|Florida Anthrax Bacterium Sequenced|
|Sequencing two Ames isolates reveals four SNPs and new genomic markers|
Edward R. Winstead
May 9, 2002
The anthrax bacteria mailed in the US attacks last fall were so closely related to other previously known anthrax isolates that investigators could not tell them apart based on their DNA fingerprints. The question then became: Could a more comprehensive analysis of the isolates using genomic tools pick up the smallest DNA differences among the bacteria? The answer is yes, according to a comparative analysis of two anthrax isolates whose sequences are published today.
Scientists found four DNA differences between two isolates and dozens of new landmarks in the genome that can be used to classify and discriminate among anthrax isolates in future studies. The study's more important message, according to the researchers, is that genome sequencing and computation methods are a "powerful new tool" for analyzing microbial pathogens and investigating outbreaks of infectious disease.
The findings about the anthrax isolates, which are reported today online in Science, were made by scientists at The Institute for Genomic Research (TIGR) in Rockville, Maryland, and Northern Arizona University in Flagstaff. The results do not by themselves make clear who or what may have been the source of the anthrax attacks that killed five people and hospitalized others in the United States last fall. However, the data have been provided to the FBI, which is investigating all aspects of the attacks. The FBI placed no restrictions on the publication of the scientific data.
Timing was an important factor in undertaking the study. When the US Centers for Disease Control first identified a rare case of inhalation anthrax in Florida last October, scientists at TIGR had nearly finished a project to sequence Bacillus anthracisnamely the chromosome of an anthrax isolate from a laboratory in Porton Down, U.K. The isolate is known to be among the 'Ames' strain of anthrax, as was the isolate that killed a photo editor in Florida.
TIGR had been interested in sequencing a second strain of anthrax, but it seemed unlikely that an agency would fund such a project, according to Timothy D. Read, who heads the anthrax sequencing program at TIGR. "There were many organisms of interest waiting to be sequenced, and we thought that doing a second strain would seem repetitive," he recalls. The letter attacks last fall obviously generated new interest in anthrax, and Read quickly secured $250,000 in funding for the comparative sequencing study from the US National Science Foundation.
In the Science paper posted online today, a team led by Claire M. Fraser of TIGR, in collaboration with the laboratory of Paul Keim at Northern Arizona University, reports four single nucleotide polymorphisms, or SNPs, between the Porton and Florida chromosomes. These differences between the two isolates show the power of genomics to identify minute changes in the organisms' DNA.
"Genome sequencing is by far the most comprehensive way to catalogue differences among microbes," says Fraser, "and the sequence gives you the finest resolution possible, down to the single base pair."
The research team also discovered 60 new 'markers' in the Bacillus anthracis genomeDNA sequences that may vary from one isolate to another. These include SNPs, insertions or deletions of DNA, and short sequences that are repeated at different lengths in the genome known as VNTRs (variable-number tandem repeats).
The researchers tested the markers by analyzing nine isolates, which they grouped into six categories based on their genomic DNA. Five isolates were of the Ames strain and had been previously indistinguishable using more conventional DNA fingerprinting methods.
The Ames strain was first isolated from a dead cow in Texas in 1981. It was subsequently sent to the US Army Medical Research Institute (USAMRIID) in Fort Detrick, Maryland, and used in the US defensive biological weapons program. The strain was also sent to other laboratories in the U.S. and Europe, including the lab in Porton Down.
In launching his sequencing project at TIGR in 1999, Read selected the Porton isolate in part because it lacked plasmids and therefore did not require special biohazard containment. Plasmids are DNA structures outside the chromosome that contain genes for virulence.
After last year's anthrax attacks, preliminary DNA fingerprinting indicated that the anthrax bacteria in the Florida, New York, and Washington, D.C. incidents were derived from the Ames strain. Like the DNA fingerprinting technique used in forensic studies of human DNA, the microbial version analyzes variable DNA sequences in different parts of the genome.
Bacillus anthracis is one of the most homogeneous species known, and it was not clear last fall that sequencing two closely related isolates would reveal any differences. Some colleagues even told the TIGR researchers that the study was a waste of time and money.
"We knew the Florida and Porton isolates could not be distinguished by DNA fingerprinting, and we thought we might not find any differences at all by sequencing them," says Read. "In fact, we did not know what to expect, and that also seemed to be a reason to do the study."
Just how little variability there is between the Florida and Porton chromosomes surprised David A. Relman, a microbiologist at Stanford University, California, and one of the two authors of the commentary accompanying the study in Science.
"It's incredible that two life forms could be so similar and have different recent histories," says Relman, adding that it wasn't surprising that most of the variation was on the plasmids. "Plasmids get around and move around, and they have mechanisms for evolving and adapting."
"This study moves us to the next level of power and resolution in revealing the degree of variation that exists in nature between different individuals of the same species," says Relman. "The research provides an approach for distinguishing among individuals and understanding the mechanisms behind variation," as well as the pressures that drive certain kinds of mutations to appear in a genome or a population.
A more comprehensive survey of this microbe in nature is needed, according to many in the field. To make sense of mutations detected through comparative sequencing and how the pathogen can evolve, scientists need to have information about the rates at which mutations occur in a species in the laboratory and in nature.
"Without a better understanding of this microbial 'background,' we stand in danger of making false accusations and wrongful incriminations," Relman and Craig A. Cummings of Stanford write in their commentary.
As the next step in the research, the National Institute of Allergy and Infectious Diseases (NIAID) is providing funding to TIGR, in collaboration with Northern Arizona University, for the sequencing of at least fourteen additional isolates of Bacillus anthracis. These will include some Ames isolates and represent geographic diversity. The idea is to examine a set of more distantly related isolates to gain a fuller picture of genetic differences in the species.
Keim's laboratory has about 1,200 anthrax isolates. Two years ago, he led a group that developed a novel molecular 'typing' system for B. anthracis based on VNTRs, i.e., loci in the genome where a pair of letters repeats at different lengths. The system was based on sequencing eight VNTRs in the B. anthracis genome.
Keim and colleagues used the method to analyze 426 B. anthracis isolates and found that the entire group could be classified into 89 distinct genotypes. But they could not make further distinctions among the isolates using the VNTRs. When their study was reported in the Journal of Bacteriology in 2000, no whole-genome sequences for B. anthracis were availableonly partial sequences of the two plasmids.
"The whole-genome sequences generated by TIGR allow us to identify hyper-variable regions very quickly," says Keim. "Now, a bright undergraduate can accomplish more in an afternoon at the computer than two postdoctoral fellows could do in three years before we had the genome sequence."
The next step, continues Keim, is to do genomic comparisons of diverse strains: "This is going to allow for novel forensic tools in the future and really address the subtle biological differences that exist among B. anthracis strains."
"One of the real contributions of the Science paper to the field are the rigorous statistical analyses that accompany their claims," says Relman. "The numbers are impressive and the probability is that, yes, the polymorphisms they identify are almost certainly real and not due to sequencing errors."
The new study introduces a statistical model that distinguishes between true genetic polymorphisms and random sequencing errors. Steven L. Salzberg, of TIGR, led the bioinformatic analyses.
One unanswerable question at this point is why the researchers found seven differences between two DNA samples of the Porton isolate obtained three years apart.
In theory the Porton organism would not have developed mutations because it was in a freezer for most of the intervening time. But the Porton laboratory does not have records of how the sample was handled or the number of times it was taken out of the freezer and 'grown up.' Without these records, the TIGR researchers can only speculate about the source of the variants.
"If this case does ever get solved, it will not be solved by science alone," says Fraser. "More traditional investigative work will be involved."
"When we do fourteen strains it may be possible to say that an isolate came from a specific lab, and the people in the lab may be cast in a suspicious light," Fraser continues. "This organism has been in the lab for twenty years, and many people may have passed through the lab in that time."
Fraser and others have proposed that a database of pathogens be created, and the idea has some support. Such a database would have multiple uses because genomic data have applications in medicine as well as drug and vaccine development, and would be a resource for epidemiological and forensic studies.
A repository of genomic information would also help engage researchers who might not normally work in this area, says Carole A. Heilman, director of the Division of Microbiology and Infectious Diseases, at the NIAID.
A pathogen database has a practical research advantage in that many laboratories are not prepared to handle pathogens but may be able to contribute through data analysis and other ways. Anthony S. Fauci, director of NIAID, has invited research proposal on a range of areas related to bioterrorism and infectious agents.
"Some of these organisms are difficult to work with and laboratories need clearances that can be difficult to come by," says Heilman. "But there is much that can be learned by understanding the proteins and the functional pathways. So genomic information would be a resource that the general community could use to begin working on bioterrorism defenses in this way."
"I am more convinced than ever that a database of thirty important pathogens is doable and can provide a resource for tracking these agents," says Fraser.
. . .