Human Nutrition and Metabolism

Kinetic Model of Folate Metabolism in Nonpregnant Women Consuming [ 2H2 ]Folic Acid: Isotopic Labeling of Urinary Folate and the Catabolite para-Acetamidobenzoylglutamate Indicates Slow, Intake-Dependent, Turnover of Folate Pools1,2 Jesse F. Gregory III,3 Jerry Williamson, Jo-Fu Liao, Lynn B. Bailey and John P. Toth Food Science and Human Nutrition Department, University of Florida, Gainesville, FL 32611– 0370 ABSTRACT In a 10-wk study of folate metabolism in nonpregnant women (21–27 y, n 55– 6 per group), subjects were fed a diet containing ;68 nmol/d (30 mg/d) folate from food. The remainder of the ingested folate was provided as folic acid in apple juice (as nonlabeled during wk 1–2, as [2H2]folic acid during wk 3–10) to yield a constant intake of 454, 680 or 907 nmol/d (200, 300 or 400 mg/d). Isotopic enrichment of total urinary folate and the primary catabolite para-acetamidobenzoylglutamate (ApABG) was determined. Isotopic enrichment of ApABG served as an indicator of labeling of tissue folates. A kinetic model consisting of fast- and slow-turnover nonsaturable pools and a saturable slow-turnover pool, with provisions for urinary and fecal excretion, catabolism and enterohepatic circulation, yielded a close fit to the data. Mean residence times for total body folate were 212, 169 and 124 d for folate intakes of 454, 680, and 907 nmol/d, respectively. The model predicted that variation in folate intake over this range had little effect on the mass of the large saturable folate pool; however, the fast-turnover nonsaturable pools increased in proportion to folate intake, whereas the slow nonsaturable pool also tended to increase. This model will aid in evaluation of folate turnover and in predicting kinetic consequences of physiologic conditions associated with altered folate requirements. J. Nutr. 128: 1896 –1906, 1998. KEY WORDS:





stable isotopes


regeneration, and in the shuttling of one-carbon units that function in many aspects of metabolism and regulation. Inadequate folate nutrition is associated with increased risk of neural tube defects (Scott et al. 1995), certain forms of cancer (Mason 1995), and many forms of vascular disease (Boushey et al 1995, Morrison et al. 1996, Rimm et al. 1998). The metabolic effects of folate deficiency involved in these disease processes may include elevation in plasma homocysteine concentration (Selhub et al. 1993), impaired nucleic acid synthesis (Wagner 1995), reduced methylation of regulatory elements of certain genes (Mason 1995), and increased DNA fragmentation due to misincorporation of uracil (Blount et al. 1997, Pogribny et al. 1997). The long-term goal in understanding of folate requirements should involve defining intakes that minimize such deleterious processes and optimize folatedependent processes in metabolism and cellular development. Investigation of the in vivo kinetics of folate provides an integrated view of the relationships among rates of intake, turnover, masses of in vivo folate pools and relative significance of excretory processes. Such information will strengthen our understanding of how changes in folate intake influence the quantity of folate available for metabolic processes and will aid in defining the nutritional requirement for folate more fully. The study of in vivo kinetics and the development of mathematical models in human subjects provide information relevant to our understanding of nutrient requirements that

Adequate nutritional status for folate depends on a longterm intake to provide concentrations of the various tetrahydrofolate (H4folate)4 coenzymes in tissues sufficient to maintain optimal metabolic function. Concentrations of folate coenzymes in vivo depend largely on the quantity and bioavailability of ingested folate and the rate of their loss by urinary and fecal routes and through catabolism, although these relationships have not been fully defined. A better understanding of these relationships among the overall rate of folate turnover, mass of various in vivo pools, intake and bioavailability may aid in defining the nutritional requirement for this vitamin more precisely. Understanding of the public health importance of adequate folate nutrition is rapidly expanding. Folate nutrition is intimately linked to cellular replication and homeostasis through the function of folates in nucleic acid synthesis, methionine

1 Supported by USDA National Research Initiative Competitive Grants Program #92–37200 –7466, National Institutes of Health Clinical Research Center grant RR00082, and funds from the Florida Agricultural Experiment Station. This is Florida Agricultural Experiment Station Journal Series No. R-06237. 2 The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ‘‘advertisement’’ in accordance with 18 USC section 1734 solely to indicate this fact. 3 To whom correspondence should be addressed. 4 Abbreviations used: ApABG, para-acetamidobenzoylglutamate (N-acetylpara-aminobenzoylglutamate); GCMS, gas chromatography-mass spectrometry; [2H2]folic acid, [39,59-2H2]folic acid; [2H4]folic acid, [glutamate-2H4]folic acid; H4folate, tetrahydrofolate; pABG, para-aminobenzoylglutamate; t½, half-life.

0022-3166/98 $3.00 © 1998 American Society for Nutritional Sciences. Manuscript received 13 March 1998. Initial review completed 13 May 1998. Revision accepted 24 June 1998. 1896


does not depend on interpretation of results from animal models, may yield greater insight into the physiology of nutrient processing and metabolism, and allows simulation and prediction of the effects of altered nutritional or physiologic conditions. As reviewed recently (Gregory and Scott 1996), the in vivo kinetics of folate metabolism and turnover have been examined in animals and humans. Studies of short-term kinetics yield useful information regarding plasma concentrations after various folate doses (Anderson et al. 1992, Bunni et al. 1989, Loew et al. 1987, Menke et al. 1993, Priest et al. 1991, Rogers et al. 1997), but they are not suitable for development of mathematical models of long-term whole-body folate kinetics. Kinetic investigations of folate turnover in animals have provided evidence of at least two identifiable pools, demonstrated by direct analysis of tissues (Bhandari and Gregory 1992, Lakshmaiah and Bamji 1981, Murphy and Scott 1979, Scott and Gregory 1996, Tamura and Halsted 1983). Several previous studies have examined folate kinetics in humans, although none has provided a full kinetic model nor has the influence of nutritional status been investigated. Krumdieck et al. (1978) administered radiolabeled folic acid to a single female subject and observed substantial catabolism and fecal excretion in folate turnover with an apparent halflife (t½) of ;100 d for the primary folate pool. Fecal excretion of labeled folate or catabolites was found to be an important excretory process, and the presence of labeled pterins in urine indicated that cleavage of the folate molecule was a catabolic process. Cleavage of the 9C-10N bond of the folate molecule is the only known mechanism of folate catabolism (Murphy et al. 1976, Murphy and Scott 1979); the major catabolic product in urine is the para-acetamido derivative of para-aminobenzoylglutamate (pABG) (McPartlin et al. 1992, Murphy et al. 1976, Murphy and Scott 1979). Additional kinetic studies in human subjects have suggested that folate turnover is accelerated by high intakes (;2–5 mg/d; Russell et al. 1983, Von der Porten et al. 1992), although the dose dependence of turnover rate at lower intakes has not been determined. This research group has reported a preliminary kinetic model of whole-body folate turnover in human subjects on the basis of chronic administration of deuterium-labeled folic acid (Stites et al. 1997). The model consisted of a small fastturnover pool in equilibrium with a large slow-turnover pool, with a provision for urinary excretion of intact folate and for other losses, i.e., fecal and catabolic. On the basis of measurement of isotopic enrichment of urinary folate, it was predicted that folate turnover would be very slow, as reflected by an apparent fractional catabolic rate of 0.008 6 0.001/d (mean 6 SEM, n 5 4) for estimated total folate intakes of 649 –1324 nmol/d (286 –584 mg/d). Several limitations of this modeling approach were as follows: 1) modeling was imprecise because calculations were based on urinary folate excretion, which comprised only 1–2% of folate turnover; 2) the relative significance of fecal excretion and catabolism could not be determined; 3) folate intake was not controlled; and 4) predicted masses of in vivo pools were directly proportional to folate intake. However, this study provided new quantitative information regarding long-term folate metabolism and demonstrated the feasibility of long-term kinetic modeling of folate metabolism using chronic administration of deuterium-labeled folic acid. We have conducted a controlled dietary study in which nonpregnant women were fed diets that provided intakes of 454, 680 or 907 nmol/d (200, 300 or 400 mg/d) for 10 wk. Diets were designed to provide only 68 nmol/d (30 mg/d) of folate from food sources, with the remainder given as synthetic folic acid in apple juice. During wk 1–2, the synthetic folic


acid was not isotopically labeled, whereas during wk 3–10, a portion of this folic acid was deuterium-labeled. Results of this study regarding relationships of dietary intake and serum and erythrocyte folate concentration, urinary folate excretion, and plasma homocysteine concentration have been reported previously (O’Keefe et al. 1995). We report here additional results of this study including isotopic excretion as urinary folate and the primary catabolite, para-acetamidobenzoylglutamate (ApABG), as a function of folate intake, along with the development of an expanded kinetic model of folate metabolism. The data derived from this study and the resulting model will provide an initial quantitative picture of whole-body folate metabolism in adequately nourished young women and will serve as a basis for additional evaluation of conditions associated with changes in folate requirements. SUBJECTS AND METHODS Overview of protocol. The details of the protocol and diet composition have been reported previously (O’Keefe et al. 1995). Nonpregnant female subjects (n 5 18, age 21–27 y, weight 47– 67 kg) had normal blood chemistry and were in good health as reflected by a medical history and examination by a physician. This study was approved by the University of Florida Institutional Review Board. Informed consent was obtained from each subject. Subjects were randomly assigned to three treatment groups, n 5 6 per group. One subject withdrew for personal reasons midway through the study. The protocol was conducted on an out-patient basis at the University of Florida Clinical Research Center. Consumption of meals and supplements was supervised by research personnel, and compliance was encouraged through daily interaction with the subjects. Adequacy of vitamin and mineral intake was ensured by administration of folatefree vitamin supplements (Fos Free, Mission Pharm, San Antonio, TX), a mineral supplement (Solgar Chelated Solamins Multiminerals, Solgar Vit, Lynbrook, NY), a potassium supplement (K-DUR 10, Key Pharm, Kenilworth, NJ) and a calcium supplement (Albertsons, Boise, NJ). Intake of dietary energy, protein, and fat and of supplemental vitamins and minerals was reported previously (O’Keefe et al. 1995). Blood and urine collections were conducted at designated intervals throughout the study. All kinetic analysis was based on urinary folate and ApABG excretion. For this purpose, 24-h urine collections were made directly into acid-washed brown plastic 2-L bottles containing 5 g of dry sodium ascorbate. Each 24-h collection was begun after the first morning void and included the first void of the following morning. These bottles were kept refrigerated (2– 4°C) during the collection period. Immediately after the 24-h collection was completed, total volume was measured, 50 mL portions were transferred to polyethylene vials, saturated with nitrogen gas and stored at 230°C until analyzed. All blood collection and analysis procedures were reported in the previous paper (O’Keefe et al. 1995). Three diet composites were collected for measurement of total folate, as described and reported previously (O’Keefe et al. 1995). Folic acid sources administered. Nonlabeled folic acid supplements were prepared from commercially available folic acid (Sigma Chemical, St. Louis, MO), whereas the [39,59-2H2]folic acid ([2H2]folic acid) used as a stable-isotopic tracer was synthesized in this laboratory (Pfeiffer et al. 1997). Each was analyzed to verify purity and identity by HPLC, proton nuclear magnetic resonance and gas chromatography-mass spectrometry (GCMS) before use (Gregory 1990). Solutions of each were prepared in 0.1 mol/L PBS (pH 7.0) and the concentration determined spectrophotometrically using the molar absorptivity coefficient of 27,600 L/(mol z cm) (Blakley 1969). Appropriate volumes of each solution were dispensed into commercial pasteurized apple juice and stored as 45-mL portions in 50-mL conical centrifuge tubes, saturated with nitrogen gas and stored at 230°C until used. During d 1–14, the supplemental folic acid consisted only of nonlabeled folic acid in apple juice (45-mL portions given at morning and evening meals). During d 15–70, the supplemental folic acid consisted of an equimolar blend of nonlabeled ([1H]) and [2H2]folic acid in apple juice (45-mL portions given at morning and



TABLE 1 Summary of folate intake by nonpregnant women during the 10-wk protocol upon which kinetic modeling was based

Total folate intake

Stage of protocol



454 (200 mg/d)

1–14 15–70 1–14 15–70 1–14 15–70

680 (300 mg/d) 907 (400 mg/d)

Folate from foods

Total synthetic folic acid

Nonlabeled folic acid

[3959-2H2]Folic acid

Isotopic enrichment of total ingested folate

386 193 612 306 839 419.5

0 193 0 306 0 419.5

0 0.425 0 0.450 0 0.463

nmol/d 68 68 68 68 68 68

386 386 612 612 839 839

1 Folate sources included endogenous food folate, nonlabeled folic acid and [39,59-2H2]folic acid.

evening meals). The concentration of these forms of folic acid in the apple juice was confirmed by HPLC (Gregory and Toth 1988), and stability was verified by analysis at several times throughout the study. A summary of the folate intake of subjects in each group is shown in Table 1. The final phase of this study was to determine whether the urinary excretion of a single bolus dose of a labeled folate could be used as a functional indicator of folate status. For this purpose, the following single-day protocol was conducted immediately after the study described in this paper. After completion of the last urine collection on d 70, a single dose of [glutamate-2H4]folic acid ([2H4]folic acid) [1134 nmol (500 mg) Gregory and Toth 1988] was administered to each subject; then each subject collected urine for 24 h. The results of this short-term study are reported in a separate communication (Gregory et al. 1998). Analytical methods. Documentation of folate nutritional status. As described previously, serum, erythrocyte and urinary folate concentrations were determined by microbiological assay with Lactobacillus casei (Tamura 1990). Total plasma homocysteine concentrations were determined by a fluorometric HPLC procedure (Vester and Rasmussen 1991). Determination of urinary folate by HPLC and preparation of urinary folate for GCMS analysis. Urinary folate concentration was determined by HPLC after affinity chromatography (Gregory and Toth 1988). This method is based on isolation and purification of urinary folate using columns packed with Affigel 10 (BioRad Laboratories, Hercules, CA) coupled to bovine milk folate-binding protein. Recovery of 5-methyl-H4folate and folic acid added to urine was typically .95%. Care was taken to maintain the quantity of total folate applied to the affinity column at ,30% of column capacity to ensure high recovery of all folates. The 5-mL fraction containing folate eluted from this column was divided as follows: 1 mL was used for HPLC analysis (Gregory and Toth 1988, Stites et al. 1997) and the remainder prepared for GCMS determination of isotopic enrichment (Pfeiffer and Gregory 1997). Preparation for GCMS analysis involves intentional cleavage of the 9C-10N bond, isolation of the resulting pABG by HPLC and derivatization with combined trifluoroacetic anhydride and trifluoroethanol (Gregory and Toth 1988) to form N-trifluoroacetyl-p-aminobenzoylglutamate lactam a-trifluoroethyl ester. Determination of total urinary pABG and preparation of ApABG for GCMS analysis. Total urinary catabolite excretion (sum of pABG and ApABG) was determined quantitatively by HPLC using a modification of the method of Wang et al. (1994). To a 170-mL portion of the folate-free effluent from the affinity chromatography column, 10 mL of [3H]ApABG (333 Bq; prepared according to McPartlin et al. 1992) and 20 mL of 8 mol/L HCl were added in a 5-mL conical screw-cap tube. The mixture was heated for 1 h to effect deacetylation; deacetylation was omitted in several trials to measure pABG alone. Two milliliters of 0.1 mol/L potassium phosphate buffer (pH 7) was added, followed by 100 mL of a solution of Fluorescamine (3 mg/mL in acetonitrile; Sigma Chemical) and mixed thoroughly. The

entire mixture was transferred to a PrepSep C-18 solid-phase extraction column (Fisher Scientific, Pittsburgh, PA) previously conditioned with 5 mL methanol and 5 mL sodium phosphate buffer (pH 6.0). After a wash with 10 mL of 0.033 mol/L sodium phosphate buffer (pH 6.0), the pABG-Fluorescamine derivative was eluted with 5 mL of 0.033 mol/L sodium phosphate buffer (pH 6.0) containing 13% (v/v) acetonitrile. Quantification was performed by HPLC using a Microsorb-MV C18 column (4.6 mm i.d. 3 100 mm, 3-mm particle size octadecylsilyl; Rainin Instrument, Woburn, MA) with a mobile phase of 0.033 mol/L sodium phosphate (pH 6.0) containing 20% (v/v) methanol and 20% (v/v) acetonitrile, pumped at 1 mL/min. Fluorescence was monitored (Model LS-5, Perkin-Elmer, Norwalk, CT) at excitation and emission wavelengths of 400 and 500 nm, respectively. Fractions (1 mL each) were collected in a fraction collector. Those fractions corresponding to the pABG derivative were mixed individually with 4 mL of scintillation fluid (ScintiVerse II, Fisher Scientific, Fair Lawn, NJ) and radioactivity determined by liquid scintillation spectrometry with correction for quenching (Model LS-9000, Beckman Instruments, Fullerton, CA). The concentration of total urinary pABG was calculated relative to the response of identically prepared standards, with correction for the recovery of the internal standard. The intra-assay recovery of the radiolabeled ApABG internal standard added to urine samples (n 5 14) was 48.2 6 5.6% (mean 6 SD), whereas the interassay recovery (n 5 6) was 44.7 6 4.8 % (mean 6 SD). A typical chromatogram is shown in Figure 1. Urinary ApABG was isolated and prepared for GCMS analysis as an indicator of the isotopic enrichment of tissue folates. Urine (pH 7) was first passed over the folate-binding protein affinity chromatography columns, as described above, and the folate-free effluent collected. Both pABG and ApABG had been shown previously to have no retention on this column. This effluent was acidified to 0.1 mol/L with HCl; then ApABG was isolated by the method of McPartlin et al. (1992). Although the McPartlin method specifies the use of tritium-labeled internal standards to compensate for low and variable recovery, internal standards were omitted to avoid complicating subsequent GCMS analysis. Thus, no effort was made to quantify ApABG in this procedure as performed preparatively for this study. This purification method initially involves application of the sample to a column packed with Dowex 50 (Sigma 50X8-400, Sigma Chemical) equilibrated in 0.1 mol/L HCl, followed by collection of the effluent and a 50 mL 0.1 mol/L HCl wash to recover the ApABG, which is not retained. This fraction is adjusted to 0.2 mol/L HCl and incubated in a boiling water bath for 1 h to deacetylate the ApABG. The resulting pABG was recovered and partially purified by application to a Dowex 50 column equilibrated in 0.2 mol/L HCl. After a wash with 50 mL 0.3 mol/L HCl, the pABG was eluted and collected. The pABG in this fraction was converted to the naphthylethylenediamine derivative, then applied to a C18 Sep-Pak solid phase extraction column (Waters Division, Millipore, Milford, MA), washed with 10 mL water, then eluted with 4 mL methanol and evaporated to dryness under nitrogen gas. The naphthylethylenedia-


FIGURE 1 Typical chromatogram from fluorometric HPLC determination of total para-aminobenzoylglutamate (pABG) in urine. This analysis measures the sum of free pABG and its acetylated derivative acetamidobenzoyglutamate (ApABG).

mine moiety was removed by zinc-HCl treatment, continuing as described by McPartlin et al. (1992). This entire solution was subjected to preparative HPLC, and the pABG peak collected and evaporated to dryness (Pfeiffer and Gregory 1997). Derivatization for GCMS analysis was then performed as described above (Gregory and Toth 1988). Gas chromatography-mass spectrometry analysis of pABG. GCMS analysis of derivatized pABG (derived from both urinary folate and urinary ApABG) was performed as previously described in electroncapture negative ionization mode with selected-ion monitoring at mass-to-charge ratios (m/z) 426 and 428 (Gregory and Toth 1988). All analyses were performed using a Hewlett-Packard Model 5989 GCMS system (Palo Alto, CA) with methane as reagent gas. Working standard response curves were prepared by using known mixtures of labeled [2H2] and nonlabeled pABG (prepared from known mixtures of [2H2] and nonlabeled folic acid) to determine the relation between ratios of observed peak areas in GCMS analysis by selectedion monitoring and the actual molar ratios of labeled and nonlabeled folates. All standard mixtures and samples were analyzed in duplicate or triplicate, and ratios of labeled and nonlabeled forms of pABG were determined using simultaneous equations that corrected for the natural abundance of isotopomers. Kinetic modeling. All modeling was conducted using the compartmental analysis module of SAAM II software, version 1.1 (SAAM Institute, University of Washington, Seattle, WA; Foster et al. 1994) on a personal computer. The model developed in this study was based on that reported previously (Stites et al. 1997), with changes described below to increase its physiologic accuracy. The previous model consisted of a fast-turnover (Pool 1) and a slowturnover pool (Pool 6), the latter presumably comprised of folates associated with tissues; both of these are nonsaturable compartments. Ingested folate (Pool 5) entered Pool 1, with assumed 67% absorption. The assumed 67% bioavailability was based on the reported low bioavailability of many sources of food folate (Sauberlich et al. 1987) and the observed ;85% absorption of folic acid (dissolved in apple juice) when consumed with a light meal (Pfeiffer et al. 1997). This preliminary model had provisions for losses as urinary folate and via other routes (catabolic and fecal). The expanded model devised for this study (Fig. 2) included these pools plus the following additional characteristics: 1) A third saturable pool (Pool 4) was included that represented the major slow-turnover folate compartment. 2) Provisions were made for urinary excretion of intact folates from Pools 1 and 6. 3) Additional provisions were introduced for catabolic losses from tissue folate pools (Pools 4 and 6). 4) Losses of folate via fecal


excretion also were included from both Pools 4 and 6. 5) Secretion of folate from Pool 6 into an intestinal compartment (Pool 7) was included to represent digestive secretions such as bile and pancreatic juice. After a 0.5-d delay (compartment 8), this secreted folate entered Pool 9 from which a large fraction is reabsorbed back to Pool 1, whereas the fraction not reabsorbed from Pool 9 undergoes fecal excretion. 6) Provision also was made for incomplete bioavailability of dietary folate (again assumed to be 67% overall) by adjusting the fraction of ingested folate entering Pool 5 that underwent transfer to Pool 1. As in previous modeling, compartment 2 and compartment 3 had no anatomical or physiologic equivalent but were simply a sink for totaling catabolic losses (as ApABG) and urinary excretion of folate, respectively. Sources of folate intake were designated in SAAM II modeling as “endogenous input” (total nonlabeled folate, i.e., food folate and nonlabeled folic acid), and “exogenous input” ([2H2]folic acid), at levels shown in Table 1. The mass of Pool 4, the large, saturable, tissue folate pool, was defined by a Michaelis-Menten expression, [Bmax 3 mass(6)]/[Kd 1 mass(6)], with subtraction of first-order– based outflow to Pools 2, 6 and 10. In this expression, Bmax was estimated to be the total binding capacity of Pool 4 and Kd was an empirical constant representing the overall dependence of saturable processes and saturable binding on the mass of free folate in tissues. In this model developed with SAAM II, the exchange rate constants [k(I, J)] represented the fraction of folate in compartment J transferred to compartment I per unit time (d). The mass transferred from compartment J to compartment I per unit time was termed the flux (I, J), which represents mass(J) 3 k(I, J). The following assumptions were made in modeling on the basis of known aspects of folate metabolism: 1) The mass of the rapidturnover pool (Pool 1) is much smaller than that of the combined main tissue folate pools (Pools 4 and 6). 2) Although the total mass of body folate has not been determined directly, we assumed on the basis of folate in human liver that total body folate mass would not exceed ;90 mmol (;40 mg expressed as monoglutamyl folate equivalents). 3) Mean concentration of bile folate was estimated to be ;90 nmol/L (;40 ng/mL; Lavoie and Cooper 1974), with a typical bile flow of 600 –700 mL/d (Guyton 1971) and mean total biliary folate of

FIGURE 2 Kinetic model of folate metabolism in humans. Labels: Pool 5, GIT 5 gastrointestinal tract (into which all folate initially enters); Pool 1, Fast 5 rapid-turnover pool; Pool 6, Slow-F 5 slow-turnover free folate pool; Pool 4, Slow-B 5 slow-turnover bound folate pool; Pool 2, Uapabg 5 urinary para-acetamidobenzoylglutamate (ApABG); Pool 3, Ufolate 5 urinary folate; Pool 7, SI 5 small intestinal pool into which biliary folate enters, followed by a 0.5-d delay (Pool 8); Pool 9, SI 5 small intestinal pool from which biliary folate is reabsorbed to Pool 1; Pool 10, Fecal 5 fecal folate derived from endogenous sources (e.g., digestive secretions and sloughed mucosal cells). Compartment 8 constitutes a delay affecting the rate of folate passage through enterohepatic circulation.



TABLE 2 Serum and erythrocyte folate concentration, plasma total homocysteine concentration, and urinary folate excretion by nonpregnant women after controlled folate intake for 10 wk1,2 Folate intake

Serum folate

nmol/d 454 (200 mg/d) 680 (300 mg/d) 907 (400 mg/d)

Erythrocyte folate nmol/L

6.4 6 0.76a 7.3 6 1.1a 14.3 6 2.0b

357 6 19.4 440 6 47.1 567 6 83.0

Plasma homocysteine

Urinary folate



12.6 6 1.67b 8.40 6 1.21a 7.72 6 0.667a

3.37 6 0.24a 6.07 6 1.78a 24.5 6 10.0b

1 Values are means 6 SEM, n 5 5 for 454 nmol/d group, n 5 6 for 680 and 907 nmol/d groups. Within a column, values followed by a different superscript letter are significantly different, P , 0.05. Urinary folate values were subjected to logarithmic transformation before ANOVA to improve normality and compensate for unequal variance. 2 Data within these columns were presented previously (O’Keefe et al. 1995) and are presented to document folate status of subjects used in kinetic modeling. All values are for the final blood draw or urine collection during wk 10 of the protocol described in this paper.

58 nmol/d. 4) The Michaelis-Menten behavior of the major saturable folate pool was assumed on the basis of published studies in which tissue folate and whole-body folate exhibited such a relationship to dietary folate intake (Keagy 1982). 5) Folate catabolism was assumed to occur only in tissue pools, and urinary ApABG was the only catabolite included in modeling. Acetylation of pABG catalyzed by arylamine N-acetyltransferases (Minchin 1995) occurs in the cytosolic fraction of human tissues; thus isotopic enrichment of urinary ApABG was assumed to reflect the isotopic enrichment of tissue folate pools undergoing catabolism. The origin of free (nonacetylated) urinary pABG has not been determined. Formation of nonacetylated pABG may arise from instability of urinary folate before or after excretion. Free pABG is a minor component of total excretion of these catabolites (Caudill et al. 1998, McPartlin et al. 1992 and 1993). The model was initially fit to the experimental data that were entered as isotopic enrichments of urinary folate and urinary ApABG. Transfer rate constants were manually altered in systematic fashion until a reasonable fit to the data was obtained and results were consistent with the above assumptions. The iterative “fit” function was then used to solve and adjust rate constants until the model’s “objective function” was minimized, which indicated optimal fit. To visualize relative significance of urinary folate excretion and catabolic processes in overall folate turnover more easily, the data were then converted to express urinary [2H2]folate and urinary [2H2]ApABG as nanomoles excreted per day, followed by the same modeling process. Model-based simulations. Several simulations were conducted based on the model developed in this study to estimate the effect of further variation in folate intake. These simulations evaluated in vivo folate mass, as follows: 1) effects of elevated folate intake (1500, 2000 and 4000 nmol/d) were evaluated using rate constants derived from the 907 nmol/d intake of this study; and 2) effects of lower folate intake (100, 175 and 300 nmol/d) were evaluated using rate constants derived from the 454 nmol/d intake of this study. Statistical analysis. Differences among dietary groups with respect to rate constants and masses of folate pools were evaluated using one-way ANOVA, with multiple comparisons made by the StudentNewman-Keuls procedure using SigmaStat Version 1.0 software (Jandel, San Rafael, CA). Repeated measures one-way ANOVA was used to assess differences among groups with respect to total pABG excretion. Finally, the strength of relationships among all indicators of folate nutritional status was evaluated using the Pearson productmoment correlation procedure. These analyses were performed as described by Glantz (1992). Differences with P , 0.05 were considered to be significant.

RESULTS Folate nutritional status, excretion of folate and total p-aminobenzoylglutamate. Descriptive indicators of folate status among treatment groups of this study have been reported previously (O’Keefe et al. 1995). A brief summary is presented

in Table 2. Significant differences were observed in serum folate and plasma homocysteine concentrations and urinary folate excretion (P , 0.05). Essentially all urinary folate was 5-methyl-H4folate in HPLC analysis, with little or no unchanged folic acid regardless of folate intake. No difference was seen among dietary groups with respect to total pABG excretion. Mean values for total urinary pABG excretion for intakes of 454, 680 and 907 nmol/d were 341 6 75, 298 6 41 and 342 6 63 nmol/d, respectively (means 6 SEM; four analyses per subject of samples collected over 4 wk midway through isotopic administration period). These results indicate that total pABG excretion greatly exceeded excretion of intact urinary folate and that the fraction of intake that it comprised varied inversely with the intake level. When acid hydrolysis was omitted in randomly selected samples, the pABG peak was reduced to ,20% of that seen when using the hydrolysis reaction. Thus, ApABG constituted .80% of the total catabolite excretion, consistent with data reported by McPartlin et al. (1992 and 1993) and Caudill et al. (1998). This suggests that mean excretion of ApABG was ;260 nmol/d [i.e., 80% of the mean total pABG excretion (327 nmol/d)]. This value is consistent with previous reports that ApABG excretion was much greater than folate excretion. Isotopic enrichment of urinary folate and p-acetamidobenzoylglutamate. Isotopic enrichment of urinary folate and urinary ApABG increased gradually throughout the study. Maximum values were observed in the 907 nmol/d intake group and approached isotopic enrichments of 0.3. Relative to the calculated isotopic enrichment of total ingested folate of 0.425– 0.465 ([2H2]folic acid/total ingested folate; Table 1), these results indicate that isotopic equilibrium of body folate pools was not reached within the 8-wk period of [2H2]folic acid administration. The patterns of labeling for all groups were qualitatively similar at each folate intake. The major difference observed between groups was a greater initial rise in enrichment at the high intake. Kinetic modeling. The model shown in Figure 2 provided good fit to the isotopic enrichment data for all subjects (for example, Fig. 3). Analogous modeling was then conducted after transformation of the data from isotopic enrichment to excretion (nmol/d) of urinary [2H2]folate and [2H2]ApABG to allow evaluation of the relative extent of isotopic excretion by each of the primary routes (i.e., urinary [2H2]folate, urinary [2H2]ApABG and apparent fecal excretion). Because direct measurement of urinary ApABG concentration was not performed in this study, we employed an assumed



ApABG excretion of 220 nmol/d for all subjects for the calculation of urinary [2H2]ApABG excretion from isotopic enrichment. This assumed value was chosen as intermediate between the estimated 260 nmol/d ApABG excretion in this study and values of 95 nmol/d (Caudill et al. 1998) and ;160 nmol/d (McPartlin et al. 1993) for ApABG excretion in nonpregnant women. Urinary excretion of [2H2]folate was calculated from measured urinary folate excretion and isotopic enrichment values. Application of the model to these excretion data yielded good fit. Representative results for each level of folate intake are shown in Figure 4. A summary of the model-derived estimates of rate constants, overall fractional catabolic rates and overall mean residence times for each level of folate intake is presented in Table 3, and estimates of masses are shown in Table 4. The estimated mass of total body folate in this model was controlled primarily by the capacity and binding constant of the saturable pool. Several observations made during modeling merit comment. Urinary folate data could be accurately fit only when including output of urinary folate from both Pools 1 and 6 in the model. Similarly, urinary ApABG data could be accurately fit only when output was from both Pools 4 and 6. This model is fit to data for urinary excretion of [2H2]folate and [2H2]ApABG; thus, losses by fecal and/or other routes are estimated. A steady state is attained only when rate constants for fecal excretion are in the range shown in Table 3, which provides sufficient flux via the fecal route to keep the model in balance. With respect to overall folate turnover, as reflected by fractional catabolic rates, the main characteristic found in modeling was very slow turnover of whole-body folate. Modelpredicted mean residence times were .100 d for all subjects and were .200 d for most subjects at the lowest folate intake. Finally, estimated masses of Pool 1, the rapid turnover pool, ranged from ;0.6 to 1.2 mmol for all subjects in this study. Dividing these mass(1) values by an assumed 3-L plasma volume yields an apparent concentration greater than that of plasma folate. This observation suggests that rapidly exchanging folates in certain tissues contribute to the rapid turnover folate pool in addition to plasma folate, although it also is possible that the model overestimates the mass of this pool. The total folate intake of treatment groups was found to influence several kinetic parameters (Table 3). The rate constants for secretion of urinary folate, k(3,1) and k(3,6), at the highest folate intake were significantly greater than those of the lower two intake levels (P ,0.05). A trend (P 5 0.122) was observed for an effect of folate intake on the rate constant for catabolism from Pool 6 [k(2,6)]. Rate constants for fecal excretion of folate from Pool 6, k(10,6), were significantly greater at the higher two levels of folate intake. Whole-body fractional catabolic rates increased with increasing folate intake, with significant differences between values at each intake level (P , 0.05). This effect corresponded to large differences in mean residence times for whole-body folate (i.e., means of 212, 169 and 124 d for intakes of 454, 680 and 907 nmol/d, respectively). With respect to predicted masses (Table 4), a significant effect of folate intake was observed only for Pool 1, the small, rapid-turnover pool. However, trends that did not achieve significance that related folate intake with the mass of Pool 6 (P 5 0.119) and total mass (P 5 0.102) were found.

FIGURE 3 Isotopic enrichment of urinary folate and urinary paraacetamidobenzoylglutamate (ApABG) excreted by nonpregnant women consuming [2H2]folic acid during chronic total folate intake of 454, 680 or 907 nmol/d. The solid and dashed lines are model predictions for isotopic enrichment of urinary folate and ApABG, respectively.



TABLE 3 Model-derived estimates of transfer rate constants, fractional catabolic rates and mean residence times of body folate in nonpregnant women with total folate intakes of 454, 680 or 907 nmol/d1,2,3 Folate intake Rate constants (d21) k(1,9) 3 101 k(10,4) 3 103 k(10,6) 3 104 k(2,4) 3 104 k(2,6) 3 104* k(3,1) 3 104 k(3,6) 3 105 k(4,6) 3 102 k(6,4) 3 103 Overall fractional catabolic rate (d21) 3 103 Mean residence time (d)

454 nmol/d (n 5 5)

680 nmol/d (n 5 6)

907 nmol/d (n 5 6)

8.20 6 0.80 5.30 6 0.20 2.80 6 1.80a 2.22 6 0.46 9.12 6 1.73 9.13 6 2.03a 7.11 6 3.48a 2.10 6 0.24 3.08 6 1.13

8.33 6 0.67 18.9 6 1.19 36.5 6 14.1b 7.14 6 5.64 17.9 6 2.5 9.47 6 2.03a 7.82 6 1.59a 2.17 6 0.40 3.11 6 1.13

7.00 6 0.89 9.40 6 1.9 26.5 6 12.5b 4.14 6 1.84 21.4 6 5.6 15.8 6 12.3b 46.2 6 9.7b 3.15 6 0.60 4.77 6 1.89

4.74 6 0.19a

6.07 6 0.39b

8.22 6 0.46c

212 6 8c

169 6 12b

124 6 7a

1 Values are means 6 SEM. Decimal values presented as exponential for clarity; for example, a k(I,J) value of 0.00398 would be shown as k(I,J) 3 103 5 3.98. 2 Fixed values for the following rate constants were assumed in modeling (units are d21): k(1,5) 5 0.67, k(0,5) 5 0.33, k(0,2) 5 1, k(0,3) 5 1, k(1,6) 5 1e-5, k(6,1) 5 0.9, k(7,6) 5 0.005. 3 All rate constants, k(I,J), and fractional catabolic rates were logarithmically transformed before ANOVA to improve normality and compensate for unequal variances. Within a row, values followed by a different superscript letter are significantly different, P , 0.05. For k(2,6) marked with an asterisk (*), there was a nonsignificant trend (P 5 0.122) for ANOVA.

The ratio of the mass of Pool 6 to the combined masses of Pools 4 and 6 was also evaluated to determine the fraction of tissue folate in “free” (nonbound) form. The model predicted that 30 – 40% of folate in tissue pools was in free form. This fraction tended to increase (P 5 0.119) as a function of folate intake (Table 4). Because of the small number of subjects in each group (n 5 5– 6), which yielded less than optimal statistical power, these trends are reported to convey information regarding the apparent response of the model. Model-based simulation studies. The kinetic model developed here was used in several simulations to obtain preliminary predictions of the effects of several altered physiologic or nutritional conditions (Fig. 5). Simulations of higher and lower levels of folate intake showed that Pools 1 and 6 exhibited the greatest response. The saturable Pool 4 exhibited a slight reduction with decreasing folate intake and a slight increase with higher intakes. DISCUSSION The kinetic model described here represents an extension of previous kinetic analyses of human folate metabolism because of the greater physiologic relevance of this model to

FIGURE 4 Excretion (nmol/d) of urinary [2H2]folate and [2H2]paraacetamidobenzoylglutamate ([2H2]ApABG) by nonpregnant women consuming [2H2]folic acid during chronic total folate intake of 454, 580 or 907 nmol/d. The solid and dashed lines are model predictions for urinary excretion of [2H2]folate and [2H2]ApABG, respectively.


TABLE 4 Model-derived estimates of mass of folate pools and totalbody folate for nonpregnant women consuming total folate intakes of 454, 680 or 907 nmol/d Folate intake Mass(1), mmol Mass(4), mmol Mass(6), mmol* Total mass. mmol* Mass(6)/[mass(4) 1 mass(6)]*

454 nmol/d (n 5 5)

680 nmol/d (n 5 6)

907 nmol/d (n 5 6)

0.611 6 0.003a 43,5 6 0.13 20.4 6 2.2

1.02 6 0.06b 43.5 6 0.17 27.0 6 3.4

1.19 6 0.003c 43.7 6 0.07 28.1 6 2.3

64.5 6 2.3

71.5 6 3.6

73.0 6 2.4

0.316 6 0.025

0.376 6 0.028

0.388 6 0.022

1 Values are means 6 SEM. Total mass 5 mass(1) 1 mass(4) 1 mass(6). 2 For mass(1) and mass(4), values were logarithmically transformed before ANOVA to improve normality and compensate for unequal variance. Within a row, values followed by a different superscript letter are significantly different, P , 0.05. For parameters marked with an asterisk (*), there was a nonsignificant trend (P 5 0.119, P 5 0.102 and P 5 0.119) for ANOVA of mass(6), total mass, and the mass(6)/ [mass(4)1mass(6)] ratio, respectively.

known aspects, including the provisions for catabolism from tissue pools and incorporation of enterohepatic circulation. A strength of this study, relative to our preliminary model (Stites et al. 1997), is the determination of two routes of isotopic elimination (i.e., as urinary [2H2]folate and [2H2]ApABG). The use of two criteria, along with the fact that excretion of ApABG comprises a much more substantial fraction of folate turnover than does excretion of intact folate, reduces the uncertainty in modeling. Modeling would be further improved by including measurement of isotopic enrichment of plasma folate in any additional studies of this type. Measurement of isotopic enrichment of erythrocyte folate is feasible (Von der Porten et al. 1992) and would be of interest. However, data regarding erythrocyte labeling would be of little additional benefit in modeling because erythrocytes constitute a small fraction of total body folate. Also, erythrocyte folate would not parallel labeling of most other cellular folates because erythrocyte folate is deposited mainly during erythropoiesis, whereas the mean cell lifetime is at least 100 d. Although three in vivo pools are kinetically identifiable from our data, pools involving enterohepatic circulation have been included for physiologic relevance (Steinberg 1984), not kinetic identifiability at this stage. The delay included in enterohepatic circulation has little effect on the model. Very rapid aspects of folate absorption and distribution, as seen in a recent stable isotopic study of plasma kinetics (Rogers et al. 1997), have not been included because of the 24-h sampling times used in this protocol. No effort has been made to account for the metabolic interconversion and function of folates. Several modeling studies that are based on kinetic evaluation of folate metabolism and the effects of antifolates in cell culture or cell-free systems have been reported (Jackson and Harrap 1973, Seither et al. 1989, White 1979). However, the present modeling is directed primarily at examining wholebody turnover, and incorporation of such metabolic interconversions of folate coenzymes into the current model is not justified or necessary. The current model is a reasonable approximation of whole-body folate turnover and represents a starting point upon which to base additional modeling and simulation to examine other physiologic conditions (e.g., pregnancy).


In designing this study, it was originally intended to conduct modeling on the basis of urinary excretion of labeled total pABG; thus determination of ApABG concentration in urine by the method of McPartlin et al. (1992) was not conducted. Once HPLC and GCMS analyses were complete and modeling was in progress, the advantage of kinetic calculations based on excretion of [2H2]ApABG became apparent, but samples were no longer available for direct determination of urinary ApABG concentration. Conclusions of this study are reasonable on the basis of the assumed value for ApABG excretion. In the worst case, the assumed value is an overestimate of ApABG excretion, which would cause estimated rates of catabolism to be comparably overestimated. If that were the case, then fecal losses from tissue pools (Pools 4 and 6), which are already large (consistent with Krumdieck et al. 1978), would be even greater than currently seen in this model. The inclusion of a saturable pool as the major tissue folate compartment contributes to the relevance of the model. In rat liver, ;60% of cytosolic folate and 20% of mitochondrial folate are specifically bound to proteins (Zamierowski and Wagner 1977). These observations compare favorably to the model-based predictions regarding the fraction of free folate in tissue pools of this study (Table 4). The model-based prediction that the fraction of free folate in tissues increases with increasing folate intake is also consistent with the observations of Zamierowski and Wagner (1977) that free folate is depleted to a greater extent than protein-bound folate during folate deficiency. Previous studies in rats have shown that tissue folate mass and/or concentration increases in a nonlinear fashion with increasing folate intake ranging from deficient to optimal levels (Clifford et al. 1990, Keagy 1982). These observations also support the need for a major saturable pool in modeling.

FIGURE 5 Model-derived preduction of the relationship between mass of folate pools and folate intake at steady state for human subjects. Data points for total folate intakes of 454 and 907 nmol/d are model-derived estimates of folate pools in nonpregnant women in this study. Modeling simulations for intakes ,454 nmol/d and .907 nmol/d were based on rate constants determined from models of those subjects.



The total body mass of folate in adult humans has not been precisely determined, as discussed previously (Stites et al. 1977). Herbert (1987) estimated a total body folate mass of 7.5 6 2.5 mg (17 6 5.7 mmol). Hoppner and Lampi (1980) analyzed 560 human livers and reported a mean of 8.0 6 2.8 mg/g (18.1 6 6.3 nmol/g), and a similar range was reported by Whitehead (1973). Assuming that liver folate comprises half of total body folate and that human liver mass is 1400 g, then the last-mentioned two studies would suggest that total body folate would be approximately 22 mg or 50.8 mmol. The model provides slightly higher estimates (Table 4) for the nutritionally relevant folate intakes of this study. Even higher values reported previously (Stites et al. 1997) are probably overestimates because the model did not include a saturable pool. As stated above, a long-range goal of this modeling effort is to provide kinetically based information regarding the human requirement for folate. Attempts have been made to derive requirements for other vitamins on the basis of kinetic analysis, e.g., ascorbic acid, although determining a requirement on the basis of kinetic data remains problematic (Shane 1997, Young 1996). Several observations from this study, however, may provide information regarding optimal intake. Rate constants for urinary excretion of folate [k(3,1) and k(3,6)] increased substantially between intakes of 680 and 907 nmol/d (300 and 400 mg/d), consistent with elevated urinary excretion of labeled and nonlabeled folate at the 907 nmol/d intake level (Table 2). This may have been due to saturation of renal reabsorption mechanisms and possibly increased secretion of free renal folate as predicted by the increase in k(3,6). The existence of renal tubular secretion of folate has been reported previously and characterized recently (Morshed et al. 1997). These kinetic and analytical findings indicate greater loss of urinary folate between intakes of 680 and 907 nmol/d. A second line of inference involves the observation that plasma homocysteine was significantly greater in the 454 nmol/d intake group, with several subjects exhibiting concentrations of plasma homocysteine .15 mmol/L, which suggests marginal deficiency at that level of intake (O’Keefe et al. 1995). The model predicts a trend toward increases in the mass of free tissue folate (Pool 6) with increasing folate intake; the mass of the rapid turnover pool (Pool 1) also increased with increasing folate intake and apparently exceeded the mass of plasma folate. These observations may reflect an increase in the concentration of 5-methyl-H4folate available for remethylation of homocysteine at higher levels of folate intake. The elevation of plasma homocysteine at the 454 nmol/d intake (similar to the 1989 Recommended Dietary Allowance), whether or not related to this prediction of mass(1) and mass(6), is strong functional evidence that this level of intake is not sufficient for all subjects. The reductions in plasma and erythrocyte folate and increase in plasma homocysteine at the 454 nmol/d intake seen in this protocol (O’Keefe et al. 1995) were important considerations in the selection of a higher value for the Recommended Dietary Allowance for folate (Institute of Medicine 1998). To evaluate the association of the masses of in vivo folate pools with indicators of folate status, Pearson product-moment correlation analysis was performed (Table 5). Although the values for masses of Pools 1, 4 and 6 and total folate mass were not significantly associated with erythrocyte folate, the masses of Pools 1 and 6 and total folate were significantly correlated with serum folate. Further, the correlation of the mass of Pool 4 and serum folate approached significance (P 5 0.0825). In spite of the fact that tissue folates would participate in reactions governing plasma homocysteine concentration, only the mass of Pool 1 was significantly correlated with plasma homo-

TABLE 5 Evaluation of linear correlation between predicted masses of folate pools and criteria of folate nutritional status using Pearson product-moment correlation procedure1 Variable

Serum folate

Erythrocyte folate

Plasma homocysteine

r Mass(1) Mass(4) Mass(6) Mass(total)

0.507 (0.0378) 0.433 (0.0825) 0.504 (0.0391) 0.570 (0.0169)

0.405 (0.107) 0.322 (0.208) 0.150 (0.566) 0.170 (0.513)

20.607 (0.00983) 20.0893 (0.733) 20.236 (0.361) 20.250 (0.332)

1 Values shown are Pearson product-moment correlation coefficients, with P-values shown in parentheses. Coefficients with P , 0.05 were considered to indicate a significant linear relationship. Positive correlation coefficients indicate that both variables increase together, whereas negative values indicate an inverse relationship. Correlation coefficients for associations among folate pools were: mass(1) vs. mass(4), r 5 0.188 (P 5 0.470); mass(1) vs. mass(6), r 5 0.553 (P 5 0.0213); mass(1) vs. mass(total), r 5 0.570 (P 5 0.0169); mass(4) vs. mass(6), r 5 0.692 (P 5 0.00203); mass(4) vs. mass(total), r 5 0.707 (P 5 0.00152); mass(6) vs. mass(total), r 5 0.999 (P , 0.0001).

cysteine. We speculate that it is a fraction of the folate in Pool 6 that comprises the 5-methyl-H4folate actually involved in homocysteine remethylation. The results of this study highlight the important role of catabolism as a route of folate turnover. With the small sample size and variability of total pABG excretion, no significant differences were seen as a function of folate intake. However, the observed [2H2]ApABG excretion (Fig. 4) as well as steadystate estimates of catabolic flux [calculated as the respective rate constants, k(4,2) and k(6,2), for ApABG formation 3 mass of Pools 4 and 6] do indicate that ApABG excretion is influenced by folate intake. Similarly, in a study of supplementation with higher levels of folic acid in addition to a low folate diet (120 mg dietary folate plus folic acid to yield either 450 or 850 mg/d), we have shown that the daily excretion of ApABG and the minor nonacetylated form pABG does increase significantly with increasing folate intake (Caudill et al. 1998). Fecal excretion accounts for a substantial proportion of human folate turnover (Krumdieck et al. 1978). The model predicts that fecal losses from tissues (Pool 6) increase with increasing folate intake (i.e., not simply unabsorbed dietary folate). This conclusion is based on the substantially (although not significantly) greater values of k(10,6) at the higher two folate intakes. Unfortunately, direct determination and interpretation of fecal [2H2]folate, labeled catabolic products, or even total deuterium are not possible because of the confounding bacterial synthesis of folate and the high natural abundance of deuterium. Thus, one cannot calculate a requirement on the basis of isotopic balance. Also, one cannot use these kinetic results to estimate a minimally adequate pool mass or the quantity that must be replaced daily. Predictions of wholebody folate in this study indicated means of 64.5 6 2.3, 71.5 6 3.6 and 73.0 6 2.4 mmol for intakes of 454, 680 and 907 nmol/d, with fractional catabolic rates of 0.00474, 0.00607, and 0.00822/d, respectively. On the basis of the observations regarding homocysteine concentrations, one might infer that a


predicted whole-body folate mass of ;70 mmol is adequate to maintain this aspect of folate-dependent one-carbon metabolism. Studies involving more precise modeling and the use of additional functional indicators of folate status are required to resolve such issues. In this model, the compartments that would correspond primarily to metabolically active folates in tissues (i.e., Pools 4 and 6) are comprised of a number of chemical forms of the vitamin that exhibit similar turnover kinetics. As stated previously, Pool 1 may also consist of some metabolically active folates in tissues. Within each of these pools, folate molecules may undergo metabolic interconversions and transport between cytosol and mitochondria or other organelles. On the basis of the kinetic data and model of this study, one cannot interpret precisely the function of various folates in these pools. Mean residence times for whole-body folate were 212 6 8, 169 6 12 and 124 6 7 d for intakes of 454, 680 and 907 nmol/d, respectively. These findings illustrate the difficulty in designing and interpreting folate nutritional studies as a result of the very long time required to fully achieve a new steadystate level in response to a dietary change or supplementation. The 10-wk feeding periods used in this study would not have allowed the subjects to fully attain new steady-state levels with respect to folate pool masses, although this protocol was sufficient for substantial differences to be observed. Assuming the kinetic rule-of-thumb that five times t½ is required to achieve ;95% of the steady-state level (and t½ of whole body folate 5 0.693 3 mean residence time), we predict that a study of ;500 d would be required to be fully assured that subjects had achieved constant body pools if using a low folate intake (;454 nmol/d). It is interesting to note that the mean residence time decreases with increasing folate intake, which suggests that newly absorbed folate molecules compete for binding sites in tissues and thus accelerate turnover. In summary, the kinetic study and compartmental model reported here have extended our quantitative understanding of folate metabolism in humans. The kinetic data and model reported complement observations reported previously regarding nutritional status and folate intake (O’Keefe et al. 1995). Simulations based on this model permit a preliminary prediction of the effect of nutritional and physiologic conditions that are neither experimentally feasible nor practical. The results of this study give a baseline from which to compare additional kinetic studies to evaluate kinetic results of such conditions associated with altered folate requirements. LITERATURE CITED Anderson, J. H., Kerr, D. J., Setanoians, A., Cooke, T. G. & McArdle, C. S. (1992) A pharmocokinetic comparison of intravenous versus intra-arterial folinic acid. Br. J. Cancer 65: 133–135. Bhandari, S. D. & Gregory, J. F. (1992) Folic acid, 5-methyl-tetrahydrofolate, and 5-formyl-tetrahydrofolate exhibit equivalent intestinal absorption, metabolism, and in vivo kinetics in rats. J. Nutr. 122: 1847–1854. Blakley, R. L. (1969) The Biochemistry of Folic Acid and Related Pteridines. Wiley, New York, NY. Blount, B. C., Mack, M. M., Wehr, C. M., MacGregor, J. T., Hiatt, R. A., Wang, G., Wickramasinghe, S. N., Everson, R. B. & Ames, B. N. (1997) Folate deficiency causes uracil misincorporation into human DNA and chromosomal breakage: implications for cancer and neuronal damage. Proc. Natl. Acad. Sci. U.S.A. 94: 3290 –3295. Boushey, C. J., Beresford, S.A.A., Omenn, G. S., & Motulsky, A. G. (1995) A quantitative assessment of plasma homocysteine as a risk factor for vascular disease. Probable benefits of increasing folic acid intakes. J. Am. Med. Assoc. 274: 1049 –1057. Bunni, M. A., Rembiesa, B. M., Priest, D. G., Sahovic, E. & Stuart R. (1989) Accumulation of tetrahydrofolates in human plasma after leucovorin administration. Cancer Chemother. Pharmacol. 23: 353–357. Caudill, M. A., Gregory, J. F., Hutson, A. C. & Bailey, L. B. (1998) Folate catabolism in pregnant and nonpregnant women consuming controlled folate intakes. J. Nutr. 128: 204 –208.


Foster, D. M., Barrett, P.H.R., Bell, B. M., Beltz, W. F., Cobelli, C., Golde, H., Jacquez, J. A. & Phair, R. D. (1994) SAAM II: simulation, analysis and modeling software. BMES Bull. 18: 19 –21. Glantz, S. A. (1992) Primer of Biostatistics, 3rd ed. McGraw-Hill, Inc., New York, NY. Gregory, J. F. & Scott, K. C. (1996) Modeling of folate metabolism. In: Advances in Food and Nutrition Research (Coburn, S. P. & Townsend, D. W., eds.), vol. 40, pp. 81–93. Academic Press, New York, NY. Gregory, J. F. & Toth, J. P. (1988) Chemical synthesis of deuterated monoglutamate and in vivo assessment of urinary excretion of deuterated folates in man. Anal. Biochem. 170: 94 –104. Gregory, J. F., Williamson, J., Bailey, L.B. & Toth, J. P. (1998) Urinary excretion of [2H4]folate by nonpregnant women following a single oral dose of [2H4]folic acid is a functional index of folate nutritional status. J. Nutr. 128: 1907–1912. Guyton, A. C. (1971) Textbook of Medical Physiology, 4th ed., p. 762. W. B. Saunders Co., Philadelphia, PA. Herbert, V. (1987) Recommended dietary intakes (RDI) of folate in humans. Am. J. Clin. Nutr. 45: 661– 670. Hoppner, K. & Lampi, B. (1980) Folate levels in human liver from autopsies in Canada. Am. J. Clin. Nutr. 33: 862– 864. Institute of Medicine (1998) Dietary Reference Intakes: Thiamin, Riboflavin, Niacin, Vitamin B6, Folate, Vitamin B12, Pantothenic Acid, Biotin, and Choline. Standing Committee on the Scientific Evaluation of Dietary Reference Intakes, Institute of Medicine, National Academy of Sciences. National Academy Press, Washington, DC. Jackson, R. C. & Harrap, K. R. (1973) Studies with a mathematical model of folate metabolism. Arch. Biochem. Biophys. 158: 827– 841. Keagy, P. M. (1982) Integrated saturation model for tissue micronutrient concentrations applied to liver folacin. J. Nutr. 112: 377–386. Krumdieck, C. L., Fukushima, K., Fukushima, T., Shiota, T. & Butterworth, C. E., Jr. (1978) A long term study of the excretion of folate and pterins in a human subject after ingestion of 14C folic acid, with observations on the effect of diphenylhydantoin administration. Am. J. Clin. Nutr. 31: 88 –93. Lakshmaiah, N. & Bamji, M. S. (1981) Half-life and metabolism of 3H folic acid in oral contraceptive treated rats. Horm. Metab. Res. 13: 404 – 407. Lavoie, A. & Cooper, B. A. (1974) Rapid transfer of folic acid from blood to bile in man, and its conversion into folate coenzymes and into a pteroylglutamate with little biological activity. Clin. Sci. Mol. Med. 46: 729 –741. Loew, D., Eberhardt, A., Heseker, H. & Ku¨bler, W. (1987) Zur plasmakinetik und elimination von folsa¨ure. Klin. Wochenschr. 65: 520 –524. Mason, J. (1995) Folate status: effects on carcinogenesis. In: Folate in Health and Disease (Bailey, L. B., ed.), pp. 329-361. M. Dekker, Inc., New York, NY. McPartlin, J. M., Courtney, G., McNulty, H. Weir, D. G. & Scott, J. M. (1992) The quantitative analysis of endogenous folate catabolites in human urine. Anal. Biochem. 206: 256 –261. McPartlin, J., Halligan, A., Scott, J. M., Darling, M. & Weir, D. G. (1993) Accelerated folate breakdown in pregnancy. Lancet 341: 148 –149. Menke, G., Mu¨ller, J., Weimann, H.-J. & Schuster, O. (1993) Bioverfu¨gbarkeit von folsa¨ure nach intramuskula¨rer und intraveno¨ser gabe an gesunde versuchspersonen. Arzneim.-Forsch. 43: 22–27. Minchin, R. F. (1995) Acetylation of p-aminobenzoylglutamate, a folic acid catabolite, by recombinant human arylamine N-acetyltransferase and U937 cells. Biochem. J. 307: 1–3. Morrison, H. I., Schaubel, D., Desmeules, M. & Wigle, D. T. (1996) Serum folate and risk of fatal coronary heart disease. J. Am. Med. Assoc. 275: 1893–1896. Morshed, K. M., Ross, D. M. & McMartin, K. E. (1997) Folate transport protein mediates the bidirectional transport of 5-methyltetrahydrofolate in cultured human proximal tubule cells. J. Nutr. 127: 1137–1147. Murphy, M., Keating, M., Boyle, P., Weir, D. G. & Scott, J. M. (1976) The elucidation of the mechanism of folate catabolism in the rat. Biochem. Biophys. Res. Commun. 71: 1017–1024. Murphy, M. & Scott, J. M. (1979) The turnover, catabolism and excretion of folate administered at physiological concentrations in the rat. Biochim. Biophys. Acta 583: 535–539. O’Keefe, C. A., Bailey, L. B., Thomas, E. A., Hofler, S. A., Davis, B. A., Cerda, J. J. & Gregory, J. F. (1995) Controlled dietary folate affects folate status in nonpregnant women. J. Nutr. 125: 2717–2725. Pfeiffer, C. M. & Gregory, J. F. (1997) Preparation of stable-isotopically labeled folates for in vivo investigation of folate absorption and metabolism. Methods Enzymol. 281: 106 –116. Pfeiffer, C. M., Rogers, L. M., Bailey, L. B. & Gregory, J. F. (1997) Absorption of folate from fortified cereal-grain products and of supplemental folate consumed with or without food determined using a dual-label stable-isotope protocol. Am. J. Clin. Nutr. 66: 1388 –1397. Pogribny, I. P., Muskhelishvilli, L., Miller, B. J. & James, S. J. (1997) Presence and consequence of uracil in preneoplastic DNA from folate/methyl-deficient rats. Carcinogenesis 18: 2071–2076. Priest, D. G., Schmitz, J. C., Bunni, M. A. & Stuart, R. K. (1991) Pharmacokinetics of leucovorin metabolites in human plasma as a function of dose administered orally and intravenously. J. Natl. Cancer Inst. 83: 1806 –1812. Rimm, E. B., Willett, W. C., Hu, F. B., Sampson, L., Colditz, G. A., Manson, J. E., Hennekens, C. & Stampfer, M. J. (1998) Folate and vitamin B6 from diet and supplements in relation to risk of coronary heart disease among women. J. Am. Med. Assoc. 279: 359 –364. Rogers, L. M., Pfeiffer, C. M., Bailey, L. B. & Gregory, J. F. (1997) A dual-label



stable-isotopic protocol is suitable for determination of folate bioavailability: evaluation of urinary excretion and plasma folate kinetics of intravenous and oral doses of [13C5] and [2H2]folic acid. J. Nutr. 127: 2321–2327. Russell, R. M., Rosenberg, I. H., Wilson, P. D., Iber, F. L., Oaks, E. B., Giovetti, A. C., Otradoved, M. A., Karwaski, P. A. & Press, A. W. (1983) Increased urinary excretion and prolonged turnover time of folic acid during ethanol ingestion. Am. J. Clin. Nutr. 38: 64 –70. Sauberlich, H. E., Kretsch, M. J., Skala, J. H., Johnson, H. L. & Taylor, P. C. (1987) Folate requirement and metabolism in nonpregnant women. Am. J. Clin. Nutr. 46: 1016 –1028. Scott, K. C. & Gregory, J. F. (1996) The fate of [3H]folic acid in folate adequate rats. J. Nutr. Biochem. 7: 261–269. Scott, J. M., Weir, D. G. & Kirke, P. N. (1995) Folate and neural tube defects. In: Folate in Health and Disease (Bailey LB, ed.), pp. 329-360. M. Dekker, Inc., New York, NY. Seither, R. L., Trent, D. F., Mikulecky, D. C., Rape, T. J., & Goldman, I. D. (1989) Folate-pool interconversions and interconversions of biosynthetic processes after exposure of 1210 leukemia cells to antifolates. Experimental and network thermodynamic analyses of the role of dihydrofolate polyglutamates in antifolate action in cells. J. Biol. Chem. 264: 17016 –17023. Selhub, J., Jacques, P. F., Wilson, P. W., Rush, D. & Rosenberg, I. H. (1993) Vitamin status and intake as primary determinants of homocysteinemia in an elderly population. J. Am. Med. Assoc. 270: 2693–2698. Shane, B. (1997) Vitamin C pharmacokinetics: it’s de´ja` vu all over again. Am. J. Clin. Nutr. 66: 1061–1062. Steinberg, S. E. (1984) Mechanisms of folate homeostasis. Am. J. Physiol. 246: G319 –G324. Stites, T. E., Bailey, L. B., Scott, K. C., Toth, J. P., Fisher, W. P. & Gregory, J. F. (1997) Kinetic modeling of folate metabolism using chronic administration of deuterium-labeled folic acid in adult men. Am. J. Clin. Nutr. 65: 53– 60.

Tamura, T. (1990) Microbiological assay of folates. In: Folic Acid Metabolism in Health and Disease (Picciano, M. F., Stokstad, E.L.R. & Gregory, J. F., eds.), pp. 121-137. John Wiley & Sons, Inc., New York, NY. Tamura, T. & Halsted, C. H. (1983) Folate turnover in chronically alcoholic monkeys. J. Lab. Clin. Med. 101: 623– 628. Vester, B. & Rasmussen, K. (1991) High performance liquid chromatography method for rapid and accurate determination of homocysteine in plasma and serum. Eur. J. Clin. Chem. Clin. Biochem. 29: 549 –554. Von der Porten, A., Gregory, J. F., Toth, J. P., Cerda, J. J., Curry, S. H. & Bailey, L. B. (1992) In vivo folate kinetics during chronic supplementation of human subjects with deuterium-labeled folic acid. J. Nutr. 122: 1293–1299. Wagner, C. (1995) Biochemical role of folate in cellular metabolism. In: Folate in Health and Disease (Bailey, L. B., ed.), pp. 23– 42. M. Dekker, Inc., New York, NY. Wang, C., Song, S., Bailey, L. B. & Gregory, J. F. (1994) Relationship between urinary excretion of p-aminobenzoylglutamate and folate status of growing rats. Nutr. Res. 14: 875– 884. White, J. C. (1979) Reversal of methotrexate binding to dihydrofolate reductase by dihydrofolate. Studies with pure enzyme and computer modeling using network thermodynamics. J. Biol. Chem. 254: 10889 –10895. Whitehead, V. M. (1973) Polyglutamyl metabolites of folic acid in human liver. Lancet 1: 743–745. Young, V. R. (1996) Evidence for a recommended dietary allowance for vitamin C from pharmacokinetics: a comment and analysis. Proc. Nat. Acad. Sci. U.S.A. 93: 14344 –14348. Zamierowski, M. M. & Wagner, C. (1977) Effect of folacin deficiency on folacin-binding proteins in the rat. J. Nutr. 107: 1937–1945.