SIV infection induces alterations in gene expression and loss of interneurons in Rhesus Macaque frontal cortex during early systemic infection

Abstract

Understanding the neurobiological mechanisms underlying HIV-associated neurocognitive decline in people living with HIV is frequently complicated by an inability to analyze changes across the course of the infection and frequent presence of comorbid psychiatric and substance use disorders. Preclinical non-human primate simian immunodeficiency virus (SIV) models help address these shortcomings. However, SIV studies frequently target protracted endpoints, limiting our understanding of the neuromolecular alterations during the early post-infection window. To begin to address this knowledge gap, we utilized single nuclei transcriptomics to examine frontal cortex samples of rhesus macaques 10- and 20-days post-SIV infection, compared to non-infected controls. We identify and validated a decrease in inhibitory neurons during the early post infection window, representing a potential substrate of longer-term injury and neurocognitive impairment in people living with HIV. Differential expression identified alterations in cellular subtype gene expression that persisted over the 20-day time course and short-lived differences only detected at 10-days post-SIV infection. In silico predicted regulatory mechanisms and dysregulated neural signaling pathways are presented. Analysis of cell-cell interaction networks identify altered signal pathways in the frontal cortex that may represent regional alterations in cell-cell communications. In total, these results identify cell type-specific molecular mechanisms putatively capable of underlying long-term neurocognitive alterations in persons living with HIV.

Fig. 1. Selective loss of interneurons.

a After quality controls, 69,151 nuclei transcriptome profiles were used for unbiased clustering and are depicted as a uniform manifold approximation and projection (UMAP) dimension reduction plot. b Cell types were annotated by known expression markers. The diameter and color of the dots are proportional to the percentage of nuclei expressing the gene (Pct. Exp.) and the average expression level of the gene (Avg. Exp.), respectively, for each cell type. The cell type number and label color are aligned with those of the UMAP. The bar graphs for each cell type depict the proportion of nuclei that originated from the control (black), 10-days post-infection (DPI) (light grey), or 20 DPI (dark grey) treatment group for that cell type. The bottom bar depicts the proportion of all post-QC nuclei originating from each treatment group. c Proportion testing at 10 and 20 DPI was used to identify cell types with altered representation amongst the cortical composition. An increase in the number of microglia and a decrease in the proportion of pericytes and interneuron subtypes was observed at both timepoints. While this was seemingly accompanied by an increase in the proportion of an excitatory neuron population (cluster #2), repeating the proportion testing at 10 and 20 DPI in the absence of the inhibitory neuron populations (panel d demonstrated that the perceived increase in excitatory neurons was an artifact of the decrease of inhibitory neurons. Loss of inhibitory neurons was validated by immunohistochemistry for GAD67 e and isotype-matched IgG1 control f antibodies (n = 3/condition). 10 and 20 DPI there is a decrease in the percentage of GAD67 positive pixels in both grey matter g and white matter h.

Fig. 2. Differential expression.

a 10 days post-infection (DPI) there were 3490 cell type-specific differentially expressed genes (DEGs) with the majority being downregulated (2390 downregulated and 1100 upregulated, represented on the volcano plot by the blue and red dots, respectively). b By 20 DPI, the total number of DEGs was reduced to 1282, but the bias for downregulation of gene expression persisted (1164 downregulated and 118 upregulated). c 10 and 20 DPI transcriptome alterations include shared and timepoint-specific alterations in gene expression, with the vast majority of overlapping gene expression being down regulated. d Venn diagrams distinguishing the number genes that were different expressed at both 10 and 20 DPI for each cellular subtype. The percentage of non-overlapping altered genes varied by timepoint and cell type. The scatter plots depict the magnitude and direction (log fold change) of differential expression of gene differentially expressed at 10 DPI (x-axis) and 20 DPI (y-axis). All cell type specific genes with altered expression at both timepoints were altered with the same direction of effect, suggesting that these transcriptome alterations represent an enduring change throughout the examination timeframe.

Fig. 3. Cell-cell interaction networks.

a Alteration of known ligand (left column) and receptor (center column) pairs were utilized to identify altered interaction vectors (right column) at 10 DPI (top row) and 20 DPI (bottom row). b Circos plot depiction of the differing types of altered interaction vectors and 10 and 20 DPI. Colors around the outside of the Circos plot indicate the cluster contain the ligand or receptor for each interaction vector. The color of the interaction vector (with reduced opacity) indicates the cluster producing the ligand for each interaction vector.