In both normal and hyperlipidemic individuals, serum concentrations of lipids are reduced when polyunsaturated (POLY)⁵ fat diets are eaten; saturated (SAT) fat diets cause the inverse to occur (Mattson and Grundy 1985). When compared with SAT fat, monounsaturated (MONO) fat consumption results in lower serum total and LDL cholesterol concentrations but either no change or decreases in HDL cholesterol (Mensink 1994). Some mechanisms involved in this response of serum lipids to dietary fats have been characterized, including changes in fecal excretion rate of neutral sterols and bile acids, exogenous cholesterol absorption rate, serum lipoprotein composition and catabolism, apolipoprotein synthesis or catabolism rates, and hepatic LDL receptor number (Mazier and Jones 1991). These mechanisms are thought to account for the large and immediate changes seen in serum cholesterol concentration following fat intake modification.

Dietary fat composition may alter cholesterol synthesis rates, but until recently, no simple procedures were available for monitoring short-term changes in humans. The deuterium incorporation method, however, offers a safe, practical tool for investigations into human cholesterol metabolism; it has been compared with other methods and its diagnostic accuracy validated (Jones et al. 1992 and 1993). The method has been used to examine how cholesterol synthesis rates react to changes in diet fat saturation in elderly and hypercholesterolemic individuals (Jones et al. 1994a and 1994b), but to date no work with normal individuals has been conducted to see if their synthesis rates change in response to shifts in diet fat saturation. In addition, it is not known whether diet fat and feeding state interact to alter synthesis rates.

Our objectives were to determine whether in normolipidemic men 1) cholesterol synthesis rates are sensitive to consumption of diets containing either MONO, POLY or SAT and 2) these rates differ if subjects are fed vs. unfed. A study of cholesterogenesis in normal healthy individuals measured using deuterium incorporation was combined with cholesterol turnover measured by the specific activity of radioactively labeled serum cholesterol. The former method is thought to measure de novo cholesterol synthesis in the body's central or M(1) cholesterol pool (Goodman et al. 1973). This pool consists of cholesterol that equilibrates rapidly with plasma cholesterol and includes that in erythrocytes, liver, intestine and other viscera such as lung, pancreas, spleen and kidney. The latter method yields useful measurements such as cholesterol pool sizes and rates of synthesis.

SUBJECTS AND METHODS

For determination of rates of cholesterol synthesis, subjects were randomly assigned to three groups and consumed all meals during each diet period under supervision for 13 d. Subjects were free living but came into the metabolic research unit three times daily; they were expected to eat only what was provided at these meals. On d 13, baseline blood samples were drawn from fasting subjects, followed by pre-breakfast administration of a single oral deuterium oxide dose (99.8 atom percent excess, ICN Biomedicals, Montréal, Québec) of 0.7 g deuterium oxide/kg estimated body water. Body water was estimated as 73% of the fat-free mass, assessed using bioelectrical impedance analysis (model 101, RJL Systems, Detroit, MI) (Kushner and Schoeller 1986). Blood was drawn at 12-h intervals for 48 h. This corresponded to a fed day, when three meals were consumed, followed by a day when no meals were consumed. During the 48-h test period, body water deuterium oxide enrichment was maintained by consumption of lightly labeled water (1.4 g deuterium oxide/kg H₂O). Diet periods were separated by 8-wk intervals, allowing body levels of deuterium oxide to normalize. Diets were given to each subject in random order. All diet periods occurred during the 9-mo ¹⁴C study.

Cholesterol synthesis determinations during each test period. Fractional synthesis rates (FSR), defined as the proportion of the central or M(1) pool replaced daily by newly synthesized cholesterol, were calculated as the change in product deuterium enrichment over time divided by the maximum possible enrichment, based on a linear rate of uptake of label into cholesterol over time (Jones et al. 1988 and 1993). Cholesterol FSR was calculated for each 12-h period in question. The equation used was where del_cholesterol and del_{body water} are differences in deuterium enrichment of each tissue expressed as parts per thousand vs. standard mean ocean water (SMOW). The amending factors in the equation's denominator correct for the absolute ratio of carbon to hydrogen atoms within the cholesterol molecule. Deuterated water enters cells readily and equilibrates quickly with intracellular water. Little unlabeled water is generated intracellularly, allowing the cell precursor pool enrichment to equal that of plasma (Jones et al. 1988). The calculated rate of cholesterol synthesis depends upon the rate of label incorporation per molecule of cholesterol. During the synthesis of cholesterol, hydrogen atoms from water are incorporated into the sterol molecule in three different ways. Seven atoms of hydrogen are incorporated directly from water, 15 atoms from NADPH, and eventually, hydrogen atoms from water are incorporated into the acetyl CoA pool, which can be used as a cholesterol precursor (Dietschy and Spady 1984). If the assumptions are made that over a 48-h period there is complete equilibration of deuterated water with plasma water and with NADPH, but that the acetyl CoA pool is not yet labeled, then 81% of the hydrogen atoms per cholesterol carbon atoms, the H/C ratio, will be incorporated into the molecule from ²H₂O (Dietschy and Spady 1984). Other assumptions that must be incorporated with use of this method include 1) the amount of recycling of label into other pools, such as acetate, during the time of interest; 2) the form of the mathematical equation, usually accepted as monoexponential over short periods of time, of D incorporation over time; and 3) the theoretical and actual maximum plasma cholesterol enrichment (Dietschy and Spady 1984, Jones et al. 1993). The number 2 in the numerator converts the 12-h FSR to a 24-h FSR for each period. Daily de novo total unesterified cholesterol synthesis was calculated as the product of FSR and the mass of the rapidly exchangeable M(1) pool which is unesterified cholesterol. This is defined as one-third of the total M(1) pool, which contains both esterified and unesterified cholesterol (Goodman et al. 1973). This was verified by measuring the proportion of serum cholesterol that was esterified and that was not. Because the M(1) pool cholesterol is rapidly exchangeable, the proportion of esterified to nonesterified cholesterol in serum is thought to reflect that of other segments of the pool (Goodman et al. 1973).

Table 1. Age, height, body weight and body mass index (BMI) of men at the beginning and end, and daily energy intake during each 2-wk diet period1 [View Table]

Table 2. Composition of diet1 [View Table]

Serum samples obtained at 12-h intervals during each 48-h test period were used for measurement of cholesterol synthesis (Jones and Schoeller 1990, Jones et al. 1993). Lipids were extracted from 2 to 4 mL of serum and separated using thin layer chromatography. Unesterified cholesterol bands were transferred to pre-annealed Pyrex (Corning Glass works, Corning, NY) combustion tubes (18 cm × 6 mm) containing 500 mg of cupric oxide (BDH Chemicals, Toronto, Ont.) and 2 cm of 1-mm diameter silver wire. Tubes were sealed under vacuum and heated at 520°C for 4 h. The resulting combustion water was vacuum-distilled into pre-annealed Pyrex tubes containing 60 ± 5 mg of zinc (Biogeochemical Laboratories, Indiana University, Bloomington, IN). Tubes were sealed and heated at 520°C for 30 min to reduce combustion water to gas. Plasma water deuterium oxide-enriched samples were diluted sevenfold to reduce the enrichment to within the range of working standards. Microcapillary tubes containing 2 µL of sample were distilled into Pyrex tubing sections containing zinc and heated at 520°C for 30 min. Enrichment of the hydrogen gas obtained from all samples was determined by isotope ratio mass spectrometry (VG Isomass, 903D, Cheshire, England), with an internal analytical error of 0.17 parts per thousand () relative to SMOW. The mass spectrometer was calibrated daily using SMOW, standard light Antarctic precipitation (SLAP), and Greenland ice sheet precipitation (GISP). Samples were analyzed in duplicate. The overall precision of analysis in this study was determined by averaging replicate SD. The average SD of deuterium oxide enrichment for body water and serum unesterified cholesterol were 1.7 and 8.0, respectively, relative to SMOW.

Table 3. Serum lipid concentrations in men during each 2-wk diet period1 [View Table]

Table 4. Serum triglyceride concentrations in men during each 2-wk diet period measured at 12-h intervals1 [View Table]

Table 5. Three-pool cholesterol model: comparison of the values obtained by Goodman et al. (1973) and those obtained in this study [View Table]

The Gauss-Markov F test was used to find the number of terms in a multiexponential equation that would best fit each specific radioactivity decay curve (Goodman et al. 1973). The percent improvement in residual error when using a three-term instead of a two-term equation, a measure of closeness of fit to the equation, was also calculated for each data set. When equations are generated with CONSAM, the best fit was obtained by a program using the Marquardt nonlinear least squares fitting technique (Marquardt 1963). Because of concern that the specific diets would alter the curves and equations generated, data were generated both with and without the points obtained during diet periods.

Table 6. Cholesterol turnover parameters obtained by fitting each subject's 9-mo specific radioactivity decay data to a multicompartmental model1 [View Table]

RESULTS

Meals did not differ in protein, fat, carbohydrate or cholesterol content (Table 2). For fatty acids, diets differed in 8:0, 10:0, 14:0, 18:0, 18:1 and 18:2 (P  0.05). Diets did not differ in total 16:0 (P = 0.064). Meals offered within each diet period did not differ in fatty acid composition (data not shown).

Serum cholesterol concentrations of each subject when they consumed each diet are listed in Table 3. Total cholesterol concentrations varied with the diet consumed (P = 0.001). Ratios of serum unesterified to esterified cholesterol were not affected by diet fat intake (P = 0.256) (data not shown); one-third was unesterified and two-thirds was esterified. A diet effect was also noted for serum HDL cholesterol concentrations (P = 0.046). Serum triglyceride concentrations showed an effect of time (P = 0.001) and diet (P = 0.026) (Table 4).

Data from the specific radioactivity decay study were next examined; sample decay curves are presented in Figure 1. In all cases except one (Subject F), a three-term equation gave the best fit (P = 0.05). This allowed interpretation of our data based on the model developed by Goodman et al. (1973), which we accomplished using SAAM/CONSAM. Estimates of rates of cholesterol synthesis, total cholesterol produced daily, and mass of cholesterol pools for each subject were generated. Comparisons of the values obtained in an earlier study (Goodman et al. 1973) and those from this investigation are provided in Table 5. Statistical evaluation was not attempted because subjects in our study were normolipidemic, but only half of those who participated in the study of Goodman et al. (1973) were so. Individual M(1) pool sizes are listed in Table 6, along with daily cholesterol production rates when subjects were not consuming test diets. Cholesterol production rates include daily synthesis as well as exogenous or dietary cholesterol (Goodman et al. 1973). Pool M(2) and M(3) sizes are also listed; those from the Goodman study are listed as ranges for pool size as calculated by the authors. Those for the present study were calculated by CONSAM as a single pool size. No significant differences in curves or equations were detected between data sets resulting from the use of all points collected and those where points collected during specific diets were omitted from the 9-mo study.

Total body water deuterium oxide enrichment (Fig. 2) remained at a constant level throughout each 48-h test period; no effect of either diet or time was noted. Unesterified cholesterol deuterium oxide enrichment (Fig. 3) was lower when subjects received the SAT diet than when they received the POLY diet (P = 0.043). Additionally, deuterium oxide enrichment at 0 h and at 12 h was lower than at 24, 36 or 48 h (P = 0.001).

Fractional synthesis rates were greater when men consumed the POLY diet rather than the SAT diet; rates when they consumed the MONO diets did not differ significantly from the other two (Appendix 1) (P = 0.045). Absolute synthesis did not differ among diets, whether expressed as milligrams per day (Appendix 2) (P = 0.769), or milligrams per kilogram of body weight per day (Appendix 3) (P = 0.789). Both FSR (P = 0.001) and ASR expressed as either milligrams per day (P = 0.02) or milligrams per kilogram of body weight per day (P = 0.04) were lower in the last two 12-h periods than in the second, with that in the first 12 h being the lowest.

DISCUSSION

Previous studies using deuterium oxide incorporation to measure cholesterol synthesis in humans are strongly supportive of the present findings. In hypercholesterolemic subjects, synthesis rates were greater in those fed a corn oil-based diet than in subjects receiving either a baseline diet or an olive oil-based diet, but were not different from rates of synthesis in subjects fed a canola oil-based diet (Jones et al. 1994a). In mildly hypercholesterolemic elderly people, rates of synthesis were greater in subjects fed diets containing corn oil than in those fed diets containing beef tallow (Jones et al. 1994b).

Mechanisms have been suggested that may be responsible for the varied effect of different fats on rates of unesterified cholesterol synthesis. Enhanced LDL fractional catabolic rates may be responsible for the low serum cholesterol concentrations seen in subjects consuming POLY or MONO diets (Shepherd et al. 1980). Others have reported enhanced fecal sterol excretion in subjects receiving POLY diets compared with those fed SAT diets (Connor et al. 1969, Nestel et al. 1973); some have suggested that this indicates increased synthesis rates (Jones et al. 1994a and 1994b, Oh and Monaco 1985). Polyunsaturated fats may enhance hepatic cholesterol elimination, up-regulate removal of circulating sterol and thus invoke higher rates of synthesis. The results of the present study support this hypothesis, but not all studies concur. Cholesterol balance studies on hyperlipidemic subjects receiving liquid diets containing butter, safflower oil or sunflower oil were unable to demonstrate appreciable differences in rates of cholesterol synthesis amongst the three diet groups (Grundy and Ahrens 1970, McNamara et al. 1987), and not all studies observed changes in fecal steroid excretion when POLY diets were fed (Grundy and Ahrens 1970, Shepherd et al. 1980, Spritz et al. 1965). Additionally, in guinea pigs fed olive oil-enriched diets, 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) reductase was less active than in those fed either beef tallow-or corn oil-based diets, suggesting a diminished rate of synthesis in guinea pigs fed olive oil diets (Fernandez and McNamara 1994). Although this has not yet been replicated in humans, guinea pigs are thought to be a good model for human cholesterol metabolism.

Deuterium incorporation data in this study indicated that rates of cholesterol synthesis were greater during the fed state than during the food-deprived state. Such results are in agreement with earlier reports comparing rates of cholesterol synthesis between fed and non-fed periods in humans (Jones et al. 1988). The process whereby cholesterol synthesis rates are altered by feeding state is unknown. During periods of food restriction, cholesterol precursors such as acetyl-CoA derived from stored triglyceride may be used as energy substrates, thus limiting their availability as substrates for HMG-CoA reductase and limiting sterol synthesis. Additionally, the activity of HMG-CoA reductase is decreased by glucagon and glucocorticoids, both of which would be present in elevated concentrations during fasting. Work completed by Jones and Schoeller (1990) suggests that insulin may be a factor regulating synthesis during periods of varying food intake. Clearly, cholesterol synthesis rates are potentially sensitive to feeding state. Fractional synthesis rates were low in the first 12 h of each test period; this either indicates that body water deuterium enrichment may not have reached the plateau value or that the shorter time period of availability limited the amount of labeling of various intermediates in the cholesterol biosynthetic pathway. Because this is a common occurrence and not limited to a particular diet intake, the 12-h data can still be used for comparisons of cholesterol synthesis among subjects fed different diets.

At present, the proportion of whole-body cholesterol synthesized in the central pool is unknown. This makes it difficult to compare the rates of cholesterol synthesis obtained when subjects were fed the three specific diets to that obtained over the 9-mo period, when the subjects consumed unrestricted diets: theoretically, the two methods may not measure the same thing. When looking at whole-body synthesis, much attention has been devoted to hepatic synthesis, a source of M(1) de novo cholesterol: it accounts for 40-50% of whole-body newly synthesized cholesterol in selected primates, 70% in baboons and 10% in cynomolgus monkeys (Dell et al. 1985, Spady and Dietschy 1983, Turley et al. 1995). If hepatic synthesis represented the only source of M(1) de novo synthesis in humans, deuterium oxide incorporation data would not represent a complete look at the effect of fat saturation on the rate of whole-body cholesterol synthesis in humans. Other tissues and organs, however, also contribute to this pool's newly synthesized cholesterol, and these contributions are measured by and accounted for with the deuterium oxide incorporation method. In addition, equilibration of deuterium oxide across the central pool is rapid, with over 60% of hepatic pool unesterified cholesterol exchanging with total plasma cholesterol per hour (Schwartz et al. 1993). It is therefore unlikely that we are experiencing any time lag in our measurements of cholesterol synthesis using this method.

The deuterium oxide incorporation method has been criticized, both because there is no physiological basis for a model based on linear regression and because occasionally negative FSR are obtained (Foster et al. 1993). Presently we consider a negative FSR to be an indication that rates are lower than in other periods being compared, but we cannot yet partition out the effects of substrate recycling or incoming cholesterol from other pools. Additionally, although there may not be a physiological basis for a model based on linear regression, the initial short-term deuterium oxide incorporation rate is linear (Jones et al. 1988 and 1993). This linear uptake rate is unaffected by flux rates of other, unlabeled material into the system and can be taken to represent a direct measure of synthesis independent of the total whole-body production rate (Jones et al. 1994a and 1994b). Furthermore, in this study the estimates of daily cholesterol synthesized (Appendix 3) are in agreement with those derived using other techniques, such as sterol balance (Turley et al. 1995). Consequently, the conclusion from this study is that in a simple and short-term one-pool model, where entry of one tracer into one pool is examined and data points are spaced at 12-h intervals, the deuterium oxide incorporation method can yield reasonable estimates of rates of cholesterol synthesis.

In summary, in a study looking at cholesterol metabolism in normal healthy individuals, the deuterium oxide incorporation data confirmed that unesterified cholesterol synthesis is greater when subjects are fed compared with periods of no food consumption. Additionally, rates of unesterified cholesterol synthesis are greater when POLY fats are consumed than when SAT fats are consumed, with rates when subjects consume MONO diets being intermediate. This occurred even though serum cholesterol concentrations were clearly greater when the SAT diet was consumed compared with either of the other two diets. This effect of POLY fats on rates of cholesterol synthesis had previously been shown in hypercholesterolemic and elderly individuals but not in normolipidemic people. Present findings show that although SAT diets do cause a marked increase in serum cholesterol concentrations in normal healthy humans, the increase is not due to a rise in cholesterol synthesis while such diets are consumed. These results support the present dietary recommendations that SAT fat should not be a major component in the diet of normal individuals.