Barcoding Neurones

(December 6th, 2016) New ways of exploiting high throughput sequencing are emerging every day, but imagine our surprise when we heard of RNA technology being used for – wait for it – neuroanatomy.





When you are wielding a hammer like DNA technology, we perfectly understand it when everything looks like a nail. Anthony Zador and colleagues at Cold Spring Harbour laboratories have reported in Neuron how you can use barcoding of neurones with RNA to trace neuronal projections at a high-throughput scale – a method they call MAPseq.

The idea behind barcoding neurones with RNA is actually very simple. You inject a construct into a part of the brain that uniformly infects neurones with random stretches of RNA. Then you slice the brain and amplify up the RNA. Count the number of individual barcodes in each slice, and from there you work out the projections of single neurones.

Neurones can have several broad patterns of innovation – one neurone might innervate several targets, for instance, or there might be a one-to-one mapping with all the neurones from one point innervating the same targets, or there might even be divergence, with neurones from several locations arriving at the same targets.

Just how does barcoding tell you about innervation patterns? Take, for example, if the same barcode comes up in different target locations: then you know there is a divergent pattern of innervation. On the other hand, if barcodes turn up in restricted, but quite different, target locations, that tells you there are complex and specific patterns of innervation. Neither of these can be done with traditional staining.
But can it be as easy as that? What happens, say, if the same barcode ended up in more than one neurone? And what about if more than one barcode gets taken up by a neurone?

It turns out that probability actually works heavily in your favour, provided you get the injection concentrations right. Using the same maths that is used to work out the chance of 2 children in a class of 30 having a birthday on the same day (which is according to Wikipedia, about 65%), Zador worked out that hitting the 1,000 neurones in the mouse brain region called locus coeruleus (LC) with a million barcodes will mean that 99.9% of all neurones would be labeled uniquely. All the same, to be on the safe side, Zador amplified up the barcode RNA from several isolated cells, and found that about 80% of the neurones had one, and only one, barcode. In Zador’s MAPseq, each random RNA barcode is 30 nucleotides long, giving a theoretical diversity of some 1018 possible sequences – plenty to go around when you have only 108 or so neurones (as mice, the subjects of Zador’s experiments, do).

So that’s alright, then. But what if the RNA doesn’t spread all along the axon? To deal with this issue, the barcode RNA is built into a payload construct that also includes four copies of a nλ RNA binding domain plus a protein (pre-mGRASP) that targets to presynaptic terminals. Once delivered virally into the cell, the construct makes the barcode RNA under one promoter, and the pre-mGRASP/nλ under another promoter. The nλ snags the barcode onto the pre-mGRASP protein which then transports the barcode to the presynaptic terminals. 

The final step of the technique is stunningly elegant. To overcome possible biases in the amplification of the RNA, you need a method to count the individual RNA molecules. Zador did this by annexing a unique 12-nucleotide sequence to each RNA molecule recovered from the target tissue. In addition, they also added a slice-specific 6 nucleotide sequence. With that magic combination, you can mix all the recovered RNA together, amplify it up, and use the molecule- and slice-specific tags to work out how many of each barcode was in each slice.

MAPseq will not replace established methods of neuronal tracing. What it gains in throughput, it loses in resolution. And it involves a lot of nucleic acid manipulation and titration steps (tweaking). However, it does open the way to high-throughput neuroanatomy, especially where neurones from one structure innervate a wide region of the central nervous system.

Steven Buckingham

Picture: www.publicdomainpictures.net/Karen Arnold




Last Changes: 01.12.2017



Information 4


Information 5


Information 6