Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2016 Jul 26;16(4):1126-1137.
doi: 10.1016/j.celrep.2016.06.059. Epub 2016 Jul 14.

Cellular Taxonomy of the Mouse Striatum as Revealed by Single-Cell RNA-Seq

Affiliations
Free PMC article

Cellular Taxonomy of the Mouse Striatum as Revealed by Single-Cell RNA-Seq

Ozgun Gokce et al. Cell Rep. .
Free PMC article

Abstract

The striatum contributes to many cognitive processes and disorders, but its cell types are incompletely characterized. We show that microfluidic and FACS-based single-cell RNA sequencing of mouse striatum provides a well-resolved classification of striatal cell type diversity. Transcriptome analysis revealed ten differentiated, distinct cell types, including neurons, astrocytes, oligodendrocytes, ependymal, immune, and vascular cells, and enabled the discovery of numerous marker genes. Furthermore, we identified two discrete subtypes of medium spiny neurons (MSNs) that have specific markers and that overexpress genes linked to cognitive disorders and addiction. We also describe continuous cellular identities, which increase heterogeneity within discrete cell types. Finally, we identified cell type-specific transcription and splicing factors that shape cellular identities by regulating splicing and expression patterns. Our findings suggest that functional diversity within a complex tissue arises from a small number of discrete cell types, which can exist in a continuous spectrum of functional states.

Figures

Figure 1
Figure 1. Reverse engineering of mouse striatum by single-cell RNAseq
A) Workflow for obtaining and sequencing cDNA from single cells. Striatal slices from D1-tdTom/D2-GFP and Aldhl1-GFP mice were dissociated and cells collected by FACS or MACS. Cells were then captured, imaged, and cDNA amplified in microfluidic chips. B) Unbiased clustering of ten major classes of cells using t-distributed stochastic neighbor embedding (tSNE), which distributes cells according to their whole-transcriptome correlation distance. Each single cell is represented as a dot and colored by a clustering algorithm (DBSCAN). C) Box-and-whisker plots showing total number of genes detected per cell for major cell types. D) Expression of putative marker genes for each of 10 major cell types. Scaled expression of marker genes is shown by the color of the cell points. Each tSNE cluster is enriched for one marker, and we were able to cells to one of 10 major cell types. E) Heatmap of top 50 genes most highly correlated to each cell type. Each row is a single cell and each column is a single gene. Bar on the right shows the experimental origin of cells. Bar on the left shows DBSCAN clustering of cells and the bottom shows the cell type assignment for each set of 50 genes. Within each 50-gene set, the genes are ordered by increasing p-value of correlation to that cell type from left to right.
Figure 2
Figure 2. Characterization of discrete MSN subtypes
A) Hierarchical clustering of the highest pairwise gene correlations in MSNs shows two strongly anticorrelated clusters of genes which include known MSN subtype markers for D1- and D2-MSNs. B) rPCA of MSNs, Cells (rows) are ordered by their projection onto PC1 and genes (columns) by their positive (left) or negative (right) contribution to PC1. This identifies three molecularly distinct populations, assigned to D1-MSN (red), D2-MSN (green) and D1/2 hybrid MSN (yellow). Bar on the left shows the experimental origin of cells. C) The distribution of single MSNs projected onto the D1-D2 scores. The D1 and D2 score is calculated by summing the scaled expression values of the genes shown in Fig 2B. D1- and D2-MSNs form distinct peaks, and the novel MSNs are distributed between the 2 peaks. D) Biplot of D1- and D2-MSN cells by their expression of D1 genes (y-axis) and of D2 genes (x-axis). Scaled expression of Drd1 is shown by the color of the cell points. E) Sagittal brain section of a D2-GFP/D1-TdTom double reporter mouse showing tdTom-GFP double positive novel MSNs in both nucleus accumbens and dorsal striatum. F) Confocal imaging of striatal slices of D2-GFP/D1-TdTom double reporter mice demonstrating the existence of D1/2 hybrid MSNs in striatum. G) rPCA of MSNs in D1 part of Figure 2D using all expressed genes, cells (rows) are ordered by their projection onto PC1 and genes (columns) by their positive (left) or negative (right) contribution to PC1. This identifies two molecularly distinct populations, assigned to Major (Foxp1) D1-MSN (red), and Pcdh8-MSN (yellow). H) Biplot of major D1- and Pcdh8-MSN cells by their expression of D1 genes (y-axis) and of Pcdh8 genes (x-axis). Scaled expression of Penk is shown by the color of the cell points. I) The distribution of MSNs projected onto Pcdh8-Foxp1 scores. Major-D1 and Tacr1-MSNs form distinct peaks J) Box plots showing specific Pcdh8 and Tacr1 expression in Tacr1-MSNs compared to other cell types in MSN. K) rPCA of MSNs in D2 part of Figure 2D, cells (rows) are ordered by their projection onto PC1 and genes (columns) by their positive (left) or negative (right) contribution to PC1. This identifies two molecularly distinct populations, assigned to D2-MSN (green), and Htr7-MSN (light green). L) Biplot of D2- and Htr7-MSN cells by their expression of Htr7- genes (y-axis) and of Synpr- genes (x-axis). Scaled expression of Tac1 is shown by the color of the cell points. M) The distribution of MSNs projected onto Htr7-Synpr scores. Major D2- and Htr7-MSNs form distinct peaks N) Box plots showing specific Htr7 and Agtr1a expression in Htr7-MSNs compared to other cell types in MSN.
Figure 3
Figure 3. Identification of heterogeneity within MSN subtypes
A) rPCA of major D1-MSNs using all expressed genes, reveals a molecularly distinct sub-population of D1-Dner MSNs. Bar on the left shows the experimental origin of cells. B) The distribution of D1-MSNs projected onto Meis2-Dner gene group scores. C) Biplot of D1- and D1-Dner MSNs by their expression of D1-Meis2 genes (y-axis) and of D1-Tnnt1 genes (x-axis). Scaled expression of Meis2 and Tnnt1 are shown by the color of the cell points D) rPCA of D1-MSNs using all expressed genes, reveals a continuous transcriptional gradient marked by opposing expression gradients of Cnr1 and Wfs1. Bar on the left shows the experimental origin of cells. E) The distribution of D1-MSNs projected onto Cnr1-Wfs1 gradient scores. F) Biplot of D1-MSNs by their expression of Wfs1 gradient genes (y-axis) and of Cnr1 gradient genes (x-axis). Scaled expression of Cnr1 and Wfs1 are shown by the color of the cell points. G) rPCA of D2-MSNs using all expressed genes, reveals a sub-population of D2-Cartpt MSNs. Bar on the left shows the experimental origin of cells. H) The distribution of D2-MSNs projected onto Calb1-Cartpt gene group scores. I) Biplot of D2-Calb1 and D2-Cartpt MSNs by their expression of Cartpt group genes (y-axis) and of Calb1 group genes (x-axis). Scaled expression of Calb1 and Cartpt are shown by the color of the cell points. J) rPCA of D2-MSNs using all expressed genes, reveals a continuous transcriptional gradient marked by opposing expression gradients of Cnr1 and Crym. Bar on the left shows the experimental origin of cells. K) The distribution of D1-MSNs projected onto Cnr1-Crym gradient scores. L) Biplot of D1- MSNs by their expression of Crym gradient genes (y-axis) and of Cnr1 gradient genes (x-axis). Scaled expression of Cnr1 and Crym are shown by the color of the cell points M) rPCA of Pcdh8-MSNs using all expressed genes, reveals a continuous transcriptional gradient marked by opposing expression gradients of Cnr1 and Wfs1 similar to 2D. Bar on the left shows the experimental origin of cells N) The distribution of Pcdh8-MSNs projected onto Cnr1-Wfs1 gradient scores. O) Biplot of Pcdh8-MSNs by their expression of Wfs1 gradient genes (y-axis) and of Cnr1 gradient genes (x-axis). Scaled expression of Cnr1 and Wfs1 are shown by the color of the cell points. P) rPCA of Htr7-MSNs, reveals a continuous transcriptional gradient marked by opposing expression gradients of Cnr1 and Th. Bar on the left shows the experimental origin of cells Q) The distribution of Htr7-MSNs projected onto Cnr1-Th gradient scores. R) Biplot of Htr7-MSNs by their expression of Th gradient genes (y-axis) and of Cnr1 gradient genes (x-axis). Scaled expression of Cnr1 and Th are shown by the color of the cell points.
Figure 4
Figure 4. Characterization of vascular cells, astrocytes, and oligodendrocytes
A) Hierarchical clustering of the highest pairwise gene correlations within vascular cells reveals two large clusters of subtype specific transcripts. B) rPCA of 43 striatal vascular cells identifies two molecularly distinct populations, assigned to vascular smooth muscle cells (VSMC) that express Myl9 and Tagln and endothelial cells that express Ly6e and Pltp. C) A histogram of vascular cells’ PC1 scores shows a clearly bimodal distribution and confirms the existence of two distinct subtypes. D) Hierarchical clustering of the highest pairwise gene correlations within immune cells reveals two large clusters of interconnected transcripts. E) rPCA of 119 immune cells (microglia and macrophages) identifies two molecularly distinct populations, assigned to microglia that express Sparc and macrophages that express Mrc1. F) The distribution of single immune cells along the first PC1 of rPCA. Cells form two distinct peaks. G) Hierarchical clustering of the highest pairwise gene correlations within oligodendrocytes reveals two large clusters of interconnected transcripts. H) rPCA of 43 striatal oligodendrocytes identifies two distinct oligodendrocyte populations: mature oligodendrocytes (MO) which express Klk6 and Sec11c, and newly formed oligodendrocytes (NFO) which express Nfasc and Ckb. There are also a significant number of oligodendrocytes with intermediate expression of both sets of genes, which are likely transitioning between NFO and MO. I) The distribution of single oligodendrocytes along the first PC1 of rPCA. MO and NFO cells form two distinct peaks with transitioning oligodendrocytes bridging the two peaks. J) Hierarchical clustering of the highest pairwise gene correlations within astrocytes did not reveal interconnected transcripts, instead showing low correlation values between genes. J) rPCA of 107 single striatal astrocytes reveals transcripts that increase or decrease continuously without defining distinct subpopulations. Cells (rows) are ordered by their PC1 scores and genes (columns) are ordered by their PC1 loading. The continuum of astrocyte transcriptional states on one side is marked by higher expression of transcripts related to synaptic communication (Slc6a11, Slc6a1, Slc6a9, Gria1 and Gria2) and on the other side transcripts related to translation (Rpl9, Rpl14, Rps5 and Rpsa) and cell polarity regulators (Cdc42). L) The distribution of single astrocytes along PC1 is unimodal indicating striatal astrocytes exhibit continuous transcriptional variation within one discrete subtype.
Figure 5
Figure 5. Cell type specific transcriptional factors identified by single-cell transcriptome analysis
A) Correlogram visualizing correlation of single-cell gene expression between cell type-specific TFs. B) Scaled expression of the TFs that are most specific to discrete cell types. Scaled expression of marker genes is shown by the color of the cell points.
Figure 6
Figure 6. Single-cell transcriptome analysis reveals splicing factors specific to neurons
A) Hierarchical clustering of pairwise gene correlations for cell type specific splicing factors (SF). B) tSNE plot of single cells colored by the scaled expression selected splicing factors: the neuronal-enriched splice factors Celf4 and Rbfox1 and the oligodendrocyte-enriched splice factor Pcbp4.
Figure 7
Figure 7. Differential splicing analysis reveals several modes of single-cell splicing regulation
A) Expression of selected splice sites that have significant cell type-specific regulation. The number of reads per cell (log10) aligning to the major and minor variants are plotted on the x- and y-axes, respectively. Many single cells are located on the diagonal of the plots, indicating they express both the major and minor variants of that splice site (compound splice sites). B) Box plots showing the total number of compound splice sites detected per cell across cell types.

Similar articles

See all similar articles

Cited by 78 articles

See all "Cited by" articles

MeSH terms

LinkOut - more resources

Feedback