Tamoxifen-Induced Anorexia Is Associated With Fatty Acid Synthase Inhibition in the Ventromedial Nucleus of the Hypothalamus and Accumulation of Malonyl-CoA

We investigated body weight changes in the RMH trial (3) (online appendix [available at http://diabetes.diabetesjournals.org]). The effects of TMX or placebo on body weight were analyzed in 2,329 subjects for whom BMI data were available and who did not develop cancer during the study. The trial was approved by the RMH Ethics Committee (3).

We used Wistar rats (450–500 g; Charles River Laboratories), lean Zucker rats [Crl:(ZUC)-faBR, 300–350 g; The Jackson Laboratories], and obese Zucker rats [Crl:(ZUC)-faBR, 475–525 g; The Jackson Laboratories] age-matched (11–12 weeks) with lean Zucker rats. Rats were housed in a temperature-controlled room, with a 12-h light/dark cycle (lights on at 8:00 a.m. and off at 8:00 p.m.). All experiments were conducted in accordance with Home Office Guidelines and the Ethics Committee of the University of Santiago de Compostela. Male rats were used to exclude the confounding effects of the estrous cycle on food intake (22,23).

TMX (Sigma) was dissolved in sesame oil containing 1% benzyl alcohol (5). The control group received a daily subcutaneous injection of vehicle and was provided standard rodent diet ad libitum. The TMX group received daily subcutaneous injections of TMX and was also provided ad libitum food. To monitor the effects of food intake, a pair-fed group was also included; for 5 days, the pair-fed group received the average amount of food eaten by the TMX-treated rats and received vehicle injections. For the hypothalamic studies, a dose of 0.5 mg · kg⁻¹ · day⁻¹ TMX was administered for 5 days. On day 5, rats were killed 4 h after TMX treatment. Zucker lean and obese rats (n = 10–12 animals/group) were also treated subcutaneously with TMX (0.5 mg · kg⁻¹ · day⁻¹) for 5 days. On the final day of the protocol, rats were given a final dose of TMX (9:00 a.m.) and killed 4 h later. To test the effect of TMX treatment on refeeding, rats were fasted for 24 h before TMX administration. Next, the rats (n = 9–10 animals/group) were treated with vehicle or TMX (0.5 mg/kg) and allowed to eat ad libitum. In all the protocols, TMX was administered subcutaneously at 9:00 a.m.

TMX citrate was used for intracerebroventricular experiments because it can be dissolved in aqueous solutions. Chronic (intracerebroventricular) cannulae were stereotaxically implanted under ketamine-xylacine anesthesia (50 mg/kg i.p.) in the lateral ventricle as previously described (24). Correct localization of the cannulae was confirmed by histological analysis. Before TMX citrate administration, rats (n = 9 animals/group) were fasted for 24 h in order to adjust the rats to the same pretreatment levels of stimulated appetite response. Rats received either a single administration of TMX citrate (10 μg in 10 μl 2-hydroxypropyl-β-cyclodextrin; Sigma) or vehicle (10 μl, control rats). Since there are no reported studies about intracerebroventricular treatments with TMX, we selected this dose on the basis of previous literature based upon intracerebroventricular administration of other estrogenic compounds (25). A third group of animals was treated with an equimolar dose of citrate (Sigma) to that of the TMX citrate animals.

Five days before the test, rats were allowed 2 h daytime access to water. On the day of the experiment, rats (n = 16 rats/group) were given access to 0.15% sodium saccharin rather than water for 30 min. Immediately afterward, one experimental group was injected intraperitoneally with 0.15 mol/l lithium chloride (LiCl) in saline (26,27) and subcutaneously with sesame oil containing 1% benzyl alcohol (vehicle). The second experimental group was injected intraperitoneally with saline and subcutaneously with TMX (0.5 mg/kg). A third experimental group (control group) was injected intraperitoneally with saline and subcutaneously with vehicle. Twenty-four hours later, rats were given 2 h access to a two-bottle choice test of 0.15% saccharin versus water. Data were expressed as percent saccharin preference ratio (100 × saccharin intake/[saccharin intake + water intake]).

For the intracerebroventricular-conditioned taste aversion studies, one group was injected intraperitoneally with 0.15 mol/l LiCl in saline and intracerebroventricularlly with vehicle (10% 2-hydroxypropyl-β-cyclodextrin); the second group was injected intraperitoneally with saline and intracerebroventricularlly with TMX citrate (10 μg). A third group (control group) was injected intraperitoneally with saline and intracerebroventricularlly with vehicle. The experiment proceeded as the subcutaneous treatment (n = 16 rats/group).

Rats were fasted for 24 h (n = 8–12 animals/group). To evaluate the effect of melanocortin receptors on the anorectic actions of TMX, the rats received a single intracerebroventricular administration of saline or the melanocortin-3/-4 receptors antagonist SHU9119 (3 nmol in saline; Phoenix) (28,29). To test the effect of the inhibition of ACC on the anorectic actions of TMX, the rats were treated with a single intracerebroventricular administration of vehicle (DMSO) or an ACC inhibitor: 5-(tetradecyloxy)-2-furoic acid (TOFA; 10 μg) (14). Subsequently, these rats received either vehicle or TMX (0.5 mg/kg) administered subcutaneously.

Kits for the measurement of cholesterol, glucose, and triglycerides were obtainded from Roche (21,30). Free fatty acids were measured using the NEFA-C colorimetric kit (Wako Chemicals) (21,30). Plasma leptin and insulin levels were measured by ELISA kits (Crystal) (21,30).

The mRNA levels of estrogen receptor β and progesterone receptor were studied by using real-time PCR (TaqMan) (online appendix Table 1 and online appendix methods) as previously described (30). We used six to eight rats per group.

Coronal brain sections (16 μm) were probed with specific antisense oligos (online appendix Table 2). In situ hybridizations were performed as previously published (24,31,32) (online appendix methods). We used 10–19 rats per group.

Diaminobenzidine immunohistochemistry was performed as described (33) (online appendix methods) using a c-FOS antibody (Santa Cruz). We used six to eight rats per group.

Hypothalamus total protein lysates were subjected to SDS-PAGE, electrotransferred on a polyvinylidine fluoride membrane, and probed with the indicated antibodies: ACC, pACC-Ser⁷⁹, AMP-activated protein kinase (AMPK)α1, and AMPKα2 (Upstate); pAMPK-Thr¹⁷² (Cell Signaling); and β-actin (Abcam). For protein detection, we used horseradish peroxidase–conjugated secondary antibodies and chemiluminescence (Amersham) (online appendix methods). We used 8–12 rats per group.

Hypothalamic protein extracts were incubated at 4°C with the corresponding antibody conjugated to protein G-Sepharose (34). Phosphotransferase activity toward the AMARA-peptide for AMPK activity (35) or LKB1tide-peptide (Upstate) was then measured (36) (online appendix methods). We used 8–12 rats/group.

Malonyl-CoA was measured by radioisotopic method in neutralized perchloric acid filtrates (37,38) (online appendix methods). We used 8–10 rats per group.

Data were expressed as means ± SE and analyzed by using StatView 4.57 (Abacus Concepts). Statistical significance was determined by ANOVA and post hoc Bonferroni test. P < 0.05 was considered significant. Protein and mRNA levels were presented as percentage change in relation to control group (vehicle treated or fed ad libitum).

Subcutaneous administration of TMX to rats markedly decreased food intake (Fig. 1A and online appendix Fig. 1A) and body weight (Fig. 1B and online appendix Fig. 1B). To determine whether the anorexigenic effect of TMX was dependent on feeding state, subcutaneous TMX was administered to 24-h fasted rats. TMX significantly attenuated refeeding at 6, 12, and 24 h (Fig. 1C).

We detect no differences in plasma glucose, triglycerides, and free fatty acids among the experimental groups (online appendix Table 3). Both TMX-treated and pair-fed rats showed significant decreases in insulin and leptin when compared with vehicle-treated rats (online appendix Table 3). Moreover, as it has been previously demonstrated (21,39), serum total, HDL, and LDL cholesterol were all reduced in the TMX group when compared with the ad libitum and pair-fed groups (online appendix Table 3).

Since TMX is insoluble in an aqueous solution, TMX citrate was used for intracerebroventricular experiments. We confirmed that TMX citrate subcutaneous administration, at an equimolar dose to TMX (0.5 mg · kg⁻¹ · day⁻¹), recapitulated the effects observed with TMX (online appendix Fig. 1C and D). Intracerebroventricular administration of TMX citrate to 24-h fasted rats decreased refeeding at 4, 12, and 24 h after treatment (Fig. 1D). This anorectic effect was specific for TMX because administration of citrate alone induced no changes in food intake (4 h: control, 14.2 ± 0.9 vs. citrate, 14.1 ± 1.2 g). Moreover, intracerebroventricular administration of TMX citrate did not affect plasma glucose, insulin, or leptin levels (online appendix Table 4), suggesting that the anorectic effect of TMX was not the result of peripheral secondary effects of TMX. Furthermore, we showed that subcutaneous administration of TMX at an equivalent dose to the intracerebroventricular treatment (10 μg i.c.v. = 25 μg · kg⁻¹ · day⁻¹ s.c.) does not affect feeding, body weight (online appendix Fig. 1E and F), plasma glucose, insulin, or leptin levels (online appendix Table 5). These data make any peripheral effect of centrally administered TMX diffusing from the central nervous system unlikely.

To confirm that TMX was bioactive at hypothalamic levels, we studied the mRNA levels of estrogen receptor β and progesterone receptor of TMX-treated rats. As previously reported, the mRNA expression of estrogen receptor β (40) and progesterone receptor (41) was increased in the hypothalamus after TMX treatment (online appendix Fig. 2). These results indicate that the anorexic action of intracerebroventricular TMX citrate is mediated by central mechanisms.

We evaluated whether the anorectic effect of TMX was mediated by conditioned taste aversion (Fig. 1E). As expected, LiCl-treated rats drank significantly less saccharin solution (LiCl-paired flavor), indicating that they developed conditioned taste aversion (26,27,42). Rats treated subcutaneously with TMX or intracerebroventricularly with TMX citrate showed significantly increased saccharine preference ratios compared with LiCl-treated rats. The preference ratios for subcutaneous TMX (55.0 ± 4.4) and intracerebroventricular TMX citrate (86.1 ± 8.5) did not show significant differences when compared with their respective controls (control for the subcutaneous experiment: 65.4 ± 6.0; control for the intracerebroventricular experiment: 96.0 ± 5.1). Additionally, the preference ratios for TMX and TMX citrate were >50%, indicating the absence of aversive effects (26,42).

TMX-treated rats recovered their normal feeding pattern after treatment was discontinued (Fig. 1F), indicating that TMX did not induce a permanent state of wasting (42). Furthermore, we did not observe changes in other commonly used parameters of health status: temperature (Fig. 1G), skin aspect, stool consistency, or obvious abnormal behavior, as was evident in the LiCl-treated animals. These data suggest that the powerful anorectic effect of TMX, when administered intracerebroventricularlly or subcutaneously, was not related to aversive effects, illness, or malaise.

We examined whether the anorexigenic effect of TMX could overcome the hyperphagic state associated with a defective leptin receptor signaling. Zucker rats were treated with TMX (0.5 mg · kg⁻¹ · day⁻¹) for 5 days. TMX markedly decreased food intake in obese Zucker rats (which lack a functional long form of leptin receptor) and their lean littermates (online appendix Fig. 3A and B) and prevented body weight gain in both animal models (online appendix Fig. 3C and D).

The levels of CART and proopiomelanocortin (POMC) mRNAs in the ARC were increased in TMX-treated rats when compared with pair-fed animals (Fig. 2A and B). The mRNA levels of both neuropeptides were similar to the levels in untreated fed animals, suggesting that these anorexigenic signals were inappropriately elevated for the level of feeding in TMX-treated animals. Agouti-related protein mRNA was unaffected (Fig. 2C). We detected increased levels of neuropeptide Y mRNA in the ARC (Fig. 2D) and melanin-concentrating hormone mRNA levels in the lateral hypothalamic area (LHA) (Fig. 2E) in the TMX-treated animals. These data suggest a compensatory upregulation of orexigenic neuropeptides against the reduction in feeding induced by TMX. Of note, these effects of TMX were specific since no changes were detected in the mRNA levels of CART in the PVN and in the LHA, corticotrophin-releasing hormone in the PVN, and thyrotrophin-releasing hormone in the PVN (online appendix Fig. 4).

Our data suggested that TMX could inhibit feeding through a mechanism involving the POMC neurons in the ARC. To test this hypothesis, we used SHU9119, a specific antagonist of the melanocortin-3 and -4 receptors (28,29). Intracerebroventricular administration of SHU9119 to rats before subcutaneous administration of TMX significantly reversed the anorectic response to TMX (Fig. 2F). SHU9119 did not have any effect on feeding when administered alone. These data suggest a direct effect of SHU9119 blocking TMX action on the melanocortin system, rather than a direct orexigenic effect. Similarly, these data indicate that POMC and melanocortin-4/-3 receptors are likely downstream mediators of TMX anorectic effects.

It has been recently shown that pharmacological inhibition of hypothalamic FAS inhibits feeding (13,19,20). We have recently reported that TMX decreases the expression and activity of FAS in liver (21). Using in situ hybridization analysis, we analyzed the FAS mRNA levels in hypothalami from rats treated subcutaneously with TMX (0.5 mg · kg⁻¹ · day⁻¹ for 5 days). We also included control groups of fasted rats (24 and 48 h) and refed rats (24 and 48 h). Our data revealed a 40% decrease in FAS mRNA levels in the VMN after TMX treatment (Fig. 3A and B). The expression of FAS in other hypothalamic nuclei (ARC, PVN) (Fig. 3C and D), as well as in the cortex, hippocampus, and thalamus (data not shown), was not affected. Hence, TMX specifically reduced FAS expression in the VMN. Furthermore, TMX exerted this effect centrally since intracerebroventricular administration of TMX citrate also decreased (37%) FAS gene expression in the VMN (Fig. 4A).

Our fasted and refed groups also revealed that FAS gene expression was specifically decreased in the VMN after 24 and 48 h of fasting and that this effect was reverted by refeeding (Fig. 3A and E). This physiological nutritional regulation of FAS expression was specific to the VMN since FAS mRNA expression in the ARC and PVN did not change in response to any of these conditions (Fig. 3F and G).

To further characterize the specificity of the effect of TMX in the VMN, we assessed the hypothalamic levels of c-fos expression, a marker of neuronal activation (13,27). Consistent with our FAS mRNA data (Fig. 3A–C), TMX-treated rats showed significantly increased c-fos immunoreactivity in the VMN, but not in the ARC, compared with control and pair-fed rats (Fig. 4B–D).

We found that the anorectic effect of TMX was associated with hypothalamic accumulation of malonyl-CoA (Fig. 5A). This effect was specific for TMX since the hypothalamic malonyl-CoA content in pair-fed and fasted rats was decreased (Fig. 5A and B). Levels of hypothalamic malonyl-CoA recovered after refeeding for 24 or 48 h (Fig. 5B). We also studied if the inhibitory effect of TMX on refeeding (Fig. 1C) correlated with hypothalamic malonyl-CoA content. Refed rats subcutaneously treated with TMX showed a further increase in hypothalamic malonyl-CoA content when compared with refed rats treated with vehicle (Fig. 5C). These results strongly suggest that a rise in malonyl-CoA levels may mediate the inhibitory effect of TMX on refeeding.

The control of malonyl-CoA synthesis is regulated to a large extent by ACC. The activity of ACC can in turn be controlled by AMPK (43). Therefore, parallel changes in FAS and AMPK/ACC could explain the presence of altered malonyl-CoA levels in our TMX-treated rats. However, our results showed no differences in the protein levels of AMPKα1, AMPKα2, or pAMPKα (Fig. 6A and B). Similarly, no differences in AMPKα1 and AMPKα2 activities (Fig. 6D and E) or in ACCα, ACCβ, and pACC levels (Fig. 6A and C) were detected. LKB1 activity was not affected by either TMX or pair-feeding (Fig. 6F). These data suggest that TMX selectively inhibits fatty acid synthesis at the level of FAS activity.

To establish a mechanistic link between the effects of TMX, inhibition of FAS, accumulation of malonyl-CoA, and anorectic effects, we investigated the effect of TOFA, an inhibitor of ACC. Intracerebroventricular administration of TOFA completely prevented the anorectic effect of subcutaneous TMX on food intake during a 24-h refeeding experiment (Fig. 7A). This dose of TOFA, when administered alone, did not have any effect on feeding, excluding the possibility of a direct orexigenic effect. Furthermore, administration of TOFA prevented a TMX-induced increase in malonyl-CoA, suggesting that alterations in malonyl-CoA levels mediate the anorectic effect of TMX (Fig. 7B).

The potential relevance of our data were assessed by reanalyzing data from the RMH trial (3). When subjects were stratified according to their body weight at the beginning of the treatment, we found that obese (BMI ≥30 kg/m²) women treated with TMX initially lost weight and remained significantly lighter than those receiving placebo at 3, 24, 48, and 72 months (Fig. 8).

TMX-treated rats experience anorexia and weight loss (5,6). However, the mechanism for this anorectic effect has not been identified. We have recently reported that TMX inhibits FAS gene expression and activity in rat liver (21). As pharmacological inhibition of hypothalamic FAS elicits an anorectic response through a malonyl-CoA–dependent mechanism (13,19,20), we hypothesized that decreased food intake in response to TMX may be the result of FAS inhibition and hypothalamic malonyl-CoA accumulation. Here, we demonstrated that TMX administration decreases FAS mRNA expression, specifically in the VMN, despite FAS being globally expressed in the brain in addition to other hypothalamic nuclei, like the ARC or PVN. This nucleus specificity of TMX was further supported by c-fos expression data that show specific VMN-neuron activation in response to TMX treatment. These data represent the first evidence of a nucleus-specific regulation of hypothalamic FAS mRNA associated with changes in food intake. Our fasting-refeeding experiments further supported a specific role for VMN and FAS expression in the regulation of feeding, since mRNA levels were only reduced in the VMN during fasting. These observations suggest that FAS in the VMN may act as a nutritional sensor to modulate food intake.

Next, we investigated whether TMX induced changes in hypothalamic malonyl-CoA. Although the decrease in FAS mRNA was restricted to just one nucleus, changes in malonyl-CoA levels were detectable when we analyzed the whole hypothalamus. It may be possible that inhibition of FAS activity could be a more generalized effect than mRNA regulation; however, it is also possible that the changes in malonyl-CoA levels in the VMN are significant enough to be detectable when the whole hypothalamus is assayed.

If FAS expression within the VMN has a role in regulating food intake, then the fact that FAS mRNA levels decrease in response to both fasting and TMX treatment is apparently contradictory. The explanation for this apparent paradox is found by examining malonyl-CoA levels. We demonstrate that the anorectic effect induced by TMX is similar to that of C75 in that it requires the accumulation of hypothalamic malonyl-CoA (14). The anorectic effect of TMX was reverted by pharmacological inhibition of ACC that prevented accumulation of malonyl-CoA. Thus, specific inhibition of FAS activity by TMX in the presence of maintained ACC activity (see below) facilitates accumulation of malonyl-CoA. Conversely, during fasting the decrease in FAS mRNA expression is associated with decreased levels of malonyl-CoA, suggesting that early steps of the hypothalamic lipogenic pathway are physiologically inhibited during fasting. Supporting this argument are recent reports demonstrating that during fasting, activation of hypothalamic AMPK phosphorylates and inhibits ACC (44,45).

We also investigated the effects of TMX on other components of the fatty acid biosynthetic pathway. AMPK plays a critical role in the hypothalamic regulation of feeding through its effects on the lipogenic pathway and neuropeptides (44,45). In situations of energy demand such as fasting, AMPK is phosphorylated by LKB1 at Thr¹⁷² (and then activated), resulting in phosphorylation and subsequent inhibition of ACC, thereby inactivating the fatty acid biosynthetic pathway (43). Under conditions of energy surplus, ACC is not phosphorylated, facilitating production of malonyl-CoA (43,45). Our data showed that FAS inhibition induced by TMX is not associated with changes in either AMPK protein levels/activity or phosphorylated ACC levels. These data suggest that the hypothalamic AMPK/ACC axis is not affected by TMX. Hence, the accumulation of malonyl-CoA induced by TMX is the result of specific FAS inhibition in the presence of normal ACC activity (online appendix Fig. 5).

Food restriction is associated with reduction in the mRNA levels of CART and POMC in the ARC (7–9,12). This expected reduction in CART and POMC levels was blocked by TMX. Furthermore, pharmacological inhibition of melanocortin-3/-4 receptors by SHU9119 prevented the TMX anorectic effects, indicating that inhibition of feeding by TMX involves the melanocortin system. CART/POMC neurons receive neural projections and excitatory inputs from neurons in the VMN (46,47); therefore, changes in FAS/malonyl-CoA in the VMN might be linked to CART/POMC changes in the ARC. However, this mechanism still needs to be clarified.

TMX is a drug widely used for the treatment of estrogen receptor–positive breast cancers (3,4). Despite its widespread use, the action of TMX on human body weight is controversial. Data from the WHEL (Women’s Healthy Eating and Living) clinical trial show that obese women with breast cancer under TMX treatment had a decreased BMI (48). However, measurement of BMI in this study occurred at varying times after breast cancer diagnosis and thus did not necessarily reflect prediagnosis BMI. The results of this study may also have been confounded by the concurrent diagnosis of cancer. Hence, we reanalyzed the body weight data from the RMH trial (3). Since these subjects were free of cancer during the study, changes in body weight are more likely to be a direct effect of TMX. We found that obese (BMI ≥30 kg/m²) women treated with TMX were significantly lighter than those receiving placebo. Although this is a post hoc analysis of the data, we suggest that the large size of the cohort studied strongly supports the finding that low-dose TMX can induce a sustainable body weight decrease in obese humans. Although the RMH trial was not an obesity prevention/treatment trial, the finding that the obese women lost weight initially for 4 years and on average had maintained their initial weight by the end of the study is a highly significant deviation from the expected (49,50).

In conclusion, this study demonstrates that the anorectic actions of TMX involve a specific decrease in FAS mRNA levels in the VMN and hypothalamic malonyl-CoA accumulation. Specific regulation of FAS in the VMN by TMX or fasting suggests that fatty acid metabolism in this nucleus may play a physiological role in feeding control. Altogether, our data indicate that altering FAS in the VMN may be a suitable strategy for altering food intake and body weight homeostasis in experimental animals and obese subjects.

This work has been supported by grants from the Medical Research Council (to A.J.V.-P.), the Welcome Trust (to A.J.V.-P.), the Spanish Ministry of Education (to C.D.), the Xunta de Galicia (to C.D.), and the European Union (LSHM-CT-2003-503041) (to C.D.). A.K.S. is funded by the U.S. Public Health Service (DK 19514) and a grant from the Juvenile Diabetes Research Foundation. M.L. is funded by the Marie Curie Program (QLK6-CT-2002-51671).

We are grateful to Keith Burling, Mark Campbell, and Janice Carter for their technical assistance, to Dr. Anthony P. Coll (University of Cambridge, Cambridge, U.K.), to Dr. Benjamin Challis (University of Cambridge, Cambridge, U.K.), to Prof. Manuel Ros (University Rey Juan Carlos, Madrid, Spain), and to Prof. Manuel Tena-Sempere (University of Cordoba, Cordoba, Spain) for their useful comments and criticism. We thank Prof. Len Storlien (AstraZeneca, Mondal, Sweden) for supplying us TOFA and Prof. Dario Alessi (University of Dundee, Dundee, U.K.) for the LKB1 antibody.