Single-cell transcriptomics reveals distinct inflammation-induced microglia signatures

Sousa, Carole; Golebiewska, Anna; Poovathingal, Suresh K; Kaoma, Tony; Pires-Afonso, Yolanda; Martina, Silvia; Coowar, Djalil; Azuaje, Francisco; Skupin, Alexander; Balling, Rudi; Biber, Knut; Niclou, Simone P; Michelucci, Alessandro

Minor comments: Some of the figures contain inaccuracies in the labeling. For example Figure 3C log10FDR should be -log10FDR. Moreover, labeling of differentially expressed pathways in all figures often does not contain the full name of the pathway shown. The authors argue that microglial subpopulations exist that show differential responses to acute inflammation. By selecting population-specific cell surface markers, they were able to effectively isolate only brain-resident microglia, based on Cd11b+/Cd45int expression, as opposed to other intrinsic or peripheral cells (Fig. 1) . The authors show that upon an acute LPS injection or upon in vitro LPS stimulation, microglia lose their homeostatic and phagocytosing gene expression profile, whilst pro-inflammatory genes are upregulated (Fig. 2) . This data was confirmed using a single-cell sequencing approach, showing that microglia exposed to LPS show a distinct reactive phenotype compared to saline-injected control mice (Fig. 3) . By analysing the sequencing data more closely, the group could identify a specific microglia sub-population ("subset LPS") that showed an intermediate phenotype compared to control and reactive microglia (Fig. 4) . They postulate based on differential gene analysis that these cells might either be damaged cells, recovering from their activation status or a cell population with specific DNA repair properties. Lastly, in Fig. 5 the authors compared the gene signatures of LPS-exposed microglia to that of "disease-associated microglia" (DAM) that has been described by Keren-Shaul et al. (2017) We acknowledge the reviewer for his comments and criticisms regarding the relevance of the study for neurodegenerative diseases. Taking them into account, we strengthen the results regarding the acute inflammatory model and put less emphasis on the comparison with neurodegenerative conditions. Specifically: -we amended the title: "Single-cell transcriptomics reveals distinct inflammatory and neurodegenerative microglia signatures" now reads "Single-cell transcriptomics reveals distinct microglia signatures under inflammation" -we shortened the text describing the comparison with DAM as recommended below (minor point 7c "The text describing Fig. 5 should be more to the point and less descriptive") -we placed the results regarding the comparison with DAM in supplementary data (in the revised version, Fig. 5 has been moved to Fig. EV3 ) -we built up the analysis of our acute inflammatory model performing additional experiments and adding further results taking into consideration the suggestions from both reviewers. As a side note regarding the reviewer's statement "The differences which the authors describe could simply reflect the distinct microglia properties inherent to the respective model", we would like to point out that several relevant studies are conducted by comparing data obtained on different mouse strain backgrounds showing comparable results. Similar observations are also applicable to the reviewer's assertion "To be able to effectively compare the two populations it would be necessary to re-run the RNASeq analysis comparing the effects of LPS injection on microglia phenotypes with states of microglia in the Alzheimer model in the same laboratory with the same platform". For these aspects, please refer, for example, to the following article: Holtman et al (2015) Induction of a common microglia gene expression signature by aging and neurodegenerative conditions: a c