Marcos Sotomayor

Deep learning approaches have revolutionized protein structure prediction. These tools are trained using experimental data and recapitulate reported conformations, but there is great interest in predicting conformations that may be functionally relevant although experimentally underrepresented. Since many modern structure prediction tools use generative artificial intelligence diffusion models, we reframe the search for alternative molecular conformations as that of sampling from a diffusion distribution conditioned using any arbitrary Bayesian likelihood. We implement a twisted diffusion sampler in Boltz-2 to sample this conditioned distribution and demonstrate the utility of this approach, which does not require any additional training of the neural network, by implementing a diffusion analog of steered molecular dynamics simulations applied to mechanical systems. We can reproduce predicted stretched states of fragments of DNA, the muscle protein titin, and the inner-ear protocadherin-15 protein, as well as open states of the MscL ion channel consistent with experimental results. We expect that steered structure predictions will help sample underrepresented and non-equilibrium conformations for many macromolecular systems.

Scaffolded DNA origami has become a valuable nanoscale tool for applications in biomedical and physical sciences. Critical to leveraging the modular and programmable properties of DNA origami nanodevices is access to the scaffold strand, a long single-stranded DNA (ssDNA) of precise length and sequence, which is folded into a compact shape via piecewise base-pairing with many staple strands, short ssDNA oligonucleotides. Current methods to produce and manipulate long ssDNA scaffolds can be costly, time-consuming, and cumbersome. In contrast, methods to produce and manipulate the sequence of double-stranded DNA (dsDNA) are efficient and scalable. Here, we present a method for the rapid isolation of target ssDNA sequences from a variety of dsDNA sources using oligonucleotides as blocking strands that bind continuously to the undesired strand, thereby releasing the target scaffold strand. We report successful ssDNA isolation from linear and supercoiled dsDNAs of various sequences and lengths, ranging from 769 to 15,101 nucleotides. In addition to isolating ssDNA, we demonstrated this approach enables folding of DNA origami directly from dsDNA templates using both blocking and staple strands in a single-pot thermally controlled reaction. Furthermore, we explore multi-scaffold and gene-encoding DNA origami structures, expanding the framework for application-based designs. Graphical

Molecular dynamics simulation is a powerful computational technique used for predicting and understanding the dynamic behavior of biomolecular systems. Steered molecular dynamics (SMD) simulations enable the study of force-induced processes in biomolecules, effectively mimicking single-molecule force spectroscopy experiments probing protein unfolding and receptor-ligand unbinding. Given the stochastic nature of these mechanical events, accurately exploring the dynamic behavior of biomolecules and extracting accurate physical information requires several in-silico experiments. This includes performing many pulling simulations at different velocities or force loading rates. The large amount of data obtained from these simulation sets requires efficient automated data processing tools. We present PySteMoDA, a novel Python package with a user-friendly graphical interface specifically designed for constant-velocity SMD data analysis. The automated force peak detection methods reduce user bias, improve accuracy, and accelerate data analysis. The package also allows identification of residues involved in mechanical events through computation of the time-dependent mechanical work and correlation factors between residue pairs. This package not only addresses automated data processing in SMD simulations and accurate parameter extraction, but also significantly enhances accessibility and usability. Through PySteMoDA, users can efficiently analyze simulation data without the barrier of coding, facilitating a wider range of investigations and insights in the field of computational biochemistry and biophysics.

Lipid bilayers are essential to life as they surround most cells and membrane-bound organelles. The integrity and fate of cells depend on the asymmetric makeup of lipid bilayers with various membrane proteins regulating the lipid composition of a bilayer’s two leaflets. Lipids scramblases are one of the primary regulators of lipid asymmetry in bilayers, spontaneously transferring lipids between membrane leaflets. Members of the TMEM16, OSCA/TMEM63, and TMC families have been suggested to be lipid scramblases. Despite significant differences, these proteins share a common structural architecture that features a membrane-exposed groove. The “credit card” mechanism proposes that lipids switch leaflets by moving their polar head groups either inside (partially dry) or on the surface of (wet) membrane-exposed, open hydrophilic grooves. However, emerging evidence of closed-groove scrambling challenges this model. Given the sequence diversity of groove-lining amino acids in TMEM16, OSCA/TMEM63, and TMC proteins, we hypothesized that lipid scrambling is primarily determined by groove architecture. To test this hypothesis and the credit card mechanism, we used coarse-grained molecular dynamics simulations of experimental structures and AlphaFold-generated models of six different scramblases in closed and open states. In these simulations, we observed little scramblase activity in most closed-state configurations but robust scrambling by all open-state models. We then built simplified TMEM16-based scramblases with only three bead types uniformly set for solvent-facing, transmembrane, and groove regions. We used this and further simplified models to vary groove surface hydrophilicity, groove surface geometry, and groove architecture. Our models support the partially dry and wet credit card mechanisms and suggest that groove architecture plays a more important role in facilitating lipid scrambling than the detailed sequence of groove-lining amino acids. STATEMENT OF SIGNIFICANCE Many biological processes, including blood clotting and apoptosis, are triggered by exposure of specific lipids on the outer leaflet of a lipid bilayer. Spontaneous diffusion of lipids between leaflets is slow, but scramblases can speed up transbilayer lipid exchange, a key step in signaling cascades. Here we used coarse-grained simulations to isolate and quantify the key biophysical properties of scramblases that modulate lipid movement across the bilayer. We found that the architecture of scramblases’ membrane-exposed groove is key to their function. This can be used to generate hypotheses about disease-causing mutations in known scramblases, guide drug design, and suggest favorable properties for de novo designed scramblases.

This paper, on the role of calcium and integrin-binding protein 2 and 3 in the hair-cell in the mechano-electrical transduction (MET) apparatus, is a mix of confirmatory studies with new and potentially important data. Some parts, such as zebrafish studies, the modelling and simulations, are regarded as necessary and convincing. Other parts of the paper do not have the same novelty. Both Liang et al. (2021) and Wang et al. (2023) had previously demonstrated a role for CIB2/CIB3 in auditory and vestibular cells in mice. Moreover, there are also data in Riazuddin et al. (2012) paper that demonstrates the importance of CIB2 in zebrafish and Drosophila. Breaking the manuscript up to focus on specific aspects of the problem might alleviate the limitations of this multi-faceted study.

Calcium and integrin-binding protein 2 (CIB2) and CIB3 bind to transmembrane channel-like 1 (TMC1) and TMC2, the pore-forming subunits of the inner-ear mechano-electrical transduction (MET) apparatus. These interactions have been proposed to be functionally relevant across mechanosensory organs and vertebrate species. Here we show that both CIB2 and CIB3 can form heteromeric complexes with TMC1 and TMC2 and are integral for MET function in mouse cochlea and vestibular end organs as well as in zebrafish inner ear and lateral line. Our AlphaFold 2 models suggest that vertebrate CIB proteins can simultaneously interact with at least two cytoplasmic domains of TMC1 and TMC2 as validated using nuclear magnetic resonance spectroscopy of TMC1 fragments interacting with CIB2 and CIB3. Molecular dynamics simulations of TMC1/2 complexes with CIB2/3 predict that TMCs are structurally stabilized by CIB proteins to form cation channels. Overall, our work demonstrates that intact CIB2/3 and TMC1/2 complexes are integral to hair-cell MET function in vertebrate mechanosensory epithelia.

Mechanical forces play key roles in biological processes such as cell migration and sensory perception. In recent years molecular force sensors have been developed as tools for in situ force measurements. Here we use all-atom steered molecular dynamics simulations to predict and study the relationship between design parameters and mechanical properties for three types of molecular force sensors commonly used in cellular biological research: two peptide-and one DNA-based. The peptide-based sensors consist of a pair of fluorescent proteins, which can undergo Förster resonance energy transfer (FRET), linked by spider silk (GPGGA) _n or synthetic (GGSGGS) _n disordered regions. The DNA-based sensor consists of two fluorophore-labeled strands of DNA that can be unzipped or sheared upon force application with a FRET signal as readout of dissociation. We simulated nine sensors, three of each kind. After equilibration, flexible peptide linkers of three different lengths were stretched by applying forces to their N-and C-terminal Cα atoms in opposite directions. Similarly, we equilibrated a DNA-based sensor and pulled on the phosphate atom of the terminal guanine of one strand and a selected phosphate atom on the other strand in the opposite direction. These simulations were performed at constant velocity (0.01 nm/ns – 10 nm/ns) and constant force (10 pN – 500 pN) for all versions of the sensors. Our results show how the force response of these sensors depends on their length, sequence, configuration and loading rate. Mechanistic insights gained from simulations analyses indicate that interpretation of experimental results should consider the influence of transient formation of secondary structure in peptide-based sensors and of overstretching in DNA-based sensors. These predictions can guide optimal fluorophore choice and facilitate the rational design of new sensors for use in protein, DNA, hybrid systems, and molecular devices. STATEMENT OF SIGNIFICANCE: Biomolecular structures involved in various biological processes, including muscle function and sensory perception, generate, convey, and respond to mechanical forces. In-vivo accurate measurement of these forces is challenging but needed to understand biological function. Here we present a comprehensive computational analysis of three different types of molecular force sensors used to report pico-Newton level forces in biomolecular systems. Our atom-level simulation predictions provide mechanistic insight that can facilitate experimental data interpretation, selection of sensor design parameters, and the development of new force sensors tailored to specific applications and environments.

Protocadherin-15 is a core protein component of inner-ear hair-cell tip links pulling on transduction channels essential for hearing and balance. Protocadherin-15 defects can result in non-syndromic deafness or Usher syndrome type 1F (USH1F) with hearing loss, balance deficits, and progressive blindness. Three rationally engineered shortened versions of protocadherin-15 (mini-PCDH15s) amenable for gene therapy have been used to rescue function in USH1F mouse models. Two can successfully or partially rescue hearing, while another one fails. Here we show that despite varying levels of hearing rescue, all three mini-PCDH15 versions can rescue hair-cell mechanotransduction. Negative-stain electron microscopy shows that all three versions form dimers like the wild-type protein, while crystal structures of some engineered fragments show that these can properly fold and bind calcium ions essential for function. In contrast, simulations predict distinct elasticities and nano differential scanning fluorimetry shows differences in melting temperature measurements. Our data suggest that elasticity and thermal stability are key determinants of sustained hearing rescue by mini-PCDH15s.