Mapping Ion Channel mRNA Copy Number to Channel Conductance

Question

At least in invertebrates, it appears that within identified cell types, there are robust correlations between ion channel mRNA copy number (as identified by single-cell qRT-PCR) and the maximal whole-cell conductance for that ion channel species. For an example, see the following paper:

• Variable channel expression in identified single and electrically coupled neurons in different animals (Schulz et al., 2006)

These results suggest that in general, we can model the relationship between ion channel mRNA copy number $m$ and conductance $g$ as

$$ g = \alpha + \beta m + \epsilon $$

where $\epsilon$ is some noise term and $\alpha$ and $\beta$ depend on cell type and ion channel species. In general, one would need to fit these parameters from data. I have two questions:

1. Due to issues with capture efficiency, etc., is it reasonable to expect there to be similar linear relationships between UMI counts of ion channel transcripts from scRNA-seq and ion channel conductances? Or is only qRT-PCR sufficiently accurate for estimating copy number?

2. If one does not have conductance data available to fit a regression model, are there theoretical models (e.g. involving gene regulatory networks) from which one could derive expected mappings between ion channel copy number and conductance?

Answer 1

I don't have access to the paper so I can't fully evaluate the relationship you describe. I'm also neither an electrophysiologist nor a

However to answer your questions:

1. This to me is a solid "probably?". It is reasonable to expect a correlation, but that doesn't mean that one will exist for any given mRNA species, especially given the rather noisy nature of scRNA-seq data. (Maybe it's gotten better more recently, I haven't tracked the tech closely). Briefly, RT-PCR will almost certainly be more accurate, but scRNA-seq will probably approximate the same answer (one possibly helpful though slightly old reference).
2. I'd be pretty skeptical of such a model constructed without regard to empirical conductance measurements. I can readily find people claiming to have accomplished something like this. These models are directly based on conductance measurements so they are likely somewhat trustworthy, and their fit to data appears to be pretty good (see figure; I have only the vaguest idea of what's being plotted here, so read the paper!)