Single-Cell Analysis of Neuroinflammatory Responses Following Intracranial Injection of G-Deleted Rabies Viruses

Abstract

Viral vectors are essential tools for the study of neural circuits, with glycoprotein-deleted rabies viruses being widely used for monosynaptic retrograde tracing to map connectivity between specific cell types in the nervous system. However, the use of rabies virus is limited by the cytotoxicity and the inflammatory responses these viruses trigger. While components of the rabies virus genome contribute to its cytotoxic effects, the function of other neuronal and non-neuronal cells within the vicinity of the infected host neurons in either effecting or mitigating virally-induced tissue damage are still being elucidated. Here, we analyzed 60,212 single-cell RNA profiles to assess both global and cell-type-specific transcriptional responses in the mouse dorsal raphe nucleus (DRN) following intracranial injection of glycoprotein-deleted rabies viruses and axonal infection of dorsal raphe serotonergic neurons. Gene pathway analyses revealed a down-regulation of genes involved in metabolic processes and neurotransmission following infection. We also identified several transcriptionally diverse leukocyte populations that infiltrate the brain and are distinct from resident immune cells. Cell type-specific patterns of cytokine expression showed that antiviral responses were likely orchestrated by Type I and Type II interferon signaling from microglia and infiltrating CD4⁺ T cells, respectively. Additionally, we uncovered transcriptionally distinct states of microglia along an activation trajectory that may serve different functions, which range from surveillance to antigen presentation and cytokine secretion. Intercellular interactions inferred from transcriptional data suggest that CD4⁺ T cells facilitate microglial state transitions during the inflammatory response. Our study uncovers the heterogeneity of immune cells mediating neuroinflammatory responses and provides a critical evaluation of the compatibility between rabies-mediated connectivity mapping and single-cell transcriptional profiling. These findings provide additional insights into the distinct contributions of various cell types in mediating different facets of antiviral responses in the brain and will facilitate the design of strategies to circumvent immune responses to improve the efficacy of viral gene delivery.

Keywords: G-deleted rabies virus; brain infiltration; microglia; neuroinflammation; scRNA-seq.

Figure 1

Single-cell transcriptional analysis of responses to viral infection in the brain. (A) Experiment schematic. Unpseudotyped SADΔG B19 rabies viruses (RbV) were injected into 8–10 week old C57BL/6J mice. Each animal received a pair of injections into two different regions innervated by dorsal raphe nucleus (DRN) 5-HT neurons (nucleus accumbens (NAc) and substantia nigra (SN) in the schematic shown). Tissue containing the DRN was dissected 7 days post-injection and dissociated into whole-cell suspensions. scRNA-seq libraries were generated using the microfluidic-based inDrop platform. Age-matched uninjected animals were used as the Control group. (B) UMAP plot of merged dataset containing 60,212 cells. Control and RbV datasets were merged using canonical correlation analysis (CCA)-based dataset alignment methods. Individual points representing single cells are color-coded by cell class/type as shown in (E). (C) UMAP plot of the merged dataset with cells color-coded by experimental condition (RbV or Control). (D) Stacked bar plot showing the relative proportion of each cell type in RbV and Control groups. Cell class/type categories are color-coded following the same scheme as (B). (E) Left: dendrogram with cell class/type labels corresponding to the cluster labels in (B). Right: dot plot showing expression of example genes (columns) used to identify the major cell classes/types (rows). The color of each dot represents the average log-scaled expression of each gene across all cells in a given cluster, and the size of the dot represents the fraction of cells in the cluster in which transcripts for that gene were detected.

Figure 2

Rabies infection induces both global and cell-type-specific transcriptional changes. (A) Scatter plot comparing averaged expression of each gene in simulated “bulk” RNA-seq for RbV and Control groups. Several of the genes most enriched in the RbV group are labeled in red. The gray line indicates the line of unity (x = y). (B) Dot plot of the top MSigDB hallmark gene sets that were significantly enriched and up-regulated in the RbV group (5% FDR, Benjamini-Hochberg correction, Normalized Enrichment Score > 0). (C) Bar plot showing the number of differentially expressed genes (DEGs) that were significantly changed (388 DEGs with Q value < 0.01 and |log₂ fold-change| > 1). differential expression (DE) genes were either a part of global response programs (35 genes significant in ≥ 9 resident cell types) or cell type-specific responses (204 genes significant in 1 of 12 resident cell types). (D) The number of Reactome pathways that are either up-regulated (blue) or down-regulated (red) in the RbV group compared to controls. (E) Dot plot showing Reactome pathways that are altered in each type of resident cell. The size of each dot represents the number of genes in each cell type (columns), and the color of each dot indicates the normalized enrichment score of each pathway (rows). Pathways that were not significantly enriched (Q value ≥ 0.05, Benjamini-Hochberg correction) are not displayed.

Figure 3

Type I interferon responses are induced by IFNβ produced in microglia. (A) UMAP plot of the immune cell subset. Individual points represent single cells, which are color-coded by their assigned cell-type identity. (B) UMAP feature plot showing Ifnb1 expression, in which the color of each cell represents the log-scaled UMI counts. Ifnb1 transcripts are only detected in a small subset of microglia. (C) UMAP feature plot for Ifng expression. CD4⁺ T cells are the primary source of Ifng. (D) Dot plot showing the expression of selected cytokine genes (columns) in each immune cell type (rows). The dot color represents the average log-scaled expression of each gene in a given cell cluster, and the size of the dot represents the fraction of cells in that cluster in which transcripts for that gene were detected.

Figure 4

Microglia are found in distinct transcriptional states along an activation trajectory. (A) UMAP plot showing subclusters of microglia found using shared nearest-neighbors (SNN) graph-based clustering (see “Materials and Methods” section). Points are single cells color-coded by subcluster assignment. (B–E) UMAP feature plots showing the expression of genes that are differentially expressed between microglia subclusters. Cells are color-coded by their log-scaled UMI counts for the indicated gene. (F) Dot plot showing expression of several genes (rows) that are differentially expressed between microglia subclusters [columns—color-coded according to (A)]. (G) The inferred trajectory of microglial activation. Genes that were differentially expressed between microglia subclusters were used for trajectory inference. Individual cells were arranged along a single unbranched trajectory based on their pseudo-time values. Top: cells are color-coded by subclusters, following the color scheme in (A). Cells from subclusters I-II are on the left-most end of the trajectory, while cells from subcluster VI are on the opposite end. The remaining subclusters were distributed along the middle of the inferred trajectory. Middle/Bottom: inferred graphs with cells color-coded by their expression of the indicated genes, quantified as log₁₀(counts+0.1), which are differentially expressed between microglia sub-clusters.

Figure 5

The structure of intercellular interactions is altered by the immune response. Heatmap showing the change (RbV—Control) in the number of interactions between cell types resulting from RbV infection. Interactions were inferred from single-cell expression data using CellPhoneDB. Red colors indicate an increase in the number of interactions between a pair of cell types in the RbV group relative to Control, whereas blue colors indicate a decrease in the number of interactions.