80 Chapter 4 with the in vitro data, we show that in AD, ALS and MS, and in marmoset EAE, TSPO protein expression does not increase in CNSmyeloid cells that express a pro-inflammatory phenotype, while expression is markedly increased in activated myeloid cells in all mouse models of these diseases. With exploration of the relative expression of TSPO in publicly available CNS single cell RNA sequencing (scRNAseq) data from brains of the human diseases and rodent models, we again show an increase in microglial TSPO gene expression in mice with proinflammatory stimuli, but not humans. Finally, using functional studies and examination of transcriptomic co-expression networks, we find that TSPO is mechanistically linked to classical proinflammatory myeloid cell function in rodents but not humans. These data suggest that the commonly held assumption that TSPO PET is sensitive to microglial activation is true only for a subset of species within the Muroidea superfamily of rodents. In contrast, in humans and other mammals, it simply reflects the local density of inflammatory cells irrespective of the disease context. The clinical interpretation of the TSPO PET signal therefore needs to be revised. Methods Meta-analysis of TSPO gene expression Datasets were searched using the search terms “Macrophage/Monocyte/Microglia” and filtered for ‘Homo sapiens’ and ‘Mus musculus’. Datasets with accessible raw data and at least three biological replicates per treatment group were used. To avoid microarray platformbased differences only datasets with Affymetrix chip were used. Raw microarray datasets were downloaded from ArrayExpress (https://www.ebi.ac.uk/arrayexpress/) and RMA normalisation was used. The ‘Limma v.3.42.2’ R package was used to compute differentially expressed genes, and the resulting P-values are adjusted for multiple testing with Benjamini and Hochberg’s method to control the false discovery rate22. Meta-analysis was performed using R package ‘meta v.5.1.1’. A meta P-value was calculated using the random-effect model. ChIP-seq data processing and visualisation ChIP-seq datasets were downloaded fromGSE6659423 (human) and GSE3837724 (mouse). Raw fastq sequences were aligned with Bowtie2 v.2.2.925 to the human reference genome hg19 or to mouse reference genome mm9, annotated SAM files are converted to tag directories using HOMER v.4.11.126 using the makeTagDirectory module. These directories are further used for peak calling using-style histone parameter or converted to the bigWig format normalized to 106 total tag counts with HOMER using the makeUCSCfile module with -fsize parameter set at 2e9. For the analysis of histone ChIP-seq data input samples were utilized as control files during peak detection, whereas IgG control files were used during peak correction of the PU.1 ChIP-seq data. Peaks were visualised using UCSC genome browser27 Multiple sequence alignment and phylogenetic tree construction We have retrieved the TSPO promoter region starting from 1 Kbp upstream and 500 bp downstream of the putative transcription start site (TSS) of 34 rodent and non-rodent mammals from ENSEMBL genome database (http://www.ensembl.org/index.htmls). The full list can be found in Supplementary File 2. The multiple sequence alignment was performed using the T-Coffee (v13.45.0.4846264) multiple sequencing tool with the parameter -mode=procoffee which is specifically designed to align the promoter region28,29. The sequence alignment and the phylogenetic tree were visualised using Jalview (v 2.11.1.6)30. Phylogenetic tree
RkJQdWJsaXNoZXIy MTk4NDMw