Biology subjects

Savinov, A.

Publications and source records attributed to Savinov, A..

6 recordsLinked to original sources

bioRxiv2026

AI-discovered protein fragments as generalizable regulators of biomolecular condensates

Biomolecular condensates are a major driver of cellular organization; however, we lack a predictable and systematic approach to modulate their underlying multivalent interactions. Here, we demonstrate a generalizable AI-driven method for designing protein fragments to control condensate formation, applying this approach across G3BP1, SARS-CoV-2 nucleocapsid, TDP-43, and focal adhesion kinase (FAK). Computationally screening 2,235 fragments, we selected 18 for experimental investigation, attaining a 50% success rate. Furthermore, predicted fragment binding modes align with their activities, revealing known and novel interactions driving condensate formation. For example, a fragment which suppresses FAK condensates in mammalian cells uncovered an interdomain interaction required for phase separation. Together, our results establish AI-guided protein fragment discovery as a generalizable strategy to dissect and control the molecular interactions that govern biomolecular condensates.

biophysics↗

bioRxiv2025

BoltzGen: Toward Universal Binder Design

We introduce BoltzGen, an all-atom generative model for designing proteins and peptides across all modalities to bind a wide range of biomolecular targets. BoltzGen builds strong structural reasoning capabilities about target-binder interactions into its generative design process. This is achieved by unifying design and structure prediction, resulting in a single model that also reaches state-of-the-art folding performance. BoltzGens generation process can be controlled with a flexible design specification language over covalent bonds, structure constraints, binding sites, and more. We experimentally validate these capabilities in eight diverse design campaigns with functional and affinity readouts across 26 targets. In our experiments, binder modalities span from nanobodies to disulfide-bonded peptides, and targets from disordered proteins to small molecules. In particular, we identify nanobody binders for novel targets with low similarity to proteins with already known bound structures. We release model weights, data, and both inference and training code at: https://github.com/HannesStark/boltzgen.

bioengineering↗

bioRxiv2023

High-throughput computational discovery of inhibitory protein fragments with AlphaFold

Peptides can bind to specific sites on larger proteins and thereby function as inhibitors and regulatory elements. Peptide fragments of larger proteins are particularly attractive for achieving these functions due to their inherent potential to form native-like binding interactions. Recently developed experimental approaches allow for high-throughput measurement of protein fragment inhibitory activity in living cells. However, it has thus far not been possible to predict de novo which of the many possible protein fragments bind to protein targets, let alone act as inhibitors. We have developed a computational method, FragFold, that employs AlphaFold to predict protein fragment binding to full-length proteins in a high-throughput manner. Applying FragFold to thousands of fragments tiling across diverse proteins revealed peaks of predicted binding along each protein sequence. Comparisons with experimental measurements establish that our approach is a sensitive predictor of fragment function: Evaluating inhibitory fragments from known protein-protein interaction interfaces, we find 87% are predicted by FragFold to bind in a native-like mode. Across full protein sequences, 68% of FragFold-predicted binding peaks match experimentally measured inhibitory peaks. Deep mutational scanning experiments support the predicted binding modes and uncover superior inhibitory peptides in high throughput. Further, FragFold is able to predict previously unknown protein binding modes, explaining prior genetic and biochemical data. The success rate of FragFold demonstrates that this computational approach should be broadly applicable for discovering inhibitory protein fragments across proteomes. Significance StatementPeptides can regulate protein interactions by binding to specific interfaces, and fragments of larger proteins have high potential to function in this manner. Recently developed experimental methods allow massively parallel measurement of protein fragment-based inhibition in vivo. However, we have lacked comparable computational methods to predict which protein fragments act as inhibitors and how they bind. Here we report a new approach, FragFold, which leverages high-throughput AlphaFold predictions of protein - fragment binding to tackle these problems at scale. FragFold is successful at predicting inhibitory protein fragments and their binding modes across diverse protein structures and functions. This new approach stands to enable proteome-wide discovery of inhibitory protein fragments and aid the interpretation of high-throughput experimental measurements of inhibitory activity. ClassificationBiological Sciences / Biophysics and Computational Biology

biophysics↗

bioRxiv2022

Mapping functional regions of essential bacterial proteins with dominant-negative protein fragments

Massively-parallel measurements of dominant negative inhibition by protein fragments have been used to map protein interaction sites and discover peptide inhibitors. However, the underlying principles governing fragment-based inhibition have thus far remained unclear. Here, we adapt a high-throughput inhibitory fragment assay for use in Escherichia coli, applying it to a set of ten essential proteins. This approach yielded single amino acid-resolution maps of inhibitory activity, with peaks localized to functionally important interaction sites, including oligomerization interfaces and folding contacts. Leveraging these data, we perform a systematic analysis to uncover principles of fragment-based inhibition. We determine a robust negative correlation between susceptibility to inhibition and cellular protein concentration, demonstrating that inhibitory fragments likely act primarily by titrating native protein interactions. We also characterize a series of trade-offs related to fragment length, showing that shorter peptides allow higher-resolution mapping but suffer from lower inhibitory activity. We employ an unsupervised statistical analysis to show that the inhibitory activities of protein fragments are largely driven not by generic properties such as charge, hydrophobicity, and secondary structure, but by the more specific characteristics of their bespoke macromolecular interactions. AlphaFold computational modeling of peptide complexes with one protein shows that the inhibitory activity of peptides is associated with their predicted ability to form native-like interactions. Overall, this work demonstrates fundamental characteristics of inhibitory protein fragment function and provides a foundation for understanding and controlling protein interactions in vivo. Significance StatementPeptide fragments derived from protein sequences can inhibit interactions of their parental proteins, providing a promising avenue for drug development. Here we employ a massively-parallel assay to measure in vivo inhibition by fragments that tile the full sequences of ten essential bacterial proteins. We leverage these data to decipher principles of fragment-based inhibition, showing how parental protein concentration drives activity and how protein fragment length interplays with activity and specificity. We employ statistical analysis to parse the roles of biophysical properties in fragment-to-fragment variation, and AlphaFold modeling to determine the relationship between measured inhibitory activity and predicted native-like binding. These results provide a path towards rational design of peptide inhibitors and broader principles of protein-protein interactions in living cells.

microbiology↗

bioRxiv2021

Use of Induced Pluripotent Stem Cells to Build Isogenic Systems and Investigate Type 1 Diabetes

Type 1 diabetes is a disease that arises due to complex immunogenetic mechanisms. Key cell-cell interactions involved in the pathogenesis of T1D are activation of autoreactive T cells by dendritic cells (DC), migration of T cells across endothelial cells (EC) lining capillary walls into the islets of Langerhans, interaction of T cells with macrophages in the islets, and killing of {beta}-cells by autoreactive CD8+ T cells. Overall, pathogenic cell-cell interactions are likely regulated by the individuals collection of genetic T1D-risk variants. To accurately model the role of genetics, it is essential to build systems to interrogate single candidate genes in isolation during the interactions of cells that are essential for disease development. However, obtaining single-donor matched cells relevant to T1D is a challenge. Sourcing these genetic variants from human induced pluripotent stem cells (iPSC) avoids this limitation. Herein, we have differentiated iPSC from one donor into DC, macrophages, EC, and {beta}-cells. Additionally, we also engineered T cell avatars from the same donor to provide an in vitro platform to study genetic influences on these critical cellular interactions. This proof of concept demonstrates the ability to derive an isogenic system from a single donor to study these relevant cell-cell interactions. Our system constitutes an interdisciplinary approach with a controlled environment that provides a proof-of-concept for future studies to determine the role of disease alleles (e.g. IFIH1, PTPN22, SH2B3, TYK2) in regulating cell-cell interactions and cell-specific contributions to the pathogenesis of T1D.

immunology↗

bioRxiv2021

Effects of sequence motifs in the yeast 3' untranslated region determined from massively-parallel assays of random sequences

The 3' untranslated region (UTR) plays critical roles in determining the level of gene expression, through effects on activities such as mRNA stability and translation. The underlying functional elements within this region have largely been identified through analyses of the limited number of native genes. To explore the effects of sequence elements when not present in biologically evolved sequence backgrounds, we analyzed hundreds of thousands of random 50-mers inserted into the 3' UTR of a reporter gene in the yeast Saccharomyces cerevisiae. We determined relative protein expression levels from the fitness of a library of transformants in a growth selection. We find that the consensus 3' UTR efficiency element significantly boosts expression, independent of sequence context; on the other hand, the consensus positioning element has only a small effect on expression. Some sequence motifs that are binding sites for Puf proteins substantially increase expression in this random library, despite these proteins generally being associated with post-transcriptional downregulation when bound to native mRNAs. Thus, the regulatory effects of 3' UTR sequence features like the positioning element and Puf binding sites appear to be strongly dependent on their context within native genes, where they exist alongside co-evolved sequence features. Our measurements also allowed a systematic examination of the effects of point mutations within efficiency element motifs across diverse sequence backgrounds. These mutational scans reveal the relative in vivo importance of individual bases in the efficiency element, which likely reflects their roles in binding the Hrp1 protein involved in cleavage and polyadenylation.

genomics↗