Finally, selected SAGs are more fully sequenced. JCVI barcodes and pools up to 90 single cell genomes samples per Illumina lane, says Karen Nelson, director of the JCVI's Rockville campus. At JGI, researchers pool 12 cells per lane in an Illumina HiSeq run, producing more than 4 gigabases of sequence per cell.
One recent study highlights the power of this approach. Ramunas Stepanauskas, director of the Bigelow Laboratory Single Cell Genomics Center, used primers to genes “mediating carbon fixation and chemolithotrophic energy production”— chiefly, RuBisCO—to identify microbes in the deep ocean capable of fixing carbon (2). These cells, Stepanauskas explains, may power the energy-intensive process not with sunlight —there is no sunlight at 800 meters—but by chemical reactions, such as oxidizing sulfur compounds.
That conclusion would have been tough to make using metagenomics, Stepanauskas says. RuBisCO genes and proteins have been seen in the deep ocean before, but were assumed to originate from dead algae sinking from the ocean surface. However, using single cell genomics, his team demonstrated unambiguously that these RuBisCO genes belong to microbes that are native to and ubiquitous in the dark ocean itself.
That finding, Stepanauskas says, defies conventional wisdom. The RuBisCO-containing cells represent at least 20% of the microbial population in the deep ocean, enough to significantly impact global carbon cycles. “Our discovery suggests that the carbon cycle is much more complicated.”Lysis and bias and contamination, oh my!
The 4 Gb of sequence JGI generates per cell is enough theoretically to read a typical bacterial genome 1,000 times over. Yet even that extensive coverage isn't enough to produce complete assemblies, which is not surprising since operating at the single cell level produces significant and vexing new challenges.
Cell lysis, for instance, is easy enough when working with samples in bulk, because researchers can use harsh chemicals like phenol to remove proteins and membranes and then subsequently eliminate those reagents. But that isn't possible with single cells, explains Woyke.
Lysis conditions need to be gentle enough to maintain DNA integrity, but harsh enough to be effective. It's a difficult balance to strike: At JGI lysis (using potassium hydroxide) typically only works for about 20% of the cells.
DNA yield is another issue. Most microbial cells contain only a single chromosomal copy and femtograms of DNA, far too little to be directly sequenced, so it must be amplified. The traditional whole-genome amplification option, PCR using random primers, introduces too much sequence bias and results in short fragments, says Lasken. But in 2002, Lasken, then at biotech firm Molecular Staging, and colleagues introduced a new method called multiple displacement amplification (MDA).
MDA is a random-primed, isothermal amplification process that employs the highly processive phi29 polymerase to produce fragments on the order of tens of kilobases in length. According to Woyke, the key to phi29 is that it doesn't stop upon hitting new double-stranded DNA, but rather keeps moving. Newly synthesized fragments thus become templates themselves, producing a hyperbranched structure sprouting multiple replication forks.
The process is highly efficient: From femtograms of starting material MDA can produce micrograms of long fragments, typically 20-40 kb in length. But MDA has issues, too. Many reagents, it turns out, contain minute quantities of contaminating DNA that can significantly impact the process, Woyke says; after all, it takes only the nucleic acid equivalent of a few bacterial cells to swamp one cell's worth of DNA. Woyke's team has devised several techniques, including UV irradiating MDA reagents and performing cellular sorts twice, to minimize contamination issues.
Blainey took another approach. Prior to joining the Broad Institute in early 2012, Blainey was a postdoc with Steven Quake at Stanford University. There his job was to develop methods to sequence the genomes of unculturable “dark matter” microbes “that were from the deepest, darkest regions of the phylogenetic tree.”
To do that, Blainey devised a microfluidics device that could separate up to 48 individual cells using optical tweezers, lyse them, and produce SAGs on-chip. Blainey developed his chip specifically to minimize contamination, he says. By reducing reaction volumes down to the nanoliter level, Blainey figured he could shift the balance of DNA to contaminants in favor of the former. “We've still got one cell, but we've got a lot less of that other junk that comes with the reagents,” he explains.