Characterising processing conditions that artifactually bias human brain tissue transcriptomes
Preprint
- 23 October 2024
- preprint
- Published by Springer Nature in Research Square
Abstract
Autopsy-derived brain tissue analysis is vital for exploring the complex landscape of neurobiology in health and disease but processing conditions during post-mortem handling can lead to significant technical artifacts affecting data interpretation. Here, we define brain transcriptomic signatures from healthy adult human brain tissue that was snap frozen in under 1 hour (Mean: 31min, Min: 10min, Max 55min) of extraction (hereafter referred to as ~ 0 hours) and compared it to brain autopsy tissue with either typical shorter (Mean: 6 hours, Min: 4h, Max: 14h) or longer (Mean: 36 hours, Min: 17h, Max: 70h) post-mortem intervals. We found a large number of differentially expressed genes in post-mortem tissue compared to snap frozen tissue, even with the shorter post-mortem intervals. These differences allowed us to define a general “artifactual” gene signature from adult human brain autopsies that arise as a result of post-mortem processing (termed B rain A rtifact G ene ( BAG ) Signatures). We subjected the snap frozen brain samples to different times and temperatures mimicking those typical with autopsy material to determine how these common variables influence brain gene expression. Using this approach, we discovered a set of regulated genes that we defined as “ T ime and T emperature R esponse genes U nderlying T ranscriptional H eterogeneity (TTRUTH)” genes. Using deep learning approaches, we then developed a model capable of modelling the extent to which individual brain autopsy samples from non-neurological disease control donors express artifactual transcripts associated with post-mortem interval time and temperature. This allowed us to assign TTRUTH scores to each individual brain autopsy sample. Moreover, using single nuclear RNA sequencing on paired samples, we identified that neuronal populations are the initial expressers of these artifactual transcripts. As tissue remains at room temperature for extended periods, oligodendrocytes emerge as the predominant cell types expressing artifactual genes. Finally, we provide an Open Science website tool for others to use to determine whether their samples are subject to similar artifacts. Using this tool, the brain autopsy research community can now assign TTRUTH scores to human brain autopsy RNAseq datasets, to provide an additional quality control measure to better standardise datasets, allow additional sample stratification across experimental groups and enhance data interpretation.Keywords
All Related Versions
This publication has 54 references indexed in Scilit:
- Characterisation of premature cell senescence in Alzheimer’s disease using single nuclear transcriptomicsActa Neuropathologica, 2024
- Impact of variation in tissue preparation methodology on the functional outcome of neocortical mouse brain slicesBrain Research, 2020
- Single-Cell Multi-omic Integration Compares and Contrasts Features of Brain Cell IdentityCell, 2019
- DoubletFinder: Doublet Detection in Single-Cell RNA Sequencing Data Using Artificial Nearest NeighborsCell Systems, 2019
- Differential transcriptional response profiles in human myeloid cell populationsClinical Immunology, 2018
- iPSC-Derived Human Microglia-like Cells to Study Neurological DiseasesNeuron, 2017
- Metabolomics and neuroanatomical evaluation of post-mortem changes in the hippocampusBrain Structure and Function, 2017
- RNA integrity in post-mortem samples: influencing parameters and implications on RT-qPCR assaysInternational journal of legal medicine, 2011
- Prospective collection of tissue samples at autopsy in children with diffuse intrinsic pontine gliomaCancer, 2010
- The Sun Health Research Institute Brain Donation Program: Description and Eexperience, 1987–2007Cell and Tissue Banking, 2008