THE HUMAN BRAIN MICROBIOME IN HEALTH AND PARKINSONS's DISEASE MUCH ADO ABOUT NOTHING

Background: Several studies of PD and AD patients' brains and experimental models indicate the presence of inflammatory processes in both diseases. This was recently linked to remnants of poly-microbial infections in brain tissue - linking the observed neuro-inflammation to putative pathogens. Reports on residing bacteria in healthy brains even suggested the existence of a 'brain microbiome'. Indeed, microbiome research in the last decade has considerably increased our knowledge of organ – specific microbiomes in humans. Using 16S rRNA gene amplicon sequencing, we tested the hypothesis of a bacterial brain microbiome, using brain samples from healthy and PD human subjects, as well as murine ones. In line with state-of-the-art recommendations, several negative and positive controls were included in the analysis and estimation of the total bacterial biomass was assessed by 16S rRNA gene qPCR. Results: Bacterial biomass detected via 16S qPCR was very low in all biological samples, but human samples seemingly contained more bacterial biomass. However, stringent reanalysis revealed that a combination of exogenous DNA contamination (54.81%) and false positive amplification of host DNA (34.24%, 'off-targets') could explain this signal in both, amplicon sequencing and qPCR based analysis. This 'off-target' amplification was identified as a major confounder in this low-biomass/high-host-DNA setting and false positive results were observed with the majority of tested 16S pipelines, despite stringent contaminant filtering. Following up several leads of seemingly brain enriched microbes, in-depth checks revealed all to be false positive findings. Conclusions: Taxonomic signals obtained from (very) low biomass samples in 16S rRNA gene sequencing and appearing solely or enriched in biological samples have to be carefully scrutinized to exclude the possibility of 'off-target' amplifications. They have to be explicitly matched against any possible background genomes available in large quantities (i.e. the host genome in host-associated samples). Our data refutes the hypothesis of a bacterial infection in PD brains, and a healthy 'bacterial brain microbiome' is unlikely to exist for bacteria.