A while ago, during SfN, I wrote about an interesting poster on clustering cell types by single cell RT PCR. The paper is out and I just wanted to share some details with you (In case you're too lazy to read it yourself ;-)).
The cells in the vestibular nuclei (and in the cerebellar nuclei I can tell you) are hard, if not impossible to distinguish electrophysiologically. So, if you want to find out what different cell types are doing during behavior you're going to have a hard time. No way to distinguish the glutamatergic projection neurons from the glycinergic ones and no way of telling if you're listening to an interneuron or to a GABAergic projection neuron. But Kodama et al used expression profiles of transmitter-related genes, ion channels and marker genes based on the allen brain atlas.
To be able to compare the results to previous studies they used three mouselines characterized before: YFP-16 (excitatory neurons), GIN (somatostatin, inhibitory neurons), GlyT2 (glycine transporter 2). Only five genes for neurotransmitters and genes related to neurotransmitters were used (VGluT1/2, glycine transporter 2 and Gad1/2). These genes clustered nicely on the different mouselines. Interestingly, the clusters are not perfect, proving that you always have some sort incompleteness and bleed-through with transgenetic animals.
Now the interesting part is if you can match the expression profile of ion-channel related genes to the physiology. For example: fluorescent neurons in YFP-16 animals have narrow action potentials. And GIN neurons show less rebound firing than YFP-16 neurons do. So, you would expect differences in ion-channels mediating action potential shape and differences in T-type calcium channels and H-channels. Indeed, these differences are reflected in the expression profiles. Genes for NaV1.1 and NaV1.6 are upregulated in YFP-16 neurons as compared to GIN and GlyT2 neurons. The same goes for the hyperpolarizing currents: Kcnc1, 2 and Kcnc3 were all upregulated in YFP-16 neurons. Also the differences seen in postinhibitory rebound firing were reflected in the expression profiles. HCN and combined T-type channel expression were upregulated in YFP-16 neurons.
Now six classes of neurons can be distinguished by marker genes.
Exc1: Vglut2/ Secreted phosphoprotein 1
Exc2: Vlugt1/ Corticotropin releasing hormone
Inh1: Nav beta4/ GlyT2
Inh2: Coagulation factor C homolog
Inh3: Corticotropin-releasing factor-binding protein
By doing in-situ hybridization combined with tracer injections, the authors were able to pinpoint the roles of some of the classes. Exc1 neurons project to the motor nuclei, Exc2 neurons project to the cerebellar cortex as mossy fibers, Inh1 neurons project to the motor nuclei as well, Inh3 neurons project to the vestibular nuclei and Exc3 and Inh2 neurons could not be traced. (Nucleo-olivary?)
The tactic used here to classify neurons has some clear advantages. Even neurons that cannot be clustered (easily) on the basis of electrophysiology alone can be identified using genetic expression clustering. Also, if specific markers are known, transgenic mouselines can be generated specifically for each cluster.
There are also a few things that worry me a bit about the paper. The spike-in RNAs used to quantify the expression profiles do not show the linear relationship that you would expect (fig 1E). In other words, it is not clear whether the results from the genetic profiles are compared to the linear fit and how the outlier is handled. Another concern is that only the MVN was used and only the central part of the MVN. What about the other nuclei and the periphery of the MVN? This is especially a concern since different neuronal morphologies are not uniformly present throughout the nucleus. So, maybe there is only a subsampling of the neurons in the present study.
Some more research is needed to address these issues. Still, I think the paper is a big leap forwards for cerebellar research.
Kodama T, Guerrero S, Shin M, Moghadam S, Faulstich M, & du Lac S (2012). Neuronal Classification and Marker Gene Identification via Single-Cell Expression Profiling of Brainstem Vestibular Neurons Subserving Cerebellar Learning. The Journal of neuroscience : the official journal of the Society for Neuroscience, 32 (23), 7819-31 PMID: 22674258