Development and applications of small-angle scattering-based structure modeling tools for proteins and nucleic acids

Small-angle scattering (SAS) of X-rays and neutrons allows the study of biological macromolecules in solution, at close to native conditions. The rapidly increasing popularity of the technique is attributed to both improvement in experimental facilities and continuous development of SAS data analysis and structure modeling tools. ATSAS, a software suite developed at the EMBL, is arguably the most comprehensive and utilized computer package for SAS data analysis worldwide. I present here the development of three computational tools, two of which have already been integrated into ATSAS: (1) SAS-guided normal mode analysis in torsion angle space (TNMA); (2) the use of sequence coevolution to reduce the ambiguity of SAS-based modeling; and (3) computation of anomalous scattering (ASAXS) effects in the context of SAXS data. Further, this PhD work contains the results of integrative structural biology projects in collaborations with user groups of the ESRF and EMBL Hamburg SAXS beamlines, where the newly developed methods were utilized. In normal mode analysis, macromolecular motion is approximated as collective, low frequency harmonic oscillations around an initial, equilibrium structure. NMA in Cartesian space (CNMA) has been demonstrated to reasonably approximate conformational changes for a large set of proteins, and was thus used as the basis for SREFLEX, a method in the ATSAS suite to morph crystallographic structures to fit SAS data. However, it was shown in this work that SAS-guided CNMA results in stereochemically broken structures when applied to RNA. In comparison, SAS-guided TNMA of the same RNA structures resulted in improved models, in terms of both accuracy and stereochemistry. An implementation of SAS-guided TNMA, NMATOR, was thus developed and made available in the latest ATSAS v3.0.0 package. NMATOR was also used to generate SAXS-based solution structure models of Alu RNA, and the condensin HEAT-repeat protein Ycg1, and the ISC proteins HscA and IscU. The solution properties and structure of Ycg1, as determined through SAXS, have been published (Manalastas-Cantos et al, 2019). SAS modeling ambiguity was also tackled in this work and two ways of ameliorating it through the generation of distance constraints were discussed: (1) experimentally, through anomalous scattering effects; and (2) bioinformatically, by evaluating sequence coevolution. A program to account for energy-dependent anomalous effects when computing SAXS data from macromolecular models was written and is available in ATSAS version 3.0.0, for planning and analyzing ASAXS experiments. In addition, sequence coevolution analysis and the integration of identified coevolving pair distance constraints into SAXS-guided modelling, was shown to improve heterodimer modeling accuracy. Sequence coevolution was utilized to generate distance constraints for HscB-IscU heterodimer modeling