Johns Hopkins Magazine -- June 1997
Johns Hopkins Magazine

JUNE 1997

S C I E N C E    &    T E C H N O L O G Y

New promise for muscle-damaging diseases... liquid jets that make light... an alarming rate of coral decline... clue-giving curves... hardwired for language

Mighty mice
Through genetic engineering, Johns Hopkins scientists have crafted what might be called a biotech version of Mighty Mouse.

"The mice are fairly spectacular," says Se-Jin Lee, assistant professor of molecular biology and genetics. "They are normal in every respect, except their muscles are two to three times larger than normal." "They look like Schwarzenegger mice."

The extraordinary mice promise to be new tools for studying muscle-damaging diseases such as muscular dystrophy. Agricultural researchers may also apply the technique to create Goliaths in other species. "If you could take cows, chickens, pigs, and do the analogous manipulation, you would have two to three times more meat," says Lee. Picture bulging drumsticks and mammoth pork chops.

The project began when Lee, working with graduate student Alexandra McPherron, discovered a gene called myostatin that is expressed only in skeletal muscle, the tissue used for voluntary movements (in contrast, smooth muscle tissue, such as that found in the heart, controls involuntary movements). To elucidate myostatin's function, they "knocked out" the gene in embryonic mouse cells. Ann Lawler, assistant professor of gynecology and obstetrics, then inserted the cells into a host mouse embryo, and implanted the embryo into a female mouse. The miniature Schwarzeneggers resulted.

In the May 1 Nature, the scientists conclude that the protein coded for by the myostatin gene (which also goes by the name myostatin) dampens the growth of skeletal muscle. Thus, deleting the myostatin gene allows muscles to grow larger.

Lee's team is now crossbreeding the new mice with mice that have an inherited form of muscular dystrophy, a disease that causes muscle fibers to degenerate. The researchers will then examine offspring that inherit both muscular dystrophy and the myostatin deletion. Lee hopes that the lack of myostatin in these offspring will confer greater muscle growth that will compensate for muscular dystrophy.

Information garnered through such experiments may also lead to therapies for people with cachexia, the muscle-wasting that occurs in cancer, AIDS, and other diseases. "How to use these mice in clinical settings is one of our major goals," notes Lee. Such clinical research will probably target the myostatin protein rather than the gene, says Lee, which is much easier to do.

Of course, the finding raises thorny ethical issues. Might steroid-popping athletes switch to myostatin inhibitors? Such possibilities clearly make Lee uncomfortable. "Everybody from the FDA to physicians," he says, "will have to work together to regulate the use of myostatin." --Melissa Hendricks

What's behind a bubble's glow
Imagine a tiny bubble in a flask of liquid. Now, if the flask is placed in a field of high-pitched sound waves, the bubble will shrink and expand, shrink and expand, shrink and--glow!

Mysteriously, sound energy is transformed into light energy, a process that concentrates energy by a factor of more than a trillion.

Scientists first observed this phenomenon, called sonoluminescence, in 1934. But still, says Andrea Prosperetti, "nobody has a clear, watertight proof of what's going on."

Prosperetti, the Charles A. Miller Jr. Distinguished Professor of Mechanical Engineering, recently proposed that the sound waves force a tiny jet of liquid to push through the bubble at superfast speeds (perhaps 4,000 miles per hour), and slam into the other side. The impact is so abrupt that the fluid molecules do not have time to flow away. Instead, they shatter, and this fracturing--for some, as yet unknown reason--yields light.

"Fractoluminescence" can be seen in other sytems, notes Prosperetti. Try crunching Wint-O-Green lifesavers in a darkened room.

Scientists have proposed a dozen other theories to explain sonoluminescence. Some scientists have been excited by the possiblity that sonoluminescent bubbles might be laboratories of nuclear fusion, the clean and inexpensive energy source that has proved so elusive. According to this theory, sonoluminescence is caused by shock waves within the shrinking bubble that produce temperatures of more than 2 million degrees Fahrenheit, near the temperature required for fusion.

However, Prosperetti's theory, which he describes in the April Journal of the Acoustical Society of America, bursts that bubble, if you'll pardon the pun. His theory--yet to be confirmed--holds that sonoluminescent bubbles heat to only 10,000 degrees F--nowhere near the tremendous temperatures required for fusion. --MH

Toxicologist notes coral decline
A sphere of coral the size of a softball rests on biologist Gary Ostrander's desk. It is as blindingly white as a starlet's capped teeth--white the way coral should not be, says Ostrander, associate dean for research at the
Krieger School of Arts and Sciences.

White hard coral have lost symbiotic algae that normally inhabit their tissues and provide vital nutrients and coloration. Without algae, hard coral--the type that builds reefs--dies and looks bleached. Over the past two decades, coral bleaching has increased at an alarming rate, says Ostrander, who is part of a large unaffiliated group of scientists monitoring the health of coral reefs around the world. When coral is lost, so is its ecosystem: thousands of species of fish live near and feed off corals. Ostrander is now collecting some of the first data showing declines in fish populations associated with coral bleaching.

Bereft of life-giving algae, a section of coral reef bleaches out and dies (below), an increasing phenomenon.

Coral are animals related to jellyfish and sea anemones. Reef-building coral begin life as larvae, and mature to hydra-like polyps. The polyps deposit a hard calcium carbonate skeleton, to which they permanently attach. Coral continuously build onto the calcium carbonate skeleton, enabled by single-celled photosynthetic algae that live inside their cells and provide fats and carbohydrates. In return, the coral protects the algae (called zooxanthellae) from predators.

One symbiont cannot survive without the other. If the zooxanthellae leave, the coral dies, leaving only the bleached skeleton. Small amounts of bleached coral can regenerate. But off the coast of San Salvador, where Ostrander has spent the last several years monitoring coral reefs, another variety of algae takes over. Macroalgae grow over the surface of the coral, and prevent the regrowth of new polyps or the rehabitation of zooxanthellae, Ostrander has found.

Twice a year, Ostrander flies to the Bahamian Field Station on the island of San Salvador. From there he scuba dives to several undersea plots on coral reefs, which are marked by stainless steel pins. He and his assistants record the location and identity of everything they see: coral, algae, sponges, sand, rock. Even without analyzing the data, Ostrander says he has seen more and more bleached patches. In the past few years, bleaching has increased by about 5 percent around San Salvador, he estimates. "This may seem minuscule, but the problem is that 10 years ago there was none," he says.

On one of the reefs, Ostrander has also found 50 percent fewer fish and 30 percent fewer fish species.

San Salvador is far from the worst region for coral bleaching. Colleagues at 26 field stations around the world have documented similar declines. At the Galapagos Islands, for example, 90 percent of the coral is gone, says Ostrander.

In a worst case scenario, an entire coral reef dies, and leaves a void bereft of marine life, save for a few bottom-dwelling scavenger fish. Coral reefs are like oases in an ocean that otherwise is much like a desert, says Ostrander.

Coral bleaching is a warning sign that there are stresses on the undersea environment, says Ostrander. In the April 25 Journal of Toxicology and Environmental Health, Ostrander and doctoral student William Meehan call upon toxicologists and environmental scientists to help figure out exactly what those stresses are.

If scientists could point their fingers at a single environmental culprit--global warming, for instance--understanding coral bleaching would be that much simpler. "But the causes of coral bleaching are equivocal," says Ostrander. Some scientists have linked the bleaching to rising ocean temperatures. But others find evidence that cooler temperatures are a cause. Still others blame oil spills, herbicides, increasing salt concentration, or sedimentation.

One thing is almost certain, however, says Ostrander: Humans are accelerating the process. He believes that many different processes are at play, and weeding out one from the other in the huge mixing bowl of the ocean will be extremely difficult. --MH

Categorizing the curvaceous
Imagine you open a box of chocolates. Without cheating--don't taste or smell, how do you tell which is your favorite nut cluster and which the yucky mint cream?


Curvature, along with other visual cues such as color, help your brain to identify things, such as a bumpy nut chocolate. Could a robot make the same distinctions?

It turns out that recognizing objects based on curvature is a relatively complex task. But computer science doctoral student Elli Angelopoulou developed a computer vision technique that does just that. Her system can tell apart different types of coffee mugs, disposable razors, children's toys, and, yes, chocolates. Angelopoulou, who will report on recent results of her system this month in Puerto Rico at a meeting of the Institute of Electrical and Electronics Engineers (IEEE), says that the system is correct roughly 80 percent of the time. "We're working on improving that," she adds.

If she succeeds, the vision system could conceivably be incorporated into an industrial robot, such as one that sorts and packages of toys or chocolates, or perhaps one used to search for survivors following a nuclear explosion.

"Elli has done a very elegant analysis," says her advisor, professor of computer science Larry Wolff, who heads the Computer Vision Laboratory at the Whiting School of Engineering. While computer scientists have developed systems that recognize colors and shapes, they have found the design of systems that analyze curvature to be much more complex, says Wolff.

"There are two or three other systems that try to work on curved objects," says Angelopoulou, "but they're not as information-rich. They only check the curves on the borders of objects, not over the whole surface. Ours examines every part of the surface."

When it comes to chocolates, Elli Angelopoulou's object recognition system has discriminating tastes.
For her technique, Angelopoulou places three lights roughly equidistant around an object, such as a child's toy. She shines one of the lights on the toy, positions a video camera in front of the light, and films the object. Then she shuts off the light, and repeats the process with each of the other two lamps, creating three images of the toy taken from different vantage points. Each image contains roughly 10,000 pixels, the tiny points on a video screen in various shades of gray.

Angelopoulou's computer program next converts the triplet of images into a special value for each pixel, and compresses the 10,000 values into a 10-number "object signature."

Each object should have a unique signature, says Angelopoulos. A computer vision system would compile a large database of object signatures, to which it would refer to identify a novel object. "Theoretically, a robot could use this technique" to home in on an object, says Angelopoulos. She hopes that robotics experts will now continue the work.

Unfortunately for Angelopoulou, a confessed chocoholic, "Nobody from Godiva came to fund this study." --MH

Hardwired for language
Parents do not sit down with a copy of The Elements of Style to teach their infants that a subject is followed by a verb. But babies, remarkably, learn grammar relatively quickly.

Many linguists believe that the brain--perhaps from birth or even earlier--is hardwired for a universal language, says Hopkins professor of cognitive science Paul Smolensky. Despite diversity as vast as Greek and Swahili, Japanese and Gaelic, everyone may have the same hardwiring.

Now, in the March 14 Science, Smolensky and Alan Prince, a linguist at Rutgers University, propose an idea called "optimality theory," which, they say, can help elucidate hardwiring.

"The basic goal of optimality theory, and all linguistic theory, is to try to understand what the grammar (as opposed to the words) of languages have in common," Smolensky explains. "On the surface, it seems like anything goes. But if you delve deeper, you find a lot of commonalities."

Smolensky proposes that grammatical sentences abide by universal constraints. "When we hear speech and analyze it, the structures we impose to encode meaning and understand what we hear are subject to these constraints," he explains.

According to optimality theory, constraints are ranked by hierarchy. In a given language, some are stronger than others, and when two constraints cannot both be satisfied, the "optimal" structure will be the one that obeys the stronger constraint. An example might help.

In English, sentences follow the word order subject-verb-object. That is one constraint. But another constraint says that a question-word such as "what" comes first, even if it is an object ("What did John see?"). In this case, having a question-word come first outweighs having a subject-verb-object sentence order.

While all languages have the same constraints, the constraints have different emphases in different languages. In English, for example, the constraint requiring all sentences to have subjects takes precedence over a constraint requiring that all words in a sentence contribute to its meaning. Thus, we say, "It rains," even though "it" does not signify or modify a thing, action, or place.

In Italian, however, the equivalent sentence is "Piove" ("Rains"). The constraint requiring that all words have meaning outranks the constraint that all sentences have a subject.

These universal constraints may be represented by connections among neurons, says Smolensky, but exactly what those "neural networks" are biologically is for the neuroscientists to figure out. The universal constraints in our neural networks may be present at birth, or even earlier.

If optimality theory is confirmed, it will help settle a long-running debate among linguists on how language is learned.

One group of linguists takes an empirical approach, and believes that different languages have different grammatical rules, and that people learn the rules through experience.

Other linguists say that empiricism cannot explain how babies learn grammar so readily, given the relative dearth of grammatical information they receive, a concept called the poverty of the stimulus. Some of these linguists, most notably the distinguished Noam Chomsky, propose that something biological allows us to comprehend language. If infants indeed know the universal constraints, it lends credence to the view of the Chomskyites.

With Hopkins professor of psychology Peter Jusczyk, Smolensky is testing his theory in infants. In a first series of tests, 20 six-month-old babies listened to recordings of a female voice speaking nonsense sounds. Babies, they found, were more attentive when listening to sounds that abided by the universal grammar. They reported these findings in Baltimore last month, at an international meeting on optimality theory sponsored by Hopkins and the University of Maryland. --MH