Liver fatty acid (FA)-binding protein (L-Fabp), a cytoplasmic protein expressed in liver and small intestine, regulates FA trafficking in vitro and plays an important role in diet-induced obesity. We observed that L-Fabp−/− mice are protected against Western diet-induced obesity and hepatic steatosis. These findings are in conflict, however, with another report of exaggerated obesity and increased hepatic steatosis in female L-Fabp−/− mice fed a cholesterol-supplemented diet. To resolve this apparent paradox, we fed female L-Fabp−/− mice two different cholesterol-supplemented low-fat diets and discovered (on both diets) lower body weight in L-Fabp−/− mice than in congenic wild-type C57BL/6J controls and similar or reduced hepatic triglyceride content. We extended these comparisons to mice fed low-cholesterol, high-fat diets. Female L-Fabp−/− mice fed a high-saturated fat (SF) diet were dramatically protected against obesity and hepatic steatosis, whereas weight gain and hepatic lipid content were indistinguishable between mice fed a high-polyunsaturated FA (PUFA) diet and control mice. These findings demonstrate that L-Fabp functions as a metabolic sensor with a distinct hierarchy of FA sensitivity. We further conclude that cholesterol supplementation does not induce an obesity phenotype in L-Fabp−/− mice, nor does it play a significant role in the protection against Western diet-induced obesity in this background.
- diet-induced obesity
- hepatic triglyceride
- cholesterol metabolism
- saturated fatty acid
- polyunsaturated fatty acid
the consumption of diets rich in fat and cholesterol has been linked to the dramatic increase in obesity in Western societies in the last several decades, although the contribution of individual genetic variation to obesity susceptibility is being increasingly recognized. There is intense interest in the pathways and mechanisms that influence the uptake and utilization of individual dietary components (for review see Ref. 24), both as potential drug targets and as sites of genetic polymorphisms. There is abundant evidence that alterations in fatty acid (FA) metabolism play a key role in many of the local and systemic manifestations of obesity, including insulin resistance, pancreatic beta cell injury, dyslipidemia, hepatic steatosis, and other features of the metabolic syndrome (2, 13, 22).
Our understanding of the metabolic pathways that modulate FA uptake, utilization, and selective metabolic compartmentalization has been advanced through findings that distinct families of genes exhibit intersecting roles in each of these steps. Recent work has focused attention on the intracellular lipid trafficking functions of cytoplasmic FA-binding proteins (FABPs), which represent a large family of ∼15-kDa proteins capable of binding hydrophobic lipid ligands at high specificity (for review see Ref. 3). Liver FABP (L-Fabp) is a member of the FABP multigene family and is a highly abundant protein expressed in mammalian enterocytes and hepatocytes (7, 14, 15). Studies using recombinant protein-binding assays and in vitro cell culture systems demonstrate that L-Fabp plays an important role in metabolic trafficking of a number of hydrophobic ligands, particularly polyunsaturated FA (PUFA) and saturated FA (SFA), as well as cholesterol and bile acids (4, 5, 15, 25–27). The physiological functions of L-Fabp have also been explored through loss-of-function studies in vivo (1, 18, 20). Targeted deletion of L-Fabp revealed that L-Fabp−/− mice were protected against hepatic steatosis accompanying a prolonged fast (20). More recent findings demonstrated that L-Fabp−/− mice fed a high-SF, high-cholesterol (0.125%) Western diet were protected against diet-induced obesity and hepatic steatosis (21). These findings, in conjunction with other studies demonstrating that L-Fabp deletion in hepatocytes disrupted net FA uptake and utilization (10, 11, 23), strongly suggest that L-Fabp plays an important physiological role in substrate compartmentalization and metabolic trafficking of FA.
Independent studies have demonstrated no overt growth delay or failure to gain weight in chow-fed L-Fabp−/− mice, suggesting that the protection against Western diet-induced obesity represents a distinctive program of diet-gene interactions (18, 20). This interpretation, however, was complicated by a striking obesity phenotype in female L-Fabp−/− mice fed a semisynthetic diet supplemented with cholesterol (16). To resolve this discrepancy and to address the importance of dietary cholesterol vs. dietary fat content and FA composition in modulating the obesity phenotype and hepatic steatosis, we fed female L-Fabp−/− mice low-fat diets containing supplemental cholesterol or high-fat diets containing SFA or PUFA as the predominant source of triglyceride (TG), but without supplemental cholesterol. The findings suggest a dramatic interaction between SFA feeding and L-Fabp−/− deletion. However, in contrast to previous findings (16), our results do not support a role for cholesterol supplementation in promoting obesity or hepatic steatosis in L-Fabp−/− mice.
MATERIALS AND METHODS
Biochemical assays for TG, cholesterol, free FA (FFA), glucose, and phospholipid (PL) levels in serum and tissue were performed using commercially available kits (Wako Chemicals, Richmond, VA).
L-Fabp−/− mice were generated previously in our laboratory (20). C57BL/6J congenic mice (backcrossed >10 generations to C57BL/6J animals) were used for all studies. Age-matched C57BL/6J mice (Jackson Laboratory) were used as controls. Female mice were used for all experiments. Mice were housed in a full-barrier facility with a 12:12-h light-dark cycle and maintained on standard chow [PicoLab Rodent Diet 20, 4.5% fat (0.8% saturated fat), 0.015% cholesterol] with free access to food and water unless otherwise noted. All animal protocols were approved by the Washington University Animal Studies Committee and conformed to criteria outlined in the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
All mice were weighed and bled before the start of any dietary intervention and weighed weekly thereafter. The 2% cholesterol chow diet (diet no. 904691, MP Biomedicals, Solon, OH) contained 12% calories from fat. The 1.25% cholesterol semisynthetic diet and the low-cholesterol control semisynthetic diets were the same formulation as the diet used by Martin et al. (16) [diet nos. D11243 (control) and D01091702 (cholesterol), Research Diets, New Brunswick, NJ] and contained 12% of calories from fat (3.9 kcal/g). The high-SF and high-PUFA diets (diet nos. 960242 and 960244, respectively, MP Biomedicals) contained 41% of calories from fat (4.4 kcal/g). Mice were started on the cholesterol-containing diets at ∼8 wk of age and maintained on the diet for 10–12 wk. Mice were started on the SF and PUFA diets at ∼12 wk of age and maintained on the diet for 20 and 36 wk, respectively.
Food consumption and fecal fat determinations were performed as described elsewhere (21) 8–12 wk after the mice were placed on the respective diets. Food intake was measured and feces were collected for ≥3 days for all animals. Fecal fat content was determined gravimetrically as described elsewhere (28). Briefly, 1 g of dried feces was solubilized overnight in 10 ml of water and extracted in 40 ml of chloroform-methanol (2:1). Five milliliters of the organic phase were aliquoted into a preweighed vial, dried under nitrogen, and reweighed to determine lipid mass. Lipid mass was normalized to food consumption and dietary fat content for determination of percent fat absorption. At the conclusion of dietary studies, the mice were killed after a 4-h fast. Body, liver, and gonadal fat pad weights were measured. Serum was collected, aliquoted, and frozen at −80°C or stored overnight at 4°C before fast pressure liquid chromatography (FPLC). Tissues were flash frozen in liquid nitrogen and stored at −80°C for later analysis.
Analysis of tissue lipid content.
Frozen tissue (100–200 mg) was homogenized in PBS (1.6 ml), and protein concentration was determined by DC Protein Assay (Bio-Rad, Hercules, CA). Extractions were performed as described elsewhere (20) using 5 ml of chloroform-methanol (2:1) and 0.5 ml of 0.1% sulfuric acid per 0.2–0.5 ml of homogenate. An aliquot of the organic phase was collected, dried under nitrogen, and resuspended in 2% Triton X-100. TG, FFA, cholesterol, and PL content were determined using commercially available kits and normalized to protein content of the homogenate. For analysis of the lipoprotein profile of mice on the distinct cholesterol diets, serum from three to four animals per genotype was pooled (total volume ∼150 μl) and fractionated on tandem Superose 6 columns using a Pharmacia FPLC instrument. Fractions were stored overnight at 4°C before determination of cholesterol content using a commercially available kit (Wako Chemicals).
Energy utilization studies.
O2 consumption, heat production, respiratory exchange ratio (RER, or respiratory quotient), and CO2 production rate were determined using an Oxymax indirect calorimeter (Columbus Instruments, Columbus, OH). Mice were housed individually in the Oxymax chamber and allowed to acclimate for several hours before the start of data collection. O2 and CO2 levels were sampled every 2 min for ∼20 h, and the data for each 30-min interval (15 samplings) were averaged. Mice were studied after ∼14 wk on the SF diet and after ∼27 wk on the PUFA diet.
FABP5 mRNA expression.
RNA was isolated from frozen liver tissue (4 animals per genotype per dietary modification) using TRIzol reagent (Invitrogen, Carlsbad, CA) as directed by the manufacturer. For Western diet samples, RNA was isolated from the livers of female animals fed Western diet for ∼18 wk, as described previously (21). Total RNA (10 μg) was treated with DNase I (DNA-Free, Ambion) and then with SuperScript II reverse transcriptase (Invitrogen) to prepare cDNA. Real-time quantitative PCR was performed on an SDS 7000 (Applied Biosystems) using 2× SYBR Green Master Mix (Applied Biosystems). Expression levels were normalized to 18S, and relative gene expression was determined using the comparative threshold cycle method (Applied Biosystems User Bulletin 1). Real-time PCR primer sequences are as follows: 5′-CGACAGCTGATGGCAGAAAA-3′ and 5′-CCCATTGCTGGTGCTGG-3′ (REF) for Maxwell, 5′-GGAAGGAGAGCACGATAACAAGA-3′ and 5′-GGTGGCATTGTTCATGACACA-3′ (REF) for Hoekstra, and 5′-GAAGATGATCGTGGAGTGTGTCA-3′ and 5′-GCCCTCATTGCACCTTCTCA-3′ (identified using Primer Express, Applied Biosystems) for Fapb5.
Statistical significance was determined using an unpaired, two-tailed Student's t-test performed by Microsoft Excel. Unless otherwise noted, values are means ± SE.
L-Fabp−/− mice exhibit reduced weight gain after dietary cholesterol supplementation.
Body weight of female L-Fabp−/− mice fed a standard chow diet supplemented with 2% cholesterol for up to 12 wk was significantly lower that of C57BL/6 control mice (Fig. 1A, Table 1), and weight gain was reduced compared with C57BL/6 control mice (Fig. 1B). Further characterization of these animals revealed reduced hepatic TG content and comparable abundance of other lipid species, including FFA (Fig. 2A). These findings contrast with previous results in female L-Fabp−/− mice fed a cholesterol-supplemented diet, where significant increases in hepatic TG, cholesterol, and FFA were noted (16). FPLC of serum cholesterol distribution revealed a subtle but distinct shift in lipoprotein particle distribution in L-Fabp−/− mice to larger, HDL particles, without an overall change in serum lipid content (Fig. 2B, Table 1).
One possible source of the discrepancy between these findings and those reported previously by Martin and colleagues (16) was the composition of the base diet (i.e., standard rodent chow) to which the cholesterol supplement was added. To address this concern, we turned to a semisynthetic phytol- and phytoestrogen-free diet supplemented with 1.25% cholesterol formulated as previously described (16). However, our results were essentially the same as those outlined above: female L-Fabp−/− mice tended to weigh less and gained significantly less weight (Fig. 3), whereas hepatic (Fig. 4A) and serum lipid content (Table 2) was comparable between the two genotypes. FPLC analysis again revealed a subtle shift in cholesterol distribution toward larger HDL particles in L-Fabp−/− mice (Fig. 4B).
These findings collectively suggest that dietary cholesterol supplementation does not produce an obesity phenotype in female L-Fabp−/− mice, nor does it promote excess hepatic TG or cholesterol accumulation. To resolve whether cholesterol supplementation alters the obesity phenotype previously observed in Western diet-fed L-Fabp−/− mice, we conducted further studies using defined high-fat diets.
L-Fabp−/− mice are protected against obesity when fed a high-SF diet.
Weight gain was dramatically reduced in female L-Fabp−/− compared with C57BL/6J mice fed a high-SF diet for 20 wk (Fig. 5). Hepatic TG and PL contents in L-Fabp−/− mice were significantly lower, whereas cholesterol content was comparable to that in wild-type controls (Fig. 6). Female L-Fabp−/− mice also manifested reduced adiposity, as well as reduced serum TG and FFA levels (Table 3). These findings generally recapitulate the results obtained in female L-Fabp−/− mice fed a Western diet, although in those mice we observed increased hepatic cholesterol content in addition to reduced TG and PL (21).
Additional studies were performed to explore some of the potential mechanisms that might underlie the protection against high-fat diet-induced obesity and hepatic steatosis in L-Fabp−/− mice fed an SF diet. We confirmed no difference in absorption of dietary fat (98.67 ± 0.08 and 98.58 ± 0.09% in C57BL/6 and L-Fabp−/− mice, respectively, n = 4–5/genotype, P = 0.449) and no significant difference in food intake (2.73 ± 0.09 and 2.50 ± 0.09 g/day in C57BL/6 and L-Fabp−/− mice, respectively, n = 8–9/genotype, P = 0.093) in wild-type and L-Fabp−/− mice, although there was a trend toward decreased food consumption in the L-Fabp−/− mice. In addition, animals were subjected to indirect calorimetry to evaluate O2 consumption and CO2 production. From these values, heat production and RER were calculated and used as surrogate measures of energy expenditure. L-Fabp−/− mice display significantly reduced heat production during light and dark periods (Table 4), suggesting that the protection against obesity cannot be attributed to increased energy expenditure. RER values were not different between the genotypes, indicating that the reduced serum FFA levels in L-Fabp−/− mice (Table 3) do not result from increased FA oxidation. These data indicate that neither decreased dietary fat absorption nor increased energy expenditure in L-Fabp−/− mice accounts for the reduced weight gain in L-Fabp−/− mice fed the SF diet.
L-Fabp−/− mice fed a high-PUFA diet display characteristics indistinguishable from those of wild-type controls.
The findings with the SF diet prompted us to examine the effects of feeding a high-PUFA diet. In view of the known binding preference for L-Fabp toward unsaturated FA (4, 5, 15), we predicted that the protection against high-fat diet-induced obesity in L-Fabp−/− mice might be even more pronounced with high-PUFA feeding. However, this was not the case. PUFA diet-fed animals of both genotypes exhibited comparable weight gain (Fig. 7), with comparable liver and fat tissue weights (Table 5) and similar hepatic lipid content (Fig. 8). We observed no difference in dietary fat absorption (97.40 ± 0.22 and 96.92 ± 0.16% in C57BL/6 and L-Fabp−/− mice, respectively, n = 7–9) or food consumption (2.22 ± 0.23 and 2.09 ± 0.10 g/day in C57BL/6 and L-Fabp−/− mice, respectively, n = 7–9) between the wild-type and L-Fabp−/− mice fed the PUFA diet, consistent with no difference in weight gain.
Changes in hepatic Fabp5 expression: diet- and gene-specific effects.
Several recent studies have shown that expression of Fabp5 (epidermal Fabp or mal1) may be regulated by changes in dietary lipid content in mouse liver (8, 19). Maxwell and colleagues (19) noted a threefold downregulation of Fabp5 mRNA abundance after 1 wk of 0.5% cholesterol feeding. More recently, Hoekstra and colleagues (8) noted a striking upregulation of Fabp5 and Fabp5-related transcripts in livers of mice fed a Western diet for 2–6 wk. These studies suggest that members of the FABP multigene family may be cholesterol- and fatty acid-regulated targets. We therefore examined the Fabp5 mRNA expression in livers of wild-type and L-Fabp−/− mice fed chow or high-cholesterol (1.25%) semisynthetic diets. Fabp5 mRNA was significantly higher in chow-fed L-Fabp−/− mice than in chow-fed C57BL/6 controls (Fig. 9A), suggesting that there may be some adaptation in Fabp gene family expression in the liver after L-Fabp deletion. However, in contrast to previous findings (19), we observed no significant difference in Fabp5 expression between cholesterol- and chow-fed mice in either genotype, although the data were somewhat variable (Fig. 9, A and B). Similar findings were obtained using three distinct primer sets, including those used in the studies cited above, as well as a set designed independently in our laboratory (not shown). We also examined hepatic Fabp5 expression in mice fed the high-SF diet, the PUFA diet, and the Western diet. As shown in Fig. 9C, Fabp5 expression was markedly increased in the livers of wild-type and L-Fabp−/− mice fed the SF diet. Interestingly, Fabp5 expression was significantly increased in the livers of L-Fabp−/− mice fed a Western diet, but not in C57BL/6 mice, compared with chow-fed controls. These data suggest that expression of Fabp5 may be upregulated in mice fed a high-fat diet, although the magnitude of the induction may be dependent on the FA composition and length of feeding.
Considerable data, including the finding that L-Fabp colocalizes with peroxisome proliferator-activated receptor-α and may function in the selective delivery of ligands (including FAs and fatty acyl CoA) for interaction with nuclear receptors (9), support a role for L-Fabp in FA uptake, compartmentalization, and metabolism. These intracellular transport functions inferred from studies in cell culture have been extended to investigate the physiological function of L-Fabp in rat liver, where the data suggest that uptake of long-chain FA (particularly palmitate) generally parallels the abundance of L-Fabp (10, 11). These findings were complemented and extended by studies demonstrating reduced hepatic FA uptake and binding capacity in L-Fabp−/− mice (18, 20). Additional studies demonstrated that L-Fabp−/− mice are protected against obesity when fed a high-cholesterol, high-SF Western diet (21).
The findings in this study demonstrate a dramatic diet-gene interaction effect with regard to SFA feeding and L-Fabp in the C57BL/6J congenic background. Our findings strongly suggest that protection against high-fat diet-induced obesity is strikingly dependent on the fat composition. PUFA-diet-fed L-Fabp−/− mice were not protected against obesity, whereas SF diet-fed mice demonstrated the same protection observed previously in Western diet-fed mice (21). These findings are somewhat counterintuitive, since recombinant L-Fabp displays higher binding affinity for polyunsaturated than for saturated long-chain FAs (4, 5, 15), and we had anticipated that the effects on weight gain in PUFA diet-fed L-Fabp−/− mice would be comparable to, if not greater than, those observed in SF diet-fed mice. In considering the FA species specificity of this diet-gene interaction, it is worth noting that the Western diet studied previously contains 21% milk fat, including ∼60% SFA and ∼30% unsaturated FA, whereas the SF diet contains >80% SFA. This observation does not answer the following question: Does the protection against obesity and hepatic steatosis in L-Fabp−/− mice reflect simply the abundance of dietary SF alone or the relative proportions of SFA vs. PUFA or other FAs?
As we found in our studies with L-Fabp−/− mice fed a Western diet (21), it is not easy to explain the protection against diet-induced obesity and steatosis in L-Fabp−/− mice fed the high-SF diet. Western diet-fed L-Fabp−/− mice exhibited decreased weight gain, with no malabsorption of dietary fat, no increase in heat production, and no difference in food consumption. However, energy balance calculations showed that a subtle difference in food consumption and/or heat production, i.e., within the standard error of the observed values, could be sufficient to explain the difference in weight gain between the two genotypes.
The situation appears to be similar in the case of mice fed a high-SF diet. L-Fabp−/− mice gain an average of 0.26 g/wk less than C57BL/6 mice (0.53 ± 0.05 vs. 0.27 ± 0.03 g/wk, n = 8–9 per genotype, P = 0.033), which reflects an energy differential of ∼2.3 kcal/wk (0.26 g/wk × 9 kcal/g fat = 2.34 kcal/wk, with the assumption that the differential is due to differences in fat weight). We observed no difference in fecal fat content and decreased heat production in L-Fabp−/− mice (Table 4). However, there is a trend toward reduced food consumption (an average of 0.23 g/day less) in the L-Fabp−/− mice, although the difference is not significant. It is worth noting that a decrease in food consumption of as little as ∼0.07 g/day could explain the 2.3-kcal energy differential between the genotypes (2.3 kcal/wk ÷ 4.4 kcal/g SF diet ÷ 7 days/wk = ∼0.07 g/day). This value is within the standard error of the observed values and beyond the limits of detection with the current methodology. Thus a subtle difference in food consumption or feeding behavior may be sufficient to explain the protection against diet-induced obesity in SF diet-fed L-Fabp−/− mice. Although the precise mechanisms accounting for this divergent physiological response to high-fat feeding in L-Fabp−/− mice requires further study, our findings represent a direct example of diet-gene interaction and support a role for L-Fabp in the metabolic channeling of SFAs that contributes to the pathogenesis of hepatic steatosis and diet-induced obesity.
In regard to the effect of cholesterol feeding, our results are in direct contrast to other studies with L-Fabp−/− mice (16). These previous findings demonstrated exacerbation of weight gain and obesity in female L-Fabp−/− mice fed a 1.25% cholesterol diet for 5 wk, with a twofold increase in fat tissue mass, accompanied by increased hepatic TG and cholesterol content (16). The present results suggest, if anything, less weight gain and reduced hepatic TG content in female L-Fabp−/− mice fed a cholesterol-supplemented, low-fat diet. Moreover, our findings demonstrate a subtle but distinctive response to cholesterol feeding in female L-Fabp−/− mice, with a trend to less hepatic TG accumulation and a subtle, yet reproducible, shift in HDL size. The implications of this latter finding are unclear, particularly since there was no difference in total serum cholesterol or TG levels between the genotypes (Tables 1 and 2).
The origin of the altered response to cholesterol supplementation in terms of weight gain is not immediately apparent but, nevertheless, raises crucial implications for understanding the role of L-Fabp in modulating diet-gene interactions that might contribute to obesity and hepatic steatosis. It is conceivable that the genetic background of the two lines could at least partially account for this discrepancy, since the line of Martin and colleagues (18) was not fully crossed into the C57BL/6J background, whereas the mice used in our cholesterol dietary studies were fully congenic (≥10 generations). Differences in generation number or initial targeting strategy might also play a role. Other explanations, such as differences in sex or dietary regimens, seem less likely, since we attempted to replicate the conditions specified by Martin and colleagues in their studies. These were important considerations, since there was no apparent obesity phenotype or augmented hepatic TG content in male L-Fabp−/− mice fed a cholesterol-supplemented diet (17).
There is less information in regard to a physiological role for L-Fabp in modulating hepatic cholesterol metabolism, although earlier in vitro studies clearly established a role for L-Fabp in regulating cholesterol uptake and metabolism in mouse L cells (12) and in mobilization of sterol from the membrane bilayer (27). Evidence that L-Fabp plays a role in hepatobiliary cholesterol metabolism has been inferred from the striking upregulation of L-Fabp gene expression in the livers of chow-fed Scp-2−/− mice, coupled with an abrogation of the increase in biliary cholesterol secretion after lithogenic diet feeding (6). These findings were collectively interpreted as suggesting that the compensatory mechanisms that result in impaired biliary cholesterol secretion in Scp-2−/− mice may involve alterations in L-Fabp expression (6). More recently, studies in male L-Fabp−/− mice demonstrated alterations in hepatobiliary bile acid content after cholesterol feeding (17).
Taken together, our findings raise important and testable implications for cytoplasmic lipid-binding proteins in the integrated regulation of hepatic steatosis and obesity after fat and cholesterol feeding. Dietary cholesterol supplementation has been shown to regulate a number of genetic programs, including sterol response element-binding protein- and liver X receptor-dependent pathways, each of which activates a range of downstream targets with integrative roles in hepatic lipogenesis (8, 19). Further study is required to elucidate the pathways by which the divergent response to SFA vs. PUFA feeding is mediated in L-Fabp−/− mice, but our results suggest an important diet-gene interaction with respect to FA content. Consistent with this is the marked difference in hepatic Fabp5 expression between SF diet-fed mice and PUFA diet-fed mice (Fig. 9). Moreover, our findings indicate that the protection against Western diet-induced obesity in L-Fabp−/− mice is due primarily to altered metabolism of SFA. On the other hand, we conclude that cholesterol supplementation does not produce an obesity phenotype in L-Fabp−/− mice, nor does it enhance hepatic TG accumulation.
This work was supported by National Institutes of Health Grants to N. O. Davidson, including Grants HL-38180 and DK-56260 and Digestive Disease Research Core Grant DK-52574. B. T. Sternard was supported by the Biomedical Research Apprenticeship Program at Washington University. The authors also acknowledge the support of Diabetes Research and Training Center Grant P60 DK-020579-30.
The authors acknowledge the help of Trey Coleman and the Clinical Nutrition Research Unit for assistance with the Oxymax studies (National Institute of Diabetes and Digestive and Kidney Diseases Grant P30 DK-56341).
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.
- Copyright © 2008 the American Physiological Society