Identification of novel glial genes by single-cell transcriptional profiling of Bergmann glial cells from mouse cerebellum

Abstract

Bergmann glial cells play critical roles in the structure and function of the cerebellum. During development, their radial processes serve as guides for migrating granule neurons and their terminal endfeet tile to form the glia limitans. As the cerebellum matures, Bergmann glia perform important roles in synaptic transmission and synapse maintenance, while continuing to serve as essential structural elements. Despite growing evidence of the diverse functions of Bergmann glia, the molecular mechanisms that mediate these functions have remained largely unknown. As a step toward identifying the molecular repertoire underlying Bergmann glial function, here we examine global gene expression in individual Bergmann glia from developing (P6) and mature (P30) mouse cerebellum. When we select for developmentally regulated genes, we find that transcription factors and ribosomal genes are particularly enriched at P6 relative to P30; whereas synapse associated molecules are enriched at P30 relative to P6. We also analyze genes expressed at high levels at both ages. In all these categories, we find genes that were not previously known to be expressed in glial cells, and discuss novel functions some of these genes may potentially play in Bergmann glia. We also show that Bergmann glia, even in the adult, express a large set of genes thought to be specific to stem cells, suggesting that Bergmann glia may retain neural precursor potential as has been proposed. Finally, we highlight several genes that in the cerebellum are expressed in Bergmann glia but not astrocytes, and may therefore serve as new, specific markers for Bergmann glia.

Figure 1. Harvesting of individual Bergmann glia and quality control of single-cell cDNA.

Ai, a live cerebellar slice obtained from an adult GFAP-GFP transgenic mouse imaged under phase contrast optics (left) and fluorescence illumination (right). The Bergmann glia (see arrowheads) are the cells with the most fluorescence. Aii, freshly dissociated cells include putative astrocytes, which are devoid of processes and show relatively weak GFP fluorescence (arrowheads), and putative granule neurons, which have small, round, GFP-negative cell bodies (arrows). Aiii, a freshly dissociated Bergmann glia (arrowhead in left panel) can be distinguished from other cells by the bushy processes that emanate from one side of the soma–these are the long Bergmann glial processes that have partially retracted or been sheared off during tissue dissociation. In addition, Bergmann glia display strong GFP fluorescence (arrowhead in right panel), with the mean GFP intensity of their cell bodies 2.9±0.8 fold that of astrocytes; n = 17 cells). Aiv, a single Bergmann glia being washed by placement in a new dish containing fresh buffer, before being picked again with a new microelectrode. This step is performed to exclude contaminating cells or mRNAs. Scale bar, 70 µm in Ai, 25 µm in Aii and Aiii, 40 µm in Aiv. B, left panel, agarose gel electrophoresis of cDNAs generated from single GFP+ and GFP− cells. The gels show that most of the cDNA lies between 300 and 1000 bases. Right panel, agarose gels showing PCR with primer pairs directed towards β-actin (Actb), ornithine decarboxylase (Odc), Gfap, and neurofilament light chain (Nefl). The results show that the single cell cDNAs from Bergmann glia (GFP+) and neurons (GFP−) contain both high and low abundance transcripts (Actb and Odc, respectively). Bergmann glia are positive for the astroglial marker Gfap and negative for the neuronal marker Nefl, whereas neurons are negative for Gfap and positive for Nefl. C. Southern blot analysis of cDNAs from two putative Bergmann glial cells (GFP+) and a putative neuron (GFP−) shows presence of the high, medium, and two low abundance markers (Actb, Actg, and Odc and Ppp1ca, respectively). In addition, the GFP+ cells are positive for Gfap, Fabp7 (BLBP), Sept4 and Slc1a3 (GLAST), confirming their glial identity, whereas the GFP− cell lacks all these markers. Conversely, the GFP+ cells are absent for the neuronal markers Nefl and Mtap2 whereas the GFP− cell is positive. These results confirm the preservation of low to high abundance transcripts after the single-cell RT-PCR amplification, and also confirm the cell identity of the Bergmann glia used for microarray analysis.

Figure 2. Expression profiles of cell type control genes confirm purity of Bergmann glial cDNA generated by single cell RT-PCR.

Mean expression levels of well established markers for astroglia, neurons, oligodendrocytes and microglia were analyzed in the transcriptional profiles of ten Bergmann glia (five each from P6 and P30) using GeneSpring GX 7.3 software. For genes represented by multiple probe sets, the averaged expression of all probe sets were used. Astroglial genes were robustly expressed whereas markers of other cell types were absent or extremely low, confirming the astroglial identity of Bergmann glia and the absence of contaminating mRNAs from other cell types during cell harvesting. Error bars represent ± SEM.

Figure 3. Single cell RT-PCR of individual Bergmann glia is sufficiently accurate for comparison of expression profiles by age.

A, Scatter plots of raw gene expression level compared between two cells from the same age (P6-1 vs. P6-2; left panel) and between ages (P6-1 vs. P30-1; right panel) (a.u.: arbitrary units). The samples from the same age show high similarity (mean within-age correlation coefficient = 0.81), which is comparable to what is reported between individual cells of a glioblastoma cell line (0.86) . The high level of concordance between samples of the same age increases the reliability of comparisons between ages. Samples of different ages are significantly more divergent (mean P6 vs. P30 correlation coefficient = 0.66). B, Dendrogram and sample clustering of individual Bergmann glia. Unsupervised hierarchical clustering based on overall gene expression profiles reveals two distinct clusters corresponding to the two ages, P6 and P30. This suggests that the samples from the two ages do indeed represent two statistically distinct populations suitable for valid comparison.

Figure 4. Identification of a developmentally regulated GPCR that is Bergmann glia-specific in the cerebellum.

in situ hybridization with a ³³P-labeled probe for Gpr126, a little known GPCR of the adhesion family, reveals signal specifically in the Purkinje cell layer at P7 (arrows, middle panels), consistent with expression in Bergmann glia. Gpr126 expression is developmentally regulated, and becomes undetectable in the adult (right panel). Unlike most classic Bergmann glial markers, which are also expressed by cortical radial glia, Gpr126 is specific to Bergmann glia and not detected in cortical radial glia at E15 (arrow in top left panel). Labeling is seen in the ventricular zone of the developing cerebellar anlage at E15 (arrowheads in left panels), suggesting that Gpr126 may be expressed in progenitors of Bergmann glia. All sections are oriented with rostral to the right. Scale bar, upper panels: left, 3 mm; center, 2 mm; right, 2.5 mm; lower panels: left, 100 µm; center, 1 mm; right, 1.4 mm.