrome b562 (Rd-apocyt b562, 106 amino acids), by fusing the C-terminal helix of the B-domain of protein A with the N-terminal helix of Rd-apocyt b562 without changing their hydrophobic core residues. The success of the design was confirmed by determining the structure of the engineered protein with multidimensional NMR methods. Kinetic studies showed that the logarithms of the folding/unfolding rate constants of the engineered protein are linearly dependent on concentrations of guanidinium chloride in the measurable range from 1.7 to 4 M. Their slopes (m-values) are close to those of Rd-apocyt b562. In addition, the 1H-15N HSQC spectrum taken at 1.5 M guanidinium chloride reveals that only the Rd-apocyt b562 domain in the designed protein remained folded. These results suggest that the two domains have weak energetic coupling. Interestingly, the redesigned protein folds faster than Rd-apocyt b562, suggesting that the fused helix stabilizes the rate-limiting transition state.