XIENCE NM of the week

Protein Mav Tie Obesitv to Diabetes J

The high incidence of obesity among people with type I1 diabetes suggests a connection between the two conditions. Scientists have sought a link by studying insulin resistance, the trademark s y m p tom of type 11, or adult-onset, diabetes. But they still don’t know why cells in people with insulin resistance ignore in- sulin’s signals to process blood glucose for use by muscles and other tissues.

Researchers working with mice have now identified a hormone, called resistin, that is secreted by fat cells and appears to play a direct role in type II diabetes. Healthy mice given doses of extra re- sistin for 2 days develop insulin resist- ance, researchers at the University of Pennsylvania in Philadelphia report in the Jan. 18 NATURE.

Interestingly, obese mice naturally produce copious resistin, says study coauthor Mitchell A. Lazar, a molecular endocrinologist at Pennsylvania. When given drugs that inhibit the effects of re- sistin, these overweight mice process glucose more efficiently, he says.

The Pennsylvania researchers have identified the human gene that encodes resistin, but they haven’t yet gauged the hormone’s effects in people.’

Roughly four out of five people with type II diabetes are obese. The new find- ings “indicate that resistin may form at least part of the missing link between

J obesity and diabetes,” says Jeffrey S. Fli- er of Beth Israel Deaconess Medical Cen- ter in Boston in the same issue of NATURE.

Lazar’s team found resistin in mice while monitoring the effects of a diabetes medication in the family of drugs called TZDs, or thiazolidinediones. Earlier stud- ies in rodents had shown that TZDs slow type I1 diabetes even though they spur the creation of fat cells, a seemingly con- tradictory action. The drugs work by acti- vating a receptor molecule, called PPAR- gamma, in fat cells.

When PPAR-gamma becomes active, production rates change for some pro- teins in the cells. Although most of these rates rise in the presence of TZDs, the r e searchers focused on the few protiens having rates that dropped. Among these, Lazar and his colleagues identified one that they dubbed resistin.

Further tests showed that TZDs indeed reduce the concentration of resistin in the blood of mice.

“This is a big deal,” says Allen M. Spiegel, director of the National Institute of Diabetes and Digestive and Kidney Dis- eases in Bethesda, Md. Measuring resistin concentrations in blood could help physi- cians diagnose type II diabetes, both Spiegel and Lazar say.

Resistin shares some qualities with an- other protein secreted by fat cells and as- sociated with obesity, the hormone l e p

Scheduled random walks skirt collisions Juggling competing demands in a

network of feverishly calculating com- puters drawing on the same memory resources is like trying to avert colli- sions among blindfolded, randomly zigzagging ice skaters.

Among networked computers, some sort of software scheduler must act as a referee to regulate data flow. Proving that a given scheduler not only prevents conflicts but also performs its duties ef- ficiently can be surprisingly difficult, however. Computer scientists have found that analyzing even simple data- sharing cases can be troublesome.

One important scheduling problem is equivalent to moving two tokens, each one at the start of a different infinite string of random digits. At each tick of a clock, the scheduler chooses which of the tokens to move one digit to the right, always making sure that the two tokens never end up on the same number at the same time. Can the scheduler keep the tokens going forever along their own strings without ever having both tokens simultaneously on, say, a 7.

36

Despite the problem’s apparent sim- plicity, no one knows the answer yet, says mathematician Peter M. Winkler of Bell Laboratories in Murray Hill, N.J. Nonetheless, recent advances suggest that the answer is probably yes, he re- marked at the Joint Mathematics Meet- ings held last week in New Orleans.

Instead of following tokens along infi- nite tracks, computer scientists find it convenient to study random walks on a two-dimensional array of points con- nected by lines. At each tick of the clock, a token moves randomly from one point to another connected point.

The behavior of random walks by a single token on such an array is well un- derstood, Winkler says. “Much less is known about the behavior of multiple, simultaneous random walks,” he notes.

The scheduling question takes the fol- lowing form: If two tokens take random walks on the same array and every step of each walk is known in advance, can the scheduler keep the tokens from ever colliding by choosing, on each tick of the clock, which token moves?

SCIENCE NEWS, VOL. 159

tin. This hormone, discovered in 1995, seems to regulate food intake.

Establishing that fat cells secrete re- sistin and leptin confirms that these cells are more than just “oily stuff in the body,” Spiegel says. Fat in the body “is an en- docrine gland, a hormone-producing sub- stance involved in a dialogue with the brain, liver, and muscle in a complex [process] of nutrient metabolism,” he says.

Leptin doesn’t appear to have a straightforward association with diabetes. In rodents, a leptin deficiency causes se- vere insulin resistance, but people with type I1 diabetes actually have high concen- trations of leptin in their blood. Some re- search points instead to a compound called tumor necrosis factor alpha as a trigger for insulin resistance, Flier says.

Resistin “is almost certainly a piece of the puzzle,” Lazar says. “Resistin may ac- tually play a big role in explaining why having too many fat cells can induce in- sulin resistance.”

The Pennsylvania researchers have al- ready devised an antibody to resistin, which they used in the mouse tests to in- hibit the newfound substance’s effects. However, they still haven’t found the m e lecular receptor that allows resistin to bind to cells. Identifying this molecule could give drug makers a target by which to chemically block the effects of resistin, Spiegel says. -N Seppa

Winkler has solved a variant of the problem that turns out to be easier to handle than the original. In the variant, the scheduler doesn’t know the paths in advance but instead has the option of immediately retracting an erroneous move by sending a token backwards along its walk.

For this fickle scheduler, Winkler can specify precisely which arrays of points and connections lend themselves to col- lision-free navigation by two randomly walking tokens.

Using different methods, Bela Bol- lobas of the University of Memphis in Tennessee and his collaborators have independently obtained similar results.

So far, researchers have failed in their efforts to use these techniques to prove that a clairvoyant scheduler who can’t re- tract a move can promise collision-free navigation on any array, Winkler says. R e cent theoretical work by Peter Gats of Boston University and computations by John Tromp of the National Research In- stitute for Mathematics and Computer Science in Amsterdam have provided clues that may yet point to a solution, he adds. 4. Peterson

JANUARY 20,2001