Regulatory, Integrative and Comparative Physiology

Characterization of weight loss and weight regain mechanisms after Roux-en-Y gastric bypass in rats

Ana Guijarro, Susumu Suzuki, Chung Chen, Henriette Kirchner, Frank A. Middleton, Sergiy Nadtochiy, Paul S. Brookes, Akira Niijima, Akio Inui, Michael M. Meguid


Roux-en-Y gastric bypass (RYGB) is the most effective therapy for morbid obesity, but it has a ∼20% failure rate. To test our hypothesis that outcome depends on differential modifications of several energy-related systems, we used our established RYGB model in Sprague-Dawley diet-induced obese (DIO) rats to determine mechanisms contributing to successful (RGYB-S) or failed (RYGB-F) RYGB. DIO rats were randomized to RYGB, sham-operated Obese, and sham-operated obese pair-fed linked to RYGB (PF) groups. Body weight (BW), caloric intake (CI), and fecal output (FO) were recorded daily for 90 days, food efficiency (FE) was calculated, and morphological changes were determined. d-Xylose and fat absorption were studied. Glucose-stimulated vagal efferent nerve firing rates of stomach were recorded. Gut, adipose, and thyroid hormones were measured in plasma. Mitochondrial respiratory complexes in skeletal muscle and expression of energy-related hypothalamic and fat peptides, receptors, and enzymes were quantified. A 25% failure rate occurred. RYGB-S, RYGB-F, and PF rats showed rapid BW decrease vs. Obese rats, followed by sustained BW loss in RYGB-S rats. RYGB-F and PF rats gradually increased BW. BW loss in RYGB-S rats is achieved not only by RYGB-induced decreased CI and increased FO, but also via sympathetic nervous system activation, driven by increased peptide YY, CRF, and orexin signaling, decreasing FE and energy storage, demonstrated by reduced fat mass associated with the upregulation of mitochondrial uncoupling protein-2 in fat. These events override the compensatory response to the drop in leptin levels aimed at conserving energy.

  • food efficiency
  • gut and adipose hormones
  • mitochondria
  • obesity

bariatric operations are currently the only treatment for morbid obesity (defined as a body mass index of 39 or greater or weight at least 50% above normal). Of the 180,000–200,000 bariatric operations performed in 2006, ∼80% were Roux-en-Y gastric bypass (RYGB). RYGB also reverses and ameliorates the major cardiovascular and metabolic risk factors, including Type 2 diabetes mellitus and hyper- and dyslipidemia (12, 72), reduces the long-term mortality and morbidity associated with obesity, and decreases health care costs (15).

Despite the success of the RYGB operation, 20% of patients “fail to maintain long-term weight loss.” Non-surgically related failure usually occurs within the first three postoperative years (12), suggesting a metabolic-endocrine compensatory etiologic mechanism.

Weight loss induced by RYGB occurs biphasically. Initially there is rapid weight loss lasting 1–3 yr, followed by prolonged weight stabilization (39). Both occur despite a gradual increase in caloric intake during the same period (72). Our RYGB model in diet-induced obese Sprague-Dawley rats (52) also shows a postoperative biphasic weight change pattern. In this model, ∼10–13 rat days is equivalent to 1 human year (62), thus allowing us to follow rats postoperatively for a considerably long time compared with human subjects.

Conventional surgical belief is that weight loss following RYGB occurs because of 1) the small gastric pouch that limits caloric intake and 2) the Roux-en-Y loop of hindgut that short-circuits the remaining gastrointestinal tract, decreasing nutrient absorption. However, a large body of evidence suggests that after RYGB hormonal components play a major role in reducing food intake and decreasing weight (6, 17).

Other mechanisms are initiated in the creation of the gastric pouch because it divides vagal parasympathetic and sympathetic fibers (60), therefore influencing the afferent signals to the brain, while efferent signals can also be disrupted, which may play a role in the outcome of this operation. However, these mechanisms do not sufficiently account for the maintenance of body weight loss or regain after RYGB, suggesting that nutrient-hormone changes could define a successful or failed operative outcome.

Mitochondria play a fundamental role in thermogenesis and energy regulation. Major modifications in oxidative phosphorylation and mitochondrial metabolism occur in obesity (53). Proton leak across the mitochondrial inner membrane, which is catalyzed by uncoupling proteins (26) and bypasses ATP production, may account for 25% of the standard metabolic rate in the rat (66). Hepatic mitochondrial proton leak is positively correlated with standard metabolic rate (10) and is affected by thyroid status, being fourfold greater in hyperthyroid rats (33). Skeletal muscle mitochondria are also relevant in the context of energy homeostasis. Thus a decrease in skeletal muscle mitochondrial mass and function is associated with accelerated rate of fat recovery (catch-up fat) and insulin resistance, which are characteristic features of weight recovery after caloric restriction (16).

We postulate that the outcome after RYGB depends on differential modifications of several energy-related systems. To test our hypothesis we studied changes in body weight, food intake, nutrient absorption, fecal output, food efficiency, body composition, and metabolically relevant nutrients and hormones in diet-induced obese Sprague-Dawley male rats and examined the changes in electrical activity of the efferent vagal fibers innervating the gastric pouch. In addition, we measured numerous transcripts in the hypothalamus and subcutaneous fat involved in the modulation of food intake, metabolism, and energy homeostasis as well as mitochondrial function in skeletal muscle.


Study Design

The Committee for the Humane Use of Animals at the State University of New York Upstate Medical University approved the experiments, and animal care was in accordance with guidelines established by the National Institutes of Health.

The preoperative protocol to induce obesity in the Sprague-Dawley male pups, our operative procedures in which the subdiaphragmatic vagal trunks are preserved, and the postoperative management of the rats were described previously (52). To induce obesity, pups were fed a high-energy diet (D12266, Research Diets, New Brunswick, NJ) for 16 wk (74). After obesity was induced, obese rats were stratified according to body weight to ensure similar average starting body weight before the following surgical procedures: 1) sham operation on obese controls (Obese rats), which continued to eat ad libitum; 2) Roux-en-Y gastric bypass (RYGB rats); and 3) sham operation on obese rats that were pair fed (PF rats) by being given the same amount of rat chow to eat as consumed the previous day by paired RYGB rats. After surgery, rats were fed with rat chow (Rat Chow Diet no. 5008; Ralston Purina, St. Louis, MO).

After the procedure, rats were followed for 90 postoperative days and then euthanized, in a nonfasting state, starting at 9 AM. Because eight rats were euthanized in 1 day and the harvesting of blood and tissue took ∼60 min, this would result in fasting rats for different periods that would introduce the variable of “fast duration,” which has metabolic consequences of relevance to our investigation. Therefore to avoid this variable, the rats were not fasted. The groups consisted of Obese (n = 10), RYGB [n = 12; consisting of RYGB rats that successfully lost weight (RYGB-S, n = 8) and RYGB rats that failed to sustain weight loss by regaining weight (RYGB-F, n = 4)], and PF (n = 8) rats (see Statistical Analysis).

Because of the observed weight loss pattern derived from the above study, only Obese (n = 10) and RYGB (n = 12) rats had absorption studies done on postoperative days 30, 60, and 90.

Body Weight, Caloric Intake, Fecal Output, Food Efficiency, and Body Composition

Postoperatively, daily body weight, food intake, and fecal output were measured gravimetrically for 90 days. Food intake was computed as caloric intake, because between postoperative days 1 and 8 the rats had a liquid diet. They also had liquid stool, and thus fecal weight was measured only after solid food was given. Fecal output is expressed as fecal weight per 1 g of food intake × 100. Food efficiency, an index of nutrient assimilation into body mass, was calculated as a ratio of change in body weight to cumulative caloric intake and is expressed as a percentage.

At 90 days the rats were anesthetized with isoflurane and decapitated, and muscle (representing combined weight of left gastrocnemius plus soleus), fat depots (abdominal subcutaneous, mesenteric, epididymal, and retroperitoneal), and visceral organs (liver, heart, and pancreas) were dissected out and weighed, reflecting morphological changes and body composition. Fat mass was the sum of the weights of the fat depots, while fat-free mass was calculated by subtracting fat mass from body weight. The above parameters were expressed as a percentage of body weight.

d-Xylose and Fat Absorption

Before testing, rats were food deprived for 16 h. d-Xylose solution [0.5 g/kg; 20% (wt/vol)] was gavaged via a transoral tube into the gastric pouch of RYGB rats or the stomach of Obese rats without anesthesia. Urine was collected for 5 h to measure the amount of excreted d-xylose (23). d-Xylose absorption (%) was calculated as urine d-xylose (g)/administered d-Xylose (g) × 100. To measure fat absorption, stool was collected for 72 h before postoperative days 30, 60 and 90 and its fat content extracted and measured (76): fat absorption (%) per day = fat intake (g) − stool fat (g) (44). A group of PF rats was not included at the time of the study, because their gastrointestinal tract was not surgically altered and was thus anatomically similar to that of the Obese rats. We made the assumption that the absorption would thus be the same, based on supporting data from our lab in a different experiment, showing no differences in the disaccharidase activity in proximal and distal jejunum and ileum in PF rats compared with Obese rats.

Metabolically Relevant Nutrients and Hormones

After decapitation, mixed arterial and venous blood in nonfasted rats was collected into EDTA-rinsed tubes. It was centrifuged at 3,000 rpm for 10 min at 4°C for plasma. Samples were stored for a limited time at −80°C before analysis in duplicate. Plasma glucose, total cholesterol, triglycerides, and free fatty acids (FFA) were measured by enzymatic colorimetric kits (WAKO, Richmond, VA). Plasma polypeptide YY (PYY) (DSL, Webster, TX), cholecystokinin (CCK), ghrelin (Phoenix, Belmont, CA), insulin (ALPCO, Windham, NH), leptin (R&D Systems, Minneapolis, MN), and free triiodothyronine (T3) (Alpha Diagnostic, TX) were measured with enzyme immunoassay kits.

Efferent Vagal Nerve Firing Rates of Gastric Pouch

In a smaller separate set of rats [Obese (n = 3) and RYGB (n = 4)] repeated measurements of vagal efferent nerve firing rates of the gastric pouch were calculated before and after a glucose challenge at 90 days. The methodological details of similar studies were previously described in detail (54). Rats were food deprived for 6 h and anesthetized by intraperitoneal injection of urethane (1 g/kg body wt). The abdomen was opened, and the vagal gastric efferent nerve was isolated and divided with a dissection microscope. The central cut end was placed on a silver wire recording electrode to record efferent discharges. Glucose (5% in 5 ml water) was gavaged into the gastric pouch of RYGB rats or the stomach of Obese rats. The mean number of vagal efferent firing impulses per 5 s was determined in 10 successive 50-s periods. This measurement was repeated every 30 min for 180–210 min in response to the 5% glucose solution. These results were compared between the groups after conversion of raw data to standard pulses by a window discriminator, which separated study-generated discharges from background noise.

Blue-Native Gel Electrophoresis of Skeletal Muscle Mitochondrial Protein Complexes

To determine the relative amounts of mitochondrial respiratory complexes contained within skeletal muscle (gastrocnemius and soleus) samples at 90 days, frozen tissue (75 mg) was homogenized (glass Dounce homogenizer) in 2 ml of extraction buffer (mM: 440 sucrose, 20 MOPS, 1 di-Na+ EDTA, 0.2 PMSF, pH 7.2 at +4°C) and centrifuged at 500 g for 2 min. The pellet was discarded, and 1 ml of supernatant was centrifuged at 14,000 g for 10 min. The pellet from this spin (nominal mitochondrial fraction) was resuspended in 200 μl of buffer [0.75 mM aminocaproic acid, 50 mM Bis-Tris, 1% (wt/vol) n-dodecyl-β-d-maltoside, pH 7 at 4°C] and kept on ice for 10 min with occasional mixing. The mixture was then centrifuged at 14,000 g for 10 min, and 9 μl of Coomassie blue suspension (0.5 M aminocaproic acid, 5% Coomassie blue) was added to the supernatant (100 μl) before loading directly onto gel. Blue-native gel electrophoresis was performed as previously described with minor modifications (11). Gels were run at a constant 40 V for 1 h with Hi-Blue cathode buffer (50 mM tricine, 15 mM Bis-Tris, 0.02% Coomassie blue, pH 7 at +4°C) and then at 20 V overnight with Lo-Blue cathode buffer (50 mM tricine, 15 mM Bis-Tris, 0.002% Coomassie blue, pH 7 at +4°C). Gels were stained in a solution of Coomassie blue R-250 and G-250 [0.05% (wt/vol) each one in 10% acetic acid and 25% isopropanol, pH 7 at +4°C] for 3 h and then destained in 10% acetic acid and 25% isopropanol. The relative amount of mitochondrial respiratory complexes contained within skeletal muscle samples was assessed by optical densitometry with Scion Image software (Scion, Frederick, MD).

Short Tandem Repeats

We genotyped eight short tandem repeats (STRs) that are highly polymorphic between Sprague-Dawley and other closely related strains (Wistar, Wistar-Kyoto, spontaneously hypertensive). The STRs were amplified, using hepatic DNA from one RYGB-S and three RYGB-F rats. The hepatic DNA was purified with the MasterPure kit (Epicentre Biotechnologies, Madison, WI), and PCR was performed to amplify the STR sequences with primers based on the Rat Genome Database records for specific STR sequences (and provided by Drs. R. Hoopes and S. Scheinman at SUNY Upstate Medical University). The STRs were chosen to interrogate three different chromosomes (13) at approximately every 60 megabases. The eight STRs included were D1RAT196, D1RAT193, D1MIT32, D2RAT6, D2RAT88, D2RAT171, D3MGH16, and D3MIT13. The PCR products of these reactions were resolved on a 4% agarose/ethidium bromide-stained gel by electrophoresis (80 V, 1 h) and compared with the simple sequence length polymorphism expected size database for different rat strains available on the Rat Genome Database.

Microarray of Whole Hypothalamus and Peripheral Fat

To examine the possibility that changes in expression of specific genes and gene groups might underlie the failure of certain rats to sustain weight loss after RYGB, we performed gene expression analysis with oligonucleotide microarrays (RAE230 2.0, Affymetrix) on three groups of male rats: RYGB-S (n = 8), RYGB-F (n = 3), and PF controls (n = 3). For each animal, total RNA was isolated from homogenates of an entire hypothalamus from one hemisphere and a sampling of subcutaneous abdominal fat with the RNeasy kit (Qiagen). We had a large number of successful animals (RYGB-S rats) with tissue available. Thus we chose to use a partial pooling design for establishing the baseline expression values for the RYGB-S rats that showed the expected outcomes. For RYGB-F and PF groups, the RNA from three individual animals was used for microarray analysis. For the RYGB-S group, the eight total RNA samples for each tissue type were divided into three separate pools (each of which contained RNA from 2 or 3 rats) for microarray analysis using an equal quantity of RNA from each animal. In total, 18 GeneChip hybridizations were performed comparing the three treatment groups (RYGB-S, RYGB-F, PF) and the two tissue types (hypothalamus and fat), using three independent replicates for each. Amplification and labeling of RNA were performed with the WT-Ovation RNA Amplification System (NuGen), and hybridization, washing, staining, and scanning of the GeneChips were performed according to a standard protocol (GeneChip Expression Analysis Technical Manual 701021 Rev. 5, Affymetrix). After scanning, the microarray images were processed with GeneChip Operating System (GCOS) software to obtain performance metrics, and prepared for data analysis using the robust multiarray analysis (RMA) normalization method (36). After RMA-based estimates of gene expression levels were obtained, individual transcripts in each of the three experimental groups and two tissue types were compared with a two-way analysis of variance (2 × 3 ANOVA) comparing tissue type (hypothalamus or fat) and treatment group (RYGB-S, RYGB-F, PF). Because our primary aim was to determine possible causes of the failure to sustain weight loss in the RYGB-F animals, we focused the present analysis on a subset of 667 probes that were determined to be involved in feeding, digestion, or general carbohydrate, lipid, and protein metabolism, using the online NetAffx tool (Affymetrix). A list of these probes is posted at A table of the genes showing a significant main effect of treatment in either tissue type (uncorrected P < 0.05) was generated, and post hoc analyses [Fisher's protected least significant difference (LSD)] were performed to determine the specific comparisons showing significant effects.

Statistical Analysis

After the operation, time series plots of daily body weight, food intake, and fecal output were plotted for each individual rat to show changing patterns. To determine the change or inflection point (change point day) of each daily characteristic during the rapid and gradual weight loss periods after operation for 90 days, a switching quadratic trend analysis was performed, to identify the change point day of mean body weight, caloric intake, and fecal output in Obese, RYGB, and PF rats. The exact time point was estimated based on the smallest average error(s) and the highest adjusted R2 for each possible time period between t = 0 and t = 90 days. When the switching quadratic trend analysis was applied to the body weight series of each individual rat and the mean body weight series of each of the study groups, it revealed two separated RYGB cohorts: RYGB-weight loss (RYGB-S) and RYGB-weight regain (RYGB-F). Thus further comparisons and analyses were conducted separately with respect to the four experimental groups: Obese, RYGB-S, RYGB-F, and PF. In the RYGB-S, RYGB-F, and PF rats a switching quadratic trend model was used to optimally analyze the data, while a switching linear trend model fitted the Obese rat data. A linear trend model for change point analysis as used for mean daily caloric intake and fecal output fitted the postoperative Obese, RYGB-S, RYGB-F, and PF rat data more optimally. Comparisons of daily body weight, caloric intake, fecal output, feeding efficiency, and organ weight in the four study groups were analyzed by one-way ANOVA and Tukey's pairwise multiple-comparison test with the join error rate equal to 0.05. To assess the differences in d-xylose and fat absorption between Obese and RYGB rats and with time, one-way ANOVA and Tukey's pairwise multiple-comparison test with the join error rate equal to 0.05 were performed; data are reported as means ± SE. For vagal nerve firing rates of the gastric pouch, differences in efferent firing rates were evaluated by one-way ANOVA and t-test; data are reported as means ± SE. Microarray data were analyzed by two-way ANOVA, and post hoc analyses (Fisher's protected LSD) were performed to determine the specific comparisons showing significant effects. Densitometric data from blue-native gels were analyzed by two-way ANOVA, with P < 0.05 being considered significant.


Body Weight and Caloric Intake

These interrelated indexes are plotted together in Fig. 1. The preoperative body weight of all RYGB rats (n = 12) was 765.76 ± 8.10 g. The switching quadratic trend analysis of daily body weight showed that after RYGB operation 4 of the 12 rats, after an initial period of body weight loss, began to regain weight, i.e., the RYGB operation failed (RYGB-F, n = 4), so that by postoperative day 90 they weighed more (P < 0.001) than the remaining RYGB rats that successfully continued to lose weight (RYGB-S, n = 8), as shown in Fig. 1A.

Fig. 1.

Body weight pattern (A) and caloric intake (B) in sham-operated obese control (Obese), successful and failed Roux-en-Y gastric bypass (RYGB-S and RYGB-F), and pair-fed (PF) rats from operation day (day 0) to 90 days after surgery. Data are represented as means ± SE.

The change point day as well as the quantitative changes in mean body weight and mean daily change in body weight before and after change point day are shown in Table 1. In the immediate postoperative period the Obese rats lost 7.0 ± 0.7% of their initial body weight until change point day on day 9 (Table 1). With the switch to solid chow, the Obese rats gained weight until the end of the study, when they weighed 4.3 ± 3.5% more than their initial body weight. After the operation RYGB-S rats also lost weight rapidly (15.4 ± 0.6% of their initial body weight) but, despite the switch to chow, continued to lose weight until change point day at day 22 (Table 1), after which body weight loss stabilized, decreasing by 33.5 ± 2.6% compared with their starting body weight on day 90. Initially, the RYGB-F and the PF rats rapidly lost body weight until change point day at days 24 and 23, respectively (at which point RYGB-F rats had lost 10.8 ± 1.4%, while PF rats had lost 7.7 ± 1.5%). This was followed by a 10.9 ± 2.3% regain in body weight for RYGB-F rats and an 8.8 ± 5.0% regain in PF rats. By day 90, RYGB-S rats had lost a greater percentage of body weight (P < 0.001) than RYGB-F rats (14.2 ± 1.1%) and PF rats (13.4 ± 3.5%). The body weight pattern of RYGB-F and PF rats was similar, with no difference in the percentage of body weight loss (P = 0.997), which, however, was lower than their initial body weight by day 90.

View this table:
Table 1.

Mean BW and mean change in daily BW during 90 postoperative days in Obese, RYGB-S, RYGB-F, and PF rats

The preoperative caloric intake was 101.4 ± 6.1 kcal/day. After the operation, when rats were started on a liquid diet, the Obese rats ate 46.1 ± 4.7 kcal/day, while RYGB-S rats ate 16.9 ± 2.0 kcal/day and RYGB-F rats ate 15.9 ± 1.9 kcal/day (Fig. 1B). On day 9 chow was provided and caloric intake increased in all groups, as shown in Fig. 1B. This occurred during the same time that body weight was decreasing until their respective change point days. As shown in Table 2, the change point day was day 25 in Obese rats, day 34 in RYGB-S rats, day 30 in RYGB-F rats, and day 34 in PF rats. After each change point day, caloric intake remained relatively stable until the end of the study.

View this table:
Table 2.

Mean CI and mean change in daily CI during 90 postoperative days in Obese, RYGB-S, RYGB-F, and PF rats

The quantitative changes in mean caloric intake and in mean daily change in caloric intake before and after change point day are shown in Table 2. Although caloric intake increased between day 9 and change point day in all groups, it remained significantly lower in RYGB-S and PF rats than in Obese rats (P = 0.005 and P = 0.006, respectively). Caloric intake in RYGB-F rats was lower vs. Obese rats, but it was not statistically different (P = 0.105; Table 2). Between change point day and day 90 mean caloric intake in RYGB-S rats was lower than in the remaining groups (P < 0.001), while in RYGB-F rats it was higher than that in both RYGB-S and PF rats (P < 0.001), although it was not different from that in Obese rats (P = 0.920). Caloric intake in PF rats was lower than in Obese rats (P < 0.001). Throughout the study there were no differences (P > 0.05; Table 2) in the daily mean change in caloric intake among the groups.

Fecal Output

Because immediately after the operation rats were on a liquid diet, their liquid stools were not quantified. After the introduction of solid chow feces became formed and were quantitated, and, as shown in Table 3, from day 9 to change point day, there were no differences in the ratio of fecal weight per 1 g of food intake among groups.

View this table:
Table 3.

Percentage of fecal weight per 1 g of food intake during 90 postoperative days in Obese, RYGB-S, RYGB-F, and PF rats

However, from change point day to the end of the study, RYGB-S and RYGB-F rats put out significantly greater amounts of feces than both the Obese and PF cohorts, in whom the gastrointestinal tract was intact (P < 0.001). RYGB-S rats had also a greater fecal output than RYGB-F rats (P < 0.001). No differences were found between PF and Obese rats (P = 0.752).

d-Xylose and Fat Absorption

At day 30, d-xylose absorption (Fig. 2A) decreased in RYGB-S vs. Obese rats (P = 0.001), while in RYGB-F rats it was similar to Obese rats (P = 0.725). d-Xylose absorption at days 60 and 90 was not significantly different in both RYGB groups vs. Obese rats (P > 0.05), while there was no differences between RYGB-S and RYGB-F rats at any time point (P > 0.05).

Fig. 2.

d-Xylose (A) and fat (B) absorption in Obese, RYGB-S, and RYGB-F rats at postoperative days 30, 60, and 90. Data are expressed as means ± SE. ***P < 0.005 vs. Obese rats.

In contrast, fat absorption (Fig. 2B) was significantly lower in both RYGB-S and RYGB-F rats vs. Obese rats at postoperative days 30, 60, and 90 (P < 0.001), but no differences were found between RYGB-S and RYGB-F rats (P > 0.05) at any time point or within each group with time (P > 0.05).

Food Efficiency

Between days 0 and 8 the food efficiency was negative in all groups (Fig. 3A), and in RYGB-S rats it was significantly lower than in Obese (P < 0.001), RYGB-F (P = 0.046) and PF (P = 0.019) rats. Both RYGB-F (P = 0.022) and PF (P = 0.002) rats had a lower food efficiency than Obese rats, but there were no significant differences between them (P = 0.999). From day 9 to change point day (Fig. 3B) no differences were found among groups (P > 0.05), although RYGB-S rats had a lower food efficiency compared with the remaining groups, which was not statistically different (P > 0.05). From change point day to day 90 (Fig. 3C) food efficiency improved, becoming positive in Obese, RYGB-F, and PF rats while remaining negative and significantly lower in RYGB-S rats compared with Obese (P = 0.002), RYGB-F (P = 0.029) and PF (P = 0.011) rats.

Fig. 3.

Food efficiency from days 0 to 8 (A), from day 9 to change point day (B), and from change point day to postoperative day 90 (C). Data are expressed as means ± SE. *P < 0.05, ***P < 0.005 vs. Obese rats; #P < 0.05 vs. RYGB rats.

Body Composition

At day 90, in RYGB-S (P < 0.001), RYGB-F (P < 0.001), and PF (P = 0.007) rats total fat mass (expressed as % of body weight) was lower than in Obese rats (Table 4). In RYGB-S rats, total fat weight was also lower than in PF rats (P < 0.001), while there were no differences between RYGB-F and PF rats (P = 0.543). The percentage of fat mass in RYGB-S rats was lower than in RYGB-F rats, but the difference was not statistically significant (P = 0.110). The reduction in the percentage of fat mass relative to Obese rats was 71.5 ± 3.4% in RYGB-S rats, 47.1 ± 11.1% in RYGB-F rats, and 27.4 ± 5.5% in PF rats.

View this table:
Table 4.

Body composition (fat mass, fat-free mass, and organ weights expressed as percentage of body weight) and plasma nutrient and hormone concentrations at day 90 in Obese, RYGB-S, and RYGB-F, and PF rats

The percentage of fat free mass (Table 4) was higher in RYGB-S (P < 0.001), RYGB-F (P < 0.001), and PF (P = 0.007) rats compared with Obese rats. In RYGB-S rats, the percentage of fat free mass was higher than in PF rats (P < 0.001), while there were no statistical differences between RYGB-F and PF rats (P = 0.263) or between RYGB-S and RYGB-F rats (P = 0.110).

At day 90, liver weight, as a percentage of body weight (Table 4), was lower in PF rats than in Obese (P = 0.016), RYGB-S (P < 0.001), and RYGB-F (P = 0.002) rats. There was no difference in RYGB-S vs. Obese rats (P = 0.423), RYGB-F vs. Obese rats (P = 0.277), or RYGB-S vs. RYGB-F rats (P = 0.877). Pancreas weight, expressed as a percentage of body weight, did not change with RYGB or caloric restriction (P > 0.05; Table 4). As shown in Table 4, the percentage of heart weight was higher in RYGB-S rats compared with both Obese (P < 0.001) and PF (P = 0.023) rats. No differences were found among the remaining groups (P < 0.05). The percentage of the skeletal muscle weight was not different among groups (P < 0.05; Table 4).

Metabolically Relevant Nutrients and Hormones

Relative to Obese rats plasma glucose was decreased in RYGB-S and RYGB-F rats (P < 0.001) and PF rats (P = 0.010), as shown in Table 4. Plasma glucose was also decreased in RYGB-S vs. PF rats (P = 0.042) but was not different between the RYGB groups (P = 0.999) or between RGYB-F and PF rats (P = 0.204).

Total cholesterol was lower in RYGB-S vs. Obese rats (P < 0.001; Table 4) and it was lower in RYGB-F and PF rats vs. Obese rats, but not statistically different. No differences were found among the remaining groups (P > 0.05). A decrease in plasma triglycerides occurred in RYGB-S (P = 0.001), RYGB-F (P = 0.018), and PF rats (P = 0.005) compared with Obese rats. FFA concentrations were similar in all groups (P > 0.05).

No differences were detected among groups in total and acylated ghrelin and CCK, although acylated ghrelin in both RYGB groups was slightly lower than in Obese and PF rats (P > 0.05). PYY concentrations in RYGB-S rats were more than 3.7-fold higher than in Obese rats (P < 0.001) and 2.7-fold higher than in PF rats (P = 0.012). There were no differences between RYGB-F and PF rats (P = 0.977).

A lower plasma insulin concentration in RYGB-S rats (P = 0.001; Table 4) and PF rats (P = 0.002) vs. Obese rats was measured. There were no differences between PF and RYGB-F rats or between RYGB-S and RYGB-F rats (P = 0.354 and P = 0.993, respectively).

Relative to that in Obese rats, plasma leptin was lower in RYGB-S (P < 0.001), RYGB-F (P = 0.001), and PF (P = 0.020; Table 4) rats. Plasma leptin levels in RYGB-S rats were 14.8-fold lower than in Obese rats and 8.7-fold lower compared with PF rats (P = 0.004). In the RYGB-F and PF rats leptin concentration was 2.8-fold and 1.7-fold lower than in Obese rats, respectively. No differences were found between RYGB-F and PF rats (P = 0.504) or between RYGB-S and RYGB-F rats (P = 0.209).

Plasma T3 was significantly lower in RYGB-S rats than in Obese (P = 0.003), RYGB-F, and PF rats (P < 0.001), while no differences were found among Obese, RYGB-F, and PF rats (P > 0.05).

Efferent Vagal Nerve Firing Rates of Gastric Pouch

Figure 4, A and B, top, show representative tracings of upper gastrointestinal vagal efferent firing rate before and after a 5% glucose gavage in one Obese rat and one RYGB rat. These data are shown quantitatively in Fig. 4, A and B, bottom, and the mean values for Obese and RYGB rats are shown in Fig. 4C. A significant increase in vagal celiac efferent firing rate occurred 60 min after glucose gavage and peaked at 120 min, remaining elevated for the duration of the study in Obese rats (Fig. 4, A and C). In contrast, in RYGB rats a 5% glucose gavage into the gastric pouch did not change the firing rate (Fig. 4, B and C).

Fig. 4.

Vagal nerve firing rates of gastric pouch. A and B: representative tracings (top) of upper gastrointestinal vagal efferent firing rate before and after a 5% glucose gavage in 1 Obese rat (A) and 1 RYGB rat (B) and their corresponding quantitative representation (bottom). C: mean values for Obese and RYGB rats. Data are represented as means ± SE. *P < 0.05 vs. Obese rats.

Blue-Native Gel Electrophoresis of Mitochondrial Protein Complexes

Figure 5A shows a representative blue-native gel of mitochondrial complexes from skeletal muscle of RYGB-S and RYGB-F rats studied at day 90, while Fig. 5B reveals a significant increase in the levels of complexes I, II, and IV of mitochondrial oxidative phosphorylation in RYGB-F rats. Because samples were prepared from identical quantities of tissue, these data are indicative of greater mitochondrial mass per unit of tissue in the RYGB-F group.

Fig. 5.

Mitochondrial respiratory complex content in skeletal muscle. Representative picture of mitochondrial respiratory complexes (Cx) separated by blue-native gel electrophoresis (A) and quantitative analysis of mitochondrial respiratory complex content (B). Data are represented as means ± SE. #P < 0.05 vs. RYGB-S rats.

Short Tandem Repeats

The lengths of the STRs from RYGB-S and RYGB-F rats were all in the expected ranges for Sprague-Dawley-derived rat strains except for the D1RAT193 STR of one RYGB-F rat. Notably, however, evidence for heterogeneity was found (at least 2 different STRs) at five of the eight loci (D1RAT193, D1MIT32, D2RAT88, D3MGH16, and D3MIT13). These data suggested that the success or failure of RYGB was not simply related to genetic background differences that might have been present between the animals in the random-bred groups.

Microarray of Whole Hypothalamus and Peripheral Fat


We performed a total of 18 GeneChip hybridizations comparing 2 tissue types (hypothalamus and subcutaneous abdominal fat) and 3 treatment groups (RYGB-S, RYGB-F, and PF), using 3 independent replicate samples for each comparison group. Overall, the 18 microarray hybridizations performed equally well, with % present call rates, scale factors, RawQ, 3′-to-5′ ratios for housekeeping genes, and average signals of present and absent genes highly comparable within each tissue type. Thus all arrays were included in the subsequent RMA-based data analysis.

Single gene results.

We initially used a 2 × 3 ANOVA to test a specific list of 667 probes involved in feeding or metabolic function for significant effects of treatment group in either tissue type and subsequently performed post hoc analyses (Fisher's protected LSD) to determine the source of the main effect in RYGB-F and PF animals compared with RYGB-S animals. A total of 60 of 667 tested transcripts demonstrated a significant main effect of treatment for either tissue type (Table 5). Of these, 44 transcripts showed significant main effects in the hypothalamus only, 14 transcripts showed significant main effects in the fat tissue, and 2 transcripts showed significant main effects in both tissues in this paradigm. The entire set of data for the 667 genes of interest is available at the following link: All of the microarray data generated in this study have been published at the NCBI Gene Expression Omnibus database (accession no. GSE8314).

View this table:
Table 5.

Feeding/metabolic function-related genes showing differential expression in RYGB-S, RYGB-F, or PF rats

Examination of the specific transcripts showing significant changes in expression indicated several involved in aerobic or anaerobic glycolysis (e.g., aldolase, enolase, lactate dehydrogenase, glyceraldehyde-3-phosphate dehydrogenase), as well as fat metabolism, fat cell differentiation, or fatty acid transport [e.g., multiple isoforms of fatty acid desaturases and fatty acid transporters, hydroxysteroid (17-β) dehydrogenase, adipose differentiation-related protein]. In addition, numerous transcripts involved in signaling mechanisms that regulate food intake or metabolism were also observed as significantly changed [e.g., β1-adrenergic receptor, cannabinoid receptor-1, CCK, corticotrophin-releasing hormone receptor-2, orexin, orexin receptor-2 (OX2R), leptin receptor, PYY, ionotropic glutamate receptor. and thyroid hormone receptor-α]. Some examples are shown in Fig. 6. Notably, many of these genes are the same as those (or directly related to those) showing significant changes in the plasma of RYGB-F or PF animals (e.g., glucose, triglycerides, PYY, leptin, and T3; see Table 4) and thus the two data sets can be considered to provide independent validation of these effects at the transcriptional and protein levels.

Fig. 6.

Expression changes of selected feeding/metabolic function-related genes in the hypothalamus and subcutaneous abdominal fat of RYGB-S, RYGB-F, and PF rats. Data are represented as means ± SE. #P < 0.05 vs. RYGB-S rats.


The successful and progressive decline in body weight and thus the weight loss pattern after RYGB in diet-induced obese rats is remarkably similar to the two-stage weight loss pattern defined in humans after successful RYGB for the morbidly obese (39), as well as that achieved in response to behavior and/or lifestyle modification (24), treatment with antiobesity drugs (8), or a very low-calorie diet (34), and to that previously observed in our 60-day RYGB study (63). In our model, as well as in patients undergoing successful RYGB, a greater magnitude of weight loss is achieved than with nonpharmacological or currently available pharmacological treatments, including endocannabinoid receptor blockade therapy (34, 57). This rapid weight loss is followed by a weight loss plateau, which after RYGB in successful patients is maintained for life (72). Since 10–13 rat days are approximately equivalent to 1 human year (62) weight loss is maintained for the equivalent of ∼7–9 years in our model and represents a successful loss of primarily fat mass (Table 4), indicative of lipid oxidation (total cholesterol and triglycerides; Table 4).

In our study, a biphasic weight change was observed in all groups. The initial decrease in body weight is attributed to the combined effects of operative stress and suboptimal caloric intake due to the postoperative liquid diet. However, this is a common factor in all groups. The provision of solid chow on day 9 after the operation led to an increase in caloric intake in all groups despite which RYGB and PF rats continued to lose weight during this time, suggesting that both the effects of the operation and the caloric restriction by itself (PF group) triggered peripheral metabolic signals, which continued to induce weight loss even during a simultaneous increase in caloric intake. Because in this group of rats, switching quadratic trend analysis revealing the differences between those rats that successfully lost weight and those that started to regain weight was performed at day 90, we did not have a PF cohort for each RYGB group. Thus the information gained from these PF rats poses some limitations on the interpretation of their data.

In comparison to other therapies such as very low-calorie diet (34), the RYGB-induced rapid weight loss phase is even more striking, because it occurred 1) when a concurrent and continuous increase in caloric intake occurred and 2) during a similar time frame, i.e., 10–35 post-RYGB days, which is equivalent to the initial 2–3 yr after RYGB in humans (Fig. 1). In both RYGB-S rats and human RYGB outcome, weight loss and caloric intake stabilized and plateaued on approximately the same timescale, suggesting that this stabilization in energy intake may also account in part for the plateau in body weight. In our model, the lower mean caloric intake in RYGB-S rats compared with Obese, RYGB-F, and PF rats (Table 2) partly explains the greater degree of weight loss, but caloric restriction may not be the sole cause of this difference. PF rats had a significantly lower caloric intake than RYGB-F rats but had a similar degree of weight reduction during the body weight plateau. Furthermore, in RYGB-F rats caloric intake was the same as in Obese rats but their body weight was lower, suggesting that other factors contribute to explain the differences. The effect of caloric restriction on body weight observed in PF rats is similar to that described in mice, in which the initial period of weight loss is followed by a period of weight regain (25).

Another factor contributing to the greater degree of weight loss in RYGB-S vs. both RYGB-F and PF rats is their higher fecal output (Table 3) despite the fewer calories ingested by RYGB-S rats (Table 2). The greater fecal loss found in both RYGB groups vs. Obese and PF rats may be explained by the rearranged gastrointestinal tract in RYGB, which reduces the length of the gastrointestinal tract in which nutrient absorption occurs, as shown by a decrease of both d-xylose and fat absorption (Fig. 2). The greater fecal output measured in RYGB-S rats might reflect their lower food efficiency. While RYGB is thought to cause weight loss in part via malabsorption, although this is transiently true (9), once the gut has adapted malabsorption is not observed. When the temporal changes in small bowel absorption after RYGB were compared, d-xylose absorption during the rapid weight loss period was less than during the gradual weight loss period, suggesting that small bowel adaptation with time increased the efficiency of absorption. Could this contribute to the slowing of weight loss manifested by the plateau? An increase in PYY was associated with intestinal cell proliferation after intestinal resection and during adaptation to short bowel syndrome (2) leading to enhanced absorption and delayed intestinal transit time (4, 74). Together the above findings suggest that the rise in PYY as previously reported by us (74) contributes to the temporal adaptation in d-xylose absorption.

The weight loss plateau in the RYGB-S rats reflects compensatory processes that oppose weight loss to the ∼80% fat mass loss, as shown in Fig. 3 (40). A similar response also occurred in RYGB-F (∼44%) and PF (∼41%) rats, although to a lesser degree. In RYGB-F and PF rats, which started to regain weight, the compensatory processes opposing weight loss are more robust than in RYGB-S rats, leading to a lower body weight loss. In obese subjects the degree of fat mass loss induced by the RYGB operation exceeds the 10% reduction in body weight (46), which is accompanied by a reduction in energy expenditure beyond that predicted by loss of body mass (21). In these obese study subjects sympathetic nervous system (SNS) tone and concentrations of leptin and thyroid hormones are decreased, while parasympathetic nervous system (PNS) tone is increased.

The plateau in weight loss seen in behavior lifestyle modification and with antiobesity drugs is due to a reduction in resting energy expenditure, reflecting the compensatory processes that oppose weight loss. These processes, also referred to as the “starvation response,” are the result of a decline in leptin levels, which counterbalances the decreased caloric intake (50). This response is aimed at conserving energy through an increase in food efficiency (68). In our model we found that the decrease in plasma leptin was accompanied by enhanced food efficiency during the plateau in body weight loss in both RYGB-F and PF rats, but this response was less robust in RYGB-S rats, whose food efficiency remained negative despite the even greater reduction in plasma leptin. This lower food efficiency can explain the lower weight of RYGB-S rats compared with the remaining groups and is the reason why mean daily body weight change remained negative in this group of rats.

Does successful RYGB reset the leptin “set point” [concentration of plasma leptin perceived by the hypothalamus as a state of energy sufficiency (27)] by lowering it? If that is the case, in RYGB-S rats the decrease in leptin is not perceived as a state of energy insufficiency, and consequently food efficiency remains negative, while in RYGB-F and PF rats the reduced plasma leptin triggers an increase in food efficiency that promotes an increase in weight. At such lower leptin concentrations it could also activate the SNS by acting on the hypothalamus (70). How the leptin set point is reset by successful RYGB is a crucial question that remains to be clarified. The increased leptin receptor expression in the hypothalamus of RYGB-S rats (Fig. 6) may account, in part, for this hypothetical lower leptin set point, although it does not rule out other changes including a higher affinity of these receptors. However, taking into account that RYGB-S rats had a decrease in T3 levels and in thyroid hormone receptor-α in hypothalamus, we would expect a decrease in resting energy expenditure and therefore an increase in food efficiency. However, food efficiency remains negative in these rats.

In finding a decrease in T3 in RYGB-S rats, we expect a lower metabolic rate, because T3 is involved in the regulation of energy expenditure, modulating both obligatory and adaptive thermogenesis (71). T3 increases metabolic rate by upregulating uncoupling protein (UCP) expression in skeletal muscle, heart, and white adipose tissue and modulating UCP-1 in brown adipose tissue (BAT) (45), leading to an increase in heat production and a decrease in the thermodynamic efficiency of ATP synthesis (71). In caloric restriction a decrease in T3, thyroxine, and TSH plasma levels occurs, while intracerebroventricular leptin increases T3 production in normal rats (19). Therefore, the lower RYGB-S leptin levels along with their lower caloric intake may contribute to the reduction in T3 plasma levels. On the other hand, T3 induces a rise in the expression of genes coding for enzymes of lipogenesis (45, 55) and stimulates food intake by acting directly on the ventromedial nucleus (41). This orexigenic effect is also achieved by the inhibition of leptin production (51). Taking the sum of these effects into account, the lower T3 levels in RYGB-S rats would explain the reduction in fat pad mass, caloric intake, and body weight. Furthermore, the reduction in T3 could be part of the compensatory processes dampening the excessive increase in energy expenditure during the negative energy state induced by successful RYGB (32).

A more plausible interpretation of our data is that in RYGB-S rats leptin set point does not change but the compensatory processes that oppose weight loss triggered by the decreased leptin levels are offset by the increase in PYY levels in these rats (Table 4), because in high-fat diet-induced obese mice the administration of PYY3–36 reduces food intake and body weight, decreases food efficiency and adiposity, increases fatty acid oxidation, and increases fecal energy loss (1). In contrast, in both RYGB-F and PF rats PYY was not different from that in Obese rats, and consequently there is no compensation for the reduction in plasma leptin, contributing to a lower weight loss than in RYGB-S rats. Moreover, we have found a higher hypothalamic PYY expression in RYGB-S rats compared with RYGB-F rats (Fig. 6), suggesting an inhibition of the neuropeptide Y (NPY)/agouti-related protein neurons in the arcuate nucleus (ARC) and an activation of the proopiomelanocortin/cocaine-amphetamine-regulated transcript neurons leading to an increase in α-MSH. Confirmation of this suggestion is provided by our previous report of a significant decrease of NPY in ARC and in both parvocellular and magnocellular paraventricular nuclei (PVN) with a simultaneous increase in α-MSH and 5-HT receptors 10 days after RYGB (67).

The sum of these changes decreases food intake and increases SNS activity, causing an increase in energy expenditure in peripheral tissues (3). In addition, the higher hypothalamic expression of CRF receptor-2 in RYGB-S rats (Fig. 6) counterbalances this compensatory response to decreased leptin levels in RYGB-S rats, since CRF, when administered intracerebroventricularly or microinjected into preoptic area or dorsomedial hypothalamus (DMH), elicits a sustained increase in sympathetic nerve activity to interscapular BAT and its temperature, and also increases expired CO2 and heart rate (13) and may also account for the lower food intake found in RYGB-S rats. Similarly, intracerebroventricular administration of orexin-A induces an increase in body temperature (79) and enhances metabolic rate and alters respiratory quotient in mice (49). In our study, both orexin mRNA and OX2R mRNA were elevated in the hypothalamus of RYGB-S rats. Therefore, in RYGB-S rats we would expect an increase in metabolic rate and in both physical activity and body temperature, leading to greater energy expenditure and to lower body weight. However, the reduction found in ionotropic glutamate receptor expression in the hypothalamus of RYGB-S rats would argue against a role of increased CRF and orexin activity inducing an increase in energy expenditure in these rats, because the effects of both neuropeptides on thermogenesis are mediated by ionotropic glutamate receptors (13). Such a discrepancy could be due to measurement of the expression of these receptors in the whole hypothalamus, whereas the glutamate receptors that mediate CRF and orexin effects on thermogenesis are localized in the DMH. Furthermore, l-glutamate has either stimulatory or inhibitory effects on sympathetic innervation to BAT depending on the hypothalamic nucleus into which it is injected (78).

Both central and peripheral administration of anandamide increases food intake and body weight in rodents (37) by acting on cannabinoid receptor 1 (CB1R) (14). Could the RYGB procedure, therefore, also affect the endocannabinoid system? Although we found higher hypothalamic expression of CB1R in RYGB-S vs. RYGB-F rats, they had a marked elevation in hypothalamic expression of arachidonate 15-lipoxygenase, an enzyme involved in the degradation of both anandamide and 2-arachidonylglycerol (42). The CB1R antagonist SR-141716 induces long-lasting weight loss, independent in part of food intake, by deceasing fatty acid synthesis (56) and increasing metabolic activities (38, 47) and/or energy expenditure as suggested by the increase in basal oxygen consumption in ob/ob mice after chronic intraperitoneal administration (47), while CB1R activation increases the hepatic gene expression of the lipogenic enzymes in mice (56) and induces long-lasting hyperthermia (64).

By 90 days after operation, we documented significant loss of fat mass in RYGB-S rats (Table 4). This reduction was associated with a decrease in insulin levels leading to the mobilization of muscle and liver substrates such as glycogen or triglycerides. In fact, we found a reduction in hepatic (77) and intramyocellular (unpublished data) triglycerides in RYGB rats, which is in accordance with an increase in fat catabolism, documented by the decrease in plasma triglycerides and cholesterol found in the present study. The reduction in caloric intake did not affect the weight of the liver, pancreas, and skeletal muscle in RYGB rats but decreased the weight of the liver in PF rats in accordance with what was reported by other investigators, who found that a 40% caloric restriction in rats diminished the weight of the liver, kidneys, and heart (75). The liver and the heart, along with the brain and the kidney, are the most metabolically active organs (31), and although these organs make up <10% of total body weight they account for the majority of basal energy expenditure (28). Therefore, drastic reduction in the weight of the liver in PF rats could be interpreted as an attempt to reduce energy expenditure aimed at conserving energy and avoiding a further reduction in body weight. On the contrary, the RYGB operation seems to overcome this effect of caloric restriction given that the weight of the liver in RYGB rats showed a slight, although not statistical, increase compared with Obese rats and was markedly higher than in PF rats. Furthermore, in RYGB-S rats the weight of the heart was higher than in Obese and PF rats. This could contribute to higher energy expenditure in RYGB rats, particularly in RYGB-S rats, leading to a greater weight loss.

In Obese, PF, and RYGB-F rats the increase in caloric intake that followed the diminished food intake (secondary to the stress of the operation and the lower caloric intake during liquid diet intake) was accompanied by an increased weight, leading to a higher fat mass compared with RYGB-S rats. The process of accelerated rate of fat recovery (catch-up fat) is a recognized risk factor for the development of adult obesity after dieting. After semistarvation, rats refed the same amount of a low-fat diet as controls have lower energy expenditure because of reduced thermogenesis favoring accelerated fat deposition and show normal glucose tolerance but higher plasma insulin after a glucose load at a time when their body fat and plasma FFA did not exceed those of controls. Isocaloric refeeding on a high-fat diet results in even lower energy expenditure and thermogenesis and greater increased fat deposition and leads to even higher plasma insulin and elevated plasma glucose after a glucose load (22). Because UCP-2 and UCP-3 increase lipid utilization (69), these proteins may confer resistance to weight gain. Although not statistically significant in our microarray comparisons (which used a small number of samples), we found a biological increase in UCP-2 expression in both hypothalamus (1.5-fold) and abdominal subcutaneous fat (2.8-fold) in RYGB-S rats (data not shown), which would explain the absence of weight regain in these rats. The higher hypothalamic expression of UCP-2 in RYGB-S vs. RYGB-F and PF rats is relevant in the regulation of energy homeostasis. UCP-2 is coexpressed with CRF in the PVN, with vasoactive peptide and oxytocin in the supraoptic nucleus, and with NPY in the ARC (35) neurons expressing these proteins. Moreover, all these neurons and those expressing melanin-concentrating hormone and orexin are targeted by axon terminals containing UCP-2, while UCP-2-producing neuronal perikarya express receptors for leptin in the ARC (35). The mitochondria of these UCP-2-producing neurons have increased proton leak generating heat that controls the neuronal activity in the hypothalamic peptidergic systems involved in energy homeostasis.

An increase in SNS activity, via activation of α2-adrenergic receptors on β-cells, may account for the reduction in insulin levels in RYGB-S rats (48). Also, the stimulation of SNS driven by increased PYY contributes to their greater decrease in fat mass by promoting lipolysis in adipocytes via activation of β3-adrenergic receptor that causes an increase in both the expression and the activity of UCP-1 and UCP-2 (48). The higher UCP-2 expression in subcutaneous fat, along with elevated plasma PYY and hypothalamic PYY mRNA, found in RYGB-S rats is in accordance with the enhanced activation of SNS in RYGB-S rats.

The decrease in plasma glucose levels is due to the lower caloric intake, because it also occurs in PF rats. But the fact that the glucose levels in RYGB-S rats were lower than in PF rats suggests another mechanism, such as the activation of β2-adrenergic receptor in skeletal muscle by SNS stimulation leading to an increase in glucose oxidation (5), concurring with the higher plasma PYY levels and the simultaneous higher hypothalamic PYY expression in RYGB-S rats.

We did not find changes in FFA, probably because of the dual effect of SNS activation. On one hand, the stimulation of hormone-sensitive lipase leads to the breakdown of the triglycerides (48), decreasing plasma triglyceride concentration and thus increasing FFA levels, while on the other hand, the SNS activation also stimulates fatty acid oxidation in skeletal muscle (5), leading to the absence of net changes in plasma FFA. Similar changes in peripheral nutrients and hormones have been reported by other investigators in humans and in biliary pancreatic models of weight loss (6, 29). However, the data concerning the changes in plasma ghrelin remain inconsistent because they may be related to the size of the gastric pouch, afferent limb length, current body mass index, degree of weight loss, and the division of some vs. all autonomic fibers innervating the stomach and the foregut (18).

The vagus is considered a sensory rather than an effector nerve since it contains ∼95% afferent and 5% efferent fibers (61). Although we try to preserve the vagus nerve in our model the possibility exists that in RYGB-F rats this nerve and its gastric fibers were unintentionally divided or injured. Therefore, the satiety signals borne by the afferent branch, such as CCK, which enhances the process of satiation by activating the abdominal vagal afferent neurons that innervate the stomach and the duodenum (73) and acting synergistically with leptin (58), would not reach the nucleus of the solitary tract and the area postrema in the brain stem. Moreover, some of the vagal afferent fibers innervating the stomach are also distension responsive (65), so that if the vagal afferent pathway was divided or seriously interrupted in our RYGB-F rats, these vagally dependent satiety signals would not reach the brain, thus leading to the greater caloric intake observed in RYGB-F rats. The putative loss of these vagal signals would contribute to the RYGB operation failure inducing a degree of weight reduction in RYGB-F rats comparable to that achieved in RYGB-S rats (Fig. 1). If that was the case, the weight reduction in RYGB-F rats could be attributed to just a reduction of caloric intake. Supporting this explanation is the fact that the degree and pattern of weight loss were the same in RYGB-F and PF rats.

On the other hand, the stimulation of the efferent vagus promotes energy storage by opposing the effects of SNS activity (50). In our study, vagal celiac efferent firing rate in RYGB rats did not change after glucose gavage into the gastric pouch (Fig. 4), suggesting that the RYGB operation leads to the loss of the vagal efferent signal from the brain, resulting in decreased peristalsis, nutrient absorption, and partitioning of nutrients into adipose tissue, and thus lower energy storage (43). This interpretation concurs with our previous finding (74) showing that RYGB delays gastric emptying and increases intestinal transit time. Therefore, we postulate that in RYGB-S rats the nerve pathway coming from the dorsal motor nucleus of the vagus nerve was divided or injured, eliminating or attenuating these energy-preservative effects of efferent vagus, while in RYGB-F rats it was preserved, explaining why they had a greater body weight compared with RYGB-S rats. However, if this was the case, we should also have observed higher insulin levels in RYGB-F rats compared with RYGB-S rats, because vagal stimulation activates β-cells to produce more insulin.

A decrease in skeletal muscle mitochondrial mass and function is associated with the accelerated rate of fat recovery (catch-up fat) and insulin resistance that are characteristic features of weight recovery after caloric restriction (16). Our findings are consistent with this fact, because in RYGB-S rats lower caloric intake was associated with lower mitochondrial mass than in RYGB-F rats (Fig. 5). The lower number of mitochondria in RYGB-S rats did not lead to lower energy expenditure and thus to a greater fat mass. Instead, a lower fat mass was found in RYGB-S rats, suggesting that a disruption in the fat storage process or alterations in adipogenesis in RYGB-S rats had occurred, as supported by our finding of the higher expression of adipose differentiation-related protein in subcutaneous fat in these rats (Fig. 6). The impaired adipogenesis in RYGB-F and PF rats potentially leads to a progressive adipocyte hypertrophy, and an increased deposition of fat in skeletal muscle, liver, myocardium, and pancreas (20).

Could it be that successful RYGB induces changes in mitochondria morphology and function [smaller mitochondria have lower oxidative capacity and lower ATP production (30)] and thus induces lower energy storage and anabolic capacity? Fiber composition in skeletal muscle depends on both developmental factors and physiological cues such as patterns of innervation, functional demands, or hormonal signals (59). RYGB could induce muscle plasticity aimed at increasing the proportion of glycolytic fibers (fast-twitch fibers), which have lower dependence on lipids and higher capacity to shift between glucose and lipids as fuel substrates than slow-twitch glycolytic fibers. This postulate is supported by the fact that while fast-twitch fibers lack rich mitochondrial networks, slow-twitch fibers have the highest mitochondrial content of any muscle-type fibers and could be adapted to the depleted fat stores in RYGB-S rats that become relatively more dependent on glucose utilization, as suggested by our finding of significantly higher levels of mRNA of 1-α enolase, an enzyme involved in glycolysis, in RYGB-S (Fig. 6).

In skeletal muscle, UCP-2 and UCP-3 expression are upregulated during starvation and downregulated during refeeding (69), emphasizing the role of UCPs in the regulation of lipids as fuel, rather than just as mediators of thermogenesis (69). Thus their suppression would account for the increase in metabolic efficiency and for the subsequent energy conservation to overcome the energy deficit resulting from starvation. The decrease in mitochondrial mass in RYGB-S rats, which have a lower caloric intake and greater loss of fat mass vs. RYGB-F rats, which have a higher caloric intake and body weight and therefore a lower energy deficit, occurs as a natural response to the state of energy deficiency sensed by the RYGB-S rats. Supporting this idea is the fact that during the body weight plateau, food efficiency is higher than during the previous period (Fig. 2). However, we can also invoke other mechanisms in RYGB-S rats overriding the enhanced efficiency of body fat deposition resulting from the decrease in UCP-2 and UCP-3 expression. One of these mechanisms would be an increase in the expression and/or activity of UCP-1 in BAT dissipating energy as heat (7) and resulting from enhanced sympathetic neural modulation, associated with the higher plasma PYY, and the increased hypothalamic expression of PYY, CRF receptor-2, OX2R, and orexin. The reduction in fat mass in RYGB-S results in a higher ratio of surface area to volume and reduced insulation leading to higher thermoregulatory needs.

Although we expected an increase in mitochondrial mass in RYGB-S vs. RYGB-F rats, the discrepancy may be due to an increase in the expression and/or activity of UCP-2 and UCP-3 in skeletal muscle whose magnitude was sufficiently large to overcome the decrease in mitochondrial mass. The increase in the expression of UCP-2 in subcutaneous fat (2.8-fold) and hypothalamus (1.5-fold) in RYGB-S vs. RYGB-F rats (data not shown) would support this argument and would result in increased mobilization of lipids and their utilization (69), accounting for the lower fat mass in the RYGB-S rats.

Finally, to determine whether the success or failure of the RYGB operation was not due to different genetic backgrounds in Sprague-Dawley rats, we screened for the possibility of genetic admixture effects in these random-bred Sprague-Dawley rats. Because the lengths of the STRs from the RYGB-S and RYGB-F rats were all in the expected ranges for Sprague-Dawley-derived rat strains (except for the D1RAT193 STR of 1 RYGB-F rat but not the other rats; data not shown), these results indicate that it is unlikely that failure of RYGB was due to genetic background differences. However, evidence for heterogeneity was found at least at two different STRs at five of the eight loci (D1RAT193, D1MIT32, D2RAT88, D3MGH16, and D3MIT13). The sum of these data indicate that there was considerable variation between rats of even the same treatment group, and strongly suggest that the demonstrated failure of RYGB in some rats was due to common physiological or epigenetic adaptations associated with the experimental paradigm.

The present findings provide evidence that success of RYGB in inducing a sustained weight loss engages many different systems and physiological processes that effectively oppose the compensatory processes to weight loss (starvation response). Specifically, this success is achieved not only by RYGB-induced decreased caloric intake and increased fecal output, but also via an activation of the SNS, driven by increased PYY, CRF, and orexin signaling, decreasing food efficiency and energy storage, demonstrated by reduced fat mass associated with the upregulation of mitochondrial UCP-2 in fat. These events override the compensatory response to the drop in leptin levels aimed at conserving energy, which is more robust in those rats that failed RYGB. Also, our data suggest an important role of the vagus in the outcome of RYGB surgery, whose contribution to the weight loss in this procedure has not been adequately explored.


M. M. Meguid is supported, in part, by the American Society for Bariatric Surgery, the American Diabetes Association, and a SUNY Upstate Intramural Fund grant. P. S. Brookes is funded by National Heart, Lung, and Blood Institute Grant HL-071158.


We thank Dr. A. Krol and Dr. Robert Ploutz-Snyder for their scientific advice and Karen Hughes and Karen Gentile for their technical assistance.


  • 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