MATERIALS AND METHODS
Data collection. This research effort proceeded by mining existing data and literature available online. mRNA expression information was drawn from microarray and transcriptome datasets available on the Allen Brain Atlases portal (https://portal.brain-map.org). Data were not available for mouse brain development (http://developingmouse.brain-map.org) , so we analyzed the prenatal and postnatal expression FKBP5 mRNA only in the rhesus macaque (Macaca mulatta ) and human (Homo sapiens ) brain. Regions of interest (ROIs) including prefrontal cortex, amygdala, hippocampus, and striatum were selected on the basis of their involvement in MDD and ELS (Criado-Marrero et al, 2020; Troubat et al., 2021; Grogans et al., 2022; Pizzagalli and Roberts, 2022). The primary visual cortex was included as an independent region in the figure as an example of a non-ROI, to highlight the differences between gene expression for an individual region which is not relevant to MDD, and those which are. Adult mouse FKBP5 mRNA expression data were also retrieved from the Allen Brain Atlases portal (http://mouse.brain-map.org/gene/show/14006; experiment 70525820, probe name RP_050317_04_G06). For the re-analyses of FKBP5 mRNA expression in human and mouse MDD, data were retrieved from the original publications. Human MDD RNA sequencing (RNAseq) data (Matosin et al., 2023) consisted of PFC (area BA9) samples from 179 control subjects and 144 MDD cases, previously described in Tao et al. (2017). RNAseq data from a mouse model of MDD were retrieved from https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE109315. In this study, the authors performed RNAseq on prefrontal cortex and hippocampus from adult C57BL/6N control mice and C57BL/6N mice subjected to chronic social defeat stress (CSDS); data were normalized using voom and adjusted for sequencing (vHPC) and library preparation batches (vHPC, mPFC) with ComBat (Laine et al., 2018).
Data analysis. The human developmental transcriptome and non-human primate (NHP) microarray microdissection data for the FKBP5 gene were analyzed with Matlab (Mathworks) using the toolbox cbrewer2 version 1.0.0.1 (Lowe, 2023) to generate heat maps. The log2 reads per kilobase per million (RPKM) values of transcriptome data and z-score normalized values of microarray data (probe MmugDNA.10638.1.S1_at) were averaged for donors of the same age, and also across substructures to achieve the macrostructures (for example, the V1 macrostructure is composed of substructures including layer I, layer II, etc.). Supplementary Table 1 reports a full list of the components included in each macrostructure. In the case of the human data, any time point which lacked results for more than three macrostructures was excluded. Supplementary Table 2 summarizes the number of subjects per time point for the human data, while Supplementary Table 3 reports the same information for the corresponding non-human primate data points. FKBP5 mRNA expression data from the MDD mouse model (Laine et al. 2018; GEO accession number GSE109315) were analyzed with Matlab (Mathworks). FKBP5 mRNA expression data from RNAseq human datasets were re-plotted from Figure 1a of the original paper (Matosin et al., 2023) and analyzed using Prism 8.0 (GraphPad).
Code Accessibility. The code described in this paper is freely available online on Zenodo (https://doi.org/10.5281/zenodo.7525970 ) and GitHub (https://github.com/BozziNDRLab/Strelevitz_et_al_2023.git ). The specific toolbox we used in the code was obtained from Lowe (2023).