On the Trail of Salmonella

In early 2010, about 300 people in the U.S. became ill with S. Montevideo, with victims in almost every state. Early in the investigation, it was thought the outbreak could be traced to contaminated pistachios, which had been implicated in a 2008 Salmonella outbreak.

Of all foodborne pathogens, Salmonella is one of the most difficult to isolate because of its homogeneity. Strains like Salmonella enteritidis and Salmonella Montevideo are, genetically speaking, almost indistinguishable from one another using conventional tools of forensic microbiology.

In fact, if you were to test a contaminated food sample with pulsed-field gel electrophoresis (PFGE), the standardized DNA fingerprinting used by the Centers for Disease Control and Prevention’s (CDC) PulseNet program to detect foodborne disease case clusters and identify common sources of outbreaks, S. enteritidis and S. Montevideo would appear to be the same pathogen.

That hamstrings investigators trying to accurately define the scope of an outbreak involving Salmonella. But next-generation genomic sequencing now under investigation at the U.S. Food and Drug Administration (FDA) could break that bottleneck.

In early 2010, about 300 people in the U.S. became ill with S. Montevideo, with victims in almost every state. Early in the investigation, it was thought the outbreak could be traced to contaminated pistachios, which had been implicated in a 2008 Salmonella outbreak. PFGE analysis couldn’t detect a difference between the strains involved in the two outbreaks—ultimately, field investigation narrowed the source of contamination to a spiced meat rub used in salami.

But which company produced the tainted product? Only in retrospect, using next-generation sequencing, were investigators at the FDA able to pinpoint the precise source of contamination—a single processing plant.

“This is the first time this technology has been used in a foodborne outbreak investigation, with enough strains to truly define the scope of an outbreak,” said Steven Musser, PhD, director of the FDA’s Office of Regulatory Science (ORS). Previously, next-generation sequencing had been used to differentiate among a handful of strains of an organism: in a Canadian tuberculosis study and an outbreak of cholera in Haiti. But in this case, researchers were attempting to distinguish among 30-40 Salmonella strains.

Dr. Musser explained the limitations of PFGE this way: “It chops DNA into very large pieces, separated electrophoretically into a pattern of about eight to 10 bands—sometimes more, sometimes less. This is a very nice technology for saying yes, it looks like we have a cluster of cases, but you may have a pattern that consistently shows up that you cannot differentiate further. It’s a high-level snapshot.”

But if PFGE is Google Earth at satellite-level view, next-generation sequencing is Google Earth at street level.

With a longer-lasting outbreak—for example, Salmonella in peanut butter or eggs—new genomic sequencing could help define the outbreak.

“We sequence all the DNA of a clinical strain,” Dr. Musser said. “We have every single base, and we know exactly how they define that particular organism. Even though 99% of the genetic information is very similar, this technology allows us to find the very small differences that define whether something has a high or low probability of being in the cluster. So with whole genome sequencing, we could easily say that the pistachios weren’t involved in the new outbreak.”

Next-generation sequencing can also include or exclude a particular patient from involvement in a specific outbreak. “That can be of great benefit to the epidemiologist, to help them determine whether they’re looking at multiple sources of contamination or one large, single outbreak,” Dr. Musser said.

Despite its advantages, whole genome sequencing will be limited primarily to retrospective investigations for the foreseeable future. That’s because of two major limitations: time and money.

The current sequencing platforms used by FDA investigators—they have two—allow them to sequence eight genomes a week. A typical outbreak investigation would require the sequencing of 40-50 genomes in all. “That takes a couple of months, which is far too long,” said Dr. Musser. “And it costs $800 to $1,000 per sequence we run, which would be about $40,000 in total. That’s too much for many clinical labs.”

But times are changing. “It looks like that technology will soon become available for about $100 or less per strain, and it will take about half a day to do the analysis,” he said. “At that point, it will become very practical for people to do in clinical labs, food testing labs, and regulatory labs. Within five years, I can envision laboratories routinely using this tool.”

But it may take longer than that to integrate such an approach into real-time outbreak investigation at the federal level, Dr. Musser cautioned. “I may be able to show how this works in my lab, or several others, but integrating it into a national system like PulseNet requires a lot of coordination and validation. These labs have been doing things the same way for a number of years, and it’s going to take a long time to change. But ultimately, I think this is one of the technologies that will form the basis of future surveillance activities.”

And it’s likely that Salmonella outbreaks may be among the first for which next-generation sequencing is used in real time. “If you have something like E. coli 0157:H7 in lettuce or produce, those tend to be rapid. They quickly come and go. We wouldn’t be able to do a meaningful analysis in one of those cases,” Dr. Musser said. “But if you have something like the Salmonella in peanut butter or eggs that went on for months and months, then we could use this technology right now to help define the outbreak.”

Whole genome sequencing may also become the technology of choice for compliance. Recently, Dr. Musser’s team worked with the FDA’s Office of Compliance to trace a sample taken from the environment of a plant production floor to a specific food sample in the plant. “This allows the Office of Compliance to go beyond a PFGE link and establish a more defined relationship between an environmental isolate and a processed food—in this case, a Salmonella strain in animal treats at a plant in the Midwest,” said Eric Brown, PhD, acting director of the ORS division of microbiology.

That’s a very valuable application, said Michael Doyle, PhD, Regents Professor of food microbiology and director of the Center for Food Safety at the University of Georgia in Griffin. “Many processing plants have resident Salmonella and Listeria strains. If you do the sequencing and isolate strains, then, should a plant have an outbreak or just do some routine retail testing and find the strain that matches a particular fingerprint, you can trace it back to the specific source of the contaminant.”

But the technology may not be quite as powerful a fingerprint as the FDA thinks, Dr. Doyle cautioned. “Bacteria mutate. More information has to be gathered relative to how powerful the assay is with the same strain after many transfers, once it’s been in the field a few years,” he said. “It’s too early to know if there are going to be a lot of mutations that would foul the system, or if there are other issues that would lead to misidentification of the source. They’ve had good results so far, but time will tell.”

It won’t be a magic bullet, agreed John Besser, PhD, deputy chief of the enteric diseases laboratory branch at the CDC. “It’s not like DNA fingerprinting in humans, which could be used to establish guilt or innocence of a crime. While the methods are similar—or even identical—the problem is somewhat different. As pathogens multiply in the environment and are passed from person to person, and from person to animal, animal to food, food to person, and back again, they will change in ways we don’t fully understand. In some cases, the DNA sequence match will be very significant, but in others, it won’t.”

Even when it cannot be used in real time, whole genome sequencing provides very important support for the field epidemiologist. “When there’s a very common allele and they can’t tell everything apart, that means the investigation is fully supported only by the epidemiology,” said Marc Allard, PhD, research area coordinator for the ORS’s genomics program. “This approach gives the epidemiologists independent confirmation of their findings.”

And that is a critical step forward, said Dr. Besser. “Ultimately, next-generation sequencing will strengthen the signal that there is something wrong in the food supply, or in the midst of an outbreak investigation, it will give us new tools to work with.”

Gina Shaw is a food safety writer based in Montclair, N.J.

Pinpointing Salmonella’s Virulence

Structure of the needle complex of Salmonella, embedded in a cellular context (artist's interpretation based on original data).

One of the reasons Salmonella is such a virulent pathogen is the crafty infection system it shares with the bacteria that cause dreaded diseases like plague, typhoid, and cholera: a needle complex. These hollow, needle-shaped structures found on the surface of the organism inject their host cell—that is, the human body—with bacterial toxins called effector proteins, which reprogram the cell to allow the infectious bacteria to swamp its defenses. (Imagine an invading army shooting bullets that would persuade the defending soldiers to lay down their arms and stand aside for the enemy.)

Understanding the function of Salmonella’s needle complex could prove crucial to more effective prevention of infectious outbreaks. “The secretion system is essential for virulence—consequently, reducing or inhibiting its function might be a good strategy for therapeutic intervention,” said biochemist and biophysicist Thomas Marlovits, PhD, a group leader at the Vienna Institutes IMP (Research Institute of Molecular Pathology)-IMBA (Institute of Molecular Biotechnology) in Austria.

Dr. Marlovits and his team have recently developed the most detailed three-dimensional images ever created of Salmonella’s needle complex, using an advanced technique called cryo-electron microscopy (cryoEM). “CryoEM allows the direct visualization of large molecules (such as the needle complex) in a near native-like environment,” he explained. “The sample preparation involves rapid freezing in a buffer of choice, which renders a fully hydrated, near native-like state of the needle complex particles, and is in sharp contrast to conventional preparation methods that usually include staining procedures of the molecules. We are then able to observe the sample material under this fully hydrated state, thanks to the special capabilities of the microscope, such as keeping the sample at -180 degrees Celsius and using low-dose conditions, which means that very few electrons interact with the molecule of interest. All together, these procedures keep the material intact in its finest details.”

Dr. Thomas Marlovits and his team have recently developed the most detailed three-dimensional images ever created of Salmonella’s needle complex by using cryo-election microscopy, or cryoEM.

Conventional electron microscopy, on the other hand, would use staining, which can easily destroy the fine details; high-vacuum conditions, which present the sample in conditions far from its native state; and a high dose of electrons, which would destroy the sample in fractions of a second without recovering high-resolution information.

The need to use low-dose conditions, however, presents scientists with a new visualization challenge: low contrast. It’s almost like trying to make out a person’s face in very dim light. Dr. Marlovits and his team have used specialized software to overcome this problem. “It allows us to use statistical methods to improve the signal-to-noise ratio in the images by ‘finding’ all the particles that are observed in a specific orientation relative to the electron beam,” he said. “Once found, we put all these particles in one basket and can generate an ‘average’ image, which now has a good signal-to-noise ratio.”

These images have provided, for the first time, the blueprint of Salmonella’s needle complex, including how its proteins are organized—in three dimensions. “From that, we could predict what we might call interaction sites of neighboring molecules,” Dr. Marlovits said. “Based on such a prediction, we introduced mutations at an interaction site and could show that modification of this site prevents the needle complex from being assembled—and, consequently, the infection cannot occur.”

The needle complex is a highly dynamic yet stable structure that undergoes a great deal of rearrangement during assembly. In continuing research, Dr. Marlovits aims to decipher that process, which involves the assembly of more than 200 protein molecules in a highly regulated and coordinated system. “Once we better understand this process, one could think about interfering with the substrate selection in order to make the infection system nonfunctional, or to harness its secretion power to deliver special, yet-to-be-designed biomolecules as therapeutic molecules into specific cells.”

—By Gina Shaw