obesity the integrated roles of environment and genetics

8/14/2019 Obesity the Integrated Roles of Environment and Genetics

1/16

WALTHAM International Science Symposium:Nature, Nurture, and the Case for Nutrition

Obesity: The Integrated Roles of Environment and Genetics1,2

John R. Speakman3

Aberdeen Centre for Energy Regulation and Obesity, Division of Energy Balance and Obesity, RowettResearch Institute, Aberdeen AB21 9SB, Scotland and School of Biological Sciences, University ofAberdeen, Aberdeen AB24 2TZ, Scotland

ABSTRACT Obesity represents one of the most serious global health issues with ;310 million people presentlyaffected. It develops because of a mismatch between energy intake and expenditure that results from behavior(feeding behavior and time spent active) and physiology (resting metabolism and expenditure when active). Both ofthese traits are affected by environmental and genetic factors. The dramatic increase in the numbers of obese peoplein Western societies reflects mostly changing environmental factors and is linked to reduced activity and perhapsalso increased food intake. However, in all societies and subpopulations, there are both obese and nonobese

subjects. These differences are primarily a consequence of genetic factors as is revealed by the high heritability forbody mass index. Most researchers agree that energy balance and, hence, body weight are regulated phenomena.There is some disagreement about exactly how this regulation occurs. However, a common model is the lipostaticregulation system, whereby our energy stores generate signals that are compared with targets encoded in the brain,and differences between these drive our food intake levels, activity patterns, and resting and active metabolisms.Considerable advances were made in the last decade in understanding the molecular basis of this lipostatic system.Some obese people have high body weight because they have broken lipostats, but these are a rare minority. Thissuggests that for the majority of obese people, the lipostat is set at an inappropriately high level. When combined withexposure to an environment where there is ready availability of food at low energy costs to obtain it, obesity develops.The evolutionary background to how such a system might have evolved involves the evolution of social behavior, theharnessing of fire, and the development of weapons that effectively freed humans from the risks of predation. Thelipostatic model not only explains why some people become obese whereas others do not, but also allows us tounderstand why energy-controlled diets do not work. Drug-based solutions to the obesity problem that work with thelipostat, rather than against it, are presently under development and will probably be in regular use within 510 y.

However, several lines of evidence including genetic mapping studies of quantitative trait loci associated with obesitysuggest that our present understanding of the regulatory system is still rudimentary. In particular, we know nothingabout how the target body weight in the brain is encoded. As our understanding in this field advances, new drugtargets are likely to emerge and allow us to treat this crippling disorder. J. Nutr. 134: 2090S2105S, 2004.

KEY WORDS: obesity genetics environment review

Background

It is an ironic and rather sad fact that the two majornutritional problems that presently face the world are that

;600 million people face severe energy deficits and starvationwhile at the same time, ;310 million people face a problem ofchronic energy surplus and obesity (1). In some cases, theseproblems exist alongside one another (2), but obesity is

predominantly a problem of Westernized societies. In the lastfew years it has become an increasing problem throughout mostof the world (e.g., 36). By 2000 the obesity problem hadalready grown to such an extent that the World HealthOrganization declared it was the greatest health threat facingthe West (7). Obese people carry around excessive amounts ofbody fat, which in the absence of more direct measures, isgenerally estimated by combining measures of height andweight. The most common method of combining thesemeasures is to divide body weight in kilograms (W) by theheight in meters (h) multiplied by itself (i.e., W/h2). This is

1

Presented as part of the WALTHAM International Science Symposium:Nature, Nurture, and the Case for Nutrition held in Bangkok, Thailand, October2831, 2003. This symposium and the publication of the symposium proceedingswere sponsored by the WALTHAM Centre for Pet Nutrition, a division of Mars, Inc.Symposium proceedings were published as a supplement to The Journal ofNutrition. Guest editors for this supplement were DAnn Finley, James G. Morris,and Quinton R. Rogers, University of California, Davis.

2 My work on obesity has been generously supported by the ScottishExecutive, the Biotechnology and Biological Sciences Research Council of theUK (Grant 1/S12830), and the Wellcome Trust. I am grateful to the organizers ofthe 2003 WALTHAM International Symposium for the invitation to present this workas a plenary address.

3 To whom correspondence should be addressed. E-mail: [email protected].

0022-3166/04 $8.00 2004 American Society for Nutritional Sciences.

2090S

byonOctober8,2008

jn.nutrition.org

Downloadedfrom
http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/http://jn.nutrition.org/


2/16

called the body mass index (BMI)4. On the BMI scale, peoplewith an index .25 but ,30 are said to be overweight, andpeople with an index.30 are defined as obese. Although thesedefinitions have been adopted by the WHO, there are severalwell-recognized problems with this index. In particular, it doesnot reflect body fatness changes very well when a person is alsochanging his or her height over time. Consequently, BMIcannot be used to reliably gauge the body fatness of children. In

addition, bodybuilders and some athletes, who have developedlarge amounts of muscle tissue, may also be misclassified asobese. Generally, however, in most adults, BMI correlatesreasonably closely to body fatness (which is measured by moresophisticated scanning and imaging devices), particularly ifcombined with measures such as the waist circumference (8).

Despite its limitations, the preferred use of BMI is justifiedon historical grounds. During the 1930s, U.S. life insurancecompanies analyzed the factors that influence the probabilitythat someone would redeem a life insurance policy (whichgenerally meant a person had died). Based on redemptions of;5,000,000 policies issued in the 1930s, the factor thecompanies decided was the most effective predictor of mortalitywas BMI. In fact, the life insurance companies discovered that

mortality was lowest for individuals with BMIs between 20 and25. Above this range, mortality increased, and it increaseddramatically when BMI was .30. Conversely, mortality alsoincreased when BMI fell to ,20. This pattern of increase athigh and low BMIs is at least partly influenced by theheterogeneous nature of the population involved, whichincluded both smokers and nonsmokers (9). Smokers domi-nated the low-BMI classes throughout the 1940s1960s, beforethe health effects of smoking were widely acknowledged. Whenonly nonsmokers were examined, the dramatic rise in mortalityat low BMI was much reduced although not entirely removed.The normal BMI range for minimal mortality in nonsmokers ispresently suggested to be from 18.5 to 24.9 (7).

Effects of obesity on mortality

Among both smokers and nonsmokers, mortality increasesas BMI increases .25. This is because there is a direct linkbetween body fatness and susceptibility to many degenerativediseases (6,10). The most important of these is type 2 diabetes.The risk of developing type 2 diabetes for someone with a BMIof 35 compared to someone with a BMI of 22 is 4090-fold andappears to be greater in obese females than males (11).Diabetes increases the risk of circulatory disorders andcardiovascular disease and is the most common cause ofblindness in people aged ,60 y. Increased body fatness is alsoan independent risk factor for hypertension and cardiovasculardisease. Some cancers (particularly breast cancer) are increased

in relation to body fatness, although whether this is a problemof impaired early detection using existing techniques in theobese patient is not clear. Obesity is also a risk factor for severalrespiratory disorders including sleep apnea (12). The obesepatient is more likely to be depressed and taking antidepressantmedications, more likely to suffer joint problems, and also morelikely to suffer from skin disorders.

Because of the medical complications associated withobesity, there is a statistically significant increased risk ofmortality once BMI increases .25. Because the original datawere derived for mostly white persons living in the U.S., therehas been recent discussion over the relevance of this cut-offpoint for different ethnic groups (13). In particular, in Asiansand Chinese, mortality has already started to increase at a BMIof 2324. If a person has a BMI .30, then on average, his or her

expected life span is reduced by 9 y compared to someone witha BMI of 2022 (14). There are ;30,000 premature deaths inthe UK annually that can be attributed to obesity (14). Forcomparison, this is about the same number of individuals whodie of lung cancer (;33,000 annually). Throughout Europe,the deaths attributable to obesity exceed 300,000 annually,which is ;1 in every 13 recorded deaths (15).

The health consequences of obesity combined with the factthat large numbers of people are afflicted by it has significanteconomic consequences. In the UK, these costs were quantifiedin 2000 by the National Audit Office (13) to be;500 millionannually in direct health care costs and a further 2 billionannually in wider costs to the economy. In the U.S., the totaleconomic costs of obesity in 19971998 were estimated at

;$92.6 to 99.2 billion annually when normalized to 2002dollars (1618). Similar data are available for most Westernsocieties that indicate the present costs of obesity generally runat ;58% of heath-care spending (1822).

The effects of obesity on health outcome appear to bereversible if the person in question loses weight. This isimportant, because it is not entirely clear yet what themechanisms are by which obesity predisposes someone toa particular disease. Based on the correlative evidence alone,obesity might be linked to elevated disease risks by a geneticpleiotropy; that is, a genetic problem leads a person not only todevelop obesity, but the same problem, in association witha Western lifestyle, independently elevates the disease risk. Ifthis were true, simply reducing the level of body fatness would

not remove the underlying cause of the elevated disease risk.However, several recent studies consistently show im-provements in associated health outcomes for people who wereobese but have sustained weight loss for protracted periods.Consequently, it is obesity itself and not a shared underlyinggenetic lesion that causes elevated disease risk.

The exact mechanisms by which obesity leads to thedevelopment of disease is an area of intense study. This isbecause it is not obesity itself but rather the healthconsequences that are the issue. If the nature of the causallink between obesity and health problems could be elucidated,then developing therapies that break that link might bea feasible solution to the problem. This is obviously attractive,because it would mean people could continue to pursue

a Western lifestyle while avoiding the negative consequences.There are two basic alternatives that have been proposed tolink obesity with diabetes. The first suggests that becauseobese people have much higher circulating levels of free fattyacids, in muscles, these might compete with circulatingglucose. This leads to persistently elevated circulating glucoselevels, elevated insulin secretion, and ultimately, insulinresistance. The alternative suggestion is that some productssecreted by fat tissue actively interact with insulin to generateinsulin resistance or promote insulin sensitivity. Someimportant secretions from adipose tissue in this respect areTNF-a (23,24), adiponectin (25,26), and resistin (26,27).Blocking the signals that link obesity to diabetes may be aneffective therapy for obesity-mediated diabetes, but this

approach is only likely to address one disease consequence ata time. Hence, a much more useful approach is to actually try

4 Abbreviations used: AgRP, Agouti-related protein; a-MSH, a-melanocytestimulating hormone; ARC, arcuate nucleus; BMI, body mass index; BMR, basalmetabolic rate; CART, cocaine and amphetaminerelated transcript; db/db, mutantmouse lacking functional leptin receptors; fa/fa, mutant rat lacking functional leptinreceptors; GABA, g-amino butyric acid; MC1-R, MC3-R, and MC4-R, melano-cortin-1, -3, and -4 receptors; NPY, neuropeptide Y; ob/ob, mutant mouse lackingability to produce leptin; POMC, pro-opiomelanocortin; PVN, paraventricularnucleus; PYY3-36, peptide YY3-36; QTL, quantitative trait loci; RMR, restingmetabolic rate; TNF-a, tumor necrosis factor-a; tub/tub, fat mutant mouse.

2091SOBESITY: GENETIC AND ENVIRONMENTAL INTERACTION

byonOctober8,2008

jn.nutrition.org

Downloadedfrom


3/16

to eliminate all the disease consequences at once by treatingobesity itself.

Energy balance and obesity

Obesity is a problem of imbalance between energy intakeand expenditure. Total energy demands can be partitioned into

several different compartments. The requirement to sustaingeneral cellular processes is known as the basal metabolic rate(BMR). BMR is measured under strict conditions whereby thesubject is completely inactive, is not digesting food, and is ata thermoneutral temperature. If we sit quietly, we burn moreenergy than BMR, and this is generally called the restingmetabolic rate (RMR). After food ingestion, our metabolic raterises further; when we start to move around, it increases evenmore and is termed active energy expenditure. To balance ourtotal energy expenditure, we eat food. In the short term,expenditure and intake do not match closely, because we feedin discreet bouts, whereas expenditure is a continuous process.Because of the laws of thermodynamics, which state that energycan neither be created nor destroyed, the imbalance between

intake and expenditure requires that we also have the capacityto temporarily store energy. Generally, the equation is foodenergy intake5 energy expenditure1 energy storage. We storeenergy as fat, because it is much denser than carbohydrate andalso does not require large amounts of water for storage. Giventhe energy-balance equation, it is clear that obesity results fromeither food intake being too high, expenditure being too low(through low RMR and/or activity expenditures), or a combi-nation of both.

The obesity phenomenon is usefully considered as havingtwo different facets. The first is the temporal trend in theprevalence of overweight and obese people over time,particularly over the last 25 y. Most people alive today (in2004) were also alive in 1980. Consequently, the expansion of

the prevalence of obesity between 1980 and 2004 must reflectchanges that have occurred in our behavior (activity and foodintake) that are dependent on the changing environment

rather than the effects of population genetics. The second facetis the differential susceptibility of individuals to the problem.Whatever environment or time period we choose, we find a mixof obese and lean people. These differences may reside in themicrostructure of the environment experienced by eachindividual or may depend on genetic differences. An importantpoint to recognize is that genetics and environmental factorsexert their effects on energy balance and obesity via effects on

our behavior and physiology ( Fig. 1). There is no direct linkbetween our genes and our body weight or fatness. This isfrequently misunderstood as is evidenced by discussions aboutwhether obesity is caused by our genes or our behavior (e.g.,28). As Figure 1 makes clear, these are not alternative causalagents. Behavior and genes are different levels of the samecausal framework. The question of genes or behavior makesno more sense than the question of behavior or energybalance. Consequently, although we might say that obesity hasa large genetic component to it, this doesnt mean that theobese have somehow miraculously deposited enormous quan-tities of body fat without eating too much food, or expendingtoo little energy, or doing both.

Facet one: trends in obesity over time

The prevalence of overweight and obesity in people hasincreased dramatically in the last 3050 y. For example, in1980, 7% of the UK population was classified as obese. By 2000this was 20%. Overweight subjects comprised ;20% of the UKpopulation in 1980. By 2000 this was 45%. Consequently, themajority of current UK society is either obese or overweight (7).The UK numbers are on the high side but are typical of manyEuropean nations (1). In the U.S., the progression of obesity is;34 y ahead of the problem in Europe; 2003 estimates suggestthat 31% of the population is obese (BMI . 30) (29).Moreover, obesity has even greater prevalence in some

subpopulations. For example, the prevalence of obesity inblack populations across the entire U.S. is ;5% higher than incolocalized white populations (30), and for black females,

FIGURE 1 The major causal linkagesamong genetics, environmental effects,physiology, behavior, and energy balance.Multiple causal routes exist by which bothenvironmental and genetic effects may exerttheir influence. The diagram clarifies thatbehavior is not an alternative mechanism togenetics,they are different levels of the samephenomenon. Perceptions of energy imbal-ance feed back into behavior and physiologyto influence both intake and expenditure. Thenature of these feedback links is disputed,and they are omitted from the above frame-work. The different feedback models areaddressed in Figure 2.

2092S SUPPLEMENT

byonOctober8,2008

jn.nutrition.org

Downloadedfrom


4/16

prevalence is even higher with ;50% obese. Socioeconomicclass and education also appear to be important factors, withthe lowest educated groups having ;5% higher prevalencethan more educated subpopulations (31,32). Similar demo-graphic patterns are reported for almost all Westernizedsocieties. In addition, there are signs that this problem hasachieved a truly global distribution during the last decade;many Asian, South American, and African nations report that

obesity rates are increasing rapidly (36,33).It is easy to identify some correlated societal trends to the

obesity epidemic. One of the factors believed to be important istelevision viewing (3439). By the early 1970s, televisions werealready a virtually ubiquitous household feature in the U.S. andmost of Europe. What has changed is the amount ofprogramming. We now have a choice of many differentchannels to watch 24 h/d: there is always something to catchour attention and potentially keep us watching. By the late1990s it was estimated that 20% of children in the U.S. werewatching .6 h of television every day. Parental ratings of howmuch television a child watches are positively associated withthe childs BMI (35), and having a television in a childsbedroom is a significant risk factor for obesity (37); also,

television viewing is negatively associated with engagement inphysical activity (38,40). However, significant differencesappear to exist in the associations within different ethnicsubpopulations, which suggest that the linkages are complex(39). In adults, it was suggested that the association betweentelevision viewing and BMI is stronger in women than in men(36,41), but other studies have not reported this difference(42).

Many other changes in activity patterns have occurred overthe last 50 y or so. In the 1950s most people in Westernsocieties walked around a range of small local shops to do theweeks shopping. This has been almost completely replaced byone-stop, out-of-town hypermarkets, where one can parkconveniently close to the store, buy everything in a large cart

that doesnt have to be carried around, and then drive it allhome. There is no doubt that these changes were largelyfacilitated by the spread of car ownership. Out-of-townsupermarkets could not have been a commercial propositionin the 1950s, because only a small sector of the populationcould have used them. Another factor in temperate zoneswhere it is cold in the wintertime is the work involved inheating ones home. The spread of central heating systems hasreplaced the hard work that it took to keep coal fires burning.The domestic appliance market has also made enormousinroads into the labor that typified housework in the 1950s and1960s. Washing machines, vacuum cleaners, dishwashers, spindryers, and tumble dryers have all reduced domestic workchores. Our addiction to domestic appliances is such that we

now have devices that take any effort out of even the mosttrivial of tasks, such as electric toothbrushes and carvingknives.

At the same time that there have been changes in activitypatterns, there have also been large changes in eating habits.Key dietary behavior shifts include greater consumption of foodaway from the home (43) and large increases in total energyfrom salty snacks, soft drinks, and pizza (44,45). Despite thesechanges, some studies suggest that total daily energy intake hasactually declined over the last 30 y (46,47), but other studiessuggest an increase (44), and yet others find no significanttrends (48,49). At national levels, however, food-sales statisticsshow that for almost all food types, the same amount of food ormore is presently being purchased per individual compared with

20 y ago. This is particularly the case for confectionary items.Over the last two decades, fat consumption has declined in

parallel with the increased prevalence of obesity in both theU.S. (50) and Europe (48), and the decline is matched by aparallel increase in carbohydrate consumption (48,49).

It is clear that social changes involved activity reductions,and this was potentially combined with an increase in energyconsumption. We should be cautious, however, about drawingsimple cause-and-effect arguments from these correlations withthe spread of obesity. For example, enrollments at health clubs

have also expanded over the same period that obesity hasincreased. In addition, the stimulation of television and theavailability of cheap electric lighting means that on average,individuals spend 2 h less each day asleep compared with the1920s (51). Time spent sleeping is positively associated withobesity (42), so this change should be protective. Despite thesecaveats, some intervention studies have assessed the roles ofdifferent factors. Studies in the U.S. were performed to try tointervene in childrens television viewing habits to determinethe impact on body fatness. These interventions show asignificant effect on body weight. Surprisingly however, this wasnot because the children adopted more active lifestyles whendeprived of their viewing time. Rather, the major differenceswere in the consumption of snack foods: the children ate

these when watching television, but not when engaged inalternative pursuits (even sedentary ones).

Facet two: individual differences in susceptibility

The second facet of the obesity problem is that despite livingin similar environments, not everyone gets obese. This effectcan be traced to individual differences in behavior andphysiology. Millions of people eat regularly at burger bars, yetothers never visit them. In contrast, some people exerciseregularly, whereas others almost never do. Some of thesedifferences reflect substructuring of the environment in whichwe live. For example, middle-class areas may have betterprovision of parks and cycle paths, which encourage recrea-

tional activity. This is only part of the explanation, however,because whatever environment or sector of the community ischosen, it is populated by a mix of obese and lean people. Oneexplanation for these differences is that for everyone, there aresubtle differences in lifetime-accumulated experience andsocial factors that lead to differences in behavior andphysiological energy expenditure. This experience may includethe period spent in utero. An alternative viewpoint is that thesebehavioral differences have a large genetic component to them.This model suggests that, when immersed in the sameenvironment, some people develop obesity because theirgenetic constitution is such that they are more likely to adoptbehaviors or have physiologies (such as low RMRs) that leadthem into positive energy balance.

Calculations indicate that the heritability of BMI is high.Accordingly, studies of monozygotic and dizygotic twins raisedapart and together and pairs of adopted and nonadoptedvirtual twins indicate the variation in body mass that isexplained by genetic factors is also very high. A meta-analysisrecently indicated that the average variation explained bygenetic factors was 77%, and that shared environmental factorswere insignificant (52). Faced with this information, it isimportant to recognize what exactly is being said. A commonresponse to the suggestion that shared environments areunimportant is to point out that if one were to take a pair oftwins and keep one of them locked up with minimal food for

years on end, the deprived twin would not develop the sameBMI as the twin given free access to food; hence it is obvious

that environment must be important. However, although thishypothetical experiment would probably yield the suggested


byonOctober8,2008

jn.nutrition.org

Downloadedfrom


5/16

result, such extreme environments are seldom encountered inWestern society. It appears that once certain environmentalrequirements are met, additional variability in the environmentbecomes relatively unimportant and genetic factors predomi-nate. A counterexample to the starved-twin scenario clarifiesthis point: if I had an enormous refrigerator in my housecontinuously stocked full of food, it would probably have noeffect on my actual food intake if I were provided with a second,

similarly continuously stocked refrigerator. Once I have meta certain minimal environmental requirement, additionalvariation in the environment is irrelevant, and this appears tobe the present situation in Western society. Illustrating thispoint further, the average income in the U.S. (in 2002 figures)was $27,000 annually; yet it is possible to purchase ones entiredaily energy requirements for ;$5/d (,5% of the averageannual income).

In this context, the changing prevalence of obesity over timemust be a gene-environment interaction effect. Addressing thegenetic side of this interaction is potentially a far more tractableproblem than addressing the environmental component byreengineering society, because the level at which interventionsmight ultimately be made is the individual rather than the

society as a whole. There is a clear need, therefore, tounderstand the genetic bases of food intake, energy expendi-ture, and hence, energy-balance variations.

Regulation of energy balance

There is a generalized opinion that energy balance, andhence, body weight and/or fatness, are phenomena that areregulated (e.g., 5357). The origins of this idea seem to residein calculations of the potential impacts of imbalances in ourenergy intake and expenditure. For example, on average,a typical candy bar contains ;250 kcal (1 MJ) of energy. If overthe last 23 y I had eaten a candy bar more than my energyexpenditure every day, I would have accumulated 23 3 365

3 1000 kJ 5 8395 MJ of extra energy. Because 1 kg of fattissue contains ;33 MJ of energy (58), the 8395 MJ would beequivalent to deposition of;254 kg of fat tissue! Even if I hadeaten a candy bar above my requirements only once each week,I would have accumulated 36 kg of body fat. If such smallimbalances in our intake and expenditure can have such majorimpacts on body weight, the argument is that these phenomenanormally must be regulated in some manner. However, thiscalculation involves several dubious assumptions. It is assumedthat all of the excess energy is deposited as fat, and that thedeposited fat tissue expends no energy. More realisticassumptions about the nature of the deposited tissue and theenergy expenditure of that tissue lead to the conclusion thattrivial imbalances in our energy balance generally lead to trivial

changes in our body weight (see, for example, 5961).Development of obesity actually requires quite severe im-balances in ones energy budget. Nevertheless, despite thewidespread acceptance that the trivial-imbalance argument iswrong, it is still widely believed that energy balance, energyintake, and body weight are regulated phenomena. This iscertainly not an outdated concept. A search on the ISI Web ofScience bibliographic database (Sept. 2003) on the term bodyweight regulation revealed 432 articles with this phrase in thetitle or abstract published between 1981 and 2003, of which 61(14%) were published in the first 9 mo of 2003 (3.4% of thetotal time).

Several different models are available for the manner inwhich energy balance may be regulated (Fig. 2). In the simplest

model (Fig 2A), daily food intake and daily energy expenditurevary as independent stochastically variable traits. The differ-

ence between the intake and expenditure generates a dailyenergy storage term that is either positive or negative de-pending on the contributing factors. If the storage is positive,

energy is deposited; if the term is negative, energy is withdrawn(61,62). The subject withdraws or deposits energy into or fromeither stored fat or lean tissue depending on a utilization ratiothat is generally called the p ratio (59). All tissue expendsenergy, but the rate of energy expenditure in lean tissue greatlyexceeds that in fat tissue. If there is an energy surplus, thegrowth of the lean and fat tissues increases energy expenditure.A sustained energy surplus will therefore result in sustainedgrowth of both lean and fat reserves until the expenditurebalances the intake. The differential energy expenditure of leanand fat tissue, however, means that the amount of fat that isdeposited before the subject regains energy balance dependscritically on the p ratio. If this ratio is low, which reflectsdiversion of energy into lean tissue, stability is regained at a

lower body fatness than if the p ratio is high. The key questionwith respect to this model is whether the p ratio is an

FIGURE 2 Three alternative models of energy-balance regulationinvolve feedback from energy imbalance into energy intake andexpenditure. In one model, food intake and energy expenditure arestochastically variable from day to day and independent of each other (A).The resultant variations in energy balance translate into storage of lean

and fat tissue controlled by the p ratio. Lean and fat tissue also directlyaffect the levels of energy expenditure. In the second model, there ismemory of the historical energy balance that feeds back onto food intakeand energy expenditure (B). If an animal spends a period in energy deficitunder this model, it will feed back an impetus to increase intake andreduce expenditure. Alternatively overconsumption will lead to conversefeedback, reduced intake, and increased expenditure. In the third model,the same feedbacks relative to historical energy balance come from thefat and lean tissue reserves rather than historical memory (C).

2094S SUPPLEMENT

byonOctober8,2008

jn.nutrition.org

Downloadedfrom


6/16

individually defined parameter under genetic control orwhether it is instead variable and itself dependent on bodycomposition such that when subjects have large fat reserves,they preferentially mobilize and deposit fat, but during periodsof negative energy balance, they reduce the p ratio and start tomobilize protein as well and thus autoregulate (to some extent)their body composition (60). Several arguments suggest it isunlikely that the p ratio is individually fixed (63).

The p-ratio model can explain many features of energybalance without the need to invoke any regulatory signalsgenerated by lean or fat tissues, particularly if the p ratio isfixed genetically (59). The model, however, is an inadequatedescription of phenomena related to energy regulation (6063). The main difficulty faced when using this model isexplaining patterns of intake that occur after periods of fooddeprivation. During food deprivation, subjects tend to reducetheir expenditure more than can be accounted for by the lossof metabolizing tissue (6466). On refeeding, they also tend toexhibit post-restriction hyperphagia. These patterns would notemerge if intake and expenditure were independent stochasticvariables linked only to storage by the p ratio. The salient issueis this: When we have been food deprived, where do the

feelings of hunger come from that cause us to subsequentlyovereat and reduce our energy expenditures to redress theprevious energy imbalance? Two basic possibilities exist. First,there may be a sensing mechanism that monitors the extent ofenergy imbalance by the fluxes of nutrients related to intakeand expenditure. The glucostatic model of food intakeregulation (67) is this type of energy balanceregulationmodel. By this hypothesis, feeding behavior is regulated byfluctuations in circulating levels of glucose, which are sensitiveto patterns of both energy expenditure and intake. Thus ifglucose-sensing centers in the brain detect a drop incirculating glucose levels, feeding is stimulated, and feedingis then terminated by the increase in circulating glucose.Glucose status, therefore, acts as a sensor for immediate energy

balance. This is, however, a signal that monitors only theshort-term flux of energy. It is likely that animals, includinghumans, have also evolved mechanisms to monitor patterns ofenergy imbalance over much more protracted periods. Thiscould be an accumulated memory of energy balance (Fig. 2B)or, an alternative source of such signals is the energy storesthemselves.

The third class of model therefore invokes a feedback loop(Fig. 2C) generated from the stored energy (fat and lean tissuestores) (61,6870). This could work in several different ways,but two models predominate. One mechanism is that energyintake and expenditure rates are tied into the sizes of the energystores by an amplification mechanism (71). When stores arelarge, the signals from the stores do not stimulate feeding very

much but have a strong effect on expenditure. Conversely,when the stores get smaller, the effect on expenditure isdiminished and the effect on feeding is increased. The secondway this might operate is by reference to a target store size thatis encoded centrally (7275). Because most energy is stored asfat, the suggestion is that the regulated body component is bodyfatness (with lean body mass dragged along by some unknownmechanism in the p ratio). Other models are also feasible wherelean mass is directly regulated (68). The regulation of bodyfatness by reference to a centrally encoded target is generallyattributed to Kennedy (69) and is called the lipostatic modelof body weight regulation.

At present, the exact nature of the regulatory model forenergy balance and body weight remains uncertain, and

separating the alternatives models is not easy. In particular,the feedback from an accumulated memory of the status of

energy balance (model 2) and feedback from the stored reserves(model 3) might be difficult to resolve. This is because in mostcircumstances, the two factors move in parallel. It is seldom thecase that someone could accumulate an energy deficit but notaffect levels of their stores and vice versa. To test between thesemodels, therefore, requires an experimental design thatseparates the different effects. If the system works as envisagedin model 2 (Fig. 2B), with a memory of energy deficit feeding

back to drive intake and expenditure, then it seems likely thatthis memory will decay over time. For example, if I starved youfor 24 h, you would feel hungry the next day. If I gave you freeaccess to food the day after you had starved, you would overeatto compensate for the missing intake. However, if I stopped youfrom overeating for a week after your starvation day, but gave

you enough to balance your energy demands, and then gaveyou free access to food, you would probably not feel as hungry as you did the day after you had starved. This is becausethe memory of the energy deficit had receded because of theintervening time spent in energy balance. This decline in theforce of memories of energy deficit over time could be used toseparate regulation by model 2 from model 3 (Fig. 2C). Thedesign is as follows: a set of subjects is placed on an energy-

restricted diet that provides fewer calories in food than theywere expending with a resultant loss of body weight (fat andlean tissue). At the same time, they accumulate memory ofenergy deficit. After a suitable period on restriction (of say,3 mo), thesubjects are randomized into one of twogroups. Group1 is taken off the diet immediately and given free access to food.Group 2 is fed enough food to keep the body mass constant andis regulated on a daily basis for an additional interventionperiod of 3 mo and is then given free access to food. At the endof this period, group 2 will have the same body fat and leanstores as group 1, but will have been fed in energy balance for3 mo. Hence, if it is the size of the stores that drives thehyperphagic responses, the extent of hyperphagia in the releasephase of the experiments will be the same for the two groups.

In contrast, if it is the memory of energy deficit that drivesthe hyperphagic response, group 1 will have a much strongerhyperphagia than group 2, because the memory of deficitwill have receded over the 3 mo that group 2 was fed in energybalance. In general, the more time spent feeding in energybalance, the lower will be the predicted hyperphagicresponse.

These experiments have not yet been performed, but wouldbe extremely informative. It seems likely that elements of allthree different models are important. First, short-term signalsprobably play a critical role in switching feeding behavior onand off and in regulating short-term activity patterns. The pratio is an important aspect of nutrient allocation that governshow energy surpluses and deficits are translated into body

composition, whereas long-term signals that potentially interactwith centrally encoded targets provide an overall orchestrationof the short-term regulation mechanisms.

There is some experimental support for this regulatoryframework. Many studies show that when subjects are fooddeprived, they compensate for the deprivation by reducing theirenergy expenditure. Although less frequently performed, theconverse, increasing expenditure in response to overfeeding, isalso observed (e.g., 76,77). Perhaps the most comprehensivestudy was that of Leibel et al. (78), who placed obese and leansubjects in a protocol where they were systematically under- oroverfed to produce differences in their body weights by ;10%.The researchers then examined the compensatory responses inthe subjects energy-expenditure patterns. As anticipated by

the model, when subjects were underfed, they responded to theenergy deficit by decreasing their expenditure more than could


byonOctober8,2008

jn.nutrition.org

Downloadedfrom


7/16


8/16

(a-MSH), whereas NPY/AgRP neurons secrete AgRP. a-MSHis an agonist, and AgRP is an antagonist of both MC4-R andMC3-R (109). When leptin molecules from the periphery dockwith their receptors on NPY/AgRP neurons in the ARC, thereis suppression of NPY release (110,111) and reduction ing-aminobutyric acid (GABA) release at the synapses where theNPY/AgRP neurons meet the POMC/CART neurons. Underthe GABA-mediated disinhibition and combined with direct

leptin stimulation, the POMC/CART neurons release a-MSHat the MC4-R on PVN neurons at the same time that theantagonist AgRP derived from the leptin-stimulated NPY/AgRP neurons is reduced. Secretion of a-MSH depends onsuccessful cleavage of the POMC molecules, which is broughtabout by the proconvertase-1 enzyme. Additionally, thestimulatory effect of a-MSH appears to depend critically onits acetylation status, which is also under regulatory control ofacetylase-deacetylase enzymes. The increased a-MSH andreduced AgRP levels stimulate the MC4-R with two down-stream effects. First, the sympathetic system is stimulated andreleases noradrenaline at peripheral cells that express b-adren-rgic receptors. In brown adipose tissue in rodents, this producesan immediate stimulation of metabolic rate. At the same time,

food intake is inhibited. The resultant negative energy balancecauses leptin levels to decrease. This reduced signal has theopposite effects in the ARC nucleus: NPY/AgRP cells arestimulated to release NPY; POMC/CART cells are inhibited(directly and indirectly), and at the MC4-R in the PVN, AgRPrelease is enhanced, whereas a-MSH level is reduced. Thisleads to reduced sympathetic activity, reduced energy expen-diture, and stimulated feeding behavior.

There is now considerable evidence that the leptin systeminterplays with other signals that may also be involved in long-and short-term regulation. The best-characterized of these sofar is insulin. Like leptin, the secretion of insulin is proportionalto the magnitude of the body fat stores. Neurons in the ARCthat express the leptin receptor also often express the insulin

receptor IR-1. The insulin receptor signals via thephosphoinositide 3-kinase pathway, and there is suggestion(112) that intracellular interaction between leptin signalingand insulin signaling may exist. Insulin is known to inhibit NPY(113), and animals with reduced numbers of insulin receptorshave elevated food intake consistent with increased NPY levels(114). Insulin may act as a long- and short-term signal given itsresponsivity to altered glucose levels after feeding. Similarshort-term signals include glucose and free-fatty acids them-selves and peptides secreted from the gut such as cholecysto-kinin (115), ghrelin (116), and peptide YY (PYY,117),receptors for which also colocalize to ARC cells.

The situation in humans is likely to be more complex than inrodents because of the input of cortical-level processes (118).

Moreover, we are aware that endogenous cannabinoids and theserotonergic and opioid-dopaminergic (119) systems are alsoinvolved, but the details of the interactions presently remainobscure. Nevertheless, this model potentially explains whymany people are able to regulate their body weight over periodsof many years. Whenever one puts on too much weight, leptin,insulin, and probably other signals are generated that reducefood intake and increase energy expenditure until the signalsand weight are normalized. Whenever one loses weight, theopposite occurs: the multiple signals decrease in magnitude,food intake is stimulated, and energy expenditure is suppressedto regain the missing weight.

One major argument against the lipostatic theory of weightcontrol is the fact that so many people are presently obese. If

such a system operates, the argument goes, then clearly itcannot be very effective. However, even if lipostats exist,

obesity may develop for two reasons. First, obese people mayhave broken lipostats; or second, the lipostats in obese peoplemay be set at very high target levels. After the discovery ofleptin, there was immediate interest in whether obesity wasa result of deficient leptin productiona broken lipostat.Obese people might have abnormalities in their ability toproduce leptin in much the same way as the leptin-deficient ob/ob mouse. Imagine, for example, if my leptin production was

only half of the normal rate. My brain would think that I wasonly half as fat as I actually am, and it would encourage me toeat more food and expend less energy until my fat content washigh enough to redress the leptin-production deficiency. If thiswere the case, the potential to cure obesity would be in sight:one could treat people with leptin (or drugs to mimic its actionor stimulate natural leptin production) to trick the brain intothinking that one was fatter than was actually the case. Theregulatory system would then downregulate food intake andincrease energy expenditure to redress the balance. The onlydownside is that this would need to be a lifelong treatment,because discontinuation would lead to decreased leptin levelsand resultant weight regain. However, the euphoria over thepotential of leptin was short lived, because it was found (120

124) that in general, fat people have very high leptinconcentrations, as might be expected from their high bodyfatness if their leptin-production systems were workingcorrectly. The vast majority of obese people do not haveleptin-production deficiencies or defects in their leptin re-ceptors (125).

An interesting exception was two children from Cambridge,UK (126). These children were ,10 y old when they werediscovered, but they already had ravenous, insatiable appetites.This had resulted in extreme obesity that had forced theirparents to seek hospital treatment for them. These childrenwere massively obese; for example, the older child, when aged10 y, weighed 110 kg. Both children had no detectable levels ofcirculating leptin in their blood. Additional investigations

showed they had a recessive mutation in their leptin gene thatprevented leptin production. Once the defect was diagnosed,the children were treated with recombinant leptin. With dailyinjections, their dramatic weight-gain trajectories and compul-sive-eating behavior were reversed (127). Since this time, otherfamilies have also been described with leptin-productiondeficiencies including a family of three people from Turkeywhose body weights almost halved when they received dailyleptin injections for 10 mo.

An interesting subpopulation of the obese are individualsthat are heterozygous for the leptin deficiency. These in-dividuals have lowered leptin production for their body fatnesslevels, and as a consequence they are on average more obesethan their homozygous wild-type relatives (128). These

heterozygous individuals are also extremely rare but providean important proof of principle regarding the functioning of thelipostatic control mechanism. Some individuals have beenidentified with loss-of-function mutations for the leptinreceptor Ob-Rb. A small group of individuals has also beenidentified with mutations in the POMC gene, which makesthem unable to secrete a-MSH (129,130). Because thesesubjects do not secrete a-MSH onto the MC4-R in the PVN,they have a perpetually stimulated appetite and developobesity. A second consequence of this mutation is the subjectshave bright-red hair (130) because of an absence of a-MSHinteracting on MC1-R in the periphery (131). However, themost important single-gene defects of which we are presentlyaware concern polymorphisms of the MC4-R itself (132134).

Many polymorphisms have been identified in this gene (135)most of which do not have any functional significance (136).


byonOctober8,2008

jn.nutrition.org

Downloadedfrom


9/16

Some key polymorphisms of MC4-R are associated with obesity,and in most populations that have been studied to date,mutations of this gene account for;35% of all morbid obesity(133).

Overall then, completely broken lipostats explain onlya small portion of the total numbers of obese and overweightpeople. What about the genetic mis-setting of lipostats athigh targets, which then interact with the environment to

generate the temporal and spatial patterns of susceptibility thatwe presently observe? Is this a feasible alternative explanationfor the high prevalence of obesity? The negative-feedbackmodel of body weight regulation has many similarities withthermostatic temperature regulation in a house (137). In thatsystem, a room-temperature sensor compares the actual anddesired room temperatures, and an effector system (theradiators) can be modulated to bring actual and desired statesinto balance. We can understand the interplay between genesand the environment by using this thermostatically heatedhouse analogy. Imagine houses have a whole range ofthermostatic temperature set points from cold to hot. If thehouses were all in the arctic and were poorly insulated, then allthe house temperatures would be cold because even those with

high set points (or broken thermostats) couldnt get as hot astheir thermostats want them to be because the capacity of theirheating systems wouldnt be able to match the heat lossthrough their roofs. This is equivalent to the genetic systembeing environmentally constrained. However, if the houseswere all in the tropics, where the environmental temperature ismuch warmer, then those with high target temperatures (orbroken systems) would reach their high temperatures. This isequivalent to the genetic system being released from environ-mental constraint.

The key point is that actual house temperatures depend onboth the thermostat characteristics and the environment inwhich the houses are located. Our genes may predispose us toobesity, but genes can only be expressed in environments, and

it is the gene-environment interaction that is most important.Obese people might have high target body fatness settings intheir lipostats that result in actual high body fatness in Westernsocieties where there is ready access to high energycontentfoods. Presently in rural, primitive farming communities, and inthe past in Western societies, people might have (or have had)high target settings or breakage in their regulation systems.However, if they had to work 12 h/d digging fields or chasingprey, they may have been unable to ever consume sufficientfood to reach this target.

Evolutionary context

One difficulty with this interpretation is understanding the

evolutionary processes that might result in systems evolvingwith such variable individual lipostat settings. Our knowledgeabout the evolution of body-weightregulation mechanisms inhumans is poor. Generally, arguments tend to paint simplisticadaptive scenarios emphasizing that storing large amounts of fatmight at some point have been selectively advantageous (e.g.,because it would provide increased resistance to starvation) theso-called thrifty gene hypothesis. The main problem with suchadaptive scenarios, however, is understanding why such anadvantageous genotype would not rapidly replace the dis-advantageous non-thrifty genotype, and make everyone inmodern society susceptible to obesity. This problem is seldomrecognized in such adaptive landscapes.

Studies of wild animals suggest that stored body fat is

actually regulated under stabilizing selection as part of a dynamictradeoff between the risks of starvation, which promote fat

storage, and the risks of predation, which promote leanness(138147). Because early humans likely also faced thesecontrasting selective pressures, they probably also evolveda regulatory system (i.e., a lipostatic system) that promotes fatstorage to avoid starvation but also prevents excessive fatstorage to avoid predation (148). The set point of this systemmay have been temporally variable because of the differingselective pressures at different times of year. Several key events

in human evolution likely have dramatically reduced the risksof predation. These include the evolution of social behavior,which allows groups of humans to drive away predators andalert each other of the presence of danger. Such behaviors arealso observed in modern-day groups of apes and monkeys suchas vervet monkeys, which have evolved complex signals toindicate the approach of different types of predators (149152).The harnessing of fire ;1.8 million y ago (153) and theconstruction of tools to actively aid defense would furtherenhance protection against the threat of predation. Under thisrelease from constraint, it is feasible to imagine that target setpoints might drift upward at random because they would haveno selective consequences as long as actual body weights wereconstrained by food supply. This would remove the strong

selection that imposed an upper limit on fat storage, but defenseagainst energy deficits would still be strongly selected for. Thisasymmetry accords with modern experience (154). In thisevolutionary scenario, the capacity to deposit enormous fatstores is not seen as adaptive; hence there is no need to explainwhy the trait did not spread through the entire population.Several hundred generations later, when faced with theWestern lifestyle, the continued absence of strong selectionon the lipostat target would result in a diversity of targets andthe consequent diversity of body-fatness phenotypes that wepresently observe. One additional prediction from thishypothesis is that the greatest release from the constraints atthe upper boundary of regulation should occur in thosepopulations that have been most liberated from the risks of

predation. There is no lower risk than living in a completelypredator-free environment, and although such environmentsare uncommon, small remote islands generally have nopredators that are likely to pose a risk to humans. It is perhapssignificant that the greatest levels of obesity in modernpopulations occur in Pacific Islanders, where some islands have.70% of people in the obese category (e.g., Tonga; 155).

Why conventional behavioral (energy restriction)approaches to weight loss dont work

The most-common treatment prescribed for obesity andoverweight is to engage in an energy-restricted diet. Indeed inmany countries, before obese subjects can be prescribed drugs

to assist them, they need to have tried this option first.Moreover, this is a method of choice for attempting to reduceweight by overweight subjects (and people in the normal weightrange). Recent surveys suggest that at any one time, 40% ofpeople in Western societies may be engaged in some form ofenergy restriction. Vast numbers of options are open to people,and a large dieting industry caters to these needs. Manyorganized dieting clubs promote calorie restriction as themethod of choice for regaining control of body weight.Unfortunately, almost all of these approaches provide thepromise of short-term success coupled with long-term failure.

If one consumes a low calorieintake diet that provides lessenergy than one expends, the result is reduced energy storagethat is manifested as a loss of body tissue, some of which is fat

and some of which is lean, depending on ones p ratio, whichvaries with how fat one already is and perhaps also the rate of

2098S SUPPLEMENT

byonOctober8,2008

jn.nutrition.org

Downloadedfrom


10/16

weight loss. Whatever is claimed to be the mechanism, allsuccessful weight-reduction diets work on the principle ofenergy deficit. The major differences between diets are theeffects that they have on our short-term perceptions of hungerand their side effects on other systems. Presently, popular dietssuch as the Atkins diet rely on large quantities of dietaryprotein almost to the exclusion of carbohydrate. The scientificbasis for this diet is that there appears to be a hierarchy of

macronutrient effects on perceptions of hunger: fat is leastsatisfying, carbohydrates next, and finally, the most satiatingmacronutrient is protein (156). A diet that provides an energydeficit, where most of the calories come from protein, thereforeleaves one feeling less hungry than if the same deficit came froma fiber-dominated diet. But weight loss is the same for the sameenergy deficit. The difference is that because perceptions ofhunger are different, it may be easier to adhere to a high-protein diet than a high-fiber diet, and the longer one stays ona diet, the more weight is lost.

Potentially major problems exist with high-protein/low-carbohydrate diets. First, these diets are generally deficient inmicronutrients (which can be rectified by taking a vitaminsupplement) and also, possibly, in carbohydrates that are

important for optimal function of the central nervous system.Second, a high protein intake is generally coupled with a highsaturated fat intake and a high serum cholesterol level, both ofwhich are implicated as causal factors in cardiovascular disease,cancer, and stroke independent of the obesity they engender.Urinary calcium excretion increases with high protein intake.Moreover, bowel cancer is strongly negatively correlated withdietary fiber content, and so excluding fiber from the dietenhances the risk of developing bowel cancer. Processing highprotein loads may also increase the risk of kidney failurebecause of the concomitant high rate of urea production.

These high-protein/low-carbohydrate diets are consequentlyonly appropriate for the short term. Even then, the benefits ofweight loss need to be balanced against the negative side

effects. During 2001, the American Heart Association and theWorld Cancer Research Fund both came down heavily againstthese diets as aids in the fight against obesity. A meta-analysisof high-protein diet interventions and their effects concludedthat high-protein diets do result in greater weight loss thanconventional diets, but that one of the largest potentialproblems they pose is increased calcium excretion in urine,which may ultimately contribute to excess bone loss (157).

The most beguiling aspect of all energy-controlled diets isthat in the short term, they do result in significant weight loss. Ifone is in caloric deficit, then ones feedback mechanismsrespond by attempting to minimize the magnitude of thedifference between what is being eaten and what the bodyrequires. One becomes lethargic, and activity level and resting

metabolic rate are reduced (78). Consequently the rate ofweight loss slows as the time under restriction progresses, andeventually a steady state is reached where the reduced intake ismatched by reduced energy expenditure. To sustain additionalweight loss requires an even more stringent reduction in energyintake. The weight that has been lost is maintained only withsustained dietary control.

The only way to permanently lose weight on a diet is to staypermanently on the diet or replace it with some alternativeform of caloric intervention like exercise (158). This is becauseonce the diet is stopped, body fat content is considerably lowerthan the bodys target fatness; the body produces low levels ofleptin and other signals to indicate it is under target. Themajor impetus of the lipostatic regulation system is to redress

this imbalance by impelling the individual to eat more food. Toemphasize this, it is noteworthy that an almost 100% successful,

permanent treatment for obesity is surgical intervention such asgastric banding. This works because it forces people to complywith calorie restriction forever.

We can understand why calorie-restricted diets dont workforweight control by reconsidering the analogy of thethermostatin a house (137). Imagine your thermostat is set very high and

your house is hot. You arent happy with this situation, butbecauseyou dont know how the thermostat works, you call in an

expert who opens the sitting-room window. There is animmediate improvement in the house temperature. Over thenext few days, the house cools to a more comfortable level, but itis still a bit too warm; you need to open another window to keepthe temperature decreasing. Eventually you reach a desirabletemperature. Somebody comes by and says that you might haveexposed yourself to some risks by opening your windows likethat: your kids might get colds in the drafts, and you might getburgled. You decide after a while that having the windows openis not a great idea, and you close them. Immediately thetemperature starts to rise again, and soon you are back to yourboiling-hot house. Energy-restriction diets are like attempts tofix your house temperature by opening a few windows. They areeffective, but they only work as long as you stay on them, and

they sometimes have undesirable side effects. Usually, dis-continuing the diet results in you regaining the original weight(159). In fact, the diet foodproducts industry is a multibillion-dollar enterprise because diets work in the short term, but thepopulation of potential customers is never seriously diminishedby the diets themselves.

Alternative approaches: working with the lipostatrather than against it

If leptin works in leptin-deficient subjects (127), then couldit be used in other obese people to trick their brains intothinking they are even fatter than they are? After all, leptinadministration to mice that had no genetic problems in

producing leptin caused weight loss (91), and individuals thatare heterozygous for the loss-of-function mutation respond asexpected from their lower leptin levels (128). Moreover,transgenic mice overexpressing leptin almost completely losetheir body fat (97), and appetite loss in subjects at high altitudeappears to be linked to elevated leptin concentrations (160).Clinical trials with leptin started in the late 1990s. The largesttrial (161) included ;75 obese patients who were given dailyleptin injections as different doses. The subjects were alsoprescribed a diet that over the 6-mo study period was expectedto result in weight loss of;12 kg. The study revealed thatsubjects on placebo treatment lost on average only 1.4 kg,whereas those on the highest doses of leptin (0.3 mg/kg) lostfive times as much (7.1 kg). Although this modest weight loss is

lower than anticipated (12 kg) given the prescribed diet, it isclear that the people on leptin treatment found it easier toadhere to the dietary prescription. The major problem,however, is the dose of leptin required to obtain this resultinvolved an injection of;30 mg of leptin each day (161). Withthe present (2004) price of recombinant leptin of approximately$100/mg, this treatment would cost $3,000/d. This is unlikelyto be the sort of treatment that could be envisaged for largenumbers of patients in general therapy, because it would costmore than treating the health consequences of the obesity. Inaddition, ;15% of the patients on the trial receiving leptinexhibited moderate to severe allergic reactions at the injectionsites.

Although leptin has not provided the solution it once

promised, it does provide some reassurance that the system inhumans is not radically different to that in rodents. Using


byonOctober8,2008

jn.nutrition.org

Downloadedfrom


11/16

animal models, therefore, is the best hope we have of findinggeneral solutions to the obesity problem. If leptin production isnot the problem, then other possibilities are that the leptin failsto cross the blood-brain barrier (162) or the signal is not readproperly by the receptor system (both contributing to so-calledleptin resistance). In the last couple of years, there has beenenormous interest in other signals from the periphery that mayplay a role in signaling energy status to the brain and may

therefore be putative targets for drug development. The twomajor ones are ghrelin, which is a growth hormone secreta-gogue produced by the stomach, and PYY3-36, which is alsoa gut-derived signaling compound. The effects of thesecompounds remain controversial, and a vigorous debate ispresently occurring over their effects on body-weight control.

The future

Although we have seen spectacular advances in ourunderstanding of the molecular basis of body-weight regulationduring the decade since the discovery of leptin, and within thenext decade we are likely to see drugs on the market that workwith this system rather than against it to produce weight loss,

there is still a great deal that we do not understand. Threesources of information serve to quantify the extent of ourpresent ignorance about the control system and its interrela-tionship with obesity, as follows: 1) genetic mapping studies,which identify the specific regions of the genome wherepolymorphic loci are associated with variation in body weight;2) actions of drugs that produce effects on body weight forwhich the mechanism is unclear including drugs that haveweight gain as an unexpected side effect; and 3) examination ofthe lipostatic model itself, so we can see areas where ourunderstanding is deficient.

Genetic mapping studies aim to identify quantitative traitloci (QTL) where important genes are located that influencebody weight. QTL mapping involves studying the genomes of

families where there is divergence in the BMIs of the twoparents. When such individuals produce offspring, thesechildren inherit half their genetic material from their fatherand half from their mother. Each of these contributions isa mixture of the DNA that the parents inherited from theirparents. By using genetic markers along the genome, it ispossible to trace which bits of chromosome have been inheritedby which children and associate this with the development ofobesity in those children. Across many such pedigrees, it ispossible to identify regions of the genome where variationsgenerate variability in BMI. These regions are known as QTL.Chagnon et al. (163) summarized the QTL-mapping studiesperformed to date and identified 68 different QTL regions ofthe human genome where significant associations were found

between body fatness (normally BMI) and polymorphicvariation. These are spread across all chromosomes exceptthe Y chromosome. Seven of these loci were confirmed inmultiple (between two and five) studies. Interestingly, few ofthese map closely to candidate genes for compounds presentlyknown to be component parts of the lipostatic system. That wedo not know the candidate genes at these major loci is a strongmessage that our understanding of this system and the mannerin which it works is still rather rudimentary.

The second line of evidence concerning our ignorancederives from drug effects on appetite and energy balance, andthat we do not fully understand how they work relative to thelipostat system. First, one of the two pharmaceuticals licensedfor the treatment of obesity falls into this class. Sibutramine is

a 5-HTreuptake inhibitor that produces significant weightloss. There is an indication that the actions of leptin and

serotonin interact (164), but as yet this is poorly understood.Second, a phenomenon that is reported by marijuana users isthat smoking the drug stimulates appetite. The discovery ofendogenous cannabinoid receptors (CB-1), which are putativedrug targets, has led to the development of agonist andantagonist drugs that bind to these receptors. There are alreadyCB-1 antagonist drugs that appear to generate significantreductions in appetite and weight loss (165). The details of how

they work, however, beyond interacting with the CB-1receptor, is unclear. Finally, there is a whole class of drugsthat have weight gain as a side effect of their use. These includemany of the antipsychotic treatments for schizophrenia such asclozapine and olanzapine (166168). Significant weight gain onthese drugs appears to include a combination of effects onexpenditure and intake (166). The drugs are known to beactive at many types of receptors in the brain includinghistamine, dopamine, and 5-HT receptors (169). Their mode ofaction relative to the established network downstream fromleptin is unclear.

If we consider the model of the lipostatic system, it is evidentthat there are large blanks in our understanding of how thissystem actually works. We know some of the signals that

emanate from the periphery to indicate energy status, but wedont know how many other signals there might be and thetimescales over which they are important. How partitioning ofenergy (the p ratio) is controlled also remains unclear. Therehas been considerable debate about the nature of transportprocesses for the peripheral signals across the blood-brainbarrier (162) and whether defects in these systems interferewith efficient signaling, but little progress has been made alongthese lines. Intracellular signaling once the main compoundshit their target receptors has been actively studied, andconsiderable progress has been made in understanding howleptin and insulin modulate gene-expression profiles in theirtarget neurons. Initial events downstream from leptin arestarting to become clear. However, the interactions beyond the

MC4-R in neurons within the PVN are far less understood.Generally published models have arrows that emerge from thispoint to feeding behavior and energy expenditure, but thedetailed mechanics of how feeding behavior and expenditureare controlled are vague. As our understanding of this areablossoms, there will likely be a whole range of new targets towhich pharmaceutical intervention may be possible. In theeffector system, the molecular basis of variability in restingmetabolism (170) and the way that activity levels are regulatedremain virtually unknown.

One area that we still know very little about is the nature ofthe encoding of the lipostatic target weight. Several small-animal models appear to be particularly suitable for investigat-ing this phenomenon including animals that alter their body

weight in response to changing seasons (171,172). We do knowthat these changes are signaled by altered photoperiod: byswitching the photoperiod, it is possible to move animals fromone level of regulation to another. An example of such a switchoccurs in the bank vole Clethrionomys glareolus (173). Whenbank voles maintained in cold conditions are exposed to a shortphotoperiod, they have a low body mass and relatively little fatstorage. If the photoperiod is altered to yield a long day (16 h oflight), the animals put on weight for a period of;2 wk untilthey are;10% heavier (Fig. 3A). Most of this weight gain is fattissue. They then reestablish control at this higher level. Thispattern of change appears to involve a shift in the target of thelipostatic system.

By extracting RNA from the hypothalami of short- and long-

day bank voles, we can compare the gene-expression profiles ofthe major genes known to be involved in the lipostatic system.

2100S SUPPLEMENT

byonOctober8,2008

jn.nutrition.org

Downloadedfrom


12/16

The results of this comparison are illustrated in Figure 3B. Thiscomparison shows that some genes are differentially regulated,in particular, the CART gene is upregulated in the long-daygroup (173). This potentially indicates that CART has someinvolvement in the lipostatic regulatory set point. Anothersmall-animal model system that is extensively characterizedwith respect to responses to photoperiod is the Siberian hamsterPhodopus sungorus. On exposure to short photoperiods,hamsters lose body weight and reestablish control at a muchlower level where ;30% of the body weight has been lost (Fig.4A). Gene-expression profiles that compare long- and short-photoperiod hamsters are also shown in Figure 4B (172). Some

parallel changes in the patterns for bank voles and hamsters areapparent. In particular, the upregulation of CART is observed.Interestingly CART was recognized in candidate-genescreenings in humans as a potential candidate for influencingobesity (163), although other studies of polymorphism linkagesbetween the CART gene and obesity suggest that the effectsare mostly on regional fat distribution (163a). These studies arein their infancy, but the potential is clear. Once we know whichgenes are involved in setting the lipostat target, we can start tothink about manipulating the system. These manipulationscould include production of agonists and antagonists to thegene products and their receptors to turn someone with a hightarget weight into someone with a low target weight.

How far in the future is this dream of manipulating body

weight by target resetting? Inevitably, progress depends onmaking key breakthroughs in gene identification and control

and the development of suitable pharmaceuticals. Because drugevaluation and testing generally takes about a decade, thiscould become a reality in 1520 y. Drugs that mimic or interferewith other components of the lipostatic system are already inphases 2 and 3 of development and should be available tohealth care practitioners within the next 510 y. Meanwhile,the main alternative strategy is willpower. This does work, butfor a permanent solution, this needs permanent lifetimecommitment and struggle. The evidence suggests that fewpeople are able to sustain this. At present, the only alternativesolution that has almost guaranteed success is forced adherenceto energy restriction by surgical intervention. Surgical in-

tervention in 20% of the population is, however, beyond thecapacity of health care systems to deliver.

Drugs that manipulate the lipostatic system will providea valuable advance in the fight against obesity. Using thesedrugs at the same time as calorie restriction means that patientswill no longer need to perpetually fight against the impetus oftheir regulatory systems. Rather, we will use the system thatcontrols body weight to take it to a level that we want instead ofan arbitrary setting that is based on genetic inheritance. Theeffect of treatment would be gradual, but unlike conventionalapproaches that fight against ones genes, an individual wouldwork with his or her genes, which will make the entire processconsiderably easier. However, it should be clear that whatevermanipulation is envisaged, its success will only last as long as

the manipulation continues. If we administer drugs that lowerthe lipostatic set point, they will only help suppress body fatness

FIGURE 3 Body mass changes after transfer from short (8 h light:16h dark) to long (16 h light:8 h dark) days in the bank vole Clethrionomysglareolus (A). Gene-expression profiles in the hypothalamus for a panelof neuropeptides known to be associated with energy regulation incomparing short to long dayexposed animals (B) (redrawn from originaldata in 173). LD, longday; SD, short day; AgRT, Agouti-related transcript;MC3-A and -D, melanocortin-3 receptors in the arcuate and dorsomedialnuclei, respectively; CRF, corticotrophin releasing factor.

FIGURE 4 Body mass changes after transfer from long to shortdays in the Siberian hamsterPhodopus sungorus (A). Gene-expressionprofiles in the hypothalamus for a panel of neuropeptides that are knownto be associated with energy regulation comparing short to long dayexposed animals (B) (redrawn form original data in 172).


byonOctober8,2008

jn.nutrition.org

Downloadedfrom


13/16

as long as the set point is altered. Once the drugs arediscontinued, the set point will (presumably) readjust to theoriginal level and the system will provide an impetus to regainbody weight. The same is true for all of the drugs that are basedon mimicking peripheral signals. Once the signal has gone, theimpetus to replace it by increasing fatness will return. Althoughthey are valuable, such drugs will need to be lifelong treatmentsif they are to be effective. The ultimate holy grail solution is

to permanently switch the system so that the effects are alsopermanent. Whether this is even possible without some form ofgene therapy (174,175) that may be politically unacceptable ispresently unknown.

Conclusion

Obesity is a serious condition that presently affects ;310million people; an additional 800 million people are overweight.Differences among individuals in their susceptibility to obesityprobably reflect, in large part, their genetic constitutions.Attempts to fight against ones genetic legacy are difficult tosustain, so although most calorie-restricted diets are successfulin the short term, they are unsuccessful in the long term.

Presently research efforts are geared toward identifying andultimately manipulating the control system for body weight sothat solutions can work in harmony with (instead of against)ones genes. In the coming decade, we are likely to see furtherdramatic advances in our understanding of this regulatorysystem and novel drugs developed on the back of this improvedunderstanding.

LITERATURE CITED

1. International Obesity Task Force [Online]. http://www.iotf.org [accessedJanuary 2004].

2. Maire, B., Delpeuch, F., Cornu, A., Tchibindat, F., Simondon, F.,Massamba, J. P., Salem, G. & Chevassus-Agnes, S. (1992) Urbanization and

nutritional transition in sub-Saharan Africa: examples from Congo and Senegal[in French]. Rev. Epidemiol. Sante Publique 40: 252258.

3. Arroyo, P., Loria, A., Fernandez, V., Flegal, K. M., Kuri-Morales, P., Olaiz,G. & Tapia-Conyer, R. (2000) Prevalence of pre-obesity and obesity in urbanadult Mexicans in comparison with other large surveys. Obes. Res. 8: 179185.

4. Dorosty, A. R. & Reilly, J. J. (2000) The effect of using different BMIreference data on obesity prevalence in Iranian children. Obes. Res. 8: D18 (abs.).

5. Kain, J., Uauy, R. & Burrows, R. (2001) Basic determinants of theincrease in obesity prevalence among Chilean children. Obes. Res. 9: 135S (abs.).

6. Likitmaskul, S., Kiattisathavee, P., Chaichanwatanakul, K., Punnakanta,L., Angsusingha, K. & Tuchinda, C. (2003) Increasing prevalence of type 2diabetes mellitus in Thai children and adolescents associated with increasingprevalence of obesity. J. Pediatr. Endocrinol. Metab. 16: 7177.

7. World Health Organization (1998) Obesity: Preventing and Managingthe Global Epidemic. Report 894. World Health Organization, Geneva, Switzer-land.

8. Caterson, I. D. & Gill, T. P. (2002) Obesity: epidemiology and possibleprevention. Best Pract. Res. Clin. Endocrinol. Metab. 16: 595610.

9. Engeland, A., Bjorge, T., Selmer, R. M. & Tverdal, A. (2003) Heightand body mass index in relation to total mortality. Epidemiology 14: 293296.

10. Kopelman, P. G. (2002) Obesity as a medical problem. Nature 404:635643.

11. Chan, J. M., Rimm, E. B., Colditz, G. A., Stampfer, M. J. & Willett, W.C. (1994) Obesity, fat distribution and weight gain as risk factors for clinicaldiabetes in men. Diabetes Care 17: 961969.

12. Quilliot, D., Petit, F. X., Cornette, A. & Ziegler, O. (2002) Do you reallyknow sleep apnea in obese patients? Ann. Endocrinol. (Paris) 63: S38S51.

13. World Health Organization (2000) The Asia-Pacific Perspective: Re-defining Obesity and Its Treatment. World Health Organization, Geneva, Switzer-land.

14. Bourn, J. (2001) Tackling obesity in England. Report HC 220 of theNational Audit Office. ISBN 0102814015. The Stationery Office, London, UK.

15. Banegas, J. R., Lopez-Garcia, E., Gutierrez-Fisac, J. L., Guallar-Castillon,P. & Rodriguez-Artalejo, F. (2003) A simple estimate of mortality attributable toexcess weight in the European Union. Eur. J. Clin. Nutr. 57: 201208.

16. Wolf, A. M. & Colditz, G. A. (1998) Current estimates of the economiccost of obesity in the United States. Obes. Res. 6: 97106.

17. Wolf, A. M. (1998) Impact of obesity on healthcare delivery costs. Am.J. Manag. Care 4: S141S145.

18. Finkelstein, E. A., Fiebelkorn, I. C. & Wang, G. J. (2003) Nationalmedical spending attributable to overweight and obesity: how much, and whospaying? Health Aff. 22: 219226.

19. Levy, E., Levy, P., Lepen, C. & Basdevant, A. (1995) The economiccost of obesitythe French situation. Int. J. Obes. 19: 788792.

20. Schneider, R. (1996) Relevance and cost of obesity in Germany [inGerman]. Ernahrungs-Umschau 43: 369373.

21. Swinburn, B., Ashton, T., Gillespie, J., Cox, B., Menon, A., Simmons, D. &Birkbeck, J. (1997) Health care costs of obesity in New Zealand. Int. J. Obes.21: 891896.

22. Oppert, J. M. & Rolland-Cachera, M. F. (1998) Prevalence, timetrends, and economic cost of obesity [in French]. Med. Sci. (Paris) 14: 939943.

23. Fernandez Real, J. M., Gutierrez, C., Ricart, W., Casamitjana, R.,Fernandez Castaner, M., Vendrell, J., Richart, C. & Soler, J. (1997) The TNF-alpha gene Nco I polymorphism influences the relationship among insulinresistance, percent body fat, and increased serum leptin levels. Diabetes 46:14681472.

24. Gutierrez, C., Ricart, W., Casamitjana, R., Biarnes, J., FernandezCastaner, M., Vendrell, J., Richart, C., Soler, J. & Fernandez Real, J. M.(1997) A TNF-alpha gene polymorphism is related with insulin resistance, percentbody fat and increased leptin. Diabetologia 40: 1198 (abs.).

25. Savage, D. B., Sewter, C. P., Klenk, E. S., Segal, D. G., Vidal-Puig, A.,Considine, R. V. & ORahilly, S. (2001) Resistin/Fizz3 expression in relation toobesity and peroxisome proliferator-activated receptor-gamma action in humans.Diabetes 50: 21992202.

26. Silha, J. V., Krsek, M., Skrha, J. V., Sucharda, P., Nyomba, B. L. G. &Murphy, L. J. (2003) Plasma resistin, adiponectin and leptin levels in lean andobese subjects: correlations with insulin resistance. Eur. J. Endocrinol. 149: 331335.

27. Vidal-Puig, A. & ORahilly, S. (2001) Resistin: a new link betweenobesity and insulin resistance? Clin. Endocrinol. (Oxf.) 55: 437438.

28. Chisholm, D. J., Samaras, K., Markovic, T., Carey, D., Lapsys, N. &Campbell, L. V. (1998) Obesity: genes, glands or gluttony? Reprod. Fertil. Dev.10: 4953.

29. Centers for Disease Control and Prevention [Online]. U. S. Dept. of Healthand Human Services, http://www.cdc.gov [accessed January 2004].

30. Gordon-Larsen, P., Zemel, B. S. & Johnston, F. E. (1997) Secularchanges in stature, weight, fatness, overweight, and obesity in urban AfricanAmerican adolescents from the mid-1950s to the mid-1990s. Am. J. Hum. Biol. 9:675688.

31. Gutierrez-Fisac, J. L., Regidor, E., Banegas, J. R. B. & Artalejo, F. R.(2002) The size of obesity differences associated with educational level in Spain,1987 and 1995/97. J. Epidemiol. Community Health 56: 457460.

32. Ball, K., Mishra, G. D. & Crawford, D. (2003) Social factors andobesity: an investigation of the role of health behaviours. Int. J. Obes. 27: 394403.

33. Walker, A. R. P., Adam, F. & Walker, B. F. (2001) World pandemic ofobesity: the situation in Southern African populations. Public Health 115: 368372.

34. Hernandez, B., Gortmaker, S. L., Colditz, G. A., Peterson, K. E., Laird,N. M. & Parra-Cabrera, S. (1999) Association of obesity with physical activity,television

obesity the integrated roles of environment and genetics

Documents