Killers in the kitchen: DNA helps track down foodborne disease

David Pacchioli
February 11, 2008

A rash of recent headlines hits close to home: Salmonella in peanut butter. E. coli in spinach and ground beef. In the U.S. alone, foodborne illnesses cause hundreds of thousands of hospitalizations and 5,000 deaths a year.

Most unsettling to consumers are the hidden outbreaks, by-product of our far-flung food distribution system. "There may be sick people scattered all over the United States, a case here, a case there," says Steve Knabel. "It can be very widespread, but no local health worker would know there's an outbreak occurring." To track these outbreaks to their sources, researchers rely increasingly on molecular epidemiology.

"You can think of it as DNA fingerprinting," says Knabel, professor of food science at Penn State. Knabel's specialty is Listeria monocytogenes, a common bacterium that, when found in ready-to-eat foods such as hotdogs, lunch meats, and soft cheeses, as well as in milk and raw meats, can cause listeriosis, a rare but deadly illness that sickens 2,500 people a year, killing 500.

The Centers for Disease Control's current "gold standard" for typing strains of pathogens like Listeria is pulsed-field gel electrophoresis, or PFGE, he explains. The microorganism's DNA is sliced into fragments with specialized enzymes, then placed in a gel, where pulses of electric current separate the DNA molecules by weight. The result is a characteristic banding pattern, "like a bar code in the grocery store."

"This technology has allowed us to see what were previously hidden outbreaks-ones that were ongoing but nobody had detected them using conventional methods," Knabel says. "But it's still not ideal." PFGE requires highly skilled technicians, for one thing. And results can be hard to standardize and interpret.

Fortunately, advances in DNA sequencing now allow a more accurate approach. "Because of our ability to sequence entire genomes, we can do sequence-based subtyping," Knabel explains. By comparing specific genes present in two or more strains of a given pathogen, researchers can detect differences at the level of single nucleotides, or bases: the As, Ts, Gs, and Cs that are DNA's fundamental building blocks.

Seven years ago, Knabel's Ph.D. student Wei Zhang set about finding a sequence-based method for typing Listeria. The method then favored, says Knabel, relied on comparing housekeeping genes, present in virtually all cells. Because these genes are essential to a cell's survival, they don't change much from one generation-or one organism-to the next. This relative stability makes the changes that do occur easier to detect and compare.

Housekeeping genes "work great for getting a big-picture view of how an organism has evolved," Knabel allows. But for tracking down disease outbreaks, he argues, you need even more discriminatory power. "You need to get down to the tips of the evolutionary tree." Instead of housekeeping genes, Zhang chose to focus on virulence genes-whose presence or absence determines whether or not a particular strain of bacteria can actually cause illness. "In Listeria monocytogenes there are about 15 known virulence genes," Knabel says, "and probably many more that we don't know about yet." Zhang chose six of these genes and used them to compare a set of 28 diverse L. monocytogenes strains Knabel had on hand in the lab.

The results were outstanding. "He was able to separate all 28 isolates by differences in their virulence genes," Knabel remembers, "where PFGE could only separate about three-quarters of them."

For the next test, another of Knabel's doctoral students, Yi Chen, requested outbreak strains of Listeria monocytogenes from the Centers for Disease Control (CDC). "By pure chance," Knabel says, the strains the agency sent were from two well-publicized recent outbreaks: one that killed 21 people in 1998 and was traced to hot dogs; and another, from 2002, in which listeria-infected poultry products killed 7 people. "We took these two outbreaks and threw both methods at them," Knabel says.

Listeria monocytogenes
Courtesy Penn State College of Agricultural Sciences

Listeria monocytogenes

As expected, the PFGE patterns showed slight differences, identifying the two separate outbreak strains. The sequence-based method, however, showed something that hadn't been seen before. "We had something like 2,400 bases from the six virulence genes for each isolate," recalls Knabel, "and they were absolutely identical. I remember staring at the computer when Yi first pulled them up. We looked at each other and thought, 'What the heck is going on here?'" Their method had proved something the CDC hadn't suspected: that the 1998 and 2002 outbreaks were two strains from the same epidemic. Applying this method to additional outbreak strains they were able to very accurately differentiate all four known listeriosis epidemics.

Still needing a way to tell the strains apart, Knabel and Chen had what he calls "another stroke of luck." They stumbled on something called a prophage sequence, a stretch of viral DNA patched into the bacterial chromosome. "At some point in its evolution a virus inserted its DNA into the Listeria chromosome," he explains. "And it turns out this sequence was perfect for what we needed-it's variable enough to separate outbreak strains, but also stable, because it's replicating along with the Listeria genome. By combining virulence gene and prophage sequencing we now have a complete sequence-based strategy, with results superior to PFGE."

"This has worked so well with Listeria," he adds, "that I am now working with.Ed Dudley here in food science and with others at Penn State to see if we can adapt it to track other pathogens," including E. coli 0157:H7, the culprit in a number of high profile food-poisoning outbreaks, and methicillin-resistant Staphylococcus aureus or MRSA.

He hopes that eventually this sequence-based strategy will be even more widely applicable. "Something like 300 bacterial genomes have been completely sequenced now," Knabel says, "and of those around 100 are pathogens."

"There's so much sequence information out there, it's nice to be able to actually use it to improve public health."

Stephen J. Knabel, Ph.D., is professor of food science in the College of Agricultural Sciences; Results reported above were published in the Journal of Clinical Microbiology and Applied and Environmental Microbiology.

Last Updated February 11, 2008