Courtesy of Mary Ann Moran
An outbreak of Salmonellosis erupted in Ohio, Michigan, Georgia, and Alabama during the winter of 1981. Frustrated epidemiologists could find no common link, until they finally realized what all the victims had in common: marijuana.1 Samples of pot used by some patients in Michigan were tainted with Salmonella muenchen, which was phenotypically unremarkable, except for the presence of two low-molecular-weight plasmids. Genetic fingerprinting of these non-chromosomal DNAs tied the cases together, and even uncovered new incidents that investigators did not know were linked to the outbreak.
Twenty years later, this marriage of molecular biology and microbiology continues to evolve, as have the tools of the trade. In today's microbiology laboratory, loops and Bunsen burners share bench space with real-time PCR machines, automated DNA sequencers, and microarray readers. Every branch of the field has benefited. Vigilant microbiologists and epidemiologists are spotting disease outbreaks earlier, and microbial ecologists seeking novel microbes in soil and seawater have developed a deeper, richer appreciation for the diversity of life. The increased sensitivity of these technologies means scientists can even study organisms that were previously deemed unculturable. But some investigators caution that reliance on technology is a dangerous crutch, that there is no substitute for a sharp mind and thorough bench work.
Courtesy of Lawrence Livermore National Laboratories
BIOWEAPON SENSING At the Lawrence Livermore National Laboratory, Pat Fitch heads a team charged with developing biosensors and biodetection technology. The group has focused on real-time PCR-based approaches, some of which the LLNL licensed to Cepheid, a Sunnyvale, Calif.-based company that is developing biosensors for the US Postal Service. A recent success is HANAA, the "handheld nucleic acid analyzer." Designed to fit in the pants pocket of a specific special forces uniform, the brick-sized HANAA can identify an organism in less than 10 minutes, says Fitch, who won't comment on whether HANAA is currently deployed in the Persian Gulf.
HANAA employs sequence-specific TaqMan probes that light up in the presence of a particular agent. Finding appropriate bioweapons signatures, says Fitch, is a "non-trivial" exercise. Each sequence must be common to all strains that are potential threats--for example, all known isolates of Yersinia pestis--yet also must not exist anywhere else in nature. Often such probes will come from pathogen-specific genes, such as those encoding toxins.
Another LLNL system, called BASIS, monitored the 2002 Winter Olympic Games in Salt Lake City. As skaters spun and skiers slalomed, a filter sampled the air. Couriers then retrieved these filters periodically, brought them to a relocatable field laboratory, and tested for a range of biological threats, using Cepheid real-time thermal cyclers and probes developed at LLNL.
As biodetection platforms, both HANAA and BASIS require somebody to supply processed samples. But LLNL has also developed autonomous biosensors, one of which was field-tested in Albuquerque, NM, last fall. The device, says Fitch, ran unattended for four days, and could send reports hourly. "It just sits in the lobby of a building and just runs," he says. These instruments can run either TaqMan PCR or Luminex-based antigen-detection schemes.
Catherine Fenselau, professor of chemistry and biochemistry, University of Maryland, is part of a team developing an entirely different approach to biosensing, one that's based on mass spectrometry (MS). Each bacterial species expresses a unique galaxy of proteins, and in 1975 Fenselau demonstrated that she could use mass spectrometry on whole organisms for microbial identification. Because such spectra are complex and variable, her team has adopted selective solubilization strategies. So now team members like postdoc Bettina Warscheid treat Bacillus spores (their testbed) with acid to extract small acid-soluble spore proteins prior to analysis.
Courtesy of Carl Woese
The group's automated biosensor collects air samples on a reel-to-reel film akin to videotape, forwards the tape into a processing station for acid treatment, and then into the mass spectrometer. The method requires no sample clean up, and unlike PCR or antibody-based approaches, is entirely unbiased; that is, it makes no assumptions about what it will find. The approach "always gives some information, because everything has a mass," explains Fenselau. Thus, MS can identify new organisms and potential threats long before the organisms have been cultured for detailed study.
But the system suffers from some limitations. First, the spectra are of little value without genome sequence data for comparison; Fenselau's team is developing computational workarounds for this problem based on machine learning. Another problem: mass spec, which can detect as few as 3,000 spores, is still orders of magnitude less sensitive than PCR. Finally, the technique has trouble sifting through the spectra produced by spore mixtures, as it is difficult to distinguish which spectral peak arises from which organism. To combat this last problem, Fenselau and Warscheid modified their approach, supplementing whole protein spectra with individual peptide microsequencing by post-source decay, to unambiguously identify the protein species.
Klaus Nüsslein, a microbiologist at the University of Massachusetts, Amherst, is working on yet another approach. His biosensor is made by mixing the bacteria to be detected with monomers that then polymerize around the organism. When the bacteria are removed, what remains is a surface that is pockmarked with the organism's imprint. Nüsslein then couples the detection chip to a vibrating detection device, such that interaction with the target bacteria changes the vibration frequency, signaling the event. According to Nüsslein, the sensor has performed well in initial tests, and can detect cells at "environmental levels"--around 500 cells per ml.
Other molecular biodetection strategies include microarrays, containing hundreds or thousands of species- or strain-specific sequences; antibody-based sensing; and DNA sequencing. Though the US Departments of Defense and Homeland Security clearly are interested in all such technologies, biosensors and detectors have numerous civilian applications. They could be installed in pediatrician's offices, emergency rooms, and food-processing facilities, for instance. But first they will need to become easier to use and more reasonably priced; Fitch puts the price of LLNL's automated instrument at under $100,000 (US). "The ultimate goal," he says, "is, [to] make it like a smoke alarm, where you buy one that's less than $20, it costs you a 9-volt battery once a year, and unless something happens, you ignore it." Such capability, he predicts, is years away.
FOODBORNE PATHOGEN DETECTION At the US Centers for Disease Control and Prevention in Atlanta, Bala Swaminathan, chief, foodborne and diarrheal diseases laboratory section, uses a blend of molecular biology and computer science to keep up with emerging outbreaks. In 1993, an outbreak of Escherichia coli O157:H7 food poisoning hit the western United States. It took public health officials 39 days to identify the outbreak; by the time it was all over, nearly 800 people had fallen ill, and four had died. A unique chromosomal fingerprint in bacterial isolates recovered from infected individuals helped officials tie the cases to contaminated hamburger meat.
As a result, the CDC in 1996 established PulseNet, a nationwide network of public health laboratories that perform bacterial genomic fingerprinting on clinical isolates of Campylobacter, E. coli O157, Listeria, Salmonella, and Shigella. Genomic DNA is isolated, digested with specific restriction enzymes, and resolved using pulsed-field gel electrophoresis. The sizes of the resulting fragments are then recorded in the database along with epidemiological evidence.
The database receives about 25,000 patterns per year, says Swaminathan, helping the CDC detect outbreaks as they develop. When an O157 outbreak flared up in Colorado in 2002, Swaminathan says, the CDC spotted the cluster in 18 days, and only 25 individuals were infected. Now the CDC is working with officials in Canada, Europe, and Asia to implement PulseNet internationally.
But Swaminathan also concedes that PulseNet has drawbacks. First, it records only DNA fragment length, not sequence. And second, though the system is highly standardized, small fluctuations in experimental conditions can decrease each fingerprint's comparative value. So the CDC is developing new approaches. In multilocus sequence typing, a series of eight to 20 genes are individually sequenced to produce a more accurate fingerprint. Alternatively, organisms can be typed based on the location and number of tandem repeats in their genomes.
Sequencing technology will be in the hands of public health officials in about five years, predicts Swaminathan. But "the ultimate goal of our project," he says, "is to be able to detect, identify, and subtype bacteria without isolating the bacteria." This technological shift would increase sensitivity and speed, he says.
STUDYING UNCULTURABLE MICROBES Though some microbiologists would prefer not to culture the bacteria they investigate, others have no choice--they study "uncultivatable" organisms. The problem is immense: Perhaps only 0.1% of soil microbes can be grown in the lab. Before the application of molecular biological techniques, such microbes would have been completely beyond the grasp of researchers, who typically have to grow bacteria to study them. But today, these scientists are developing clever strategies to access this untapped biodiversity.
One popular approach is to examine highly conserved sequences in ecosystems such as soil or ocean water. Such genes, explains Mary Ann Moran, associate professor of marine sciences, University of Georgia, "have a clock speed that's just slow enough and just fast enough so that it keeps track with how the organisms are diverging into different species over time."
Any conserved gene will work; researchers have used genes for nitrogen fixation and DNA repair, for instance. But the most widely used marker encodes 16S rRNA. According to Nüsslein, this gene is especially popular because every organism carries it, and because it is so well conserved. Thousands of examples are archived in repositories such as GenBank and Michigan State University's Ribosomal Database Project. Scientists can therefore use PCR to amplify the bulk 16S rDNA population from a field sample using universal primers, compare the resulting sequences against known examples, apply some bioinformatic wizardry, and obtain not only counts of new and known organisms, but a phylogenetic tree as well. Such information, it is hoped, will fetch microbiologists the ultimate prize: the ability to culture otherwise unculturable organisms. The new species' position in a phylogeny, for instance, could provide clues to essential nutrients and culture conditions.
One drawback is the inherent bias introduced, says Nüsslein, because each approach assumes that all rRNA genes look like those already isolated. "We won't use a primer for a sequence we've never seen before," he says. And, adding new sequences to the database that are based on existing primers may actually harden that bias, he notes, by reinforcing an errant assumption. Some researchers, therefore, have taken a dramatically different approach to study microbial diversity in a less biased "discovery mode," called metagenomics or environmental genomics. As Moran puts it, metagenomics answers the question, "What's out there?"
In the metagenomics approach, bulk genomic DNA from an ecosystem is fragmented into large, 40-50-kb pieces and cloned into bacterial artificial chromosomes (BACs). The resulting library is then transformed into E. coli and screened for interesting genes, much like an expression library.
Edward Delong, of the Monterey Bay Aquarium Research Institute, demonstrated the power of metagenomics when he identified a novel bacterial photorhodopsin in seawater.2 Prior to his report, such molecules had been observed in Archaea and Eukarya, but never in Bacteria. Delong was able to demonstrate that the gene was bacterial because it was located on a fragment that also contained a previously unknown bacterial 16S rRNA gene. "That, I think, stands in most people's mind as the zenith of what this technology can do," says Jo Handelsman, professor of plant pathology, University of Wisconsin-Madison, whose research group coined the term metagenomics. Handelsman has had some metagenomics successes of her own. Her lab, for example, in collaboration with plant pathologist Robert Goodman and chemist Jon Clardy, recently isolated novel antibiotics from a soil metagenomic library.3
With so many new tools to work with, microbiologists seem poised to tackle most any problem. But Abigail Salyers, professor of microbiology, University of Illinois at Urbana-Champaign and former president of the American Society for Microbiology, cautions that gadgetry is no substitute for intuition and good microbiological know-how. "I think we're fooling ourselves if we think technology's going to be a fix for thinking."
Jeffrey M. Perkel can be contacted at email@example.com.
1. D.N. Taylor et al., "Salmonellosis associated with marijuana: A multistate outbreak traced by plasmid fingerprinting," N Engl J Med, 306:1249-53, 1982.
2. O. Beja et al, "Bacterial rhodopsin: Evidence for a new type of phototrophy in the sea," Science, 289:1902-6, 2000.
3. D.E. Gillespie et al., "Isolation of antibiotics turbomycin A and B from a metagenomic library of soil microbial DNA," Appl Environ Microbiol, 68:4301-6, 2002.