103 TSPO in neurodegeneration unique to microglia from a subset of species from the Muroidea superfamily of rodents. The increase in TSPO expression is likely dependant on the AP1 binding site in the core promoter region of TSPO. Finally, we showed that TSPO is mechanistically linked to classical proinflammatory myeloid cell function in mice but not humans. This finding fundamentally alters the way in which the TSPO PET signal is interpreted, because it implies that the microglial component of the TSPO PET signal reflects density only, rather than a composite of density and activation phenotype. For example, in Parkinson’s Disease (PD) there is evidence of activated microglia in the postmortem brain but minimal change in microglial density77. Three well designed studies using modern TSPO radiotracers found no difference in TSPO signal between PD and controls groups78-80. The lack of increase in the TSPO PET signal is consistent with the data presented here, and should therefore not be interpreted as evidence for lack of microglial activation in PD. Our study has several limitations. First, we have only examined microglia under certain pro-inflammatory conditions and cannot exclude the possibility that other stimulation paradigms would increase TSPO in human myeloid cells. However, the in vitro stimuli which were examined included a broad range of pro-inflammatory triggers, and the three human diseases are diverse with respect to the mechanisms underlying the activation of microglia. Second, the measurements of cellular TSPO expression we used in brain tissue are semiquantitative. However, the same IHC quantification methods were used in all human and mouse comparisons, and these methods consistently detected cellular TSPO increases in mouse microglia despite not detecting analogous changes in human microglia. Furthermore, where IMC and immunofluorescence were used, the quantitative data were consistent with IHC. The neuropathology protein quantification was also consistent with gene expression measured by scRNAseq. Third, for RNAseq analysis, we were restricted to single cell rather than single nucleus experiments. This is because TSPO is detected in only 5-12% of microglial nuclei36,81-83 but ~80% of microglial cells37-42. Fourth, the in vitro assay which most closely mimics in vivo PET data is radioligand binding, which quantifies the binding of the radioligand to the binding site itself. Here, we quantified expression of the TSPO gene or protein rather than radioligand binding site density. However, we have previously shown that for TSPO, gene expression, protein expression and radioligand binding site data closely correlate15. Finally, whilst we present data correlating inducible TSPO expression with the presence of the AP1 binding site in the TSPO core promoter region, to demonstrate causation the AP1 binding site would need to be knocked out from the mouse or rat, and knocked in to a non-Muroidea rodent. Furthermore, although we were able to find array expression data for a range of non-rodent mammals that show TSPO is not induced upon myeloid cell activation, we were unable to find array expression data for those rodents that lack the AP1 binding site, such as squirrel or naked mole rat. In summary, we present in vitro expression and sequence alignment data from a range of species, as well as ex vivo data from neurodegenerative and neuroinflammatory diseases and associated animal models. We show that inflammation-induced increases in cellular TSPO expression are restricted to microglia from a subset of species within the Muroidea superfamily of rodents, and that TSPO is mechanistically linked to classical pro-inflammatory myeloid cell function in mice, but not humans. This challenges the commonly held view that TSPO provides a readout of microglial activation in the human brain and shows that the TSPO PET signal likely reflects the local density of inflammatory cells irrespective of phenotype. The interpretation of TSPO PET data therefore requires revision.
RkJQdWJsaXNoZXIy MTk4NDMw