AlphaFold predictions of fold-switched conformations are driven by structure memorization.

Recent work suggests that AlphaFold (AF)-a deep learning-based model that can accurately infer protein structure from sequence-may discern important features of folded protein energy landscapes, defined by the diversity and frequency of different conformations in the folded state. Here, we test the limits of its predictive power on fold-switching proteins, which assume two structures with regions of distinct secondary and/or tertiary structure. We find that (1) AF is a weak predictor of fold switching and (2) some of its successes result from memorization of training-set structures rather than learned protein energetics. Combining >280,000 models from several implementations of AF2 and AF3, a 35% success rate was achieved for fold switchers likely in AF's training sets. AF2's confidence metrics selected against models consistent with experimentally determined fold-switching structures and failed to discriminate between low and high energy conformations. Further, AF captured only one out of seven experimentally confirmed fold switchers outside of its training sets despite extensive sampling of an additional ~280,000 models. Several observations indicate that AF2 has memorized structural information during training, and AF3 misassigns coevolutionary restraints. These limitations constrain the scope of successful predictions, highlighting the need for physically based methods that readily predict multiple protein conformations.

Sequence clustering confounds AlphaFold2.

Though typically associated with a single folded state, some globular proteins remodel their secondary and/or tertiary structures in response to cellular stimuli. AlphaFold21 (AF2) readily generates one dominant protein structure for these fold-switching (a.k.a. metamorphic) proteins2, but it often fails to predict their alternative experimentally observed structures3,4. Wayment-Steele, et al. steered AF2 to predict alternative structures of a few metamorphic proteins using a method they call AF-cluster5. However, their Paper lacks some essential controls needed to assess AF-cluster's reliability. We find that these controls show AF-cluster to be a poor predictor of metamorphic proteins. First, closer examination of the Paper's results reveals that random sequence sampling outperforms sequence clustering, challenging the claim that AF-cluster works by "deconvolving conflicting sets of couplings." Further, we observe that AF-cluster mistakes some single-folding KaiB homologs for fold switchers, a critical flaw bound to mislead users. Finally, proper error analysis reveals that AF-cluster predicts many correct structures with low confidence and some experimentally unobserved conformations with confidences similar to experimentally observed ones. For these reasons, we suggest using ColabFold6-based random sequence sampling7-augmented by other predictive approaches-as a more accurate and less computationally intense alternative to AF-cluster.

ColabFold predicts alternative protein structures from single sequences, coevolution unnecessary for AF-cluster.

Though typically associated with a single folded state, globular proteins are dynamic and often assume alternative or transient structures important for their functions1,2. Wayment-Steele, et al. steered ColabFold3 to predict alternative structures of several proteins using a method they call AF-cluster4. They propose that AF-cluster "enables ColabFold to sample alternate states of known metamorphic proteins with high confidence" by first clustering multiple sequence alignments (MSAs) in a way that "deconvolves" coevolutionary information specific to different conformations and then using these clusters as input for ColabFold. Contrary to this Coevolution Assumption, clustered MSAs are not needed to make these predictions. Rather, these alternative structures can be predicted from single sequences and/or sequence similarity, indicating that coevolutionary information is unnecessary for predictive success and may not be used at all. These results suggest that AF-cluster's predictive scope is likely limited to sequences with distinct-yet-homologous structures within ColabFold's training set.

Identification of a covert evolutionary pathway between two protein folds.

Although homologous protein sequences are expected to adopt similar structures, some amino acid substitutions can interconvert α-helices and ß-sheets. Such fold switching may have occurred over evolutionary history, but supporting evidence has been limited by the: (1) abundance and diversity of sequenced genes, (2) quantity of experimentally determined protein structures, and (3) assumptions underlying the statistical methods used to infer homology. Here, we overcome these barriers by applying multiple statistical methods to a family of ~600,000 bacterial response regulator proteins. We find that their homologous DNA-binding subunits assume divergent structures: helix-turn-helix versus α-helix + ß-sheet (winged helix). Phylogenetic analyses, ancestral sequence reconstruction, and AlphaFold2 models indicate that amino acid substitutions facilitated a switch from helix-turn-helix into winged helix. This structural transformation likely expanded DNA-binding specificity. Our approach uncovers an evolutionary pathway between two protein folds and provides a methodology to identify secondary structure switching in other protein families.

Distinguishing features of fold-switching proteins.

Though many folded proteins assume one stable structure that performs one function, a small-but-increasing number remodel their secondary and tertiary structures and change their functions in response to cellular stimuli. These fold-switching proteins regulate biological processes and are associated with autoimmune dysfunction, severe acute respiratory syndrome coronavirus-2 infection, and more. Despite their biological importance, it is difficult to computationally predict fold switching. With the aim of advancing computational prediction and experimental characterization of fold switchers, this review discusses several features that distinguish fold-switching proteins from their single-fold and intrinsically disordered counterparts. First, the isolated structures of fold switchers are less stable and more heterogeneous than single folders but more stable and less heterogeneous than intrinsically disordered proteins (IDPs). Second, the sequences of single fold, fold switching, and intrinsically disordered proteins can evolve at distinct rates. Third, proteins from these three classes are best predicted using different computational techniques. Finally, late-breaking results suggest that single folders, fold switchers, and IDPs have distinct patterns of residue-residue coevolution. The review closes by discussing high-throughput and medium-throughput experimental approaches that might be used to identify new fold-switching proteins.

AlphaFold2 has more to learn about protein energy landscapes.

Recent work suggests that AlphaFold2 (AF2)-a deep learning-based model that can accurately infer protein structure from sequence-may discern important features of folded protein energy landscapes, defined by the diversity and frequency of different conformations in the folded state. Here, we test the limits of its predictive power on fold-switching proteins, which assume two structures with regions of distinct secondary and/or tertiary structure. Using several implementations of AF2, including two published enhanced sampling approaches, we generated >280,000 models of 93 fold-switching proteins whose experimentally determined conformations were likely in AF2's training set. Combining all models, AF2 predicted fold switching with a modest success rate of ~25%, indicating that it does not readily sample both experimentally characterized conformations of most fold switchers. Further, AF2's confidence metrics selected against models consistent with experimentally determined fold-switching conformations in favor of inconsistent models. Accordingly, these confidence metrics-though suggested to evaluate protein energetics reliably-did not discriminate between low and high energy states of fold-switching proteins. We then evaluated AF2's performance on seven fold-switching proteins outside of its training set, generating >159,000 models in total. Fold switching was accurately predicted in one of seven targets with moderate confidence. Further, AF2 demonstrated no ability to predict alternative conformations of two newly discovered targets without homologs in the set of 93 fold switchers. These results indicate that AF2 has more to learn about the underlying energetics of protein ensembles and highlight the need for further developments of methods that readily predict multiple protein conformations.

AlphaFold2 fails to predict protein fold switching.

AlphaFold2 has revolutionized protein structure prediction by leveraging sequence information to rapidly model protein folds with atomic-level accuracy. Nevertheless, previous work has shown that these predictions tend to be inaccurate for structurally heterogeneous proteins. To systematically assess factors that contribute to this inaccuracy, we tested AlphaFold2's performance on 98-fold-switching proteins, which assume at least two distinct-yet-stable secondary and tertiary structures. Topological similarities were quantified between five predicted and two experimentally determined structures of each fold-switching protein. Overall, 94% of AlphaFold2 predictions captured one experimentally determined conformation but not the other. Despite these biased results, AlphaFold2's estimated confidences were moderate-to-high for 74% of fold-switching residues, a result that contrasts with overall low confidences for intrinsically disordered proteins, which are also structurally heterogeneous. To investigate factors contributing to this disparity, we quantified sequence variation within the multiple sequence alignments used to generate AlphaFold2's predictions of fold-switching and intrinsically disordered proteins. Unlike intrinsically disordered regions, whose sequence alignments show low conservation, fold-switching regions had conservation rates statistically similar to canonical single-fold proteins. Furthermore, intrinsically disordered regions had systematically lower prediction confidences than either fold-switching or single-fold proteins, regardless of sequence conservation. AlphaFold2's high prediction confidences for fold switchers indicate that it uses sophisticated pattern recognition to search for one most probable conformer rather than protein biophysics to model a protein's structural ensemble. Thus, it is not surprising that its predictions often fail for proteins whose properties are not fully apparent from solved protein structures. Our results emphasize the need to look at protein structure as an ensemble and suggest that systematic examination of fold-switching sequences may reveal propensities for multiple stable secondary and tertiary structures.

Intrinsically disordered proteins/regions and insight into their biomolecular interactions.

Proteins may vary from being rigid to having flexible regions to being completely disordered, either as an intrinsically disordered protein (IDP) or having specific intrinsically disordered regions (IDRs). IDPs/IDRs can form complexes otherwise impossible, such as wrapping around the binding partner, hence providing the plasticity needed for achieving assemblies with specific functions. IDRs can exhibit promiscuity, using the same region in the sequence to bind multiple partners, and act as hubs in protein-protein interaction network (an essential part of the cell signalling network). Disorder-to-order transition on binding provides specificity with affinity, optimum for reversibility of the binding, thus offering suitability for regulation and signalling processes. IDRs interactions may be modulated by the environment or covalent modifications; mis-signalling or their unnatural or non-native folding may lead to diseases. This article aims to provide an overview of structural heterogeneity, as seen in IDPs/IDRs, and their role in biological recognition, binding and function.

Plausible blockers of Spike RBD in SARS-CoV2-molecular design and underlying interaction dynamics from high-level structural descriptors.

COVID-19 is characterized by an unprecedented abrupt increase in the viral transmission rate (SARS-CoV-2) relative to its pandemic evolutionary ancestor, SARS-CoV (2003). The complex molecular cascade of events related to the viral pathogenicity is triggered by the Spike protein upon interacting with the ACE2 receptor on human lung cells through its receptor binding domain (RBDSpike). One potential therapeutic strategy to combat COVID-19 could thus be limiting the infection by blocking this key interaction. In this current study, we adopt a protein design approach to predict and propose non-virulent structural mimics of the RBDSpike which can potentially serve as its competitive inhibitors in binding to ACE2. The RBDSpike is an independently foldable protein domain, resilient to conformational changes upon mutations and therefore an attractive target for strategic re-design. Interestingly, in spite of displaying an optimal shape fit between their interacting surfaces (attributed to a consequently high mutual affinity), the RBDSpike-ACE2 interaction appears to have a quasi-stable character due to a poor electrostatic match at their interface. Structural analyses of homologous protein complexes reveal that the ACE2 binding site of RBDSpike has an unusually high degree of solvent-exposed hydrophobic residues, attributed to key evolutionary changes, making it inherently "reaction-prone." The designed mimics aimed to block the viral entry by occupying the available binding sites on ACE2, are tested to have signatures of stable high-affinity binding with ACE2 (cross-validated by appropriate free energy estimates), overriding the native quasi-stable feature. The results show the apt of directly adapting natural examples in rational protein design, wherein, homology-based threading coupled with strategic "hydrophobic ↔ polar" mutations serve as a potential breakthrough.

Structural motifs in protein cores and at protein-protein interfaces are different.

Structures of proteins and protein-protein complexes are determined by the same physical principles and thus share a number of similarities. At the same time, there could be differences because in order to function, proteins interact with other molecules, undergo conformations changes, and so forth, which might impose different restraints on the tertiary versus quaternary structures. This study focuses on structural properties of protein-protein interfaces in comparison with the protein core, based on the wealth of currently available structural data and new structure-based approaches. The results showed that physicochemical characteristics, such as amino acid composition, residue-residue contact preferences, and hydrophilicity/hydrophobicity distributions, are similar in protein core and protein-protein interfaces. On the other hand, characteristics that reflect the evolutionary pressure, such as structural composition and packing, are largely different. The results provide important insight into fundamental properties of protein structure and function. At the same time, the results contribute to better understanding of the ways to dock proteins. Recent progress in predicting structures of individual proteins follows the advancement of deep learning techniques and new approaches to residue coevolution data. Protein core could potentially provide large amounts of data for application of the deep learning to docking. However, our results showed that the core motifs are significantly different from those at protein-protein interfaces, and thus may not be directly useful for docking. At the same time, such difference may help to overcome a major obstacle in application of the coevolutionary data to docking-discrimination of the intramolecular information not directly relevant to docking.

How to choose templates for modeling of protein complexes: Insights from benchmarking template-based docking.

Comparative docking is based on experimentally determined structures of protein-protein complexes (templates), following the paradigm that proteins with similar sequences and/or structures form similar complexes. Modeling utilizing structure similarity of target monomers to template complexes significantly expands structural coverage of the interactome. Template-based docking by structure alignment can be performed for the entire structures or by aligning targets to the bound interfaces of the experimentally determined complexes. Systematic benchmarking of docking protocols based on full and interface structure alignment showed that both protocols perform similarly, with top 1 docking success rate 26%. However, in terms of the models' quality, the interface-based docking performed marginally better. The interface-based docking is preferable when one would suspect a significant conformational change in the full protein structure upon binding, for example, a rearrangement of the domains in multidomain proteins. Importantly, if the same structure is selected as the top template by both full and interface alignment, the docking success rate increases 2-fold for both top 1 and top 10 predictions. Matching structural annotations of the target and template proteins for template detection, as a computationally less expensive alternative to structural alignment, did not improve the docking performance. Sophisticated remote sequence homology detection added templates to the pool of those identified by structure-based alignment, suggesting that for practical docking, the combination of the structure alignment protocols and the remote sequence homology detection may be useful in order to avoid potential flaws in generation of the structural templates library.

Structural changes in DNA-binding proteins on complexation.

Characterization and prediction of the DNA-biding regions in proteins are essential for our understanding of how proteins recognize/bind DNA. We analyze the unbound (U) and the bound (B) forms of proteins from the protein-DNA docking benchmark that contains 66 binary protein-DNA complexes along with their unbound counterparts. Proteins binding DNA undergo greater structural changes on complexation (in particular, those in the enzyme category) than those involved in protein-protein interactions (PPI). While interface atoms involved in PPI exhibit an increase in their solvent-accessible surface area (ASA) in the bound form in the majority of the cases compared to the unbound interface, protein-DNA interactions indicate increase and decrease in equal measure. In 25% structures, the U form has missing residues which are located in the interface in the B form. The missing atoms contribute more toward the buried surface area compared to other interface atoms. Lys, Gly and Arg are prominent in disordered segments that get ordered in the interface on complexation. In going from U to B, there may be an increase in coil and helical content at the expense of turns and strands. Consideration of flexibility cannot distinguish the interface residues from the surface residues in the U form.

Determining the roles of a conserved tyrosine residue in a Mip-like peptidyl-prolyl cis-trans isomerase.

The FKBP22 and the related peptidyl-prolyl cis-trans isomerases dimerize using their N-terminal domains. Conversely, their C-terminal domains possess both the substrate and inhibitor binding sites. To delineate the roles of a conserved Tyr residue at their N-terminal domains, we have studied a FKBP22 mutant that carries an Ala in place of the conserved Tyr at position 15. We have demonstrated that the Tyr 15 of FKBP22 is indispensable for preserving its dimerization ability, catalytic activity, and structure. The residue, however, little contributed to its inhibitor binding ability and stability. The mode of action of Tyr 15 has been discussed at length.

Changes in protein structure at the interface accompanying complex formation.

Protein interactions are essential in all biological processes. The changes brought about in the structure when a free component forms a complex with another molecule need to be characterized for a proper understanding of molecular recognition as well as for the successful implementation of docking algorithms. Here, unbound (U) and bound (B) forms of protein structures from the Protein-Protein Interaction Affinity Database are compared in order to enumerate the changes that occur at the interface atoms/residues in terms of the solvent-accessible surface area (ASA), secondary structure, temperature factors (B factors) and disorder-to-order transitions. It is found that the interface atoms optimize contacts with the atoms in the partner protein, which leads to an increase in their ASA in the bound interface in the majority (69%) of the proteins when compared with the unbound interface, and this is independent of the root-mean-square deviation between the U and B forms. Changes in secondary structure during the transition indicate a likely extension of helices and strands at the expense of turns and coils. A reduction in flexibility during complex formation is reflected in the decrease in B factors of the interface residues on going from the U form to the B form. There is, however, no distinction in flexibility between the interface and the surface in the monomeric structure, thereby highlighting the potential problem of using B factors for the prediction of binding sites in the unbound form for docking another protein. 16% of the proteins have missing (disordered) residues in the U form which are observed (ordered) in the B form, mostly with an irregular conformation; the data set also shows differences in the composition of interface and non-interface residues in the disordered polypeptide segments as well as differences in their surface burial.

Effects of Small Molecule Calcium-Activated Chloride Channel Inhibitors on Structure and Function of Accessory Cholera Enterotoxin (Ace) of Vibrio cholerae.

Cholera pathogenesis occurs due to synergistic pro-secretory effects of several toxins, such as cholera toxin (CTX) and Accessory cholera enterotoxin (Ace) secreted by Vibrio cholerae strains. Ace activates chloride channels stimulating chloride/bicarbonate transport that augments fluid secretion resulting in diarrhea. These channels have been targeted for drug development. However, lesser attention has been paid to the interaction of chloride channel modulators with bacterial toxins. Here we report the modulation of the structure/function of recombinant Ace by small molecule calcium-activated chloride channel (CaCC) inhibitors, namely CaCCinh-A01, digallic acid (DGA) and tannic acid. Biophysical studies indicate that the unfolding (induced by urea) free energy increases upon binding CaCCinh-A01 and DGA, compared to native Ace, whereas binding of tannic acid destabilizes the protein. Far-UV CD experiments revealed that the α-helical content of Ace-CaCCinh-A01 and Ace-DGA complexes increased relative to Ace. In contrast, binding to tannic acid had the opposite effect, indicating the loss of protein secondary structure. The modulation of Ace structure induced by CaCC inhibitors was also analyzed using docking and molecular dynamics (MD) simulation. Functional studies, performed using mouse ileal loops and Ussing chamber experiments, corroborate biophysical data, all pointing to the fact that tannic acid destabilizes Ace, inhibiting its function, whereas DGA stabilizes the toxin with enhanced fluid accumulation in mouse ileal loop. The efficacy of tannic acid in mouse model suggests that the targeted modulation of Ace structure may be of therapeutic benefit for gastrointestinal disorders.

Identification of the target DNA sequence and characterization of DNA binding features of HlyU, and suggestion of a redox switch for hlyA expression in the human pathogen Vibrio cholerae from in silico studies.

HlyU, a transcriptional regulator common in many Vibrio species, activates the hemolysin gene hlyA in Vibrio cholerae, the rtxA1 operon in Vibrio vulnificus and the genes of plp-vah1 and rtxACHBDE gene clusters in Vibrio anguillarum. The protein is also proposed to be a potential global virulence regulator for V. cholerae and V. vulnificus. Mechanisms of gene control by HlyU in V. vulnificus and V. anguillarum are reported. However, detailed elucidation of the interaction of HlyU in V. cholerae with its target DNA at the molecular level is not available. Here we report a 17-bp imperfect palindrome sequence, 5'-TAATTCAGACTAAATTA-3', 173 bp upstream of hlyA promoter, as the binding site of HlyU. This winged helix-turn-helix protein binds necessarily as a dimer with the recognition helices contacting the major grooves and the ß-sheet wings, the minor grooves. Such interactions enhance hlyA promoter activity in vivo. Mutations affecting dimerization as well as those in the DNA-protein interface hamper DNA binding and transcription regulation. Molecular dynamic simulations show hydrogen bonding patterns involving residues at the mutation sites and confirmed their importance in DNA binding. On binding to HlyU, DNA deviates by ∼68º from linearity. Dynamics also suggest a possible redox control in HlyU.

Actin-curcumin interaction: insights into the mechanism of actin polymerization inhibition.

Curcumin, derived from rhizomes of the Curcuma longa plant, is known to possess a wide range of medicinal properties. We have examined the interaction of curcumin with actin and determined their binding and thermodynamic parameters using isothermal titration calorimetry. Curcumin is weakly fluorescent in aqueous solution, and binding to actin enhances fluorescence several fold with a large blue shift in the emission maximum. Curcumin inhibits microfilament formation, which is similar to its role in inhibiting microtubule formation. We synthesized a series of stable curcumin analogues to examine their affinity for actin and their ability to inhibit actin self-assembly. Results show that curcumin is a ligand with two symmetrical halves, each of which possesses no activity individually. Oxazole, pyrazole, and acetyl derivatives are less effective than curcumin at inhibiting actin self-assembly, whereas a benzylidiene derivative is more effective. Cell biology studies suggest that disorganization of the actin network leads to destabilization of filaments in the presence of curcumin. Molecular docking reveals that curcumin binds close to the cytochalasin binding site of actin. Further molecular dynamics studies reveal a possible allosteric effect in which curcumin binding at the "barbed end" of actin is transmitted to the "pointed end", where conformational changes disrupt interactions with the adjacent actin monomer to interrupt filament formation. Finally, the recognition and binding of actin by curcumin is yet another example of its unique ability to target multiple receptors.

Flexibility in the N-terminal actin-binding domain: clues from in silico mutations and molecular dynamics.

Dystrophin is a long, rod-shaped cytoskeleton protein implicated in muscular dystrophy (MDys). Utrophin is the closest autosomal homolog of dystrophin. Both proteins have N-terminal actin-binding domain (N-ABD), a central rod domain and C-terminal region. N-ABD, composed of two calponin homology (CH) subdomains joined by a helical linker, harbors a few disease causing missense mutations. Although the two proteins share considerable homology (>72%) in N-ABD, recent structural and biochemical studies have shown that there are significant differences (including stability, mode of actin-binding) and their functions are not completely interchangeable. In this investigation, we have used extensive molecular dynamics simulations to understand the differences and the similarities of these two proteins, along with another actin-binding protein, fimbrin. In silico mutations were performed to identify two key residues that might be responsible for the dynamical difference between the molecules. Simulation points to the inherent flexibility of the linker region, which adapts different conformations in the wild type dystrophin. Mutations T220V and G130D in dystrophin constrain the flexibility of the central helical region, while in the two known disease-causing mutants, K18N and L54R, the helicity of the region is compromised. Phylogenetic tree and sequence analysis revealed that dystrophin and utrophin genes have probably originated from the same ancestor. The investigation would provide insight into the functional diversity of two closely related proteins and fimbrin, and contribute to our understanding of the mechanism of MDys.

Proline substitutions in a Mip-like peptidyl-prolyl cis-trans isomerase severely affect its structure, stability, shape and activity.

FKBP22, an Escherichia coli-specific peptidyl-prolyl cis-trans isomerase, shows substantial homology with the Mip-like virulence factors. Mip-like proteins are homodimeric and possess a V-shaped conformation. Their N-terminal domains form dimers, whereas their C-terminal domains bind protein/peptide substrates and distinct inhibitors such as rapamycin and FK506. Interestingly, the two domains of the Mip-like proteins are separated by a lengthy, protease-susceptible α-helix. To delineate the structural requirement of this domain-connecting region in Mip-like proteins, we have investigated a recombinant FKBP22 (rFKBP22) and its three point mutants I65P, V72P and A82P using different probes. Each mutant harbors a Pro substitution mutation at a distinct location in the hinge region. We report that the three mutants are not only different from each other but also different from rFKBP22 in structure and activity. Unlike rFKBP22, the three mutants were unfolded by a non-two state mechanism in the presence of urea. In addition, the stabilities of the mutants, particularly I65P and V72P, differed considerably from that of rFKBP22. Conversely, the rapamycin binding affinity of no mutant was different from that of rFKBP22. Of the mutants, I65P showed the highest levels of structural/functional loss and dissociated partly in solution. Our computational study indicated a severe collapse of the V-shape in I65P due to the anomalous movement of its C-terminal domains. The α-helical nature of the domain-connecting region is, therefore, critical for the Mip-like proteins.