Neuronal classification and marker gene identification via single-cell expression profiling of brainstem vestibular neurons subserving cerebellar learning - PubMed (original) (raw)

Neuronal classification and marker gene identification via single-cell expression profiling of brainstem vestibular neurons subserving cerebellar learning

Takashi Kodama et al. J Neurosci. 2012.

Abstract

Identification of marker genes expressed in specific cell types is essential for the genetic dissection of neural circuits. Here we report a new strategy for classifying heterogeneous populations of neurons into functionally distinct types and for identifying associated marker genes. Quantitative single-cell expression profiling of genes related to neurotransmitters and ion channels enables functional classification of neurons; transcript profiles for marker gene candidates identify molecular handles for manipulating each cell type. We apply this strategy to the mouse medial vestibular nucleus (MVN), which comprises several types of neurons subserving cerebellar-dependent learning in the vestibulo-ocular reflex. Ion channel gene expression differed both qualitatively and quantitatively across cell types and could distinguish subtle differences in intrinsic electrophysiology. Single-cell transcript profiling of MVN neurons established six functionally distinct cell types and associated marker genes. This strategy is applicable throughout the nervous system and could facilitate the use of molecular genetic tools to examine the behavioral roles of distinct neuronal populations.

PubMed Disclaimer

Figures

Figure 1.

Figure 1.

Molecular anatomy of the MVN. A, Examples of marker gene candidates. Using the Allen Mouse Brain Atlas (Lein et al., 2007), genes expressed in subpopulations of MVN neurons were identified as marker candidates. In situ hybridization images of exemplar marker gene candidates Col5a1, Spp1, and Crhbp are shown. The images were taken from the boxed region in the schematic drawing of the mouse coronal brainstem section (left,∼−5.9 mm from bregma). The distinct distribution of hybridization signal across the three genes imply their potential use as marker genes. Scale bar, 300 μm. B, Representative MVN fluorescent images of YFP-16, GIN, and GlyT2 mouse lines. The single-cell cDNAs were prepared from fluorescent protein-expressing MVN neurons in the three lines. MVNp, parvocellular MVN; MVNm, magnocellular MVN; NPH, nucleus prepositus hypoglossi; 4V, fourth ventricle (darkened for contrast). Scale bar, 200 μm. C, A representative image of MVN after punching out its central portion for single-cell cDNA amplification. Scale bar, 200 μm. D, qPCR results for Gapdh before and after the 10 cycle amplification. Data from all of the single-cell cDNAs (n = 167) are plotted. The linear relationship indicates no systematic distortion in abundance representation of transcripts by the 10 cycle amplification. E, The abundance representation of the spike-in RNAs in the single-cell cDNAs (the 10 cycle PCR product). The mean Ct values in qPCR are plotted against the original copy number of the spike-in RNAs (log-scaled). The cases where Ct value was not determined (no transcript was detected) are excluded from averaging. Lys, Phe, and Thr were on a regression line in the order of copy number, indicating that abundance representation of the three spike-in RNAs was preserved. Dap is not on the same regression line, which is at least partly due to the lower amplification rate of Dap in qPCR (1.79) than the other three (Lys, 2.01; Phe, 1.97; Thr, 2.00). _Y_-intercept of the regression line is 26.2. Error bar indicates SD. F, The heat map of the single-cell cDNAs (columns) based on the transcript profiles of the transmitter-related genes (rows). Color in the first row indicates the mouse lines from which the single-cell cDNAs were prepared (YFP-16: n = 61, orange; GIN: n = 52, blue; GlyT2: n = 54, green). In the heat map showing qPCR results, Ct values lesser and greater than the mean Ct of spike-in RNA Thr (five copies per sample) are coded in red-yellow and blue-gray gradation, respectively. Excitatory (_Slc17a6_-positive) and inhibitory (_Gad1_-positive) cell groups are defined clearly by the transcript profiles. As expected, YFP-16 neurons are both excitatory and inhibitory, while GIN and GlyT2 cells are only inhibitory. Note the coexpression of Gad1, Gad2, and Slc6a5 in the inhibitory cell group.

Figure 2.

Figure 2.

Ion channel transcript profiles correspond with electrophysiological properties of MVN neurons. A, Dendrogram and heat map of single-cell cDNAs from each neuron (columns) based on the ion channel transcript profiles (rows). The color coding of the mouse lines and the color scale of heat map are the same as in Figure 1_F_. Ion channel transcript profiles divide the single-cell cDNAs into two groups, bisected by the vertical dotted line; one predominantly comprises YFP-16 cells, and the other largely comprises GIN and GlyT2 cells. The bar graph at right indicates MI of the ion channel genes, reflecting their contribution to the primary dichotomy of the dendrogram. Genes in the heat map are sorted by MI; genes ranking high in MI are expressed at higher levels in the YFP-16-rich branch and include ion channels essential for sustained firing at high rates. Ct values represented as gray-blue indicate very low expression levels and were not used in the dendrogram calculations to exclude the influence of stochastic error in gene expression regulation. B, Comparison of NaV, KV3, and BK channel conductances recorded in acutely dissociated MVN neurons, g(NaV), g (KV3), and g(BK), respectively, with corresponding transcript levels (qPCR Ct values). The transcript profiles are congruent with electrophysiological data and reveal the ion channel subtypes underlying the electrophysiologically measured conductances. Significant differences in conductances and transcript levels between YFP-16 and GIN neurons are indicated by *p < 0.05 and ***p < 0.0001. The conductance data were reanalyzed from Gittis and du Lac (2007) (for NaV: YFP-16, n = 39, GIN, n = 35; for KV3: YFP-16, n = 20, GIN, n = 19; for BK: YFP-16, n = 10, GIN, n = 5). C, Action potential waveforms from 2 GlyT2 neurons are shown above a plot of maximum firing rate evoked during 1 s of intracellular depolarization versus action potential width (measured at half-height) for neurons recorded from three mouse lines. As predicted from ion channel gene expression profiles, GlyT2 neurons range widely in electrophysiological properties. YFP-16: n = 40, orange squares; GIN: n = 43, blue triangles; GlyT2, n = 41, green open circles. YFP-16 and GIN data are replotted from Bagnall et al. (2007). Scale for action potential waveform: 40 mV and 4 ms for cells 1 and 2 and 40 mV and 1 ms for the overlay.

Figure 3.

Figure 3.

Identification of marker genes associated with the neurochemically and electrophysiologically homogeneous cell types. The dendrogram of the single-cell cDNAs (columns) was computed using the transcript profiles of the transmitter-related genes, the ion channel genes, and the marker gene candidates selected using Allen Mouse Brain Atlas (Lein et al., 2007). The color code rows under the dendrogram indicate mouse lines as in Figure 1_F_ and Figure 2_A_. In the dendrogram, the single-cell cDNAs are divided into six cell types (E1–3, I1–3) within which expression profiles of the transmitter-related and ion channel genes are generally homogeneous. The heat map shows the transcript profiles (qPCR Ct values) of the transmitter-related genes (n = 5), the ion channel genes that ranked high MI in Figure 2_A_ (n = 8) or differentially expressed between E1 and E2 (E1 vs E2, n = 2) or I2 and I3 (I2 vs I3, n = 3), and the marker gene candidates. Those associated with the defined cell types are labeled as “marker genes” (n = 4), those serving as marker in combination are labeled as “intersectional” (n = 2), and the others are labeled as “other candidates” (n = 8).

Figure 4.

Figure 4.

Validation of identified marker genes with double fluorescent in situ hybridization. In each part, an in situ hybridization image of a coronal MVN section is shown in the top row (i), and magnified images of the white-boxed area are shown in the bottom row (ii–iv). A, Slc17a7 (magenta) and Crh (inset, green). The in situ hybridization signals of Slc17a7 and Crh were sparsely distributed over the MVN, and greatly overlapped each other. The arrowhead indicates the granule cell layer in the cerebellum. B, Adcyap1 (magenta) and Slc17a6 (green). _Adcyap1_-positive cells were scattered in the MVN, especially near the fourth ventricle. In the majority of cases, they colabeled with Slc17a6. C, Coch (magenta) and Gad1 (green). _Coch_-positive cells were nearly always colabeled by Gad1. The hybridization signal was also observed in the choroid plexus in the fourth ventricle. D, Crhbp (magenta) and Gad1 (green). _Crhbp_-positive cells were also colabeled by Gad1 (green, iii and iv), forming a dense cluster near the fourth ventricle and showing distinct distribution from that of _Coch_-positive cells. Scale bar: 150 μm for the top row images (i) and the inset images in A; 50 μm for the bottom images (ii, iii, iv).

Figure 5.

Figure 5.

Projection target and Purkinje cell innervation of molecularly defined cell types. A, Retrograde labeling of precerebellar neurons followed by in situ hybridization (top) or immunostaining against Spp1 (bottom) demonstrates that precerebellar neurons largely express Slc17a7, but not Spp1, indicating that E2 corresponds to precerebellar neurons. For retrograde labeling, FG and glycoprotein-deleted rabies virus are used for in situ hybridization and immunostaining, respectively. B, Excitatory premotor neurons retrogradely labeled by BDA. The top and bottom row show the cells labeled by injection into the contralateral abducens (cAbd) and the contralateral OMN (cOMN), respectively. The left, middle, and right columns show BDA signal, Spp1 immunoreactivity, and merged images, respectively. The majority of the retrogradely labeled neurons are immunostained against Spp1, indicating that E1 corresponds to premotor neurons. C, A representative image of MVN neurons densely innervated by Purkinje cell terminals (*). Pcp2 immunoreactivity visualizes Purkinje cell terminals. Spp1+/GFP + cells in the GlyT2 line mostly include cells densely innervated by Purkinje cell terminals. D, A representative distribution map of Spp1 immunoreactivity positive cells (Spp1+). Filled green dots and open circles indicate the presence and absence of GFP signal in the GlyT2 line, which presumably correspond to inhibitory and excitatory cells, respectively. Pink stars indicate cells densely innervated by Purkinje cell terminals (dFTN; Shin et al., 2011), which also express GFP. Note that Spp1-positive inhibitory cells in the ventrolateral MVN are likely to belong to I1, which include some dFTNs. dFTNs near the fourth ventricle might belong to I2, because they are in the territory of cells expressing Coch (Fig. 4_C_, I2 marker). The image shown in C corresponds to the boxed area.

Figure 6.

Figure 6.

Summary of marker genes and projection targets of the six cell types identified with our molecular classification strategy. E1–3 are excitatory neurons; I1–3 are inhibitory neurons.

Similar articles

Cited by

References

    1. Aizenman CD, Linden DJ. Regulation of the rebound depolarization and spontaneous firing patterns of deep nuclear neurons in slices of rat cerebellum. J Neurophysiol. 1999;82:1697–1709. - PubMed
    1. Bagnall MW, Stevens RJ, du Lac S. Transgenic mouse lines subdivide medial vestibular nucleus neurons into discrete, neurochemically distinct populations. J Neurosci. 2007;27:2318–2330. - PMC - PubMed
    1. Bagnall MW, Zingg B, Sakatos A, Moghadam SH, Zeilhofer HU, du Lac S du. Glycinergic projection neurons of the cerebellum. J Neurosci. 2009;29:10104–10110. - PMC - PubMed
    1. Balaban CD, Beryozkin G. Organization of vestibular nucleus projections to the caudal dorsal cap of kooy in rabbits. Neuroscience. 1994;62:1217–1236. - PubMed
    1. Barmack NH, Fagerson M, Errico P. Cholinergic projection to the dorsal cap of the inferior olive of the rat, rabbit, and monkey. J Comp Neurol. 1993;328:263–281. - PubMed

Publication types

MeSH terms

Substances

LinkOut - more resources