Decoding Bacterial Methylomes

Upon revisiting the data—and applying the new software—Schadt and colleagues discovered that along with the Shiga toxin, this particular E. coli had also adopted a methylase from the same phage. This methylation-laying enzyme resulted in a complete epigenetic makeover, the team learned. The group is still working on characterizing the effects of these epigenetic modifications, but Schadt said that the various pathways that were upregulated and downregulated in the bacterium, including changes in swarming and growth patterns, could have contributed to making it more virulent.

The technology is transforming the study of bacterial genome modification, said Richard Roberts, chief scientific officer at New England Biolabs, who started collaborating with PacBio to investigate bacterial epigenetics in 2010. In addition to simply mapping the epigenomes of hundreds of bacteria species, including emerging pathogens, Roberts is adding to his library of knowledge on how bacteria use methylation to protect their genomes from the restriction enzymes they release to cut up invading viral DNA. Other researchers are working on understanding the role of methylation in the cell cycle.

“It’s like you’ve been in a closed room for a long time, and you open the window and look out,” said Roberts. “And there’s a whole lot of stuff out there, and you don’t know where to look.”

The PacBio RS sequencerLAWRENCE BERKELEY NATIONAL LAB, ROY KALTSCHMIDT

Fortuitous discovery

PacBio was founded in 2004, and it commercially released its first sequencer, the PacBio RS, in April 2011. The PacBio RS II was released last month (April 2013). But since 2009, scientists at the company had suspected that its technology, in addition to sequencing nucleotides, could also detect DNA modifications.

The most commonly studied type of eukaryotic modification, methylation of the carbon-5 position of cytosine, is relatively easy to identify through bisulfite sequencing. Treating DNA with bisulfite replaces all nonmethylated cytosines with uracil, making it possible to detect methylation through sequencing the treated DNA. But bacteria are often methylated at the nitrogen-6 position of adenine and the nitrogen-4 position of cytosine as well, and there have been no good techniques for locating these modifications in the genome.

PacBio’s technology, SMRT sequencing, functions by detecting fluorescently labeled nucleotides in real time as DNA polymerase adds them to the DNA template strand. The original purpose of this alternative type of sequencing was to produce longer reads of DNA than traditional next-generation sequencing methods can. For instance, Illumina’s popular sequencer, which detects fluorescently labeled nucleotides that reversibly terminate DNA synthesis, can read 250 base pairs in a row, while SMRT sequencing produces reads of 3,000 to 5,000 base pairs on average, and can even generate reads as many as 20,000 base pairs in length. But PacBio scientists realized that SMRT sequencing had another advantage: DNA polymerase would synthesize DNA at slightly different speeds depending on whether the template strand was epigenetically modified or not.

Sure enough, the nucleotides emit pulses of fluorescent light as they are added to the DNA, and by calculating the lengths of the pulses and distances between them, it is possible to identify not only carbon-5-methylated cytosines, but also nitrogen-6-methylated adenines and nitrogen-4-methylated cytosines. PacBio got funding from the Human Genome Research Institute to develop software to translate these pulses into information about DNA modification, and in May 2010 they published a paper in Nature Methods introducing the method.

“There was no technique until PacBio came along,” said Bart Weimer, who studies foodborne pathogens at the University of California, Davis. Now, researchers can begin to use methylation patterns to differentiate closely related bacterial strains, he said, or shed light on variation among pathogens that gene sequencing has failed to explain.

“We now systematically include this type of information in our characterizations of pathogenic strains of bacteria,” Schadt said.

The epigenetics of pathogens

The 2010 paper immediately caught the attention of Roberts, a 1993 winner of the Nobel Prize in Physiology or Medicine for the discovery that genes can be discontinuous, divided by stretches of noncoding DNA called introns. He contacted PacBio asking how he could try out the technology, then began testing it on a few sets of methylation patterns he already knew well. He quickly found the technique to be highly accurate. In fall 2012 he and collaborators at PacBio published a summary of the methylomes of six bacterial species, from the pathogenic Campylobacter jejuni to the metal-reducing Geobacter metallireducens, discovering many new methylation motifs. He has collaborated with 10 different groups and is beginning to work with four more. So far, they have characterized methylation patterns for 150 species of bacteria, with 50 additional species currently being analyzed.

Roberts and others are also tantalized by the idea that the PacBio sequencer can detect DNA modifications besides methylation. According to current PacBio chief scientific officer Jonas Korlach, the kinetics of DNA polymerase change during sequencing as it passes over a diverse set of modifications, including one to DNA’s phosphate backbone called phosphorothioation in bacteria, formyl and carboxyl modifications in eukaryotes, and various signatures of DNA damage. However, software is not yet available to automatically identify and read off these modification types. Researchers have also noticed delays in DNA synthesis that may represent yet-undiscovered types of DNA modifications, Roberts said. “Maybe it will lead us into new ways of modifying DNA.”

But the most commonly used application of the new technology is in understanding and tracking disease. The US Department of Agriculture, for example, is investigating the epigenomes of bacteria involved in bovine respiratory disease, while researchers at the Allegheny-Singer Institute in Pennsylvania are trying to understand the epigenetics of an antibiotic-resistant Streptococcus pneumoniae strain called PMEN1 that that was first identified in the 1980s and has sickened people across the globe. Marc Allard of the US Food and Drug Administration (FDA) has published methylation details for 12 serovars of Salmonella enterica and found that their methylation patterns vary considerably, even between closely related strains. And Weimer and his colleagues are now incorporating methylation data into the 100K Foodborne Pathogens project, which he heads in partnership with the FDA and Agilent Technologies. They will look at methylomes of about 1 percent of the 100,000 genomes they plan to sequence, he said, and have already selected Listeria, Salmonella, Campylobacter, and a species of Vibrio involved food poisoning from shellfish to focus on first.

Weimer says he and others are in the very earliest stages of their research, but he is convinced the methylation data will yield copious new information on why closely related pathogens vary in virulence and pathogenicity. “This has never been done at this scale before,” said Weimer, who will participate this weekend (May 18) in a Microbial Epigenetics Workshop at the American Society for Microbiology annual meeting in Denver. “[This new technique] will enable huge amounts of work by lots of folks around the world to be done.”

Roberts agreed that the possibilities for future research are bountiful. “I feel like a kid in a candy store looking for interesting stuff to follow through on at the moment.”

Correction (May 15): This story has been updated from its original version to reflect that DNA polymerase synthesizes DNA at different speeds depending on epigenetic modification of the template strand, not the primer strand.