“With MAPseq, we’re trying to figure out where individual neurons end up,” said Justus Kebschull, the study’s first author and a graduate student in the laboratory of Anthony Zador at Cold Spring Harbor Laboratory. Zador’s group was motivated to develop MAPseq by the lack of high-throughput methods to track individual neurons.
In the new study, the researchers injected a deactivated virus containing a large number of RNA molecules into a selected brain region. Each RNA molecule has its own unique “barcode” sequence, and each one ends up in a neuron and gets carried through axons. After a couple of days, the researchers dissected the brain and collected and sequenced the RNA barcodes from different regions.
By matching the RNA barcodes from the injected region to those in various other regions in the brain, the researchers can trace where individual neurons end up. “This is the first example of using sequencing and barcodes for high-throughput neuroanatomy,” said Kebschull.
Mapping Connections Quickly
Figuring out how neurons are connected is a key part of understanding how the brain works. “If you think of the brain like a computer, we need to know what’s connected to what, for example whether the hard drive is connected to the processor,” Kebschull said.
In many current brain mapping techniques, researchers express a fluorescent marker in neurons and use a microscope to see which brain regions light up. But “every area of the brain has thousands and hundreds of thousands of neurons,” said Kebschull. These “bulk tracing” methods can’t differentiate between individual neurons, and as a result, researchers can’t distinguish whether two source neurons end up in the same region or in different regions. That’s important information. “Different neurons may have different properties, and may carry different information.”
Other brain mapping techniques can trace single neurons, but these methods are both labor-intensive and time-consuming. “It takes weeks to trace a single one,” said Kebschull. “If you want to look at hundreds of thousands, you’re basically screwed.” In contrast, MAPseq takes about a week to map a couple of thousand neurons from a single injection site, about a thousand-fold improvement in speed.
Adapting Existing Tools
Zador and his team were inspired by previous brain mapping techniques, such as the Brainbow approach in which researchers randomly expressed fluorescent markers to label different neurons with a wide array of colors (2). But there’s a limit to the number of colors that can be easily distinguished using microscopy, and for practical purposes, researchers could trace at most ~10 neurons at a time in those models.
“With MapSeq, we thought of switching from colors to random nucleic acid sequences,” Kebschull said. Using nucleic acid barcode sequences would provide a much higher diversity of markers, and these could be quickly and relatively inexpensively read using current high-throughput sequencing methods. Starting with a sequence of 30 nucleic acids, the researchers would end up with 1018 possible sequences, 10 orders of magnitude more than the number of neurons in a mouse brain.
Zador first suggested using barcode sequencing for high-throughput brain mapping in 2012, and the researchers then had to figure out the technical details (3). They decided to use a deactivated Sindbis virus, a commonly used tool in neuroscience, to transport the barcode RNA sequences into mouse neurons. But they immediately ran into a complication.
“The virus that people thought was non-replicating was actually replicating,” Kebschull said. As a result, viral particles propagated beyond the primary infected neuron to label other neurons; this would complicate mapping analysis. For the new technique, Kebschull created a recombinant Sindbis virus that infected neurons but didn’t propagate, paving the way for further refinements (4).
Once the RNA barcodes were delivered into the neurons, the researchers had to ensure that they were transported along axons to their terminals. They added an engineered protein to the barcode RNA for this purpose, and showed that it worked efficiently. In addition, the researchers wanted to ensure that each labeled neuron received a unique barcode, and they calculated how many viruses had to be injected in order to achieve this goal. Kebschull and Zador also figured out how to accurately identify the barcodes using high-throughput sequencing (5).
Testing the Technique
Zador’s team tested MAPseq in a part of the mouse brain called the locus coeruleus (LC), which serves as a source of the hormone noradrenaline. Noradrenaline modulates alertness and attention, and it was still unclear whether neurons from the LC stretch all over the cortex or only to specific regions. “The LC was a nice test case,” said Kebschull.
Using MAPseq, the researchers found a wide variety of neuronal projection patterns from the LC, with some neurons going to one specific target and others spreading out more widely. As a result, noradrenaline from the LC could travel to the whole cortex, as well as have more acute effects on specific regions, consistent with previous findings.
The current study used just one viral injection into the LC, and Zador and his team are working on increasing that number. Eventually they hope to use MAPseq to simultaneously trace neurons all across the cortex. “We’re scaling up this method to do more high-throughput neuroanatomy,” Kebschull said. The researchers also plan to use MAPseq in different mouse models of disease. “We can look at different mouse models of autism or developmental diseases, to see how the connectivity of the brain is miswired.”
1. Kebschull JM, Garcia da Silva P, Reid AP, Peikon ID, Albeanu DF, Zador AM. High-Throughput Mapping of Single-Neuron Projections by Sequencing of Barcoded RNA.
Neuron. 2016 Aug 17. pii: S0896-6273(16)30421-4. doi: 10.1016/j.neuron.2016.07.036. [Epub ahead of print]
2. Livet J, Weissman TA, Kang H, Draft RW, Lu J, Bennis RA, Sanes JR, Lichtman JW. Transgenic strategies for combinatorial expression of fluorescent proteins in the nervous system. Nature. 2007 Nov 1;450(7166):56-62.
3. Zador AM, Dubnau J, Oyibo HK, Zhan H, Cao G, Peikon ID. Sequencing the connectome. PLoS Biol. 2012;10(10):e1001411. doi: 10.1371/journal.pbio.1001411. Epub 2012 Oct 23.
4. Kebschull JM, Garcia da Silva P, Zador AM. A New Defective Helper RNA to Produce Recombinant Sindbis Virus that Infects Neurons but does not Propagate. Front Neuroanat. 2016 May 24;10:56. doi: 10.3389/fnana.2016.00056. eCollection 2016.
5. Removing distortions from high-throughput sequencing data: Kebschull JM, Zador AM. Sources of PCR-induced distortions in high-throughput sequencing data sets. Nucleic Acids Res. 2015 Dec 2;43(21):e143. doi: 10.1093/nar/gkv717. Epub 2015 Jul 17.