Weight regain after sustained weight reduction is accompanied by suppressed oxidation of dietary fat and adipocyte hyperplasia

Matthew R. Jackman, Amy Steig, Janine A. Higgins, Ginger C. Johnson, Brooke K. Fleming-Elder, Daniel H. Bessesen, Paul S. MacLean


A dual-tracer approach (dietary 14C-palmitate and intraperitoneal 3H-H2O) was used to assess the trafficking of dietary fat and net retention of carbon in triglyceride depots during the first 24 h of weight regain. Obesity-prone male Wistar rats were allowed to mature under obesogenic conditions for 16 wk. One group was switched to ad libitum feeding of a low-fat diet for 10 wk (Obese group). The remaining rats were switched to an energy-restricted, low-fat diet for 10 wk that reduced body weight by 14% and were then assessed in energy balance (Reduced group), with free access to the low-fat diet (Relapse-Day1 group), or with a provision that induced a minor imbalance (+10 kcal) equivalent to that observed in obese rats (Gap-Matched group). Fat oxidation remained at a high, steady rate throughout the day in Obese rats, but was suppressed in Reduced, Gap-Matched, and Relapse-Day1 rats though 9, 18, and 24 h, respectively. The same caloric excess in Obese and Gap-Matched rats led to less fat oxidation over the day and greater trafficking of dietary fat to visceral depots in the latter. In addition to trafficking nutrients to storage, Relapse-Day1 rats had more small, presumably new, adipocytes at the end of 24 h. Dietary fat oxidation at 24 h was related to the phosphorylation of skeletal muscle acetyl-CoA carboxylase and fatty acid availability. These observations provide evidence of adaptations in the oxidation and trafficking of dietary fat that extend beyond the energy imbalance, which facilitate rapid, efficient regain during the relapse to obesity.

  • acetyl-CoA carboxylase
  • metabolic inflexibility
  • postobese
  • adipocyte cellularity

over 60% of adults in the united states are now overweight or obese, and close to 17% of children and adolescents are also afflicted (42). Our current therapeutic strategies have minimal success for long-term weight reduction (10), as weight regain continues to plague efforts to maintain the weight-reduced state. There is growing evidence to suggest that weight regain after weight loss is due, at least in part, to metabolic adaptations in the homeostatic control system that regulates body weight (18, 27, 31). The most prominent of these adaptations is a large gap between appetite and expenditure (9, 12, 19, 26, 47). We have used a rodent experimental paradigm modeling human weight regain to show that this energy gap expands as the time-of-weight reduction is extended with an energy-restricted, low-fat diet, that the energy gap is greatest early in the relapse process, and that it does not completely resolve even after the lost weight has been regained (3133). Failure to maintain a restriction on intake exposes peripheral tissues to a large excess of calories that must be either oxidized or stored. Clarifying how these ingested fuels are metabolized during relapse is critical to our understanding of this persistent energy gap, as these nutrients feed back into the neural centers of the brain and can affect the regulation of appetite and energy expenditure (58).

Weight reduction leads to an improvement in whole body insulin sensitivity and a concordant enhancement in the regulation of both carbohydrate (CHO) and lipid metabolism (23, 24). Although this improvement in metabolic flexibility is considered to be beneficial from a health perspective, some have speculated that it may facilitate the process of weight regain by promoting the trafficking of ingested fuels to adipose tissue (53, 59). Consistent with this assertion, weight-stable, weight-reduced subjects have suppressed postprandial and 24-h fat oxidation (24, 25, 45), even though clearance of dietary fat is enhanced (15, 54). These data suggest that in the weight-reduced state, CHO becomes the preferred fuel for energy needs and that dietary fat is diverted to adipose tissue storage. If sustained, this would not only promote energetically efficient repletion of adipose tissue, but it may also promote subsequent meal initiation by depleting CHO reserves. At present, it has not been confirmed that this shift in fuel utilization persists during the relapse to obesity, when the energy gap of weight-reduced subjects is realized, and they experience a large caloric excess. Moreover, it is unclear that if this preferential use of fuel persists during relapse, whether it is a result of peripheral adaptations to weight reduction (i.e., enhanced metabolic flexibility) or simply a consequence of excess calories.

In the present study, we have used an experimental paradigm modeling human weight regain to examine the pattern of intake, absorption, oxidation, and trafficking of dietary fat in obese, weight-reduced, and relapsing rats, over 24 h. In one group of relapsing rats, we limited the food provision, so that their positive energy imbalance matched the energy gap in the obese (Gap-Matched) rats. This comparison allowed us to examine the differences in fuel metabolism over a 24-h period when obese and weight-reduced subjects experience the same caloric excess. These studies show that sustained weight reduction results in a dynamic, fluctuating state of metabolism consistent with an enhancement in metabolic flexibility, reflected concomitantly in the pattern of food intake, the availability of exogenous fuels, and the oxidation and trafficking of ingested fat. The shift in fuel metabolism is a consequence not only of the overfeeding, but also of enhanced regulation of CHO and fat oxidation that likely involves acetyl-CoA carboxylase (ACC) in skeletal muscle. When the large energy gap is realized during the relapse to obesity, peripheral tissues preferentially oxidize CHO for energy needs, traffic ingested fat toward storage, and favor the repletion of visceral adipose tissue.


Experimental paradigm of weight regain.

Seventy-two male Wistar rats (125–150 g) were purchased from Charles River Laboratories (Charles River Laboratories, Wilmington, MA), from which 24 obesity-prone rats were identified by their weight gain response to a high-fat diet (HF; 46% kcal fat) (Research Diets, New Brunswick, NJ; RD# 12344), a response that has previously been shown to predict future weight gain (21). Obesity-prone rats were then matured in obesogenic conditions (free access to HF diet, housed individually in cages that limit activity) for 16 wk to induce obesity (∼35% body fat). Eighteen obese rats were then placed on an energy-restricted, low-fat diet (LF; 12% kcal fat, Research Diets; RD# 11724), equal to ∼60% of the calories eaten ad libitum by obese rats, to induce weight loss. This weight loss regimen was maintained for ∼2 wk, causing a 10–15% loss in body mass. For 8 subsequent weeks, this reduced weight was maintained by providing a limited provision of the LF diet at the beginning of each dark cycle. The remaining six obese rats were switched to LF diet ad libitum for 10 wk. The obese and weight-reduced rats then began the dual-tracer protocol. Rats were individually housed in the University of Colorado Denver Center for Comparative Medicine and the Center for Human Nutrition Satellite Facility (22–24°C; 12:12-h light-dark cycle) with free access to water. All procedures were approved by the UCDHSC Animal Care and Use Committee.

Intake, expenditure, and fuel utilization.

A metabolic monitoring system was used to assess energy balance and fuel utilization. This system was developed by the Energy Balance Core Laboratory at the University of Colorado Clinical Nutrition Research Unit, as described previously (33). It consists of a multichamber indirect calorimeter designed for the continuous monitoring of up to four rats simultaneously, obtaining measurements of V̇o2 and V̇co2 from each chamber every 6 min. Chambers allow for the collection of urine, feces, and food spillage. Metabolic rate (MR) was calculated with the Weir equation (MR = 3.941·V̇o2 + 1.106·V̇co2 −2.17·N) (55), and extrapolated throughout 24 h to acquire estimates of total energy expenditure (TEE). Respiratory exchange ratio (RER) was calculated as the ratio of CO2 production to O2 consumption (V̇co2/V̇o2). Before the tracer protocol, obese and weight-reduced rats were examined while in energy balance, so that estimates of whole body substrate oxidation could be calculated from V̇o2, V̇co2, and from measurements of urinary nitrogen (N), using derivations of Weir's equation: CHO disappearance = (4.57·V̇co2) − (3.23·V̇o2) − (2.6·N); lipid disappearance = (1.69·V̇co2) − (1.69·V̇co2) − (2.03·N); protein disappearance = 6.25·N. While in energy balance, calculations of substrate disappearance are a good reflection of substrate oxidation. The net disappearance of substrate via ketogenesis, lipogenesis, and gluconeogenesis without subsequent oxidation would be minimal and, therefore, should not affect measurements of V̇o2 and V̇co2 (17).

Experimental design and dual-tracer protocol.

Rats were placed in the monitoring system for three days before the tracer study to acclimatize them to the environment and to characterize aspects of energy balance [energy intake (EI) and TEE] and RER. The dual-tracer approach was loosely based upon the work of Commerford et al. (8) and is diagrammed in Fig. 1. Two hours before the beginning of the third day, an intraperitoneal injection of 250 μCi 3H2O was given. This injection equilibrates with body water within 2 h and remains steady over the next 24-h period (8). Incorporation of tritium into the lipid and glycogen pools provides an estimate of the net retention of carbons into these pools but does not provide any information as to the source of those carbons. A 1-[14C]palmitate tracer was blended with the LF diet to achieve a specific activity of 2.0 μCi/kcal of dietary lipid. This tracer allowed us to examine the trafficking and oxidation of ingested fat during the testing period. Obese rats had ad libitum access to this diet during the 24-h monitoring period. A group of weight-reduced rats were given a limited provision of the diet that matched their energy expenditure (Reduced rat group), to examine the animals in the weight-reduced state. A second group of weight-reduced rats were provided free access to the diet (Relapse-Day 1 rat group), to examine animals during the initial day of the relapse to obesity. During the lead-in period, we observed that Obese rats experienced a modest positive energy imbalance. This persistent energy gap between appetite and expenditure averaged ∼10 kcal/day, and the third group of weight-reduced rats was given a provision of food that would induce the same positive energy imbalance (Gap-Matched rats). To account for a small induction of thermogenesis from the overfeeding, we were required to add a total of ∼12 kcal over their weight-reduced energy expenditure. As indicated in Fig. 2A, we achieved an average imbalance of 9.1 ± 1.0 kcal/day in Gap-Matched rats. Matching the energy gap of Obese rats allowed us to examine how an obese rat and a relapsing rat differentially ingest and metabolize nutrients when overfeeding, given an equivalent caloric excess.

Fig. 1.

Experimental design for the dual-tracer study. A metabolic monitoring chamber was used to assess the fuel utilization and lipid trafficking of exogenous fat during the first day of weight regain. A single injection of 3H-H2O was given 2 h prior to the study to allow time for its equilibration in total body water. A 14C-palmitate tracer was incorporated into the low-fat diet with a specific activity of 2μCi/kcal fat. The absolute amount of 14C ingested was reflective of the absolute amount of dietary fat, and therefore total calories, consumed over the 24-h period. O2 consumption and CO2 production were monitored every 6 min, and CO2 was trapped at regular intervals to assess oxidation rates of “ingested” lipid. Urine was collected and analyzed to estimate protein disappearance for the dark (shaded bar in timeline) and light (open bar in timeline) cycles. At the end of the 24-h period, the rats were euthanized and tissues were examined for the amount of 14C and 3H in each lipid extracts. Bolded lines indicate continuous monitoring. Bolded hatches indicate discrete collections and/or measurements.

Fig. 2.

Energy balance, gap-matching, and the ingestion and absorption of dietary fat. Data related to energy balance during the 24-h tracer experiment are shown for the four groups of rats and are expressed as means ± SE. A: difference between intake and expenditure is indicated above the bars. When no asterisk is present, the imbalance represents an energy gap between intake and expenditure under ad libitum conditions. When an asterisk is present, a limited provision of the low-fat diet is provided to maintain energy balance (Reduced rat group) or to match the energy gap of obese rats (Gap-Matched rat group). B: diurnal energy intake (EI) and total energy expenditure (TEE) is expressed for the 12-h dark cycle and the 12-h light cycle, with the average energy balance indicated above the EI-TEE pairs. C: fat intake (a direct reflection of total energy intake) was tracked every 3 h.

Although the different groups were provided different amounts of food, and thus different amounts of the 14C tracer, the specific activity of the diet was constant. The absolute amount of tracer consumed reflected the amount of dietary lipid ingested over the 24-h period. EI was assessed every 3 h, while O2 consumption and CO2 production was monitored every 6 min by indirect calorimetry. When not being monitored by the calorimetry system, an inline stopcock was used to draw off effluent CO2 from the individual chambers over 4.5-min intervals in 3.0-ml aliquots of a 2:1 mixture of methanol and methylbenzethonium (hyamine) hydroxide (Sigma). The 14C content of these samples was then measured with a Beckman LS6500 scintillation counter. Background activity, determined by counting a sample containing only scintillation fluid and hyamine hydroxide, was subtracted from experimental values. Urine was collected at 12 and 24 h for the estimation of protein disappearance.

At the end of the 24-h period, rats were anesthetized with isoflurane. A sample of blood was obtained from the vena cava, and rats were then euthanized via exsanguination. Tissues were collected for determination of 14C and 3H accumulation in the lipid fraction of gastrocnemius muscle, liver, and in inguinal, mesenteric, epididymal, and retroperitoneal adipose tissues. To determine total tracer content within the gastrointestinal tract, the entire gastrointestinal (GI) tract was removed, stripped completely of mesenteric fat, weighed, and homogenized in 0.9% saline. Duplicate samples (0.25 ml) of the homogenate were then digested with 0.5 ml of tissue solubilizer overnight (Solvable, NEN) at 50°C and bleached with hydrogen peroxide, and the 14C content was determined by scintillation counting. Serum 14C content was determined in a similar manner. 14C and 3H content within liver, muscle, and the individual fat pads was determined after extraction of lipid with chloroform-methanol (2:1, vol/vol). Phases were separated with the addition of H2SO4 and centrifugation. The lower phase was collected and allowed to dry overnight under nitrogen, and 14C and 3H content within the lipid fraction was measured by scintillation counting. An additional portion of the individual tissues were treated similarly, and total triglyceride content was then determined from glycerol release following acid hydrolysis (Kit 320-a Sigma, St. Louis, MO).

The 14C and 3H content within the lipid fraction of each tissue was calculated from the measured disintegrations/min (dpm) per gram of lipid/gram of tissue multiplied by the total weight of the tissue. Serum 14C content was calculated as the measured 14C activity/ml of serum × 0.0385 (%body mass accounted for by serum) × body mass as previously described (6). Total body skeletal muscle tracer content within lipid was calculated by multiplying the 14C and 3H contents/gram lipid of the gastrocnemius by its lipid content and extrapolating this to an estimate of total skeletal muscle calculated as 61% of fat-free mass, as determined by dual-energy X-ray absorptiometry (DXA). The 14C and 3H content in visceral adipose tissue was estimated by the sum of 14C and 3H contained in lipid extracts of retroperitoneal (RP), mesenteric (MES), and epididymal (EPI) fat pads. The 14C and 3H content in subcutaneous adipose tissue was estimated by extrapolating the content in lipid extracts of inguinal fat pads to an estimate of subcutaneous adipose mass, which was determined as the difference between fat mass and the summed weight of the visceral pads.

Body composition analysis.

Body composition was measured by DXA using the Lunar DPX-IQ (GE Lunar, Madison, WI), with Lunar's Small Animal Software ver. 1.0. Corrected fat mass and fat free mass were calculated from DXA data and carcass weights, as previously described (31).

Adipocyte cellularity.

RP, MES, and EPI fat pads were removed and weighed. A portion of RP, MES, and subcutaneous inguinal pads from each rat was immediately processed for cellularity by methods described previously (31). In short, a small portion of the pad was digested with collagenase and adipocytes (150–250 cells/sample) were imaged using an Olympus BX60 microscope and a C-mounted Canon Power Shot G5 digital camera. Images were processed with Matlab software (courtesy of Dr. Williams Betts, UCHSC). The number of cells per fat pad was calculated with the average diameter, a density conversion factor (0.915 g/cc), and the mass of the fat pads (38).

Metabolite analyses.

Urinary nitrogen and creatinine were measured by a colorimetric assay (Thermo Electron, Melbourne, Australia). Plasma was assayed for insulin and leptin with commercially available ELISAs (ALPCO Diagnostics, Windam, NJ). Glucose, free fatty acid (FFA), total cholesterol, and triglycerides (TGs) were measured via colorimetric analysis (34). Liver and skeletal muscle glycogen were determined using the method described by Chan and Exton (7). 3H and 14C counts associated with liver and skeletal muscle glycogen were determined on a portion of the extracts used for determination of tissue glycogen.

Western blots of AMPK and ACC.

Gastrocnemius homogenates were prepared in cold lysis buffer (1:10 wt/vol) containing 20 mM Tris·HCl (pH 7.4), 150 mM NaCl, 1% IPEGAL, 20 nM NaF, 2 mM EDTA, 2.5 nM NaPP, 20 mM beta glycerophosphate, 10% glycerol, 4 mM PEFA Bloc SC, 143 mM complete mini-EDTA-free and 4 mM phosphatase inhibitor cocktail 2. Samples were separated by 10% SDS PAGE (Bis-Tris NUPAGE; Invitrogen, Carlsbad, CA) and transferred to a PVDF membrane (Bio-Rad Laboratories, Hercules, CA). The membranes were probed for phospho-AMPK (p-AMPK, Th-172), total AMPK, phospho-ACC (p-ACCα/β, Ser-79), and total ACCα/β (Cell Signaling Technology, Danvers, MA), or GAPDH (Santa Cruz Biotechnologies, Santa Cruz, CA). A conjugated IgG-horseradish-peroxidase secondary antibodies and enhanced chemiluminescence reagent (Amersham Biosciences, NJ) were used to detect binding via film. Bands were visualized using an HC Precision Scan Pro and quantified using Quantity One Analysis software (Bio-Rad Laboratories).

Statistical analysis.

Data were analyzed using SPSS software ver. 15.0 by ANOVA with Duncan's post hoc analysis for homogeneous groups whenever a significant main effect was observed. Pearson and partial correlation coefficients were calculated to examine the relationships between the rate of fat oxidation, factors in skeletal muscle involved in regulating fat oxidation, and fatty acid availability. A multivariate linear regression model was used to examine the combined contribution of several of these parameters to the rate of dietary fat oxidation at 24 h. Statistical significance was assumed when P < 0.05.


Body composition.

Body composition data are depicted in Table 1. Obesity-prone rats matured under obesogenic conditions of ad libitum feeding of a high-fat diet with limited physical activity and achieved a weight of 690 ± 11 g. In response to the energy-restricted low-fat diet, the rats lost 98.0 ± 6 g, or 14 ± 1%, of their body weight. Obese rats continued to gain weight, but at a reduced rate after being switched to ad libitum feeding of the LF diet. After 2 wk of weight loss and eight subsequent weeks of weight maintenance, the Reduced rats were randomly assigned to one of three groups, 1) Reduced, 2) Relapse-Day 1, or 3) Gap-Matched (matched to the energy gap of Obese rats), for the dual tracer experiment.

View this table:
Table 1.

Morphometric characteristics

Energy balance and gap-matching.

Daily EI and TEE are depicted in Fig. 2A. Obese rats exhibited an average caloric excess of 10.0 ± 3.9 kcal/day. Reduced rats were to remain in energy balance for the tracer protocol, so the food provision (EI) matched the energy requirements (TEE) of these animals. In ad libitum-fed Relapse-Day 1 rats, intake was 130 ± 5 kcal, inducing a positive energy imbalance of 54 kcal, roughly five times greater than that observed in Obese rats. To compare the consumption and metabolism of ingested fat of Obese and relapsing rats without the influence of this difference in their energy imbalance, Gap-Matched rats were given a limited provision of the LF diet so that the difference between their intake and expenditure (9.1 ± 0.7 kcal) matched the energy gap in Obese rats. EI and TEE in the light and dark cycles for each group are depicted in Fig. 2B. Obese rats consumed their food in regular amounts throughout both dark and light cycles, while Reduced, Relapse-Day 1, and Gap-Matched rats exhibited a biphasic feeding pattern in which they fed at a high rate until all of their provision was gone or until their own feedback systems (around 12 h) prevented further consumption for the remainder of the 24-h period. The time course of this pattern of EI is shown in Fig. 2C and is expressed as cumulative intake of dietary fat. A greater portion of ingested fat remained in the gut of Obese and Relapse-Day 1 rats (∼19%) at the time of death than remained in the gut of Reduced and Gap-Matched rats (∼8%).

Oxidation of dietary fat.

The oxidation rate of dietary fat in Obese rats increased over the first 6 h and then remained relatively stable over the remainder of the 24-h study period (Fig. 3A). In contrast, the rate of fat oxidation was suppressed during the first 9 h in Reduced rats (P < 0.001), during the first 18 h in Gap-Matched rats (P < 0.001), and during the entire 24-h period in Relapse-Day 1 rats (P < 0.001). The propensity of weight-reduced and relapsing rats to suppress fat oxidation during their dark cycle, when most of their dietary fat was consumed, was apparent whether expressed in absolute terms (Fig. 3B) or as a percentage of total dietary fat absorbed (Fig. 3C). A critical observation in this analysis was that less fat was oxidized in Gap-Matched rats than in Obese rats, even though they experienced an equivalent positive energy imbalance and metabolizeable energy.

Fig. 3.

Dietary fat oxidation and respiratory exchange ratio. The oxidation of dietary fat and whole body fuel utilization are expressed as means ± SE, and bars with the same letter designation are not significantly different (P < 0.05). A: oxidation of dietary fat was assessed by measuring 14C-CO2 in expired air over 4.5 min at each time point, while consuming a diet in which the 14C-palmitate tracer reflects the caloric amount of dietary lipid (2μCi/kcal). B: regular measurements of the rate of dietary fat oxidation were extrapolated to determine the absolute amount of ingested fat oxidized during the dark and light cycles. C: data in B were adjusted for the fat that was not absorbed by the end of the study. The percentage of absorbed fat that was oxidized in the 12-h dark cycle and 12-h light cycle is shown. D: respiratory exchange ratio (RER; V̇co2/V̇o2) derived from indirect calorimetry measurements is expressed for the 12-h dark and 12-h light cycle.

To pursue a better understanding of this difference in fat metabolism, the phosphorylated and total protein of acetyl-CoA carboxylase (ACCα/β) and AMP-activated protein kinase (AMPKα) were examined in skeletal muscle homogenates (Fig. 4, A and B). Total ACC was similar, but the p-ACC (inactive) was lowest in the Relapse-Day 1 animals. The ratio of p-ACC to total ACC was significantly lower in Relapse-Day 1 rats, and this ratio was significantly related to the rate of dietary fat oxidation at the end of the study (Fig. 4C). Compared with all other groups, Relapse-Day 1 animals had a 33% lower p-AMPK/GAPDH (P < 0.05) and tended to have a lower p-AMPK/total AMPK ratio than the rest of the animals (P = 0.08).

Fig. 4.

ACC, AMPK, and the rate of dietary fat oxidation. A: representative Western blots of total and phosphorylated forms of AMP-activated protein kinase α (AMPK and p-AMPK) and acetyl-CoA carboxylase α/β (ACC and p-ACC) are shown. B: phosphoprotein/total protein ratios for AMPK and ACC are expressed as means ± SE. Groups with a similar letter designation are not significantly different (P < 0.05). The rate of dietary fat oxidation at hour 24, measured from 14C-CO2 in expired air over 4.5 min was significantly related to the p-ACC/total ACC ratio (C) and to plasma free fatty acid (FFA) (D).

Whole body fat oxidation.

Before the tracer protocol, the diurnal pattern of whole body fat oxidation was significantly different between Obese (10 ± 1 kcal/12-h dark cycle vs. 11 ± 1 kcal/12-h light cycle) and Reduced (5 ± 1 kcal/12-h dark cycle vs. 13 ± 2 kcal/12-h light cycle). Both an effect of light cycle (P < 0.001) and an interaction between group and light cycle (P < 0.001) were observed. These data indicate that, as with the oxidation of ingested fat, whole body fat oxidation is relatively constant throughout the day in Obese rats, whereas Reduced rats exhibit a cyclical pattern of suppressed and enhanced fat oxidation. This differential diurnal pattern of fuel utilization between Obese and Reduced rats was also observed for RER on the day of the tracer protocol (Fig. 3D). For Reduced animals that are in energy balance, this diurnal fluctuation in RER reflects substrate oxidation, even though there is net retention of tracer into lipid pools (Tables 2 and 3).

View this table:
Table 2.

3H incorporation into tissue lipid

View this table:
Table 3.

Tissue distribution of ingested fat in lipid pools

In contrast to Reduced rats, the other groups were in positive energy balance, making it difficult to estimate whole body fat oxidation from V̇o2, V̇co2, and N. De novo lipogenesis and the storage of dietary fat in Relapse-Day 1 rats are likely to be significant confounders, as the net retention of both tracers is more than twice as high than in Reduced rats (Tables 2 and 3). Because we have not measured the accumulation of lipid from de novo lipogenesis in absolute terms, we cannot accurately calculate whole body fat oxidation in these groups. However, it may be reasonable to assume that the relative trends in the oxidation of dietary fat in Fig. 3, AC are a good reflection of the relative trends in whole body fat oxidation. The suppression of whole body fat oxidation and the induction of lipogenesis would, in combination, explain the elevated RER (Fig. 3D) in Relapse-Day 1 rats that surpasses 1.0 during the light cycle on the tracer day (17). This deviation of substrate disappearance from substrate oxidation is much less pronounced in Obese and Gap-Matched rats, because there is less lipid accumulation in the body and a lower retention of dietary fat via de novo lipogenesis. In these groups, the diurnal pattern of RER is more consistent with oxidation of the dietary tracer.

Humoral characteristics and tissue glycogen levels.

Several plasma metabolites and endocrine factors, as well as hepatic and muscle glycogen levels, are shown in Table 4. Obese and Relapse-Day 1 rats had higher insulin, leptin, and triglycerides than Reduced or Gap-Matched rats. FFAs and TGs represent available substrates for skeletal muscle fat oxidation and adipose tissue uptake and storage. Reduced and Gap-Matched rats had elevated FFA and suppressed TGs compared with Obese rats. In contrast, Relapse-Day 1 rats had suppressed FFA and similar TGs compared with Obese rats. The oxidation of dietary fat at the end of the study was related to insulin (r = −0.58, P < 0.005), plasma FFA (r = 0.57, P < 0.005) (Fig. 4D), TGs (r = −0.74, P < 0.001), and leptin (r = −0.56, P < 0.05). Hepatic glycogen levels were suppressed in Reduced and Relapse-Day 1 rats compared with Obese and Gap-Matched rats.

View this table:
Table 4.

Humoral characteristics and tissue glycogen content

Trafficking to storage, a preference for visceral adipose tissue.

An examination of the trafficking of dietary fat and net retention of ingested fuel indicates that relapse is characterized by the preferential repletion of adipose tissue. The concentration of dietary fat in adipose tissue at the end of the experiment was lower in both subcutaneous and visceral adipose tissue compartments in Obese rats, compared with the weight-reduced and relapsing rats (Fig. 5A). These data suggest that the pathways involved in the uptake and/or storage of circulating lipid are likely induced to a greater extent in the depots of Reduced, Relapse-Day 1, and Gap-Matched rats, compared with Obese rats. The concentration of 3H was also higher in the visceral adipose depots of these rats (data not shown), suggesting that the pathway of de novo lipogenesis was also induced in these depots or in other tissues, such as the liver, from which the synthesized lipid is subsequently delivered.

Fig. 5.

Net retention of dietary fat in visceral and subcutaneous adipose lipid. Net retention of dietary fat was examined in lipid extracts of visceral (RP, MES, and EPI pads) and subcutaneous (inguinal pad) adipose tissue. Data in each panel are expressed as means ± SE. A: net retention of 14C in the depots is expressed as calories of dietary fat per gram of depot lipid. B: net retention of 14C in the depots is expressed as kcalories dietary fat retained in all visceral and subcutaneous (fat mass and visceral fat) adipose tissue within the animal. Groups with similar letter designations are not significantly different, P < 0.05.

Extrapolating these data to the amount of accumulation in the entire body revealed a greater amount of trafficking of dietary fat to subcutaneous adipose tissue of Obese rats compared with Reduced rats (Fig. 5B), because the overall mass of this depot was more than twice as high (Table 1). In contrast, the mass of visceral adipose tissue was about twice as high in Obese rats, but the absolute amount of dietary fat trafficked to this depot was similar to their Reduced counterparts. This preferential trafficking to visceral adipose tissue was more apparent in Gap-Matched rats that experienced the same caloric excess as the Obese, while Relapse-Day 1 rats exhibited an elevated retention of dietary fat in both visceral and subcutaneous compartments compared with the Obese. When expressed as a percentage of the dietary fat that was absorbed, this preferential trafficking to visceral fat was still apparent (Table 3). A similar preference for retaining substrate in the visceral compartment via de novo lipogenesis was observed in weight-reduced and relapsing rats (Table 2), even though the overall mass of this tissue was greater in the Obese (Table 1). Obese rats exhibited a larger subcutaneous/visceral ratio of the net retention of 3H to adipose tissue lipid pools than Reduced, Relapse-Day 1, or Gap-Matched rats.

Adipose tissue cellularity and weight regain.

RP pads from Relapse-Day 1 rats exhibited a larger number of cells per pad and a smaller average diameter than the other three groups (Fig. 6A) and had a greater frequency of adipocytes that were less than 20 μm in diameter (Fig. 6B). Besides those cells less than 20 μm in diameter, all groups exhibited a relatively normal distribution in the size frequency in RP adipose tissue, with the zenith appearing between 60 and 100 μm for Reduced, Relapse-Day 1, and Gap-Matched rats, and a zenith between 100 and 120 μm for Obese rats. Obese rats had a greater frequency of cells that were over 100 μm in diameter and fewer between 60 and 80 μm in diameter compared with all other rats. Weight-reduced rats, regardless of their relapse state, had a smaller average MES diameter (79 ± 5 vs. 58 ± 3 μm, P = 0.0001) than Obese rats, had a greater number of MES cells that were less that 40 μm in diameter (P < 0.01) and tended to have a greater number of MES adipocytes (71 ± 6 vs. 53 ± 11 × 106 cells/pad, P = 0.14). In contrast to these visceral depots, the subcutaneous depot had a substantial number of adipocytes that were less than 40 μm in diameter in all groups, and there was no apparent elevation in the number of small adipocytes in any of the weight-reduced or relapsing rats (Fig. 6C).

Fig. 6.

Cellularity profiles of visceral and subcutaneous adipose tissue. A: cell diameter and cell number per pad are shown for the RP depot. Groups with different letter designations are significantly different in cell size (P < 0.05), while groups with different symbols are significantly different for cell number per pad (P < 0.05). The size distribution frequency in 20-μm increments is presented for RP adipose tissue (B) and inguinal subcutaneous adipose tissue (C). Significant differences within each size range are indicated with letter designations, P < 0.05: aObese vs. Reduced; bObese vs. Relapse Day 1; cObese vs. Gap-Matched; dReduced vs. Relapse Day 1; eReduced vs. Gap-Matched; and fRelapse Day 1 vs. Gap-Matched.

Correlates of fat metabolism.

In a multivariate regression analysis, skeletal muscle p-ACC/total ACC, plasma FFA, and plasma TG could account for 68% of the variation in the rate of dietary fat oxidation at the end of the study. Controlling for the variation in insulin in a partial correlation analysis eliminated the relationship between fat oxidation and p-ACC/total ACC and reduced the strength of the relationship between fat oxidation and plasma FFA (r = 0.47, P < 0.05). Finally, as would be expected, the oxidation of dietary fat was inversely related to the trafficking of dietary fat to visceral (r = −0.60, P < 0.001) and subcutaneous (r = −0.51, P < 0.001) adipose tissue. The TG/FFA ratio was the strongest correlate of the trafficking of dietary fat to visceral adipose (r = 0.70, P < 0.001) and subcutaneous adipose (r = 0.57, P < 0.005). Whereas insulin was unrelated to the trafficking of dietary fat to the liver, it was highly related to the net retention of 3H into hepatic lipid (r = −0.77, P < 0.001). These data suggest skeletal muscle ACC regulation and the availability of plasma FFA may independently contribute to the changes in the oxidation rate of dietary fat but that insulin likely plays a significant role in regulating both parameters in this experimental paradigm.


The novel observations from this study are that sustained weight reduction is accompanied by adaptations in peripheral tissues that work in concert with changes in the energy balance control centers within the central nervous system (CNS) to facilitate rapid efficient weight regain during the early stages of relapse. In addition to the well-known energy gap between intake and expenditure, we observed an enhanced ability to regulate the oxidation of dietary fat, enhanced trafficking, and retention of ingested fuel to visceral adipose tissue, and the formation of small, presumably new, visceral adipocytes. In response to a caloric excess, rats that have been weight reduced suppress fat oxidation, preferentially burn CHO, and traffic dietary fat to adipose tissue lipid pools. We have shown that this energetically efficient shift in metabolism during relapse is not simply due to excess calories, as the same caloric excess resulted in a difference in the fuel metabolism of Obese and Gap-Matched rats. We also have observed that these adaptations in the regulation of peripheral fuel metabolism are not a transient phenomenon that dissipates after the first meal. Rather, these adaptations persist at least through the first 24 h of relapse to facilitate weight regain and may impart a permanent expansion in the storage capacity of visceral adipose tissue. Thus, coordinated changes in both the CNS (28, 39, 40) and in peripheral tissues work together to promote rapid, efficient regain.

These observations are consistent with previous reports that weight reduction is accompanied by an increased energy gap between appetite and expenditure that leads to an exceptionally large caloric excess in Relapse-Day 1 rats (9, 12, 19, 26, 3133, 47). As has been shown in these studies, obesity-prone rats consume ∼60% of their calories in the dark cycle and ∼40% of their calories in the light cycle. The minor favoritism for dark-cycle food intake is greatly exaggerated after weight reduction, owing to both the reported propensity of weight-reduced rats to overeat during acute feeding bouts (41) and the fact that the diet provision was provided at the beginning of the dark cycle. This drive to eat excessively is thought to be due, at least in part, to the hypoleptinemia and reduced insulin levels that accompany weight reduction (29, 31, 48). Insulin and leptin are well-known adiposity signals, but after weight reduction, they underestimate peripheral fat stores. The low levels of these adiposity signals are likely to contribute to the restructuring of neural control centers in the brain, and hence, reduce their sensitivity to satiety signals (28, 39, 40). The consequence is an exceptionally high rate and amount of energy intake and a suppression of energy expenditure, which results in a large caloric excess if the animals have free access to food. Although we know that the energy gap persists even after the lost weight is regained (31), insulin and leptin return to levels found in the Obese rats within 24 h. As such, other signals are likely involved in sustaining the energy gap throughout the relapse process. Because the normalization of satiety as the animals relapse is likely gradual and complex, a better understanding of this resolution will require further studies of the many putative peripheral and central sensing systems in this experimental paradigm.

The diurnal fluctuation in exogenous fuel availability that occurs in the weight-reduced state is accompanied by a coordinated shift in whole body and dietary fat oxidation. Our data extend previous reports that postprandial fat oxidation is suppressed after weight reduction (3, 4, 45, 51) by showing that this shift in fuel utilization is sustained with overfeeding and can result in marked changes in the oxidation and trafficking of dietary fat over 24 h. Some reports have shown that, even while in energy balance, 24-h RER is elevated in postobese individuals (2, 25). Our experimental design allowed us to compare how the same caloric excess over a 24-h period is metabolized in the obese and weight-reduced states. Obese and Gap-Matched rats were similar in fat-free mass and absorbed similar amounts of dietary fat, but the former used more of this dietary fat as a fuel for energy needs. In addition, the large caloric excess in Relapse-Day 1 rats led to a suppression in fat oxidation that persisted over the entire 24-h period, indicating that the magnitude and persistence of this suppression in the oxidation of dietary fat, and likely whole body fat oxidation, is related to the magnitude of the caloric excess.

Our present findings indicate that the enhanced regulation of fat oxidation may be mediated, at least in part, by enhanced control of skeletal muscle ACC. Insulin and glucose have been shown to coordinately suppress fat oxidation in skeletal muscle via ACC-mediated production of malonyl-CoA (49), which potently inhibits carnitine palmitoyl transferase I and prevents the entrance of acyl-CoA moieties into the mitochondria. In skeletal muscle, insulin and glucose are thought to activate ACC not only by inhibiting AMPK (56), which would attenuate its inhibitory phosphorylation of ACC at Thr172, but also by increasing cytosolic citrate concentrations (49). While insulin and glucose are elevated in both Obese and Relapse-Day 1 rats at the end of the study, p-ACC is much lower in the latter group. These data are consistent with the impaired regulation of the muscle AMPK-ACC system observed in diet-induced obese mice (35) and suggests that weight reduction attenuates the impairment. Relapse may be a reflection of refeeding after a short period of fasting, which leads to the activation of ACC and an increase in malonyl-CoA levels in both muscle and liver (57). This similarity to refeeding after a fast may, however, be limited to mature adults. In young rats that are still depositing lean body mass, refeeding after calorie restriction that is involved in “catch-up” growth is accompanied by impaired regulation of AMPK and ACC by insulin (52). This striking difference between the “catch-up” growth phenomenon in young, growing animals and weight regain in the relapse to obesity in mature animals emphasizes the impact that age and maturity can have on the peripheral adaptations to caloric restriction.

Besides the regulation of ACC, our data indicate that substrate availability plays an important role in determining the oxidation rate of relapsing animals. At 24 h, Obese and Relapse-Day 1 rats had similar levels of plasma TGs, but very different levels of plasma FFA. Given the negative relationship between fat oxidation and TG, particularly in the three groups of weight-reduced rats, one might speculate that these TGs did not represent a significant source of substrate for β-oxidation in the weight-reduced state. Refeeding after calorie restriction or fasting has been shown to increase adipose LPL activity and reduce skeletal muscle LPL activity (13, 14, 43, 44). Tissue-specific expression of LPL could serve to traffic ingested lipid, released into circulation primarily as chylomicron TGs, away from skeletal muscle and toward adipose tissue storage in relapsing animals. In addition, the striking difference in plasma FFA between Obese and Relapse-Day 1 rats, given similar levels of hyperinsulinemia, may be a consequence of enhanced regulation of lipolysis and re-esterification by insulin that has been observed after weight reduction (20), which likely involves hormone-sensitive lipase. As such, a likely scenario exists, whereby tissue-specific expression of LPL and enhanced control of lipid turnover in adipose tissue contributes to the preferential trafficking of dietary fat toward storage. More studies are needed to verify not only whether these mechanisms are involved, but also to investigate whether a more pronounced effect on LPL and turnover control in visceral adipose tissue explains the preferential trafficking to this depot. Given that skeletal muscle p-ACC, plasma FFA, and plasma TG accounted for close to 70% of the variation in fat oxidation at 24 h, our data would suggest that concomitant adaptations in adipose tissue, skeletal muscle, and liver work together to direct dietary fat away from oxidation and toward storage.

Preferential use of CHO for energy needs while storing fat is an energetically efficient way to gain weight (16, 50). Some estimates indicate that up to 25% of the available CHO load must be used to pay for the expensive process of converting glucose into fat via de novo lipogenesis, while the cost to store an equivalent excess of fat is less than 2% (50). The implications of this fuel preference are that the magnitude of the caloric restriction must be greater to maintain the weight-reduced state than it would be if fat, rather than CHO, could be oxidized for energy needs. In addition, the amount of weight regained for a given caloric excess is more than it would be during relapse if fat, rather than CHO, is the fuel for oxidation. Consistent with this assertion, Gap-Matched rats tended to gain over twice as much weight during the 24-h tracer study than Obese rats (P = 0.09), an effect that is likely to become more substantial when relapsing on a diet higher in fat.

The elevated net retention of 3H in the lipid pools of adipose tissue in Relapse-Day 1 rats would suggest that the induction of de novo lipogenesis is a critical part of the early stages of relapse, even though this may be a less efficient way to store fat. However, RER in these animals did not surpass 1.0 until later in the day (during the light cycle), indicating that this is a delayed response to the caloric excess. This delayed response could be, at least in part, mediated by the insulin's stimulation of sterol response element binding protein 1c and the lipogenic enzymatic profile, which has been shown to occur with a similar time course in response to refeeding after short-term energy restriction (5, 22). The relevance of this observation to regain in humans is unclear, as rodents are thought to be inherently more lipogenic. However, some studies in humans would suggest that de novo lipogenesis is physiologically relevant under conditions of sustained CHO overfeeding (1, 36, 50). Our observations would suggest that weight reduction primes the body for a more profound induction of lipogenesis, particularly in visceral adipose tissue. Obesity in humans has been shown to blunt the lipogenic response to forced CHO overfeeding (36), and it may be that the improvement in insulin's action after weight reduction alters this response to more reflect the profound induction observed in the lean. Regardless, the timing of this response would suggest that it may serve to dissipate the excess CHO that remains after glycogen stores are filled and energy needs are met. Alternatively, there is surprising data from overfed humans to suggest that the excess CHO is preferentially cleared to adipose lipid storage at the expense of glycogen repletion (36). Our observation of suppressed glycogen levels in Relapse-Day 1 rats would tend to support this assertion. The conversion to fat in lieu of glycogen repletion may hasten the initiation of subsequent meals.

The last significant observation from this study was an alteration in visceral adipose tissue cellularity. The appearance of small (<20 μm) adipocytes in the RP pad within 24 h of relapse suggests a regain-induced proliferation of preadipocytes and/or differentiation of new adipocytes early in the regain process in this depot. In contrast to RP fat, hyperplasia may have also occurred in MES pads of all weight-reduced rats, regardless of their relapse state. Although we did not have the power to statistically detect an elevation in cell number (∼34% higher), this effect would be consistent with other reports on EPI pads of calorie-restricted and relapsed rats (31, 60) and in subcutaneous fat of weight-reduced humans (29). These observations indicate that depots may vary in their sensitivity to the signals promoting hyperplasia after weight reduction. Visceral pads are known to vary among themselves and compared with subcutaneous fat in their inherent proliferative capacity (inguinal > retroperitoneal > epididymal > mesenteric) (11), so it would not be surprising if these depots display a difference in the way they respond to the flexible, dynamic state of weight reduction and regain. Regardless of the apparent discrepancies, in a previous report we observed hyperplasia in both EPI and RP depots that persisted after all of the lost weight had been regained (MES pads were not assessed) (31). Because the distribution frequency was similar for obese and relapsed animals, it appears that hyperplasia occurring early in relapse persists throughout the regain process and that the small, presumably new, adipocytes preferentially accumulate fat relative to their large adipocyte counterparts. Smaller adipocytes in the weight-reduced state have a reduced rate of basal and catecholamine-stimulated lipolysis (30) and are more sensitive to the antilipolytic effect of insulin (37), so the accumulation of dietary fat in and preferential hypertrophy of smaller fat cells would not be unexpected.

Our observations support the hypothesis that the enhancement in metabolic flexibility that occurs with weight reduction facilitates weight regain. Weight reduction in mature, obese subjects in this experimental paradigm and in others improves insulin sensitivity (31, 46). The predicted impact on the way muscle, liver, and adipose tissue metabolize ingested fuels during relapse is consistent with our observations of the oxidation and storage of ingested fuels in the present study. Weight reduction leads to a dynamic diurnal shift in metabolism that integrates the meal pattern, fuel preference, storage, and remobilization of ingested fuels. This is in stark contrast to the static state of metabolism in Obese rats, which were characterized by a slower, steady availability of exogenous nutrients, a constant fuel preference, and, presumably, a sustained trickle of exogenous fuels into storage depots. This distinct difference between Obese and Reduced rats may reveal a broader, more practical expression of metabolic inflexibility (23, 24) and its alteration with long-term weight reduction (53). In this context, Reduced rats have become metabolically flexible, suppressing fat oxidation when exogenous nutrients elicit a postprandial insulin response and promoting the uptake and storage of these ingested nutrients in adipose tissue. The hyperinsulinemia and available glucose fails to suppress fat oxidation and promote storage in visceral adipose tissue in the metabolically inflexible Obese rats, while the same insulin and glucose levels during relapse have a profound effect on fat oxidation and visceral adipose tissue storage.

Perspectives and Significance

The model employed in this study was designed to reflect adult humans that attempt to maintain a weight-reduced state by restricting their intake of a low-fat diet without implementing a regular program of exercise. The advantage of this experimental paradigm is that we have produced a successful weight maintenance period that would approximate 3 to 6 years in humans. Keeping weight off for this period of time in the free-living human and even in well-controlled clinical studies has proven to be close to impossible without the addition of regular exercise. The metabolic adaptations to weight reduction observed in this and other reports that appear to facilitate weight regain may be the reason for the difficulty. In future studies, we plan to characterize how regular exercise or proactively adjusting the feeding pattern to 3 or 4 times per day impacts these metabolic adaptations. It is important to note, however, that given the difference in life span, a day in the life of a rat may equate to an entire month in a human. As such, the shift in metabolism observed in the present study may reflect a more subtle shift that occurs intermittently or spread out over a period of weeks or months in a relapsing human. In any case, our observations serve to reiterate the complexity and integrative nature of these metabolic adaptations. It is obvious that there is no single factor or tissue that alone can be blamed for the biological drive to regain weight. Rather, weight loss is accompanied by changes in a number of regulatory factors in several tissues, involving both the CNS and the periphery. Given the complexity of this metabolic state, it is not surprising that the majority of therapeutic approaches for preventing weight regain, particularly those with specific targets in specific organs, have limited efficacy. Overcoming this complex multiorgan, adaptive response to weight reduction may require equally complex therapeutic strategies that target several key regulatory aspects of this metabolic state in a comprehensive approach to promote long-term weight reduction.


This work was supported by grants from the National Institutes of Health (DK-38088 and DK-67403 to P. S. MacLean). We also acknowledge the generous support from Dr. Jed Friedman and the Energy Balance and Metabolic core laboratories within the Colorado Clinical Nutrition Research Unit, which also receives funding from the National Institutes of Health (DK-48520).


We appreciate the valuable consultation with Dr. Michael Pagliassotti from Colorado State University in the early planning stages of this study.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract