Developing Spatially Resolved Molecular Drug-Repurposing Assays for Treating Age-Related Frailty

As people age, they lose skeletal muscle tissue. Muscle mass plays an important role in regulating an individual's metabolism and loss can lead to an increased risk of developing obesity and type two diabetes. Maintaining muscle mass is especially critical for determining physical function in everyday life. The only effective strategy for maintaining muscle mass and function is exercise therapy. Typically, weight-lifting is the preferred method for building back muscle mass however other types of brisk exercise (e.g. high-intensity cycling) will help. There is one major issue, however, and that is that for every 100 people that follow a tailored supervised physical exercise program only 40 will demonstrate robust gains in muscle mass, 30 will demonstrate modest improvements while 30 will not improve at all. The reason for the highly variable clinical response is unknown, however the evidence from completed clinical trials indicates that it is unlikely to be caused by lack of protein in a normal diet nor is it related to the type of training protocol. In fact, it is most likely caused by genetic factors and interactions with additional environmental factors, including preclinical disease and potentially prescribed drugs. We have completed the genomic profiling of thousands of human muscle biopsy samples and compared their profiles to those we can create by treating cells in the lab with drugs. This has allowed us to identify old drugs that might be useful for treating aging related diseases and also some that may interfere with the benefits of exercise. Muscle tissue is actually composed of several different types of cells and each plays a role in determining the responses to exercise or drug treatment. In the present study, we wish to study the genomic responses in each type of cell using a new technology called 'spatial transcriptomics'. This allows us to understand the activation status of each type of cell, following exercise, and relate that activation status to the gains (or lack of) noted after supervised resistance training. This will enable us to understand if one or more cell type is failing to respond normally to the exercise program. This method also allows us to better understand the molecular pathways activated by exercise, as our previous work relies on averaging the measurements across different cell types because the tissue biopsy is pulverised, mixing all the cells together. This new method should allow us to create more accurate signatures of the pathways responsible for effective gains in muscle mass (as well as other health endpoint points) and better match those signatures to drug-signatures that might help over-come low responder status. The same signatures can also provide a readout of drugs that might interfere with exercise rehabilitation responses (e.g., both paracetamol and metformin have been found to limit training responses in clinical trials) and older individuals can therefore, when being rehabilitated from illness, falls or surgery, be temporarily taken off those medications while completing their physical therapy. The application of the new knowledge - both practical and potential biotechnological application - has the potential to enable aging populations, in the UK and globally, to live a healthier fulfilling life.

We will apply machine learning methods to single-cell spatial imaging, splice-specific transcriptomics and clinical physiology to provide new insight into the drivers of the highly variable functional responses to skeletal muscle rehabilitation in the elderly. To date, using our existing bulk muscle biopsy (mixed-cell) tissue genome-wide transcriptomic data (>1,200 profiles), we established that mTOR signalling declines from the 3rd to 6th decade of life; a process linked with insulin resistance, muscle growth and atrophy pathways. These signatures also allowed us to screen for drugs in silico that have the potential to modulate muscle rehabilitation responses i.e., for drug-repurposing and pharmacovigilance. Identifying commonly prescribed drugs or combinations, which likely impair normal responses to physical activity could be impactful in the near term. As muscle is composed of several cell types (endothelial; myocyte-subtypes), we hypothesise that cell-specific factors govern the clinical responses in the elderly. From a candidate list of >2,000 genes, we will custom design 500 assays for the Multiplexed Error Robust Fluorescence In Situ Hybridization (MERFISH) method (MERSCOPE platform). We will profile high vs low treatment responder samples from our completed muscle rehabilitation of the elderly study, which captured the expected treatment heterogeneity (70yr, n=83, MRI lean mass gains -5% to +15%; men/women). To support our MERSCOPE cell-specific mTOR results we will use our existing a) muscle tissue from healthy adults treated with Rapamycin (n=8, 16wk), b) primary human muscle cell full transcriptomes (n=16) treated with 100nM Rapamycin), and we will generate c) transcriptomes from human endothelial cells treated with Rapamycin. Machine learning will be applied to stratify treatment outcomes by MERSCOPE profiles and separately to optimise cell-type specific transcript selection for in silico drug repurposing/pharmacovigilance assay design.