Dietary ganglioside decreases cholesterol content, caveolin expression and inflammatory mediators in rat intestinal microdomains

Abstract

Membrane microdomains rich in cholesterol and sphingolipids, including gangliosides (GGs), are known to be important regions for cell signaling and binding sites for various pathogens. Cholesterol depletion inhibits the cellular entry of pathogens and also reduces inflammatory signals by disrupting microdomain structure. Our previous study showed that dietary gangliosides increased total ganglioside incorporation while decreasing cholesterol in the intestinal mucosa. We hypothesized that diet-induced reduction in cholesterol content in the intestinal mucosa disrupts microdomain structure resulting in reduced pro-inflammatory signals. Male weanling Sprague-Dawley rats were fed semipurified diets for 2 weeks. Experimental diets were formulated to include either ganglioside-enriched lipid (GG diet, 0.02% gangliosides [w/w of diet] ) or polyunsaturated fatty acid (PUFA diet, 1% arachidonic acid and 0.5% docosahexaenoic acid, w/w of total fat), in a control diet containing 20% fat. Levels of cholesterol, GG, caveolin, platelet activating factor (PAF), and diglyceride (DG) were measured in the microdomain isolated from the intestinal brush border. The GG diet increased total gangliosides by 50% with a relative increase in GD3 and a relative decrease in GM3. Cholesterol content was also reduced by 23% in the intestinal microdomain. These changes resulted in a significant decrease in the ratio of cholesterol to ganglioside. The GG diet and the PUFA diet were both associated with reduction in caveolin, PAF, and DG content in microdomains, whereas no change occurred in the ganglioside profile of animals fed the PUFA diet. Dietary gangliosides decrease the cholesterol/ganglioside ratio, caveolin, PAF and DG content in microdomains thus exerting a potential anti-inflammatory effect during gut development.

Introduction

Microdomains, generally called lipid rafts, caveolae, or glycosphingolipid-signaling domains, are important regions for signal transduction and lipid and protein trafficking (Brown, 1998; Hakomori and Handa, 2000). Microdomains have been recently recognized as one of the sites for cellular entry of bacterial and viral pathogens (Bavari et al., 2002; Wolf et al., 2002). For example, the entry of filoviruses occurs at sites of lipid rafts (Bavari et al., 2002). Cholera toxin enters the cell by endocytosis and requires GM1 for retrograde trafficking into host cells via association with lipid rafts (Wolf et al., 2002).

Physiological and functional roles of microdomains are dependent on cholesterol and sphingolipids, including gangliosides. Reduction of cholesterol inhibits pathogen entry by disrupting the structure of microdomains (Samuel et al., 2001; Wolf et al., 2002) and impairs inflammatory signaling (Wolf et al., 2002). Cholesterol up-regulates the expression of caveolin, a protein marker of caveolae (Hailstones et al., 1998). Sphingolipid depletion inhibits the intracellular trafficking of GPI-anchored proteins (Kasahara and Sanai, 1999), suggesting that lipid–protein interaction directly modulates gene expression and cellular trafficking important for cell development and behavior.

The neonatal intestine has endocytic and enzymatic transport systems for absorption of nutrients and immunoglobulins (Moxey and Trier, 1979; Wilson et al., 1991) but is susceptible to pathogen entry because of higher permeability than adult intestine (Koldovsky, 1994). Gangliosides from mothers’ milk may act as receptors for viral and bacterial toxins to protect against entry of pathogens into the neonatal enterocyte (Rueda et al., 1998). During development, membrane permeability gradually decreases (Koldovsky, 1994), whereas peptidases and glycosidases become functionally active and enriched in microdomains (Danielsen and van Deurs, 1995). Many digestive and absorptive enzymes, such as alkaline phosphatase, aminopeptidase N and A, and sucrase-isomaltase are also increased in apical membrane microdomains (Danielsen, 1995). A recent study demonstrates that the annexin-2 and caveolin-1 complex in microdomains regulates intestinal cholesterol transport (Smart et al., 2004). These results suggest the importance of microdomains in intestinal apical membranes for nutrient uptake and metabolism.

Polyunsaturated fatty acids (arachidonic acid or docosahexaenoic acid; PUFA) can accumulate in microdomains and displace functional proteins by changing the lipid composition of the microdomain (Stulnig et al., 2001). This observation highlights the importance of dietary lipids in modulating physiological and biological properties of lipids and proteins in the microdomain. Whether dietary gangliosides affect the lipid profile and protein components of microdomains during gut development is not known.

Our previous study (Park et al., 2005) showed that dietary ganglioside significantly increased total amount of gangliosides and decreased cholesterol content in the intestinal mucosa. Cholesterol depletion inhibits inflammatory signaling by disrupting microdomain structure (Samuel et al., 2001; Wolf et al., 2002). Thus, it was hypothesized that diet-induced cholesterol reduction in the microdomain, would disrupt microdomain structure and reduce pro-inflammatory mediators such as diglyceride (DG) and platelet activating factor (PAF). DG derived from phospholipids by phospholipase C (PLC) binds to protein kinase C (PKC) to phosphorylate targeted proteins, such as the epidermal growth factor receptor (Smart et al., 1995; Sciorra and Morris, 1999). PAF, 1-O-alkyl-2-acetyl-sn-glycero-3-phosphorylcholine, stimulates inflammatory cells such as leukocytes (Prescott et al., 1990) and activates phospholipase A2 (PLA2) in the intestinal tissue to release arachidonic acid (Okayasu et al., 1987). Increased arachidonic acid and derivatives of the lipoxygenase pathway such as LTB4 and 5-HETE, simultaneously stimulate PAF synthesis by activation of acetyl-CoA acetyltransferase (Peplow and Mikhailidis, 1990). PAF binds its receptor to increase intracellular calcium and inositol triphosphate (IP3) production and PKC activation for inflammation (Flickinger and Olson, 1999). It is unknown whether PAF also localizes in the microdomain. Because several studies reported that sphingomyelin (SM) has an inhibitory effect on PLA2 activity (Koumanov et al., 1997) and that PLA2 localizes in microdomains (Murakami et al., 1999), it is of interest to determine whether dietary ganglioside also decreases PAF synthesis either by increasing sphingolipids or by disrupting microdomain structures in the developing intestine. We also examine whether dietary ganglioside reduces DG content in the microdomain because sphingolipids inhibit PKC and PLC (Hannun and Bell, 1989; Katoh, 1995; Daniele et al., 1996), and these two enzymes exist in microdomains (Liu et al., 1997).

Dietary SM and cerebrosides are metabolized by sphingomyelinase and ceramidase through the stomach, small intestine, and colon in rats and mice (Schmelz et al., 1994; Nyberg et al., 1997). The process of digestion and absorption seems to be slow and poor in the intestine (Schmelz et al., 1994). SM or cerebrosides fed to animals are found in feces as metabolites and as the intact form (25–43%) (Nyberg et al., 1997). Hydrolyzed sphingolipids are transported to the circulation (Imaizumi et al., 1992). Although there is not direct information on digestion and absorption of gangliosides intact, several studies indicate that dietary gangliosides are hydrolyzed by sialidase in the gastrointestinal tract from birth as the enzyme exists in saliva, gastric, intestinal secretion and mucosa as well as in human milk (Schauer et al., 1976; Tram et al., 1997).

Neonates consume sphingolipids, including gangliosides, from mothers’ milk (Berger et al., 2000). Gangliosides are known to act as receptors for viruses and toxins (Laegreid and Kolsto Otnaess, 1987; Rolsma et al., 1998), activators for T-cells (Ortaldo et al., 1996) and stimulators for Th-1 and Th-2 cytokine-secreting lymphocytes in neonates (Vazquez et al., 2001). Gangliosides are also one of the major lipid components in microdomains. It is not known whether dietary ganglioside changes the lipid profile and structure of the intestinal microdomain and modulates inflammatory signaling mechanisms in the developing intestine. Arachidonic acid and docosahexaenoic acid are added into infant formulas as important lipid components for early development. It is not known how PUFA alters the lipid profile and caveolin content of intestinal microdomains. Thus, the PUFA diet was chosen as a second control to compare with the control and the GG diet. The objective of this study was to determine the effects of dietary ganglioside and PUFA on microdomain structure and pro-inflammatory mediators, DG and PAF, in the developing gut.

Result

Animal growth and tissues

The initial and final body weights of animals after 2 weeks feeding of experimental diets were not significantly different among control, GG, and PUFA groups. Individual food intake was similar between experimental groups. Intestinal mucosal weight and intestinal length were not affected by dietary treatment (data not shown).

Ganglioside content and composition

Animals fed the GG diet had up to a 50% increase in total ganglioside content in intestinal microdomains (Figure 1). Relative content of GM3 in microdomains decreased significantly from 83.7 to 77.7%, whereas GD3 increased from 4.4 to 8.3% (Table I). These changes in ganglioside profile were accompanied by a significant increase in total GD3 content (Figure 2), but the total GM3 content remained the same (data not shown). This finding agrees with our previous study (Park et al., 2005), in which it was found that dietary ganglioside significantly increased total ganglioside content in developing rat intestinal mucosa resulting in compositional changes such as decreased GM3 and increased GD3. The content of GM1, GD1a, GD1b, and GT1b was not changed by either experimental diet (Table I). The PUFA diet had no effect on total ganglioside content or relative composition.

Sphingomyelin content

Animals fed the GG diet showed a 57% increase in SM content in microdomains (Figure 1), whereas the PUFA diet had no effect on SM.

Cholesterol content of microdomains

Both the GG diet and the PUFA diet resulted in a significant lowering of microdomain cholesterol levels, but the effect of the P UFA diet was slightly less pronounced (Figure 1).

Ratios of cholesterol to gangliosides and cholesterol to sphingomyelin

Animals fed the GG diet showed a 40% reduction in the ratio of cholesterol to total gangliosides, from 31.3 to 18.9 (Figure 3), whereas the PUFA diet had no effect on this relationship. Feeding the GG diet also resulted in a 56% reduction in the ratio of cholesterol to SM. The ratios were 143, 117, and 63 for control, PUFA, and GG diets, respectively (Figure 3).

Caveolin content in microdomains

Both experimental diets were associated with significantly lower expression of caveolin protein (70 and 60% reduction for the PUFA diet and GG diet, respectively, compared to animals fed the control diet) in intestinal microdomains (Figure 4). The caveolin-1-alpha (upper) and caveolin-1-beta (lower) isoforms were found to be present in intestinal microdomains.

Content of pro-inflammatory mediators (diglyceride and platelet activating factor)

After feeding the GG diet, total DG and 1,2-DG were reduced by 43 and 44%, respectively (Table II). The PUFA diet also caused a reduction in these two lipids but to a lesser extent (32 and 33% for total DG and 1,2-DG, respectively). The GG and PUFA diets decreased PAF content by 65 and 50%, respectively (Figure 2) compared to the animals fed the control diet.

Discussion

Diet-induced change in the microdomain ganglioside content is significant because GD3 and GM3 are involved in a variety of cellular functions. GM3 is co localized with signaling molecules such as c-Src, Rho, and Fak in microdomains (Iwabuchi et al., 2000). Thus, reduction in GM3 by the GG diet may alter signaling pathways related to these molecules. The elevation of GD3 by the GG diet may enhance immune function and gut protection during development because GD3 activates T-cells (Ortaldo et al., 1996) and has an anticarcinogenic effect in the mouse colon (Schmelz et al., 2000). The present results showing accumulation of dietary gangliosides in microdomains are supported by a previous study demonstrating that administration of [3H] GM3 to Neuro 2a cells showed enrichment of [3H]GM3 in microdomains (Prinetti et al., 1999). These observations suggest that an exogenous supplement of ganglioside is directly incorporated into the microdomain.

Cholesterol is an important lipid involved in compartmentalizing lipids and proteins into microdomains (Incardona and Eaton, 2000). Recent studies found that viral pathogens and cholera toxin appear to reach the ER by caveolae (Lencer et al., 1995; Majoul et al., 1996). Cholesterol reduction in cell membranes inhibits the invasion of HIV-1 cholera toxin (Wolf et al., 2002) and malarial parasite (Samuel et al., 2001) by disrupting microdomain structure. Cholesterol depletion by drugs has been shown to down-regulate caveolin gene expression (Hailstones et al., 1998). A similar effect is reported here in that cholesterol depletion by diet modification created a reduction in caveolin protein expression. Decreased caveolin expression may induce a dissociation of the complex of annexin-2 and caveolin-1 necessary for cholesterol trafficking in the intestine (Smart et al., 1995). Thus, by reducing microdomain cholesterol content leading to lower caveolin expression and disruption of potential invasion sites, a possible anti-infective effect of dietary ganglioside is demonstrated.

Our study provides new evidence to indicate that dietary ganglioside and PUFA decrease pro-inflammatory mediators PAF and DG in the intestinal microdomain of developing animals. The ganglioside-enriched diet is more effective at reducing these factors than the PUFA-enriched diet. It is known that GM3 stimulates PLA2 activity, leading to arachidonic acid release in human peripheral blood lymphocytes (Garofalo et al., 1998). Thus, the decrease in GM3 observed in the microdomain may be linked to reduction in PLA2 activity which in turn leads to decreased PAF content (Table I and Figure 2B). Because it is known that PAF binds to a PAF receptor in cell membranes to initiate inflammatory signaling events (Flickinger and Olson, 1999), this study suggests that PAF receptors localize in the intestinal microdomain. In this study, we presented the PAF content as a percentage of control instead ng/mg protein. Recent studies have demonstrated that GD3 has a stronger inhibitory effect against PKC activation than GM3 (Katoh, 1995) and that gangliosides in general appear to be inhibitory to PLC activity (Daniele et al., 1996). The increase in total gangliosides and GD3 observed in association with the GG diet may be related to the reduction in DG content through inhibition of PKC or PLC. Microdomain DG might be a critical lipid component for modulating the structure and function of microdomains through activation of PKC (Hannun and Bell, 1989), because PKC regulates a cyclic transition of microdomains from the membrane to a vesicle that then returns to the cell surface (Smart et al., 1995).

The potential for dietary supplementation of physiologic doses of dietary ganglioside to exert anti-inflammatory effects in the gut of developing animals has been described. These results suggest that dietary ganglioside is bioavailable in the gut, and thus potential for protective roles in neonatal intestine should be investigated.

Materials and methods

Animals and diets

This study was approved by the University of Alberta Animal Ethics Committee. Male Sprague-Dawley rats (18-day-old, n = 8 per diet), with average body weights of 41.6 ± 1.6 g, were randomly separated into 3 groups of 8 with the rats separated into groups of 3, 3 and 2 and housed in each polypropylene cage. Animals were maintained at a constant temperature of 23°C and a 12 h light/dark cycle. Animals had free access to water and one of three semipurified diets containing 20% (w/w) fat for 2 weeks. The composition of the basal diets fed has been previously reported (Table III) (Clandinin and Yamashiro, 1980). Animal body weight and food intake were recorded every other day throughout the experiment. The control diet fat was a blend of triglyceride, which reflected the fat composition of an existing infant formula. Fatty acids of the control diet (Jumpsen et al., 1997) were composed of about 31% saturated fatty acids, 48% monosaturated fatty acids, and 21% PUFA providing a ratio of 18:2n-6 to 18:3n-3 of 7 to 1. The polyunsaturated fatty acid diet (PUFA diet) was formulated by adding arachidonic acid (20:4n-6, 1% [w/w] of total fat) and docosahexaenoic acid (22:6n-3, 0.5% [w/w] of total fat) (Martek Biosciences, Columbia, MD) to the fat blend of the control diet. The ganglioside diet (GG diet) was formulated by adding ganglioside-enriched lipid (dairy gangliosides, Fontera, Cambridge, New Zealand) to the control diet at a level of 0.02% (w/w) gangliosides in the total diet. Ganglioside-enriched lipid consisted of about 45–50% (w/w) phospholipids and 15–20% (w/w) gangliosides. The cholesterol content of total lipid in the ganglioside fraction was negligible (<0.002% [w/w] of total fat). The ganglioside fraction contained approximately 80% (w/w) GD3, 9% GD1b, 5% GM3, and 6% other minor gangliosides (GM2, GM1, and GT1b). The ganglioside-enriched lipid also contained 60–70% lactose and 10–12% minerals, and the level of these amounts was adjusted in the basal diet (Table III). Over all, the fatty acid composition of diets was not different between the control and the GG diet. For diet formulation, ganglioside-enriched lipid was mixed first with the triglyceride (20% w/w) to be added to the semipurified diet and then mixed into the basal diet (80% w/w). Triglyceride and basal diet is mixed well by a diet mixer (Hobart, Australia). All mixed diets were kept in airtight containers and stored in –25°C during the experimental period.

Collection of samples

After anaesthetizing animals with halothane, the small intestine (jejunum to ileum) was excised. The intestine was washed with ice-cold 0.9% saline solution to remove visible mucus and dietary debris, opened and moisture was carefully removed with a paper towel to measure mucosa weight. Intestinal mucosa was scraped off with a glass slide on an ice-cold glass plate. All mucosa samples were kept at –70°C until extraction.

Sucrose gradient separation of microdomains

Intestinal microdomains were prepared by ultra-centrifugation of a discontinuous sucrose gradient (Igarashi and Michel, 2000). Intestinal mucosa (about 1.5g) was suspended with 4 mL TME (10 mM Tris–HCl, 1 mM MgCl₂, 1 mM EGTA) solution containing 1 mM phenylmethyl sulfonyl fluoride, 0.001% (w/v) apotinin, and 2% (v/v) Triton X-100 for 30 min in ice and homogenized with 15 strokes of a Dounce homogenizer with a tight-fitting pestle (Wheaton Scientific, Millville, NJ). The homogenate was adjusted to 45% (w/v) sucrose by adding an equal volume of 90% (w/v) sucrose and then homogenized again with 5 strokes of the Dounce homogeniser. A 6 mL 5% and 14 mL 35% discontinuous sucrose gradient was overlaid on the homogenate in 45% (w/v) sucrose, which left a 45-35-5% sucrose gradient from the bottom. After 16 h centrifugation at 70,000 × g at 4°C in a Beckman SW 28 Ti rotor, the interface between 5 and 35% sucrose was collected as the microdomain fraction. Microdomains were washed with TME solution (1% Triton X-100) and centrifuged twice at 100,000 × g for 1 h at 4°C to remove sucrose. The pellet was re-washed with phosphate buffer solution (PBS) to remove Triton X-100 and re-suspended in PBS for protein and lipid analysis. Enrichment of microdomains was confirmed by analysing the amounts of caveolin, cholesterol, and gangliosides in the intestinal microdomain compared to the protein pellet, which was soluble in detergent solution. Immunoblotting and thin-layer chromatography (TLC) development was used to confirm microdomain purification (Figure 5). About 25 µg protein or equal amount of lipids per mg protein from the microdomain fraction and the detergent soluble protein pellets were loaded on 15% SDS–PAGE for caveolin detection and on TLC plate for gangliosides and cholesterol, respectively. Two samples of each microdomains (lanes 1 and 2) or detergent soluble proteins (lanes 3 and 4) are shown in the Figure 5. The content of caveolin-1 and cholesterol was 4–5 times higher in the intestinal microdomain fraction than the detergent soluble protein pellets (Figure 5A and C). Ganglioside GM3, which is the major ganglioside in the rat intestine, constituted about 80–90% of total gangliosides and was exclusively found in the microdomain, but not in the detergent soluble protein pellet (Figure 5B).

Western blotting for caveolin content

Protein content from microdomains was measured by QuantiPro BCA Assay Kit (Sigma-Aldrich, Oakville, ON). Approximately, 25 µg protein was dissolved with SDS reducing sample buffer and loaded onto 15% SDS–PAGE minigels. After transferring proteins onto nitrocellulose (Amersham Pharmarcia Biotech, Piscataway, NJ), membranes were blocked with 5% nonfat dried milk in TBS-T (20 mM Tris; pH 7.6; 137 mM NaCl; 0.1% Tween 20) for 1 h at room temperature. The primary antibody which specifically recognizes caveolin (anti-caveloin-1 antibody, Mouse IgG, BD Bioscience, Ontario, Canada), was diluted 1:1000 in TBS-T with 1% nonfat dried milk and incubated for 90 min at room temperature. The membrane was washed three times for 10 min each time in TBS-T. The secondary antibody (goat anti-mouse IgG-HRP conjugate; Bio-Rad, CA) was diluted 1:2000 in 1% nonfat dried milk in TBS-T and incubated for 1 h at room temperature. After washing the membrane with TBS-T three times for 10 min each time, the caveolin protein was developed by enhanced chemiluminescence (ECL) detection reagent according to the protocol supplied by Amersham Pharmarcia Biotech, UK. Relative blot intensity of caveolin was measured from five animals from each diet group using an Imaging Densitometer (Bio-Rad, Hercules, CA).

Ganglioside extraction and purification

Total lipid was extracted from the microdomain fraction using the Folch method (Folch et al., 1957). For extraction of gangliosides (Svennerholm, 1964), the upper phase was collected into a test tube and the lower organic phase was washed twice with the Folch upper phase solution (chloroform:methanol:water, 3:48:47 by vol.). The upper phase gangliosides were pooled and purified by passage through Sep–Pak C₁₈ cartridges (Waters Corporation, Milford, MA,) preconditioned with 10 mL of methanol, 20 mL of chloroform:methanol (2:1, v/v), and 10 mL of methanol (Williams and McCluer, 1980). Cartridges were washed with 20 mL of distilled water. Gangliosides were eluted with 5 mL of methanol and 20 mL of chloroform:methanol (2:1, v/v), dried under N₂ gas and then redissolved in 500 µL of chloroform:methanol (2:1, v/v). Gangliosides were stored at –70°C until analysis.

Analysis of total and individual ganglioside content

Total ganglioside content was measured as N-acetyl neuraminic acid (NANA) as described by Suzuki (Suzuki, 1964). An aliquot of purified ganglioside sample was dried under N₂ gas and dissolved with 0.5 mL distilled H₂O and 0.5 mL resorcinol-HCl in screw-capped Teflon-lined tubes (Svennerholm, 1957). The purple-blue color developed by heating was extracted into butylacetate:butanol (85:15, v/v). Optical density was read at 580 nm.

Individual gangliosides were identified by silica gel high performance thin layer chromatography (HPTLC; Whatman, Clifton, NJ) using ganglioside standards GM3, GM2, and GD3 and bovine brain ganglioside mixture (Alexis, San Diego, CA) in a solvent system of chloroform:methanol:0.2% (w/v) CaCl₂·2H₂O (55:45:10, by vol.). Individual gangliosides were recovered and measured as described above. The percentage of ganglioside was calculated by NANA content in individual ganglioside.

Cholesterol assay

Cholesterol analysis was completed by measuring cholesterol ester and free cholesterol together using a test kit (Sigma, St. Louis, MO).

Analysis of sphingomyelin and platelet activating factor

To determine SM content, total lipid extracted from microdomains was applied onto a silica gel ‘H’ TLC plate and separated using chloroform:methanol:2-propanol:0.2% KOH: triethylamine (45:13.5:37.5:9:27, by vol). PAF was separated from the total lipid fraction on a silica gel ‘G’ TLC plate (Fisher Scientific, CA) using chloroform:methanol:water, (65:35:6, by vol) for PAF. Commercial standards of SM, PAF, and lyso-PC (Sigma, MO) were spotted onto the plate for identification. TLC plates were visualized with 0.1% ANSA (anilino naphthalene sulfonic acid) under UV exposure. Lipids were recovered and lipid phosphate was measured according to the method of Itoh et al. (1986). PAF content was presented as a percentage of control.

Analysis of diglyceride

To measure DG content, total lipid was separated on a silica gel ‘G’ TLC plate using petroleum ether:diethyl ether:acetic acid (80:20:1, by vol). 1,2-DG and 1,3-DG were exposed to 0.1% ANSA and identified under UV light by comparison with commercial standards. Cholesterol co migrated with 1,3-DG, so these were recovered in a single fraction. 1,2-DG and 1,3-DG was methylated with a known amount of heptadecanoic acid (C17:0) as an internal standard. To remove cholesterol from 1,3-DG after methylation, fatty acid methyl esters (FAME) were applied onto a silica gel ‘G’ plate and developed with toluene. The purified FAME fraction was collected, extracted with hexane and injected into a gas liquid chromatograph (GLC, Varian Model 3400 CX, Varian Canada Inc., Mississauga, ON) to measure total fatty acid content in DG. The GLC was equipped with a flame ionization detector and a 25 m BP-20 fused capillary column (SGE, Ringwood, Australia).

Statistical analysis

Values shown are means ± standard deviation (SD). Significant difference between the control group and experimental groups was determined by one-way analysis of variance (ANOVA) with SAS. Significant effects of diet treatment were determined by a Duncan multiple range test at a significance level of P < 0.05.

Acknowledgments

We thank Dr. Goh for technical support. This research was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and Wyeth Nutrition.

References

Author notes

³Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton, Alberta, Canada T6G 2P5; ⁴Department of Medicine, University of Alberta, Edmonton, Alberta, Canada T6G 2P5; ⁵Wyeth Nutrition, Collegeville, PA; and ⁶Alberta Institute for Human Nutrition, University of Alberta, Edmonton, Alberta, Canada T6G 2P5