81 TSPO in neurodegeneration was constructed using MEGA11 using Maximum Parsimony method with 1000 bootstrap replication. The MP tree was obtained using the Tree-Bisection-Regrafting (TBR) algorithm31. Motif finding and motif enrichment We have used SEA (Simple Enrichment Analysis) from the MEME-suite (v 5.4.1) to calculate the relative motif enrichment between Muroidea family species and non-Muroidea mammals32,33. We set the TSPO promoter sequences for the three Muroidea species (Mouse, Rat, Chinese Hamster) as the input sequence and the rest of species as the control sequence. We set the E-value ≤ 10 for calculating significance. We used the motifs for AP1, ETS and SP1 from JASPAR motif database (https://jaspar.genereg.net/). Multi-species TSPO expression in macrophage and microglia Datasets were searched using the search terms “Macrophage/Monocyte”, “Microglia” and “LPS”. Dataset featuring stimulation less than 3 hours were excluded. Datasets with accessible raw data and at least three biological replicates were used. Microarray datasets were analysed as the same way described in section “Meta-analysis of TSPO gene expression”. Raw gene count data for the RNAseq datasets were downloaded from either ArrayExpress or GEO (https://www.ncbi.nlm.nih.gov/geo/) and differential expression was performed using DESeq2 v.1.26.034. For S1a, the mouse Tspo expression (GEO ID: GSE38371) fold change was directly used from the respective study since biological replicates were not publicly accessible35. Human and mouse scRNAseq analysis of microglia We assessed alterations in gene expression of TSPO in human and mouse activated microglia in publicly available scRNAseq datasets. Postmortem human brain samples are predominantly studied using single nucleus RNA sequencing (snRNAseq) rather than single cell RNAseq (sc) RNAseq because the latter requires intact cells which cannot be recovered from frozen brain tissue samples. However, TSPO is detected in a very low percentage of nuclei from snRNAseq experiments which prevents accurate assessment of differential expression of TSPO across disease or microglial states36. For this reason, we searched MEDLINE for human scRNAseq experiments involving AD, MS and ALS donors and mouse brain scRNAseq datasets derived from the respective mouse models, as well as of pro-inflammatory activation with LPS treatment. We found three human studies involving donors with AD37 and MS38,39. Where microglia from CSF samples were analysed with scRNAseq. We found no studies with ALS donors. We found three mouse studies: an LPS activated model40 an AD model41 and acute EAE42. A fourthmouse scRNAseq dataset was identified from LPS-treatedmice43, however, due to its small size (less than 400microglial cellswere sequenced), this dataset was discarded from further analysis. Raw count matrices were downloaded from the Gene Expression Omnibus (GEO) with the following accession numbers: GSE13011942, GSE11557140, GSE9896941, GSE13826638 and GSE13457837. Data were processed with Seurat (v3)44 or nf-core/scflow45. Quality control, sample integration, dimension reduction and clustering were performed using default parameters as previously described36,46. Microglial cells (mouse datasets) and microglia-like cells were identified using previously described cell markers. Differential gene expression analysis was performed using MAST47 implemented in Seurat to perform zero-inflated regression analysis by fitting a fixed-effects model. Disease vs control group comparisons were performed for all datasets, except for the Keren-Shaul dataset where the AD-associated microglia phenotype was compared to the rest of the microglial population in 5XFAD mice. In all cases, we assessed expression of activated microglial markers. Gene
RkJQdWJsaXNoZXIy MTk4NDMw