This post was triggered by my recent post on the role of in vitro and animals experiments in biomedical research. It is offered as another example of how data from animal experiments, if uncritically applied to people, can lead to false beliefs about human biology.
In the late 1970s and 1980s, the notion that defective thermogenesis (heat generation) in a tissue called brown fat was a major cause of obesity became very popular:
“Recent studies on brown adipose tissue have shown that a defect in this tissue is one probable cause of obesity” Jean Himms-Hagen (1979).
Brown fat is a tissue which unlike relatively inert, white storage fat has a high blood flow, a high metabolic rate and is capable of generating considerable amounts of heat mainly by burning fats. This process of heat generation in brown fat is known to be important in enabling small animals like mice and rats to keep warm in a cold environment. Mice clearly cannot significantly reduce their heat losses in the cold by reducing skin blood flow because they are so small; any point within the mouse is just a couple of centimetres or so from the surface. Once they have adapted to a cold environment, rats and mice do not shiver noticeably but they rely on extra heat generation in brown fat (non-shivering thermogenesis) to keep them warm. Brown fat thermogenesis is also an important means for hibernating animals to generate the heat needed to arouse them from their hibernation; during hibernation their body temperature can drop to just above freezing point.
In a very influential paper in Nature published in 1979, Mike Stock and Nancy Rothwell argued that brown fat could also “burn off” surplus calories when animals and perhaps people overate. They called this burning off of surplus energy, diet-induced thermogenesis; the term adaptive thermogenesis has also been used to clearly differentiate it from inevitable increase in energy expenditure caused by digesting, absorbing and assimilating more food. This theory would suggest that some animals and people could remain lean despite overeating because they were able to burn off any surplus food calories in their brown fat; differences in the ability of individuals to do this could explain differences in susceptibility to weight gain and obesity. The impact of this paper was enhanced by the screening at about the same time, of a BBC Horizon programme The Fat in the Fire that showcased Rothwell and Stock’s ideas and presented a very slick and persuasive, but one-sided, account of the supporting evidence.
This adaptive thermogenesis theory also encouraged the idea that drugs that stimulated heat production in brown fat could be used to burn off calories and thus aid in the management of overweight and obesity. It prompted a spike of research activity searching for these so-called thermogenic (heat-generating) drugs. Several novel compounds that specifically switched on brown fat in rats (ADBR3 agonists) were developed by scientists working for Beecham Pharmaceuticals in 1984 but these proved to be ineffective in obese people.
Brown fat had previously been considered vestigial in adult humans because we rely largely upon heat conservation rather than heat generation to keep warm. When extra heat production is needed in people it is largely generated in muscles by shivering. The idea that adult human brown fat is largely vestigial is now once again the predominant scientific belief. There has, more recently, been apparent confirmation that many people do have very small deposits of brown fat that can be activated by cold stress. The potential contribution of this tiny amount of tissue to total body energy expenditure is negligible although the possibility that other fat cells might be artificially triggered to become brown fat cells has been mooted.
Brown fat is thought to have a role in new-born human babies. Heat production in brown fat is more important in babies because their small size makes heat conservation less effective and they have not yet fully developed the ability to shiver.
Hypothermia in genetically obese mice interpreted as a “thermogenic defect”
Observations and experiments on mutant mice played an important part in supporting this theory of defective brown fat thermogenesis as a cause of obesity. One type of mutant mice is referred to as ob/ob; these are mice where both copies of a then unidentified gene termed the ob gene are defective. Another type of mutant is referred to as db/db and although it was clear that a different gene is involved these mice have essentially the same characteristics as the ob/ob mice.
These mutant mice are extremely obese and over 60% of their body weight is fat. They eat more than their lean siblings but they also have lower than normal body temperatures, are very inactive and unlike their lean siblings they are very susceptible to cold; they rapidly die if exposed to a temperature of 4oC (normal fridge temperature). Ordinary, non-mutant mice will adapt to fridge temperature and survive indefinitely if they have food and bedding. Although these mutant mice normally eat much more than their lean siblings, it was found that they still got heavier and fatter even when their food intake was restricted and exactly matched to that of these lean siblings. This suggested that they were more efficient in their use of food energy because they were getting much fatter despite eating exactly the same amount of food. For many years, these mutant mice were classified as having a metabolic obesity caused by more efficient metabolic use of energy as opposed to a regulatory obesity caused by simple overeating. The low body temperature and poor cold tolerance of these mice was generally interpreted at the time as being caused by what was termed a thermogenic defect. In other words, these mice were cold because of something wrong with their heat generating mechanism in brown fat which made them not only susceptible to hypothermia but also metabolically susceptible to obesity because they “wasted” less calories keeping warm. They were hypothermic and cold intolerant because of a perceived inability to generate heat fast enough to maintain normal body temperature. This also meant that if brown fat was involved in adaptive thermogenesis or the burning off of surplus calories, then they would be unable to do this effectively, again making them prone to excess weight gain.
The hypothermia of obese mice as an adaptation to perceived starvation rather than a thermogenic defect
As early as 1982, my research group questioned whether the hypothermia and poor cold tolerance of these genetically obese mice was really due to a failure of thermogenesis. We found that even normal lean mice are not always homeothermic and do not always maintain a standard 37oC body temperature.
When deprived of food, one of the ways that even normal mice can respond is to become torpid i.e. they slow metabolism and decrease their core body temperature. They lower their core temperature to just above ambient temperature with a minimum of just above 20oC. Mice can spend several hours in this torpid state when fasting and this saves them from wasting their energy reserves on trying to keep warm. Maintaining a standard 37oC body temperature is a very energy expensive process for something as small as a mouse. We found that when we fasted some of our laboratory mice for 48h (lean mice from our ob/ob generating colony) and remotely monitored their core body temperature with surgically implanted temperature sensors they underwent two periods of deep torpor. At the end of these periods of torpor they aroused and warmed up spontaneously without any human interference and before their food supply was restored (see figure 1). Lowering of body temperature was therefore a normal adaptive response of these mice to starvation which enabled them to conserve energy when short of food. We found that ob/ob mice were more prone to torpor than lean mice and they had deeper and longer bouts of fasting-induced torpor than their lean siblings. Indeed it could be said that a semi-torpid state of low body temperature and profound inactivity is the usual state of ob/ob mice and sometimes we recorded ob/ob mice going into full torpor even when food was present.
Figure 1* Records of continuous body temperature records in 2 C57Bl/6 mice over a three-day period that included a 48h fast.
*Adapted from Webb, GP (1992) Viewpoint II: Small animals as models for studies on human nutrition. In Nutrition and the consumer. Ed. Walker, AF and Rolls, BA London: Elsevier Applied Science. Pp. 279-297.
It had been suggested in the 1950s that an undiscovered “satiety hormone” perhaps produced by fat tissue might be involved in long-term regulation of body weight. According to this so-called lipostat theory, when fat stores in adipose tissue were plentiful then this hypothetical hormone would be released in large quantities and act on a part of the brain called the hypothalamus to reduce appetite and food intake. This hypothetical satiety hormone was identified in 1994 and is now called leptin. If fat stores became depleted then less leptin would be released, the hypothalamus would register this and cause the mice to be hungrier and eat more to enable them to replenish their fat stores. As the brain (hypothalamus) was monitoring the amount of fat stored in the body then it gave a plausible mechanism for how body weight could be regulated in the long-term:
- Too little fat – less leptin released from adipose tissue – hypothalamus responds by increasing hunger and food intake
- Too much fat – more leptin released – hypothalamus responds by decreasing hunger and food intake.
It is otherwise difficult to explain how long-term weight stability could be achieved if appetite is solely controlled by short-term responses to things like amounts of food in the gut or changes in blood glucose level (the glucostat theory).
An ingenious, although perhaps rather grisly, series of experiments published in the 1970s , had suggested that perhaps ob/ob mice failed to produce this still theoretical leptin and that db/db mice with very similar characteristics did not respond to leptin. The db/db might be unresponsive because of a defect in the gene for the leptin receptor and without this receptor they would be unable to respond to it. Thus ob/ob mice got fat because they could not make leptin and db/db mice because they could not respond to it. These 1970s experiments involved a technique called parabiosis where mice are surgically joined together so that they share compounds present in each other’s blood. For example, when an ob/ob mouse was paired with a db/db mouse then the ob/ob mouse lost a lot of weight. This was explained by suggesting that the db/db partner of the pair was producing lots of leptin because it had large fat stores and could make the hormone but not respond to it. When this surplus leptin enters the blood of the ob/ob partner it reduced appetite and caused weight loss; although ob/ob mice cannot produce leptin they are very sensitive to it.
We suggested, way back in 1982 that if this lipostat theory were correct then because they produced no leptin, the brains of ob/ob mice would always respond as if the mice had no fat stored in their adipose tissue and normal starvation responses would be permanently switched on i.e. continuous hunger and overeating, reduced activity, and reduced energy expended on heat production and a tendency to become fully torpid without food deprivation. The low body temperature and permanent semi-torpid state of these mice would be the animal’s normal response to falsely perceived starvation. Thus the hypothermia and cold sensitivity of the ob/ob mouse would represent not a thermogenic defect but an adaptation to its perceived state of permanent starvation and lack of fat stores resulting in the brown fat being “switched off” to save energy. Lack of use would make the brown fat of these ob/ob mice become atrophied and so that they might be unable to respond to sudden cold exposure and so die of hypothermia when exposed to fridge temperature. Of course, in 1982 leptin was still only a theoretical hormone as was the notion that ob/ob mice could not make it and db/db mice were unresponsive to it.
The discovery of leptin confirms our explanation
In 1994, a protein hormone produced almost exclusively in adipose tissue was identified in both mice and in people; and the name leptin has been adopted. It was confirmed that ob/ob mice have a mutation that means they fail to produce this hormone in active form and that db/db mice have a mutation in the gene for its receptor and so they produce it but cannot respond to it. Later studies found rare examples of people who like ob/ob mice could not produce active leptin and people who like the db/db mice had receptor defects that made them unresponsive to it. These people are very obese and eat voraciously but their thermoregulation is normal. This indicates that, as we had suggested, the hypothermia observed in ob/ob and db/db mice was a peculiarly mouse way of responding to perceived starvation. People do not become torpid when fasted and so their thermoregulation is not obviously affected by lack of leptin. The notion that the primary genetic defect in these obese mice was in their brown fat has now clearly been shown to be false.
The human equivalents of ob/ob and db/db mice were initially found amongst close-knit immigrant communities in Britain and France where marriage within the group and between related individuals, like cousins, was the norm. In such communities rare diseases caused by recessive genes are more likely to show up as the full disease. Related individuals are much more likely to both have any defective gene and thus their offspring are more likely to receive two copies of the defective gene i.e.one from each parent (most societies forbid marriage between first degree relatives e.g. siblings or parents and their children). This also suggests that these abnormal genes are rare in humans and only responsible for a tiny fraction of cases of human obesity.
The discovery of leptin has not proved to be the expected great leap forward in obesity management that persuaded a drug company to pay millions of dollars for the patent on the leptin gene. Most obese people already have high blood leptin levels because they have high fat stores and injection of still more leptin is ineffective in causing significant weight loss. Most obese people seem to have become less sensitive to leptin just as type 2 (adult onset) diabetics become less sensitive to insulin. Leptin is effective in treating those few individuals who because of a genetic problem cannot produce their own leptin (like ob/ob mice) but of course it has no effect on those people who have a defective leptin receptor (like db/db mice). Leptin has to be administered by injection because it is a protein that would be digested if given orally.
Flawed interpretation of animal experiments helped drive this brown fat theory of human obesity
This brown fat theory of obesity seems thus to be based on very human-orientated (anthropocentric) interpretation of animal experiments and their application to a human problem. These leptin deficient mice are not hypothermic and cold intolerant because of a primary failure of their heat generating system (brown fat). Rather they are hypothermic because their brains have switched off their brown fat to save energy because of what the brain, in the absence of leptin signals, wrongly perceives as a permanent state of body fat depletion or starvation. Torpor is not a usual human response to food shortage, real or perceived, and so hypothermia is not seen in leptin deficient people.
It is sometimes easy with the benefit of hindsight to criticise scientific interpretations and see the reason for past mistakes (e.g. in the protein gap case-study) but in this case I can claim to have been “wise before the event”. I suggested in print several times prior to the discovery of leptin in 1994 that the general “thermogenic defect” interpretation of the hypothermia seen in ob/ob mice might be wrong and thus to imply that people with a similar gene defect to ob/ob or db/db mice would not exhibit any abnormalities in their body temperature regulation i.e. they would not suffer from permanent hypothermia.
A footnote about cafeteria-feeding
Although observations upon genetically obese mice played a pivotal role in the development of their brown fat theory of obesity, Rothwell and Stock did most of their own experimental work on rats. They used a procedure called cafeteria-feeding to encourage their rats to overeat. This involves offering rats a wide variety of appealing, high-calorie human foods in addition to their normal pellet diet. It had been previously used as a reversible experimental obesity because animals return to normal weight when cafeteria-feeding is stopped. In their Nature paper, Rothwell and Stock suggest that considerable variability had been reported in the extent of the obesity caused by cafeteria-feeding which had often been attributed to failure to cause overeating. However, in some experiments where food intake was monitored, they suggested that animals did not gain much weight despite considerable apparent overeating and they suggested that this might be due to increased heat production i.e. adaptive thermogenesis or as they termed it diet-induced thermogenesis.
The key experiment in their Nature paper involved a comparison of 6 rats fed on a cafeteria-diet for 21 days with 6 control rats just fed standard pellet diet. They reported that despite getting heavier and fatter than the controls, the gain in carcass energy was much less than would be expected from the extra calories consumed and they proposed that many of these extra calories had been “burnt off” by diet-induced thermogenesis in brown fat. They also gave other data showing that cafeteria-fed animals had: higher metabolic rate than controls in short sampling periods, increased metabolic rate in response to a drug known to switch on brown fat and that their brown fat was heavier than in controls and in some ways resembled that of cold-acclimatised animals. They also gave some very preliminary and indirect evidence for the presence of active brown fat in two human volunteers.
It must be said that for a major Nature paper that had such an impact, there were distinct limitations on the amount of data presented (just 6 test animals in the key experiment) and some understandable limitations in the technical rigour of the experiments described. Rothwell and Stock consistently argued that the potential likely errors from these limitations were very small in relation the large amount of food energy that seemed to be lost by diet-induced or adaptive thermogenesis. Did the referees overlook some of the paper’s technical limitations because of the originality of the very clever and exciting new theory that seemed to link together and explain findings from several different areas of research? If it had proven to be correct this theory could have had a major practical impact on strategies used to investigate and manage the huge problem of human obesity.
Professor G Romaine Hervey and Graham Tobin then based at the University of Leeds were persistent critics of Rothwell and Stock’s theory of diet-induced thermogenesis. They argued that a number of problems with their technical methods and assumptions could provide alternative explanations of the apparent adaptive thermogenesis reported by Rothwell and Stock. Hervey and Tobin used a much more controlled but arguably “less natural” way of overfeeding rats; they fed them by stomach tube. When they overfed rats in this way they could account for essentially all of the surplus energy supplied to the rats without the need for any adaptive thermogenesis:
- The extra energy, mainly extra fat. stored in the rats’ bodies
- The energy cost of digesting, absorbing and assimilating the extra food which they calculated from measurements of energy expenditure in the immediate postprandial period
- The energy costs of synthesising the extra body fat
- The cost of maintaining and moving a heavier body.
They could also similarly account for essentially all of the extra energy consumed by cafeteria-fed rats.
They argued that Rothwell and Stock had not fully allowed for all of these predictable extra energy costs of overfeeding. They also suggested that Rothwell and Stock may have overestimated the extent of the extra calorie intake of cafeteria-fed rats because of things like:
- Difficulties in fully accounting for greasy spilled food
- Not fully allowing for water loss from spilled, recovered food
- The use of standard food tables to estimate the energy yield of cafeteria foods
- Reliance on manufacturers data on the energy yield of the pellet diet.
They also suggested that the method of estimating how much extra energy was stored during the cafeteria-feeding period was prone to considerable potential error. Rothwell and Stock killed and analysed six matched, control animals at the start of the cafeteria-feeding period and then used this to generate baseline values to calculate changes in the carcass composition of those rats actually used in the feeding experiment.
Hervey and Tobin claimed that whether using tube-feeding (40-50%) or cafeteria-feed (20-40%) the extra energy expenditure induced by overeating was consistent in different strains and ages of rats whereas Rothwell and Stock suggested that the degree of adaptive thermogenesis was affected by age and strain and even varied between animals of the same strain from different suppliers. This variable response they argued could be used to explain differences in susceptibility to obesity. It could also explain conflicting results in different labs or between different experiments.
For several years there was an ongoing and sometimes acrimonious debate between the pro diet-induced thermogenesis school of Rothwell and Stock and the anti-school of Hervey and Tobin. In 1983, the two pairs of opposing protagonists were invited by the journal Clinical Science to present their opposing views. Hervey and Tobin produced an extensive, detailed and closely argued 12 page account of their case against diet-induced thermogenesis. Rothwell and Stock give a much shorter 4 page statement of their case which in my opinion tends to dismiss the opposing arguments rather than give a detailed point by point rebuttal of the criticisms. An obituary for Mike Stock written by Professor Paul Trayhurn alludes to the hostility of this debate.
I gratefully acknowledgement the intellectual and practical contribution of my friend and ex-colleague, Dr Mike Jakobson to my earlier work on torpor and obesity.