Lab Matters - Winter 2013

DNA laboratory staff, Newborn Screening Program, Wadsworth Center, N YS Department of Health. (Pictured left to right: April Parker, Jason Isabelle, Lea Krein, Allison Young, Lisa DiAntonio, Carlos Saavedra-Matiz, Matthew Nichols.)

Last year, William Wolfgang, PhD, at Wadsworth’s Bacteriology Laboratory, initiated a retrospective study—partnering with the Connecticut Public Health Laboratory, FDA and Cornell University’s Department of Food Science—to demonstrate how WGS might have contributed to a 2010 Salmonella Enteritidis outbreak investigation, had it been available in real time. The outbreak was centered in a Connecticut long-term care facility, and epidemiological data showed a strong link to cannoli purchased at a New York bakery.

Salmonella Enteritidis has always been problematic for outbreak investigators; it is a clonal organism with little genetic variability. “Most has exactly the same PFGE pattern,” said Wolfgang. “But PFGE only samples dozens of sites [on the genome], nowhere near the five million nucleotides we probe in WGS.” WGS of contemporary Salmonella Enteritidis isolates showed that a number of community residents had also been infected with the outbreak organism. Wolfgang said, “If epidemiologists had been aware of this at the time, they could have questioned more people outside of the long-term care facility. Probably the bakery wasn’t the ultimate source; it was probably eggs, which are commonly associated with Salmonella Enteritidis.” The value of WGS for foodborne disease surveillance has not gone unnoticed by CDC. Efrain Ribot, PhD, team lead of PulseNet USA, said the agency has long recognized the problem with PFGE for clonal bacterial strains. One early attempt to solve this problem— before the advent of NGS—was the CDC-funded development of multiple-locus variable number tandem repeat analysis (MLVA). But MLVA has its own problems, being highly species- and serotype-specific, so testing protocols cannot be generalized.

Ribot said, “In five years, I envision that NGS platforms will replace MLVA and are likely to replace PFGE too.” The trick will be to develop protocols that generate actionable data at least as fast as PFGE does, especially during outbreaks.

“With WGS,” said Ribot, “we haven’t gotten the timing yet, because data analysis is the bottleneck. Then we will be faced with the issue of establishing the proper infrastructure. How do you store the data? . . . . We’ll need data warehousing and pipelines that are going to allow us to do the surveillance in real time. We also have to try our best to ensure that the NGS model we set up for PulseNet is one that can be executed by personnel typically found in PHLs, which may have little or no experience in bioinformatics. It’s all doable, it’s going to take time and resources.”

‘exciting, but scary too’

In the meantime, those PHLs with next-gen platforms are continuing to explore the utility of the technology.

The Arizona Public Health Laboratory is one of several labs funded by CDC’s Public Health Emergency Preparedness Grant to use pyrosequencing to monitor the emergence of antiviral drug resistance in influenza viruses.

One Wadsworth project focuses on adenoviruses, prompted by a number of outbreaks in New York state.

Kirsten St. George, PhD, chief of Wadsworth’s viral diseases unit, said epidemiologic data indicated that not all of the outbreaks were connected. Yet, she said, “Available molecular techniques could not find differences among [the outbreak organisms], and sequencing individual genes did not necessarily indicate significant differences among them either.” So scientists used the Ion Torrent to sequence entire adenovirus genomes.

“We were just amazed at the data it generated,” said St. George. “There were a number of differences sprinkled throughout. It really did take sequencing across the whole genome to get a collective set of mutations.” St. George said WGS will be especially useful for studying intertypic recombinance—for example, recombination of a Type 3 and Type 7 adenovirus— that muddies the overall surveillance picture. “One method may say Type 3 and one may say Type 7. What’s moving around the country? What’s actually causing disease?” The use of WGS for newborn screening holds both promise and peril. With three billion base pairs per human genome, and potentially millions of variants between individuals, the data analysis is mind-boggling, not to mention the ethical considerations. Yet WGS has already been used, under research conditions, to help diagnose children in the neonatal intensive care unit suspected of having genetic diseases—relying heavily on each child’s symptoms to focus the data analysis.