Article pubs.acs.org/jpr
Urinary Phenotyping Indicates Weight Loss-Independent Metabolic Effects of Roux-en‑Y Gastric Bypass in Mice Florian Seyfried,*,†,§,⊗ Jia V. Li,*,‡,⊗ Alexander D. Miras,§,# Nina L. Cluny,∥ Matthias Lannoo,+ Wiebke K. Fenske,¶ Keith A Sharkey,∥ Jeremy K. Nicholson,‡ Carel W. le Roux,§,○,⊥ and Elaine Holmes‡,⊥ †
Department of General- and Visceral, Vascular- and Pediatric Surgery, University Hospital of Wuerzburg, Germany Department of Investigative Medicine, Imperial College London, London, United Kingdom ‡ Department of Surgery and Cancer, Faculty of Medicine, Imperial College London, London, United Kingdom # Molecular and Metabolic Imaging Group, Clinical Sciences Centre, Imperial College London, Hammersmith Hospital, London, United Kingdom ∥ Hotchkiss Brain Institute and Snyder Institute for Chronic Diseases, Department of Physiology and Pharmacology, University of Calgary, Calgary, Alberta, Canada + Department of Abdominal Surgery, University of Leuven, Belgium ¶ Department of Internal Medicine, University Hospital of Wuerzburg, Germany ○ Experimental Pathology, Conway Institute, School of Medicine and Medical Sciences, University College Dublin, Ireland §
ABSTRACT: Patients with a body mass index (BMI) above 35 kg/m2 with metabolic diseases benefit from Roux-en-Y gastric bypass (RYGB) independently of their final BMI and the amount of body weight lost. However, the weight loss independent metabolic effects induced by RYGB remain less well understood. To elucidate metabolic changes after RYGB, 1 H NMR spectroscopy-based urine metabolic profiles from RYGB (n = 7), ad libitum-fed sham (AL, n = 5), and bodyweight-matched sham (BWM, n = 5) operated mice were obtained. Gut morphometry and fecal energy content were analyzed. Food intake and body weight of RYGB mice were significantly reduced (p = 0.001) compared to sham-AL. There was a strong tendency that BWM-shams required less food to maintain the same body weight as RYGB mice (p = 0.05). No differences were found in fecal energy content between the groups, excluding malabsorption in RYGB animals. Unlike RYGB-operated rats, gut hypertrophy was not observed in RYGB-operated mice. Urinary tricarboxylic acid cycle intermediates were higher in the sham groups, suggesting altered mitochondrial metabolism after RYGB surgery. Higher urinary levels of trimethylamine, hippurate and trigonelline in RYGB mice indicate that the RYGB operation caused microbial disturbance. Taken together, we demonstrate for the first time that there are RYGB specific metabolic effects, which are independent of food intake and body weight loss. Increased utilization of TCA cycle intermediates and altered gut microbial-host co-metabolites might indicate increased energy expenditure and microbial changes in the gut, respectively. KEYWORDS: gastric bypass, bariatric surgery, mouse model, nuclear magnetic resonance, metabolic profiling
■
INTRODUCTION The worldwide prevalence of obesity is increasing constantly with an estimated 1.5 billion adults which are overweight and 500 million which are obese.1 Obesity, along with its associated comorbidities, has emerged as one of the major socioeconomic burdens of the 21st century. Bariatric surgery has been shown to be safe, cost-effective and the most successful therapeutic option to induce and maintain weight loss and to improve obesity-related morbidity and mortality.2,3 Bariatric surgery has also been given the term “metabolic surgery” due to its favorable effects on various aspects of metabolism and its increasing use for the treatment of metabolic disorders independent of obesity.4 Understanding the mechanisms by © 2013 American Chemical Society
which metabolic surgery causes weight loss and resolution of type 2 diabetes (T2D) will provide new avenues for exploring less invasive methods for attaining weight loss and improving health. Roux-en-Y gastric bypass (RYGB) is the most commonly performed bariatric procedure.5 The rapid time course and disproportionate impact on glycaemia, compared to other weight loss strategies, suggest that the mechanism of action of RYGB in part may be independent of or additive to weight loss.6 The RYGB changes several anatomical and physiological Received: October 2, 2012 Published: January 21, 2013 1245
dx.doi.org/10.1021/pr300909v | J. Proteome Res. 2013, 12, 1245−1253
Journal of Proteome Research
Article
mobilized. The stomach was divided 3 mm below the gastroesophageal junction in order to create a small pouch. The left gastric vessels and vagal fibres were gently mobilized aborally to avoid ischemia of the remnant stomach. After the stomach remnant was closed, the aboral jejunum was anastomosed to the small pouch in an end-to-side fashion. The jejunum was followed aborally to create an alimentary limb of 10 cm. Here, an end-to-side jejunojejunostomy (interrupted Dafilon 9-0) between the biliopancreatic and alimentary limb was performed to create the common channel. Sham-Surgery. A midline laparotomy was performed and the small bowel and the gastro-esophageal junction were mobilized. A gastrostomy (5 mm) was then performed on the anterior wall of the stomach and a jejunostomy performed 10 cm aboral to the pylorus with subsequent closure. The abdominal wall was closed using Prolene 6-0 (continuous suturing). The skin was closed with Monosyn 7-0 (intracutaneous continuous suturing).
parameters which include (i) reduction of stomach size; (ii) bypassing of the stomach and first 30−50 cm of the proximal small bowel; (iii) bile flow alteration; (iv) earlier contact of undigested food with the mid jejunum; and (v) vagal manipulation. Rodent models of gastric bypass surgery have been proven to be valuable and reliable tools in the study of the physiological changes after RYGB.7 We recently established and phenotyped a mouse model of RYGB, by mimicking the human anatomical rearrangements in C57Bl6 mice.8 A significant and sustained reduction in food intake and weight compared to sham-operated ad libitum-fed mice was consistent with results obtained from humans and rat models post RYGB. Investigation of metabolic changes in rats after RYGB via spectroscopic profiling has demonstrated that the urinary tricarboxylic acid cycle intermediates, such as succinate, 2oxoglutarate, citrate, and fumarate were depleted after surgery, while the urinary excretion of a wide range of host−microbial co-metabolites including phenylacetylglycine, 4-cresyl glucuronide, 4-cresyl sulfate and 4-hydroxyphenylacetate increased.9 These alterations, together with elevated levels of fecal amine metabolites in the RYGB rats indicated that significant changes in energy, amine metabolism, and modulation of host−gut microbial co-metabolism take place after surgery.9,10 Serum metabolite profiles, as indicated by altered serum levels of 4cresyl sulfate and indole-3-acetic acid, also suggested that a profound change of the gut microbiota occurs after surgery.11 However, it was not determined from these studies which of the aforementioned metabolic alterations are attributable to the surgery per se or which are simply a consequence of weight loss. Additionally, it has been shown that initial body mass index (BMI) and postoperative weight loss are poor predictors of the efficacy of RYGB in patients with diabetes, dyslipidaemia and the metabolic syndrome.12,13 We hypothesized that the intestinal rearrangements after RYGB induce weight loss-independent metabolic changes, which in turn contribute to the improved metabolic status after surgery. Using a NMR spectroscopy-based metabonomic platform, we investigated the urinary profiles of mice that underwent RYGB and sham operations. We also investigated the weight loss dependent and independent metabolic alterations, by comparing the RYGB group with both a shamoperated, ad libitum fed group and a sham body weight matched group, which was food restricted to achieve the same body weight as the RYGB group.
■
Experimental Protocol and Sample Collection
After 1 week of acclimatization, male C56BL6 mice (24-week old, body weight (BW) 28.8 ± 0.9 g) were randomized into three groups: (1) RYGB-operated mice (RYGB, n = 7); (2) sham-operated mice on ad libitum feeding (AL, n = 5); (3) sham-operated mice with food restriction to match the body weight of the RYGB group (BWM, n = 5). BWM-shams were therefore offered a limited amount of food twice a day (7 a.m. and 7 p.m.). Prior to surgery, mice were food deprived for 6−8 h. On days 1 and 2 post operatively, animals were offered liquid diet (Ensure) only, followed by normal chow diet from day 3 onward. BW and food intake were measured daily for 60 days. Urinary samples from all animals were collected on postoperative day 55 between 9:00 and 11:00 a.m. At least 30 μL of urine was collected directly into a Petri dish from each mouse, by gently rubbing the abdomen. The urine sample was then transferred into an Eppendorf tube and stored at −40 °C. Fecal Energy Content Analysis
Feces were collected over 24 h on postoperative days 15 and 48 from all animals, dried in an oven and weighed (Table 1). Calorie content was measured using ballistic bomb calorimetry.14 Table 1. Summary of Various Methods for Sample Analysis type of analysis time point
METHODS
Day Day Day Day
Animals and Surgery Procedure Description
Male C57BL/6 (H-2b) mice from Charles River were housed individually under a 12/12 h light/dark cycle at a room temperature of 21 ± 2 °C. Water and standard chow were available ad libitum, unless otherwise stated. All experiments were performed under a license issued by the Home Office UK (PL 70-6669). Surgery was performed according to a standardized protocol reported previously.8 Briefly, anesthesia was induced and maintained with Isoflurane/oxygen. Caprofen and Buprenorphine were injected subcutaneously for pain relief and amoxicillin was given intraperitoneally as a prophylaxis. RYGB Surgery. Following a midline laparotomy, the jejunum was transected 10 cm aboral to the pylorus. The gastroesophageal junction was exposed and the esophagus was
15 48 55 60
fecal content analysis
metabolic profiling analysis
morphology analysis
24-h feces 24-h feces -
urine -
gut tissues
Sample Preparation and 1H NMR Spectroscopy Analysis
A total of 30 μL urine of each urine sample was mixed with 25 μL of 0.2 M sodium phosphate buffer (pH = 7.4, 0.01% sodium 3-(trimethylsilyl) propionate-2,2,3,3-d4 (TSP), 3 mM Na3N, 25% D2O) and centrifuged at 11 000g for 10 min. A total of 50 μL of supernatant was transferred into a microtube of 1.7-mm outer diameter pending NMR analysis. 1 H NMR spectra of all samples were obtained using a Bruker DRX 600 MHz spectrometer (Rheinstetten, Germany) with a 5-mm TXI probe operating at 600.13 MHz. D2O solvent in the 1246
dx.doi.org/10.1021/pr300909v | J. Proteome Res. 2013, 12, 1245−1253
Journal of Proteome Research
Article
Figure 1. (Left) Body weight change for RYGB (▲) (n = 7), sham-operated ad libitum fed (□) (n = 5) and sham-operated body weight matched (◇) (n = 5). RYGB vs sham ad libitum fed, **p < 0.01. Sham ad libitum fed vs sham-operated body weight matched °°°p < 0.001. Statistical differences determined by GEE. Data are shown as mean values ± SD. (Right) Average daily food intake over 60 days for RYGB (black), sham ad libitum (white), sham BWM (gray) data are shown as mean ± SD (**p < 0.01 and ***p < 0.001, one-way ANOVA with posthoc Bonferroni test.
phosphate buffer was used to lock the field. A standard onedimensional (1-D) NMR pulse [recycle delay (RD)-90°-t1-90°tm-90°-acquire free induction decay (FID)] was employed for the acquisition of all spectra. The water peak was suppressed by selective irradiation during RD of 2 s and mixing time (tm) of 100 ms and t1 was fixed to 3 μs. The 90° pulse length was adjusted to approximately 10 μs. A total of 256 scans were recorded into 64 k data points with a spectral width of 20 ppm. An exponential function was applied to the FID prior to the Fourier transformation to achieve a line broadening of 0.3 Hz.
Gut Morphometry
Small bowel segments (2 cm) of the alimentary, biliopancreatic limb and common channel from the RYGB mice and corresponding segments of jejunum, duodenum, and ileum from the sham-AL mice were collected. Segments were opened on the mesenteric border, washed with phosphate buffered saline (PBS) and fixed overnight at 4 °C in Zamboni’s fixative (2% paraformaldehyde, 15% picric acid, pH 7.4). Transverse segments from each region were incubated in 20% sucrose in PBS overnight at 4 °C and then embedded in optimum cutting temperature compound. Sections of intestine (12 μm) were cut on a cryostat, thaw mounted onto slides coated with poly-Dlysine and stored at −20 °C until use. Sections were then processed for hematoxylin and eosin staining. Sections were immersed in Ehrlich’s Alum Hematoxylin for 4 min, rinsed in distilled water and then dipped 2 or 3 times in 0.5% acid alcohol and rinsed in distilled water. Sections were then soaked in Scott’s Blueing for 30 s before rinsing in distilled water for 30 s. Next, the sections were dipped once in Eosin Y acid, and again rinsed in distilled water. Slides were then covered by a coverslip sealed with bicarbonate-buffered glycerol and the sections were examined for morphometric analysis. Slides were viewed under a Zeiss Axioplan (Zeiss, Jena, Switzerland) microscope and images were captured using a Retiga 2000R digital camera (QImaging, Surrey, BC, Canada). Muscle thickness, mucosal thickness, crypt depth and villus height were measured from 5 crypt-villus units per tissue and the muscle layer directly below using ImageJ (NIH) software.
Data Analysis
Statistical Analysis of Body Weight, Food Intake, and Fecal Energy. All data were normally distributed and are expressed as mean ± SD. One-way ANOVA with Bonferroni correction for multiple comparisons was used to compare the food intake data obtained from the three experimental groups. Body weight changes were analyzed using generalized estimating equations (GEE) in commercial statistical software (Stata 9.1, Statacorp, College Station, TX). P < 0.05 was considered significant. Spectral Data Reduction and Multivariate Statistical Data Analyses. Urinary spectra were automatically phased, baseline corrected (polynomial) and calibrated to the TSP peak at δ 1H 0.00 in TopSpin 3.0 (Bruker, Rheinstetten, Germany). Spectra ranging from 0 to 10 ppm were digitized in MATLAB 7.14 platform (MathWorks, R2012a) with a chemical shift segment width of 0.0005 ppm, resulting into 20 000 data points. Spectral regions, containing water (δ 4.62−5.17) and urea (δ 5.57−6.15) peaks, and noise only (δ 0−0.35), were removed. Spectral alignment15 and probabilistic quotation normalization were applied to the remaining spectral data prior to employing multivariate data analysis approaches. Principal component analysis (PCA) and partial least-squares 2-discriminant analysis (PLS2-DA)16 with a unit variance (UV) scaling method were performed in SIMCA 13.0 (Umetrics) and orthogonal-partial least-squares-discriminant analysis (O-PLSDA) was carried out in MATLAB with one predictive and one orthogonal component calculated based on the UV-scaled spectral data. OPLS regression analysis was used to correlate urinary metabolic profiles with intestinal morphometric data in RYGB-operated and SHAM-AL mice, respectively. Metabolite assignments were made by 2-dimensional NMR experiments (e.g., 1H−1H correlation spectroscopy and 1H−1H total correlation spectroscopy) applied to selected samples.17
■
RESULTS
Body Weight and Food Intake
RYGB led to a significant and sustained weight loss compared to AL-sham mice (AL-sham: 30.5 ± 0.7 g vs RYGB 28.0 ± 1.4 g, p = 0.0012 eight weeks after surgery) (Figure 1). RYGB mice consumed less normal chow than sham AL controls per day (sham: 4.6 ± 0.1 g/day vs RYGB: 4.3 ± 0.2 g/day, p < 0.01) during the 60-day follow up period. Food intake of body weight matched BMW-shams was lower in comparison to what RYGB animals consumed (BMW-sham 4.1 ± 0.1 g/day vs RYGB 4.3 ± 0.2 g/day, p = 0.05) (Figure 1). Fecal Energy Content
No difference was observed in dry 24-h fecal mass between RYGB (0.87 ± 0.2 g) and either of the sham operated groups (BWM-sham: 0.75 ± 0.02; AL-sham: 0.76 ± 0.05 g, p = 0.397). Although RYGB-animals consumed more food compared to BWM-sham mice, there was no difference in their fecal energy 1247
dx.doi.org/10.1021/pr300909v | J. Proteome Res. 2013, 12, 1245−1253
Journal of Proteome Research
Article
Figure 2. 1H NMR urinary spectra of sham-AL (black), sham-BWM (green) and RYGB (red)-operated mice (A); a PCA scores plot (B) and a PLSDA scores plot (C, Q2Y = 0.47, R2Y = 79.3%) of urinary NMR spectral data obtained from RYGB (red), sham-BWM (green) and sham-AL (black) mice 55 days after surgery. Results from the permutation test (D) for the RYGB group suggest a valid PLS-DA model. The vertical axis is the R2Y and Q2Y values of each model and the horizontal axis represents the correlation coefficient between permuted Y-vectors and the original Y-vector.
content as assessed by bomb calorimetry (Sham-AL: 0.59 ± 0.02 kcal/g; sham-BWM: 0.58 ± 0.01 kcal/g; RYGB: 0.58 ± 0.03 kcal/g, p = 0.932).
performed for each group (RYGB, sham-AL and sham-BWM) and the results suggested that the PLS-DA model was significant in distinguishing RYGB from sham groups (Figure 2D). However, the permutation tests on sham groups indicate no significant difference between either of the sham groups and the remaining two groups. To further investigate metabolic differences, pairwise comparisons between each two of the groups were subsequently carried out using O-PLS-DA analysis. An O-PLS-DA analysis between sham-AL and sham-BWM groups (R2X = 42.5%; Q2Y = 0.55; R2Y = 90.1%) showed significantly higher levels of urinary hippurate (p < 0.05) and putatively assigned 3-(3-hydroxphenyl) propionate (p < 0.01) in the heavier body weight group (Sham-AL) compared to the Sham-BWM group, although ANOVA of cross-validated residuals (CV-ANOVA) gave a p-value of 0.32 for this model. The pairwise OPLS-DA comparisons between sham-AL or BWM-sham and RYGB are shown in Figure 3, in which upward pointing peaks indicate higher concentrations of metabolites in the sham groups and vice versa and where the color of a peak
RYGB Operation Significantly Alters Urinary Metabolism in Mice
Three representative 1H NMR urinary spectra obtained from sham-AL, sham-BWM and RYGB-operated animals are shown in Figure 2A. A metabolic shift was observed between the RYGB mice and the two sham control groups as shown in the PCA scores plot (Figure 2B). Separation of the RYGB from the other groups in the first principal component (PC1) is mainly influenced by the altered urinary taurine concentrations and tricarboxylic acid cycle intermediates including citrate, 2oxoglutarate, succinate and fumarate. A more subtle metabolic difference between the Sham-AL and Sham-BWM groups, reflecting different body weight due to caloric restriction in the BWM-sham, was characterized in PLS-DA scores plots (Figure 2C) where differentiation of the control groups occurred in the second component. A number of 500 permutations were 1248
dx.doi.org/10.1021/pr300909v | J. Proteome Res. 2013, 12, 1245−1253
Journal of Proteome Research
Article
Figure 3. O-PLS-DA coefficient loading plots of pairwise comparisons of urinary spectra of Sham-AL and RYGB (upper, Q2Y = 0.85, R2X = 37.8%, R2Y = 98.4%), and Sham-BWM and RYGB (lower, Q2Y = 0.82, R2X = 35.7%, R2Y = 97.4%). Peaks pointing upward represent higher concentrations of these metabolites in either Sham-AL or Sham-BWM group compared with RYGB-operated animals, and vice versa. The colors of peaks demonstrate their correlation coefficients to the classes, rising from blue (no correlation) to red (highly correlated with discrimination between classes).
Table 2. Significance Levels of Urinary Biomarkers Distinguishing between Different Classes Including Sham-AL, Sham-BWM and RYGBa student t test (p-values)
a
metabolites [spectral regions, ppm]
AL vs. BWM
AL vs. RYGB
BWM vs. RYGB
ANOVA (p-values) All Groups
2-oxoglutarate [2.429−2.466] succinate [2.41−2.417] citrate [2.532−2.591] trimethylamine [2.864−2.92] taurine [3.28−3.288] trigonelline [4.437−4.444] phenylacetylglycine [7.407−7.445] hippurate [7.816−7.855] 3-(3-hydroxyphenyl)propionate [6.755−6.773]
0.931 0.835 0.948 0.181 0.786 0.101 0.184 0.013 0.002
0.113 0.011 0.052 8.7 × 10−6 0.001 0.012 0.026 0.204 6.1 × 10−5
0.166 0.032 0.136 3.1 × 10−6 0.036 0.001 0.143 0.014 0.315
0.128 0.008 0.063
