Leibel's laboratory at Rockefeller University is conducting research aimed at developing a better understanding of the relationships among food intake behavior, energy expenditure and the biochemistry of fat tissue in man and experimental animals. In the course of researching "Gaining on Fat" for the August 1996 issue of Scientific American, staff writer W. Wayt Gibbs spoke at length with Liebel. Below is a transcript of their conversation.

SA: Recent research by you and others seems to lend support to the theory that we are born with some genetically programmed body weight, the so-called set point. Could you summarize the evidence that this is so?

RL: The reason for believing that there is a set point to human body weight is based on some animal experiments, some epidemiology and some clinical medical observations. The animal observations were made in the 1940s and 1950s by an experimental physiologist who manipulated the brains of rodents and some higher mammals with electrolytic and knife lesions, showing that there are areas of the brain that have very powerful effects on appetite and weight regulation in animals. These are the famous experiments of Hetherington and Branson.

What they came up with is a somewhat simplified model in which there are two centers in the hypothalamus at the base of the brain: one for satiety called the ventromedial hypothalamic region (VMH) and the other a hunger region which is lateral of that, called the lateral hypothalamus. [They created] lesions with knives, electrolytic techniques or chemical techniques that actually ablated the neurons in these regions. They showed that when you knocked out the VMH, the animals became hyperphagic [i.e., ate uncontrollably] and gained weight. Then people realized after doing those experiments that what was really being regulated was not appetite so much as body weight. If you took a VMH-lesioned animal that was already at its maximum body weight and then fasted it so that it lost weight, it actually would eat its way back to its starting weight. So it was not as if you had permanently altered food intake, because the animal at this new weight plateau wasn't eating at the rate that it was when it gained. It was eating at a rate consistent with maintenance of its higher body weight. As they refined these experiments they could show that if you knock out part of the VMH, you get part of the way to maximum obesity-you see a dose-dependent response to the hypothalamic injury.

Once the animals reached whatever weight plateau they were going to achieve, they would then defend that in a metabolic and behavioral sense precisely the same way as an animal that had never had [a VMH injury]. So this is where the idea of set point originated, and there are a number of experimental physiologists who established the notion that what these centers are doing is not regulating appetite per se but rather trying to keep the animal at a certain level of body fatness. The food intake obviously had to occur in order to let the animal raise its body fatness-the animal cannot do photosynthesis or something. But that behavior is in service of a more primordial biological variable, which is body fatness.

That has interesting implications. It was pointed out at the time that appetite, which up until then had been regarded as a purely behavioral phenotype, is really a phenotype that is in service of some more proximal regulatory process: in this case, the regulation of body fat content. If you think about it, that is something of a sea change in the way that appetite and food intake is looked at. There is a fork there in the historical road. One group now pursues the physiological and metabolic aspects of food intake. And right at the same time the psychoanalytic school starts looking at this as somehow being a behavior in the older model that says behavior is isolated from the soma-a sort of Cartesian dichotomy that they want to set up.

I think Hetherington and Branson in one very important way pointed the biology of food intake and energy expenditure into the area of metabolism and physiology and away from the effort to separate the mind from the body. That is very important in terms of the [intellectual] history of this field.

With that as a backdrop, now you have a series of very interesting epidemiological observations that look at body weight in humans over long periods of time, as in the Framingham study (which was not done primarily to study body fat but rather to look at cardiovascular risk factors), and found that over 30 years-say between 25 and 55 years of age-actual weight gain on average is about 20 pounds. Some people have looked at that and said, 'gee, that's a lot of weight.' But if you figure that the average adult ingests 900,000 to 1 million calories a year, and you calculate what the energy cost of that additional 20 pounds is-in other words the enthalpy of the molecules stored in the body-you actually come up with a calculation that shows that the error, that is the positive balance, over that 30 year period is about tenth of a percent. [The actual figure is about 0.3 percent-Ed.] In other words, one-tenth of one percent of the calories ingested are in fact being stored. But 99.9 percent are being expended.

That degree of control or balance is extraordinary if you think about it that way. The real question is not 'how much weight do you gain?' but 'what is the denominator of energy intake over which that weight gain occurs?' So when you think about it that way, you come to the conclusion that there might very well be some sort of regulatory mechanism that, as indicated by those animal experiments, might also be regulating body weight or body fat content in humans.

Then there is a medical observation that under certain circumstances-either when neurosurgeons operate on certain parts of the brain or [brain] injuries that occur as a result of various sorts of trauma-you actually can find in humans homologs of these injuries that are produced experimentally in rodents. For example, you can damage the ventromedial hypothalamus in humans-inadvertently, of course-during surgery for certain types of tumors, and what you get is a hyperphagic, very obese human who will then maintain that higher body weight in pretty much the same way that an injured rodent will.

So that sort of observation, plus the epidemiology, again points to the idea that the physiologies may be quite similar [between humans and rodents]. The miraculous thing about human body weight is not that people gain 20 pounds over 30 years, but that that is all they gain given the environment we live in, which is promoting energy intake and not particularly promoting energy expenditure-that is, not requiring large amounts of physical activity.

Then there is the final set of clinical observations. The diet industry or weight control industry-either [drugs] or diet books, Weight Watchers, any of these commercial outfits-do an awful lot of repeat business. If, for example, the treatment of obesity were like an appendectomy, where you go to the doctor, you get your appendix taken out, and that's it, then what we would hear is that people went to such-and-such a place, got their obesity treated and it never came back again. That is definitely not the case. In fact, when you look at the numbers-and they're hard to get, because for a number of reasons, many of the commercial enterprises don't necessarily want to publicize this-the recidivism rate to obesity following what would be considered a successful weight reduction is probably over 95 percent. That can be interpreted a number of ways, but in the context of this biological picture of the process, what it tells you is that it is not so easy to perturb individuals from [a stable weight level] by the expedient of dietary management.

One of the things that is not well recognized is the fact that weight loss itself-that is, the actual ability to reduce body weight-is not a particularly difficult problem. If you put human beings on an 800 calorie or 1000 calorie diet, they lose weight. There do not appear to be a large number of people who have bizarre responses to a 1000 calorie diet-like they don't lose weight. These are more or less apocryphal tales that when tested in formal research paradigms don't hold up. In other words, there are not people who are totally resistant to weight loss. But what does characterize the vast majority of humans is that they are very resistant to the maintenance of body weight below whatever "normal" for them is.

Continuing with the history of this, in about 1986, based on those animal and human observations indicating that there is probably a set point and that it is probably in the brain, we looked at some of the other animal models of obesity that are not induced by brain injury or chemical injury but are instead due to mutations-that is, to genetics.

Described at the time were five mutations in mice and two in rats. All except for one were autosomal recessive mutations. Several of them lead, in these animals, to something that looks an awful lot like an animal whose hyptothalamus is not regulating body weight at a normal level: this is the ob/db dyad, so to speak, ob being (as is now well known) a mutation on chromosome six and db a mutation on chromosome four. The interesting thing about these animals is that their phenotypes are virtually identical on the same genetic background. So an ob mutation and a db mutation engrafted in a genetic sense on a black-6J background [strain] look exactly the same. This led Doug Coleman in the 1960s to propose that ob was encoding a signal to the brain. [That signal] probably was in the circulation, because of some parabiosis studies that he did where he surgically joined the animals. The db was the receptor for this signal. That notion-that ob is a secreted protein and that db is its receptor-fits perfectly with all of these other observations that suggested that an animal regulates its body weight based on some central set point mechanism in the brain and something that signals how much body fat there is in the body. The ob gene appeared to be mutant for the signal and the db gene appeared to be mutant for the receptor, because the animals look identical to each other. And [surgically joining the circulatory systems of] a db animal to a normal animal actually causes the normal to reduce its food intake and die of starvation.

The other thing you could do is to take a mouse or a rat and damage the hypothalamus, then [surgically join] the injured animal to a normal one, and the normal animal will die of starvation. That suggests that the mechanical or chemical injury has knocked out the receptor and that the ligand [i.e., the ob signal] is very high in the circulation, crossing the [surgical] union. Coleman's experiments all fit with this model: that adipose tissue is making a signal that is received in the brain and that ob might be the secreted protein and db its receptor.

That was the reason for choosing those two animals [the ob and db mutant strains] when Friedman and I started the project to clone these two mutations. They fit the paradigm that body weight is very likely to be regulated, that there is very likely to be a hypothalamic center encoding whatever the set point of an animal is, and that these mutations represent disruptions of this system in two of its critical components: the signal and the receptor.

SA: When did you start those experiments?

RL: Both of those projects were started in 1986. We took the approach of cloning the genes by virtue of their position in the genome, not by virtue of their physiologic product, even though they were picked based on physiology. At the time, that was an interesting approach.

We knew that these animals were very likely to hold some of the secrets of the signaling system that regulates body weight. In parallel with that, we began experiments to look at human responses to perturbations in body weight-but not in the way that this is done when you informally treat a patient for obesity, seeing if you can get them to lose weight and then look at them with some cursory metabolic studies.

We did the following experiment in a very controlled environment to test the hypothesis that an obese individual and a lean individual, as adults at their usual body weight, will behave metabolically in response to weight perturbation in precisely the same way. This is predicted by the idea of set point, by the genetic animal models and by the injury experiments. If we're right about the way the system works in the animals and we're right about a set point, then the argument is that an obese person will defend, in a metabolic sense, their weight in the same way as a lean person because the obese person's body is behaving as if they were at their "normal" body weight.

The experiment in humans involved testing the defense mechanisms at both ends of the spectrum. Normally obese people were forced up in weight by 10 percent by overfeeding and were reduced in body weight by 10, 20 and 30 percent decrements. These patients were in this hospital [Rockefeller University Hospital] during the whole study. The shortest period any patient ever spent here to have these studies done was something like 110 days. The longest period was slightly over two years-continuously hospitalized in this clinical research center. These experiments are not feasible in any other setting I'm aware of.

SA: How did you get lean patients to admit themselves for so long?

RL: The 110 days is three to four months. We have generally enrolled students who want both to participate in clinical research studies and to learn something about molecular biology or physiology. We admit them to the third floor of the hospital, and while they are participating in these studies they live there but work in this lab. One of our students has actually taken a year off from school to do this study. Most of his time, when he's not being studied, is spent doing lab research. He's working on mapping some of the mouse genes that produce obesity in humans.

SA: But all the food he eats, he eats here?

RL: Yes, as a matter of fact, all of the food he eats during the periods when we are studying his metabolism is a liquid formula diet that we make here. It is precisely compounded and its caloric content checked by bomb calorimetry at very regular intervals, because we need to know exactly how many calories it takes to maintain somebody's body weight over a period of something like 45 days.

At the end of each of these weight plateau periods, we do very intensive studies of the energy metabolism of the individuals. These studies include how many calories it takes to keep their body weight constant for a period of about 30 days. If you know how many calories a person is taking and you know that their weight is absolutely stable, that tells you over a 45-day period exactly how many calories their body used.

SA: Is their no variance in the amounts of nutrients that people excrete?

RL: That's an interesting question, and we have actually looked at this by collecting all of the stool production of all of these patients for periods of eight days while they are doing this. There is no significant variation in the amount of absorption of these calories, no matter what plateau you're at. When you hear the results of this study, you might assume that if a person gains 10 percent of their body weight, they are going to lose more calories in their stool. But that's just not so.

SA: So everyone has a digestive system of essentially equal efficiency?

RL: Yes. Obesity or the lack thereof is not the result of variation in the variation of food absorption. I can absolutely assure you of that, having done the experiments the hard way. The very hard way. That possibility is out.

At the end of these periods of weight stability, these studies that we put these patients through are really quite extraordinary. We use heavy isotopes of water. Here we give the patient two isotopes of water to drink. They are not radioactive; they just have different numbers of neutrons in the nuclei so that they can be separated by mass spectrometry. We give them deuterated water [also known as heavy water] and O18 water. So one is tagged on the hydrogen and one is tagged on the oxygen.

The interesting thing is that when you give somebody water like this, the deuterium comes out of the body which is determined by water turnover in the individual. The O18 is in equilibrium with carbon dioxide, so the O18 comes out by two mechanisms: first with normal water by transpiration, perspiration and urine, but also in the breath. The difference between those two decay curves (the O18 comes out faster), which we obtain by getting urine from these patients every day for 10 days-that gap is proportional to carbon dioxide production in that individual. By doing this, we can figure out how much carbon dioxide this person made over a period of 10 days. Knowing that, and knowing what the so-called diet quotient is-in other words, what the ratio of carbohydrates to fat in their diet is-you can back-calculate the amount of oxygen used to produce that amount of carbon dioxide. So by some simple algebra using the rate of carbon dioxide excretion, you can actually calculate how much oxygen their body used in the process of oxidative metabolism. That is a very critical number because it tells you how much energy they burned. Oxygen consumption can be immediately converted into calories.

So we measure caloric expenditure both by figuring out how many calories it takes to make their body weight absolutely stable, and checking that number by also using this double-doped water excretion technique using mass spectroscopy. It's quite expensive: the isotopes to do such a study cost about $500, not including the spectroscopy.

Then we take the individual and we measure their body composition-how much fat is in the body-by different techniques. We weigh them in air, then weigh them in water, using Archimedes' principle. We do a scan of the body with low-energy x-rays. And we also do it by isotope distribution, since when we administer the doped water, it gets distributed in the body's water space, not in the fat. So by looking at the partitioning of that water we can get another measure of body composition. So we very carefully document the amount of body fat in these people at the end of these periods of weight stability.

Then we put them through a series of metabolic studies: looking at how they metabolize glucose, how much insulin the pancreas produces, what thyroid hormone is doing, what the catecholamines are doing-in other words, how much epinephrine and norepinephrine they're producing-and how much dopamine they're producing.

We also use several techniques for studying the autonomic nervous system. We use a technique called spectral analysis, in which you deconvolute the heart rate and also by drug blockade techniques, where we give doses of atropine and esminol sufficient to totally lyse the activity of one limb of the autonomic nervous system. By then studying heart rate in these people we can actually tell whether either their sympathetic or parasympathetic nervous system has been cranked up or cranked down as a result of changing their body weight. The autonomic nervous system is of extraordinary interest because in the animal models, it has been shown that the genes for ob and db affect food intake and the autonomic nervous system.

Finally, these people are put through a series of measures of exercise physiology. We look at how skeletal muscle converts energy into work. The reason for this is that again in the animals there is evidence of reduced physical activity and reduced energy expenditure in physical activity in both ob and db mutants. We want to know whether we see in humans a related change in the ability of skeletal muscle to convert fatty acids and glucose to mechanical work. This is done by bicycle odometers and treadmills and also by putting one the of the large muscles into a nuclear magnetic resonance (NMR) device. The leg is put into a very strong, high Tesla magnet, to look at the NMR spectrum of high-energy phosphates in both phosphocreatine, ATP and inorganic phosphate. By tuning the receiver, you can actually watch the change in the amount of creatine phosphate, which a form of high-energy storage in muscle, as it is converted to inorganic phosphate. That indicates the amount of high-energy phosphate consumed in the process of a given amount of muscle activity. The muscle activity is measured by having them push a pedal at a fixed rate and certain resistance. So we can actually look at the efficient with which skeletal muscle converts high-energy phosphate to mechanical work.

You can see what we're trying to do: We're trying to break down energy expenditure into all of its constituent components. So in addition to all these other studies, we also measure the energy expenditure of the patient at rest. We put a hood over their head and measure the rate of oxygen consumption while they are resting. We measure the energy expenditure that occurs when they ingest a fixed number of calories-this is called the thermic effect of feeding. Ingestion requires a certain amount of energy just to take the molecules that are produce by [digestion] of food and either put it into storage or use it for immediate energy requirements.

After doing all of these studies, we can tell you for each individual how much energy they are using for resting, how much for the thermic effect of feeding, what the energy cost of physical activity is for that person, the efficiency at which the individual converts chemical energy to physical work. These are all critical to getting the energy balance just right.

When you do these studies you find that when you force an individual's weight up 10 percent, they require more energy to maintain that higher body than you would predict based on their requirements at usual body weight. We know from the body composition studies how much metabolic mass they have. If I know what your lean body mass is, I can predict what your energy expenditure will be [at a stable weight] with very high precision.

SA: The amount of fat tissue you have is not relevant to this?

RL: Not except that if you have a very large amount of fat tissue, it requires a little more energy to carry it around. It acts like a knapsack in that sense. But it does not use much energy at all-adipose tissue stores energy but doesn't use much. In that sense, it is a very good storage device. I mean, what you wouldn't want is to have a gas tank that also uses gas; you'd get very bad mileage that way.

You can account for virtually all the energy expenditure in a human by knowing their lean body mass and their physical activity. So we found that a gain in body weight causes the individual to spend more energy than you would predict based on their new lean body mass. In other words, they actually fall above the regression line that they were on at usual body weight. The amount of this increase is something like 15 percent above what you would predict.

You may say, "well, 15 percent, that's not very much." But 15 percent is a lot when you consider that one-tenth of one percent is the estimated degree of balance over a long period of time. What this means is that unless this individual is able to overeat by 15 percent, or to reduce physical activity by 15 percent, they are going to fall back to their starting body weight. In fact, in all instances in which we have performed this experiment, which must number well over 40 now, the individuals fall back to their starting weight. Obese people don't hang up, lean people don't hang up there; all of them come back.

The obese respond in precisely the same way as a lean individual does. In other words, if you force an obese person up by 10 percent, they have the same increase in energy requirements as a lean individual. Incidentally, if you look at lean and obese people when they are at their usual body weights, they are right on the same regression line. Their is no difference in the energy requirements of an obese person and a lean person.

One of the myths of obesity is that somehow obese people can maintain very large amounts of body fat without taking in any extra food. It's not true. Obese people actually have larger amounts of lean tissue than normal weight individuals do. About 30 percent of the increased weight of an obese person generally is in lean body mass.

At usual body weight, if you look at energy requirement per unit of lean body mass, lean and obese individuals are not different. That is a very important observation. It is one of the points that indicates that lean and obese people are really not different in terms of their underlying metabolic status when they are at usual body weight.

SA: How heterogeneous is the degree of the increase in energy expenditure with the increase in body mass? Are some increasing their energy expenditure by 15 percent while others are increasing by 2 percent?

RL: There is some degree of heterogeneity. It is not that great, although you do occasionally see people who have very little increase in energy expenditure when they gain weight. Why somebody who has this change in body weight doesn't increase is a very interesting question. It may indicate that at their prior weight, they may have been a bit below where they "should have been". You never know for sure whether people have been holding their body weight a little lower or maybe, due to environmental circumstance, a little higher than their set point "wants" them to be. So when we perturb their weight, we may be bringing them closer to where they "ought" to be.

One definition of somebody's set point is whatever body weight puts them right on the regression line [of lean body mass vs. energy expenditure]. There is some variation, but I don't know whether that is due to noise in the system and our inability to measure the phenotype precisely enough-that's probably some of it-or whether there may people who are not quite biologically where their set point wants them to be when the study begins.

SA: Is the relationship more or less linear: when you have people drop by 10 percent does their energy expenditure drop a certain amount, and then when they drop 20 percent in weight, it decreases by twice that amount, and so on?

RL: No. That is another very interesting point. When a person goes down 10 percent in body weight, lean or obese their reduction in energy expenditure is in the 15 percent range. If you take them down by 20 percent, it doesn't get any more. So it appears that whatever this defense mechanism is, if you want to look at it teleologically like that, it kicks in quite early: 10 percent is enough to bring it out. We don't know whether five percent is, because we've never tested that small an increase in weight. But [energy expenditure] doesn't reduce any further if you drop them 20 percent, 30 percent, or even as much as 50 percent.

SA: So it functions somewhat like a switch.

RL: Correct. It's as if, once this thing gets tripped you get full metabolic compensation.

How does this happen? A fraction is due to changes in resting energy expenditure. But the majority of the change occurs in the energy cost of physical activity. This is an area we're very interested in now. We're trying to figure out the mechanism by which a change in body weight not only would cause an alteration in resting energy expenditure but also in the energy cost of physical activity. Is something happening in muscle or in the autonomic nervous supply to skeletal muscle, which influences blood flow? We're looking now at muscle fiber types, at muscle enzyme content at all this autonomic nervous activity and the efficiency of muscle's conversion of chemical to physical activity by a variety of techniques.

Meanwhile, we're also looking at changes in the level of leptin. The model predicts that when weight is altered, what's going to change, at least to some extent, is body fat. We know that and can document that change quite readily. So now one of the critical questions is: can you see in the concentrations of leptin in the blood anything that might point you towards a mechanism for changes in resting energy expenditure or even the efficiency with this skeletal muscle works. One thing you might predict is that maybe leptin has some effect on aspects of resting energy expenditure and energy expenditure during physical activity.

Clearly it is a major signal-it may not be the only signal, but it clearly is a signal from fat to the brain. I think the evidence is quite compelling for that.

Now what we're really trying to do is to bring the human physiology down to the level of molecular mechanisms. We're doing that in two ways. We're looking at the effects of change in body weight on things like leptin metabolism. But also...well, let me back up a bit.

The obese person, when obese, doesn't show the metabolic fingerprint of what led them to get obese. When we study an obese person while they are obese, they really look metabolically just like a lean person, except they are bigger. One interesting way of looking at this is to ask: what is it that a formerly obese person has-or that a normal person has when they are down in weight-that is different? Well, we know now that their energy expenditure is lower than you would expect. And we know that they are very uncomfortable. People complain that they feel cold, they feel hungry-not normal. Invariably, what happens is that both obese and lean people bounce back up to their starting body weight. We know now what the mechanism is, at least from an energy perspective: they are spending 15 percent less energy than they need, they cannot comfortably reduce their food intake indefinitely by 15 percent, so what happens is that they eat at their normal level, it is 15 percent more than their body needs to maintain weight, and they bounce back up again.

If I show you a reduced obese person of body composition identical to yours, that person will be spending 15 percent less energy than you. Presumably, that is what got them up above you in body fat at some point. They actually normalize themselves, in a metabolic sense, by the process of weight gain.

People tend to think of obesity as a deforming or a deformed phenotype, but in an energy balance sense, that is their normal energy state-that weight puts them on the line that relates lean body mass to energy expenditure. They need more fat to get onto that line.

We're trying to figure out why that is. One thing you might predict is that an obese person may be deficient in leptin or a similar protein. Well we now know, based on some earlier studies, that obese people actually have plenty of leptin. So maybe it's not leptin; maybe there is some other signal coming out of adipose tissue. Maybe it is leptin but we still don't know enough about the physiology to relate it to this system of energy expenditure.

But what you can conclude is that obese people require a larger mass of body fat to normalize their energy state. Once they've got it, they are not distinguishable from you or me or any other normal weight individual.

So the real question is now how does this system work? Leptin and the receptor for leptin are obviously of great interest. A month ago we reported that the leptin receptor, which was cloned by a group in Boston [at Millennium Pharmaceuticals], is mutant in the db animal, as you would have expected. The question is: are