Metabolic Profiling of Liver from Hypercholesterolemic Pigs Fed Rye

Hypercholesterolemic Pigs Fed Rye or Wheat Fiber and from Normal Pigs. High-Resolution Magic. Angle Spinning 1H NMR Spectroscopic Study. Hanne C...
0 downloads 0 Views 188KB Size
Anal. Chem. 2007, 79, 168-175

Metabolic Profiling of Liver from Hypercholesterolemic Pigs Fed Rye or Wheat Fiber and from Normal Pigs. High-Resolution Magic Angle Spinning 1H NMR Spectroscopic Study Hanne C. Bertram,*,† Iola F. Duarte,‡ Ana M. Gil,‡ Knud E. Bach Knudsen,§ and Helle N. Lærke§

Department of Food Science and Department of Animal Health, Welfare and Nutrition, Danish Institute of Agricultural Sciences, Box 50, DK-8830 Tjele, Denmark, and Department of Chemistry, CICECO, University of Aveiro, 3810-193 Aveiro, Portugal

This study presents the first application of a high-resolution magic angle spinning 1H NMR approach to elucidate the metabolic effects of a hypercholesterolemic condition and two high-fiber diets based on rye and wheat bread, respectively, in intact pig liver biopsy samples. Standard 1D and spin-echo 1H spectra were acquired on a total of 20 biopsy samples, and 2D total correlation spectroscopy experiments were carried out on selected samples for assignment of the observed resonances. Principal component analyses and partial least-squares regression discriminant analysis revealed differences in the hepatic lipid content and choline-containing compounds between normal and hypercholesterolemic pigs. In addition, the results demonstrated that the liver metabolite profile of hypercholesterolemic pigs fed a high-fiber rye bread differed from that of pigs fed high-fiber wheat bread with respect to both the lipoprotein fractions and the cholinecontaining compounds. These findings suggest that earlier reports on high-fiber rye diet-induced effects on plasma HDL/LDL content partially can be ascribed to effects on liver cholesterol metabolism and that the hepatic phospholipase pathways of phosphatidylcholine breakdown are affected by the high-fiber rye diet. Epidemiological studies have shown that consumption of whole-grain products reduces the risk of developing several lifestyle diseases.1,2 A reduction of plasma cholesterol is probably the most important effect responsible for reducing the risk of developing cardiovascular diseases. Hypocholesterolemic properties of oat and oat products have been widely established.3-8 In addition, animal and human studies on barley, rye, and rice and * To whom correspondence should be addressed. E-mail: hannec.bertram@ agrsci.dk. † Department of Food Science, Danish Institute of Agricultural Sciences. ‡ University of Aveiro. § Department of Animal Health, Welfare and Nutrition, Danish Institute of Agricultural Sciences. (1) Jacobs, D. R.; Meyer, K. A.; Kushi, L. H.; Folson, A. R. Am. J. Clin. Nutr. 1998, 89, 248-257. (2) Jacobs, D. R.; Meyer, K. A.; Kushi, L. H.; Folson, A. R. Am. J. Public Health 1999, 89, 322-329. (3) Anderson, J. W.; Spencer, D. B.; Hamilton, C. C.; Smith, S. F.; Tietyen, J.; Bryant, C. A.; Oeltgen, P. Am. J. Clin. Nutr. 1990, 52, 495-499.

168 Analytical Chemistry, Vol. 79, No. 1, January 1, 2007

on isolated fractions from these cereals indicate that they have similar properties.8-14 However, the underlying mechanisms associated with the grain-induced cholesterol reduction are still far from fully understood. Effects on cholesterol or lipid metabolism or on microbial fermentation in the colon are possible mechanisms,15 where the former would involve an effect on liver metabolism. Human and porcine lipoprotein and apolipoprotein profiles are similar when pigs are fed a high-fat diet,16,17 and under experimental conditions, pigs develop complex atherosclerotic lesions morphologically similar to those found in humans in just 3-8 months.18 Although pigs lack plasma cholesteryl ester transfer protein activity,17,18 cholesteryl esters can be transferred to LDL, and reverse cholesterol transport takes place in the pig as in (4) Anderson, J. W.; Gilinsky, N. H.; Deakins, D. A.; Smith, S. F.; Neal, D. S.; Dillion, D. W.; Oeltgen, P. R. Am. J. Clin. Nutr. 1991, 54, 678-683. (5) Newman, R. K.; Klopfensterin, C. F.; Newman, C. W.; Guritno, N., Hofer, P. J. Cereal Chem. 1992, 69, 240-244. (6) Ripsin, C. M.; Keenan, J. M., Jacobs, D. R.; Elmer, P. J. Welch, R. R.; Vanhorn, L.; Liu, K.; Turnbull, W. H.; Thye, F. W.; Kestin, M.; Hegsted, M.; Davidson, D. M.; Davidson, M. H.; Dugan, L. D.; Demarkwahnefired, W.; Beling, S. JAMA, J. Am. Med. Assoc. 1992, 267, 3317-3325. (7) Jenkins, D. J. A.; Kendall, C. W. C.; Vuksan, V.; Vidgen, E.; Parker, T.; Faulkner, D.; Mehling, C. C.; Garsetti, M.; Testolin, G.; Cunnane, S. C.; Ryan, M. A.; Corey, P. N. Am. J. Clin. Nutr. 2002, 75, 834-839. (8) Delaney, B.; Nicolosi, R. J.; Wilson, T. A.; Carlson, T.; Frazer, S.; Zheng, G. H.; Hess, R.; Ostergren, K.; Haworth, J.; Knutson, N. J. Nutr. 2003, 133, 468-475. (9) Behall, K. M.; Scholfield, D. J.; Hallfrisch, J. J. Am. Coll. Nutr. 2004, 23, 55-62. (10) Yang, J. L.; Kim, Y. H.; Lee, H. S.; Moon, Y. K. J. Nutr. Sci. Vitaminol. 2003, 49, 381-387. (11) Leinonen, K. S.; Poutanen, K. S.; Mykkanen, H. M. J. Nutr. 2000, 130, 164-170. (12) Rieckhoff, D.; Trautwein, E. A.; Malkki, Y.; Erbersdobler, H. F. Cereal Chem. 1999, 76, 788-795. (13) Frigård, T.; Pettersson, D.; Aman, P. J. Nutr. 1994, 124, 2422-2430. (14) Ardiansyah; Shirakawa, H.; Koseki, T.; Ohinata, K.; Hashizume, K.; Komai, M. J. Agric. Food Chem. 2006, 54, 1914-1920. (15) Riottot, M.; Lutton, C. In COST 92 Metabolic and physiological aspects of dietary fibre in food; Lairon, D., Ed.; Commission of the European Communities: Luxembourg, 1993; pp 77-82. (16) Knudsen, K. E. B.; Canibe, N. COST 92. Metabolic and physiological aspects of dietary fibre in foods; Lairon, D., Ed.; Commission of the European Communities: Luxembourg, 1993; pp 123-130. (17) Terpstra, A. H. M.; Lapre, J. A.; de Vries, H. T.; Beynen, A. C. J. Anim. Physiol. Anim. Nutr. 2000, 84, 178-191. (18) Dixon, J. L.; Stoops, J. D.; Parker, J. L.; Laughlin, M. H.; Weisman, G. A.; Sturek, M. Arterioscler. Thromb. Vasc. Biol. 1999, 19, 2981-2992. 10.1021/ac061322+ CCC: $37.00

© 2007 American Chemical Society Published on Web 11/21/2006

Table 1. Ingredients of the Standard Diet, Content in g/100 g

ingredients

standard dieta

wheat barley Triticale rape seed meal soybean meal sunflower meal oat whole grain flour vegetable oil calcium carbonate sugar beet molasses NaCl calcium phosphate vitamin mineral mixtureb

32.9 20.0 10.0 9.0 8.9 8.0 5.7 2.0 1.3 1.0 0.48 0.23 0.20

ingredients

low-fiber wheat diet

high-fiber wheat bread diet

high-fiber rye bread diet

wheat flour rye whole meal flourc Vitacell WF 600 rye brand Lacprodan 87e yeast sugar egg powder rape seed oil lard cholesterol calcium phosphate vitamin mineral mixturef

73.5 0.0 0.0 0.0 1.0 0.0 0.0 15.0 2.0 5.0 0.5 0.0 3.0

52.8 0.0 15.7 0.0 2.5 2.0 1.5 15.0 2.0 5.0 0.5 0.0 3.0

0.0 31.0 0.0 40.0 0.0 2.0 1.5 15.0 2.0 5.0 0.5 0.0 3.0

a Supplemented with 0.16% lysine in the form of 40% lysine + 60% wheat bran and 0.04% D/L-methionine in the form of 40% D/L-methionine + 60% wheat bran. b Containing 2 500 000 IU of vitamin A, 500 000 IU of vitamin D3, 30 g of vitamin E, 1.1 g of vitamin K3, 1.1 g of vitamin B1, 2 g of vitamin B2, 1.65 g of vitamin B6, 5.5 g of D-pantothenic acid, 11 g of niacin, 27.5 mg of biotin, 11 mg of vitamin B12, 25 g of Fe, 40 g of Zn, 13.86 mg of Mn, 10 g of Cu, 99 mg of I, and 150 mg of Se per kg. c Finely ground whole-kernel rye (Valsemøllen, Esbjerg, Denmark). d B3 fin, specialmade ground rye bran (Nordmills, Cerealia, Uppsala, Sweden). e Native whey protein concentrate (Arla Foods Ingredients amba, Viby, Denmark). f Containing (mg/kg) the following: 6642 Ca (PO4) , 4122 NaCl, 16580 CaCO , 286 FeSO ‚7H O, 114 ZnO, 41 Mn O , 92 CuSO ‚5H O, 0.3 KI, 0.8 2 3 3 4 2 3 4 4 2 Na2SeO3‚5H2O, 2.1 Retinoacetate, 0.03 cholexalciferol, 69 R-tocopherol, 2.52 menadione, 4.58 riboflavin, 12.59 D-pantothenic acid, 0.025 vitamin B12, 2.52 vitamin B1, 25.2 niacin, 3.78 vitamin B6, and 0.063 biotin.

humans.17 Accordingly, the pig is considered to be a suitable model for humans for studying the hypocholesterolemic effects of dietary fiber and whole-grain cereals. However, techniques for the direct study of liver metabolism are scarce. Metabolic profiling of intact tissue is now feasible using high-resolution magic angle spinning (HR-MAS) 1H NMR spectroscopy,19,20 and applications have been demonstrated on liver in animal toxicology studies,21-24 in an animal gene therapy study,25 and also for characterization of human liver biopsies from transplant patients.26 1H NMR has the advantage that it provides concurrent detection of all hydrogen-containing molecules in a sample without pretreatment, and the sensitivity is high as compared with, for example, 13C NMR. However, the major drawbacks of 1H NMR are the high cost of instrumentation and the lower sensitivity as compared with MS techniques. Quantification of metabolite concentrations is complicated in HR-MAS measurements;27 however, when combining HR-MAS measurements with multivariate data analysis, quantitative differences between samples are readily displayed. In the present study, HRMAS 1H NMR spectroscopy is used for the first time for metabolic (19) Bollard, M. E.; Garrod, S.; Holmes, E.; Lindon, J. C.; Humpfer, E.; Spraul, M.; Nicholson, J. K. Magn. Reson. Med. 2000, 44, 201-207. (20) Waters, N. J.; Garrod, S.; Farrant, R. D.; Haselden, J. N., Connor, S. C.; Connelly, J.; Lindon, J. C.; Holmes, E.; Nicholson, J. K. Anal. Biochem. 2000, 282, 16-23. (21) Waters, N. J.; Holmes, E.; Waterfield, C. J.; Farrant, R. D.; Nicholson, J. K. Biochem. Pharmacol. 2002, 64, 67-77. (22) Wu, H.; Zhang, X.; Li, X.; Wu, Y.; Pei, F. Anal. Biochem. 2005, 339, 242248. (23) Lucas, L. H.; Wilson, S. F.; Lunte, C. E.; Larive, C. K. Anal. Chem. 2005, 77, 2978-2984. (24) Garrod, S.; Humpher, E.; Connor, S. C.; Connelly, J. C.; Spraul, M.; Nicholson, J. K.; Holmes, E. Magn. Reson. Med. 2001, 45, 781-790. (25) Griffin, J. L.; Blenkiron, C.; Valonen, P. K.; Caldas, C.; Kauppinen, R. A. Anal. Chem 2006, 78, 1546-1552. (26) Duarte, I. F.; Stanley, E. G.; Holmes, E.; Lindon, J. C.; Gil, A. M.; Tang, H.; Ferdinand, R.; McKee, C. G.; Nicholson, J. K.; Vilca-Melendez, H.; Heaton, N.; Murphy, G. M. Anal. Chem. 2005, 77, 5570-5578. (27) Sitter, B.; Lundgren, S.; Bathen, T. F.; Halgunset, J.; Fjosne, H. E.; Gribbestad, I. S. NMR Biomed. 2006, 19, 30-40.

profiling of liver from normal and hypercholesterolemic pigs subjected to high-fiber, whole-grain, rye and wheat bread diets with the aim of investigating potential differences in the metabolite profile between liver from normal pigs and hypercholesterolemic pigs and as a result of diet fiber type. EXPERIMENTAL SECTION Diets. The study included three experimental diets: a standard pig feed, a high-fiber, rye bread-based diet, and a high-fiber, wheat bread-based diet (Table 1). Pigs that served as reference group (non-hypercholesterolemic) were fed the standard pig feed, while the hypercholesterolemic pigs were fed high-fiber wheat or rye bread buns. Prior to administration of the high-fiber diets (highfiber buns), noncontrol pigs were fed a low-fiber, wheat-based diet. In order to select for pigs responding to a high fat and cholesterol intake, to the low-fiber diet was added 0.5% cholesterol plus 15% egg powder, resulting in a total cholesterol content of ∼1.25% during the first two weeks. This was followed by three weeks where essentially the same diet without cholesterol was administered. However, in this second period, 6% cellulose (Vitacel WF 600, LCH A/S) was added to prevent constipation and ulceration. The experimental diets consisted of high-fiber buns either made of rye or wheat flour. The rye buns contained whole-kernel flour and rye bran, while the wheat buns contained wheat flour with added refined wheat fiber (Vitacel WF 600) in order to obtain the same dietary fiber level. Chromic oxide (0.2% as basis) was added to the experimental diets during the final 2.5 weeks of the study. The wheat buns were produced at Cerealia Unibake, Denmark. The rye buns were made at Holstebro Technical College, Denmark. Bread rolls were stored at -20 °C until consumption. Experimental Design and Animals. Pigs from the Danish Institute of Agricultural Sciences swine herd (Foulum, Denmark) were used for the study. The study was carried out in two series of experiments: a substudy with a group of reference pigs and another substudy with hypocholesterolemic pigs. Analytical Chemistry, Vol. 79, No. 1, January 1, 2007

169

Substudy 1: Eight female pigs from two litters served as reference group and were all fed a standard diet (Table 1) until euthanization at an age of ∼6.5 months (body weight of 110.0 ( 6.5 kg). Substudy 2: At a body weight of ∼70 kg, 15 female pigs were introduced to a refined, low-fiber, high-fat, wheat diet containing cholesterol (Table 1) and fed for the following 2 weeks, 2 kg/ day. After 2 weeks on this diet, a fasting blood sample was taken from each pig by venous puncture. Based on their response to the cholesterol diet, a total of 12 hyper-responders were chosen as study subjects for the remaining part of the study. The hyperresponders were selected according to a plasma cholesterol cutoff value of 3.5 mmol/L based on previous findings.16 During the following 3 weeks, the pigs were fed the wheat flour-based diet without egg powder and added cholesterol at a level of 2 kg/day for 2 weeks increasing to 2.5 kg/day in the third week. In the fourth week, the pigs were randomly divided into two groups of six animals each and introduced to the two experimental diets consisting of high-fiber rye or high-fiber wheat breads, respectively. During the experimental period, the pigs were given two meals per day, and they were initially fed 1 kg of feed per meal increasing to 1.5 kg per meal during the last 3.5 weeks of a 10week period with experimental diet. After the 10-week period, the pigs, which now ∼7 months and an average body weight of 138.3 ( 2.4 kg were euthanized with an overdose of pentobarbital and bled, and liver biopsies were taken from the left lateral lobe as quickly as possible, frozen in liquid nitrogen, and stored at -80 °C prior to NMR analysis. The animal experiments complied with the guidelines of the Danish Ministry and Justice with respect to animal experimentation and care of animals under study. NMR Measurements. HR-MAS 1H NMR data were acquired on a Bruker Avance DRX-500 spectrometer, operating at 500.13 MHz for 1H observation, at a temperature of 298 K and a spinning rate of 4 kHz. Twenty to thirty minutes prior to the NMR measurements, the liver biopsies were thawed and 20-30 mg of sample material was washed with D2O saline (0.9%) and packed into a 4-mm-diameter zirconia MAS rotor with a top insert to give a final volume of 15-25 µL. 1H NMR spectra of liver were recorded using a standard 1D pulse sequence (relaxation delay-90°-t190°-tm-90°-acquire FID) in which the water signal is irradiated during the relaxation delay (2 s) and the mixing period (tm ) 100 ms), with t1 being a short delay of 3 µs. A total of 128 transients were collected into 32K data points. A spectral width of 6000 Hz and an acquisition time of 2.73 s were used. In addition, spinecho 1H NMR spectra were obtained using the Carr-PurcellMeiboom-Gill (CPMG) pulse sequence with simple presaturation of the water peak, and a total spin-spin relaxation delay (2nτ) of 240 ms was used. A total of 256 transients were collected into 32K data points. All 1D spectra were processed with a line broadening of 0.3 Hz and a zero filling of 2. To aid spectral assignment, 2D 1H-1H total correlation (TOCSY) spectra were recorded on selected samples. The TOCSY spectra were acquired in the phase-sensitive mode using time proportional phase incrementation and the MLEV17 pulse sequence for the spin lock. For each spectrum, 4K data points with 64 transients/increment and 128 increments were acquired with a spectral width of 4192 Hz in both dimensions. The relaxation delay between successive pulse 170 Analytical Chemistry, Vol. 79, No. 1, January 1, 2007

Figure 1. 500-MHz 1H HR-MAS NMR spectra of a representative liver biopsy sample (spinning speed 4 kHz): (a) standard 1D spectrum; (b) spin-echo CPMG spectrum. Assignment: 1, lipid CH3; 2, lipid (CH2)n; 3, lipid CH2CH2CO; 4, lipid CH2CHdCH; 5, lipid CH2CO; 6, N(CH3)3 in choline, phoshocholine, glycerophosphocholine, TMAO, and betaine; 7, protons in sugars and amino acids; 8, lipid CHdCH; 9, valine and leucine; 10, alanine; 11, creatine; 12, glucose; 13, glycogen R(1-4).

cycles was 2 s, and the mixing time of the MLEV spin lock was 80 ms. Postprocessing of NMR Data. The 1H NMR spectra in the region 10.0-5.6 ppm and the region 4.0-0.0 ppm were used for further data analysis. The spectra were normalized to the whole spectrum to remove any concentration effects, and further analysis was performed using the Unscrambler software version 9.2 (Camo, Oslo, Norway). Principal component analysis (PCA)28 was applied to the centered data to explore any clustering behavior of the samples, and partial least-squares regression discriminant analysis (PLS-DA)28 was performed to explore intrinsic biochemical dissimilarities between predefined sample classes (normal vs hypercholesterolemic pigs fed rye or wheat; rye-fed vs wheat-fed). During all regressions, Martens uncertainty test29 was used to eliminate noisy variables, and all models were validated using full cross-validation.30 RESULTS Figure 1a and b shows typical standard 1H NMR and spinecho 1H NMR spectra acquired on a liver sample under HR-MAS conditions, respectively. The two pulse sequences provide different types of data. The standard 1H spectrum (Figure 1a) includes resonances from both low and high molecular weight compounds and shows principally broad resonances arising from fatty acyl chains in lipid molecules. The presence of glyceryl peaks at ∼4.1, 4.3, and 5.25 ppm reveals that the lipid molecules to a certain extent can be identified as triglycerides. Moreover, phospholipids signals and sharper glucose signals are observed in this spectrum. Contrary, the CPMG pulse sequence attenuates broad signals such as those from lipids due to their shorter T2 relaxation times, thus (28) Martens, H.; Næs, T. Multivariate calibration; Chichester: Wiley, 1989. (29) Martens, H; Dardenne, P. Chemom. Intell. Lab. Syst. 1998, 44, 99-121. (30) Martens, H; Martens, M. Food Qual. Preference 2000, 11, 5-16.

Figure 2. PCA analysis. (a) Score plot showing the first and second principal components for PCA carried out on standard 1D spectra obtained on liver samples; (b) the first X-loading of the PCA shown in (a); (c) score plot showing the second and third principal components for PCA carried out on CPMG spectra obtained on liver samples; (d) the second X-loading of the PCA shown in (c).

improving the analysis of narrow signals from the low molecular weight compounds (Figure 1b). Several signals have been assigned using the information provided by the 2D TOCSY spectra and by comparison with chemical shifts reported in the literature.26,31,32 Some of these assignments are presented in Figure 1. In order to identify possible minor metabolic differences between sample classes, a PCA was performed on both the standard and CPMG 1H spectra. The score scatter plot of PC1 versus PC2 for standard spectra shows a tendency for grouping of the three sample classes along the first score (Figure 2a). For CPMG spectra, the origin of the variance along PC1 spectra could not be identified (data not shown), while the score scatter plot of PC2 versus PC3 shows a tendency for grouping of the three sample classes along PC2 (Figure 2c). However, both for the standard and the CPMG 1H spectra, this tendency for grouping in the score scatter plot is most evident for the reference samples, (31) Bollard, M. E.; Garrod, S.; Holmes, E.; Lindon, J. C.; Humpfer, E.; Spraul, M.; Nicholson, J. K. Magn. Reson. Med. 2000, 44, 201-207. (32) Lindon, J. C.; Nicholson, J. K.; Everett, J. R. Ann. Rep. NMR Spectrosc. 1999, 38, 1-88.

while the tendency is weaker for the hypercholesterolemic samples, and the reference samples are clustered much more closely compared with wheat and rye samples from hypercholesterolemic pigs. Analysis of the loadings (Figure 2b, d) reveals that the tendency for separation into the different sample classes can mainly be ascribed to the intensity of lipid peaks at ∼0.9 and 1.3 ppm, the intensity of the peak at ∼3.25 ppm containing contribution from N(CH3)3 protons, and the region ∼3-4 ppm, which reflects glucose, glycogen, and amino acid protons. PLS-DA, which is a supervised method, was performed on the HR-MAS 1H NMR spectra to investigate the differences in liver metabolites between reference samples and from hypercholesterolemic pigs. Separate analyses were carried out for the rye and wheat diets, and results for standard 1H spectra are displayed in Figure 3. Independent of diet, the PLS-DA scores plot of the first and second components show a clear separation of reference and hypercholesterolemic samples (Figure 3a and c). The regions in the NMR spectrum that most strongly influence the separation between the reference group and hypercholesterolemic rye or wheat samples are evident in the first X-loadings, which are displayed in Figure 3b and d, respectively. In Figure Analytical Chemistry, Vol. 79, No. 1, January 1, 2007

171

Figure 3. (a) PLS-DA scores plot from analysis of standard 1D 1H HR-MAS NMR obtained on liver samples from hypercholesterolemic pigs fed a rye-based diet (closed circles) and normal pigs fed a standard diet (open circles), (b) The first X-loading of the PLS-DA shown in (a); (c) PLS-DA scores plot from analysis of standard 1D 1H HR-MAS NMR obtained on liver samples from hypercholesterolemic pigs fed a wheat-based diet (closed circles) and normal pigs fed a standard diet (open circles); (b) the first X-loading of the PLS-DA shown in (c). In both (a) and (c), a positive value of the X-loading indicates a higher level of the metabolite in hypercholesterolemic pigs compared with normal pigs, while a negative value of the X-loading indicates a lower level of the metabolite in hypercholesterolemic pigs compared with normal pigs.

3b, PC1 loadings are positive for the fatty acyl chain signals in lipidmolecules (∼0.9 and 1.3 ppm), thus indicating that these compounds are more abundant in the liver of hypercholesterolemic pigs fed rye (positive PC1 scores) than that of reference pigs (scores in negative side of PC1 axis). In contrast, the reference animals are characterized by higher hepatic levels of glucose (∼3-4 ppm) and a higher intensity of N(CH3)3 protons, which mainly reflects phospholipids. The major peaks showing differences between reference samples and samples from hypercholesterolemic pigs fed wheat are identical to the peaks discriminating between control and rye samples (Figure 3d); however, in the first loading, the CH2-methylene lipid signal at ∼1.3 ppm comprises both positive and negative values, implying that the difference between reference samples and samples from hypercholesterolemic pigs fed wheat is not solely a concentration effect but merely an effect on the chemical shift of this signal. In order to investigate the differences in liver metabolites between rye bread and wheat bread diets, PLS-DA was also performed on these two sample types without including the 172 Analytical Chemistry, Vol. 79, No. 1, January 1, 2007

reference samples, and results for standard 1H spectra are displayed in Figure 4. The two diets are clearly discriminated (Figure 4a), and this can be ascribed to a higher intensity of saturated lipid signals at ∼0.9, 1.3, and 2.2 ppm and the unsaturated lipid signals at ∼2.0 and 2.7 ppm in liver samples from ryefed pigs compared with wheat-fed pigs and a shift in the position of the ∼3.20-3.25 ppm resonances. To further elucidate the effects of diet on these signals, expanded regions of the mean spectra corresponding to each diet are shown in Figure 5. The peak at ∼3.22 ppm contains contribution from N(CH3)3 protons principally in choline, phosphocholine (PC), and glycerophosphocholine (GPC). Thus, the slight shift observed between the mean spectra (Figure 5a) may reflect different proportions of these compounds depending on type of diet. In the 0.6-2.4 ppm region (Figure 5b), all lipid signals are higher in intensity in liver samples from ryefed pigs compared with those of wheat-fed animals. DISCUSSION Atherosclerosis is a prevalent disease in the industrialized part of the world, and ∼50% of all deaths in this part of the world are

Figure 4. (a) PLS-DA scores plot from analysis of standard 1D 1H HR-MAS NMR obtained on liver samples from hypercholesterolemic pigs fed a rye-based diet (closed circles) and hypercholesterolemic pigs fed a wheat-based diet (open circles), (b) The first X-loading of the PLS-DA shown in (a). Arrows show signals that differ between the two diets: ∼0.9, 1.3, and 2.2 ppm, saturated lipid signals; ∼2.0 and 2.7 ppm, unsaturated lipid signals; ∼3.20-3.25 ppm, N(CH3)3 protons. A positive value of the X-loading indicates a higher level of the metabolite in rye samples compared with wheat samples, while a negative value of the X-loading indicates a lower level of the metabolite in rye samples compared with wheat samples.

due to some type of atherosclerosis. Atherosclerosis is a multifactorial condition with one of the major risk factors being dyslipidemia, which is characterized by an elevated serum triglyceride concentration and alterations in the levels of low-density lipoprotein (LDL) and high-density lipoprotein (HDL) fractions. Hypocholesterolemic properties of dietary intake of whole-grain cereals and products have been widely demonstrated.3-14 However, the knowledge of how the liver metabolism is affected by a hypercholesterolemic condition and how the liver reacts to a dietinduced decrease in serum cholesterol is limited. The present study is the first report on the use of HR-MAS 1H NMR spectroscopy for metabolic profiling of liver from normal and hypercholesterolemic pigs, and the first to elucidate the effects of whole-grain rye bread and wheat bread diets on liver composition and metabolism in hypercholesterolemic pigs. HR-MAS 1H NMR spectroscopy has the advantage that measurements can be performed on intact liver samples without any prior extraction, thereby displaying the metabolite profile without any misrepre-

Figure 5. Expansion of selected regions of average spectra obtained on liver samples from hypercholesterolemic pigs fed a ryebased diet (full line) or a wheat-based diet (dotted line). (a) The spectral region containing contribution from the N(CH3)3. PC, phosphocholine; GPC, glycerophosphocholine; TMAO, trimethylamine N-oxide. (b) A spectral region containing contribution from lipids.

sentation. Unsupervised PCA on 1H spectra obtained on liver from reference and hypercholesterolemic pigs under HR-MAS conditions revealed a tendency for grouping of the different sample classes, which reveals differences in the metabolic profile of the liver between the reference and the hypercholesterolemic pigs (Figure 2a, c). This result demonstrates that HR-MAS 1H NMR spectroscopy may be used for characterization of liver as function of cholesterol status. The loadings from the PCA revealed that the discrimination between liver from reference and hypercholesterolemic pigs mainly could be ascribed to effects on the ∼0.9 and ∼1.3 ppm lipid signals from methyl and methylene fatty acyl chains, respectively (Figure 2b, d). As excess hepatic cholesterol is known to be incorporated into VLDL, which will contribute to the ∼0.9 and ∼1.3 ppm lipid signals, this finding is not unexpected. In addition, the loadings revealed differences in the choline head group signal at ∼3.20-3.25 ppm between liver from reference and hypercholesterolemic pigs. This signal contains among others contribution from lipids located in the cell membrane and also cell membrane breakdown products such as GPC.33 Accordingly, (33) RuizCabello, J.; Cohen, J. S. NMR Biomed. 1992, 5, 226-233.

Analytical Chemistry, Vol. 79, No. 1, January 1, 2007

173

the present data indicate effects of a hypercholesterolemic condition on cell membrane constituents in the liver. A previous study investigating human liver as function of transplantation state also observed effects on the GPC signal,21 which suggests this signal may be an indicator of liver “functionality”. In the present study, a difference in the ∼3.25 ppm signal was evident in both standard 1H spectra and 1H spectra obtained with a spin-echo delay, and while the difference in the standard 1H spectra indicates an effect on the amount of these components, an effect also on the mobility of the components cannot be excluded from the 1H spin-echo spectra. In addition to effects of hypercholestereolemia on the lipid and the choline signals, differences between reference and hypercholesterolemic liver samples were also observed in the ∼3-4 ppm region containing contributions from several protons in glucose or glycogen, glycerol, and amino acids, as these signals were more abundant in normal liver samples compared with hypercholesterolemic liver samples. A higher amount of glycogen in the normal liver compared with hypercholesterolemic liver samples is in agreement with earlier studies showing that highfat diets are associated with decreased hepatic glycogen content34,35 and indicate that the hypercholesterolemic condition is coupled with a reduced liver gluconeogenetic activity. Individual comparison of reference liver samples and hypercholesterolemic liver samples upon either a rye bread-based or wheat bread-based diet using PLS-DA also revealed effects of the hypercholesterolemic condition on the methyl and methylene fatty acyl chains (∼0.9 and ∼1.3 ppm), the choline signal (∼3.25 ppm), and the sugar region (∼3-4 ppm) independent of diet (Figure 3). However, noteworthy the effects on the ∼0.9 and ∼1.3 ppm lipid signals were not identical for the two diets; in hypercholesterolemic liver samples fed a rye bread-based diet, the amount of these lipid signals was simply higher compared with reference pigs (Figure 3b), whereas in hypercholesterolemic liver samples fed a wheat bread-based diet, the position of these lipid signals was also shifted, which was displayed as negative values in the low-ppm part and positive values in the high-ppm part of the individual ∼0.9 and ∼1.3 ppm signals in the first X-loading (Figure 3d). The effect was also evident in a direct comparison of hypercholesterolemic liver samples from a rye-based or a wheatbased diet (Figure 4b). This shift in the position of the methyl and methylene fatty acyl chains in the 1H NMR spectrum reveals differences in the amount of fatty acyl chains bound in HDL, LDL, and VLDL lipoproteins fractions, respectively, as has been documented by LC-NMR studies.36 Accordingly, the present data reveal that the lipoprotein profile is identical in normal liver and in hypercholesterolemic liver upon a rye-based diet, whereas the lipoprotein profile is changed toward a higher level of LDL and a lower level of HDL in hypercholesterolemic liver upon a wheatbased diet as compared with a normal liver. This finding suggests that a possible decrease in the HDL/LDL fraction in the liver associated with increased cholesterol levels can be attenuated by a diet rich in rye fiber but not by a diet rich in wheat fiber. This is in line with an observed 39% reduction of plasma total and LDL (34) Lubojacka, V.; Pechova, A.; Dvorak, R.; Drastich, P.; Kummer, V.; Poul, J. Acta Vet. BRNO 2005, 74, 217224. (35) Nakayama, M.; Motoki, T.; Kuwahata, T.; Kawaguchi, Y.; Kohri, H.; Tomita, Y.; Ondera, R. Nutr. Res. 2000, 20, 1771-1782. (36) Daykin, C. A.; Corcoran, O.; Hansen, S. H.; Bjørnsdottir, I.; Cornett, C.; Connor, S. C.; Lindon, J. C.; Nicholson, J. K. Anal. Chem. 2001, 73, 10841090.

174

Analytical Chemistry, Vol. 79, No. 1, January 1, 2007

Scheme 1. Illustration of the Action of Different Phospholipases in the Hydrolysis of Phosphatidylcholine

cholesterol of hypercholesterolemic pigs fed rye compared to pigs fed wheat in spite of identical dietary fiber, fat, and cholesterol intake.37 Hence, high rye fiber, diet-induced changes in plasma LDL/HDL fractions11,12,38 can at least partly be ascribed to effects on liver cholesterol metabolism. Comparison of liver metabolite profiles from pigs fed a rye bread-based and wheat bread-based diet also revealed pronounced diet-induced effects on the lipid profile; both saturated (∼0.9, 1.3, and 2.2 ppm) and unsaturated lipid signals (∼2.0 and 2.7 ppm) were increased in liver samples from the rye-based diet compared with the wheat-based diet (Figures 4 and 5). A study on the hypocholesterolemic effects of soy and fish protein in rats reported that intake of these proteins was associated with hepatic decreases in 14:0 and 16:0 fatty acids and a hepatic increase in 18:0.39 Possibly the present findings represent an equivalent change in hepatic lipid profile with a shift toward a higher proportion of longer fatty acyl chains, which will result in an increase in the intensity of the resonances from protons in lipids. The choline signals at ∼3.25 ppm, including choline, phosphocholine, and glycerophosphocholine, also differed between the two diets (Figure 5b). This is of interest as these compounds are essential nutrients and function as substrates in many pathways. Hepatic increases in glycerophosphocholine and decreases in phosphocholine have been reported after fish oil consumption, which is consistent with a shift from the phospholipase C to the phospholipase A1/A2 pathway of phosphatidylcholine breakdown (Scheme 1).40 Most probably a similar effect on phospholipid metabolism explains the differences in the glycerophosphocholine and phosphocholine signals between the two diets in the present study, possibly promoted by the diet-induced effect on the hepatic lipid profile. In conclusion, the present study for the first time showed differences in the metabolite profile of liver from normal and hypercholesterolemic pigs and between hypercholesterolemic pigs fed either a high-fiber, rye bread-based or a high-fiber, wheat bread-based diet by use of HR-MAS 1H NMR on intact liver. The metabolic differences consisted of changes in the lipid profile and the characteristics of the choline-containing compounds. Further studies, which potentially could involve 31P NMR that enables an (37) Lærke, H. N.; Bach Knudsen, K. E.; Pedersen, C.; Mortensen M.; Theil, P. K.; Larsen, T.; Penalvo, J. Scand. J. Food Nutr. 2006, 50 (Suppl. 1), 29. (38) Lund, E. K.; Salf, K. L.; Johnson, I. T. J. Nutr. 1993, 123, 1834-1843. (39) Wergedahl, H.; Liaset, B.; Guldbrandsen, O. A.; Lied, E.; Espe, M.; Muna, Z.; Mørk, S.; Berge, R. K. J. Nutr. 2004, 134, 1320-1327. (40) Dagnelie, P. C.; Bell, J. D.; Cox, I. J.; Menon, D. K.; Sargentoni, J.; Coutts, G. A.; Williams, S. C. R. NMR Biomed. 1993, 6, 157-162.

exact phospholipid analysis,41 are needed to obtain a better understanding of the basal mechanisms behind the proposed effects on lipid and cholesterol metabolism. ACKNOWLEDGMENT The Danish Technology and Production Research Council (FTP) is thanked for financial support through the project “NMRbased metabonomics on tissues and biofluids” (project 274-05(41) Schiller, J.; Arnold, K. Med. Sci. Monit. 2002, 8, MT205-MT222.

339), and the Nordic Joint Committee for Agricultural Research is thanked for financial support through the project ‘Rye Bran for Health’ (NKJ-121). I.F.D. thanks the Foundation for Science and technology, Portugal, for funding support through the grant SFRH/BPD/11516/2002 within the III Community framework.

Received for review July 20, 2006. Accepted September 27, 2006. AC061322+

Analytical Chemistry, Vol. 79, No. 1, January 1, 2007

175