Victor S. Lamzin

Automated protein model building combined with iterative structure refinement.

In protein crystallography, much time and effort are often required to trace an initial model from an interpretable electron density map and to refine it until it best agrees with the crystallographic data. Here, we present a method to build and refine a protein model automatically and without user intervention, starting from diffraction data extending to resolution higher than 2.3 Å and reasonable estimates of crystallographic phases. The method is based on an iterative procedure that describes the electron density map as a set of unconnected atoms and then searches for protein-like patterns. Automatic pattern recognition (model building) combined with refinement, allows a structural model to be obtained reliably within a few CPU hours. We demonstrate the power of the method with examples of a few recently solved structures.

Automated refinement of protein models.

An automated refinement procedure (ARP) for protein models is proposed, and its convergence properties discussed. It is comparable to the iterative least-squares minimization/difference Fourier synthesis approach for small molecules. ARP has been successfully applied to three proteins, and for two of them resulted in models very similar to those obtained by conventional least-squares refinement and rebuilding with FRODO. In real time ARP is about ten times faster than conventional refinement. In its present form ARP requires high (2.0 A or better) resolution data, which should be of high quality and a starting protein model having about 75% of the atoms in roughly the correct position. For the third protein at 2.4 A resolution, ARP was significantly less powerful but nevertheless gave definite improvement, in the density map at least.

ARP⧸wARP and Automatic Interpretation of Protein Electron Density Maps

Publisher Summary This chapter presents phase improvement, coupled with automated map interpretation and model building, as one unified process within the framework of the automated refinement procedure (ARP/wARP) software suite. ARP/wARP is an experimental hypothesis-generating and testing procedure for placing atoms in the most likely places (according to density) and using graph-searching combined with geometric comparisons against expected stereochemical parameters to determine the most likely mainchain fragments. ARP/wARP is a software suite (copyrighted by the European Molecular Biology Laboratory) based on the paradigm of viewing model building and refinement as one unified procedure for optimizing phase estimates. The current version ARP/wARP 6.0, released in July 2002, works with density recognition-driven procedures for placing and removing atoms and therefore is limited to diffraction data extending to about 2.5 A. The iterative cycles of density modeling by the placement of atoms, unrestrained refinement of their parameters, automated model building, and restrained refinement of the hybrid model provide a powerful means of phase refinement.

ARP/wARP and molecular replacement

The aim of ARP/wARP is improved automation of model building and refinement in macromolecular crystallography. Once a molecular-replacement solution has been obtained, it is often tedious to refine and rebuild the initial (search) model. ARP/wARP offers three options to automate that task to varying extents: (i) autobuilding of a completely new model based on phases calculated from the molecular-replacement solution, (ii) updating of the initial model by atom addition and deletion to obtain an improved map and (iii) docking of a structure onto a new (or mutated) sequence, followed by rebuilding and refining the side chains in real space. A few examples are presented where ARP/wARP made a considerable difference in the speed of structure solution and/or made possible refinement of otherwise difficult or uninterpretable maps. The resolution range allowing complete autobuilding of protein structures is currently 2.0 A, but for map improvement considerable advances over more conventional refinement techniques are evident even at 3.2 A spacing.

wARP: Improvement and Extension of Crystallographic Phases by Weighted Averaging of Multiple-Refined Dummy Atomic Models

wARP is a procedure that substantially improves crystallographic phases (and subsequently electron-density maps) as an additional step after density-modification methods such as solvent flattening and averaging. The initial phase set is used to create a number of dummy atom models which are subjected to least-squares or maximum-likelihood refinement and iterative model updating in an automated refinement procedure (ARP). Averaging of the phase sets calculated from the refined output models and weighting of structure factors by their similarity to an average vector results in a phase set that improves and extends the initial phases substantially. An important requirement is that the native data have a maximum resolution beyond approximately 2.4 A. The wARP procedure shortens the time-consuming step of model building in crystallographic structure determination and helps to prevent the introduction of errors.

Auto-Rickshaw: an automated crystal structure determination platform as an efficient tool for the validation of an X-ray diffraction experiment

The EMBL-Hamburg Automated Crystal Structure Determination Platform is a system that combines a number of existing macromolecular crystallographic computer programs and several decision-makers into a software pipeline for automated and efficient crystal structure determination. The pipeline can be invoked as soon as X-ray data from derivatized protein crystals have been collected and processed. It is controlled by a web-based graphical user interface for data and parameter input, and for monitoring the progress of structure determination. A large number of possible structure-solution paths are encoded in the system and the optimal path is selected by the decision-makers as the structure solution evolves. The processes have been optimized for speed so that the pipeline can be used effectively for validating the X-ray experiment at a synchrotron beamline.

AUTOMATED REFINEMENT FOR PROTEIN CRYSTALLOGRAPHY

Publisher Summary This chapter discusses an automated refinement procedure (ARP) for proteins. The basis of ARP is the iterative use of unrestrained least-squares minimization coupled with constant updating of the model. This is comparable to the iterative least-squares/Fourier synthesis approach for small molecules. It requires X-ray data to 2.0 A, or better, to allow unrestrained refinement and improvement of the whole content of the unit cell. At lower resolutions, as a rule, only unrestrained parts of the model are expected to be improved. The quality of data and the initial phase set greatly influences the power of ARP. Applied to the refinement of a medium-size structure at 1.0 A resolution and starting from one heavy-atom position, ARP determined the complete structure in a fully automated manner. ARP uses atomicity as the main property of the structure and differs completely from, for example, direct methods that are based on atomicity through statistical relationships between amplitudes of structure factors.

Crystal Structure of Manganese Catalase from Lactobacillus plantarum

BACKGROUND Catalases are important antioxidant metalloenzymes that catalyze disproportionation of hydrogen peroxide, forming dioxygen and water. Two families of catalases are known, one having a heme cofactor, and the other, a structurally distinct family containing nonheme manganese. We have solved the structure of the mesophilic manganese catalase from Lactobacillus plantarum and its azide-inhibited complex. RESULTS The crystal structure of the native enzyme has been solved at 1.8 A resolution by molecular replacement, and the azide complex of the native protein has been solved at 1.4 A resolution. The hexameric structure of the holoenzyme is stabilized by extensive intersubunit contacts, including a beta zipper and a structural calcium ion crosslinking neighboring subunits. Each subunit contains a dimanganese active site, accessed by a single substrate channel lined by charged residues. The manganese ions are linked by a mu1,3-bridging glutamate carboxylate and two mu-bridging solvent oxygens that electronically couple the metal centers. The active site region includes two residues (Arg147 and Glu178) that appear to be unique to the Lactobacillus plantarum catalase. CONCLUSIONS A comparison of L. plantarum and T. thermophilus catalase structures reveals the existence of two distinct structural classes, differing in monomer design and the organization of their active sites, within the manganese catalase family. These differences have important implications for catalysis and may reflect distinct biological functions for the two enzymes, with the L. plantarum enzyme serving as a catalase, while the T. thermophilus enzyme may function as a catalase/peroxidase.

ARP/wARP's model-building algorithms. I. The main chain

Algorithms underlying the automatic model-building functionality of the ARP/wARP software suite are presented. Finding the most likely set of Calpha atoms from a given set of atoms is formulated as a constrained integer programming problem. The objective function is a density-weighted score for the match between observed and expected chain conformation. Graph-search algorithms are presented that find solutions to this problem in an efficient manner.

Towards complete validated models in the next generation of ARP/wARP

The design of a new versatile control system that will underlie future releases of the automated model-building package ARP/wARP is presented. A sophisticated expert system is under development that will transform ARP/wARP from a very useful model-building aid to a truly automated package capable of delivering complete, well refined and validated models comparable in quality to the result of intensive manual checking, rebuilding, hypothesis testing, refinement and validation cycles of an experienced crystallographer. In addition to the presentation of this control system, recent advances, ideas and future plans for improving the current model-building algorithms, especially for completing partially built models, are presented. Furthermore, a concept for integrating validation routines into the iterative model-building process is also presented.