Transcriptional profiling reveals potential involvement of microvillous TRPM5-expressing cells in viral infection of the olfactory epithelium

Abstract

Background: Understanding viral infection of the olfactory epithelium is essential because the olfactory nerve is an important route of entry for viruses to the central nervous system. Specialized chemosensory epithelial cells that express the transient receptor potential cation channel subfamily M member 5 (TRPM5) are found throughout the airways and intestinal epithelium and are involved in responses to viral infection.

Results: Herein we performed deep transcriptional profiling of olfactory epithelial cells sorted by flow cytometry based on the expression of mCherry as a marker for olfactory sensory neurons and for eGFP in OMP-H2B::mCherry/TRPM5-eGFP transgenic mice (Mus musculus). We find profuse expression of transcripts involved in inflammation, immunity and viral infection in TRPM5-expressing microvillous cells compared to olfactory sensory neurons.

Conclusion: Our study provides new insights into a potential role for TRPM5-expressing microvillous cells in viral infection of the olfactory epithelium. We find that, as found for solitary chemosensory cells (SCCs) and brush cells in the airway epithelium, and for tuft cells in the intestine, the transcriptome of TRPM5-expressing microvillous cells indicates that they are likely involved in the inflammatory response elicited by viral infection of the olfactory epithelium.

Keywords: Immunity; Inflammation; Microvillous cells; Mouse; Olfactory sensory neurons; Viral infection.

Fig. 1

Fluorescence activated sorting (FACS) of cells isolated from the olfactory epithelium. a TRPM5 promoter driven expression of eGFP and OMP promoter driven expression of mCherry in the olfactory epithelium. Expression of eGFP is found both in MVCs that do not express mCherry (asterisk) and in OSNs double labeled with eGFP and mCherry (arrow). i. Composite, ii. eGFP, iii. mCherry, iv. Composite magnification. Magenta: mCherry, green: eGFP. Scale bar: i-iii, 50 μm, iv, 10 μm. b Schematic of RNA-seq process from tissue to RNA extraction. Mouse OE was dissociated into single cells and sorted via FACS. RNA was extracted from each of the resulting cell populations. c Two isolated OSNs differing in eGFP expression. Magenta: mCherry, green: eGFP. Scale bar: 10 μm. d Distribution of mCherry and eGFP fluorescence intensity for FACS-sorted cells. Three cell populations were isolated for RNAseq: Cells with low OMP promoter-driven mCherry expression and high TRPM5 promoter-driven eGFP expression (MVC_eGFP cells), cells with high OMP promoter-driven mCherry and low eGFP expression (OSN_eGFP- cells) and cells with eGFP expression of the same magnitude as MVC_eGFP cells and high OMP promoter-driven mCherry expression (OSN_eGFP+ cells). The number of cells collected for this FACS run were: OSN_eGFP-s 1,500,000, OSN_eGFP+s 5336 and MVC_eGFP cells 37,178. e qPCR levels (normalized to levels 18 s RNA) for expression of transcripts encoding for OMP (i), TRPM5 (ii), eGFP (iii) and ChAT (iv). The asterisks denote significant differences tested with either t-test or ranksum with p-values below the significance p-value corrected for multiple comparisons using the false discovery rate (pFDR) [15]. pFDR is 0.033 for OMP, 0.05 for TRPM5, 0.05 for eGFP and 0.03 for ChAT, n = 8 for OMP OSN_eGFP-s, 4 for OMP OSN_eGFP+s and 4 for MVC_eGFP cells

Fig. 2

RNAseq comparison of MVC_eGFP vs. OSN_eGFP- cells. a Heatmaps showing hierarchical clustering of the top 10 upregulated and top 10 downregulated genes identified by DESeq2. b Heatmaps showing hierarchical clustering of the 550 olfactory receptor genes identified by DESeq2 as expressed in OSN_eGFP- cells. For both a and b, row and column order were determined automatically by the pHeatmap package in R. Row values were centered and scaled using ‘scale = “row”’ within pHeatmap. c Volcano plot of all olfactory receptors, demonstrating the large number of enriched olfactory receptors in the OSN_eGFP- population. d Hierarchical clustering of transcripts for taste transduction and transcripts expressed in canonical and non-canonical OSNs identified by RNAseq as significantly different in expression between MVC_eGFP and OSN_eGFP- cells. The non-canonical OSNs considered here included guanilyl-cyclase D (GC-D) OSNs [37], Trpc2 OSNs [64], Cav2.1 OSNs [72], and OSNs expressing trace amine-associated receptors (Taars) [46]. Transcripts identified by DESeq2. e Gene ontology (GO) term enrichment for synaptic vesicle or chemosensory-related GOs was calculated from differentially expressed genes using TopGO in R. An enrichment value for genes with Fischer p value < 0.05 was calculated by dividing the number of expressed genes within the GO term by the number expected genes (by random sampling, determined by TopGO)

Fig. 3

Significant differences in virally-related, immune and inflammation gene ontology lists between MVC_eGFP and OSN_eGFP-. a Gene ontology (GO) term enrichment was calculated from differentially expressed genes using TopGO in R for OSN_eGFP- vs. MVC_eGFP cells. An enrichment value for genes with Fischer p value < 0.05 was calculated by dividing the number of expressed genes within the GO term by the number expected genes (by random sampling, determined by TopGO). Heatmap show hierarchical clustering of significantly differentially expressed genes identified by DESeq2. b Significantly differences in virally-related genes within the MVC_eGFP cells compared to OSN_eGFP-

Fig. 4

RNAseq comparison of OSN_eGFP+ to both MVC_eGFP and OSN_eGFP- cells. a Heatmap showing the top upregulated genes (excluding Olfrs) that are expressed in OSN_eGFP+ cells 4 fold higher than OSN_eGFP- AND MVC_eGFP cells. Additional criteria for inclusion was mean of expression > standard deviation of expression and mean of expression greater than 100. b Heatmap showing all Olfr genes differentially expressed between OSN_eGFP+ and OSN_eGFP- cells identified by DESeq2. MVC_eGFP cells did not express Olfrs. For both a and b, row and column order were determined automatically by the pHeatmap package in R. For each data point relative expression was calculated by subtracting the average row value from each individual value. c Volcano plot of all Olfactory receptors, demonstrating the small number of enriched olfactory receptors in the OSN_eGFP+ population. d Hierarchical clustering of transcripts for taste transduction and transcripts expressed in canonical and non-canonical OSNs identified by RNAseq as significantly different in expression between the cell groups. We compared expression of transcripts involved in taste transduction, canonical olfactory transduction, and non-canonical OSNs. The non-canonical OSNs considered here included guanilyl-cyclase D (GC-D) OSNs [37], Trpc2 OSNs [64], Cav2.1 OSNs [72], and OSNs expressing trace amine-associated receptors (Taars) [46]. Transcripts identified by DESeq2

Fig. 5

In situ hybridization chain reaction finds strong TRPM5 mRNA expression in MVC_eGFP cells, but not in the nuclear OSN layer. a In situ for TRPM5 (yellow) and OMP (magenta) transcripts in the olfactory epithelium of TRPM5-GFP mice (GFP is green) shows strong label for TRPM5 in MVCs (asterisks) and sparse labeling in the OSN nuclear layer (arrows). The scale bar is 20 μm. b In situ for TRPM5 (yellow) and OMP (magenta) transcripts in the olfactory epithelium of TRPM5-GFP x TRPM5-knockout mice (GFP is green) shows no label for TRPM5 in GFP-positive MVCs (asterisks) and does show sparse labeling in the OSN nuclear layer (arrows). The scale bar is 20 μm

Fig. 6

Model depicting the role for microvillous cells involvement in the olfactory epithelium innate immune response to viral infection. 1) Secreted or cell surface glycoproteins constitute a first barrier preventing virus entry. 2) When reaching MVC_eGFP, viruses can encounter three types of membrane proteins: adhesion molecules that trigger intracellular signaling upon viral recognition (black rectangle), transmembrane proteins that block virus entry (black circle), viral receptors allowing virus entry (grey circle). 3) MVC_eGFP express numerous transcriptional factors involved in the inhibition of viral replication. 4) Cytosolic viral RNA sensing induces the production of type I interferons. 5) A possible signaling pathway leading to intracellular calcium increase, TRPM5 activation and Na⁺-mediated vesicle release. Acetylcholine can activate neighboring sustentacular cells and underlying trigeminal fibers. 6) Eicosanoids synthesis, along with IL-25 production, can recruit and activate group 2 innate lymphoid cells, which are key controllers of type 2 inflammation. 7) GPR126 activation results in NFkB activation and TNFα production. TNFα can directly activate macrophages. TNFα also induces a change in the function of horizontal basal cells, switching their phenotype from neuroregeneration to immune defense. 8) Interferons and cytokines can in turn activate antiviral immune response in neighboring MVC_eGFP