A powerful new method called computational subtraction detects foreign bacteria and viruses in human tissue samples even if the organisms haven't previously been encountered.
The microbe-hunting method relies on DNA sequence data compiled in the nearly completed Human Genome Project over the past 10 years.
Dana-Farber Cancer Institute researchers already are preparing to use the technique to investigate the causes of mysterious chronic diseases such as rheumatoid arthritis, Type I diabetes, atherosclerosis, lupus, multiple sclerosis and several types of cancer.
Undetected, possibly novel infectious agents have been suggested as possible causes of these and other illnesses.
"The technique is good for investigating all these chronic diseases of unknown origin," said Matthew Meyerson, MD, PhD, senior author of a report that is published online by the journal Nature Genetics today. It will appear in print in the journal's February issue.
Another potential use of the method is in identifying emerging infectious diseases; examples are HIV and Ebola virus, which were unknown when they first showed up in recent decades.
Microbiologists have traditionally identified pathogens (disease-causing organisms) by growing them in a laboratory dish from a sample of infected tissue. But not all pathogens can be cultured this way. Molecular tools do exist and have been used to identify some new disease organisms, but they have major limitations, Meyerson said.
Meyerson, who trained as a pathologist, has a longstanding interest in diseases whose causes remain unknown or have been wrongly linked to other factors. For example, he noted that doctors used to blame stress and diet for stomach and duodenal ulcers, which only in recent years have been shown to be caused by a bacterial infection.
Infections may be involved in a long list of inflammatory and autoimmune diseases, he said, as well as cancers including lymphomas, bladder cancer and some lung cancers.
The technique is called computational subtraction because the actual matching of DNA sequences is carried out with computers after the sample has been analyzed.
"The potential power of sequence-based computation subtraction lies in its ability to identify new nonhuman sequences in a comprehensive and unbiased manner," the authors write in their report.
"It is made possible by the sequencing of the human genome," said Meyerson, because scientists can now call up on a computer nearly the entire "instruction book" of humans -- the sequence of some 3 billion chemical "letters" that make up all the 30,000 or so human genes.
Starting with a sample of human tissue from someone with a disease, the scientists would determine the sequence of a portion of the DNA its cells contain. Using a powerful computer program called MEGABLAST, scientists can then match the sample's DNA to the sequence of DNA in the entire human genome.
"So if you sequence the DNA and compare your sequences to the human genome, you eliminate anything that matches," says Meyerson. "What's left is the DNA of the infectious agent."
Meyerson and his colleagues tested 3.2 million "expressed sequence tags," or ESTs, segments of genes collected from many humans including those with diseases. They compared these EST sequences with the sequences of the human genome itself, which should contain exclusively human DNA.
They found that about 2 percent of the EST sequences didn't match to human sequences. Some of these sequences are non-human, and turned out to be contaminating or infecting viruses, bacteria and fungi.
Among this leftover DNA, Meyerson's team found DNA from such organisms as the hepatitis B and C viruses, human papillomavirus, cytomegalovirus, Kaposi's sarcoma herpesvirus and Epstein-Barr virus.
To test whether the subtraction method worked on a sample in which the infecting agent was known, the investigators tried it in a human tissue sample known to contain a specific type of human papillomavirus (which causes cervical cancer). The technique worked as expected, highlighting the papillomavirus in the "leftover" DNA.
At times, the method likely will pinpoint an unfamiliar nonhuman DNA sequence in diseased human tissue, said Meyerson. It may be a novel microbe, and investigators will have to use other tools, such as making antibodies to the protein the DNA sequence makes, to identify it.
The other authors are Griffin Weber and Jay Shendure, MD-PhD students at Harvard Medical School; David Tanenbaum, PhD, of Dana-Farber, and George Church, PhD, of Harvard Medical School.
[Contact: Bill Schaller]