Contrastive learning in protein language space predicts interactions between drugs and protein targets.

Sequence-based prediction of drug-target interactions has the potential to accelerate drug discovery by complementing experimental screens. Such computational prediction needs to be generalizable and scalable while remaining sensitive to subtle variations in the inputs. However, current computational techniques fail to simultaneously meet these goals, often sacrificing performance of one to achieve the others. We develop a deep learning model, ConPLex, successfully leveraging the advances in pretrained protein language models ("PLex") and employing a protein-anchored contrastive coembedding ("Con") to outperform state-of-the-art approaches. ConPLex achieves high accuracy, broad adaptivity to unseen data, and specificity against decoy compounds. It makes predictions of binding based on the distance between learned representations, enabling predictions at the scale of massive compound libraries and the human proteome. Experimental testing of 19 kinase-drug interaction predictions validated 12 interactions, including four with subnanomolar affinity, plus a strongly binding EPHB1 inhibitor (KD = 1.3 nM). Furthermore, ConPLex embeddings are interpretable, which enables us to visualize the drug-target embedding space and use embeddings to characterize the function of human cell-surface proteins. We anticipate that ConPLex will facilitate efficient drug discovery by making highly sensitive in silico drug screening feasible at the genome scale. ConPLex is available open source at https://ConPLex.csail.mit.edu.

split-intein Gal4 provides intersectional genetic labeling that is repressible by Gal80.

The split-Gal4 system allows for intersectional genetic labeling of highly specific cell types and tissues in Drosophila. However, the existing split-Gal4 system, unlike the standard Gal4 system, cannot be repressed by Gal80, and therefore cannot be controlled temporally. This lack of temporal control precludes split-Gal4 experiments in which a genetic manipulation must be restricted to specific timepoints. Here, we describe a split-Gal4 system based on a self-excising split-intein, which drives transgene expression as strongly as the current split-Gal4 system and Gal4 reagents, yet which is repressible by Gal80. We demonstrate the potent inducibility of "split-intein Gal4" in vivo using both fluorescent reporters and via reversible tumor induction in the gut. Further, we show that our split-intein Gal4 can be extended to the drug-inducible GeneSwitch system, providing an independent method for intersectional labeling with inducible control. We also show that the split-intein Gal4 system can be used to generate highly cell type-specific genetic drivers based on in silico predictions generated by single-cell RNAseq (scRNAseq) datasets, and we describe an algorithm ("Two Against Background" or TAB) to predict cluster-specific gene pairs across multiple tissue-specific scRNA datasets. We provide a plasmid toolkit to efficiently create split-intein Gal4 drivers based on either CRISPR knock-ins to target genes or using enhancer fragments. Altogether, the split-intein Gal4 system allows for the creation of highly specific intersectional genetic drivers that are inducible/repressible.

Adducin Regulates Sarcomere Disassembly During Cardiomyocyte Mitosis.

BACKGROUND: Recent interest in understanding cardiomyocyte cell cycle has been driven by potential therapeutic applications in cardiomyopathy. However, despite recent advances, cardiomyocyte mitosis remains a poorly understood process. For example, it is unclear how sarcomeres are disassembled during mitosis to allow the abscission of daughter cardiomyocytes. METHODS: Here, we use a proteomics screen to identify adducin, an actin capping protein previously not studied in cardiomyocytes, as a regulator of sarcomere disassembly. We generated many adeno-associated viruses and cardiomyocyte-specific genetic gain-of-function models to examine the role of adducin in neonatal and adult cardiomyocytes in vitro and in vivo. RESULTS: We identify adducin as a regulator of sarcomere disassembly during mammalian cardiomyocyte mitosis. α/γ-adducins are selectively expressed in neonatal mitotic cardiomyocytes, and their levels decline precipitously thereafter. Cardiomyocyte-specific overexpression of various splice isoforms and phospho-isoforms of α-adducin in vitro and in vivo identified Thr445/Thr480 phosphorylation of a short isoform of α-adducin as a potent inducer of neonatal cardiomyocyte sarcomere disassembly. Concomitant overexpression of this α-adducin variant along with γ-adducin resulted in stabilization of the adducin complex and persistent sarcomere disassembly in adult mice, which is mediated by interaction with α-actinin. CONCLUSIONS: These results highlight an important mechanism for coordinating cytoskeletal morphological changes during cardiomyocyte mitosis.

S6K1 deficiency in tumor stroma impairs lung metastasis of melanoma in mice.

Accumulating data suggest that ribosomal protein S6 kinase 1 (S6K1), an effector in the mammalian target of rapamycin (mTOR) pathway, plays pleiotropic roles in tumor progression. However, to date, while the tumorigenic function of S6K1 in tumor cells has been well elucidated, its role in the tumor stroma remains poorly understood. We recently showed that S6K1 mediates vascular endothelial growth factor A (VEGF-A) production in macrophages, thereby supporting tumor angiogenesis and growth. As macrophage-derived VEGF-A is crucial for both tumor cell intravasation and extravasation across the vascular endothelium, our previous findings suggest that stromal S6K1 signaling is required for tumor metastatic spread. Therefore, we aimed to determine the impact of host S6K1 depletion on tumor metastasis using a murine model of pulmonary metastasis (S6k1-/- mice implanted with B16F10 melanoma). The ablation of S6K1 in the host microenvironment significantly reduced the metastasized B16F10 melanoma cells on the lung surface in both spontaneous and intravenous lung metastasis mouse models without affecting the incidence of metastasis to distant lymph nodes. In addition, stromal S6K1 loss decreased the number of tumor cells circulating in the peripheral blood of mice bearing B16F10 xenografts without affecting the vascular leakage induced by VEGF-A in vivo. These observations demonstrate that S6K1 signaling in host cells other than endothelial cells is required to modulate the host microenvironment to facilitate the metastatic spread of tumors via blood circulation, thus revealing its novel role in the tumor stroma during tumor progression.

TT3D: Leveraging precomputed protein 3D sequence models to predict protein-protein interactions.

MOTIVATION: High-quality computational structural models are now precomputed and available for nearly every protein in UniProt. However, the best way to leverage these models to predict which pairs of proteins interact in a high-throughput manner is not immediately clear. The recent Foldseek method of van Kempen et al. encodes the structural information of distances and angles along the protein backbone into a linear string of the same length as the protein string, using tokens from a 21-letter discretized structural alphabet (3Di). RESULTS: We show that using both the amino acid sequence and the 3Di sequence generated by Foldseek as inputs to our recent deep-learning method, Topsy-Turvy, substantially improves the performance of predicting protein-protein interactions cross-species. Thus TT3D (Topsy-Turvy 3D) presents a way to reuse all the computational effort going into producing high-quality structural models from sequence, while being sufficiently lightweight so that high-quality binary protein-protein interaction predictions across all protein pairs can be made genome-wide. AVAILABILITY AND IMPLEMENTATION: TT3D is available at https://github.com/samsledje/D-SCRIPT. An archived version of the code at time of submission can be found at https://zenodo.org/records/10037674.

Measured Steps: Navigating the Path of Oligoprogressive Lung Cancer with Targeted and Immunotherapies.

PURPOSE OF REVIEW: This review discusses the definitions, treatment modalities, management, future directions, and ongoing clinical trials of oligoprogressive disease in oncogene-driven and non-oncogene-driven NSCLC. RECENT FINDINGS: During the last decades, diagnostic and treatment modalities for oligometastatic NSCLC have advanced significantly, leading to improved survival. Additionally, our understanding of the tumor biology of oligoprogressive disease has expanded. However, despite the efforts of organizations, such as EORTC, ESTRO, and ASTRO proposing definitions for oligometastatic and oligoprogressive disease, heterogeneity in definitions persists in (ongoing) trials. Recognizing the significance of subclassification within oligoprogressive disease in NSCLC and the varying risks associated with subsequent metastatic spread, there is a call for tailored management strategies. A consensus on standardized criteria for the definition of oligoprogressive disease is urgently needed and will not only facilitate meaningful comparisons between studies but also pave the way for the development of personalized treatment plans that take into account the heterogeneous nature of oligoprogressive disease.

Preliminary investigation reveals novel pathological consequences of bluetongue virus-1 infection in the endocrine glands of pregnant Indian sheep.

Bluetongue virus (BTV), a major peril to the sheep industry, infects a wide range of the cells in the infected animals including mononuclear, dendritic and epithelial cells. However, little is known about its tropism for the secretory epithelial cells of endocrine glands and the pathogenesis it induces. The aim of the study was to assess the BTV load, antigen distribution in the tissue of the pituitary, thyroid as well as adrenal glands and associated histopathological consequences. BTV antigens were localized using immunohistochemistry in the thyroid's epithelial cells, zona fasciculata and zona reticularis cells and the anterior pituitary epithelial cells. The real-time PCR portrayed the high viral load in adrenals at 7th days postinoculation (DPI) and in thyroid and pituitary glands at 15th DPI. Serum examination revealed variation in the T-3 and T-4 of infected animals in comparison to the control group. Caspase-3 immunolocalization revealed BTV-1 induces apoptosis in the affected cells of endocrine gland of infected animals. Further, this study signifies the tropism of BTV in the novel sites (endocrine glands) of the host that might be one of the reasons for the poor performance of infected animals.

Retrieval of high-resolution aerosol optical depth (AOD) using Landsat 8 imageries over different LULC classes over a city along Indo-Gangetic Plain, India.

Aerosol optical depth (AOD) serves as a crucial indicator for assessing regional air quality. To address regional and urban pollution issues, there is a requirement for high-resolution AOD products, as the existing data is of very coarse resolution. To address this issue, we retrieved high-resolution AOD over Kanpur (26.4499°N, 80.3319°E), located in the Indo-Gangetic Plain (IGP) region using Landsat 8 imageries and implemented the algorithm SEMARA, which combines SARA (Simplified Aerosol Retrieval Algorithm) and SREM (Simplified and Robust Surface Reflectance Estimation). Our approach leveraged the green band of the Landsat 8, resulting in an impressive spatial resolution of 30 m of AOD and rigorously validated with available AERONET observations. The retrieved AOD is in good agreement with high correlation coefficients (r) of 0.997, a low root mean squared error of 0.035, and root mean bias of - 4.91%. We evaluated the retrieved AOD with downscaled MODIS (MCD19A2) AOD products across various land classes for cropped and harvested period of agriculture cycle over the study region. It is noticed that over the built-up region of Kanpur, the SEMARA algorithm exhibits a stronger correlation with the MODIS AOD product compared to vegetation, barren areas and water bodies. The SEMARA approach proved to be more effective for AOD retrieval over the barren and built-up land categories for harvested period compared with the cropping period. This study offers a first comparative examination of SEMARA-retrieved high-resolution AOD and MODIS AOD product over a station of IGP.

Exploring the presence of bovine leukemia virus among breast cancer tumors in a rural state.

PURPOSE: The bovine leukemia virus (BLV) is a deltaretrovirus that causes malignant lymphoma and lymphosarcomas in cattle globally and has high prevalence among large scale U.S. dairy herds. Associations between presence of BLV DNA in human mammary tissue and human breast cancer incidence have been reported. We sought to estimate the prevalence of BLV DNA in breast cancer tissue samples in a rural state with an active dairy industry. METHODS: We purified genomic DNA from 56 fresh-frozen breast cancer tissue samples (51 tumor samples, 5 samples representing adjacent normal breast tissue) banked between 2016 and 2019. Using nested PCR assays, multiple BLV tax sequence primers and primers for the long terminal repeat (LTR) were used to detect BLV DNA in tissue samples and known positive control samples, including the permanently infected fetal lamb kidney cell line (FLK-BLV) and blood from BLV positive cattle. RESULTS: The median age of patients from which samples were obtained at the time of treatment was 60 (40-93) and all were female. Ninety percent of patients had invasive ductal carcinoma. The majority were poorly differentiated (60%). On PCR assay, none of the tumor samples tested positive for BLV DNA, despite having consistent signals in positive controls. CONCLUSION: We did not find BLV DNA in fresh-frozen breast cancer tumors from patients presenting to a hospital in Vermont. Our findings suggest a low prevalence of BLV in our patient population and a need to reevaluate the association between BLV and human breast cancer.

Topsy-Turvy: integrating a global view into sequence-based PPI prediction.

SUMMARY: Computational methods to predict protein-protein interaction (PPI) typically segregate into sequence-based 'bottom-up' methods that infer properties from the characteristics of the individual protein sequences, or global 'top-down' methods that infer properties from the pattern of already known PPIs in the species of interest. However, a way to incorporate top-down insights into sequence-based bottom-up PPI prediction methods has been elusive. We thus introduce Topsy-Turvy, a method that newly synthesizes both views in a sequence-based, multi-scale, deep-learning model for PPI prediction. While Topsy-Turvy makes predictions using only sequence data, during the training phase it takes a transfer-learning approach by incorporating patterns from both global and molecular-level views of protein interaction. In a cross-species context, we show it achieves state-of-the-art performance, offering the ability to perform genome-scale, interpretable PPI prediction for non-model organisms with no existing experimental PPI data. In species with available experimental PPI data, we further present a Topsy-Turvy hybrid (TT-Hybrid) model which integrates Topsy-Turvy with a purely network-based model for link prediction that provides information about species-specific network rewiring. TT-Hybrid makes accurate predictions for both well- and sparsely-characterized proteins, outperforming both its constituent components as well as other state-of-the-art PPI prediction methods. Furthermore, running Topsy-Turvy and TT-Hybrid screens is feasible for whole genomes, and thus these methods scale to settings where other methods (e.g. AlphaFold-Multimer) might be infeasible. The generalizability, accuracy and genome-level scalability of Topsy-Turvy and TT-Hybrid unlocks a more comprehensive map of protein interaction and organization in both model and non-model organisms. AVAILABILITY AND IMPLEMENTATION: https://topsyturvy.csail.mit.edu. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.