Biology subjects

Rao, R.

Publications and source records attributed to Rao, R..

At least 37 records · Page 2Linked to original sources

bioRxiv2022

Language models of protein sequences at the scale of evolution enable accurate structure prediction

Artificial intelligence has the potential to open insight into the structure of proteins at the scale of evolution. It has only recently been possible to extend protein structure prediction to two hundred million cataloged proteins. Characterizing the structures of the exponentially growing billions of protein sequences revealed by large scale gene sequencing experiments would necessitate a break-through in the speed of folding. Here we show that direct inference of structure from primary sequence using a large language model enables an order of magnitude speed-up in high resolution structure prediction. Leveraging the insight that language models learn evolutionary patterns across millions of sequences, we train models up to 15B parameters, the largest language model of proteins to date. As the language models are scaled they learn information that enables prediction of the three-dimensional structure of a protein at the resolution of individual atoms. This results in prediction that is up to 60x faster than state-of-the-art while maintaining resolution and accuracy. Building on this, we present the ESM Metage-nomic Atlas. This is the first large-scale structural characterization of metagenomic proteins, with more than 617 million structures. The atlas reveals more than 225 million high confidence predictions, including millions whose structures are novel in comparison with experimentally determined structures, giving an unprecedented view into the vast breadth and diversity of the structures of some of the least understood proteins on earth.

synthetic biology↗

bioRxiv2022

Measurement of Cervical Neuronal Activity during Stress Challenge Using Novel Flexible Adhesive Surface Electrodes

This study introduces a flexible, adhesive-integrated electrode array that was developed to enable non-invasive monitoring of cervical nerve activity. The device uses silver-silver chloride as the electrode material of choice and combines it with a novel electrode array consisting of a customized biopotential data acquisition unit and integrated graphical user interface (GUI) for visualization of real-time monitoring. Preliminary testing demonstrated this novel electrode design can achieve a high signal to noise ratio during cervical neural recordings. To demonstrate the capability of the surface electrodes to detect changes in cervical neuronal activity, the cold-pressor test (CPT) and a timed respiratory challenge were employed as stressors to the autonomic nervous system. This sensor system recording, a new novel technique, was termed Cervical Electroneurography (CEN). By applying a custom spike sorting algorithm to the novel electrode measurements, neural activity was classified in two ways: 1) pre-to-post CPT, and 2) during a timed respiratory challenge. Unique to this work: 1) rostral to caudal channel position-specific (cephalad to caudal) firing patterns and 2) cross challenge biotype-specific change in average CEN firing, were observed with both CPT and the timed respiratory challenge. Future work is planned to develop an ambulatory CEN recording device that could provide immediate notification of autonomic nervous system activity changes that might indicate autonomic dysregulation in healthy subjects and clinical disease states.

physiology↗

bioRxiv2022

Combination of the Systemin peptide with the beneficial fungus Trichoderma afroharzianum T22 improves plant defense responses against pests and diseases

Trichoderma spp. are among the most widely used plant beneficial fungi in agriculture. A novel approach to enhance their effectiveness in plant defense is to use the fungi in combination with bioactive molecules including plant-derived compounds. Here, we show that plant treatment with Trichoderma afroharzianum (strain T22) and Systemin (Sys), a tomato plant peptide active in triggering plant defense, confers protection against the fungal pathogens Fusarium oxysporum, Botrytis cinerea and the insect pest Tuta absoluta. The observed defensive response was associated with increased accumulation of metabolites and transcripts involved in the Jasmonic acid (JA) pathway. Our findings suggest that the innovative combination of T. afroharzianum T22 and Sys can result in a more effective and robust control of different biotic stress agents.

genetics↗

bioRxiv2021

End-to-end learning of multiple sequence alignments with differentiable Smith-Waterman

Multiple Sequence Alignments (MSAs) of homologous sequences contain information on structural and functional constraints and their evolutionary histories. Despite their importance for many downstream tasks, such as structure prediction, MSA generation is often treated as a separate pre-processing step, without any guidance from the application it will be used for. Here, we implement a smooth and differentiable version of the Smith-Waterman pairwise alignment algorithm that enables jointly learning an MSA and a downstream machine learning system in an end-to-end fashion. To demonstrate its utility, we introduce SMURF (Smooth Markov Unaligned Random Field), a new method that jointly learns an alignment and the parameters of a Markov Random Field for unsupervised contact prediction. We find that SMURF learns MSAs that mildly improve contact prediction on a diverse set of protein and RNA families. As a proof of concept, we demonstrate that by connecting our differentiable alignment module to AlphaFold and maximizing predicted confidence, we can learn MSAs that improve structure predictions over the initial MSAs. Interestingly, the alignments that improve AlphaFold predictions are self-inconsistent and can be viewed as adversarial. This work highlights the potential of differentiable dynamic programming to improve neural network pipelines that rely on an alignment and the potential dangers of relying on black-box methods for optimizing predictions of protein sequences.

bioinformatics↗

bioRxiv2021

Evolutionary Dynamics, Evolutionary Forces, and Robustness: A Nonequilibrium Statistical Mechanics Perspective

Any realistic evolutionary theory has to consider: (i) the dynamics of organisms that reproduce and possess heritable traits; (ii) the appearance of stochastic variations in these traits; and (iii) the selection of those organisms that better survive and reproduce. These elements shape the "evolutionary forces" that characterize the evolutionary dynamics. Here, we introduce a general model of reproduction-variation-selection dynamics. By treating these dynamics as a non-equilibrium thermodynamic process, we make precise the notion of the forces that characterize evolution. One of these forces, in particular, can be associated with the robustness of reproduction to variations. The emergence of this trait in our model--without any explicit selection for it--is an example of a general phenomenon, which can be called enaptation, distinct from the well-known and studied phenomena of adaptation and exaptation. Some of the detailed predictions of our model can be tested by quantitative laboratory experiments, similar to those performed in the past on evolving populations of proteins or viruses.

evolutionary biology↗

bioRxiv2021

The Endosomal pH Regulator NHE9 is a Driver of Stemness in Glioblastoma

A small population of self-renewing stem cells initiate tumors and maintain therapeutic resistance in glioblastoma. Given the limited treatment options and dismal prognosis for this disease there is urgent need to identify drivers of stem cells that could be druggable targets. Previous work showed that the endosomal pH regulator NHE9 is upregulated in glioblastoma and correlates with worse survival prognosis. Here, we probed for aberrant signaling pathways in patient-derived glioblastoma cells and found that NHE9 increases cell surface expression and phosphorylation of multiple receptor tyrosine kinases by promoting their escape from lysosomal degradation. Downstream of NHE9-mediated receptor activation, oncogenic signaling pathways converged on the JAK2-STAT3 transduction axis to induce pluripotency genes Oct4 and Nanog and suppress markers of glial differentiation. We used both genetic and chemical approaches to query the role of endosomal pH in glioblastoma phenotypes. Loss-of-function mutations in NHE9 that failed to alkalinize endosomal lumen did not increase self-renewal capacity of gliomaspheres in vitro. However, monensin, a chemical mimetic of Na+/H+ exchanger activity, and the H+ pump inhibitor bafilomycin bypassed NHE9 to directly alkalinize the endosomal lumen resulting in stabilization of receptor tyrosine kinases and induction of Oct4 and Nanog. Using orthotopic models of primary glioblastoma cells we found that NHE9 increased tumor initiation in vivo. We propose that NHE9 initiates inside-out signaling from the endosomal lumen, distinct from the established effects of cytoplasmic and extracellular pH on tumorigenesis. Endosomal pH may be an attractive therapeutic target that diminishes stemness in glioblastoma, agnostic of specific receptor subtype. SignificanceA well-known hallmark of cancer is excessive acidification of tumor microenvironment, caused by upregulation of Na+/H+ exchanger activity on the cancer cell membrane. However, the role of organellar pH in tumor biology has not been established. This study identifies a mechanistic link between upregulation of the endosomal Na+/H+ exchanger NHE9 and stemness properties in glioblastoma, the most malignant and common brain tumor in adults. By increasing pH of the recycling endosome, NHE9 exerts a broad effect on post-translational stability and activation of multiple receptor tyrosine kinases, leading to increased stem cell-like properties of self-renewal and tumor initiation in glioblastoma models. Our findings suggest that targeting NHE9 or endosomal pH could be an effective strategy for receptor agnostic glioblastoma treatment.

cancer biology↗

bioRxiv2021

Tomato Prosystemin is much more than a simple Systemin precursor

Systemin (Sys) is an octadecapeptide which, upon wounding, is released from the carboxy terminus of its precursor, prosystemin(ProSys) to promote plant defenses. Recent findings on the disordered structure of ProSysprompted us to investigate a putative biological role of the whole precursor deprived of Sys peptide. We produced transgenic tomato plants expressing a truncated ProSys gene in which the exon coding for Sys was removed and compared their defense response with that induced by the exogenous application of the recombinant deleted ProSys[ProSys(1-178)].By combining protein structure analyses, transcriptomic analysis, gene expression profiling and bioassays with different pests we demonstrate that the truncated ProSys, that does not induce the endogenous ProSys gene, promotes defense barriers in tomato plants through a hormone independent defense pathway, likely associated with the production of oligogalacturonides (OGs). Both transgenic and plants treated with the recombinant protein showed the modulation of the expression of genes linked with defense responses and resulted protected against the lepidopteran pest Spodoptera littoralis and the fungus Botrytis cinerea. Our results suggest that the overall function of the wild type prosystemin is more complex than previously shown as it might activate at least two tomato defense pathways: the well-known Sys-dependent pathway connected with the induction of JA biosynthesis and the successive activation of a set of defense-related genes and the ProSys(1-178)-dependent pathway associated with OGs production leading to the OGs mediate plant immunity.

genetics↗

bioRxiv2021

The genome trilogy of Anopheles stephensi, an urban malaria vector, reveals structure of a locus associated with adaptation to environmental heterogeneity

BackgroundAnopheles stephensi is the most menacing malaria vector to watch for in newly urbanizing parts of the world. The fitness is reported to be a direct consequence of the vector adapting to laying eggs in over-head water tanks with street-side water puddles polluted by oil and sewage. Large frequent inversions of malaria vectors are implicated in adaptation. ResultsWe report the assembly of a strain of An. stephensi of the type-form, collected from a construction site from Chennai (IndCh) in 2016. The genome completes the trilogy with respect to a 16 Mbp inversion (2Rb) in An. stephensi associated with adaptation to environmental heterogeneity. Comparative genome analysis revealed breakpoint structure and allowed extraction of 22,650 segregating SNPs for typing this inversion. Using whole genome sequencing of 82 individual mosquitoes, we conclude that one third of both wild and laboratory populations maintain heterozygous genotype of 2Rb. The large number of SNPs are tailored to assign inversion genotype directly from 1740 exonic SNPs 80% of which are expressed in various developmental stages. ConclusionsThe genome trilogy approach accelerates study of fine structure and typing of important inversions in malaria vectors putting the genome resources for the much understudied An. stephensi, on par with the extensively studied malaria vector, Anopheles gambiae. We argue that the IndCh genome is relevant for field translation work compared to those reported earlier by showing that individuals from diverse populations cluster with IndCh pointing to significant commerce between cities, perhaps, allowing for survival of the fittest strain.

genomics↗

bioRxiv2021

Language models enable zero-shot prediction of the effects of mutations on protein function

Modeling the effect of sequence variation on function is a fundamental problem for understanding and designing proteins. Since evolution encodes information about function into patterns in protein sequences, unsupervised models of variant effects can be learned from sequence data. The approach to date has been to fit a model to a family of related sequences. The conventional setting is limited, since a new model must be trained for each prediction task. We show that using only zero-shot inference, without any supervision from experimental data or additional training, protein language models capture the functional effects of sequence variation, performing at state-of-the-art.

synthetic biology↗

bioRxiv2021

Detection of Small Fiber Neuronal Activity with Optically Pumped Magnetometers

Electrodiagnosis is routinely integrated into clinical neurophysiology practice for peripheral nerve disease diagnoses such as neuropathy, demyelinating disorders, nerve entrapment/impingement, plexopathy or radiculopathy. Measured with conventional surface electrodes, the propagation of peripheral nerve action potentials along a nerve is the result of ionic current flow which, according to Amperes Law, generates a small magnetic field that is also detected as an "action current" by magnetometers such as superconducting quantum interference device (SQUID) Magnetoencephalography (MEG) systems. Optically pumped magnetometers (OPMs) are an emerging class of quantum magnetic sensors with a demonstrated sensitivity at the 1 fT/{surd}Hz level, capable of cortical action current detection. But OPMs were ostensibly constrained to low bandwidth therefore precluding their use in peripheral nerve electrodiagnosis. With careful OPM bandwidth characterization, we hypothesized OPMs may also detect compound action current signatures consistent with both Sensory Nerve Action Potential (SNAP) and the Hoffmann Reflex (H-Reflex). In as much, our work confirms OPMs enabled with expanded bandwidth can detect the magnetic signature of both the SNAP and H-Reflex. Taken together, OPMs now show potential as an emerging electrodiagnostic tool.

neuroscience↗

bioRxiv2021

SLIT2-ROBO signaling in tumor-associated microglia/macrophages drives glioblastoma immunosuppression and vascular dysmorphia

SLIT2 is a secreted polypeptide that guides migration of cells expressing ROBO1&2 receptors. Herein, we investigated SLIT2/ROBO signaling effects in gliomas. In patients with glioblastoma (GBM), SLIT2 expression increased with malignant progression and correlated with poor survival and immunosuppression. Knockdown of SLIT2 in mouse glioma cells and patient derived GBM xenografts reduced tumor growth and synergized with immunotherapy to prolong survival. Tumor cell SLIT2 knockdown inhibited macrophage invasion and promoted a cytotoxic gene expression profile, which improved tumor vessel function and enhanced efficacy of chemotherapy and immunotherapy. Mechanistically, SLIT2 promoted microglia/macrophage chemotaxis and tumor-supportive polarization via ROBO1&2-mediated PI3K{gamma} activation. Macrophage Robo1&2 deletion and systemic SLIT2 trap delivery mimicked SLIT2 knockdown effects on tumor growth and the tumor microenvironment (TME), revealing SLIT2 signaling through macrophage ROBOs as a novel regulator of the GBM microenvironment and a potential immunotherapeutic target for brain tumors.

cancer biology↗

bioRxiv2021

MSA Transformer

Unsupervised protein language models trained across millions of diverse sequences learn structure and function of proteins. Protein language models studied to date have been trained to perform inference from individual sequences. The longstanding approach in computational biology has been to make inferences from a family of evo lutionarily related sequences by fitting a model to each family independently. In this work we combine the two paradigms. We introduce a protein language model which takes as input a set of sequences in the form of a multiple sequence alignment. The model interleaves row and column attention across the input sequences and is trained with a variant of the masked language modeling objective across many protein families. The performance of the model surpasses current state-of-the-art unsupervised structure learning methods by a wide margin, with far greater parameter efficiency than prior state-of-the-art protein language models.

synthetic biology↗

bioRxiv2020

Single Layers of Attention Suffice to Predict Protein Contacts

AO_SCPLOWBSTRACTC_SCPLOWThe established approach to unsupervised protein contact prediction estimates co-evolving positions using undirected graphical models. This approach trains a Potts model on a Multiple Sequence Alignment, then predicts that the edges with highest weight correspond to contacts in the 3D structure. On the other hand, increasingly large Transformers are being pretrained on protein sequence databases but have demonstrated mixed results for downstream tasks, including contact prediction. This has sparked discussion about the role of scale and attention-based models in unsupervised protein representation learning. We argue that attention is a principled model of protein interactions, grounded in real properties of protein family data. We introduce a simplified attention layer, factored attention, and show that it achieves comparable performance to Potts models, while sharing parameters both within and across families. Further, we extract contacts from the attention maps of a pretrained Transformer and show they perform competitively with the other two approaches. This provides evidence that large-scale pretraining can learn meaningful protein features when presented with unlabeled and unaligned data. We contrast factored attention with the Transformer to indicate that the Transformer leverages hierarchical signal in protein family databases not captured by our single-layer models. This raises the exciting possibility for the development of powerful structured models of protein family databases.1

bioinformatics↗

bioRxiv2020

A systematic analysis of rodent models implicates adipogenesis and innate immunity in pathogenesis of fatty liver disease

Animal models of human disease are a key component of translational research and yet there is often no consensus on which model is optimal for a particular disease. Here, we generated a database of 3,920 rodent models of non-alcoholic fatty liver disease (NAFLD). Study designs were highly heterogeneous therefore few models had been cited more than once. Analysis of genetic models provided evidence for the role of adipose dysfunction and perturbation of the innate immune system in the progression of NAFLD. We identified that high-fat, high-fructose diets most closely recapitulate the human phenotype of NAFLD. There was substantial variability in the nomenclature of animal models; a consensus on terminology of specialist diets is needed. More broadly, this analysis demonstrates the variability in preclinical study design, which has implications for the reproducibility of in vivo experiments.

physiology↗

bioRxiv2020

Epigenetic Modulation of SPCA2 Reverses Epithelial to Mesenchymal Transition in Breast Cancer Cells

The secretory pathway Ca2+-ATPase SPCA2 is a tumor suppressor in triple receptor negative breast cancer (TNBC), a highly aggressive molecular subtype that lacks tailored treatment options. Low expression of SPCA2 in TNBC confers poor survival prognosis in patients. Previous work has established that re-introducing SPCA2 to TNBC cells restores basal Ca2+ signaling, represses mesenchymal gene expression, mitigates tumor migration in vitro and metastasis in vivo. In this study, we examined the effect of histone deacetylase inhibitors (HDACi) in TNBC cell lines. We show that the pan-HDACi vorinostat and the class I HDACi romidepsin induce dose-dependent upregulation of SPCA2 transcript with concurrent downregulation of mesenchymal markers and tumor cell migration characteristic of epithelial phenotype. Silencing SPCA2 abolished the ability of HDACi to reverse epithelial to mesenchymal transition (EMT). Independent of ATPase activity, SPCA2 elevated resting Ca2+ levels to activate downstream components of non-canonical Wnt/Ca2+ signaling. HDACi treatment led to SPCA2-dependent phosphorylation of CAMKII and {beta}-catenin, turning Wnt signaling off. We conclude that SPCA2 mediates the efficacy of HDACi in reversing EMT in TNBC by a novel mode of non-canonical Wnt/Ca2+ signaling. Our findings provide incentive for screening epigenetic modulators that exploit Ca2+ signaling pathways to reverse EMT in breast tumors. Simple SummaryThe triple receptor negative breast cancer subtype, or TNBC, currently has no tailored treatment options. TNBC is highly metastatic, associated with high patient mortality, and disproportionately occurs in Black/African American women where it contributes to racial disparities in health outcomes. Therefore, we focused on new therapeutic approaches to TNBC. We discovered that levels of the Calcium-ATPase SPCA2 are abnormally low in TNBC and that these low levels correlate with poor survival prognosis in patients. Previously, we showed that recombinant SPCA2 prevented TNBC cells from acquiring aggressive mesenchymal properties associated with metastasis both in vitro and in vivo. These findings motivated us to search for drugs that turn the SPCA2 gene back on in TNBC cells. In this study, we show that histone deacetylase inhibitors increase SPCA2 levels, activate Ca2+ signaling and convert cancer cells to a less aggressive epithelial state. These findings could lead to new treatment options for TNBC. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=140 HEIGHT=200 SRC="FIGDIR/small/379685v1_ufig1.gif" ALT="Figure 1"> View larger version (19K): org.highwire.dtl.DTLVardef@13b2565org.highwire.dtl.DTLVardef@14317forg.highwire.dtl.DTLVardef@1bdbd34org.highwire.dtl.DTLVardef@1d0bd83_HPS_FORMAT_FIGEXP M_FIG C_FIG

cancer biology↗

bioRxiv2020

Protocol for Magnetic Resonance Imaging Acquisition, Quality Assurance, and Quality Check for the Accelerator Program for Discovery in Brain Disorders using Stem Cells

ObjectiveThe Accelerator Program for Discovery in Brain Disorders using Stem Cells (ADBS) is a longitudinal study focused on collecting and analysing clinical, neuropsychological, neurophysiological, and multimodal neuroimaging data from five cohorts of patients with major psychiatric disorders from genetically high-risk families, their unaffected first-degree relatives, and healthy subjects. Here, we present a complete description of the acquisition of multimodal MRI data along with the quality assurance (QA) and quality check (QC) procedures that we are following in this study. MethodsThe QA procedure consists of monitoring of different quantitative measurements using an agar gel and a geometrical phantom. For the already acquired data from human subjects, we describe QC steps for each imaging modality. To quantify reliability of outcome measurements, we perform test-retest reliability on human volunteers. ResultsWe have presented results from analysis of phantom data and test-retest reliability on a human volunteer. Results show consistency in data acquisition and reliable quantification of different outcome measurements. ConclusionThe acquisition protocol and QA-QC procedures described here can yield consistent and reliable outcome measures. We hope to acquire and eventually release high quality longitudinal neuroimaging dataset that will serve the scientific community and pave the way for interesting discoveries.

neuroscience↗

bioRxiv2020

Store Independent Ca2+ Entry Regulates the DNA Damage Response in Breast Cancer Cells

Although the mainstay of treatment for hormone responsive breast tumors is targeted endocrine therapy, many patients develop de novo or acquired resistance and are treated with chemotherapeutic drugs. The vast majority (80%) of estrogen receptor positive tumors also express wild type p53 protein that is a major determinant of the DNA damage response. Tumors that are ER+ and p53WT respond poorly to chemotherapy, although the underlying mechanisms are not completely understood. We describe a novel link between store independent Ca2+ entry (SICE) and resistance to DNA damaging drugs, mediated by the secretory pathway Ca2+-ATPase, SPCA2. In luminal ER+/PR+ breast cancer subtypes, SPCA2 levels are high and correlate with poor survival prognosis. Independent of ion pump activity, SPCA2 elevates baseline Ca2+ levels through SICE and drives cell proliferation. Attenuation of SPCA2 or depletion of extracellular Ca2+ increased mitochondrial ROS production, DNA damage and activation of the ATM/ATR-p53 axis leading to G0/G1 phase cell cycle arrest and apoptosis. Consistent with these findings, SPCA2 knockdown confers chemosensitivity to DNA damaging agents including doxorubicin, cisplatin and ionizing radiation. We conclude that elevated SPCA2 expression in ER+ p53WT breast tumors drives pro-survival and chemotherapy resistance by suppressing the DNA damage response. Drugs that target storeindependent Ca2+ entry pathways may have therapeutic potential in treating receptor positive breast cancer.

cancer biology↗

bioRxiv2019

What influences treatment response in animal models of non-alcoholic fatty liver disease? A meta-analysis with meta-regression.

The classical drug development pipeline necessitates studies using animal models of human disease to gauge future efficacy in humans, however, there is a comparatively low conversion rate from success in animals to in humans. Non-alcoholic fatty liver disease (NAFLD) is a complex chronic disease without any licensed therapies and hence a major field of animal research. We performed a meta-analysis of 414 interventional rodent studies (6,575 animals) in NAFLD to assess the mean difference in hepatic triglyceride content. 20 of 21 studied drug classes had similar efficacy with a mean difference of -30% hepatic triglyceride. However, when publication bias was accounted for, this reduced to -16% difference. Study characteristics were only able to account for a minority of variability on meta-regression, and we replicated previous findings of high risk of bias across 82% of cohorts. These findings build on previous work in preclinical neuroscience and help to explain the challenge of reproducibility and translation within the field of metabolism.

physiology↗