Background

Oligomeric filamentous assemblies, made of the quasi-regular repetition of macromolecular protomers, are essential components of the cell. Filaments from the cytoskeleton support cell walls or form intracellular transport networks. Other types of filaments are involved in the maintenance, segregation, and repair of DNA (Ghosal and Löwe, 2015). These processes involve dynamic interactions between two generalized helices, the DNA helix and the filament helix, generally accompanied by changes in the DNA linking number and sometimes with coordinated changes in the filament topology. Elucidation of the structure of protein filaments using structural biology methods such as X-ray crystallography or Nuclear Magnetic Resonance has long been highly challenging due to their large size and difficulty in forming crystals. With the recent advances in Cryo-Electron Microscopy, many filament structures are being unveiled; however, filament plasticity and dynamic events are still difficult to capture, and model building remains necessary to decipher how filaments respond to external stimuli. Methods that have been proposed for the modeling of filaments are mostly based on specific docking approaches (Inbar et al., 2005; Casciari et al., 2006; Karaca et al., 2010; Esquivel-Rodriguez et al., 2012) and generally take advantage of known symmetry characteristics (Eisenstein et al., 1997; Berchanski et al., 2003 and 2005; Comeau and Camacho, 2005; Pierce et al., 2005; Schneidman-Duhovny et al., 2005). These methods are not appropriate for investigating structural perturbations in the filament, either global (changes in helical characteristics) or local (symmetry disruption), that appear as responses to external stimuli.

The Heligeom module (Boyer et al., 2015) of the PTools library software (Saladin et al., 2009) has been developed to meet the challenge of modeling helical protein filaments, their interactions with other molecules such as DNA, and their structural responses to external stimuli, either diffuse (changes in longitudinal or torsional stress, following changes in salt concentration or pH), or focused (local molecule binding or interface modifications). Heligeom derives the geometry of oligomeric assemblies from the binding geometry between pairs of neighboring protomers, which offers the possibility to combine several binding modes within the same filament. The PTools library itself gathers a collection of tools to manipulate macromolecules or selected regions of these macromolecules in atomic or coarse-grained representation. While it shares several functionalities with other libraries such as MMTK (Hinsen, 2000) or the VMD Script Library (Humphrey et al., 1996), certain functionalities such as the easy use of screw transformations make it particularly well suited to dealing with helical or quasi-helical assemblies.

Here, we present the application of PTools/Heligeom to model an irregular form of the RecA nucleofilament complex that is active in homologous recombination and includes one to three DNA strands (Boyer et al., 2015). Distortion in the filament results from the introduction of an alternative binding geometry of RecA-RecA interaction at the center of a two-turn helical nucleofilament (twelve monomers). Whereas the principal RecA-RecA binding mode corresponds to the favorable binding geometry between two RecA monomers in the presence of ATP, the alternative binding mode corresponds to the geometry in the presence of ADP. Both binding geometries have been solved by crystallography. The model therefore represents the result of one ATP molecule being hydrolyzed in the filament center. All three DNA strands participate in a strand exchange process. Strand 1 is a damaged DNA strand on which the filament has polymerized; it is bound to DNA-binding site I in the filament. Strand 3 is homologous to strand 1 and occupies the secondary binding site (site II). Strand 2 is complementary to both strand 1 and strand 3. During the strand exchange process catalyzed by the filament, strand 1 captures strand 2, which was initially paired with strand 1. We model three states of the system: the initial state, where only strand 1 is bound to site I (one bound strand); the annealed state, where strand 2 has paired with strand 1 at site I (two bound strands); and the state resulting from strand exchange, where strands 1 and 2 are at site I and strand 3 is at site II (three bound strands). The protocol presented here can be used for any oligomeric system in which three-dimensional structures are available for at least two protein-protein binding geometries obtained under two different conditions, for example, with different cofactors, and present regions of structural similarity. The protocol enables exploration of transient intermediate geometries that are difficult to access experimentally. It is particularly well suited to the study of "collaborative protein filaments," as described in Ghosal and Lowe (2015).