点击显示 收起
1 From the Interdepartmental Graduate Program in Nutritional Sciences, University of Wisconsin–Madison, Madison, WI.
2 Presented in part orally at the 2002 and 2004 Experimental Biology meetings and as a poster at the 2003 International Vitamin A Consultative Group meeting. 3 Supported by Hatch-Wisconsin Agricultural Experiment Station no. WIS04533 NIHNIDDK grant no. 61973, USDA NRI grant no. 2003-35200-13754, and the University of Wisconsin Graduate School. 4 Address reprint requests to SA Tanumihardjo, Department of Nutritional Sciences, 1415 Linden Drive, Madison, WI 53706. E-mail: sherry{at}nutrisci.wisc.edu.
ABSTRACT
Background: The effects of large doses of preformed vitamin A, such as those provided in supplementation programs for the prevention of deficiency, on total serum vitamin A have been inadequately investigated.
Objective: This study characterized the time course of serum vitamin A metabolites in lactating sows after single high doses of retinyl ester.
Design: Lactating sows were fitted with jugular catheters and subsequently fed either 1.05 or 2.1 mmol retinyl ester (n = 6/group) or a corn oil vehicle (n = 3). Blood was collected at baseline and at intervals to 48 h and analyzed by gradient HPLC for retinol, retinyl esters, and metabolites.
Results: The mean (±SD) total serum vitamin A concentration peaked at 1 h (3.69 ± 4.0 µmol/L) and 2 h (7.70 ± 6.8 µmol/L) in the low- and high-dose groups, respectively (P < 0.05). Retinyl esters accounted for most of the serum vitamin A in both groups at peak time points. Mean serum retinol concentrations changed little and accounted for most of the serum vitamin A at baseline (94% and 97% for the low- and high-dose groups, respectively) but for only 22% and 14% at peak times for the low- and high-dose groups, respectively. Postdosage increases were observed for total vitamin A and retinyl esters, 4-oxoretinol, retinoyl ß-glucuronide, and retinyl ß-glucuronide but not for retinoic acid.
Conclusions: Serum retinol concentration remains relatively static after a large dose of preformed vitamin A and therefore is not an appropriate measure of intervention efficacy. The increases in ß-glucuronide metabolites and 4-oxoretinol suggest a preventive role against a rise in retinoic acid and retinol.
Key Words: Preformed vitamin A • retinyl ester • sows • time course • toxicity • vitamin A supplementation
INTRODUCTION
Administration of high doses of preformed vitamin A as retinyl esters has been a cornerstone of programs designed to prevent and address vitamin A deficiency and its related sequelae (1). Supplemental vitamin A as retinyl ester (200 000 IU; 210 µmol) is routinely provided to children aged 12 mo and to postpartum women, and 400 000 IU (420 µmol) has been provided in limited situations to postpartum women (2, 3). Mainly targeting lactating women, infants, and children in developing nations, supplementation programs have resulted in decreased morbidity and mortality related to vitamin A deficiency (1, 4-6). However, preformed vitamin A is toxic in high amounts, including those provided in supplementation programs. The US recommended dietary allowance (RDA) for vitamin A is 700 and 900 µg/d for adult women and men, respectively. The RDA for lactating women is set at 1300 µg/d, whereas the RDA for children is lower because of their lower body weight. The tolerable upper intake limit for vitamin A is 3000 µg/d (7); however, a chronic intake one-half that amount has been linked to a greater risk of hip fracture (8-10). Moreover, vitamin A supplementation (250 µg/kg) increases plasma retinoic acid concentrations in young, well-nourished men (11).
We previously showed that hepatic vitamin A in lactating sows, whose vitamin A status was in the low-normal range as compared with human standards, is readily accrued in a dose-dependent fashion after a large dose of preformed retinyl ester and that milk vitamin A remains significantly elevated for only 48 h after the dose (12). We concluded that smaller, divided doses of vitamin A or vitamin A–enriched diets (or both) that result in sustained chylomicron delivery of vitamin A to mammary tissue may be safer for the recipients and more effective in preventing vitamin A deficiency in the nursing infants than are larger, single doses.
Whereas retinoic acid is widely regarded as the form of vitamin A responsible for vitamin A's toxic effects, not much is known about its appearance in serum after the ingestion of high doses of preformed vitamin A or about the appearance of certain derivative, less toxic vitamin A metabolites in serum. These metabolites include, but are not limited to, 4-oxoretinoic acid, 4-oxoretinol, retinoyl ß-glucuronide (RAG), and retinyl ß-glucuronide (ROG). Most HPLC methods currently in use do not assess the full profile of vitamin A metabolites but focus instead solely on serum retinol or total serum esters.
Given these concerns, we concluded that the effects of large doses of preformed vitamin A on serum vitamin A metabolites have not been adequately investigated. We adapted a previously developed method (13) for quantifying an array of vitamin A metabolites from the water-soluble glucuronides to the fat-soluble esters, and we examined its precision as applied to biological samples. We hypothesized that total serum vitamin A would show an increase in retinyl esters and in other fat- and water-soluble vitamin A metabolites in lactating sows after a single high dose of retinyl ester, but that the increases may not be dose dependent for all metabolites. The doses chosen were similar to those given to lactating women in developing countries on the basis of body weight.
MATERIALS AND METHODS
Animals and diet
Animals were housed at the University of Wisconsin (UW)–Madison Swine Research and Teaching Center in Arlington, WI, and the UW–Madison Livestock Laboratory in Madison, WI. The sows were a mix of purebreds and crossbreeds, including Large White, Duroc, and Landrace. We selected apparently healthy animals that had been lactating for 7–14 d to ensure postpartum recovery and ample milk collection for the parallel study (12). Baseline characteristics and reproductive histories were obtained (Table 1). Before the experiment, center staffers fed the pigs ad libitum the standard Swine Research and Teaching Center lactation diet, which contained 5500 IU (1667 µg) retinyl acetate/kg feed. The complete nutrient composition of the diet has been published (12). The vitamin A status of the sows was in the low-normal range, compared with human standards, with hepatic vitamin A concentration in the livers of 2 control sows being 0.21 and 0.36 µmol/g liver at baseline, as previously reported in detail (12).
View this table:
TABLE 1. Baseline characteristics of lactating sows that were given either 1.05 mmol (low-dose) or 2.1 mmol (high-dose) retinyl acetate1
Jugular catheters were placed into nonanesthetized lactating sows (n = 15). The insertion into the jugular vein was made with a sterile 12-gauge needle. The catheter tubing, which was sterilized with the use of a gas autoclave before use at the UW Hospital and Clinics instrument-reprocessing unit (Madison, WI), was 5 French polyurethane tubing with an outer diameter of 1.7 mm (Access Technologies, Skokie, IL). We also used an 18-gauge stub adapter and a 0.10-mL intermittent infusion plug. Skin Bond adhesive (Owens & Minor, Waunakee, WI) was used to glue elastic tape to the skin. The tape covered the exposed tubing and usually prevented it from being pulled out. After catheterization, animals were rested and observed. The tubing was flushed periodically with sterile, heparinized saline to prevent clotting. Approximately 12 h before dosing, the animals' food was restricted. Doses of 1.05 (low-dose; LD) or 2.1 (high-dose; HD) mmol retinyl acetate, dissolved in 4.5 mL corn oil, were provided to sows in each treatment group (n = 6) by mixing the oily solution with 500 g dry feed immediately before feeding. The doses averaged slightly more vitamin A/kg body weight (4.7 and 9.4 µmol/kg, respectively) than is provided to lactating women according to International Vitamin A Consultative Group recommendations (14). This difference was meant to account for losses during feeding, because some food was invariably slopped out of the feed trough during the meal. Each animal was observed to ascertain whether the entire meal, and thus the entire dose, was consumed. Three animals were provided corn oil in their food as controls. Approval for the experiment was obtained from the University of Wisconsin (UW)–Madison Research Animal Resources Center.
Sample collection
Blood samples (10 mL) were collected via the catheter into evacuated tubes containing clot activator, kept out of direct light, and centrifuged to separate serum. Blood was collected at baseline and 1, 2, 4, 8, 16, and 24 h after the dose. Samples were also collected at 48 h from dosed animals (n = 3 from the low-dose group and n = 2 from the high-dose group) whose catheters had remained usable. All samples were protected from light and placed on dry ice in a closed container until taken within 24 h to the laboratory, where they were frozen at –80 °C until analysis.
Analysis of serum
All analyses and retinoid syntheses were conducted under yellow lights. Serum (1.0 mL) was analyzed for polar vitamin A metabolites (ie, 4-oxoretinoic acid, 4-oxoretinol, RAG, ROG, and retinoic acid), retinol, and retinyl esters by using a procedure published by Barua (13) and adapted to optimize conditions in our laboratory (15). A gradient HPLC system that included a Waters 600E multisolvent delivery system (Milford, MA) was used. The stationary phase was a Phenomenex Phenosphere 5-µm, reverse-phase, 150 x 4.6–mm C-18 ODS column (Torrance, CA) equipped with a guard column to protect the analytic column from particulate matter. The detector was a dual-wavelength absorbance detector (model 2487; Waters), which was set at 335 nm as a midpoint between the lambda maxima of retinol, ROG, and the retinyl esters (325 nm) and that of 4-oxoretinoic acid, retinoic acid, 3,4-didehydroretinyl acetate (internal standard), and RAG (ie, 347–360 nm) (16). The flow rate was set at 1.0 mL/min. The solvent system consisted of a linear gradient of 70:30 methanol:water (by vol) (10 mmol/L ammonium acetate) to 80:20 methanol:dichloroethane (by vol) (10 mmol/L ammonium acetate) over 20 min, which then ran isocratically for 10 min. After this period, the gradient was reversed to initial conditions in 5 min followed by a 10-min equilibration time before the next injection. A Shimadzu C-R7A Chromatopac data processor (Kyoto, Japan) recorded and calculated peak areas.
3, 4-Didehydroretinyl acetate (0.25 nmol), synthesized in our lab according to published procedures (17, 18), was added to each sample as an internal standard. Extraction efficiency, estimated by recovery of internal standard, was 88 ± 4.3%. The within-sample CV for this serum extraction method was calculated to be 6.4% for total vitamin A, which includes retinol, retinyl esters, and metabolites. Compounds were identified on the basis of retention times, coelution of known standards prepared in the lab (RAG was donated by AB Barua, Iowa State University), and ultraviolet spectra obtained on a subset of the serum from analysis on a Waters 996 photodiode array HPLC system equipped with a solvent delivery system (model 600; Milford, MA), which scanned from 210 to 550 nm. The same gradient HPLC solvent mixtures as described above were used.
Synthesis of retinyl ester standards
Six retinyl esters (butyrate, laurate, myristate, oleate, palmitate, and stearate) were synthesized as standards to ascertain HPLC retention times and confirm retinoid identity. Esters were synthesized by a condensation reaction of retinol with the individual fatty acid anhydrides (Sigma Chemical, St Louis, MO) in triethylamine by modifying a procedure used for the acetate ester (19). Thin-layer chromatography was used to monitor reaction progression and purification before injection onto the HPLC.
Statistical analysis
Statistical consultation was provided by the UW–Madison College of Agriculture and Life Sciences Statistical Consulting Service. For the serum, baseline values were subtracted for each sow's measurement value, and the differences (plus a small constant) were log transformed to stabilize the variance. A repeated-measures analysis of variance with fixed effects was applied by using SAS PROC MIXED software (version 8; SAS Institute, Cary, NC) to ascertain the main effects of treatment and time. To account for correlations among measurements within each sow over time, a spatial power term was used. The least-squares mean (LSM) was calculated, and the overall differences between treatment groups were ascertained by using the Tukey-Kramer adjustment. The reported treatment-by-time interaction was ascertained by testing for departure from parallel structure by removing the treatment effect from PROC MIXED, in which the postbaseline values were used. Significance at specific time points was ascertained by using LSM differences with Bonferroni correction for comparison between the 3 treatment groups. Results were considered significant at a P value < 0.05.
RESULTS
Total serum vitamin A
One sow in the HD group did not consume the dose in its entirety. Therefore, all data reported hereafter refer to 6 sows in the LD group, 5 sows in the HD group, and 3 controls. The effect on total serum vitamin A of dosing with retinyl acetate was pronounced and dose dependent. A highly significant treatment-by-time interaction occurred (P < 0.0001). Total serum vitamin A peaked at 3.7 ± 4.0 µmol/L in the LD group at 1 h, which was a significant increase from 0.77 ± 0.20 µmol/L at baseline (P = 0.042 compared with control subjects at 1 h). In the HD group, the peak occurred at 7.7 ± 6.8 µmol/L at 2 h, which was a significant increase from 0.76 ± 0.16 µmol/L at baseline (P < 0.001 compared with control subjects at 2 h). Total serum vitamin A returned to baseline by 48 h, and significant effects of the dosage disappeared after 8 h between treatments (Table 2). Confirming our hypothesis that total serum vitamin A would be significantly elevated after the dose, there was a significant effect of vitamin A dose on total serum vitamin A throughout the time course (P < 0.0001 for effect of vitamin A treatment). When we focused in on differences among treatment groups at specific time points by using LSM differences, an overall effect of vitamin A treatment was observed at 1, 2, 4, and 8 h after dose (P < 0.05). Statistical analysis using differences of LSM with the Tukey-Kramer adjustment also confirmed an overall difference between the control and LD (P = 0.0007), control and HD (P < 0.0001), and LD and HD (P = 0.0027) groups in the response of total circulating vitamin A to the dose with retinyl acetate.
View this table:
TABLE 2. Comparison by time of serum total vitamin A concentrations in lactating sows dosed with 1.05 mmol (low-dose) or 2.1 mmol (high-dose) retinyl acetate and in control sows receiving corn oil1
Serum retinol concentrations
The time course of serum retinol and retinyl ester concentrations was plotted by treatment group (Figure 1). A trend for a significant treatment-by-time interaction overall for serum retinol concentrations (P = 0.054) and a significant effect of time independent of treatment (P = 0.0069) were noted. They suggest either a diurnal pattern, a postprandial effect on serum retinol concentrations in these lactating sows, or both.
FIGURE 1.. Time course of mean (±SEM) serum retinol, circulating long-chain retinyl esters (palmitate + oleate and stearate), and circulating minor retinyl ester concentrations in lactating sows who were fed corn oil (control group; n = 3), 1.05 mmol retinyl acetate [low-dose (LD) group; n = 6], or 2.1 mmol retinyl acetate [high-dose (HD) group; n = 5]. The error bars are smaller than the points in many instances. Treatment as a function of time was significant for the long-chain esters (P < 0.0001) and for the sum of minor esters (P = 0.0095), and it was nearly significant for serum retinol concentration (P = 0.054). There was a main effect of retinyl acetate treatment on the long-chain esters (P < 0.0001) and the sum of minor esters (P = 0.016; repeated-measures ANOVA with fixed effects), and there were significant differences between all treatment groups (P < 0.05; least-squares mean differences with Tukey-Kramer adjustment for the long-chain esters). *Significantly different from control sows at specific time points, P < 0.05 (least-squares mean differences with Bonferroni correction for comparison between groups within a panel). #Significant difference between dosage groups. Note that the y axes are not the same in the 3 panels.
In contrast to total serum vitamin A, however, which was strongly affected from 1 through 8 h by dosing with retinyl acetate, the serum retinol concentration was not affected by the treatment with retinyl ester overall (P > 0.05). As a percentage of total circulating vitamin A, serum retinol accounted for most (
At baseline, total serum esters for the LD and HD groups contributed 3.6% and 2.0% of total vitamin A. Total serum esters represented 78% and 85% of total serum vitamin A in the LD and HD groups, respectively, at peak time points and fell to 4.9% and 6.1%, respectively, at 48 h. Mean ester contribution to total serum vitamin A for controls was 6.0 ± 2.2% throughout the course of the study.
Retinyl palmitate + oleate, which coeluted in this HPLC system, and retinyl stearate, hereafter referred to collectively as long-chain esters, were significantly affected by treatment-by-time interaction (P < 0.0001), treatment (P < 0.0001), and time (P = 0.0002). This indicates a dynamic system of rising and falling retinyl esters after dosing. The long-chain esters accounted for most of the serum vitamin A in both dose groups at peak time points, representing 70% of total serum vitamin A in the LD group at 1 h and 65% of total serum vitamin A in the HD group at 2 h. Figure 1 illustrates the rise in the concentration of long-chain esters after the dose. The long-chain ester concentrations differed significantly between the LD and HD groups (P < 0.05). The effect of the dose on long-chain esters was evident at specific time points through 8 h, at which time differences between controls and HD and between LD and HD groups were observed (P < 0.05).
Other retinyl esters identified, in relative declining concentrations, were retinyl linoleate, retinyl heptadecanoate, retinyl myristate or retinyl palmitoleate (or both; separation and positive identification was not achieved in this system), retinyl laurate, and an unidentified retinyl ester, which may be retinyl linolenate or retinyl pentadecanoate, according to previous reports (20, 21); these are all herein referred to collectively as minor esters. As was seen with the long-chain esters, a significant treatment-by-time interaction occurred (P = 0.0095). Overall, effects of the vitamin A dose were significant or approached significance for each of the minor esters and for the sum of the minor esters. Collectively, minor retinyl esters were significantly affected by treatment (P = 0.016) and time (P = 0.011) and showed kinetic patterns similar to those of the long-chain esters (Figure 1). There were significant differences in minor esters between the HD and LD groups at 2 h. The sum of the minor ester concentrations differed significantly between the HD group and the controls at 2 and 4 h (P < 0.05). The long-chain esters circulated for a significantly longer time after dose than did the minor esters, and there were significant effects for the LD group to 4 h and for the HD group through 8 h (P < 0.05). Thus, retinyl esters cleared slightly faster in the LD group than in the HD group (Figure 1).
Retinyl acetate concentrations (Figure 2), assumed to be derived from the dose or diet, were significantly affected by treatment (P = 0.0025), and there was a trend for an effect of time (P = 0.087), but the interaction was not significant (P = 0.12). Retinyl acetate concentrations peaked at 1 h (0.065 ± 0.066 µmol/L) and 2 h (0.053 ± 0.039 µmol/L) after dose in the LD and HD groups, respectively, and both concentrations were significantly higher than the 0.004 µmol/L concentration for the controls found by using LSM differences with the Tukey-Kramer adjustment (P < 0.0081). Serum retinyl acetate concentrations quickly dropped after 2 h, which further suggests that the retinyl acetate was derived from the dose and associated with chylomicra.
FIGURE 2.. Time course of mean (±SEM) serum retinyl acetate concentrations in lactating sows fed corn oil (control group; n = 3), 1.05 mmol retinyl acetate [low-dose (LD) group; n = 6], or 2.1 mmol retinyl acetate [high-dose (HD) group; n = 5]. The error bars are smaller than the points in some instances. Retinyl acetate concentrations were significantly affected by treatment (P = 0.0025), and there was a trend for an effect of time (P = 0.087), but treatment as a function of time was not significant (P = 0.12; all: repeated-measures ANOVA with fixed effects). The treatment groups were significantly different from the control group, P < 0.0081 (least-squares mean differences with Tukey-Kramer adjustment).
Serum retinoic acid, ß-glucuronides, and 4-oxo derivatives
Whereas we anticipated an increase in serum retinoic acid concentrations after the dose with retinyl ester, we did not observe a significantly greater effect of treatment in either dose group than in the control group. Moreover, retinoic acid concentrations did not differ significantly between the LD and HD groups at any time point (Figure 3).
FIGURE 3.. Time course of mean (±SEM) serum retinoic acid (RA) and retinoyl ß-glucuronide (RAG) concentrations in lactating sows fed corn oil (control group; n = 3), 1.05 mmol retinyl acetate [low-dose (LD) group; n = 6], or 2.1 mmol retinyl acetate [high-dose (HD) group; n = 5]. The error bars are smaller than the points in some instances. There was a significant main effect of retinyl acetate treatment on RAG concentrations (P = 0.033; repeated-measures ANOVA with fixed effects) but not on RA concentrations. Treatment as a function of time was not significant for either RA or RAG (P > 0.4). The HD group was significantly different from the control group (P = 0.026; least-squares mean differences with Tukey-Kramer adjustment). Note that the y axes are not the same in the 2 panels.
RAG and ROG, which are formed from retinoic acid and retinol, respectively, rose transiently in both dose groups after supplementation. A significant overall effect of the vitamin A dose on RAG was noted (P = 0.033; Figure 3), and the HD group was significantly different from the control group with the use of LSM differences with the Tukey-Kramer adjustment (P = 0.026). A significant treatment-by-time interaction was noted for ROG concentrations (P = 0.017), as was a significant overall effect of vitamin A dose (P = 0.0028) and time (P = 0.033) (Figure 4). Comparisons between groups showed overall differences between the control and HD groups or between the HD and LD groups by using LSM differences with the Tukey-Kramer adjustment (P < 0.014). ROG peaked at 2 h in the HD group and at 4 h in the LD group and then declined to baseline concentrations at 48 h for both groups. The effect of vitamin A treatment was significantly different between the 2 dose groups at 2 h (P < 0.05).
FIGURE 4.. Time course of mean (±SEM) serum retinyl ß-glucuronide (ROG) concentrations in lactating sows fed corn oil (control group; n = 3), 1.05 mmol retinyl acetate [low-dose (LD) group; n = 6], or 2.1 mmol retinyl acetate [high-dose (HD) group; n = 5]. The error bars are smaller than the points in some instances. A significant treatment x time interaction for ROG concentrations (P = 0.017) and a main effect of retinyl acetate treatment (P = 0.0028) and time (P = 0.033) were noted by using a repeated-measures ANOVA with fixed effects. The HD group was significantly different from the control group (P = 0.0042) and the LD group (P = 0.014) with the use of least-squares mean differences with Tukey-Kramer adjustment. *Significantly different from control sows at specific time points, P < 0.05 (least-squares mean differences with Bonferroni correction). #Significant difference between dose groups, P < 0.05 (least-squares mean differences with Bonferroni correction).
A significant treatment-by-time interaction was noted for the retinol catabolite 4-oxoretinol (P = 0.029). 4-Oxoretinol was increased by treatment; the overall effect approached significance (P = 0.061) and was driven by treatment with the higher dose of vitamin A. There were differences between the HD and LD groups and between the HD and control groups at 4 h (P < 0.0001). The values at 4 h were 0.60 ± 1.1, 1.1 ± 1.2, and 7.3 ± 16.3 nmol/L for the control, LD, and HD groups, respectively. 4-Oxoretinoic acid was detected at only one-third of all time points, and therefore it resulted in no significant treatment effects.
DISCUSSION
Before this study, we proposed that the average nursing infant may not benefit more from a 400 000 IU supplement to the mother than from a 200 000 IU dose given as a single bolus (12). This proposal was validated by determining liver reserves by using the lactating sow-nursing piglet model (22). In the current study, we evaluated the effects on serum vitamin A metabolites of a high dose of retinyl ester. The doses provided—ie, 1.05 and 2.1 mmol retinyl acetate—were calculated to emulate the 200 000 and 400 000 IU vitamin A supplements provided to lactating women in developing countries on the basis of body weight (4 and 8 µmol/kg body weight). The pig is a good model from which to extrapolate to humans because of the anatomic, physiologic, and morphologic similarity of the two species' gastrointestinal tracts. The mean porcine serum vitamin A concentrations of 0.73 µmol/L reported (23) are similar to reported values for humans (24, 25) and slightly higher than a concentration of 0.70 µmol/L, which indicated marginal liver vitamin A stores in lactating Indonesian women (26, 27). In addition, the vitamin A content of the adult pig liver—a reported range of 0.20–1.3 µmol/g—is very close to that of the adult human liver and is considered adequate for adult humans (23, 24, 28-30). The data presented here show that total serum vitamin A rose to its peak within 2 h of the dose, dropped significantly after 8 h, and returned to baseline concentrations by 48 h in both treatment groups. Overall differences in total serum vitamin A and most vitamin A metabolites were significant between treatment groups; retinol and retinoic acid were the major exceptions.
Retinol concentrations remained fairly static throughout the time course and were not affected by treatment. The increase in serum retinyl esters was dose dependent. Whereas the long-chain esters made up most of the circulating esters at all time points in both groups, an increase in other esters was also observed, and it showed a similar kinetic pattern. It is interesting that, even in control sows at baseline, significant circulating concentrations of retinyl acetate were observed, and they were confirmed via coelution with retinyl acetate standard and spectra obtained from photodiode array HPLC. This may be explained by the fact that retinyl acetate is the dietary form of vitamin A supplied in the sows' diet. As a short-chain ester, retinyl acetate may be absorbed whole in the gut and may enter the circulation intact in situations of high dietary intake. This finding concurs with previously obtained data, which showed appreciable circulating retinyl acetate concentrations in monkeys whose diets were extremely high in retinyl acetate (15) and in American women who used supplements (KL Penniston, SA Tanimihardjo, unpublished observations, 2004). In the current study, we showed that treatment had an overall effect on retinyl acetate concentrations, and yet retinyl acetate did not differ significantly between dose groups. This may seem surprising, because the HD group received twice as much retinyl acetate as did the LD group. Nonetheless, this finding suggests that both doses may have been large enough to allow the direct incorporation into micelles of retinyl acetate that had escaped hydrolysis.
Formation of RAG from retinoic acid by reaction with UDP-glucuronic acid is widely considered a mechanism for the removal of excess retinoic acid, which renders it much less toxic while it still retains some biological activity (31, 32). Less is known about ROG, which is formed by the reaction of retinol with UDP-glucuronic acid, but it is thought to be a biologically active metabolite of the retinol catabolism pathway (32, 33). Our finding of an apparent response to vitamin A dose by the glucuronides, a response that was most pronounced in the HD group, suggests that the glucuronides are metabolites that regulate serum retinoic acid and retinol concentrations. It is interesting that the postdose increase in RAG was greater in the HD group (reaching a peak at 4 h) than in the LD group, and there was a significant overall effect of treatment in the HD group only. Rapid formation of RAG from retinoic acid in the HD group within the first 4 h may have kept retinoic acid concentrations lower in the HD group. Even though retinoic acid concentrations were slightly higher at 2 h in the LD group than in the HD group, retinoic acid may not have reached high enough concentrations in the LD group to trigger significant RAG synthesis. ROG, formed from retinol, differed significantly between dose groups, which suggests that it may play a role in retinol homeostasis by preventing large increases. This possibility is supported by the relative lack of overall effect on serum retinol of the large doses of retinyl acetate, as compared with their effect on total vitamin A and retinyl esters, and the absence of differences between dose groups.
Our finding of an apparent response to vitamin A dose of 4-oxoretinol, generally considered an inactive vitamin A metabolite, could imply that one catabolic mechanism for the homeostatic regulation of serum retinol is the formation of 4-oxoretinol. In this system, we also observed 4-oxoretinoic acid, which is thought to be one of the metabolites in the retinoic acid catabolism pathway and which may regulate transcription in some species (34).
Typically, HPLC methods for measurement of vitamin A in biological tissue are tailored to analyze a specific retinoid (eg, retinol or esters) or a subset of similar retinoids (eg, polar retinoids) (16). However, limited conclusions can be drawn from assessing serum retinol or from exclusively assessing any other single retinoid (23, 25, 35, 36). Methods that simultaneously separate an array of vitamin A metabolites have not been widely used because of the difficulty in separating geometrically similar isomers of certain retinoids, the lengthy extraction procedure, and the long running time. In addition, many researchers are interested in only one or a few vitamin A compounds, and therefore they optimize the methods for isolating them exclusively. We were interested in the full profile of vitamin A metabolites and their kinetics after a large dose of vitamin A ester, and therefore we chose to use a global method, described first by Barua (13). Because the method had been previously validated only with known standards, we questioned its precision in responding to treatment that involved multiple biological samples. However, as shown by the CVs for total vitamin A and other metabolites, the method proved useful and allowed for separation and identification of a total of 9 retinyl esters, retinol, retinoic acid, RAG, ROG, 4-oxoretinoic acid, and 4-oxoretinol. Because analysis of serum retinol concentrations alone may not show the efficacy of an intervention with vitamin A, the quantification of other vitamin A metabolites may be useful.
Serum retinol metabolism may be altered during malnutrition as a result of reduced retinol-binding capacity (37). Moreover, stressed animals (eg, those with low vitamin A status, other nutrient deficiencies, or infection) may not be able to produce detoxifying metabolites to the same degree observed herein, and this failure would result in potentially higher circulating retinoic acid or retinol concentrations. Thus, one caveat of this study is that, whereas the vitamin A status of the sows may be considered to be in the low-normal range as compared with humans (0.29 µmol/g liver), the overall vitamin A nutriture and health of the sows were likely better than that of most recipients of vitamin A supplementation efforts. It is therefore conceivable that, whereas we observed significant treatment effects up to but not through 16 h after dose in this study, treatment effects may have been observed for a longer period in malnourished or otherwise compromised animals. Many of the enzymatic reactions that transform retinol and retinoic acid to their derivative retinoids, for example, require zinc as a cofactor. If zinc had been a limiting nutrient in the diet, retinol and retinoic acid concentrations may not have remained unchanged. Nonetheless, this study provides a framework for further evaluation in persons whose dietary intake of vitamin A is marginal or deficient and whose health may be compromised by infection or deficiencies that affect vitamin A metabolism and trafficking. Future investigations of the metabolism of vitamin A during high-dose supplementation programs in malnourished lactating women should proceed.
In summary, the results from this study showed significant dose-dependent effects of retinyl ester on total vitamin A and retinyl esters through 8 h. Serum retinol concentration did not increase and was blunted by a concomitant increase in derivative retinoids—namely, ROG and 4-oxoretinol. Retinoic acid concentrations were not affected, which indicated strict regulation of this teratogenic metabolite. On the other hand, an increase in the water-soluble, nontoxic derivative of retinoic acid, ie, RAG, was observed after treatment with retinyl acetate.
ACKNOWLEDGMENTS
Thomas Crenshaw provided technical expertise; the staff at the center, particularly Melissa Glenn and John Kane, provided indispensable onsite support. Undergraduates Ashley Valentine, Jordan Mills, and Sara Augustine aided in the collection and analysis of samples. The University of Wisconsin–Madison College of Agriculture and Life Sciences Statistical Consulting Service provided statistical consultation.
KLP was responsible for orchestrating the study, sample and statistical analysis, and manuscript preparation. The study was conducted in the laboratory of SAT, who was responsible for study design, oversight of the study operation, preparation of the doses, input into data analysis, and manuscript preparation and revision. Neither of the authors had any financial or personal conflict of interest.
REFERENCES