Scientists have mapped, charted, modeled and visualized the human brain in many different ways. They have marked the boundaries of the organ's four major lobes: the frontal, parietal, temporal and occipital lobes. They have divvied up the cortex into more than 50 Brodmann areas—small regions characterized by particular cell types and specific cognitive functions, such as processing speech and recognizing faces. Researchers have tagged individual neurons with fluorescent proteins, transforming gray tissue into stunning brainbows, and followed water molecules as they move through the nervous system to trace ribbons of neural tissue linking one brain region to another. More recently, some scientists have championed the importance of connectomes—detailed wiring diagrams of all the connections between neurons in a given nervous system or brain. Thoroughly understanding the brain, proponents of connectomics argue, requires precise maps of its neural circuits.
The standard way of making a connectome is serial electron microscopy—chopping up an animal's brain into thin sheets, taking photos of all the resident neurons through an electron microscope and using those photos to painstakingly reconstruct the connections between neurons. In the 1970s biologist Sydney Brenner and his colleagues began using this technique to map the 302 neurons and 7,000 neural connections, or synapses, in the nervous system of a tiny worm known as Caenorhabditis elegans. It took them more than 12 years to finish the map. So far, C. elegans is the only animal with such a thorough connectome. Since mammalian brains contain millions or billions of neurons and billions or trillions of synapses, depending on the species, researchers are searching for faster and cheaper ways to create connectomes. At Harvard University, for example, Jeff Lichtman and his colleagues have constructed an Automatic Tape-Collecting Lathe Ultramicrotome (ATLUM)—a machine that speeds up the business of slicing up brain tissue into thin sheets with conveyor-belt efficiency.
In a new essay, Anthony Zador of Cold Spring Harbor Laboratory has outlined what could be the fastest and cheapest way to construct a connectome yet—if he and his teammates surmount some significant hurdles. Their strategy, which is still in the proof of principle stage in the lab, depends on DNA barcodes and a virus that has evolved to evade the immune system by sneaking through neural highways. Zador and his colleagues call their method BOINC (Barcoding Of Individual Neuronal Connections).
The basic idea behind BOINC is to take advantage of increasingly swift and inexpensive DNA sequencing technology to make connectomes. What if instead of reconstructing the connections between neurons with thousands of digital photographs, one could instead infer how neurons are connected by analyzing their DNA? Here's how it might work.
The first step in BOINC is assigning each and every neuron in a brain a unique DNA barcode—a distinctive sequence of DNA's nucleotide building blocks, A, T, C and G. To accomplish this, Zador proposes shuffling a specific segment of the neurons' genomes with enzymes called recombinases that specialize in such genetic scrambling. If the segment of DNA is long enough, and the researchers use enough recombinases, the chance that any two neurons would end up with the same reshuffled sequence would be quite low, Zador says. He calculates that the possible permutations of a 20-nucleotide barcode could uniquely label the approximately 70 to 100 million neurons in an entire mouse brain. Once the neurons have been labeled, a different team of enzymes would excise the barcodes from the neurons' genomes and package them into plasmids—circular loops of DNA.
Analyzing the DNA at this stage would reveal nothing about the connections between neurons because the free-floating DNA barcodes would still be confined to their respective cells. What is needed is a way for connected neurons to exchange copies of these DNA barcodes. Enter the pseudorabies virus (PRV), which is actually much more closely related to the herpes virus than the rabies virus. PRV mostly infects pigs—although many other mammals are susceptible—causing fever, sneezing, coughing, constipation and severe itching. The virus hides from the immune system by crawling through neurons' interconnected branches toward the brain, hopping across synapses—the tiny gaps that separate communicating neurons.
For several decades now, scientists have tracked PRV as it moves through a nervous system in order to trace the connections between neurons. Zador proposes something entirely new: coaxing PRV to endow DNA barcodes with the properties of a functional virus, so that they too can travel from one neuron to another.
When PRV enters a cell, it brings an entourage of proteins that get right to work hijacking the cell's molecular machinery and making many copies of the PRV virus. This whole process kicks off when the ensemble of viral proteins recognize and bind to specific sequences of DNA in the virus's own genome. By weaving these genetic sequences into the free-floating plasmid DNA barcodes, Zador and his team trick the PRV's helper proteins into giving these barcodes the protein sheath that allows PRV to hop from its host cell to a connected neuron. Essentially, the DNA barcodes become viruses in their own right. Importantly, Zador must ensure that these viral DNA barcodes have one-way, single-ride tickets—that they make one hop and then stop. He thinks he can terminate a barcode's journey after one hop by restricting access to an enzyme that helps it travel across synapses.
If all goes well, the neurons become "bags of barcodes" as Zador puts it—but not grab bags of barcodes from all over the brain. Rather, since the DNA barcodes only made one hop, each neuron should have copies of its own unique barcode as well as copies of barcodes from all the neurons to which it is immediately connected, but not from any other neurons. Within each neuron, yet another enzyme named phiC31 integrase would join the barcodes from connected neurons into pairs. Each neuron would then have as many barcode pairs as it has connections to other neurons. Finally, the researchers would grind up the brain tissue, extract the DNA and make many copies of all the paired DNA barcodes, which would allow them to infer the connections between neurons. If Alpha34X's unique barcode is paired with the barcodes of Omega16P, Gamma78V and Delta23W, then those cells must be immediately connected. Zador's essay appears online October 23 in PLoS Biology.
BOINC is a series of complex steps involving subtle genetic and molecular manipulation—a lot can go wrong along the way. Although Zador has by no means overcome the many technical challenges that BOINC poses, he has been encouraged by promising first-attempts and hopes to publish experimental results soon. "We have all the steps working by themselves and proof of principle where they are all working together fairly well," he says. So far, he and his colleagues have focused on cultures of mouse brain cells and have been able to deduce a neural circuit comprised of a couple hundred neurons from DNA analysis, although they do not yet know how closely it matches the anatomical circuit in the mouse brain. Zador estimates that it would cost about $48,000 and one week of work to eventually "sequence the connectome" of a whole mouse brain.
"Without a really good hypothesis about underlying neural circuitry, you can blow a couple years of research," Zador says. "With a connectome, we can generate hypotheses and ask, 'Does this even make sense?' We can look at the map and say, 'Oh, nope, couldn't be that way.' We constrain our hypotheses tremendously if we know what the wiring diagram is. I don't expect that understanding the connectome will give us all the answers to how the brain works, but what I expect is that it will completely change how we look for answers."