SNMF: Integrated Learning of Mutational Signatures and Prediction of DNA Repair Deficiencies

Motivation

Many tumours show deficiencies in DNA damage response (DDR), which influence tumorigenesis and progression, but also expose vulnerabilities with therapeutic potential. Assessing which patients might benefit from DDR-targeting therapy requires knowledge of tumour DDR deficiency status, with mutational signatures reportedly better predictors than loss of function mutations in select genes. However, signatures are identified independently using unsupervised learning, and therefore not optimised to distinguish between different pathway or gene deficiencies.

Results

We propose SNMF, a supervised non-negative matrix factorisation that jointly optimises the learning of signatures: (1) shared across samples, and (2) predictive of DDR deficiency. We applied SNMF to mutation profiles of human induced pluripotent cell lines carrying gene knockouts linked to three DDR pathways. The SNMF model achieved high accuracy (0.971) and learned more complete signatures of the DDR status of a sample, further discerning distinct mechanisms within a pathway. Cell line SNMF signatures recapitulated tumour-derived COSMIC signatures and predicted DDR pathway deficiency of TCGA tumours with high recall, suggesting that SNMF-like models can leverage libraries of induced DDR deficiencies to decipher intricate DDR signatures underlying patient tumours.

Availability

https://github.com/joanagoncalveslab/SNMF .

Proline, a unique amino acid whose polymer, polyproline II helix, and its analogues are involved in many biological processes: a review

Proline is a unique amino acid in that its side-chain is cyclised to the backbone, thus giving proline an exceptional rigidity and a considerably restricted conformational space. Polyproline forms two well-characterized helical structures: a left-handed polyproline helix (PPII) and a right-handed polyproline helix (PPI). Usually, sequences made only of prolyl residues are in PPII conformation, but even sequences not rich in proline but which are rich in glycine, lysine, glutamate, or aspartate have also a tendency to form PPII helices. Currently, the only way to study unambiguously PPII structure in solution is to use spectroscopies based on optical activity such as circular dichroism, vibrational circular dichroism and Raman optical activity. The importance of the PPII structure is emphasized by its ubiquitous presence in different organisms from yeast to human beings where proline-rich motifs and their binding domains are believed to be involved in vital biological processes. Some of the domains that are bound by proline-rich motifs include SH3 domains, WW domains, GYF domains and UEV domains, etc. The PPII structure has been demonstrated to be essential to biological activities such as signal transduction, transcription, cell motility, and immune response.

Systemic structural analysis of alterations reveals a common structural basis of driver mutations in cancer

A major effort in cancer research is to organize the complexities of the disease into fundamental traits. Despite conceptual progress in the last decades and the synthesis of hallmark features, no organizing principles governing cancer beyond cellular features exist. We analyzed experimentally determined structures harboring the most significant and prevalent driver missense mutations in human cancer, covering 73% (n = 168178) of the Catalog of Somatic Mutation in Cancer tumor samples (COSMIC). The results reveal that a single structural element-κ-helix (polyproline II helix)-lies at the core of driver point mutations, with significant enrichment in all major anatomical sites, suggesting that a small number of molecular traits are shared by most and perhaps all types of cancer. Thus, we uncovered the lowest possible level of organization at which carcinogenesis takes place at the protein level. This framework provides an initial scheme for a mechanistic understanding underlying the development of tumors and pinpoints key vulnerabilities.

A novel Pyk2-derived peptide inhibits invadopodia-mediated breast cancer metastasis

Dissemination of cancer cells from the primary tumor into distant body tissues and organs is the leading cause of death in cancer patients. While most clinical strategies aim to reduce or impede the growth of the primary tumor, no treatment to eradicate metastatic cancer exists at present. Metastasis is mediated by feet-like cytoskeletal structures called invadopodia which allow cells to penetrate through the basement membrane and intravasate into blood vessels during their spread to distant tissues and organs. The non-receptor tyrosine kinase Pyk2 is highly expressed in breast cancer, where it mediates invadopodia formation and function via interaction with the actin-nucleation-promoting factor cortactin. Here, we designed a cell-permeable peptide inhibitor that contains the second proline-rich region (PRR2) sequence of Pyk2, which binds to the SH3 domain of cortactin and inhibits the interaction between Pyk2 and cortactin in invadopodia. The Pyk2-PRR2 peptide blocks spontaneous lung metastasis in immune-competent mice by inhibiting cortactin tyrosine phosphorylation and actin polymerization-mediated maturation and activation of invadopodia, leading to reduced MMP-dependent tumor cell invasiveness. The native structure of the Pyk2-PRR2:cortactin-SH3 complex was determined using nuclear magnetic resonance (NMR), revealing an extended class II interaction surface spanning the canonical binding groove and a second hydrophobic surface which significantly contributes to ligand affinity. Using structure-guided design, we created a mutant peptide lacking critical residues involved in binding that failed to inhibit invadopodia maturation and function and consequent metastatic dissemination in mice. Our findings shed light on the specific molecular interactions between Pyk2 and cortactin and may lead to the development of novel strategies for preventing dissemination of primary breast tumors predicted at the time of diagnosis to be highly metastatic, and of secondary tumors that have already spread to other parts of the body.

The mechanism of action of the SSB interactome reveals it is the first OB-fold family of genome guardians in prokaryotes

The single-stranded DNA binding protein (SSB) is essential to all aspects of DNA metabolism in bacteria. This protein performs two distinct, but closely intertwined and indispensable functions in the cell. SSB binds to single-stranded DNA (ssDNA) and at least 20 partner proteins resulting in their regulation. These partners comprise a family of genome guardians known as the SSB interactome. Essential to interactome regulation is the linker/OB-fold network of interactions. This network of interactions forms when one or more PXXP motifs in the linker of SSB bind to an OB-fold in a partner, with interactome members involved in competitive binding between the linker and ssDNA to their OB-fold. Consequently, when linker-binding occurs to an OB-fold in an interactome partner, proteins are loaded onto the DNA. When linker/OB-fold interactions occur between SSB tetramers, cooperative ssDNA-binding results, producing a multi-tetrameric complex that rapidly protects the ssDNA. Within this SSB-ssDNA complex, there is an extensive and dynamic network of linker/OB-fold interactions that involves multiple tetramers bound contiguously along the ssDNA lattice. The dynamic behavior of these tetramers which includes binding mode changes, sliding as well as DNA wrapping/unwrapping events, are likely coupled to the formation and disruption of linker/OB-fold interactions. This behavior is essential to facilitating downstream DNA processing events. As OB-folds are critical to the essence of the linker/OB-fold network of interactions, and they are found in multiple interactome partners, the SSB interactome is classified as the first family of prokaryotic, oligosaccharide/oligonucleotide binding fold (OB-fold) genome guardians.

Structural basis of SARS-CoV-2 spike protein induced by ACE2

Motivation

The recent emergence of the novel SARS-coronavirus 2 (SARS-CoV-2) and its international spread pose a global health emergency. The spike (S) glycoprotein binds ACE2 and promotes SARS-CoV-2 entry into host cells. The trimeric S protein binds the receptor using the receptor-binding domain (RBD) causing conformational changes in S protein that allow priming by host cell proteases. Unraveling the dynamic structural features used by SARS-CoV-2 for entry might provide insights into viral transmission and reveal novel therapeutic targets. Using structures determined by X-ray crystallography and cryo-EM, we performed structural analysis and atomic comparisons of the different conformational states adopted by the SARS-CoV-2-RBD.

Results

Here, we determined the key structural components induced by the receptor and characterized their intramolecular interactions. We show that κ-helix (polyproline-II) is a predominant structure in the binding interface and in facilitating the conversion to the active form of the S protein. We demonstrate a series of conversions between switch-like κ-helix and β-strand, and conformational variations in a set of short α-helices which affect the hinge region. These conformational changes lead to an alternating pattern in conserved disulfide bond configurations positioned at the hinge, indicating a possible disulfide exchange, an important allosteric switch implicated in viral entry of various viruses, including HIV and murine coronavirus. The structural information presented herein enables to inspect and understand the important dynamic features of SARS-CoV-2-RBD and propose a novel potential therapeutic strategy to block viral entry. Overall, this study provides guidance for the design and optimization of structure-based intervention strategies that target SARS-CoV-2.

Availability

We have implemented the proposed methods in an R package freely available at https://github.com/Grantlab/bio3d

Supplementary information

Supplementary data are available at Bioinformatics online.

DNA Helicase-SSB Interactions Critical to the Regression and Restart of Stalled DNA Replication forks in Escherichia coli

In Escherichia coli, DNA replication forks stall on average once per cell cycle. When this occurs, replisome components disengage from the DNA, exposing an intact, or nearly intact fork. Consequently, the fork structure must be regressed away from the initial impediment so that repair can occur. Regression is catalyzed by the powerful, monomeric DNA helicase, RecG. During this reaction, the enzyme couples unwinding of fork arms to rewinding of duplex DNA resulting in the formation of a Holliday junction. RecG works against large opposing forces enabling it to clear the fork of bound proteins. Following subsequent processing of the extruded junction, the PriA helicase mediates reloading of the replicative helicase DnaB leading to the resumption of DNA replication. The single-strand binding protein (SSB) plays a key role in mediating PriA and RecG functions at forks. It binds to each enzyme via linker/OB-fold interactions and controls helicase-fork loading sites in a substrate-dependent manner that involves helicase remodeling. Finally, it is displaced by RecG during fork regression. The intimate and dynamic SSB-helicase interactions play key roles in ensuring fork regression and DNA replication restart.

κ-helix and the helical lock and key model: a pivotal way of looking at polyproline II

Motivation

Polyproline II (PPII) is a common conformation, comparable to α-helix and β-sheet. PPII, recently termed with a more generic name—κ-helix, adopts a left-handed structure with 3-fold rotational symmetry. Lately, a new type of binding mechanism—the helical lock and key model was introduced in SH3-domain complexes, where the interaction is characterized by a sliding helical pattern. However, whether this binding mechanism is unique only to SH3 domains is unreported.

Results

Here, we show that the helical binding pattern is a universal feature of the κ-helix conformation, present within all the major target families—SH3, WW, profilin, MHC-II, EVH1 and GYF domains. Based on a geometric analysis of 255 experimentally solved structures, we found that they are characterized by a distinctive rotational angle along the helical axis. Furthermore, we found that the range of helical pitch varies between different protein domains or peptide orientations and that the interaction is also represented by a rotational displacement mimicking helical motion. The discovery of rotational interactions as a mechanism, reveals a new dimension in the realm of protein–protein interactions, which introduces a new layer of information encoded by the helical conformation. Due to the extensive involvement of the conformation in functional interactions, we anticipate our model to expand the current molecular understanding of the relationship between protein structure and function.

Availability and implementation

We have implemented the proposed methods in an R package freely available at https://github.com/Grantlab/bio3d.

Supplementary information

Supplementary data are available at Bioinformatics online.