Main

DHA (22:6n-3) is the major n-3 polyunsaturated fatty acid in the membrane structural lipids of the CNS, particularly in the brain and retina(1–3). Most brain and retina DHA accretion occurs during brain development(4,5), normally taken to be from the beginning of the third trimester of gestation up to 2 y of age in human(6,7). In primates, n-3 FA deficiency in perinatal life results in impaired brain function, nervous disorders, and reduced retinal response and visual acuity(8–10).

Mammals can synthesize DHA from its precursor, LNA (18:3n-3) or its metabolites(11,12). However LNA and LA (18:2n-6) are essential FA that must be obtained from the diet. The principle metabolites of LNA are the n-3 LCP: EPA (20:5n-3), DPA (22:5n-3), and DHA. DHA can be retroconverted to EPA and DPA in human skin fibroblasts(13,14).

LA competes with LNA for the desaturation and elongation enzymes required for DHA synthesis(11,15). Preformed DHA, which is the DHA in plasma or organs derived from dietary DHA, is more efficiently incorporated into brain and retina in chicks(16), piglets(17), and rats(18) than is the DHA in plasma or organs derived from dietary LNA. Brain size in subprimate species is less than 4% of body weight at the peak of the brain growth spurts, whereas it is approximately 12% for primates, including humans(19). We have recently reported that the relative efficacy of LNA or DHA as precursors for fetal brain DHA accretion is about 20:1 in the pregnant baboon(20), a figure consistent with that measured in the rat(21), and we have estimated that a dietary level of 0.45% of energy as LNA in the maternal diet is adequate to support fetal brain growth.

The major dietary omega-3 FA are LNA, EPA, and DHA. LNA is found primarily in terrestrial plants, whereas EPA and DHA are found at various ratios in marine foods. Excess EPA in infant formula, relative to n-6 fatty acids, is thought to have a negative impact on growth(22), so that generally accepted sources of DHA are the precursor, LNA, and preformed DHA. Human colostrum and milk contain DHA and EPA, and breastfed infants maintain their plasma and erythrocyte DHA levels after birth, whereas infants fed formula with only LNA as a source of omega-3 FA exhibit a drop in DHA levels, and have lower brain DHA than breast-fed infants(23,24). Although human infants can synthesize DHA from LNA(25), there is consensus that their DHA levels may be compromised with current fat blends in infant formulas(26,27). The functional consequences for formula-fed term infants, if any, are controversial(21,22,28). Although LNA is the sole omega-3 FA in term infant formula in North America and in spite of the importance of DHA in mammalian development(11,29,30), the bioactivity of LNA as a precursor for DHA synthesis and brain accretion has not been determined directly for neonates of any primate, including humans. Blood pools accessible to sampling in humans, particularly plasma and erythrocytes, appear to be inappropriate proxies for essential FA accretion in critical neural tissue, particularly brain and retina(27,31), and direct brain sampling is not possible in normal humans.

The goal of this study was to determine the bioequivalence of dietary LNA and DHA as substrates for DHA accretion in newborn term baboon brain and incorporation in retina, RPE, and liver, as well as in potential marker pools, plasma and erythrocytes, after oral administration of LNA* or DHA*; isotopic tracer detection is by high precision mass spectrometry. We hypothesize that preformed DHA is more efficiently incorporated than LNA-derived DHA into neonatal tissue. We report the data as the absolute accretion level of DHA resulting from either LNA or DHA in brain, where turnover is very slow, for both dietary forms. From these data the bioequivalence is calculated. Finally, we report the distribution of FA metabolites at 14 d post dose in each pool.

METHODS

Animals and diets. The care of animals was approved by the Cornell Institutional Animal Care and Use Committee and the facility was approved by the American Association for Laboratory Animal Care. Nine pregnant baboons were obtained either from the Southwest Foundation for Biomedical Research (San Antonio, TX) or from the University of Illinois at Chicago (Chicago, IL). Baboons had at least one previous pregnancy, and consumed standard commercial primate chow before and soon after pregnancy. After breeding at their respective colonies, pregnancy was confirmed and they were transported as soon as possible to the Laboratory for Pregnancy and Newborn Research at Cornell, arriving at 60-90 DGA. A complete veterinary examination was performed upon arrival and baboons were housed individually in cages in sight of at least one other baboon. The room temperature and humidity were maintained at 24°C and 70%, respectively, with a 14 h light and 10 h dark cycle. Adult animals consumed a commercial primate chow supplemented with fresh bananas, apples, oranges, and yams.

Neonatal baboons were born by CS at estimated conceptual ages of 174 to 189 DGA and ranged in birth weight from 720 to 970 g. Details are presented in Table 1. They were housed initially in an enclosed incubator, then transferred at about 7 d of age to individual stainless steel cages in a controlled-access nursery where the temperature and humidity were 28°C and 50%, respectively. All humans in contact with the neonates wore sterile gowns, hair nets, face masks, and shoe protectors.

Table 1 Characteristics of neonatal baboons*
Full size table

Neonatal baboons were fed a commercial human infant formula (Enfamil, Mead-Johnson, Evansville, IN) purchased from a local supermarket. All neonates thrived on the formula and, after the first week of life, increased from their birth weight of 822 ± 86 g to 1214 ± 128 g, a 48% gain by 6 wk of age. A quantitative FA profile of the formula done in our laboratory showed that 18% of total FA by weight as LA and 1.8% of total FA as LNA (LA/LNA = 10), in good agreement with values published previously(32).

Doses and sampling. [U-13C]-LNA and [U-13C]-DHA were purified from a [U-13C]-algal oil (Martek Biosciences, Columbia, MD) as described previously(33). Chemical and isotopic purity were greater than 98% as determined by gas chromatography and mass spectrometry. Either 7.5 mg LNA* or 4.4 mg DHA* as nonesterified FA was weighed in a 2-mL tared vial and 1.0 mL of prepared formula (80°C) was added and vortexed. Fatty acids are observed to dissolve rapidly upon first addition of hot formula. The vial contents were administered to the neonate as well as three rinses with cold formula. A single dose was administered once orally to each 4-wk-old neonatal baboon just before a scheduled meal. About 20 µL of each dose preparation was analyzed quantitatively and isotopically for label concentration; greater than 90% of the labeled fatty acid was found to be dissolved in formula.

A ninth baboon was used to verify baseline isotope ratios and did not receive a dose. Only one baboon was used for baseline assessment because, in the absence of artificially-enriched isotopes, baseline 13C/12C are constant within about 10 parts per million for subjects on a controlled diet(34).

It has long been known that overt omega-3 deficiency is very difficult to produce in the adult animal, leading to the conclusion that DHA turnover is very slow when dietary sources of DHA are low. Slow turnover implies that absolute amounts of [13C]-DHA should plateau after a single dose. Previous measurements support this conclusion in fetal baboons(20) and indicate that the plateau occurs about 2 wk post dose administration to a pregnant baboon. Weeks after the dose, the label can be detected in the plasma of the preformed DHA-dosed mother, so that transfer across the placenta to the fetal plasma is the slowest component of the time-to-plateau. It is expected that the time-to-plateau in the neonate is much faster because transfer of orally administered DHA to the plasma is complete within a couple of days at the normal alimentary transit times of formula. Thus, 2 wk after dose, when plateau levels were assumed, the 6-wk-old neonates were weighed, placed under halothane general anesthesia, and euthanized by exsanguination. Blood was collected with heparin as anticoagulant, and plasma and erythrocytes were separated immediately after centrifugation at 4°C. Retina and RPE were separated, collected, and stored in saline at 4°C for less than 2 h until lipid extraction occurred. The brain occipital lobes and liver were removed quickly, wrapped in aluminum foil, and frozen in liquid N2. All samples were kept in a -80°C freezer until analysis.

Lipid Extraction and FA and Tracer Analysis. Total lipids were extracted from tissue homogenate according to the Bligh and Dyer method(35), and FAME prepared using 14% BF3 in methanol. A known amount of fresh heptadecanoic acid (50 µg per 100 mg tissue or per retina or per RPE or per 500 µL plasma; 99+% pure, Sigma Chemical, St. Louis, MO) was added as an internal standard to the tissue homogenate before extraction. The purified FAME was dissolved in hexane with butylated hydroxytoluene as an antioxidant, flushed with N2, and stored in a -20°C freezer until analysis.

FAME were analyzed with a Hewlett Packard 5890 series II GC-FID using H2 carrier gas. Quantitative profiles were calculated using the internal standard and an equal weight mixture to derive response factors for each FA. GC-FID conditions and calibration details are reported elsewhere(36). Tracer analysis was performed using a high precision GCC-IRMS (Finnigan 252), described in detail previously(37,38).

Calculations. The concentration of tracer in tissues was calculated from the concentration of each FA detected by GC-FID, and the AFE of each FA determined by the GCC-IRMS. The details are described elsewhere(36,39). Briefly, high precision data are presented as the relative deviation from the international standard Pee Dee Belemnite (PDB) with a 13C/12C isotope ratio = RPDB = 0.0112372, as: Equation 1 where δ13CPDB is expressed in permil (‰) units and X refers to the sample. δ13CPDB can be directly converted to atom fraction 13C, and AFE, as: Equation 2

Based on mass balance(39), the absolute amount of tracer can be calculated by multiplying the AFE by the concentration of tracee. The concentration of tracee is calculated from gas chromatograph peak areas, using the internal standard and correcting for differential response with equal-weight standard (68A; Nuchek Prep, Elysian MN), as has been discussed in detail previously(40). To facilitate comparison between FA, all results are expressed as molar dose equivalents by adjusting for the number of mol of carbon in the tracer and analyte FA, respectively. In this way, for example, conversion of 1% of a 100 nmole LNA dose results in 1 nmole of DHA. The appropriate equation is: Equation 3 where D is dose-equivalents (in mol or g), AFE is determined by GCC-IRMS, Qf is the quantity of FA f measured by GC-FID (in mol or g), equal to the amount of labeled dose-derived FA, and C is the carbon number of f or Dose. The portion of the equation using mT (= Qf) and α (= Cf/CDose) corresponds to the general form presented previously(39). For example, the quantity of DHA* derived from LNA*-dosed animal is calculated as: Equation 4

Finally, these data were normalized to the dose size and the data presented as "% of dose" (%Dose) to adjust the different size of doses. No adjustment for body weight was made. Data are reported as %Dose found in whole brain or liver or per retina, per RPE, and per 100 mL plasma and erythrocytes. Bioequivalence of dietary LNA and DHA as substrates for tissue DHA incorporation at 2 wk post dose is calculated directly from %Dose as preformed DHA over %Dose as LNA-derived DHA in tissues.

Statistics. Data are expressed as mean ± SD. Tukey's t test was used to detect significant differences at p < 0.05 performed by MINITAB for Windows 95, version II (State College, PA).

RESULTS

Body and organ weights. Body and organ weights are presented in Table 1. No significant differences were found in brain and liver weights between LNA*-dosed or DHA*-dosed neonatal baboons. Brain weight averaged about 8% of total body weight in these 6-wk-old baboons.

FA composition of tissues. Tissue FA composition of brain, retina, RPE, liver, plasma, and erythrocytes are shown in Table 2. No statistically significant differences in FA profiles between LNA*- and DHA*-dosed neonates were detected. Therefore, they are presented in pooled form. The brain DHA composition was 9% of total FA, similar to previous findings in fetuses (8.2%)(20), and consistent with the figure of 7% in human infants(41). Retinal DHA levels were 16% of total FA, in good agreement with the human infant data(41), showing the highest concentration of DHA in the tissues investigated. In contrast, DHA constituted 1 to 4% of total FA in RPE, liver, plasma, and erythrocytes.

Table 2 Neonatal tissue FA composition*
Full size table

Brain tissue had a higher DHA concentration per g wet tissue than that of liver, erythrocytes, and plasma. The DHA concentration in brain and liver were 1.35 mg and 0.6 mg per g wet tissue, respectively. The quantities of DHA were 0.02 mg and 0.05 mg per mL of plasma and erythrocytes, respectively. There were 0.23 mg and 0.03 mg of DHA per retina and RPE, respectively.

TRACER RESULTS

Brain, retina, and RPE. The incorporation of n-3 FA derived from LNA* or DHA* dosing in the whole brain, retina, and RPE is shown in Figure 1 on a split ordinate to present the three related organs on one plot. The %Doses recovered as brain DHA were 1.71% for preformed DHA and 0.23% for LNA-derived DHA. Preformed DHA* was therefore 7-fold more efficiently incorporated and retained into brain than LNA*-derived DHA*. About 0.1% LNA* dose was detected as DPA*, whereas retroconversion of the DHA* dose was negligible.

Figure 1
figure 1

Labeled n-3 FA found in whole brain (Br, right ordinate scale), and in retina (Re) and RPE (left ordinate scale) because of a dose of either LNA* or DHA* to 4-wk-old baboon neonates, at 14 d post dose. The ratio of preformed DHA to LNA-derived DHA %Dose is 7:1 for brain, and 12:1 and 15:1 for retina and RPE, respectively. Because brain DHA turnover is slow, the ratio reflects accretion. Very low retroconversion of DHA* doses to DPA is observed as well.

Full size image

The ratio of dietary DHA* and LNA* as substrates for DHA* incorporation in retina and RPE were 12 and 15, respectively. A greater percentage of the doses was found in retina than RPE. Very small amounts of LNA* and EPA* were detected in retina and RPE in LNA*-dosed animals. Retroconversion of preformed DHA* to other FA was very low.

Liver, plasma, erythrocytes. Figure 2 presents the results for liver, plasma, and erythrocytes. The ratio of dietary DHA* and LNA* as substrates for liver DHA* incorporation was 27:1. Almost 2.4% of the preformed DHA* dose was found in liver at 14 d, while <0.1% of the LNA* dose was found in any FA form. As the liver is an active site of FA metabolism, and especially desaturation, elongation, and oxidation, these data suggest that the liver metabolizes LNA much more quickly than DHA.

Figure 2
figure 2

Labeled n-3 FA in whole liver (Li, right ordinate scale), and in plasma (Pl) and erythrocytes (Ery, left ordinate scale) due to a dose of either LNA* or DHA* to 4-wk-old baboon neonates, at 14 d post dose. The ratio of preformed DHA accretion to LNA-derived DHA accretion is 27:1 for liver, and 29:1 and 51:1 for plasma and erythrocytes, respectively.

Full size image

Figure 2 also shows the amounts of labeled n-3 FA per 100 mL plasma and erythrocytes derived from LNA*-dosed and DHA*-dosed neonatal baboons, presented again on a split scale. The recovery of both preformed DHA* and LNA*-derived DHA* or n-3 FA was higher in erythrocytes than in plasma. Total LNA* FA metabolites were 0.01-0.03%Dose in 100 mL plasma and erythrocytes. Preformed DHA* was found to be 29-fold greater in plasma lipids compared with LNA*-derived DHA*. In erythrocytes, preformed DHA* was 51-fold greater than LNA*-derived DHA*. LNA*-derived DPA* in erythrocytes was unexpectedly the highest %Dose compared with other LNA-derived metabolites in that pool. These data indicate that neither plasma nor erythrocyte total fatty acid pools reflect liver or brain levels for DHA* incorporation due to a single tracer dose. Plasma or erythrocyte data should therefore be interpreted with caution when viewed in isolation as a proxy for organ fatty acid tracer metabolism.

Distribution of labeled n-3 FA. Retroconversion of preformed DHA* to labeled FA was below 3% for all organs and tissues reported here. Figure 3 shows the % distribution of the LNA*-derived n-3 FA. Percentages for each tissue add to 100. The pattern is similar for brain, retina, RPE, liver, and plasma. In all of these pools, DHA* is the predominant labeled metabolite of LNA*, ranging from 60% in RPE and plasma to 84% in retina. DPA* is next most abundant within a concentration range of 15% in retina to 36% in RPE. The exception to the trend is erythrocytes, in which DPA* dominates at about 55%, with DHA at 26%. This observation is consistent with our measurements in heart and lung of these animals (unpublished data). LNA* is lowest in all pools.

Figure 3
figure 3

The distribution of labeled FA as a percentage of all labeled FA found in each organ or in plasma at 2 wk post dose. DHA* is found at highest concentration followed by 22:5 in all pools except erythrocytes, where 22:5 predominates.

Full size image

DISCUSSION

This is the first report of the bioequivalence of the dietary LNA and DHA as precursors for primate neonate brain DHA accretion. The measured ratio of 7 is considerably smaller than reported previously in pregnancy(20). Ratios of accretion levels at one time point reflect overall, integrated levels if turnover of the product is negligible. The turnover of brain DHA is thought to be slow, in part because of the difficulty in producing DHA deficiency in adults animals, and our recent measurements provide direct evidence that it is very low over 35 d in primate fetuses. The present study sampled a single time point at 14 d post dose, when brain DHA* accretion is hypothesized to have plateaued. To determine the minimum amount of DHA accretion over 2 wk, the measured percent of the LNA* dose recovered as brain DHA* can be treated as the tracer, and total dietary LNA as the tracee. Because all DHA must ultimately be derived from the neonate diet, this figure can be compared with the total estimated DHA accretion over this period to determine the amount of DHA retained.

Neonate LNA demand as a substrate for brain DHA synthesis can be estimated from average brain growth to be about 0.453 mg LNA/day. The neonates consumed an average of 290 mL formula per day. From the caloric density of 0.676 kcal/mL and the percent of energy as LNA (0.82%) (Diersen-Shade, personal communication), the average intake of LNA was 179 mg per day. Our measurements indicate that 0.23% of dietary LNA appeared in the brain as DHA; thus, 0.23% × 179 mg = 0.418 mg LNA per day appeared as brain DHA. The ratio 0.418/0.453 = 92% indicates that most of the LNA*-derived DHA* was present in the brain 2 wk post dose. This estimate suggests a modest 4% per week turnover of DHA in the growing neonate brain, small enough to imply that the measured ratio of preformed DHA* to LNA*-derived DHA* reflects the integrated bioequivalence.

In brain, only 0.2% and 2% of labeled n-3 FA detected in LNA*-dosed animals were LNA* and EPA*, respectively, whereas DPA* accounted for 30% and DHA* 68% of labeled n-3 FA. LNA* and EPA* in brain are significantly depleted compared with plasma. This observation supports the hypothesis that uptake of LNA and EPA from plasma is limited, or their conversion to DPA and DHA is rapid, which has been seen in other measurements(42).

There is a significant shift in DHA*/DPA* between retina (85%/14%) and RPE (60%/35%). There is very active recycling of DHA between retina and RPE, related to recycling of disks. The RPE can synthesize DHA from DPA(43), suggesting that RPE takes up DPA from the circulation and converts it to DHA for transport to the retina. DHA metabolism is highly active in the retina, and DHA is actively exchanged between RPE and retina(44). Pawloski et al. have suggested that DPA be a preferred transport form for n-3 LCP from liver to brain in cats(45). Wang and Anderson(46) suggested that supplemental DHA be transported to retina from the blood through the RPE and finally to the retina rod outer segments. The FA composition of the retina, like that of the liver, appears to be more sensitive to dietary FA composition than that of the brain(47,48).

In general, tracer experiments are indicated primarily for endogenous biomolecules requiring elucidation of a portion of a complex web of pathways. In any organ pool there are myriad biochemical pools among which the tracer is interchanged. In the present case, the relative nutritional requirements for LNA and DHA as substrates for brain DHA accretion in neonates are desired, and the labeled compounds are required to trace quantitatively the relatively short-term metabolism of nutrients consumed at a defined time.

For purposes of the tracer experiment, the dose FA, LNA* or DHA*, is the tracer whereas the corresponding endogenous dietary FA, LNA or DHA, is the tracee. The absolute concentration of tracer doses in tissue depends on the initial dose. However, the %Dose does not depend on the initial tracer dose as long as it is administered at a sufficiently low concentration that endogenous concentrations are not perturbed: halving or quartering the absolute dose will have no effect on this parameter. The %Dose does depend on the absolute concentration of tracee, because by definition it defines a pool size. %Dose is also related to all endogenous concentrations with effects on the metabolic process under study. In the present case of unsaturated FA metabolism, this certainly includes the dietary LCP and n-6 fatty acid concentrations, and possibly other nutrients as well.

A fixed diet to a normal neonate should result in a fixed FA concentration within endogenous pools in liver, plasma, and periphery, within biological variability. All these pools, and others, interact to ultimately deliver n-3 FA to developing brain. Thus, the specific results found here, particularly brain bioequivalence of 7:1, apply strictly only to diets similar to the commercial infant formula used here. However, the infant formula used is typical of North American formulas in that it is LCP-free and has an n-6/n-3 ratio of about 10. Future modifications of formula composition ratio are very likely to reduce the n-6/n-3 ratio, a change that is expected to enhance conversion of LNA to DHA as relative LA concentration decreases.

The most significant practical aspect of these results may be to suggest appropriate supplementation of DHA in cases where this step is deemed desirable based on functional outcomes. Although studies of functional outcomes with human neonates must define normality, mechanistic studies in appropriate animal models are required to confirm and extend clinical observations. In the wild, the baboon is a wide-ranging omnivore whose plasma lipid composition and behavior responds to diet in a manner parallel to humans.

The measured accretion ratio of about 7 can be compared with several studies in nonprimates. In nontracer studies, DHA was preferred 4:1 in newborn piglet CNS(17). From the data reported by Anderson et al.(16), we calculate that dietary DHA results in about 4-fold greater brain concentration of DHA than LNA in chicks at 3 wk of age (12.8 ± 3.3 versus 3.5 ± 0.7 wt% DHA in brain for diets about 3.6 wt% dietary DHA or LNA, respectively). Using orally administered [14C]FA, Sinclair(18) found that accretion of radioactivity in brain lipids was 9 times higher for DHA than for LNA in suckling rats after 22 h. Anderson et al.(49) determined that radioactivity incorporated into developing brain of hepatectomized rats after [14C]DHA injection was 1.5-fold that of [14C]LNA by injection after 30 min, supporting the important role of the liver in DHA biosynthesis. The latter two tracer studies did not distinguish between molecular forms of radioactivity. In summary, our value of 7 is within the range measured in intact subprimate species for nontracer studies.

In our previous study of pregnant baboons(20), fetal brain DHA accretion was 1.6%Dose, which is remarkably close to the 1.71%Dose found for neonate brains in this study. Maternal transport of DHA from bloodstream to fetus appears to be nearly as efficient as the transport of DHA from alimentary tract to bloodstream in the neonate, assuming the 147 DGA fetus metabolizes DHA with efficiency equal to that of the neonate. The difference in bioequivalence can be accounted for as the mother's inefficient conversion/transport of LNA to DHA.

The accretion of DHA in brain of only 1.7%Dose does not suggest strong whole body conservation of DHA for brain growth. This low figure might be taken to imply that LNA can replace DHA on a 7:1 basis in infant formula with a n-6/n-3 ratio of 10. There is some concern, however, that catabolic products of DHA, notably cholesterol and saturated FA, may serve as a carbon source for the developing brain, as has been noted in rats(50) and rhesus monkeys(51). We did not measure total excess 13C in the brain and therefore cannot establish whether DHA or other n-3 FA are converted to these products soon after uptake. Further measurements on the metabolic role of brain DHA are required to resolve this issue.