Main

DHA (22:6n-3) is the major n-3 polyunsaturated fatty acid in the mammalian brain and in retinal phospholipids, constituting approximately 25 and 50% of the total fatty acids of the aminophospholipids and phosphatidylinositol of the gray matter and retinal rod outer segments, respectively(1). The majority of DHA present in the human brain is incorporated during the brain growth spurt which starts at wk 26 of gestation and imposes high demands until about 2 y of age(2–4). The intensity and duration of nutritional demands for essential brain components, particularly n-3 fatty acids, during this period is unique in primates compared with other land mammals including small laboratory species(2).

An external source of n-3 fatty acids is essential because mammals cannot desaturate fatty acids at positions beyond the 10th carbon from the carboxyl terminus(1, 5, 6). Prominent sources for the pregnant primate are the C18 fatty acid LNA(18:3n-3), which can be elongated and desaturated in the adult animal to yield DHA(7–9), as well as preformed DHA, which is available at high concentrations in marine and animal foods but effectively absent from terrestrial plant foods.

The extra biochemical steps required for conversion of LNA to DHA are thought to reduce the efficacy with which LNA meets n-3 requirements; however, careful measurements of the bio-equivalence of LNA and DHA during development have been attempted only in chicks and pigs(10, 11) and very recently in artificially reared rats(12). Although it is known that LNA provided to rats at 0.4%en meets requirements during brain growth(13), no comparable figure is available for DHA.

Premature birth in humans is associated with a dramatic increase in neurologic and learning disorders in childhood follow-ups(14). Fatty acid nutriture is thought to be involved because accretion of DHA, normally initiated in utero and supported by placental nutrient transport, must take place for preterm infants postnatally, with nutritional support from either breast milk or formula. Preterm infants fed conventional formula without LCP show a drop in plasma and erythrocyte DHA levels and poorer electroretinogram response(6, 15) and visual acuity compared with infants fed a marine oil-supplemented formula or breast milk(16, 17). Human milk contains moderate levels of DHA, which appear to be conserved even when the diet is low in n-3 fatty acids(18–21).

The placenta is known to transport fatty acids from mother to fetus and appears to prefer transport of unsaturates. Studies of infant rhesus monkeys whose mothers were deprived of n-3 fatty acids showed a decrease inn-3 fatty acids in the CNS resulting in abnormal electroretinograms and reduced visual acuity(22, 23). Recent data show that human preterm infants are able to elongate and desaturate LNA, as well as LA (18:2n-6), to 20 and 22 carbon functionally important LCP(9), but the quantitative adequacy of this process is unknown. Human brain composition is affected by early diet, as breast-fed infants have greater DHA levels in brain than do formula-fed infants(24, 25). Information on the relative levels of interconversion and transport of LNA and DHA in the pregnant primate and its fetus, including incorporation into fetal tissues, is required to determine the optimal dietary intakes of n-3 fatty acids.

Dietary requirements for LNA are best considered under dietary conditions where LCP are absent because the only known function of LNA is as a precursor of n-3 LCP. Although rare in Western countries, LCP-free diets are relevant to freeliving situations. For instance, the strict vegan diet is characterized by an absence of LCP and is often low in fat, thus most LCP for the fetus of a vegan mother must be derived from dietary 18 carbon precursors often present at low levels. Another example is the formula-fed infant; commercial term infant formulas in the United States presently contain no LCP(26), and have an n-6/n-3 ratio of about 9:1. In spite of the importance of conversion efficiency, the nutritional bioequivalence of LNA and DHA as it is related to the developing primate brain and retina is not known. Further, brain and retinal accretion efficiency of n-3 fatty acids perinatally has not been measured in any primate species.

Using LNA* and DHA*, high precision mass spectrometry, and chronically catheterized pregnant baboons, we report the accretion of these fatty acids in fetal brain, liver, and retina after an i.v. dose to the mother. We show desaturation and elongation products of LNA appearing in the maternal and fetal plasma, and calculate the bioequivalence of dietary LNA and DHA as substrates for fetal brain DHA accretion. Finally, we estimate the minimal dietary intake of LNA and DHA needed to supply the brain with adequate DHA during the brain growth spurt.

METHODS

Care of animals. All procedures were approved by the Cornell Institutional Animal Care and Use Committee, and all facilities were approved by the American Association for Laboratory Animal Care. Fifteen female baboons (Papio cynocephalus) from the Southwest Foundation for Biomedical Research (San Antonio, TX) were used for this study. After confirmation of pregnancy, the baboons were transported to the Laboratory for Pregnancy and Newborn Research at Cornell University, and a thorough veterinary examination was performed. Animals were jacketed, connected to a flexible tether and swivel, and were housed in individual cages within view of at least one other baboon. These methods have previously been described in detail(27–29). They were acclimated to a 14-h light and 10-h dark cycle and were fed a semipurified diet devoid of LCP as shown in Table 2 (see “Results”). They were offered low fat treats such as fruits and potatoes during the day. After a minimum of 30 d of acclimation, 12 maternal baboons were instrumented under halothane general anesthesia with chronic indwelling catheters. In four baboons 7 d after the maternal catheterization, the femoral artery and vein of the fetuses were catheterized after laparotomy and hysterotomy under halothane general anesthesia.

Table 2 Fatty acid composition of LCP-free diet, maternal(n = 7) and fetal (n = 4) plasma (wt% of total fatty acids) expressed as mean± SD
Full size table

Diets. We formulated an LCP-free diet with low, but adequate, 18:2n-6 (2%en) and 18:3n-3 (0.2%en) to study metabolism of animals which must produce all LCP by elongation/desaturation or from limited body stores. Animals consumed the diet for an average of 8 wk before administration of labeled tracers.

At a mean age of 139 ± 11 (SD) dGA (term = 182 d), the pregnant animals received an i.v. dose of LNA* (n = 8) or DHA*(n = 4, “preformed DHA”) over 1 h.Table 1 indicates the weights and gestational ages of animals in the study as well as the doses administered to these animals. A maternal femoral arterial plasma sample and fetal femoral arterial plasma sample (when available) were taken before the infusion of the dose to determine baseline isotope ratio values. After the dose, maternal femoral artery plasma samples were taken from some animals hourly on the day of the dose and daily on all subsequent days. Fetal plasma samples were taken hourly from one animal dosed with LNA* and two animals dosed with DHA*. Cesarean section under halothane general anesthesia was performed at various times after the dose was administered. Fetuses were killed by exsanguination while still under halothane general anesthesia. Fetal liver, brain, and retina were weighed, flash frozen, and stored at -80°C until analysis.

Table 1 Characteristics of animals in study
Full size table

Preparation of labeled LNA or DHA. The doses of LNA* or DHA* were purified from a U-13C-labeled algal oil(Martek Biosciences, Columbia, MD). Briefly, FAME were first separated by degree of unsaturation on Ag-loaded solid phase extraction columns (Varian, Harbor City CA). DHA* was purified to 98% in this step and was used without further purification. LNA* is derived from a different oil containing appreciable contaminants, so a second step using HPLC is used to as a final purification to >98%, verified by capillary GC. The methyl esters of LNA* and DHA* were hydrolyzed to the FFA. Doses were prepared by sonicating the FFA into 5.0 mL of 10% Intralipid (KaviVitrum, Franklin, OH) diluted with 5.0 mL of sterile saline. After an overnight fast, the final solution was infused i.v. over 1 h beginning at 0830 h.

Lipid extraction and analysis. Total lipids were extracted from a whole liver lobe, a sample of brain from the occipital lobe including the visual cortex, and whole retina by the method of Bligh and Dyer(30). Butylated hydroxytoluene was added to solvents as an antioxidant, and heptadecanoate (17:0) diluted in hexane was added as an internal standard. Initially, samples were saponified by adding 1.0 mL of 0.5 N methanolic NaOH and heating in a water bath at 60°C for 10 min. Adding 1.0 mL of 14% BF3 in methanol and heating at 100°C for 2 min resulted in the formation of FAME. Hexane (1.0 mL) was added to extract the FAME, and tubes were vortexed and placed in a boiling water bath for 1 min. After the addition of 1.0 mL of saturated NaCl solution, tubes were centrifuged at 3000 rpm, the top hexane layer containing the FAME was transferred to another tube, and the hexane was dried by a stream of N2. The samples were resuspended in hexane for quantitative analysis by GC with a flame ionization detector and for isotopic enrichment by GC-combustion-IRMS. FAME were analyzed using a Hewlett Packard 5890 GC with a DB23 (J&W Scientific) fused silica capillary column (30 m, 0.32 mm inside diameter, 0.25 film thickness) with H2 carrier gas and N2 auxiliary gas. Response factors were determined daily using a standard mixture of fatty acids, and the data are expressed as percent of total fatty acids C≥14, by weight.

Isotopic analysis was performed using GC-combustion-IRMS, discussed in detail elsewhere(31). The effluent of Varian 3400 capillary GC was transferred by helium carrier gas into a furnace. This furnace is held at 850°C and loaded with a solid source of O2 to convert all organic compounds entering the furnace to CO2 and H2O. The water was removed by a water trap as chromatographic resolution was maintained. Dried CO2 entered the ion source of a Finnigan 252 high sensitivity, high precision IRMS instrument where isotope ratios were determined. This instrument has an absolute sensitivity of about 103 molecules/ion detected and has dedicated detectors/electronics for each of the major masses of CO2+, m/z = 44, 45, and 46. Three chromatograms were acquired continuously from the three channels during each GC program. Fatty acid peaks were identified in the m/z= 44 channel and extrapolated to the other channels. The m/z = 46 signal was used to adjust for the contribution of the 17O-substituted CO2 appearing at the m/z = 45 signal. Using these parameters, isotope ratios corresponding to r = 13C/12C were calculated. The average analytical precision of these measurements had a coefficient of variation = 0.05% (500 ppm).

High precision IRMS data are typically expressed as a ratio describing the relative deviation in parts per thousand from the international standard PDB. PDB is a carbonate relatively rich in 13C with an isotope ratio defined as RPDB = 0.0112372, using: eq. (1)

where δ13CPDB is expressed as “per mil”(‰) and “SPL” refers to the sample. Atom fraction, or more commonly, atom percent (AP), is more convenient than ratio-based expressions and describes the fraction or percent of 13C in each fatty acid peak. In this study, we convert the ratio data to AP or atom percent excess (APE) by subtracting baseline isotope levels, and use this value for further calculations. The contribution to isotope ratio made by the methyl C added in derivatization is taken into account by this calculation, so additional correction is not necessary.

The conversion of a labeled tracer, such as LNA* to a particular metabolite in a tissue, such as DHA*, can be determined on a molar basis by multiplying the APE of DHA by the concentration of DHA per unit lipid. This calculation yields the percentage of excess 13C in the sample DHA peak compared with the amount of 13C in the baseline DHA peak.

For these specific studies, C20-22 fatty acids synthesized from uniformly labeled LNA* are analyzed, thus only 18 of the 20-22 carbon product molecules are labeled. A correction is applied to express conversion in terms of mol of the dosed fatty acid, or LNA* equivalents, to adjust for the endogenous C added during chain elongation. This permits direct comparison of LNA and DHA dose efficacies. The equation takes the form:eq. (2)

where D* is dose-equivalents, F is the analyte fatty acid, QF is the quantity of F derived from GC-flame ionization detector measurements, and “[C]/[x]” designates the mol of C per mol of either (x=) D* or F. For the particular case of DHA (=F) derived from a dose of LNA (=D*), we have: eq. (3)

Label is reported as percent of dose found in a particular fatty acid in a particular pool, usually the whole organ, to adjust for differing dose sizes on a per kg of body weight basis. This expression can be directly applied to compare LNA-derived DHA and performed DHA and incorporated into tissues. An estimate of the percent of dietary LNA that is converted to DHA and incorporated into the brains of these animals can then be deduced.

Plasma kinetic curves are presented either on a molar concentration basis or, for LNA dosings, as the molar amount of label appearing in a particularn-3 fatty acid as a percent of the total labeled fatty acids in each sample. For the latter, the sum of labeled fatty acids reported at each time point should be 100% within experimental error.

Statistics. Pairwise significance was tested using the t test with Excel 7.0 spreadsheet (Microsoft, Seattle, WA), with significance declared at p < 0.05.

To estimate the bioequivalence of DHA and LNA in fetal organs, the data are least squares fit with trend lines using Excel. The ratio of the two trend lines is plotted and the plateau of the ratio plot is considered the bioequivalence of DHA to LNA in a particular organ.

RESULTS

Fatty acid composition. The fatty acid composition of the diet and maternal and fetal plasma is shown in Table 2 and the fatty acid composition of fetal liver, brain, and retina of baboons is shown in Table 3. The concentration of saturates and monounsaturates is identical for maternal and fetal plasma. The concentration of LA in the fetal plasma is significantly lower compared with maternal plasma, whereas that of arachidonic acid (20:4n-6) in the fetal plasma is significantly higher than its concentration in the maternal plasma. No statistically significant difference in DHA concentrations was detected, although the fetal mean concentration was greater. Increasing concentration of DHA is evident moving from maternal plasma, to fetal plasma, to liver, to brain, to retina, the latter of which are highly significant, and consistent with previous observations termed “biomagnification”(32). As expected, DHA is the most prominentn-3 fatty acid in neural tissue, with a concentration of 8.2% of total brain fatty acids. This figure compares well with the figure for human infants which remains constant at about 7%(33). Arachidonic acid was at greatest concentration of all polyunsaturated fatty acids, again in agreement with human data, with a concentration of 11.6%. LNA is characteristically low in fetal liver and retina and barely detectable in the brain.

Table 3 Fatty acid composition of fetal liver, brain, and retina (wt% of total fatty acids) expressed as mean ± SD (n = 15)
Full size table

Tracer results. Labeled fatty acids were monitored in the plasma of several mother/fetus pairs during and after the infusion of LNA* and DHA* to pregnant animals to establish gross kinetics and detect fatty acid metabolites. Figures 1 and2 show the disappearance curves in maternal and fetal plasma after administration of LNA* to one pregnant baboon and to two animals dosed with DHA*, respectively. In the LNA*-dosed animals, the absolute enrichment of LNA in maternal and fetal plasma is similar, and the maximum enrichment is achieved within 1 h after the infusion of LNA* commenced. The enrichment of LNA in both the mother and fetus declines from 9 h post-dose and approaches baseline levels by 72 h post-dose. Disappearance of DHA* in the maternal and fetal plasma of the DHA*-dosed animals was markedly different. The enrichment of DHA in the maternal plasma was greater than the enrichment of DHA* in the fetal plasma until 200 h post-dose, and always greater than maternal or fetal plasma LNA in the LNA*-dosed animals. Maximum enrichment of DHA in maternal plasma and fetal plasma is approximately 100 h post-dose. At about 10 d, the DHA enrichment is maternal, and fetal plasma begins to converge. In one pair, DHA* was detected in maternal and fetal plasma until at least 33 and 16 d post-dose, respectively.

Figure 1
figure 1

LNA plasma kinetics. Disappearance curves for plasma LNA*, in animals dosed maternally with LNA*, monitored in plasma of the mother and fetus, and expressed as percent of dose. The concentrations of LNA* are comparable in mother and fetus and drop rapidly to near baseline levels by 72 h. ▪, Fetal; □, maternal.

Full size image
Figure 2
figure 2

DHA plasma kinetics. Disappearance curves for two DHA*-dosed mothers and fetuses (pair A: ▴, fetal: ▵, maternal, and pair B: ▪, fetal; □, maternal) after administration to the mother, presented on a split x axis up to about 32 d. The maternal plasma DHA concentrations peak at about 10-fold greater than fetuses on d 1 and continue at higher levels at the last time point analyzed. There are two peaks apparent in the maternal curves, one at administration, and one at 2-3 d post-dose. Mothers and fetuses both maintained plasma DHA* concentrations significantly above baseline at all time points sampled.

Full size image

Conversion of LNA* to DHA* by the mothers was indicated by the LCPs 20:5n-3*, 22:5n-3*, and DHA* detected in the maternal plasma shown for a single animal inFigure 3a. At early times LNA* is 100% of the labeled fatty acids in plasma. By 48 h it has fallen to half total label, whereas 20:5n-3 has peaked at 20% and begins to decline to baseline as might be expected if this species functions as an intermediate. DHA and 22:5n-3 continue to rise at the last plotted point at 72 h. Shown inFigure 3b is the relative enrichment of these fatty acids in fetal plasma samples taken simultaneously. Enriched 20:5n-3 was very low in the fetal plasma of this animal, but 22:5n-3* and DHA*, as percent of total label, were present at similar levels to those measured in the maternal plasma. Saturated and monounsaturated fatty acids were also enriched in plasma at some time points (data not shown).

Figure 3
figure 3

LCP metabolite kinetics. Enrichment of plasman-3 fatty acids after LNA* administration to the mother, expressed as a percentage of the total label detected at each sampling.(a) Maternal plasma shows a peak in the 20:5n-3* curve at 48 h, consistent with this fatty acid's role as an intermediate in DHA synthesis. DHA* (22:6n-3*) and 22:5n-3* are rising at the last time point sampled at 72 h.(b) Fetal plasma DHA* and 22:5n-3* plateau at 72 h, whereas 20:5n-3 levels were very low (not presented) and LNA is near baseline.

Full size image

Fetal liver accretion. Figure 4 shows the incorporation of LNA* (% dose/organ) into fetal liver of animals whose mothers were dosed with LNA*. For all organ data, each point represents a single animal so biologic variability at each time point cannot be presented; analytical error is generally within the symbols. In these livers, LNA* was detected as early as 8 h post-dose, peaked, and approached baseline levels by 10 d. Figure 5a shows the accretion of DHA* in the fetal liver resulting from doses of either LNA* or DHA*. LNA*-derived DHA accretion rises to a plateau at about 10 d, whereas DHA* from the DHA* dose appears to peak at much higher levels within a few days. The preformed DHA peak was modeled with a polynomial trend line whereas the LNA-derived DHA data were modeled with a logarithmic trend line, both plotted through the experimental points. Ratios of trend lines were plotted for comparison of accretion levels between LNA or DHA doses. Although the ratio changes continuously for liver, we can estimate 33-fold as the maximum ratio for this organ.

Figure 4
figure 4

Fetal liver LNA concentration. Incorporation of LNA* into fetal liver in animals doses with LNA*, expressed as a percent of dose. Each point represents data from a single animal plotted against time between dose and sampling. Time points on d 1 are elevated, whereas later time points drop in concentration and approach baseline, suggesting rapid turnover. The plotted curve is an exponential least squares fit through the data.

Full size image
Figure 5
figure 5

Fetal organ DHA accretion. Whole organ DHA* accretion levels due to doses of either LNA* or DHA*, expressed as a percentage of dose, with least squares trend lines plotted through the data. Each point represents a separate animal. (a) Liver: LNA-derived DHA (18:3n-3* dose) levels are about 33-fold lower than DHA* levels resulting from preformed DHA*(22:6n-3* dose). (b) Brain: Logarithmic least squares curves are plotted through both data sets to about 34 d post-dose fit the data well (DHA dose: r2 = 0.98, LNA dose: r2 = 0.91), suggesting very slow turnover. The ratio of the fitted curves plotted against the right axis yields an average of about 20, indicating preformed DHA brain accretion is about 20-fold greater than LNA-derived DHA. (c) Retina: Although the preformed DHA* curve shows some evidence of peaking around 15-20 d, a logarithmic fit is again plotted through both data sets. The ratio of curves is about 23.

Full size image

Fetal brain and retina accretion.Figure 5b shows the incorporation of DHA* in the fetal brain along with ratio data. Once again, the incorporation of preformed DHA* was always greater than the incorporation of LNA*-derived DHA*. For both groups, the accretion tended to plateau, and accordingly they were fit with logarithmic trend lines. The ratio of the two lines plateaus at a value corresponding to about 20-fold greater accretion of preformed DHA* compared with LNA*-derived DHA. Overall about 1.6% of the preformed DHA* dose was found in the fetal brain, which was higher than that found in the liver.

Accretion in the fetal retina is presented in Figure 5c. Preformed DHA* levels were again substantially greater than for the LNA*-derived DHA*. The retina is a very small organ and may be more sensitive to experimental conditions than is the brain. We therefore plotted logarithmic trend lines through both data sets even though the preformed DHA* curved shows evidence of a peak at about 20 d. The estimated average accretion ratio for the retina is about 23-fold.

Saturated, monounsaturated, and other n-3 fatty acids were also enriched in the fetal tissues (data not shown), as we have observed previously in chow-fed rhesus monkey fetuses(34). Label detected in 16:0 indicates that carbon from the dosed LNA* and DHA* was used for de novo fatty acid synthesis. Quantitative estimates were not performed.

DISCUSSION

It is known that 9-desaturase gene expression responds rapidly to diet in rats(35), and there is limited data to suggest that elongation and desaturation of n-3 fatty acids is down-regulated in primates in response to dietary n-3 LCP within 3 wk(36). Commercial non-human primate chow usually contains fish meal as a protein source, which is an incidental source of n-3 LCP. The defined diet provided to the animals had minimal but adequate amounts of LA and LNA and no LCP. Thus, maximally efficient conversion and transport of LNA to DHA should result, because no down-regulation of desaturation/elongation is expected due to LCP via product inhibition. Intravenous introduction of a FFA dose circumvents experimental difficulties associated with quantitative oral administration to adult primates, including issues of malabsorption. At the same time the dose is introduced into the mother's bloodstream in a prominent physiologic form. Most dietary essential fatty acids are delivered to the blood stream as triglycerides within chylomicrons, where they are liberated by lipoprotein lipase as FFA, transported via albumin, and taken up by liver and peripheral tissues. Thus, administration of FFA into the bloodstream using Intralipid carrier as in these studies is for practical purposes nearly equivalent to dietary consumption.

During the period of study, the fetuses are undergoing their brain growth spurt and can be considered a model of essential fatty acid metabolism for the human brain growth spurt. The most important fetal target organ is the brain, where we find a bioequivalence of about 20:1 between preformed DHA and LNA-derived DHA.

In this study, we had access to the fetal femoral artery and vein for sampling of fetal blood in utero for several animals. The lower concentration of precursor fatty acids, LA and LNA, in the fetal plasma compared with the maternal plasma and the higher concentration of arachidonic acid and DHA in the fetus compared with the mother are consistent with previous reports in human cord blood(37). However, the LNA and DHA concentrations were not significantly different between the mother and fetus, possibly due to the transfer of n-3 LCP from the mother to the fetus during this period of rapid brain growth.

The disappearance curves in the maternal and fetal plasma indicate that LNA* equilibrated between the mother and fetus and was cleared from both plasma pools much faster than DHA*. In both animals studied, the enrichment of DHA in the maternal plasma was higher than the enrichment of DHA in the fetal plasma until approximately 10 d after the dose. Furthermore, DHA* could be detected in the plasma until the end of the study at 35 d post-dose. Overall, these data suggest that a maternal bolus of DHA* is available to the fetus for a much longer period of time than is LNA* and explains at least in part the higher levels of preformed DHA* incorporated into fetal tissues compared with LNA*-derived DHA*. The enrichment levels of LNA, 22:5n-3, and DHA are similar in the maternal and fetal plasma of the LNA*-dosed animals(Fig. 3), indicating rapid equilibration of label between these two pools. In contrast, the greater concentration of preformed DHA* in the mother compared with the fetus may indicate that a concentration gradient helps drive transport of DHA across the placenta.

The detection of LNA* and DHA* in fetal baboon tissues after a dose of LNA* or DHA* to the mother indicates that these two fatty acids are transferred intact to the fetus from the mother and incorporated into fetal tissues within 1 d of the dose. Labeled n-3 LCP appear in maternal plasma within a day and are easily detected in fetal tissues, confirming a maternal role in conversion. Because of maternal conversion, these data do not directly show that the fetus has conversion capacity, although it is expected from studies of human infants(9) and the known ontogeny of liver and brain desaturase activity in the rat(38). The higher level of preformed DHA* incorporation into fetal organs is strong evidence for a high dependence of the fetus on maternal conversion, particularly because endogenous LNA concentrations in plasma are always low compared with DHA, in this study 0.4% for LNA versus 5.7% for DHA of total fatty acids.

The brain incorporated the greatest absolute fraction of the dose at most time points for both groups, in part because of its size. At the maximum incorporation of label, the brains of the LNA*- and DHA*-dosed animals incorporated approximately 0.075 and 1.6% of the dose, respectively.

Clinical manifestations of dietary deficiency of n-3 fatty acids are most prominent in brain and retina function, and these deficiencies are usually observed in developing animals. The recognized marker forn-3 fatty acid deficiency is a concomitant rise in 22:5n-6 and fall in DHA in neural tissues(23). It has long been known that experimental induction of n-3 fatty acid deficiency by dietary means is extremely difficult to produce in adults(39) and requires long periods of deprivation. Furthermore, brain DHA concentrations are remarkably constant across species(40). These observations are evidence for very tight control and very slow turnover of brain DHA. The data ofFigure 5b are strong direct evidence in favor of this hypothesis. In contrast, the retinal DHA* accretion data ofFigure 5c may indicate more rapid turnover, although we have plotted logarithmic trend lines. Retinal DHA metabolism is extremely active and necessarily efficient, as photoreceptor membranes turnover at a very high rate(41), but recent data in rats indicate the retinal fatty acid composition is substantially more plastic than that of the brain(42).

Estimate of dietary LNA and DHA requirements. An estimate of the dietary requirement of both DHA and LNA can be made based on the tracer accretion data. The baboon brain grows very little in the first half of gestation but gains weight at an almost linear rate over the last half of gestation. At parturition after about 182 dGA, the fetal baboon brain is about 80 g(43). Our data show that baboon brain DHA is about 3.35 μmol of DHA/g of wet weight and is not related to gestational age. At term, the baboon brain has accumulated a total of 268 μmol of DHA. The second half of gestation lasts about 91 d, giving a brain demand of 2.96μmol of DHA/day.

At plateau, 0.075% of the LNA dose was incorporated as DHA into the fetal brain. The diet contained 0.24%en LNA, and we estimate, based on typical maintenance diets, an overall consumption of 2086 μmol of LNA (580 mg)/d by the pregnant baboons. Combining these factors we find that 1.56 μmol DHA/d of brain DHA accretion can be ascribed to dietary LNA, or about 53% of brain DHA, with the remaining 47% derived from maternal stores. Further, we can estimate that an increase in dietary LNA level by a factor of about 1.9, resulting in an overall LNA level of about 0.45%en, would be sufficient to meet all fetal requirements in the absence of dietary n-3 LCP. These considerations apply strictly only to the dietary n-6/n-3 ratio in the present diet. A higher ratio might be expected to reduce overall conversion/accretion in primates, even though our previous evidence shows it does not in rats(44). However, the present dietaryn-6/n-3 ratio is at the high end of the range of the industrialized free-living population and is approximately that in commercial term infant formula. Because of the competition for desaturation and elongation between LA and LNA, ratios lower than those considered, such as would result from increasing LNA alone, may result in higher levels of conversion to n-3 LCP and correspondingly lower minimum daily requirements. Finally, from our relative efficacy factor of 20:1, we can estimate that dietary DHA at about 0.03%en would meet the needs of the fetal brain.

The strict animal-protein-free vegan diet based on common terrestrial foods is practically devoid of LCP and therefore requires that individuals derive all LCP by desaturation/elongation. The LCP-free and very low fat composition of our experimental diets mirrors this requirement within the constraint that the pregnant animals were replete in DHA until starting the diet several weeks into pregnancy. Our estimate of 0.45%en is a modest requirement, and our data are therefore consistent with the hypothesis that the pregnant vegan can meet the requirements for fetal brain growth. The presence of replete body stores of DHA in our baboons would if anything suppress conversion of LNA to DHA, further indicating a fully adequate maternal/fetal desaturation/elongation system.

The higher accretion efficiency of preformed dietary DHA compared with LNA-derived DHA implies high efficiency of transfer by both the pregnant animal and the fetus. However, the role of the fetus in synthesis of DHA from LNA cannot be separated from that of the pregnant baboon from our measurements. Determination of this capacity can be ascertained only from direct dosing to the fetuses, which will permit assessment of the ontogeny of the desaturation/elongation in fetal primates. Such measurements should bear directly upon requirements of preterm infants, a group which is widely thought to be at risk for DHA deficiency(6, 17).

In conclusion, this study shows that LNA and DHA are transferred to the fetus from the mother, and preformed DHA is incorporated into fetal tissues at least one order of magnitude higher than is LNA-derived DHA. About 0.075 and 1.6% of the dose of LNA and DHA, respectively, was incorporated into the brain as DHA. Based on this level of DHA brain accretion, we estimate that LNA provided as 0.45%en in an LCP-free diet would support DHA requirements of the brain. Future investigations focusing on fetal conversion of LNA to DHA are required to determine whether the mother is the sole source of DHA in the latter half of gestation.