_Research

Projects

Systems Biology

Multiorgan pathophysiology of COVID-19

Although all COVID-19 patients admitted to the hospital have pneumonia, there are varied clinical manifestations during progression of disease in different sets of patients. As patients get critically ill, a subset of patients develops thrombosis requiring anticoagulation therapy, others have cardiovascular complications including myocarditis and still others have acute kidney injury. We do not understand the mechanistic basis for these varied pathophysiologies. We are using cellular mechanism-based approaches for the identification of multiorgan pathophysiology in COVID-19. We use host responses in multiple tissues for focusing on the cell biological mechanisms underlying viral infectivity and cytokine responses in the absence and presence of virus. Controlling the membrane turnover of ACE2, the virus receptor and internalization and processing of the ACE2-CoV complex is a feasible approach for development of therapeutics that reduce the severity SARS-CoV-2 infection. These core cell biological processes are likely to be similar in different organs although they may differ from individual to individual. We use an unbiased approach with single cell RNA Seq analyses to identify the differential capability of ACE2 turnover pathways and the genomic determinants that may control cell type and organ selective susceptibility to infection. This is done by integration of the transcriptomic data from cell-based assays of different human cell types under differing conditions with genomic data using the eQTL approaches. The availability of whole genome sequence data from the cells and patients as well as clinical data from hundreds of confirmed COVID-19 patients admitted to the different hospitals of the Mount Sinai Health System provides with a unique opportunity for data integration. The cell biological data when integrated with the clinical pathophysiology data will allow us to map the genomic data from the cell biological experiments onto the genomic data from patients to identify genomic predictors for multiorgan susceptibility in COVID-19 patients.


Computational Models of Whole Cell Function

We are developing predictive scalable models of whole cell functions. We want to explain these functions in terms of cellular components and their interactions arranged as pathways and networks. We seek to understand and predict the variability in whole cell responses using spatially resolved single cell RNA-Seq. In order to understand the dynamics of network topology, and the variable outputs that result in a range of physiological responses we focus on identifying regulatory motifs such as feedback and feedforward loops, and determining their information processing capability. We have constructed and analyzed dynamic maps of these motifs to understand how cellular signaling networks engage multiple cellular machinery to produce physiological responses to extracellular signals. Experimental data obtained from single cell transcriptomic and proteomic experiments are used to construct near comprehensive maps of cellular pathways and networks, and demonstrate how these pathways interact to produce emergent whole cell functions. We are currently using the cannabinoid-1 receptor-regulated neurite outgrowth in primary cortical neurons as a model system.


Forgetting

For over two decades we have studied, both experimentally and computationally, how interactions among cell signaling networks within neurons regulate neuronal plasticity,. As part of these studies we found that memory formation processes activates the WT1 transcriptional repressor to enable forgetting. WT1 is also involved in memory interference that arises when memory of two similar experiences get mixed up. We are now studying the relationships between molecular pathways and the cellular circuits within the hippocampus involved in forgetting and are determining the role of regulatory loops, such as feedforward loops, in controlling the balance between remembering and forgetting.


Cell Shape and Tissue Organization

We are interested in understanding how cell shape regulates cell and tissue level functions. We have found that cell shape is a distinct locus of retrievable information separate from information encoded by biochemical and biophysical signals. Cell shape information is directly transduced by some types of integrins, and cell shape modulates biochemical signaling. We are now investigating how cell shape and the spatial organization within and between cells contributes to dynamic stability of tissue level phys. For these studies, we collaborate with engineers to design biochips where we can control shape of multiple cells and the volume of extracellular spaces in tissue-like structures that can be imaged to observe and quantify biochemical signaling reactions in live cells. To decipher how the information content of cell shape regulates tissue level function, we use integrated computational modeling approaches that use neural network algorithms to run control theory models that contain spatial specification of cellular pathways using partial differential equations. These studies focus on understanding the dynamics underlying tissue integrity in physiological function.


Kidney Precision Medicine Project- the Kidney Atlas

As part of our extended interest in cell shape and tissue organization in translational research, we are participating in the Kidney Precision Medicine Project that is focused developing a kidney atlas that integrates morphology with molecular components and pathways of the different cell types in the kidney. It is anticipated that such an atlas can be used by pathologists for molecular diagnosis of kidney disease, as well as by research scientists who are focused on drug discovery for kidney disease. Our role in this project as part of the Central Hub, a collaborative enterprise, is to computationally analyze different types of omics data, and map the pathways and networks in the different renal cell types in various disease states.


Systems Pharmacology

As part of the translational research in our laboratory, we are developing systems level approaches to understanding drug action at a genome-wide level. This includes predictive toxicology, where the long-term goal is to predict the propensity for drug-induced adverse events in individuals based on their human genome. We are using systems reasoning to develop drug therapies, including mechanism-based drug combinations, for complex diseases.


Adverse Events Predictions: Cardiotoxicity of Targeted Cancer Therapeutics

Many efficacious drugs produce unwanted side effects. Often these side effects are serious resulting in adverse events both acute, such as arrhythmias, and chronic, such as heart failure. We constructed large scale networks to capture all of the known protein-protein interactions in the human genome and to computationally identify selective regions (disease neighborhoods) within the interaction space associated with specific diseases. We analyzed the relationships among drug targets and other cellular components to understand the relationship between disease neighborhood and targets of drugs used to treat other diseases. From such analyses we were able predict adverse events and explain rare adverse events reports in the FDA adverse event reporting system (FAERS) database. A successful example of our network-based prediction algorithm was identifying the relationship between loperamide and arrhythmias based on off-target binding of the drug. Subsequent clinical reports have shown an association between overuse or abuse of loperamide and arrhythmias. Mining FAERS, we have found drug combinations, often when each drug is prescribed for a distinct pathophysiology, which can mitigate drug-induced adverse events. These types of analyses showed that mapping cellular networks underlying drug action, can be used for predictive toxicology.

In a collaborative study we have developed a drug-toxicity signature generation center to discover transcriptomic signatures for the risk of heart failure associated with targeted cancer drugs such as kinase inhibitors. For this, we have used both human adult heart cells and cardiomyocytes derived from healthy human induced pluripotent stem cells (iPSCs). From transcriptomic data we have identified gene signatures that correlate well with clinical adverse event risks of different drugs. The transcriptomic signatures provide insight into the subcellular processes, as well as the structural features of drugs, that can be associated with an adverse event. Future studies are focused on obtaining transcriptomic signatures from iPSC-derived myocytes where the transcriptomic signatures can be related to genomic determinants of individuals.


Drugs for Aneurysms

We have collaborated with experts in the field of thoracic aneurysms and Marfan syndrome to utilize transcriptomic data from patients and from mice harboring the mutated Marfan gene, to identify common networks of subcellular processes in humans and animal models, which in turn can be used to identify drugs to treat the progression of aneurysms. We have identified the GABA- B receptor agonist baclofen as one such drug. Treatment of Marfan mice with baclofen, at doses that are therapeutically used in humans, leads to significant morphological and physiological improvement of the vessel wall, suggesting that baclofen could be tested in humans for mitigation of aneurysm progression. We are using such cellular mechanism-based approaches to identify other drugs that control the mechanical properties of the vessel wall for the treatment of aneurysms. Future studies will focus on identifying the cellular mechanisms associated with idiopathic aneurysms both to develop biomarkers and to identify novel therapeutic drugs.


Drugs for Neurorepair

Studies on the subcellular processes involved in neurite outgrowth have enabled us to use computational modeling and experiments to identify multiple drugs that would act on different parts of a neuron to promote the regeneration of severed axons. We have tested a four-drug combination in vivo using the rat optic nerve crush model and found that we can regenerate functional nerves from the injury site near the retinal ganglion cells to the visual cortex. We can observe morphological growth into the visual cortex in the brain and obtain light-driven electrophysiological signals after drug treatment. Future experiments are focused on improving the extant of nerve regeneration to obtain better temporal dynamics of the light-driven electrophysiological signals, and testing whether spatially specified drug combinations will work for restoration of function after spinal cord injury in rodents.