Comparative description of the mRNA expression profile of Na+ /K+ -ATPase isoforms in adult mouse nervous system

Abstract

Mutations in genes encoding Na⁺ /K⁺ -ATPase α1, α2, and α3 subunits cause a wide range of disabling neurological disorders, and dysfunction of Na⁺ /K⁺ -ATPase may contribute to neuronal injury in stroke and dementia. To better understand the pathogenesis of these diseases, it is important to determine the expression patterns of the different Na⁺ /K⁺ -ATPase subunits within the brain and among specific cell types. Using two available scRNA-Seq databases from the adult mouse nervous system, we examined the mRNA expression patterns of the different isoforms of the Na⁺ /K⁺ -ATPase α, β and Fxyd subunits at the single-cell level among brain regions and various neuronal populations. We subsequently identified specific types of neurons enriched with transcripts for α1 and α3 isoforms and elaborated how α3-expressing neuronal populations govern cerebellar neuronal circuits. We further analyzed the co-expression network for α1 and α3 isoforms, highlighting the genes that positively correlated with α1 and α3 expression. The top 10 genes for α1 were Chn2, Hpcal1, Nrgn, Neurod1, Selm, Kcnc1, Snrk, Snap25, Ckb and Ccndbp1 and for α3 were Sorcs3, Eml5, Neurod2, Ckb, Tbc1d4, Ptprz1, Pvrl1, Kirrel3, Pvalb, and Asic2.

Keywords: Na/K ATPase; expression profile; mouse brain.

FIGURE 1

Expression and proportion of transcripts encoding the Na⁺/K⁺‐ATPase subunits and isoforms in various cell classes. Each circle represents one Na⁺/K⁺‐ATPase isoform in a specific cell class. The color and size of each circle represent the average expression level and the proportion of cells expressing the corresponding isoform, respectively. (a1) and (a2) were generated using data from Saunders et al. (2018) and Zeisel et al. (2018) databases, respectively. The determinations of expression and proportion are described in Section 2.1

FIGURE 2

Expression and correlation of Na⁺/K⁺‐ATPase subunits and their isoforms within the CNS neuronal clusters. (a1,a2) Violin plots for mean expressions of Na⁺/K⁺‐ATPase subunits and isoforms within CNS neuronal clusters. Each dot represents one neuronal cluster, while each violin plot reflects the expression distribution of each individual isoform in the neurons from the mouse brain. (b1,b2) Heatmaps of mean expressions for α1, α3, and β1, the major candidates to form the minimal functional Na⁺/K⁺‐ATPase complexes. Each row represents one neuronal cluster, while the color is assigned according to the mean expression value of the gene in the corresponding cluster. More comprehensive heatmaps of all CNS Na⁺/K⁺‐ATPase subunits and isoforms are shown in Figure S3. (c1,c2) Spearman's correlation coefficients between α1, α3, and β1. The colors are assigned according to the values of correlation coefficients. Spearman's correlation coefficients between all CNS Na⁺/K⁺‐ATPase subunits and isoforms are shown in Figure S3. (d1,d2) Euclidean distance trees for the α1, α3, β1, and neurotransmitter markers. The distance refers to the divergence between gene expression profiles. The colors represent the grouping of genes based on the pairwise distances. a1, b1, c1, and d1 were generated using data from Saunders et al. (2018), while a2, b2, c3, and d3 were made using data from Zeisel et al. (2018). The average gene expression per cluster was determined as described in Section 2.1. Spearman's correlations in (b,c) and Euclidean distances in Figure 2(d) were calculated as described in Section 2.3

FIGURE 3

Identifying neuronal clusters with high expression of transcripts encoding for α1. (a1,a2) Scatter plots of the expression of α1 transcripts versus the proportion of cells expressing α1 for CNS neuronal clusters. Each circle represents one neuronal cluster. Clusters above the 95th percentile for either variable are shown in coral or cyan, the latter corresponding to those clusters with Z scores higher than 2. Red dotted lines define the 95th percentile limits. (b1,b2) Histograms for Z scores from CNS neuronal clusters. Red dotted lines define Z scores equal to 2. The number of clusters in bins with Z larger than 2 are shown above the bars. a1 & b1 and a2 & b2 were generated using data from Saunders et al. (2018) and Zeisel et al. (2018), respectively. The determinations of expression and proportion were described in Section 2.1. The gross product and Z score were calculated as described in Section 2.2

FIGURE 4

Identifying neuronal clusters with high expression of transcripts encoding for α3. (a1,a2) Scatter plots of the expression of α3 transcripts versus the proportion of cells expressing α3 for CNS neuronal clusters. Each circle represents one neuronal cluster. Clusters above the 95th percentile for either variable are shown in coral or cyan, the latter corresponding to those clusters with Z scores higher than 2. Red dotted lines define the 95th percentile limits. (b1,b2) Histograms for Z scores from CNS neuronal clusters. Red dotted lines define Z scores equal to 2. The number of clusters in bins in which Z was >2 are shown above the bars. a1 and b1, and a2 and b2 were generated using data from Saunders et al. (2018) and Zeisel et al. (2018), respectively. The determinations of expression and proportion were described in Section 2.1. The gross product and Z score were calculated as described in Section 2.2

FIGURE 5

Expression spectrum of the α and β isoforms in the mouse cerebellum. (a1,a2) Heatmaps of mean expressions for the α and β isoforms in cerebellar neuronal clusters. Each row refers to each cerebellar neuronal cluster, where its color is assigned according to its expression value. a1 and a2 were generated using data from Saunders et al. (2018) and Zeisel et al. (2018), respectively. The determinations of expression were described in Section 2.1. (b, c) Representative images of immunostaining of mouse cerebellar slices using α1 and α3 antibodies. Purkinje cells (yellow arrow), granule cells (blue arrow), Golgi cells (red arrow), basket cells (purple arrow) and glomeruli (white arrow). GL, granular layer; ML, molecular layer; PCL, Purkinje cell layer. Dashed lines represent the approximate locations of the PCL. N = 2 (animals), n = 8 (slices) for two‐month‐old WT mice. Scale = 20 μm. (d) Illustration of the cerebellar neuronal network. Filled‐colors refer to the preferential expressions of either α1 (light blue) or α3 (coral), while outline‐colors represent whether the neuron is excitatory (cyan) or inhibitory (red)

FIGURE 6

Genes correlated with the expression of α1. (a) Volcano plot of Spearman's correlation coefficients versus ‐log10 (FDR‐adjusted p values). Each symbol represents one test gene. The half‐symbols on the top edge is the default manner that R package uses to plot those genes with infinity values. Top 10 genes with highest positive correlations are marked with red color. Spearman's correlation and the corresponding log10 transformed FDR‐adjusted p values were described in Sections 2.4 and 3.5. (b) A hierarchical clustering tree that summarizes the significant biological processes. The sizes of the blue dots and their corresponding values refer to the Enrichment FDR for the specific biological process, calculated by ShinyGo software (see Section 2.5)

FIGURE 7

Genes correlated with the expression of α3. (a) Volcano plot of Spearman's correlation coefficients versus ‐log10 (FDR‐adjusted p values). Each symbol represents one test gene. Top 10 genes with highest positive correlations are marked with red color. Spearman's correlation and the corresponding log10 transformed FDR‐adjusted p values were described in Sections 2.4 and 3.5. (b) A hierarchical clustering tree that summarizes the significant biological processes. The sizes of the blue dots and their corresponding values refer to the Enrichment FDR for the specific biological process, calculated by ShinyGo software (see Section 2.5)