October 12, 2000
Researchers from the National Institute of Diabetes and Digestive and Kidney Diseases have determined the structure of a bacterial protein vital to repairing DNA. The findings, which appear in the October 12 issue of Nature, could help scientists studying a comparable, but faulty, human protein associated with a hereditary colorectal cancer.
In their paper, Wei Yang, Ph.D., and Changill Ban, Ph.D., of the Laboratory of Molecular Biology and Peggy Hsieh, Ph.D., and Galina Obmolova, Ph.D., of the Genetics and Biochemistry Branch describe the structures of the protein MutS and MutS combined with DNA that were isolated from the eubacterium Thermus aquaticus. MutS is one of several proteins that work together to correct mistakes that arise when the microbe's DNA is copied. Such repair proteins, with different names, exist in all living things.
Four bases make up DNA; when DNA replicates, mismatches sometimes occur. Normally, guanine (G) pairs with cytosine (C), and adenine (A) matches with thymine (T) along the helix. At the beginning of replication, the two DNA template strands separate and daughter strands are made to complement each template strand. For instance, if a string of bases on the parent strand reads GGATTC, the corresponding stretch on the daughter strand should read CCTAAG. If the wrong nucleotide slips in on the daughter strand, mismatch repair begins.
Scientists had long known that MutS's function in bacteria was to recognize mismatches and unpaired bases between the template and daughter strand. They found that MutS worked with another protein, MutL, to activate MutH, which then snips the daughter strand up to a thousand base pairs away from the error. MutH's cut allows a fourth protein called exonuclease to come in to take out bases, including the errors, much like a computer's backspace key.
But until recently, scientists couldn't demonstrate exactly how the repair proteins worked at the molecular level because they had no crystal structures of them. With a crystallized molecule, scientists get a three-dimensional view of the curves, twists, and indentations on a protein that indicates how and where it binds to another protein or DNA.
Yang's lab had previously determined crystal structures of MutH and MutL in Escherichia coli, a well-studied bacterium that is a good model for studying principles governing repair. MutS was more difficult to crystallize: It took nearly 5 years just to work out one structure because MutS's five domains, or sections, are quite mobile.
Shaped like a comma, MutS's multiple sections do different jobs. At domain V, two MutS copies must combine before binding to DNA can occur. Domains I and IV of both MutS subunits then bind to DNA. The junction of domains II, III, and V appears to be an area where MutS could interact with and signal MutL when an error was found.
The findings by the Yang group and a lab at the Netherlands Cancer Institute working on MutS from E. coli (also reported in Nature) demonstrate how the structure of the protein helps scientists understand the mechanism in mismatch repair.
Understanding these details of interaction in bacterial repair proteins may provide insights into their human counterparts, which have not yet been crystallized. People who inherit faulty genes for mismatch repair proteins are susceptible to cancers in various organs, including ovary, stomach, small intestine, and kidney. Errors in the gene that produces MSH2, a human version of MutS, predispose people to hereditary nonpolyposis colorectal cancer (HNPCC). This cancer accounts for 3 to 5 percent of all colorectal cancers, but people with mutations in their DNA mismatch repair genes have an estimated 80 percent lifetime risk of developing colon or rectal cancer.
"This [research] is a wonderful example of how basic science and clinical care can help each other," said Dr. Richard D. Kolodner of the Ludwig Institute for Cancer Research in La Jolla, California, in his Nature commentary on the two studies.
Yang and her colleagues plan to use what they've learned from microbial repair proteins to create models of human proteins and solve their structures.
Citation: Nature, 2000;407(Oct.12):703-710
###
CONTACT:
Anna Maria Gillis
(301) 402-7875
