Figure - available from: Nature Methods
This content is subject to copyright. Terms and conditions apply.
Electron density maps calculated from models and simulated SAXS data for samples 1 through 11 (a) Electron density maps calculated from atomic coordinates. Electron density was calculated using Chimera to a resolution indicated in parentheses. (b) Electron density maps reconstructed from simulated SAXS data using DENSS. Simulated SAXS profiles were calculated using FoXS to the maximum q value from experiment (Supplementary Table 1). Electron densities are shown as volumes colored according to density (color bar indicates electron density values in e⁻/ų).
Source publication
Using a novel iterative structure factor retrieval algorithm, here I show that electron density can be directly calculated from solution scattering data without modeling. The algorithm was validated with experimental data from 12 different biological macromolecules. This approach avoids many of the assumptions limiting the resolution and accuracy o...
Citations
... DENsity from Solution Scattering (DENSS) [57] is a new computational algorithm that reconstructs a molecule's three-dimensional electron density distribution based on the SAXS profile. This provides a big step forward in the structural fitting of SAXS data, for which only low-resolution shape reconstructions are usually possible. ...
... The process iterates by calculating a forward FFT from the flattened density to produce updated structure factors, and the cycle continues until convergence is achieved. This iterative approach allows DENSS to refine the electron [57]. DENSS calculates 3D structure factors from scattering intensities, scales them to match experimental data, and repeats the cycle until convergence. ...
... An attempt to use DQNs for solving the phase problem is discussed in Section 7.4. 57 ...
In recent years, advances in Artificial Intelligence and experimental techniques have revolutionized the field of structural biology. X-ray crystallography and Cryo-EM have provided unprecedented insights into the structures of biomolecules, while the unexpected success of AlphaFold has opened up new avenues of investigation. However, studying the dynamics of proteins at high resolution remains a significant obstacle, especially for fast dynamics. Single-particle imaging (SPI) or Flash X-ray Imaging (FXI) is an emerging technique that may enable the mapping of the conformational landscape of biological molecules at high resolution and fast time scale. This thesis discusses the potential of SPI/FXI, its challenges, recent experimental successes, and the advancements driving its development. In particular, machine learning and neural networks could play a vital role in fostering data analysis and improving SPI/FXI data processing. In Paper I, we discuss the problem of noise and detector masks in collecting FXI data. I simulated a dataset of diffraction patterns and used it to train a Convolutional Neural Network (U-Net) to restore data by denoising and filling in detector masks. As a natural continuation of this work, I trained another machine learning model in Paper II to estimate 2D protein densities from diffraction intensities. In the final chapter, corresponding to Paper III, we discuss another experimental method, time-resolved Small Angle X-ray Scattering (SAXS), and a new algorithm recently developed for SAXS data, the DENsity from Solution Scattering (DENSS) algorithm. I discuss the potential of DENSS in time-resolved SAXS and its application for structural fitting of AsLOV2, a Light-Oxygen-Voltage (LOV) protein domain from Avena sativa.
... Density maps were modelled in RAW 2.1.4 (Hopkins et al., 2017) using the DENSS electron density map feature (Grant, 2018). 20 reconstructions were run with the default set of 10000 electrons. ...
The major myelin protein expressed by the peripheral nervous system Schwann cells is protein zero (P0), representing 50% of the total protein content in myelin. This 30-kDa integral membrane protein consists of an immunoglobulin (Ig)-like domain, a transmembrane helix, and a 69-residue C-terminal cytoplasmic tail (P0ct). The basic residues in P0ct contribute to the tight packing of the myelin lipid bilayers, and alterations in the tail affect how P0 functions as an adhesion molecule necessary for the stability of compact myelin. Several neurodegenerative neuropathies are related to P0, including the more common Charcot-Marie-Tooth disease (CMT) and Dejerine-Sottas syndrome (DSS), but also rare cases of motor and sensory polyneuropathy. We find that high P0ct concentrations affect the membrane properties of bicelles and induce a lamellar-to-inverted hexagonal phase transition, which causes the bicelles to fuse into long, protein-containing filament-like structures. These structures likely reflect the formation of semi-crystalline lipid domains of potential relevance for myelination. Not only is P0ct important for stacking lipid membranes, but time-lapse fluorescence microscopy shows that it might affect membrane properties during myelination. We further describe recombinant production and low-resolution structural characterization of full-length human P0. Our findings shed light on P0ct effects on membrane properties, and with successful purification of full-length P0, we have new tools to study in vitro the role P0 has in myelin formation and maintenance.
... Based on these models, an excellent agreement between the experimental and theoretically predicted RDCs was observed (Fig. 2d, e). Furthermore, the overlay of the NMR/WAXSiS MprE7-TH1 model into the volume data of the SAXSderived solvent envelopes, calculated in Chimera, demonstrated a good correlation of 0.9 (Fig. 2f, Supplementary Fig. 15) [36][37][38] . This agreement between the two analyses provides additional support for the accuracy and consistency of the obtained room-temperature solution structural model of MprE7-TH1. ...
... Side chain of the terminal Trp residue that is brominated by SrpI is shown in stick representation. f The density from solution scattering (DENSS) 36 electron density envelope, represented as a transparent surface, generated from the SAXS dataset for MprE7-TH1 at a concentration of 523 μM. The top-ten NMR-derived models were assessed using the ATSAS CRYSOL 70 and OLIGOMER 71 programs and further refined through explicitsolvent all-atom molecular dynamics simulations with the WAXSiS server; the best model was manually superimposed onto the electron density envelope using PyMOL. ...
To biosynthesize ribosomally synthesized and post-translationally modified peptides (RiPPs), enzymes recognize and bind to the N-terminal leader region of substrate peptides which enables catalytic modification of the C-terminal core. Our current understanding of RiPP leaders is that they are short and largely unstructured. Proteusins are RiPP precursor peptides that defy this characterization as they possess unusually long leaders. Proteusin peptides have not been structurally characterized, and we possess scant understanding of how these atypical leaders engage with modifying enzymes. Here, we determine the structure of a proteusin peptide which shows that unlike other RiPP leaders, proteusin leaders are preorganized into a rigidly structured region and a smaller intrinsically disordered region. With residue level resolution gained from NMR titration experiments, the intermolecular peptide-protein interactions between proteusin leaders and a flavin-dependent brominase are mapped onto the disordered region, leaving the rigidly structured region of the proteusin leader to be functionally dispensable. Spectroscopic observations are biochemically validated to identify a binding motif in proteusin peptides that is conserved among other RiPP leaders as well. This study provides a structural characterization of the proteusin peptides and extends the paradigm of RiPP modification enzymes using not only unstructured peptides, but also structured proteins as substrates.
... The calculated electron density map with a high contour level (σ-level) displayed an isolated electron density at a location corresponding to the LID domain and an elongated shape from the NMP to the CORE domains. Next, the electron density map of the uncomplexed Adk tm was constructed from the SAXS scattering curve using DENSS (Grant, 2018), an algorithm for calculating a 3D structure from solution scattering data (Figure 7d and Figure S14). At the high σ-level, isolated electron density and elongated electron density were observed in the upper and lower part of the figure, respectively. ...
Synthetic binding proteins have emerged as modulators of protein functions through protein–protein interactions (PPIs). Because PPIs are influenced by the structural dynamics of targeted proteins, investigating whether the synthetic‐binders‐based strategy is applicable for proteins with large conformational changes is important. This study demonstrates the applicability of monobodies (fibronectin type‐III domain‐based synthetic binding proteins) in regulating the functions of proteins that undergo tens‐of‐angstroms‐scale conformational changes, using an example of the A55C/C77S/V169C triple mutant (Adktm; a phosphoryl transfer‐catalyzing enzyme with a conformational change between OPEN/CLOSED forms). Phage display successfully developed monobodies that recognize the OPEN form (substrate‐unbound form), but not the CLOSED form of Adktm. Two OPEN form‐specific clones (OP‐2 and OP‐4) inhibited Adktm kinase activity. Epitope mapping with a yeast‐surface display/flow cytometry indicated that OP‐2 binds to the substrate‐entry side of Adktm, whereas OP‐4 binding occurs at another site. Small angle X‐ray scattering coupled with size‐exclusion chromatography (SEC‐SAXS) indicated that OP‐4 binds to the hinge side opposite to the substrate‐binding site of Adktm, retaining the whole OPEN‐form structure of Adktm. Titration of the OP‐4–Adktm complex with Ap5A, a transition‐state analog of Adktm, showed that the conformational shift to the CLOSED form was suppressed although Adktm retained the OPEN‐form (i.e., substrate‐binding ready form). These results show that OP‐4 captures and stabilizes the OPEN‐form state, thereby affecting the hinge motion. These experimental results indicate that monobody‐based modulators can regulate the functions of proteins that show tens‐of‐angstroms‐scale conformational changes, by trapping specific conformational states generated during large conformational change process that is essential for function exertion.
... Solution scattering provides information on the structure and dynamics of biological macromolecules in aqueous solutions, yielding complementary information to highresolution structural techniques such as X-ray crystallography, cryo-electron microscopy (cryoEM), and nuclear magnetic resonance (NMR) (1-5). Ab initio modeling approaches have enabled the determination of low-resolution 3D shapes and density maps directly from 1D small angle X-ray scattering (SAXS) profiles (6)(7)(8)(9). We have previously described our approach for low-resolution ab initio generation of density maps from X-ray or neutron scattering data, called DENSS (9). ...
... Ab initio modeling approaches have enabled the determination of low-resolution 3D shapes and density maps directly from 1D small angle X-ray scattering (SAXS) profiles (6)(7)(8)(9). We have previously described our approach for low-resolution ab initio generation of density maps from X-ray or neutron scattering data, called DENSS (9). DENSS operates by iterating between the real space and reciprocal space domains using the Fast Fourier Transform (FFT) (10), enforcing appropriate restraints in each domain (i.e., 1D SAXS data in reciprocal space and 3D solvent flattening in real space). ...
... Finally, the algorithm presented here generates a highly accurate electron density map of a protein at high resolution that agrees with experimental SWAXS data. This will enable future modeling applications using our novel iterative structure factor retrieval algorithm implemented in DENSS (9). By combining experimental SWAXS data of full-length proteins or protein-protein complexes with high resolution density maps from known structural fragments or subunits or using predicted structures (e.g., from AlphaFold (63)), we will be able to significantly improve the resolution and accuracy of density reconstructions of unknown regions. ...
... For each dataset, models of the most populated cluster have been superposed and merged by using DAMAVER and the resulting model has been used as input for more accurate modeling (slow-mode annealing procedure in DAMMIN (Svergun, 1999). An alternative molecular envelope calculation was carried out by using the program DENSS (Grant, 2018). The similarity between molecular envelopes has been assessed by the SUPCOMB program (Kozin & Svergun, 2001), which is based on the NSD distance metrics. ...
Human aromatic amino acid decarboxylase (AADC) is a pyridoxal 5′‐phosphate‐dependent enzyme responsible for the biosynthesis of dopamine and serotonin, essential neurotransmitters involved in motor and cognitive abilities. Mutations in its gene lead to AADC deficiency, a monogenic rare neurometabolic childhood parkinsonism characterized by severe motor and neurodevelopmental symptoms. Here, for the first time, we solved the crystal structure of human holoAADC in the internal aldimine (1.9 Å) and in the external aldimine (2.4 Å) of the substrate analog L‐Dopa methylester. In this intermediate, the highly flexible AADC catalytic loop (CL) is captured in a closed state contacting all protein domains. In addition, each active site, composed by residues of both subunits, is connected to the other through weak interactions and a central cavity. By combining crystallographic analyses with all‐atom and coarse‐grained molecular dynamics simulations, SAXS investigations and limited proteolysis experiments, we realized that the functionally obligate homodimeric AADC enzyme in solution is an elongated, asymmetric molecule, where the fluctuations of the CL are coupled to flexibility at the edge between the N‐terminal and C‐terminal domains. The structural integrity of this peripheral protein region is essential to catalysis, as assessed by both artificial and 37 AADC deficiency pathogenic variants leading to the interpretation that structural dynamics in protein regions far from the active site is essential for CL flexibility and the acquirement of a correct catalytically competent structure. This could represent the molecular basis for pathogenicity prediction in AADC deficiency.
... The information on data collection and derived structural parameters is summarized in Supplemental Table S7. Ab initio calculations were performed with the DENSS suite (Grant 2018), and fourfold symmetry was imposed during shape reconstruction. Molecular models of the parallel and antiparallel coiled coils fused to SUMO domains were generated from the AlphaFold models obtained in this work and the X-ray structure of the SUMO protein (PDB 7P47) (Varejão et al. 2021) and refined in Xplor-NIH v 2.49 (Schwieters et al. 2003). ...
Meiosis-specific Rec114−Mei4 and Mer2 complexes are thought to enable Spo11-mediated DNA double-strand break (DSB) formation through a mechanism that involves DNA-dependent condensation. However, the structure, molecular properties, and evolutionary conservation of Rec114−Mei4 and Mer2 are unclear. Here, we present AlphaFold models of Rec114−Mei4 and Mer2 complexes supported by nuclear magnetic resonance (NMR) spectroscopy, small-angle X-ray scattering (SAXS), and mutagenesis. We show that dimers composed of the Rec114 C terminus form α-helical chains that cup an N-terminal Mei4 α helix, and that Mer2 forms a parallel homotetrameric coiled coil. Both Rec114−Mei4 and Mer2 bind preferentially to branched DNA substrates, indicative of multivalent protein–DNA interactions. Indeed, the Rec114−Mei4 interaction domain contains two DNA-binding sites that point in opposite directions and drive condensation. The Mer2 coiled-coil domain bridges coaligned DNA duplexes, likely through extensive electrostatic interactions along the length of the coiled coil. Finally, we show that the structures of Rec114−Mei4 and Mer2 are conserved across eukaryotes, while DNA-binding properties vary significantly. This work provides insights into the mechanism whereby Rec114−Mei4 and Mer2 complexes promote the assembly of the meiotic DSB machinery and suggests a model in which Mer2 condensation is the essential driver of assembly, with the DNA-binding activity of Rec114−Mei4 playing a supportive role.
... Solution scattering provides information on the structure and dynamics of biological macromolecules in aqueous solutions, yielding complementary information to highresolution structural techniques such as X-ray crystallography, cryo-electron microscopy (cryoEM), and nuclear magnetic resonance (NMR) (1-5) . Ab initio modeling approaches have enabled the determination of low-resolution 3D shapes and density maps directly from 1D small angle X-ray scattering (SAXS) profiles (6)(7)(8)(9). We have previously described our approach for low-resolution ab initio generation of density maps from X-ray or neutron scattering data, called DENSS (9). ...
... Ab initio modeling approaches have enabled the determination of low-resolution 3D shapes and density maps directly from 1D small angle X-ray scattering (SAXS) profiles (6)(7)(8)(9). We have previously described our approach for low-resolution ab initio generation of density maps from X-ray or neutron scattering data, called DENSS (9). DENSS operates by iterating between the real space and reciprocal space domains using the Fast Fourier Transform (FFT) (10), enforcing appropriate restraints in each domain (i.e., 1D SAXS data in reciprocal space and 3D solvent flattening in real space). ...
... The copyright holder for this preprint (which this version posted June 3, 2023. modeling applications using our novel iterative structure factor retrieval algorithm implemented in DENSS (9). By combining experimental SWAXS data of full-length proteins or protein-protein complexes with high resolution density maps from known structural fragments or subunits or using predicted structures (e.g., from AlphaFold (59)), we will be able to significantly improve the resolution and accuracy of density reconstructions of unknown regions. ...
Solution scattering techniques, such as small and wide-angle X-ray scattering (SWAXS), provide valuable insights into the structure and dynamics of biological macromolecules in solution. In this study, we present an approach to accurately predict solution X-ray scattering profiles at wide angles from atomic models by generating high-resolution electron density maps. Our method accounts for the excluded volume of bulk solvent by calculating unique adjusted atomic volumes directly from the atomic coordinates. This approach eliminates the need for a free fitting parameter commonly used in existing algorithms, resulting in improved accuracy of the calculated SWAXS profile. An implicit model of the hydration shell is generated which uses the form factor of water. Two parameters, namely the bulk solvent density and the mean hydration shell contrast, are adjusted to best fit the data. Results using eight publicly available SWAXS profiles show high quality fits to the data. In each case, the optimized parameter values show small adjustments demonstrating that the default values are close to the true solution. Disabling parameter optimization results in a significant improvement of the calculated scattering profiles compared to the leading software. The algorithm is computationally efficient, showing more than tenfold reduction in execution time compared to the leading software. The algorithm is encoded in a command line script called denss.pdb2mrc.py and is available open source as part of the DENSS v1.7.0 software package ( https://github.com/tdgrant1/denss ). In addition to improving the ability to compare atomic models to experimental SWAXS data, these developments pave the way for increasing the accuracy of modeling algorithms utilizing SWAXS data while decreasing the risk of overfitting.
Statement of Significance
Accurate calculation of small and wide-angle scattering (SWAXS) profiles from atomic models is useful for studying the solution state and conformational dynamics of biological macromolecules in solution. Here we present a new approach to calculating SWAXS profiles from atomic models using high resolution real space density maps. This approach includes novel calculations of solvent contributions that remove a significant fitting parameter. The algorithm is tested on multiple high quality experimental SWAXS datasets, showing improved accuracy compared to leading software. The algorithm is computationally efficient and robust to overfitting, paving the way for increasing the accuracy and resolution of modeling algorithms utilizing experimental SWAXS data.
... Scattering profiles were averaged, reduced, and merged from measurements at different concentrations using BioXTAS RAW software (48). Structural parameters and the distance distribution function, P(r), were calculated with GNOM (49) and the ab initio electron density was reconstructed using DENSS (50). ...
Core histones are synthesized and processed in the cytoplasm before transport into the nucleus for assembly into nucleosomes; however, they must also be chaperoned as free histones are toxic. The importin Kap114 binds and transports histone H2A-H2B into the yeast nucleus, where Ran GTP facilitates H2A-H2B release. Kap114 and H2A-H2B also bind the Nap1 histone chaperone, which is found in both the cytoplasm and the nucleus, but how Nap1 and Kap114 cooperate in H2A-H2B processing and nucleosome assembly has been unclear. To understand these mechanisms, we used biochemical and structural analyses to reveal how Nap1, Kap114, H2A-H2B and Ran GTP interact. We show that Kap114, H2A-H2B and a Nap1 dimer (Nap1 2 ) assemble into a 1:1:1 ternary complex. Cryogenic electron microscopy revealed two distinct Kap114/Nap1 2 /H2A-H2B structures: one of H2A-H2B sandwiched between Nap1 2 and Kap114, and another in which Nap1 2 bound to the Kap114•H2A-H2B complex without contacting H2A-H2B. Another Nap1 2 •H2A-H2B•Kap114•Ran GTP structure reveals the nuclear complex. Mutagenesis revealed shared critical interfaces in all three structures. Consistent with structural findings, DNA competition experiments demonstrated that Kap114 and Nap1 2 together chaperone H2A-H2B better than either protein alone. When Ran GTP is present, Kap114’s chaperoning activity diminishes. However, the presence of Nap1 2 within the Nap1 2 •H2A-H2B•Kap114•Ran GTP quaternary complex restores its ability to chaperone H2A-H2B. This complex effectively deposits H2A-H2B into nucleosomes. Together, these findings suggest that Kap114 and Nap1 2 provide a sheltered path from cytoplasm to nucleus, facilitating the transfer of H2A-H2B from Kap114 to Nap1 2 , ultimately directing its specific deposition into nucleosomes.
Significance Statement
Free histones are toxic and must be sequestered by other macromolecules in the cell. Nuclear import receptor Kap114 imports H2A-H2B into the nucleus while also chaperoning it. The histone chaperone Nap1 also chaperones H2A-H2B, but it is unclear how Nap1 and Kap114 cooperate to process H2A-H2B. We present biochemical and structural results that explain how Kap114, Nap1 and H2A-H2B assemble in the absence and presence of Ran GTP , how Nap1 and Kap114 co-chaperone H2A-H2B, and how Ran GTP and Nap1 coordinate the transfer of H2A-H2B from Kap114 to assembling nucleosomes in the nucleus.
... We observed that 0.2 M NaCl made mAb-A more compact with a smaller maximum length and globular shape than 0 M NaCl. Figure 2f shows the reconstructed threedimensional (3D) models of mAb-A at pH 2 from the SAXS profiles and P(r) by the ab initio electron density determination, DENSS (DENsity from Solution Scattering), algorithm. 25 The flattened shapes of mAb-A at pH 2 were also consistent with those obtained via the fitting analysis that used the theoretical scattering functions. In addition, these 3D models illustrated some anisotropic electron distributions, such as a lower-density region that looked like a "neck" in 0 M NaCl. ...
... (f) 3D reconstructed models of mAb-A at pH 2 using ab initio electron density determination (DENSS). 25 The canonical mAb-A models are also shown, indicating the ab initio model reconstructed from the SAXS data (light pink), overlaid with the atomic coordinates of IgG1 (red ribbons). 26 The 3D model files are available in the Supporting Information. ...
Protein denaturation is a ubiquitous process that occurs both in vitro and in vivo. While our molecular understanding of the denatured structures of proteins is limited, it is commonly accepted that the loss of unique intramolecular contacts makes proteins larger. Herein, we report compaction of the immunoglobulin G1 (IgG1) protein upon acid denaturation. Small-angle X-ray scattering coupled with size exclusion chromatography revealed that IgG1 radii of gyration at pH 2 were ∼75% of those at a neutral pH. Scattering profiles showed a compact globular shape, supported by analytical ultracentrifugation. The acid denaturation of proteins with a decrease in size is energetically costly, and acid-induced compaction requires an attractive force for domain reorientation. Such intramolecular aggregation may be widespread in immunoglobulin proteins as noncanonical structures. Herein, we discuss the potential biological significance of these noncanonical structures of antibodies.
