NEW YORK (GenomeWeb) – A team from the University of California at San Diego, the Scripps Research Institute, and Illumina has demonstrated the feasibility of a new pipeline for doing single-nuclei sequencing on nuclei from post-mortem brain samples.
Using this approach, the researchers sequenced individual nuclei from thousands of neurons in post-mortem brain samples from a deceased 51-year-old woman, and used these transcripts to characterize and classify neurons in six Brodmann areas — parts of the human cortex that are unique to humans compared with other primates. The findings, published online today in Science, revealed more than a dozen neuronal subtypes within the targeted area.
"That turns out to be a pretty daunting mathematical challenge," co-senior author Kun Zhang, a bioengineering researcher at the University of California at San Diego, told GenomeWeb. "Most groups classify cells in to cell types. In our work, all of the samples are neurons and we're classifying them into subtypes, so the signal-to-noise ratio is a lot lower than a typical single-cell dataset."
Zhang and his colleagues noted that single-neuron transcriptomic studies on human samples have mainly relied on the availability of fresh brain samples collected during neurosurgeries, which may limit the brain sites that can be profiled.
A team from the J. Craig Venter Institute, the Salk Institute for Biological Studies, and elsewhere did transcriptome sequencing on nuclei from individual neurons in freshly dissected brain tissue for a study published in Proceedings of the National Academy of Sciences in late 2013. In PNAS last year, researchers from Stanford University, led by Fluidigm co-founder and consultant Stephen Quake, described their own approach for sequencing transcripts in individual neurons isolated after surgical procedures with the help of a Fluidigm C1 chip.
In an effort to come up with methods for doing single-nucleus sequencing on neurons from post-mortem brain samples, Zhang and his colleagues made several changes to nucleus isolation and sequencing protocols, including tweaking Fluidigm C1 chip-based sample processing, with technical assistance from the company.
They also had to refine to protocol to try to increase the sensitivity of the sequencing approach, Zhang said — the nucleus contains roughly one-tenth the messenger RNA molecules found in the complete cell, and that is made up of a mixture of molecules at different stages of mRNA processing and maturity.
After isolating neuronal nuclei from postmortem brain samples — which may sit at room temperature for as long as a day — the team isolates nuclei, sorts them by flow cytometry with a neuronal nuclear antigen, and captures the cells for Illumina transcript sequencing with the Fluidigm chip.
"There's a certain degree of RNA degradation, so that means these molecules might be partially degraded — they might not have the poly-A tail, which is required for the standard Fluidigm protocol," Zhang said. "Because of technical modifications and optimizations, we were able to produce a large dataset with pretty consistent data quality and low batch effects." He further explained that the focus remained on nuclei rather than whole neurons out of necessity, noting that the long projections and interconnections of neurons in adult human brains make it tricky or impossible to isolate individual cells without damaging them.
The team applied its approach to 4,488 nuclei in the six regions in the Brodmann areas of the post-mortem brain sample from the deceased individual. After weeding out pairs of neuronal nuclei that had initially appeared as individual nuclei and doing other quality control steps, the researchers were left with 3,227 single-nucleus transcriptome datasets.
Using expression and neuronal gene marker gene patterns from the nuclei, they iteratively classified the neurons into 16 subtypes comprised of 972 inhibitory neurons — mainly interneurons — and 2,253 excitatory neurons.
Though they could not rule out the possibility transcript proportions were slightly skewed in the nuclei due to mRNA breakdown in the postmortem sample, Zhang noted that they could pick out small sets of genes that were enriched in the subtypes. And with those marker genes they could recover some of the patterns from the brain slice where the neurons originated, overlaying the new neuronal subtypes onto anatomical information such as sites where neurons originated or brain layer.
The representation and proportions of various neuronal subtypes varied depending on the brain area, Zhang said, noting that the neuronal subtype composition was quite distinct in the visual cortex, for example. And across the other areas considered, the researchers saw brain region distinctions that largely stemmed from differences in representation by one neuronal subtype.
The team is now applying its single-nuclei RNA sequencing approach to other postmortem brain samples, to look at the same brain areas in other individuals and to explore regions beyond the cortex.
"There are so many frozen brain samples sitting at different brain banks," Zhang said. "But if your method only works on fresh brain samples, you're pretty much limited to the handful of samples that you can get from neurosurgery."