AlphaFold predictions are valuable hypotheses and accelerate but do not replace experimental structure determination.

Artificial intelligence-based protein structure prediction methods such as AlphaFold have revolutionized structural biology. The accuracies of these predictions vary, however, and they do not take into account ligands, covalent modifications or other environmental factors. Here, we evaluate how well AlphaFold predictions can be expected to describe the structure of a protein by comparing predictions directly with experimental crystallographic maps. In many cases, AlphaFold predictions matched experimental maps remarkably closely. In other cases, even very high-confidence predictions differed from experimental maps on a global scale through distortion and domain orientation, and on a local scale in backbone and side-chain conformation. We suggest considering AlphaFold predictions as exceptionally useful hypotheses. We further suggest that it is important to consider the confidence in prediction when interpreting AlphaFold predictions and to carry out experimental structure determination to verify structural details, particularly those that involve interactions not included in the prediction.

A large-scale conformational change couples membrane recruitment to cargo binding in the AP2 clathrin adaptor complex.

The AP2 adaptor complex (alpha, beta2, sigma2, and mu2 subunits) crosslinks the endocytic clathrin scaffold to PtdIns4,5P(2)-containing membranes and transmembrane protein cargo. In the "locked" cytosolic form, AP2's binding sites for the two endocytic motifs, YxxPhi on the C-terminal domain of mu2 (C-mu2) and [ED]xxxL[LI] on sigma2, are blocked by parts of beta2. Using protein crystallography, we show that AP2 undergoes a large conformational change in which C-mu2 relocates to an orthogonal face of the complex, simultaneously unblocking both cargo-binding sites; the previously unstructured mu2 linker becomes helical and binds back onto the complex. This structural rearrangement results in AP2's four PtdIns4,5P(2)- and two endocytic motif-binding sites becoming coplanar, facilitating their simultaneous interaction with PtdIns4,5P(2)/cargo-containing membranes. Using a range of biophysical techniques, we show that the endocytic cargo binding of AP2 is driven by its interaction with PtdIns4,5P(2)-containing membranes.

Assessing the utility of CASP14 models for molecular replacement.

The assessment of CASP models for utility in molecular replacement is a measure of their use in a valuable real-world application. In CASP7, the metric for molecular replacement assessment involved full likelihood-based molecular replacement searches; however, this restricted the assessable targets to crystal structures with only one copy of the target in the asymmetric unit, and to those where the search found the correct pose. In CASP10, full molecular replacement searches were replaced by likelihood-based rigid-body refinement of models superimposed on the target using the LGA algorithm, with the metric being the refined log-likelihood-gain (LLG) score. This enabled multi-copy targets and very poor models to be evaluated, but a significant further issue remained: the requirement of diffraction data for assessment. We introduce here the relative-expected-LLG (reLLG), which is independent of diffraction data. This reLLG is also independent of any crystal form, and can be calculated regardless of the source of the target, be it X-ray, NMR or cryo-EM. We calibrate the reLLG against the LLG for targets in CASP14, showing that it is a robust measure of both model and group ranking. Like the LLG, the reLLG shows that accurate coordinate error estimates add substantial value to predicted models. We find that refinement by CASP groups can often convert an inadequate initial model into a successful MR search model. Consistent with findings from others, we show that the AlphaFold2 models are sufficiently good, and reliably so, to surpass other current model generation strategies for attempting molecular replacement phasing.

Ab initio solution of macromolecular crystal structures without direct methods.

The majority of macromolecular crystal structures are determined using the method of molecular replacement, in which known related structures are rotated and translated to provide an initial atomic model for the new structure. A theoretical understanding of the signal-to-noise ratio in likelihood-based molecular replacement searches has been developed to account for the influence of model quality and completeness, as well as the resolution of the diffraction data. Here we show that, contrary to current belief, molecular replacement need not be restricted to the use of models comprising a substantial fraction of the unknown structure. Instead, likelihood-based methods allow a continuum of applications depending predictably on the quality of the model and the resolution of the data. Unexpectedly, our understanding of the signal-to-noise ratio in molecular replacement leads to the finding that, with data to sufficiently high resolution, fragments as small as single atoms of elements usually found in proteins can yield ab initio solutions of macromolecular structures, including some that elude traditional direct methods.

Structure and evolution of ENTH and VHS/ENTH-like domains in tepsin.

Tepsin is currently the only accessory trafficking protein identified in adaptor-related protein 4 (AP4)-coated vesicles originating at the trans-Golgi network (TGN). The molecular basis for interactions between AP4 subunits and motifs in the tepsin C-terminus have been characterized, but the biological role of tepsin remains unknown. We determined X-ray crystal structures of the tepsin epsin N-terminal homology (ENTH) and VHS/ENTH-like domains. Our data reveal unexpected structural features that suggest key functional differences between these and similar domains in other trafficking proteins. The tepsin ENTH domain lacks helix0, helix8 and a lipid binding pocket found in epsin1/2/3. These results explain why tepsin requires AP4 for its membrane recruitment and further suggest ENTH domains cannot be defined solely as lipid binding modules. The VHS domain lacks helix8 and thus contains fewer helices than other VHS domains. Structural data explain biochemical and biophysical evidence that tepsin VHS does not mediate known VHS functions, including recognition of dileucine-based cargo motifs or ubiquitin. Structural comparisons indicate the domains are very similar to each other, and phylogenetic analysis reveals their evolutionary pattern within the domain superfamily. Phylogenetics and comparative genomics further show tepsin within a monophyletic clade that diverged away from epsins early in evolutionary history (~1500 million years ago). Together, these data provide the first detailed molecular view of tepsin and suggest tepsin structure and function diverged away from other epsins. More broadly, these data highlight the challenges inherent in classifying and understanding protein function based only on sequence and structure.

Macromolecular X-ray structure determination using weak, single-wavelength anomalous data.

We describe a likelihood-based method for determining the substructure of anomalously scattering atoms in macromolecular crystals that allows successful structure determination by single-wavelength anomalous diffraction (SAD) X-ray analysis with weak anomalous signal. With the use of partial models and electron density maps in searches for anomalously scattering atoms, testing of alternative values of parameters and parallelized automated model-building, this method has the potential to extend the applicability of the SAD method in challenging cases.

Structures of yeast mitochondrial ADP/ATP carriers support a domain-based alternating-access transport mechanism.

The mitochondrial ADP/ATP carrier imports ADP from the cytosol and exports ATP from the mitochondrial matrix. The carrier cycles by an unresolved mechanism between the cytoplasmic state, in which the carrier accepts ADP from the cytoplasm, and the matrix state, in which it accepts ATP from the mitochondrial matrix. Here we present the structures of the yeast ADP/ATP carriers Aac2p and Aac3p in the cytoplasmic state. The carriers have three domains and are closed at the matrix side by three interdomain salt-bridge interactions, one of which is braced by a glutamine residue. Glutamine braces are conserved in mitochondrial carriers and contribute to an energy barrier, preventing the conversion to the matrix state unless substrate binding occurs. At the cytoplasmic side a second salt-bridge network forms during the transport cycle, as demonstrated by functional analysis of mutants with charge-reversed networks. Analyses of the domain structures and properties of the interdomain interfaces indicate that interconversion between states involves movement of the even-numbered α-helices across the surfaces of the odd-numbered α-helices by rotation of the domains. The odd-numbered α-helices have an L-shape, with proline or serine residues at the kinks, which functions as a lever-arm, coupling the substrate-induced disruption of the matrix network to the formation of the cytoplasmic network. The simultaneous movement of three domains around a central translocation pathway constitutes a unique mechanism among transport proteins. These findings provide a structural description of transport by mitochondrial carrier proteins, consistent with an alternating-access mechanism.

ANS complex of St John's wort PR-10 protein with 28 copies in the asymmetric unit: a fiendish combination of pseudosymmetry with tetartohedral twinning.

Hyp-1, a pathogenesis-related class 10 (PR-10) protein from St John's wort (Hypericum perforatum), was crystallized in complex with the fluorescent probe 8-anilino-1-naphthalene sulfonate (ANS). The highly pseudosymmetric crystal has 28 unique protein molecules arranged in columns with sevenfold translational noncrystallographic symmetry (tNCS) along c and modulated X-ray diffraction with intensity crests at l = 7n and l = 7n ± 3. The translational NCS is combined with pseudotetragonal rotational NCS. The crystal was a perfect tetartohedral twin, although detection of twinning was severely hindered by the pseudosymmetry. The structure determined at 2.4 Šresolution reveals that the Hyp-1 molecules (packed as ß-sheet dimers) have three novel ligand-binding sites (two internal and one in a surface pocket), which was confirmed by solution studies. In addition to 60 Hyp-1-docked ligands, there are 29 interstitial ANS molecules distributed in a pattern that violates the arrangement of the protein molecules and is likely to be the generator of the structural modulation. In particular, whenever the stacked Hyp-1 molecules are found closer together there is an ANS molecule bridging them.

Likelihood-based molecular-replacement solution for a highly pathological crystal with tetartohedral twinning and sevenfold translational noncrystallographic symmetry.

Translational noncrystallographic symmetry (tNCS) is a pathology of protein crystals in which multiple copies of a molecule or assembly are found in similar orientations. Structure solution is problematic because this breaks the assumptions used in current likelihood-based methods. To cope with such cases, new likelihood approaches have been developed and implemented in Phaser to account for the statistical effects of tNCS in molecular replacement. Using these new approaches, it was possible to solve the crystal structure of a protein exhibiting an extreme form of this pathology with seven tetrameric assemblies arrayed along the c axis. To resolve space-group ambiguities caused by tetartohedral twinning, the structure was initially solved by placing 56 copies of the monomer in space group P1 and using the symmetry of the solution to define the true space group, C2. The resulting structure of Hyp-1, a pathogenesis-related class 10 (PR-10) protein from the medicinal herb St John's wort, reveals the binding modes of the fluorescent probe 8-anilino-1-naphthalene sulfonate (ANS), providing insight into the function of the protein in binding or storing hydrophobic ligands.

Automated identification of elemental ions in macromolecular crystal structures.

Many macromolecular model-building and refinement programs can automatically place solvent atoms in electron density at moderate-to-high resolution. This process frequently builds water molecules in place of elemental ions, the identification of which must be performed manually. The solvent-picking algorithms in phenix.refine have been extended to build common ions based on an analysis of the chemical environment as well as physical properties such as occupancy, B factor and anomalous scattering. The method is most effective for heavier elements such as calcium and zinc, for which a majority of sites can be placed with few false positives in a diverse test set of structures. At atomic resolution, it is observed that it can also be possible to identify tightly bound sodium and magnesium ions. A number of challenges that contribute to the difficulty of completely automating the process of structure completion are discussed.

Automating crystallographic structure solution and refinement of protein-ligand complexes.

High-throughput drug-discovery and mechanistic studies often require the determination of multiple related crystal structures that only differ in the bound ligands, point mutations in the protein sequence and minor conformational changes. If performed manually, solution and refinement requires extensive repetition of the same tasks for each structure. To accelerate this process and minimize manual effort, a pipeline encompassing all stages of ligand building and refinement, starting from integrated and scaled diffraction intensities, has been implemented in Phenix. The resulting system is able to successfully solve and refine large collections of structures in parallel without extensive user intervention prior to the final stages of model completion and validation.

Intensity statistics in the presence of translational noncrystallographic symmetry.

In the case of translational noncrystallographic symmetry (tNCS), two or more copies of a component in the asymmetric unit of the crystal are present in a similar orientation. This causes systematic modulations of the reflection intensities in the diffraction pattern, leading to problems with structure determination and refinement methods that assume, either implicitly or explicitly, that the distribution of intensities is a function only of resolution. To characterize the statistical effects of tNCS accurately, it is necessary to determine the translation relating the copies, any small rotational differences in their orientations, and the size of random coordinate differences caused by conformational differences. An algorithm to estimate these parameters and refine their values against a likelihood function is presented, and it is shown that by accounting for the statistical effects of tNCS it is possible to unmask the competing statistical effects of twinning and tNCS and to more robustly assess the crystal for the presence of twinning.

SCEDS: protein fragments for molecular replacement in Phaser.

A method is described for generating protein fragments suitable for use as molecular-replacement (MR) template models. The template model for a protein suspected to undergo a conformational change is perturbed along combinations of low-frequency normal modes of the elastic network model. The unperturbed structure is then compared with each perturbed structure in turn and the structurally invariant regions are identified by analysing the difference distance matrix. These fragments are scored with SCEDS, which is a combined measure of the sphericity of the fragments, the continuity of the fragments with respect to the polypeptide chain, the equality in number of atoms in the fragments and the density of C(α) atoms in the triaxial ellipsoid of the fragment extents. The fragment divisions with the highest SCEDS are then used as separate template models for MR. Test cases show that where the protein contains fragments that undergo a change in juxtaposition between template model and target, SCEDS can identify fragments that lead to a lower R factor after ten cycles of all-atom refinement with REFMAC5 than the original template structure. The method has been implemented in the software Phaser.

Improved estimates of coordinate error for molecular replacement.

The estimate of the root-mean-square deviation (r.m.s.d.) in coordinates between the model and the target is an essential parameter for calibrating likelihood functions for molecular replacement (MR). Good estimates of the r.m.s.d. lead to good estimates of the variance term in the likelihood functions, which increases signal to noise and hence success rates in the MR search. Phaser has hitherto used an estimate of the r.m.s.d. that only depends on the sequence identity between the model and target and which was not optimized for the MR likelihood functions. Variance-refinement functionality was added to Phaser to enable determination of the effective r.m.s.d. that optimized the log-likelihood gain (LLG) for a correct MR solution. Variance refinement was subsequently performed on a database of over 21,000 MR problems that sampled a range of sequence identities, protein sizes and protein fold classes. Success was monitored using the translation-function Z-score (TFZ), where a TFZ of 8 or over for the top peak was found to be a reliable indicator that MR had succeeded for these cases with one molecule in the asymmetric unit. Good estimates of the r.m.s.d. are correlated with the sequence identity and the protein size. A new estimate of the r.m.s.d. that uses these two parameters in a function optimized to fit the mean of the refined variance is implemented in Phaser and improves MR outcomes. Perturbing the initial estimate of the r.m.s.d. from the mean of the distribution in steps of standard deviations of the distribution further increases MR success rates.

Phaser.MRage: automated molecular replacement.

Phaser.MRage is a molecular-replacement automation framework that implements a full model-generation workflow and provides several layers of model exploration to the user. It is designed to handle a large number of models and can distribute calculations efficiently onto parallel hardware. In addition, phaser.MRage can identify correct solutions and use this information to accelerate the search. Firstly, it can quickly score all alternative models of a component once a correct solution has been found. Secondly, it can perform extensive analysis of identified solutions to find protein assemblies and can employ assembled models for subsequent searches. Thirdly, it is able to use a priori assembly information (derived from, for example, homologues) to speculatively place and score molecules, thereby customizing the search procedure to a certain class of protein molecule (for example, antibodies) and incorporating additional biological information into molecular replacement.

High-resolution structure prediction and the crystallographic phase problem.

The energy-based refinement of low-resolution protein structure models to atomic-level accuracy is a major challenge for computational structural biology. Here we describe a new approach to refining protein structure models that focuses sampling in regions most likely to contain errors while allowing the whole structure to relax in a physically realistic all-atom force field. In applications to models produced using nuclear magnetic resonance data and to comparative models based on distant structural homologues, the method can significantly improve the accuracy of the structures in terms of both the backbone conformations and the placement of core side chains. Furthermore, the resulting models satisfy a particularly stringent test: they provide significantly better solutions to the X-ray crystallographic phase problem in molecular replacement trials. Finally, we show that all-atom refinement can produce de novo protein structure predictions that reach the high accuracy required for molecular replacement without any experimental phase information and in the absence of templates suitable for molecular replacement from the Protein Data Bank. These results suggest that the combination of high-resolution structure prediction with state-of-the-art phasing tools may be unexpectedly powerful in phasing crystallographic data for which molecular replacement is hindered by the absence of sufficiently accurate previous models.

A SNARE-adaptor interaction is a new mode of cargo recognition in clathrin-coated vesicles.

Soluble NSF attachment protein receptors (SNAREs) are type II transmembrane proteins that have critical roles in providing the specificity and energy for transport-vesicle fusion and must therefore be correctly partitioned between vesicle and organelle membranes. Like all other cargo, SNAREs need to be sorted into the forming vesicles by direct interaction with components of the vesicles' coats. Here we characterize the molecular details governing the sorting of a SNARE into clathrin-coated vesicles, namely the direct recognition of the three-helical bundle H(abc) domain of the mouse SNARE Vti1b by the human clathrin adaptor epsinR (EPNR, also known as CLINT1). Structures of each domain and of their complex show that this interaction (dissociation constant 22 muM) is mediated by surface patches composed of approximately 15 residues each, the topographies of which are dependent on each domain's overall fold. Disruption of the interface with point mutations abolishes the interaction in vitro and causes Vti1b to become relocalized to late endosomes and lysosomes. This new class of highly specific, surface-surface interaction between the clathrin coat component and the cargo is distinct from the widely observed binding of short, linear cargo motifs by the assembly polypeptide (AP) complex and GGA adaptors and is therefore not vulnerable to competition from standard motif-containing cargoes for incorporation into clathrin-coated vesicles. We propose that conceptually similar but mechanistically different interactions will direct the post-Golgi trafficking of many SNAREs.

Likelihood-based signal and noise analysis for docking of models into cryo-EM maps.

Fast, reliable docking of models into cryo-EM maps requires understanding of the errors in the maps and the models. Likelihood-based approaches to errors have proven to be powerful and adaptable in experimental structural biology, finding applications in both crystallography and cryo-EM. Indeed, previous crystallographic work on the errors in structural models is directly applicable to likelihood targets in cryo-EM. Likelihood targets in Fourier space are derived here to characterize, based on the comparison of half-maps, the direction- and resolution-dependent variation in the strength of both signal and noise in the data. Because the signal depends on local features, the signal and noise are analysed in local regions of the cryo-EM reconstruction. The likelihood analysis extends to prediction of the signal that will be achieved in any docking calculation for a model of specified quality and completeness. A related calculation generalizes a previous measure of the information gained by making the cryo-EM reconstruction.

Likelihood-based docking of models into cryo-EM maps.

Optimized docking of models into cryo-EM maps requires exploiting an understanding of the signal expected in the data to minimize the calculation time while maintaining sufficient signal. The likelihood-based rotation function used in crystallography can be employed to establish plausible orientations in a docking search. A phased likelihood translation function yields scores for the placement and rigid-body refinement of oriented models. Optimized strategies for choices of the resolution of data from the cryo-EM maps to use in the calculations and the size of search volumes are based on expected log-likelihood-gain scores computed in advance of the search calculation. Tests demonstrate that the new procedure is fast, robust and effective at placing models into even challenging cryo-EM maps.

Accelerating crystal structure determination with iterative AlphaFold prediction.

Experimental structure determination can be accelerated with artificial intelligence (AI)-based structure-prediction methods such as AlphaFold. Here, an automatic procedure requiring only sequence information and crystallographic data is presented that uses AlphaFold predictions to produce an electron-density map and a structural model. Iterating through cycles of structure prediction is a key element of this procedure: a predicted model rebuilt in one cycle is used as a template for prediction in the next cycle. This procedure was applied to X-ray data for 215 structures released by the Protein Data Bank in a recent six-month period. In 87% of cases our procedure yielded a model with at least 50% of Cα atoms matching those in the deposited models within 2 Å. Predictions from the iterative template-guided prediction procedure were more accurate than those obtained without templates. It is concluded that AlphaFold predictions obtained based on sequence information alone are usually accurate enough to solve the crystallographic phase problem with molecular replacement, and a general strategy for macromolecular structure determination that includes AI-based prediction both as a starting point and as a method of model optimization is suggested.