New gene-editing system enables large-scale studies of gene function.
Despite significant progress in understanding the genetic changes in many different cancers, the diagnosis and classification of tumor type remains, at best, an imperfect art. This could change quickly, thanks to the findings of a group of researchers from the Broad Institute of MIT and Harvard, the Dana-Farber Cancer Institute, MIT and St. Jude's Children's Research Hospital in Memphis, Tenn.
The work, to be reported in the June 9 issue of Nature, focuses on microRNAs (miRNAs), small, noncoding RNA molecules that control the levels of proteins made from transcribed genes.
The scientists found a surprisingly accurate correlation between the 217 known human miRNAs and the development and differentiation of tumors. They did so by developing a new technology that could also be the basis for an easy and inexpensive diagnostic test.
"This study opened our eyes to how much more there is to learn about genomic approaches to cancer," said Todd Golub, senior author of the paper. "That microRNA profiles have such potential diagnostic utility was a big surprise to us, and one we're keen to validate in future studies." Golub is a core faculty member and director of the Cancer Program at the Broad Institute, an investigator at the Dana-Farber Cancer Institute, and a Howard Hughes Medical Institute investigator at Harvard Medical School.
MiRNAs were first identified in the worm C. elegans and were shown to control the development and differentiation of cells. When they were absent, certain cells went into abnormal rounds of cell division rather than differentiating. In light of the function of miRNAs in C. elegans, the recent discovery of at least 200 miRNAs in the human genome suggested that their expression pattern in humans could correlate with cancer, even it does not directly cause it.
But first, it was necessary to develop an accurate, fast, reproducible and inexpensive way to determine the expression pattern of all known human miRNAs. Given the small size of miRNA (approximately 21 nucleotides, or building blocks) as well as their close resemblance to each other, previous attempts to determine the pattern have been unsuccessful. In the current work, the scientists developed an ingenious method of capturing miRNA on tiny beads, with each individual bead marked with fluorescent "tags" that could tell which miRNA was bound as well as its abundance in the sample.
Testing a host of tumor samples on the miRNA-specific beads revealed that the expression patterns of miRNA not only correlated with the developmental origins of the tumor samples (e.g., epithelial cell, hematopoietic cell, etc.), but also subdivided into specific tumor types based on known genetic alterations. The team also found that miRNA levels are generally lower across tumor types than in the corresponding normal tissue, again supporting the idea that miRNA is critical to reaching and maintaining the differentiated state.
Finally, the researchers tested their discoveries against a panel of tumor samples of histologically uncertain cellular origin (but which had been determined by the anatomical location). Again, the miRNA classification provided amazing accuracy.
Although this is a preliminary study, its validation could have significant impact on the clinical diagnosis of cancer.
In addition to Golub, authors of the paper include Gad Getz, Justin Lamb and David Peck from the Broad Institute; Benjamin L. Ebert, Jun Lu and Raymond H. Mak from the Broad Institute and Dana-Farber Cancer Institute; Alejandro Sweet-Cordero from the Broad Institute and the MIT Center for Cancer Research; Howard Hughes Medical Institute Investigator H. Robert Horvitz; Eric A. Miska and Ezequiel Alvarez-Saavedra from MIT's Department of Biology; HHMI Investigator Tyler Jacks from MIT's Center for Cancer Research and MIT's Department of Biology; Adolfo A. Ferrando from the Dana Farber Cancer Institute; and James R. Downing from St. Jude Children's Research Hospital.
Jun Lu, Gad Getz and Eric Miska are listed as co-first authors on the paper.
This work was supported by the Howard Hughes Medical Institute.