Appetite Regulatory Peptides: Obesity drugs and their targets: correlation of mouse knockout phenotypes with drug effects in vivo
Obesity Reviews, Volume 7 Page 89 - February 2006, D. R. Powell
Sequencing of the human genome has yielded thousands of potential drug targets. The difficulty now is in determining which targets have real therapeutic value and should be the focus of a drug discovery effort. The available evidence suggests that knockout technology can be used prospectively to identify targets that are amenable to drug development for the treatment of a variety of diseases. This review compares the knockout phenotypes of 21 potential obesity targets with the effects of therapeutics designed for those targets on rodents and, when data were available, on humans. The phenotypes of obesity target knockouts model the effects seen when therapeutics designed for those obesity targets are delivered to rodents; of the 21 obesity targets reviewed, 16 showed a correspondence between knockout phenotype and drug effect in mice and/or rats. This suggests that, at least in terms of evaluating obesity targets, it is rare for compensatory developmental changes caused by the gene knockout to prevent detection of the relevant phenotype. In the majority of cases, the knockout phenotypes also modelled the effects seen when the relevant therapeutics were delivered to humans. Thus, it seems rational to use mouse knockout technology prospectively to identify genes that regulate body fat in vivo, and then to develop anti-obesity therapeutics by targeting the human protein products of these genes. Ultimately, the value of using this approach to identify novel targets for human anti-obesity therapies will be judged by future studies examining the anti-obesity effect, in humans, of the therapeutics that result from this approach.
Sequencing of the human genome has yielded thousands of potential drug targets. The difficulty now is in determining which targets have real therapeutic value and should be the focus of a drug discovery effort. There is a general consensus in the biomedical community that mouse knockout technology can provide powerful insight into the physiological function of mammalian genes (1). Strong support for this consensus comes from retrospective analyses which reveal that mouse knockout phenotypes for the targets of drugs correlate well with drug effects (2,3). This suggests that knockout technology can be used prospectively to identify targets, and therapeutics, for the treatment of human obesity. Ideally, the process will proceed as follows: (i) identify an obese or lean phenotype in a knockout mouse model; (ii) develop a therapeutic that specifically and appropriately targets the protein product of this gene; (iii) use that therapeutic to recapitulate the desirable phenotype in mice and other species; and (iv) use that therapeutic to recapitulate the desirable phenotype in humans.
There are two major assumptions that must be true for obesity phenotypes from mouse knockout lines to accurately predict the effects, in humans, of drugs that modulate activity of the protein product of the targeted gene: (i) compensatory effects occurring during development of knockout mice rarely alter their obesity phenotype later in life. In other words, the obesity phenotype of the knockout mice should correlate with the effects, in mice, of drugs directed against the targeted gene; and (ii) the genes that regulate body composition in mice must perform very similar functions in humans. In other words, anti-obesity drugs should have similar effects in mice and humans. This review explores the validity of these assumptions by comparing the knockout phenotypes of 21 potential obesity targets with the effects of therapeutics designed for those targets. A potential obesity target was included only if it met two criteria: (i) a generalized rather than tissue-specific knockout of the target must have been analysed for the presence of an obesity phenotype. The focus is on generalized gene knockouts because almost all currently available anti-obesity drugs modulate their protein target in all tissues where the protein is expressed, not just in specific selected tissues; and (ii) the effect of a drug designed for the target must have been tested in at least one rodent model of obesity. In most cases, differences in body weight and/or body fat constituted the knockout phenotype and the therapeutic effect of interest, but in a few instances the focus was on differences in food intake. Where possible, the effects of these therapeutics on both knockout mice and littermate controls are presented, because knockout models have shown utility in determining which effects of therapeutics are mediated through the drug target, and which are off-target effects (2,3). When possible, the effect of each therapeutic in humans is also presented and this effect is contrasted with both the knockout phenotype and the effect of that therapeutic, or a closely related therapeutic, in rodents. These data are summarized in Table 1.
The targets reviewed here are organized according to the druggable gene families to which they belong. This is to emphasize that, of the 20 000 to 25 000 genes in the human or mouse genomes, only 5000 are considered viable drug targets, or druggable, by current pharmaceutical industry standards. Almost all of these targets can be categorized as either G-protein coupled receptors (GPCRs), GPCR ligands, nuclear hormone receptors, enzymes, secreted proteins, or membrane-bound cell-surface proteins (2,3).
G-protein coupled receptors and their ligands
5-HT2c serotonin receptor
Brain serotonin levels are thought to play an important role in the neural regulation of appetite. 5-HT2c serotonin receptor –/– mice have increased body weight and fat depot weight entirely because of increased food intake, suggesting that postsynaptic 5-HT2c serotonin receptors confer the anorectic effect of serotonin (4,5). Chow-fed 5-HT2c serotonin receptor –/– mice weigh the same as +/+ littermate controls at 17 weeks of age, but weigh 29% more than these +/+ controls at 33 weeks of age. In contrast, –/– mice fed a high-fat diet (HFD) weigh 34% more than +/+ littermate controls at 17 weeks of age (5). Consistent with these findings, both m-chlorophenylpiperazine (mCPP), a high affinity 5-HT2c receptor agonist, and d-fenfluramine (d-fen), a compound that stimulates serotonin release while inhibiting serotonin reuptake into nerve terminals, acutely suppress food intake in littermate control mice. Importantly, mCPP has no effect and d-fen has only a slight inhibitory effect on food intake in 5-HT2c knockout mice, direct evidence that the full effect of these compounds requires this receptor (4,6). Furthermore, long-term treatment of mice with either fenfluramine or d-fen significantly lowers body weight, an effect that was partially, but not entirely, the result of depletion of fat stores (7–9).
Although no longer on the market, fenfluramine and d-fen have been used in the past to treat obesity. Obese patients treated with 20-mg fenfluramine three times a day for 20 weeks lost 8.4% of their body weight, significantly more than the 4.9% weight loss measured in the placebo control group (10). In 16 additional studies, patients receiving d-fen averaged 3.7% more weight loss than did those receiving placebo (11). Thus, mice and humans appear to respond in a similar manner to fenfluramine therapy.
Melanocyte-stimulating hormone (MSH)
Pro-opiomelanocortin (POMC)-derived peptides include alpha melanocyte-stimulating hormone (αMSH), a peptide that acts centrally to decrease appetite and increase energy expenditure. Mutant mice lacking all POMC-derived peptides display obesity, adrenal insufficiency and altered pigmentation, a phenotype quite similar to that of humans with POMC deficiency (12,13). The POMC –/– mice eat more than +/+ littermate controls so that by 4 months of age, chow-fed male –/– mice weighed 80% more than +/+ mice (12). Intraperitoneal injections of the stable αMSH agonist [Ac-Cys4, D-Phe7, Cys10]α-MSH4-13 given daily for 14 days led to a 46% loss of excess weight in –/– mice but did not significantly affect the weight of +/+ littermates (12). This suggests that αMSH agonists may be effective in treating the obesity syndrome present in POMC-deficient humans (13). A role for these agonists in other forms of human obesity remains to be established.
Melanocortin-4 receptor (MC4R)
Alpha melanocyte-stimulating hormone (αMSH) is considered to inhibit appetite and increase energy expenditure by activating specific central melanocortin receptors, particularly melanocortin-4 receptor (MC4R). Targeted disruption of the MC4R gene produces –/– mice that eat more than +/+ controls; in some but not all studies, –/– mice also expend less energy (14–16). When these mice are maintained on a chow diet for up to 37 weeks, MC4R –/– mice weigh at least 50% more than +/+ littermate controls, with the difference primarily because of increased fat as demonstrated by a 3- to 5-fold increase in fat pad weight of –/– mice (14,15). Characterization of the mouse MC4R –/– obesity phenotype preceded the recognition that MC4R deficiency in humans results in a similar phenotype and is the most common form of human monogenic obesity, accounting for 4% of obese patients with body mass index greater than 35 (17,18). In rats, central infusion of Melanotan II (MTII), a synthetic αMSH analogue with increased potency, led to an acute decrease in food intake and body weight. Although food intake returned to normal after 2 weeks, body weight remained low throughout the 4-week study (19). MTII appears to work primarily through the MC4R, because acute central infusion significantly inhibits food intake and body weight regain in fasted littermate control mice but not in MC4R –/– mice (20). This is supported by recent data showing that central delivery of the highly selective MC4R agonist c(1–6)suc-fRWK-NH2 acutely decreases food intake in a dose-dependent fashion (21).
Recent studies in rats suggest that, regardless of the route of administration, MTII decreases food intake by producing a strong conditioned taste aversion (CTA) (22). In general, drugs that induce vomiting in humans and other emetic species will induce a CTA in non-emetic species such as the rat (23); essentially, the drugs are causing a visceral illness that leads to avoidance behaviour. These findings are consistent with previous reports of MTII administration to humans (24–26); in one short-term, small-scale human trial, for example, nausea was reported significantly more often after administration of MTII (0.025 mg kg1 given twice subcutaneously for erectile dysfunction), but decreased appetite was not associated with MTII dosing (26). When combined with the lack of response of MC4R –/– mice to MTII, the above data suggest that, despite the central role of the MC4R in regulating body fat stores in humans and rodents, even highly selective MC4R agonists may have on-target side effects that hinder their usefulness as anti-obesity agents in at least some patients.
Cannabinoid receptor type 1 (CB1)
Cannabinoid receptor type 1 (CB1) and associated endogenous ligands, the endocannabinoids, participate in the central regulation of food intake. Two studies have examined the effects of targeted inactivation of CB1 on energy balance. In the first study, chow-fed CB1 –/– mice showed a 9% decrease in body weight, a 20% decrease in body fat, and a 3% increase in lean body mass relative to +/+ littermates at 16 weeks of age (27). In young mice, the lean phenotype was entirely the result of decreased food intake while in adult mice a metabolic component was contributory (27). In a more recent study, chow-fed CB1 –/– mice showed a 23% decrease in body weight, a 60% decrease in body fat, and a 13% decrease in total body protein relative to +/+ littermates at 20 weeks of age (28). In these –/– mice, evidence suggested that the lean phenotype was the result of both decreased food intake and increased energy expenditure (28).
Rimonabant (SR141716), a specific CB1 antagonist, decreased body weight by 25%, body fat by 60%, and body protein by 9% when given daily for 5 weeks to mice with established diet-induced obesity (DIO); pair-feeding studies showed that the weight loss was primarily, but not entirely, the result of decreased food intake (29). In two additional studies using DIO mice (28,30), rimonabant decreased body weight, fat pad weight and food intake; of interest, the weight loss induced by rimonabant was comparable with the weight loss observed when the DIO mice were switched to chow diet (30). Importantly, rimonabant had no effect on CB1 –/– mice in two of these studies (28,29), confirming that rimonabant acts by specifically inhibiting the CB1 receptor. Rimonabant also decreases food intake and body weight gain in both obese Zucker rats and lean Zucker controls; the results are consistent with the findings in DIO mice except that weight loss was entirely the result of decreased food intake (31). Further studies using AM251, a second CB1 antagonist, reported similar results; AM251 treatment decreased body and fat pad weight in DIO mice, and inhibited food intake in DIO mice and in rats (32–34). Of interest, AM251 appeared to decrease food intake in rats by producing a CTA, suggesting that food intake was decreased, at least in part, by inducing nausea (34).
Recently, a 1-year randomized, double blind clinical trial evaluated the effects of rimonabant on weight loss in 1507 overweight humans (35). Consistent with the rodent data, rimonabant induced a dose-dependent weight loss which was significantly greater than that noted in the placebo group; patients treated once a day with 20-mg rimonabant lost an average of 5 kg more than those receiving placebo. Rimonabant was, in general, well-tolerated; although treatment was associated with a dose-dependent increase in the number of patients reporting nausea, consistent with the rat CTA data, this side effect was mild, transient, and reported by less than 13% of patients receiving 20 mg d1.
Melanin-concentrating hormone-1 receptor (MCH1R)
Melanin-concentrating hormone (MCH) is an orexigenic peptide that plays a role in the central regulation of energy balance and body weight (36,37). In rodents, the effects of MCH appear to be conferred primarily through the melanin-concentrating hormone-1 receptor (MCH1R). Targeted inactivation of either MCH (36) or MCH1R (37) results in mice with decreased weight, decreased fat and increased metabolic rate. On a chow diet, MCH –/– mice weigh 25% less than +/+ littermates because of an almost 50% decrease in carcass fat (36), while MCH1R –/– mice weigh 12–15% less than +/+ littermates because of a 30–50% decrease in body fat (37). The major discrepancy between the two knockout models is food intake, which is reportedly decreased in MCH knockout mice but increased in MCH1R knockout mice. DIO rats treated for 4 weeks with SNAP-7941, a specific MCH1R antagonist, ate less and lost 26% of their body weight; additional studies suggested that the decreased food intake was not the result of a CTA (38). The SNAP-7941 findings are most consistent with the MCH phenotype, but it would be useful to determine the effects of this compound on food intake of MCH and MCH1R –/– mice. The data to date suggest that MCH1R antagonists may be useful in the treatment of obesity, but there is no information currently available on the ability of these compounds to induce weight loss in humans.
β3 adrenergic receptor (β3AR)
The development of thermogenic drugs to treat obesity has focused on adrenergic agonists because sympathetic stimulation has long been known to increase metabolic rate (39). Recent work has focused on selective β3 adrenergic receptor (β3AR) agonists because β3ARs are highly expressed in fat where they play a major role in thermogenesis, and also to avoid unwanted side effects caused by even weak stimulation of β1 and β2ARs (40–42). Evidence for the crucial role of the β3AR in body fat regulation is provided in part by studies of β3AR knockout mice. Male β3AR –/– mice showed a 35–40% increase in body fat on chow diet (43,44), and showed increased food intake associated with a 55% increase in body fat and a 10% decrease in fat-free dry mass when challenged with HFD (43); in no case was there a significant difference in body weight (43,44). In general, these data are consistent with results obtained using the β3AR-specific agonist CL-316243 in rodents. First of all, CL-316243 increases energy expenditure of littermate control mice but not β3AR –/– mice, indicating that the metabolic effects of this agonist are conferred through the β3AR (43,44). Rats treated for 2–4 weeks with CL-316243 while on chow or HFD showed modest decreases in body fat and fat depot weight, increased energy expenditure and brown adipose tissue (BAT) weight, and little change in body weight or food intake (45–47). In contrast, a single long-term study found that both control and fatty Otsuka Long-Evans Tokushima rats responded to 14 weeks of oral CL-316243 treatment with 15% and 27% decreases in body weight respectively; treated rats also exhibited marked decreases in the weights of individual fat pads, including interscapular BAT, but treatment had no effect on food intake (48). ICR and FVB mice treated for 2–4 weeks with CL-316243 showed 0–15% loss in body weight associated with a significant decrease in body fat and increase in energy expenditure (49,50). Interestingly, CL-316243 prevents excess weight gain in HFD-challenged A/J mice but not in C57BL/6J mice; in these C57BL/6J mice, HFD challenge was associated with decreased β3AR expression and function in adipocytes (51). This is consistent with the observation that the severity of obesity in a number of genetically obese rodent lines correlates with the decrease in β1AR and β3AR expression (reviewed in 40). In summary, β3AR activation in rodents results in a variable depletion of fat stores, with the extent of response depending in part on genetic background.
Selective β3AR agonists are also being developed for human use. To date, the agonist with the best selectivity and efficacy data in humans is L-796568. Although a 250-mg dose of L-796568 had no effect, a 1000-mg dose significantly increased energy expenditure by 8% over a 4-h period in obese men. In the same study, 1000 mg also significantly increased lipolysis and systolic blood pressure but resulted in no other changes consistent with β2AR stimulation (52). Treatment of obese men with 375 mg d1 of L-796568 for 28 d did not increase energy expenditure or lipolysis, but higher plasma levels of L-796568 correlated with greater decreases in fat mass (53). The authors conclude that lack of a significant chronic effect on energy expenditure and weight loss may be the result of: (i) down-regulation of β3AR by L-796568, and/or (ii) insufficient recruitment of BAT in humans. In addition, the prior study by these authors suggests a third possibility that a higher dose of this or a more selective agonist may be required to increase energy expenditure. In summary, there is some evidence that rodents and humans show a similar increase in energy expenditure in response to β3AR agonists, but additional study is required to determine if weight loss can be achieved in humans by developing more selective compounds and/or by perhaps focusing treatment on a subset of obese patients who have higher levels of adipocyte β3AR receptors during chronic therapy.
Histamine H1 receptor (H1R)
The central histaminergic system is involved in the regulation of feeding behaviour and body composition (9,54–60). Histamine can interact with four different receptor sub-types, but it is the histamine receptors H1R and H3R that are implicated in feeding and appetite regulation (54). The strongest evidence suggests that hypothalamic histaminergic neurones confer their effects on appetite through postsynaptic H1Rs (54,59,60). H1R knockout mice (61) have been used in studies relevant to this review (55–57). Central infusion of histamine for 7 d lowered food intake by 26% and body weight by 11% in H1R +/+ mice but lowered food intake by only 15% and body weight by only 5% in H1R –/– mice (57), suggesting that the effect of central histamine on food intake and body weight is conferred partly, but not completely, through the H1R. Chow-fed H1R –/– and +/+ mice studied up to 30 weeks of age did not differ in body weight, body fat or food intake; HFD challenge also did not alter body weight or food intake between the H1R –/– and +/+ mice, but body fat was increased by 20% in –/– mice after 8 weeks on HFD, and food intake was less inhibited by leptin in these –/– mice (55). In contrast, feeding a different HFD for 30 weeks resulted in a 25% increase in the body weight of H1R –/– mice relative to +/+ littermates; the increased weight was associated with a slight (5%) trend toward increased food intake (56). The observation that body weight appears to increase more than food intake in –/– mice suggests that H1R inactivation may also decrease energy expenditure, perhaps through the leptin pathway (57). Although activating postsynaptic H1R should be a viable approach to treating obesity, it is not being pursued because of difficulty in delivering a selective agonist to central H1Rs that will not also activate peripheral H1Rs resulting in undesirable effects currently treated by multiple popular H1R antagonists, the antihistamines.
Histamine H3 receptor (H3R)
Presynaptic H3Rs are constitutively active and histamine increases this activity; one end result of activation is the inhibition of histamine synthesis and release. Thus, H3R antagonists should increase histamine release resulting in postsynaptic H1R activation, essentially an indirect way to target H1Rs (54). Strong evidence for this hypothesis is provided by the ability of thioperamide, a specific H3R antagonist, to acutely decrease food intake in H1R +/+ and H3R +/+ mice, but not in either H1R –/– or H3R –/– mice (56,58). These results suggest that H3R –/– mice should be lean and eat less. Indeed, the initial description of an H3R knockout line reported that the –/– mice displayed a slightly lower mean body weight that did not achieve statistical significance (62); this appears consistent with the above data. However, a recent in-depth study of a second H3R knockout line found that –/– mice displayed an unexpectedly obese phenotype (58). By 40 weeks of age, both male and female –/– mice weighed 21% more than their respective +/+ littermates. This was associated with 23% and 6% increases in food intake by male and female –/– mice respectively, and also with significant increases in percentage body fat, and significant decreases in energy expenditure, in both the male and female –/– mice (58). Recently, DIO mice treated for 28 d with A-331440, a selective H3R antagonist, displayed decreased food intake resulting in a 20% loss of body weight and 55% loss of body fat relative to vehicle-treated control mice (9); although these data suggest that H3R inhibition leads to decreased food intake and weight loss, studies were not performed using H3R –/– and +/+ mice to determine if these effects of A-331440 were entirely conferred through the H3R. Further work is needed to reconcile the knockout data with in vivo compound effects, but it is clear that compounds targeting H3R should be pursued as potential therapeutics for obesity.
Cholecystokinin 1 receptor (CCK1R)
Cholecystokinin (CKK) is a peptide hormone and neuropeptide that has long been recognized for its ability to acutely decrease food intake in a number of vertebrate species. CCK binds to 2 receptors, CCK1R and CCK2R. Early studies with selective agonists suggested that the acute effect of CCK was conferred through CCK1R; this was confirmed with studies using knockout mice, because the CCK isoform CCK-8 acutely reduces food intake by 90% in CCK2R –/– mice but has no effect on food intake of CCK1R –/– mice (63). Interestingly, the weight of chow-fed CCK1R –/– mice is not different from that of +/+ control mice from weaning through almost 300 d of age (63). This suggests that CCK-mediated inhibition of food intake is not important in the overall regulation of body weight, a possibility that is supported by additional studies evaluating the effects of chronic CCK delivery in vivo. Rats receiving constant CCK-8 infusion by osmotic minipump show no difference in total food intake or body weight over 14 d, but they become refractory to acute CCK-8 infusions that decrease food intake of control rats receiving saline by minipump (64). Likewise, rats receiving a CCK-8 bolus at the initiation of each meal show no difference in food intake or body weight after 6 d; CCK-8 did lower meal size, but this effect was offset by an increase in meal frequency (65). Based on these data, it is possible that CCK will be an example where lack of efficacy of a potential anti-obesity therapeutic during human clinical trials was predicted by antecedent knockout data.
Neuropeptide Y Y1 receptor (NPY Y1R)
Neuropeptide Y (NPY) is a potent appetite stimulant in the central nervous system (CNS) (66). Although NPY –/– and +/+ mice exhibit comparable food intake and body weight on either chow or high fat diets (67,68), NPY-deficient ob/ob mice eat and weigh less than their ob/ob littermates (69). This suggests that, under certain conditions such as leptin deficiency, inhibition of NPY can result in decreased food intake and body weight.
Although multiple NPY receptors have been identified, the available evidence suggests that NPY-mediated effects on feeding are conferred in part through the NPY Y1 receptor (NPY Y1R): (i) NPY Y1R –/– mice eat less than littermate controls in the basal state and after fasting; (ii) NPY Y1R –/– mice are partially resistant to the appetite-stimulating effects of exogenous NPY; and (iii) NPY Y1R-deficient ob/ob mice eat less than their ob/ob littermates (70–72). Interestingly, the hypophagia noted in NPY Y1R –/– mice was associated with decreased body weight only on the ob/ob background; on the C57BL/6 background, NPY Y1R –/– mice are obese because of a marked decrease in activity and metabolic rate (70).
J-115814 is a potent NPY Y1R antagonist that is also highly selective for the NPY Y1R compared with the NPY Y2, Y4 and Y5 receptors (73). In rats, J-115814 suppressed feeding induced by centrally administered NPY, and also decreased spontaneous feeding in db/db and C57BL/6 mice. Importantly, this compound suppressed NPY-induced feeding in NPY Y5R –/– mice but not in NPY Y1R –/– mice, indicating that J-115814 is acting through NPY-Y1R (73). Thus, the hypophagia noted in NPY Y1R –/– mice is recapitulated in mice receiving J-115814, consistent with findings reported for other NPY Y1R antagonists (73,74). Although the effects of J-115814 on body weight were not reported in this study (73), a later study using the orally available and NPY Y1R-selective antagonist J-104870 showed that J-104870 suppressed food intake and body weight gain in the fa/fa (fatty Zucker) rat, which like the db/db mouse is a rodent line that develops obesity because of mutation of the long form of the leptin receptor (74). Considering the variable body composition phenotypes of NPY Y1R –/– mice on different backgrounds, it is possible that the decreased food intake induced by NPY Y1R antagonists will lead to sustained losses in body weight and fat only in selected individuals, e.g. those with leptin deficiency.
Nuclear hormone receptors
Estrogen receptor α
In rodents, oestrogen has a strong influence on body fat stores. Male and female mice lacking the oestrogen receptor α gene (ERKO mice) exhibit increased body weight and fat, decreased energy expenditure and normal food intake; 11-month-old male mice showed a 16% increase in body weight because of a greater than 2-fold increase in the weights of all tested fat depots (75). A number of studies show that various oestrogens and oestrogen analogues decrease weight, fat stores and food intake in mice and rats; in particular, the oestrogen partial agonist EM-652 reverses the dramatic gains in body weight, body fat and food intake that result from ovariectomy (76–80). Although E2 benzoate decreased the weight and food intake of ovariectomized littermate control mice, it had no effect on ovariectomized ERKO mice, confirming that the obesity-sparing effects of oestrogen are mediated through ERα (78). As expected, knockouts of the aromatase (ARKO) and follicle-stimulating hormone (FORKO) genes result in mice with oestrogen deficiency, increased body weight and increased body fat (79,80); estradiol 17β reduces the fat stores of ARKO and FORKO mice, consistent with the fact that the ERα gene is intact in these knockout mice.
In contrast to the clear response of rodents to ovariectomy and to oestrogen replacement, the effect of menopause and oestrogen replacement on human obesity is less obvious. There is little evidence suggesting that menopause is associated with a significant increase in body fat, but there is evidence that menopause is associated with a redistribution of fat to the abdomen (81,82; reviewed in 83). A meta-analysis was performed on randomized, controlled trials examining the effect of hormone replacement therapy (HRT), delivered over 3–48 months, on the body weight of thousands of perimenopausal and postmenopausal women (84). This analysis found that body weight was not significantly altered by HRT; however, limited data prevented the drawing of any conclusions regarding the effect of HRT on either total body fat or body fat distribution (84). A recent randomized, controlled clinical trial evaluated the effect of 5 years of HRT on the body weight, body fat and body fat distribution of almost 2000 postmenopausal women (85). At the end of the study, women in the HRT cohort gained significantly less weight (mean difference 0.6 kg), less fat (mean difference 0.7 kg) and less trunk fat (mean difference 0.5 kg) than the untreated cohort, suggesting that HRT was limiting visceral obesity. Additional randomized, controlled, smaller-scale studies not reviewed in the meta-analysis also suggest that HRT lowers total body fat and visceral fat in postmenopausal women on HRT; these modest decreases were statistically significant in some, but not all, instances (86–90). Interestingly, the favourable response of obese women (86) and women with an android body habitus (89) to HRT suggests that HRT might have more striking effects on body fat if specific populations were targeted for treatment.
Thyroid hormone receptors
It has long been recognized that thyroid hormone regulates energy expenditure in humans, and recent work clearly shows that resting energy expenditure correlates directly and exquisitely with changes in thyroid hormone within the euthyroid range (91,92). What is less well-recognized is that thyroid hormone also stimulates appetite and lipogenesis. This suggests that a primary function of thyroid hormone is heat production to maintain body temperature; increased food intake and lipogenesis provide the fuel source for heat production (92). Thus, excess thyroid hormone should not have a major effect on body weight unless food intake is restricted. This conclusion is supported by rodent studies. In rats treated with thyroid hormone, weight loss occurs only in association with decreased food intake (93); in addition, treating lean or obese mice with thyroid hormone increased energy expenditure but also increased food intake, resulting in a modest reduction in body fat with little change in body weight (76). Human data also support this conclusion, because the studies that showed weight loss in response to normal or supraphysiologic doses of thyroid hormone also maintained the treated patients on very-low-calorie diets (94–97; reviewed in 98). Thus, thyroid hormone is not a particularly effective weight loss medication, and off-target effects of supraphysiologic doses call into question the safety of this therapeutic.
There is no ideal knockout line to model hypothyroidism. Although the only primary developmental defect of Pax8 knockout mice is total absence of thyroid follicular cells, this is not a useful model because ligand-independent actions of thyroid hormone receptors result in perinatal lethality (99). Oxygen consumption is not decreased in mice lacking either thyroid hormone receptor α1 (TRα1) or TRβ; in fact, mice lacking TRα1 show significantly increased basal oxygen consumption relative to littermate controls, a counterintuitive result (100,101). However, the ability of exogenous thyroid hormone to increase oxygen consumption more than heart rate in TRα1 –/– mice suggests that selective TRβ agonists might be useful for treating obesity (100). Indeed, KB-141, a selective TRβ agonist, increases oxygen consumption more than heart rate in mice. In addition, providing KB-141 for 7 days to cynomolgus monkeys results in a 7% loss of body weight with no change in food intake or heart rate (100). Thus, the selective increase in metabolic rate observed in thyroid hormone-treated TRα–/– mice corresponds well to the ability of selective TRβ agonists to increase metabolic rate in mice and to lower body weight in primates.
Peroxisome proliferator-activated receptor α (PPARα)
Peroxisome proliferator-activated receptor α (PPARα) is the master regulator of fatty acid catabolism in liver and is the target of the fibrate family of lipid-lowering therapeutics (102). PPARα also appears to play a role in obesity, because recent studies revealed a subtle obesity phenotype in PPARα–/– mice (103–106). The body weight of PPARα–/– mice was increased relative to littermate controls in one study where mice were fed a HFD (103), and in most (104,105), but not all (106), studies where mice were fed a chow diet. Importantly, all studies of mice fed a chow diet reported a significant increase in gonadal fat pad weight in PPARα–/– mice (104–106); in one of these studies, five additional fat pads were analysed and all were found to weigh significantly more in male and female PPARα–/– mice (104). Food intake was increased when PPARα–/– mice were fed a HFD (103), but not when fed a chow diet (104). These observations suggest that PPARα agonists might act to lower body fat stores by inducing beta-oxidation and perhaps satiety.
Oleylethanolamide (OEA), which reduced food intake and weight gain in both lean and obese rats (107,108), is structurally similar to PPARα agonists and does in fact bind PPARα with high affinity (103). Treating PPARα–/– mice and littermate controls with OEA for 4 weeks resulted in modest, but significant, decreases in food intake and body weight of littermate control mice, but had no effect on –/– mice (103). These data suggest that OEA works through PPARα and that PPARα may be a target for obesity regulation. This hypothesis is supported by the following findings: (i) gonadal fat pad weight was significantly decreased in PPARα+/+ littermate controls, but not in PPARα–/– mice, after 7 d of treatment with the peroxisome proliferator WY-14643 (106); and (ii) chronic fibrate therapy led to modest but significant decreases in weight gain and fat depot weight, but not food intake, in HFD-challenged mice and rats (109–111).
Hepatic PPARα levels are likely 10-fold lower in humans than mice, and fibrates cause hepatic toxicity in rodents but not humans, suggesting that rodents may not be a good model for predicting the effects of PPARα modulation in humans (104,112). Nevertheless, the ability of PPARα agonists to lower lipid levels in rodents is consistent with their lipid-lowering effects in humans (103,110,111,113,114). In addition, there is more direct evidence that PPARα may play a role in human obesity. First, a gain-of-function mutation in the DNA binding domain of PPARα (L162V) was associated with significantly lower body mass index (BMI) and percentage body fat, and with trends toward lower waist circumference and total abdominal fat area as determined by computed tomography (115). In an independent study, this same mutation was associated with lower BMI in patients with type 2 diabetes (116). Also, the ability of fibrates to lower low-density lipoprotein (LDL)-cholesterol in humans was positively associated with on-treatment weight loss (114). These observations are consistent with the possibility that, similar to rodents, some humans may respond to PPARα activation with modest weight loss.
Peroxisome proliferator-activated receptor γ (PPARγ)
Thiazolidinediones (TZDs) exert their anti-diabetic effect by activating peroxisome proliferator-activated receptor γ (PPARγ). In addition to improving insulin sensitivity in rodents and humans, TZDs also promote weight gain (117–120). To better understand this process, mice were created with a targeted mutation of PPARγ (121). While PPARγ–/– mice die during embryogenesis, +/– mice are viable and show no phenotype early in life when on a low-fat diet (121). However, +/– mice on a HFD show improved insulin sensitivity and decreased food intake; by 15 weeks of age, +/– mice show a 30% decrease in fat pad weight that is likely responsible for their 15% lower body weight (117,121). When +/– mice are compared with +/+littermates and +/+ littermates treated with the TZD rosiglitazone (Rosi +/+ ), Rosi +/+ mice have higher body and fat depot weights than untreated +/+ mice, while +/– mice have the lowest body and fat depot weights; in contrast, both Rosi +/+ and +/– mice are more insulin sensitive than untreated +/+ mice (117). The mechanism behind these phenotypes is the subject of debate and recent review (117,119), but it is clear that the PPARγ+/– phenotype is desirable and predictive that PPARγ antagonists or inverse agonists can improve insulin sensitivity as do agonists, but in addition can lower body fat stores.
Recently, this prediction was supported by data obtained from mice treated with PPARγ antagonists BADGE (122) and SR202 (123). KKAy mice on HFD responded to BADGE with lower body weight, epididymal fat pad weight, adipocyte size, insulin levels and glucose levels. These effects appeared to be mediated in part through leptin-dependent pathways; among other evidence for this, db/db mice appeared less responsive to the anti-obesity and anti-diabetic effects of BADGE. Of particular interest, treatment of PPARγ+/– with BADGE for 4 weeks resulted in complete loss of visible white adipose tissue, hyperglycaemia, insulin resistance, and accumulation of triglycerides in liver and skeletal muscle; this phenotype was not seen when BADGE was given to wild-type DIO mice. These results in BADGE-treated PPARγ+/– mice are consistent with the lipodystrophy phenotype described in mice with an adipose-specific knockout of PPARγ (124). SR-202 lowered insulin levels, body weight, and weight of epididymal and brown fat depots in WT mice on either chow or HFD. However, in contrast to BADGE, SR-202 had no effect on any of these parameters when given to PPARγ+/– mice fed either chow or HFD. Taken together, these data suggest that there may be an optimal level of PPARγ inhibition, and that going beyond this level of inhibition with certain PPARγ antagonists may cause an extreme depletion of adipose tissue that is associated with a progressively more severe insulin resistance. These and other results (reviewed in 125–127) suggest the need for future studies that clarify the role of PPARγ antagonists or inverse agonists in the treatment of human obesity and type 2 diabetes.
Norepinephrine and dopamine β-hydroxylase
Many effective anti-obesity drugs act centrally to increase synaptic levels of norepinephrine (128). Phentermine, a stimulant monoamine currently approved for short-term use in the treatment of obesity, acts by this mechanism to acutely decrease food intake and body weight in mice (129). In three independent human clinical trials, patients responded to phentermine therapy with decreased appetite (130) and with 12.9% (130), 6.7% (131), and 3.4% (132) greater weight loss than was noted in patients receiving placebo. Sibutramine, the only appetite suppressant currently approved for chronic treatment of obesity, blocks the synaptic reuptake of both norepinephrine and serotonin. Both monoamines contribute to the anorectic effect of sibutramine. However, indirect activation of α1 adrenoceptors and possibly also β1 adrenoceptors, secondary to inhibition of norepinephrine reuptake, appears to play a major role in this process (133–135). Sibutramine induces weight loss by decreasing appetite in ob/ob mice (136) and DIO rats (137). In four independent human clinical trials, patients receiving 10 mg d1 of sibutramine averaged 5% more weight loss than did those receiving placebo (11).
Mice lacking functional dopamine β-hydroxylase (dbh) cannot synthesize norepinephrine or epinephrine (138). These dbh–/– mice have normal amounts of body fat and show increases in both basal metabolic rate and food intake. Because the basal metabolic rate of fasted dbh–/– mice falls to normal within hours of norepinephrine repletion by treatment with DOPS, the increased food intake observed in these mice is probably secondary to their increased metabolic rate. Thus, the complex phenotype of dbh–/– mice suggests a primary role for norepinephrine in regulating basal metabolic rate but does not clearly link norepinephrine to a primary role in regulating appetite and body weight.
Pancreatic lipase-related protein 2 (PLRP2), a protein with high triglyceride lipase activity, is secreted by the pancreas into the intestinal lumen of suckling mice to aid in digestion of dietary fat. Knockout of PLRP2 leads to weight loss in –/– pups between day 4 of life and weaning. At 23 d of age, the –/– pups weighed 23% less than littermate +/+ mice (139). The stools from these suckling –/– mice are loose, of increased volume, and contain 6-fold more fat, primarily undigested triglycerides, than do stools from littermate controls. These data are consistent with the effects of Orlistat (Xenical), an approved anti-obesity drug that inhibits pancreatic and gastric lipases. Giving mice a single dose of Orlistat results in a dose-dependent decrease in fat absorption, while providing Orlistat for 22 d to obese rats fed a HFD results in an 8% weight loss and increased food intake in the treatment group (140). In nine human clinical trials that evaluated percentage weight loss, humans also responded to daily Orlistat with 5–10% weight loss, which was always significantly greater than the 3–7% weight loss observed in the placebo control group (11,141–145).
Protein tyrosine phosphatase 1B (PTP1B)
Protein tyrosine phosphatase 1B (PTP1B) was initially implicated as a negative regulator of insulin action; among other evidence, PTP1B was shown to dephosphorylate both the insulin receptor and IRS-1 (146–148). As expected, PTP1B –/– mice are insulin sensitive, but they are also unexpectedly lean despite increased food intake (146,147). When fed a HFD, 16-week-old male –/– mice show a 38% decrease in body weight because of a 72% decrease in body fat and a 43% decrease in fat-free dry mass; additional studies suggest that an increased metabolic rate is the likely basis for the lower fat mass (146). Recent attempts to develop drugs targeting PTP1B have been hampered by the general difficulty in developing small molecule inhibitors that target a specific phosphatase. As an alternative, an antisense oligonucleotide, ISIS-113715, was developed that specifically targets PTP1B (148). In addition to improving insulin sensitivity in ob/ob mice, ISIS-113715 also had an effect on body fat. Although lower doses were ineffective, the highest ISIS-113715 dose lowered body weight by 15% and epididymal fat pad weight by 42% after 5 weeks of treatment. Under the conditions of this study, ISIS-113715 had no effect on food intake. These data suggest that ISIS-113715 lowers body fat by specifically targeting PTP1B. Additional studies showing that this antisense oligonucleotide lowers body fat in littermate controls, but not PTP1B –/– mice, would strengthen this impression.
Fatty acid synthase (FAS)
Fatty acid synthase (FAS) is the rate-limiting enzyme for de novo long chain saturated fatty acid synthesis. As such, FAS is a logical obesity target. Knockout of the FAS gene results in embryonic lethality for –/– mice and in reduced viability of +/– mice (149); clearly, FAS is essential for embryogenesis. Recent studies using C75, a FAS inhibitor, show that C75 dramatically decreases food intake and increases energy expenditure in DIO mice, resulting in a dramatic decrease in body weight because of a 73% decrease in body fat after 30 d of treatment. The effect appears to be centrally mediated with stimulation of glucose metabolism central to the mechanism (150–152). To date, the FAS +/– mouse phenotype yields no clues that suggest that FAS inhibitors might have utility in treating obesity.
In ob/ob mice, a naturally occurring mutation in the leptin gene results in loss of functional leptin while in db/db mice, a naturally occurring mutation in the leptin receptor gene results in markedly decreased hypothalamic expression of the long form of the leptin receptor (153). Both ob/ob and db/db mice rapidly gain weight and body fat because of increased food intake and decreased energy expenditure. By 7 weeks of age, cohorts of chow-fed ob/ob and db/db mice each weigh 50% more than +/? controls, with the weight difference almost entirely because of fat (154). Treatment of ob/ob mice with recombinant leptin profoundly decreases food intake while increasing energy expenditure, resulting in an almost 50% decrease in body weight that is almost entirely because of fat loss; leptin has none of these effects when given to db/db mice (155). Leptin treatment decreased body weight, body fat and food intake in some, but not all, additional studies employing lean and obese mice (156,157). Of particular interest, mice and rats with DIO may be less responsive to leptin therapy; DIO rats, perhaps a more appropriate model for the current epidemic of human obesity, become progressively more refractory to leptin therapy with age (157,158).
Leptin therapy of three adult humans with leptin deficiency dramatically lowered their body weight, body fat and food intake, while increasing their physical activity. After 18 months of treatment, body weight of these three patients dropped by 70, 48 and 58 kg, and body fat by 52, 38 and 39 kg (159). In contrast, treatment of obese, leptin-replete adults with a much higher dose of leptin resulted in a much less impressive weight loss. After 24 weeks, daily injections of 0.3 mg kg1 leptin lowered body weight by 8% (−7.1 ± 8.5 kg) while placebo injections lowered body weight by 1% (−1.3 ± 4.9 kg); the response to leptin was, obviously, quite variable (160). More recent studies show a similarly modest weight loss in obese patients treated with weekly injections of pegylated leptin (161). Thus, the data from studies of leptin therapy in obese humans are consistent with the data obtained in rodent studies.
Ciliary neurotrophic factor (CNTF)
Ciliary neurotrophic factor (CNTF) was first identified as a trophic factor for motor neurones. When CNTF was administered to patients suffering from motor neurone disease, considerable weight loss was noted in the treatment group (162). A more recent 12-week dose-ranging study in obese adults supported these findings, with the treated groups showing significantly greater weight loss than the placebo control group (163).
Ciliary neurotrophic factor (CNTF) shares a number of similarities with leptin; both cytokines signal through related receptors expressed in hypothalamic nuclei that play a role in feeding, and both activate overlapping signalling pathways (164). In ob/ob mice, leptin and CNTF induced similar decreases in food intake, body weight and body fat with little change in lean body mass (164,165). However, CNTF also induces weight loss in leptin-resistant DIO mice (164,165). Although recent studies indicate that CNTF and leptin act on different CNS sites (166), the evidence also suggests that CNTF is not acting merely as a cachectic cytokine, because CNTF does not induce muscle wasting, CTA or inflammation at doses that lead to significant weight loss (164,165).
Despite the ability of CNTF to induce weight loss in mice, CNTF –/– mice are not obese (164,167,168). Likewise, a null mutation in the human CNTF gene is not associated with obesity (164,169). These data suggest that CNTF does not play a major physiologic role in regulating mammalian body composition.
Amylin is a pancreatic beta cell peptide that is co-secreted with insulin in response to meals. Like insulin, amylin is deficient in patients with beta cell destruction, and amylin replacement can improve glucose homeostasis in these patients through a variety of effects (170,171). Amylin was also found to decrease food intake under a number of experimental conditions when delivered to mice by intraperitoneal injection (172). Peripherally administered amylin has been shown to enter the brain and bind specifically in a number of brain regions, including the hypothalamus (173). The available evidence suggests that amylin may work through the histamine H1R, because amylin does not affect food intake when given to H1R –/– mice (56).
Similar to insulin and leptin blood levels, amylin blood levels correlate directly with body fat, suggesting that amylin might play a role as a long-term regulator of body weight (173). Consistent with this hypothesis, two independent cohorts of male amylin –/– mice showed 18% and 29% increases in body weight by 18 and 40 weeks of age respectively. In addition, female –/– mice showed up to a 10% increase in body weight by 15 weeks of age (174,175). Furthermore, subcutaneous (176) or intracerebroventricular (173,177) delivery of amylin to rats resulted in a long-term decrease in food intake associated with a modest decrease in body weight. These findings fit well with the results of randomized, double-blind, placebo-controlled clinical trials in which diabetic patients were treated for at least 1 year with the amylin analogue pramlintide. In each case, pramlintide therapy was associated with a modest but significant weight loss ranging from 1.5 to 4 kg (178–181).
Obesity targets: References
The data reviewed here suggest that the phenotypes of obesity target knockouts model the effects seen when therapeutics designed for those obesity targets are delivered to rodents. Of the 21 obesity targets reviewed here, 16 showed a correspondence between the knockout phenotype and the effect of a therapeutic designed for that target. This suggests that, at least in terms of evaluating obesity targets, it is rare for compensatory developmental changes caused by the gene knockout to prevent detection of the relevant phenotype. Although this review is not exhaustive, the conclusions drawn seem warranted considering the large numbers of obesity targets reviewed, including knockouts of the known targets for most present and past obesity therapeutics, and knockouts of the known targets of many potential obesity therapeutics currently in clinical development.
The only targets where knockout phenotype did not model therapeutic effect were thyroid hormone receptors, H3R, norepinephrine, FAS and CNTF. In the case of thyroid hormone receptors, knockouts of TRα1 and TRβ do not by themselves provide clues to the role of thyroid hormone in stimulating metabolic rate. However, the ability of exogenous thyroid hormone to increase oxygen consumption more than heart rate in TRα1 –/– mice correctly models the ability of selective TRβ agonists to increase metabolic rate but not heart rate in mice, and to lower body weight in primates. Thus, the knockout phenotype did correctly model the effects of the appropriately designed therapeutic, but the –/– mice required the presence of excess thyroid hormone to display the phenotype (100,101). In the case of H3R, detailed analysis of one knockout line found that H3R –/– mice were obese (58), yet wild-type mice treated with the H3R antagonist A-331440 became quite lean (9). However, it should be noted that (i) A-331440 was not tested in H3R –/– mice to identify off-target effects; and (ii) limited data from a second H3R knockout line showed a trend for H3R –/– mice to weigh less than +/+ controls (62); thus, further study is needed to confirm that the H3R knockout phenotype does not model drug effects. In the case of norepinephrine, dbh–/– mice eat more (138); this could be considered consistent with their norepinephrine deficiency, but these mice have normal stores of body fat, a finding that is inconsistent with the fat loss induced by phentermine and related compounds (128). In the case of FAS, the knockout is embryonic lethal for –/– mice, +/– mice show reduced viability, and surviving mice provide no insight into the utility of FAS as an obesity target (149). Despite this experience with FAS, it should be noted that knockout lines that are devoid of viable –/– mice may still provide strong evidence that the gene product is an obesity target. PPARγ is a good example, because –/– mice are not viable but +/– mice show a very clear lean phenotype suggesting that PPARγ antagonists or inverse agonists may be useful in the treatment of obesity (117).
Potential obesity targets:
Of the 16 obesity targets that showed a correspondence between phenotype and therapeutic effect in rodents, nine have had therapeutics delivered to humans. The effects of orlistat, SR141716, fenfluramine, amylin and leptin in humans are consistent with the corresponding knockout phenotype in mice. The leptin data merit special mention for two reasons. First, leptin is actually the first example of a gene inactivation predicting that a new therapeutic would effectively treat some obese humans. Thus, the initial discovery that a mutation in the leptin gene resulted in morbid obesity led to the development of the leptin protein as a treatment for obesity. Leptin was first shown to lower body fat in rodents and then more recently used to effectively treat morbid obesity in leptin-deficient humans. Second, the ability of leptin to induce greater weight loss in ob/ob mice than in obese humans is often cited as evidence that mouse models are unreliable in predicting human response. In fact, both humans and mice with inactivating mutations in the leptin gene are morbidly obese and respond dramatically to exogenous leptin (154,159), while both obese humans and older DIO rodents show a limited response to leptin therapy (156,158,160,161), suggesting that the leptin response in humans is reasonably modelled in mice.
The acute effects of β3AR agonists on metabolic rate in humans also appear consistent with the knockout phenotype in mice (43,44,52). While current data suggest the chronic effects of β3AR agonists on body composition are weaker in humans than the β3AR knockout phenotype predicts (43,44,53), experience is limited. Further study is required to determine if adequate doses of highly selective β3AR agonists delivered to an optimal subset of obese humans will lead to safe and effective weight loss.
The obese phenotype of mice and humans with MC4R inactivating mutations (14–18) suggests that rodents and humans should respond in a similar manner to MC4R agonists. The acute response of humans (nausea) and rats (CTA) to the MC4R agonist MTII (22–26) suggests that both species are indeed responding in a similar manner to this MC4R agonist and that MTII-induced weight loss may well be the result of induction of visceral illness. The crucial question is whether MTII-induced visceral illness is an on-target effect in rodents and humans. Available data, although limited, suggest that rodent models will be useful both to characterize the on-target effects of future selective MC4R agonists and to predict the utility of these agonists in humans.
The trend toward a decrease in visceral body fat in postmenopausal women treated with HRT is in general consistent with the fat loss of oestrogen-deficient ARKO and FORKO mice during oestrogen replacement (79,80,85,87–90). Estrogen appears to be much more effective in depleting fat depots in rodents, but selection of specific subsets of postmenopausal women for HRT may increase the loss of total and visceral body fat (86,89). When considering the ability of PPARα agonists to induce fat loss in humans, available data are too limited to draw firm conclusions. It is possible that PPARα agonists exert very modest effects on fat loss in a subset of humans, as they do in rodents (103–111,114), but detecting such subtle effects in humans will be very difficult without carefully designed trials.
Although therapeutics based on the POMC and PPARγ knockouts have not yet been delivered to humans, existing data allow some predictions to be made. First, it seems likely that human POMC deficiency, which like leptin deficiency results in a similar obesity phenotype in mice and humans, will also respond to the appropriate exogenous protein therapeutic, in this case MSH (12,13). Second, PPARγ agonists cause obesity not only in humans but also in PPARγ+/+ mice (117,120); because PPARγ+/– mice are lean and PPARγ antagonists lower body fat in PPARγ+/+ mice, it seems quite likely that PPARγ antagonists will lower body fat in humans. The major concern with PPARγ inhibition is the potential for on-target side effects including lipodystrophy and increased risk of atherosclerosis (122–127). Wild-type DIO mice treated with the PPARγ antagonists SR-202 and BADGE did not develop lipodystrophy or appear ill (122,123), and the lean phenotype noted in PPARγ+/– mice was not associated with obvious toxicity (117,121), suggesting that partial PPARγ inhibition may be an effective treatment for obesity and type 2 diabetes in humans.
In summary, the available data suggest that knockout phenotypes of potential obesity targets model therapeutic effects in mice and, in the majority of cases, also in humans. This conclusion is in agreement with the findings of retrospective reviews comparing the effects of a large number of drugs from many therapeutic areas with the phenotype of the relevant knockout mice (2,3). Further, this conclusion is not surprising when one considers that more than 99% of mouse genes have a human ortholog (182). Because almost all genes are conserved between the two vertebrates, it is perhaps to be expected that many gene functions are also conserved. With this in mind, it seems rational to use mouse knockout technology prospectively to identify genes that regulate body fat in vivo, and subsequently to develop anti-obesity therapeutics by targeting the human protein products of these genes. PTP1B is an example where the knockout phenotype predicts that inhibiting PTP1B will lower body fat, and a therapeutic specifically targeting PTP1B subsequently recapitulated the lean phenotype in vivo (148). In a number of additional knockouts listed in Table 2, the knockout phenotypes predict a role for the gene in regulating body fat, but therapeutics specifically targeting the gene product have yet to be tested in vivo. This list is not exhaustive, and emphasizes knockouts of genes that are considered viable targets by current pharmaceutical industry standards (2,3). In addition, these conclusions have implications for the Knockout Mouse Project, which will be a co-ordinated effort to systematically knockout all mouse genes (1). It is proposed, as part of this effort, that all knockout lines be phenotyped using a limited set of broad and cost-effective screens. The data reviewed here suggest that a screen for body composition would be likely to provide real value in identifying genes that regulate, and are potential targets for the treatment of, obesity. Along these lines, it is important to note that accurate and cost-effective approaches now exist for high-throughput screening of body composition in mouse knockout lines (DR Powell et al. manuscript in preparation). Ultimately, the value of using gene knockout technology to identify novel targets for human anti-obesity therapies will be judged by future studies examining the anti-obesity effect, in humans, of therapeutics targeting the protein products of these genes.
1. Austin CP, Battey JF, Bradley A, Bucan M, Capecchi M, Collins FS, Dove WF, Duyk G, Dymecki S, Eppig JT, Grieder FB, Heintz N, Hicks G, Insel TR, Joyner A, Koller BH, Lloyd KC, Magnuson T, Moore MW, Nagy A, Pollock JD, Roses AD, Sands AT, Seed B, Skarnes WC, Snoddy J, Soriano P, Stewart DJ, Stewart F, Stillman B, Varmus H, Varticovski L, Verma IM, Vogt TF, von Melchner H, Witkowski J, Woychik RP, Wurst W, Yancopoulos GD, Young SG, Zambrowicz B. The knockout mouse project. Nat Genet 2004; 36: 921–924.
2. Zambrowicz BP, Sands AT. Knockouts model the 100 best-selling drugs – will they model the next 100? Nat Rev Drug Discov 2003; 2: 38–51.
3. Zambrowicz BP, Turner CA, Sands AT. Predicting drug efficacy: knockouts model pipeline drugs of the pharmaceutical industry. Curr Opin Pharmacol 2003; 3: 563–570.
4. Tecott LH, Sun LM, Akana SF, Strack AM, Lowenstein DH, Dallman MF, Julius D. Eating disorder and epilepsy in mice lacking 5-HT2c serotonin receptors. Nature 1995; 374: 542–546.
5. Nonogaki K, Strack AM, Dallman MF, Tecott LH. Leptin-independent hyperphagia and type 2 diabetes in mice with a mutated serotonin 5-HT2C receptor gene. Nat Med 1998; 4: 1152–1156.
6. Vickers SP, Clifton PG, Dourish CT, Tecott LH. Reduced satiating effect of d-fenfluramine in serotonin 5-HT (2C) receptor mutant mice. Psychopharmacology (Berl) 1999; 143: 309–314.
7. Carr RH, Ipaktchi M, Thenen SW. Effects of prolonged fenfluramine administration in obese and nonobese mice. Proc Soc Exp Biol Med 1977; 154: 116–120.
8. Rowland NE. Long-term administration of dexfenfluramine to genetically obese (ob/ob) and lean mice: body weight and brain serotonin changes. Pharmacol Biochem Behav 1994; 49: 287–294.
9. Hancock AA, Bennani YL, Bush EN, Esbenshade TA, Faghih R, Fox GB, Jacobson P, Knourek-Segel V, Krueger KM, Nuss ME, Pan JB, Shapiro R, Witte DG, Yao BB. Antiobesity effects of A-331440, a novel non-imidazole histamine H(3) receptor antagonist. Eur J Pharmacol 2004; 487: 183–197.
10. Weintraub M, Hasday JD, Mushlin AI, Lockwood DH. A double-blind clinical trial in weight control. Use of fenfluramine and phentermine alone and in combination. Arch Intern Med 1984; 144: 1143–1148.
11. Halpern A, Mancini MC. Treatment of obesity: an update on anti-obesity medications. Obes Rev 2003; 4: 25–42.
12. Yaswen L, Diehl N, Brennan MB, Hochgeschwender U. Obesity in the mouse model of pro-opiomelanocortin deficiency responds to peripheral melanocortin. Nat Med 1999; 5: 1066–1070.
13. Krude H, Biebermann H, Schnabel D, Tansek MZ, Theunissen P, Mullis PE, Gruters A. Obesity due to proopiomelanocortin deficiency: three new cases and treatment trials with thyroid hormone and ACTH4-10. J Clin Endocrinol Metab 2003; 88: 4633–4640.
14. Huszar D, Lynch CA, Fairchild-Huntress V, Dunmore JH, Fang Q, Berkemeier LR, Gu W, Kesterson RA, Boston BA, Cone RD, Smith FJ, Campfield LA, Burn P, Lee F. Targeted disruption of the melanocortin-4 receptor results in obesity in mice. Cell 1997; 88: 131–141.
15. Ste Marie L, Miura GI, Marsh DJ, Yagaloff K, Palmiter RD. A metabolic defect promotes obesity in mice lacking melanocortin-4 receptors. Proc Natl Acad Sci USA 2000; 97: 12339–12344.
16. Weide K, Christ N, Moar KM, Arens J, Hinney A, Mercer JG, Eiden S, Schmidt I. Hyperphagia, not hypometabolism, causes early onset obesity in melanocortin-4 receptor knockout mice. Physiol Genomics 2003; 13: 47–56.
17. Farooqi IS, Keogh JM, Yeo GS, Lank EJ, Cheetham T, O'Rahilly S. Clinical spectrum of obesity and mutations in the melanocortin 4 receptor gene. N Engl J Med 2003; 348: 1085–1095.
18. Branson R, Potoczna N, Kral JG, Lentes KU, Hoehe MR, Horber FF. Binge eating as a major phenotype of melanocortin 4 receptor gene mutations. N Engl J Med 2003; 348: 1096–1103.
19. Jonsson L, Skarphedinsson JO, Skuladottir GV, Watanobe H, Schioth HB. Food conversion is transiently affected during 4-week chronic administration of melanocortin agonist and antagonist in rats. J Endocrinol 2002; 173: 517–523.
20. Marsh DJ, Hollopeter G, Huszar D, Laufer R, Yagaloff KA, Fisher SL, Burn P, Palmiter RD. Response of melanocortin-4 receptor-deficient mice to anorectic and orexigenic peptides. Nat Genet 1999; 21: 119–122.
21. Foster AC, Joppa M, Markison S, Gogas KR, Fleck BA, Murphy BJ, Wolff M, Cismowski MJ, Ling N, Goodfellow VS, Chen C, Saunders J, Conlon PJ. Body weight regulation by selective MC4 receptor agonists and antagonists. Ann N Y Acad Sci 2003; 994: 103–110.
22. Benoit SC, Sheldon RJ, Air EL, Messerschmidt P, Wilmer KA, Hodge KM, Jones MB, Eckstein DM, McOsker CC, Woods SC, Seeley RJ. Assessment of the aversive consequences of acute and chronic administration of the melanocortin agonist, MTII. Int J Obes Relat Metab Disord 2003; 27: 550–556.
23. Bures J, Bermudez-Rattoni F, Yamamoto T. Conditioned Taste Aversion: Memory of a Special Kind. Oxford University Press: New York, 1998.
24. Wessells H, Fuciarelli K, Hansen J, Hadley ME, Hruby VJ, Dorr R, Levine N. Synthetic melanotropic peptide initiates erections in men with psychogenic erectile dysfunction: double-blind, placebo controlled crossover study. J Urol 1998; 160: 389–393.
25. Wessells H, Levine N, Hadley ME, Dorr R, Hruby V. Melanocortin receptor agonists, penile erection, and sexual motivation: human studies with Melanotan II. Int J Impot Res 2000; 12(Suppl. 4): S74–S79.
26. Wessells H, Gralnek D, Dorr R, Hruby VJ, Hadley ME, Levine N. Effect of an alpha-melanocyte stimulating hormone analog on penile erection and sexual desire in men with organic erectile dysfunction. Urology 2000; 56: 641–646.
27. Cota D, Marsicano G, Tschop M, Grubler Y, Flachskamm C, Schubert M, Auer D, Yassouridis A, Thone-Reineke C, Ortmann S, Tomassoni F, Cervino C, Nisoli E, Linthorst AC, Pasquali R, Lutz B, Stalla GK, Pagotto U. The endogenous cannabinoid system affects energy balance via central orexigenic drive and peripheral lipogenesis. J Clin Invest 2003; 112: 423–431.
28. Ravinet Trillou C, Delgorge C, Menet C, Arnone M, Soubrie P. CB1 cannabinoid receptor knockout in mice leads to leanness, resistance to diet-induced obesity and enhanced leptin sensitivity. Int J Obes Relat metab Disord 2004; 28: 640–648.
29. Ravinet Trillou C, Arnone M, Delgorge C, Gonalons N, Keane P, Maffrand JP, Soubrie P. Anti-obesity effect of SR141716, a CB1 receptor antagonist, in diet-induced obese mice. Am J Physiol Regul Integr Comp Physiol 2003; 284: R345–R353.
30. Poirier B, Bidouard JP, Cadrouvele C, Marniquet X, Staels B, O'Connor SE, Janiak P, Herbert JM. The anti-obesity effect of rimonabant is associated with an improved serum lipid profile. Diabetes Obes Metab 2005; 7: 65–72.
31. Vickers SP, Webster LJ, Wyatt A, Dourish CT, Kennett GA. Preferential effects of the cannabinoid CB1 receptor antagonist, SR 141716, on food intake and body weight gain of obese (fa/fa) compared to lean Zucker rats. Psychopharmacology (Berl) 2003; 167: 103–111.
32. Hildebrandt AL, Kelly-Sullivan DM, Black SC. Antiobesity effects of chronic cannabinoid CB1 receptor antagonist treatment in diet-induced obese mice. Eur J Pharmacol 2003; 462: 125–132.
33. Chambers AP, Sharkey KA, Koopmans HS. Cannabinoid (CB) 1 receptor antagonist, AM 251, causes a sustained reduction of daily food intake in the rat. Physiol Behav 2004; 82: 863–869.
34. McLaughlin PJ, Winston KM, Limebeer CL, Parker LA, Makriyannis A, Salamone JD. The cannabinoid CB1 antagonist AM 251 produces food avoidance and behaviors associated with nausea but does not impair feeding efficiency in rats. Psychopharmacology (Berl) 2005; 180: 286–293.
35. Van Gaal LF, Rissanen AM, Scheen AJ, Ziegler O, Rossner S, RIO Europe Study Group. Effects of the cannabinoid-1 receptor blocker rimonabant on weight reduction and cardiovascular risk factors in overweight patients: 1-year experience from the RIO-Europe study. Lancet 2005; 365: 1389–1397.
36. Shimada M, Tritos NA, Lowell BB, Flier JS, Maratos-Flier E. Mice lacking melanin-concentrating hormone are hypophagic and lean. Nature 1998; 396: 670–674.
37. Marsh DJ, Weingarth DT, Novi DE, Chen HY, Trumbauer ME, Chen AS, Guan XM, Jiang MM, Feng Y, Camacho RE, Shen Z, Frazier EGYuH, Metzger JM, Kuca SJ, Shearman LP, Gopal-Truter S, MacNeil DJ, Strack AM, MacIntyre DE, Van der Ploeg LH, Qian S. Melanin-concentrating hormone 1 receptor-deficient mice are lean, hyperactive, and hyperphagic and have altered metabolism. Proc Natl Acad Sci USA 2002; 99: 3240–3245.
38. Borowsky B, Durkin MM, Ogozalek K, Marzabadi MR, DeLeon J, Lagu B, Heurich R, Lichtblau H, Shaposhnik Z, Daniewska I, Blackburn TP, Branchek TA, Gerald C, Vaysse PJ, Forray C. Antidepressant, anxiolytic and anorectic effects of a melanin-concentrating hormone-1 receptor antagonist. Nat Med 2002; 8: 825–830.
39. Dulloo AG, Miller DS. Thermogenic drugs for the treatment of obesity: sympathetic stimulants in animal models. Br J Nutr 1984; 52: 179–196.
40. Robidoux J, Martin TL, Collins S. Beta-adrenergic receptors and regulation of energy expenditure: a family affair. Annu Rev Pharmacol Toxicol 2004; 44: 297–323.
41. Connacher AA, Lakie M, Powers N, Elton RA, Walsh EG, Jung RT. Tremor and the anti-obesity drug BRL 26830A. Br J Clin Pharmacol 1990; 30: 613–615.
42. Wheeldon NM, McDevitt DG, Lipworth BJ. Cardiac effects of the beta 3-adrenoceptor agonist BRL35135 in man. Br J Clin Pharmacol 1994; 37: 363–369.
43. Revelli JP, Preitner F, Samec S, Muniesa P, Kuehne F, Boss O, Vassalli JD, Dulloo A, Seydoux J, Giacobino JP, Huarte J, Ody C. Targeted gene disruption reveals a leptin-independent role for the mouse beta3-adrenoceptor in the regulation of body composition. J Clin Invest 1997; 100: 1098–1106.
44. Susulic VS, Frederich RC, Lawitts J, Tozzo E, Kahn BB, Harper ME, Himms-Hagen J, Flier JS, Lowell BB. Targeted disruption of the beta 3-adrenergic receptor gene. J Biol Chem 1995; 270: 29483–29492.
45. Himms-Hagen J, Cui J, Danforth E Jr, Taatjes DJ, Lang SS, Waters BL, Claus TH. Effect of CL-316,243, a thermogenic beta 3-agonist, on energy balance and brown and white adipose tissues in rats. Am J Physiol 1994; 266: R1371–R1382.
46. Ghorbani M, Claus TH, Himms-Hagen J. Hypertrophy of brown adipocytes in brown and white adipose tissues and reversal of diet-induced obesity in rats treated with a beta3-adrenoceptor agonist. Biochem Pharmacol 1997; 54: 121–131.
47. Liu X, Perusse F, Bukowiecki LJ. Mechanisms of the antidiabetic effects of the beta 3-adrenergic agonist CL-316243 in obese Zucker-ZDF rats. Am J Physiol 1998; 274: R1212–R1219.
48. Umekawa T, Yoshida T, Sakane N, Saito M, Kumamoto K, Kondo M. Anti-obesity and anti-diabetic effects of CL316,243, a highly specific beta 3-adrenoceptor agonist, in Otsuka Long-Evans Tokushima Fatty rats: induction of uncoupling protein and activation of glucose transporter 4 in white fat. Eur J Endocrinol 1997; 136: 429–437.
49. Yoshida T, Sakane N, Wakabayashi Y, Umekawa T, Kondo M. Anti-obesity effect of CL 316,243, a highly specific beta 3–adrenoceptor agonist, in mice with monosodium-L-glutamate-induced obesity. Eur J Endocrinol 1994; 131: 97–102.
50. Gavrilova O, Marcus-Samuels B, Reitman ML. Lack of responses to a beta3-adrenergic agonist in lipoatrophic A-ZIP/F-1 mice. Diabetes 2000; 49: 1910–1916.
51. Collins S, Daniel KW, Petro AE, Surwit RS. Strain-specific response to beta 3-adrenergic receptor agonist treatment of diet-induced obesity in mice. Endocrinology 1997; 138: 405–413.
52. van Baak MA, Hul GB, Toubro S, Astrup A, Gottesdiener KM, DeSmet M, Saris WH. Acute effect of L-796568, a novel beta 3-adrenergic receptor agonist, on energy expenditure in obese men. Clin Pharmacol Ther 2002; 71: 272–279.
53. Larsen TM, Toubro S, van Baak MA, Gottesdiener KM, Larson P, Saris WH, Astrup A. Effect of a 28-d treatment with L-796568, a novel beta (3)-adrenergic receptor agonist, on energy expenditure and body composition in obese men. Am J Clin Nutr 2002; 76: 780–788.
54. Hancock AA. H3 receptor antagonists/inverse agonists as anti-obesity agents. Curr Opin Investig Drugs 2003; 4: 1190–1197.
55. Masaki T, Yoshimatsu H, Chiba S, Watanabe T, Sakata T. Targeted disruption of histamine H1-receptor attenuates regulatory effects of leptin on feeding, adiposity, and UCP family in mice. Diabetes 2001; 50: 385–391.
56. Mollet A, Lutz TA, Meier S, Riediger T, Rushing PA, Scharrer E. Histamine H1 receptors mediate the anorectic action of the pancreatic hormone amylin. Am J Physiol Regul Integr Comp Physiol 2001; 281: R1442–R1448.