Howard Chang found that one long intervening noncoding RNA, called HOTAIR, is important in
breast cancer. Too much of it encourages a primary breast cancer to spread.

photograph by Ramin Rahimian

Skin Sense

An early question about aging skin led to answers on how cells code for their location.

Early in his research career, Howard Chang asked a simple question: what happens to skin cells as an organism ages? Using genetic technologies just emerging in 2000, Chang explored his question and revealed another complexity of skin—one that he's still puzzling over.

Chang's curiosity also led him into a new realm of study: long noncoding RNAs. In the past year, the HHMI early career scientist at Stanford University debuted a technique for determining RNA structures, and now he's making novel links between a specific type of RNA and human health.

Chang is a dermatologist with M.D. and Ph.D. degrees. During his clinical residency in dermatology, he worked in the Stanford research lab of HHMI investigator Patrick O. Brown, where he learned about microarrays. Chang saw the technology as an opportunity to determine the genetic differences between young and old skin cells.

There’s An App for RNA

See screenshots from Howard Chang’s iPhone app that lets the user navigate RNA structures.
Play Slideshow >>

He collected skin samples from banks of foreskin tissue taken from newborn boys. From adults, he gathered biopsies of arm, scalp, and back skin. It was widely assumed that skin was skin—identical all over the body. Chang's research revealed, on the contrary, that differences in gene activity among the tissues stemmed not only from age but also location.

"I realized these cells have functional differences based on where they are," says Chang. "It's like a built-in address code."

Chang started frequenting the morgue, collecting skin samples from different body parts of cadavers. Using microarrays, he discovered that the differences in gene activity among skin cells could be traced to Hox genes, a large family of genes already known to control positioning of body parts during development.

In 2004, Chang started his own lab. His team devised a method to look at expression of the 39 Hox genes—whether each is turned on or off—and also at expression of the DNA flanking each gene.

In humans, more than 200 of these surrounding regions were actively expressed. The researchers found this puzzling, since these regions of DNA are transcribed into RNA but never go on to produce proteins. They're called noncoding RNAs. His lab began to focus on one called HOTAIR. The researchers hypothesized that blocking cells from making HOTAIR would affect the neighboring Hox gene. But instead, the expression of a Hox gene on a completely different chromosome changed.

"This raised the idea that these RNAs can have very profound effects all over the genome," says Chang. His lab is beginning to understand these far-reaching effects.

Chang compares HOTAIR—which is a large intervening noncoding RNA, or lincRNA—to a train. It's a long, stringy molecule that carries various enzymes to Hox genes to regulate their expression. Like train cars, HOTAIR sections are modular, specific to their cargo. If Chang can figure out which RNA structures carry which enzymes, he may be able to predict the functions of lincRNAs other than HOTAIR.

But the challenge, he says, is the structures. The classic way to determine RNA structures, called reverse transcriptase PCR, is "tedious, a lot of work, and only successful on short pieces of RNA," he says. So his lab developed a way to separate lincRNAs into manageable pieces.

"If you have a chocolate bar with grooves along it, and you throw it on the ground, the pattern it breaks in will likely follow the grooves," says Chang. He uses chemicals to break the RNA at its thinnest and most exposed sections, like the grooves in the chocolate. Then his team sequences the fragments and pieces them together again. His lab used the technique, called parallel analysis of RNA structure, or PARS, to puzzle out structures of the full 3,000 RNAs in yeast cells. Their results appeared in Nature on September 2, 2010.

I realized these cells have functional differences based on where they are. It's like a built-in address code.

Howard Chang

In the spirit of data sharing, they've created an iPhone application to allow others to access the structures. Their own lab, of course, will use the data to find connections between the structures and functions of lincRNAs.

While this research sounds far removed from skin, Chang still practices dermatology, seeing patients once a week. It reminds him why HOTAIR and positional identity in skin cells are so important: they relate back to human disease.

"One of the biggest clues to a skin disease is where it occurs," he says. If a rash is on your hands and feet, that means something completely different than a rash that is only on your belly button." Understanding how cells code for their location using lincRNAs could explain why certain skin diseases affect only one area and maybe how to stop those diseases.

Recently, Chang's team found that HOTAIR is important in breast cancer. When breast cancer cells make too much HOTAIR, their address code—which should point to the breast—becomes jumbled, allowing the cancer to spread. The levels of HOTAIR in a primary breast tumor predict whether it will metastasize, the researchers concluded in an April 15, 2010, Nature article. Not something Chang expected to find when he asked a decade ago how skin cells age, he says, but no doubt a worthwhile finding.

Scientist Profile

Early Career Scientist
Stanford University
Genetics, Systems Biology