PEMT, Δ6 desaturase, and palmitoyldocosahexaenoyl phosphatidylcholine are increased in rats during pregnancy[S]

DHA is important for fetal neurodevelopment. During pregnancy, maternal plasma DHA increases, but the mechanism is not fully understood. Using rats fed a fixed-formula diet (DHA as 0.07% total energy), plasma and liver were collected for fatty acid profiling before pregnancy, at 15 and 20 days of pregnancy, and 7 days postpartum. Phosphatidylethanolamine methyltransferase (PEMT) and enzymes involved in PUFA synthesis were examined in liver. Ad hoc transcriptomic and lipidomic analyses were also performed. With pregnancy, DHA increased in liver and plasma lipids, with a large increase in plasma DHA between day 15 and day 20 that was mainly attributed to an increase in 16:0/DHA phosphatidylcholine (PC) in liver (2.6-fold) and plasma (3.9-fold). Increased protein levels of Δ6 desaturase (FADS2) and PEMT at day 20 and increased Pemt expression and PEMT activity at day 15 suggest that during pregnancy, both DHA synthesis and 16:0/DHA PC synthesis are upregulated. Transcriptomic analysis revealed minor changes in the expression of genes related to phospholipid synthesis, but little insight on DHA metabolism. Hepatic PEMT appears to be the mechanism for increased plasma 16:0/DHA PC, which is supported by increased DHA biosynthesis based on increased FADS2 protein levels.

the maternal-fetal transfer between NEFA pools (2,7,11), the placenta has lipase activity (16)(17)(18)(19)(20). Therefore, other lipid pools in plasma could be a source of DHA. Plasma lipids include different lipid classes, such as triacylglycerols (TAGs), phospholipids, and cholesteryl esters. The longchain PUFAs (LC-PUFAs), such as DHA and ARA, tend to concentrate in the phospholipids (21), which in plasma is predominantly composed of phosphatidylcholine (PC) making up the monolayer of lipoproteins (22) and hepatic lipoprotein production is increased during pregnancy (23). PC is synthesized de novo through the Kennedy pathway or by conversion of phosphatidylethanolamine (PE) through methylation by PE methyltransferase (PEMT) (24). Hepatic PE has higher DHA content, as compared with hepatic PC (25)(26)(27), and estrogen has been shown to upregulate the PE methylation pathway (28). Additionally, fetal Pemt / mice have lower levels of brain DHA (29), plasma levels of DHA in PC have been proposed as a surrogate marker of hepatic PEMT activity in humans (30), and PC derived from PEMT during pregnancy appears to be selectively transferred to the fetus (31).
In the current study, we set out to fully characterize the changes in DHA concentrations in plasma and liver in specific lipid classes before pregnancy, at 15 and 20 days of pregnancy, and 7 days postpartum in Sprague-Dawley rats. We hypothesized that hepatic PEMT would be associated with changes in DHA in plasma PC, therefore, expression and activity were examined. Hepatic enzymes involved in DHA biosynthesis were also examined across all of these time points. Based on the initial findings of a large increase in plasma DHA from 15 to 20 days of pregnancy, additional lipidomic and transcriptomic measurements were completed at these time points. This resulted in the discovery that 16:0/DHA PC was responsible for most of the increases in plasma and liver DHA. PEMT appeared to be involved in DHA mobilization during pregnancy by selectively producing 16:0/DHA PC. We confirmed that increased DHA biosynthesis from n-3 precursors likely occurs during pregnancy based on increased 6 desaturase (FADS2) protein levels. However, we cannot rule out that other mechanisms and adaptations could be involved in increasing maternal circulating DHA during pregnancy. Understanding maternal adaptations to increase DHA bioavailability for fetal delivery is necessary to determine precise maternal dietary DHA requirements in order to establish evidence-based recommendations to support pregnancy and fetal development.

Study design
All animal procedures were approved by the University of Waterloo Animal Care Committee and are in accordance with the guidelines of the Canadian Council on Animal Care. Twenty-four female Sprague-Dawley rats were purchased at 7 weeks of age and mated with 6-month-old proven male breeders (Envigo, Mississauga, Ontario, Canada) at the Central Animal Facility on campus. All rats were fed a commercial fixed-formula rodent diet (8640 Teklad 22/5 Rodent Diet; Envigo) providing 3.0 kcal/g with 17% energy from fat, 29% energy from protein, and 54% energy from carbohydrate throughout the study. The fatty acid composition of the diet was determined by gas chromatography (see details of the methods below). Pregnant rats (n = 6) were euthanized at baseline (nonpregnant rats after 7 days of acclimatization), day 15 and day 20 of pregnancy, and 7 days postpartum by exsanguination following anesthesia using isoflurane after an overnight fast. Exsanguinated blood was collected in the presence of EDTA and plasma was isolated by centrifugation at 1,500 g and the liver was excised, washed in saline (0.9% v/v), and weighed. Samples were flash-frozen in liquid nitrogen and stored at 80°C for various analyses.

The 17-estradiol concentrations
Plasma 17-estradiol concentrations were determined by ELISA (Estradiol EIA kit; Cayman Chemical, Ann Arbor, MI). Briefly, 17-estradiol was extracted from plasma with methylene chloride, and then reconstituted in EIA buffer, loaded to the 96-well plate in duplicate with standards and absorbance at 420 nm wavelength, measured by spectrophotometry, and concentrations determined by comparison to a standard curve.

Gene expression and protein determinations
Gene expression was determined by quantitative real-time PCR and protein levels were determined by immunoblotting according to previous procedures (32). For mRNA expression, liver samples were homogenized in Trizol reagent (Invitrogen, Frederick, MD) and chloroform was used to separate RNA from other cellular components. The integrity of isolated RNA was assessed by measuring intact 18S and 28S rRNA after agarose gel electrophoresis with ethidium bromide. The concentration of the samples was then confirmed using absorbance at 260 nm, while the 260/280 ratios were used to determine sample purity, as determined by a NanoDrop spectrophotometer. Only RNA samples with high purity (260/280 ratio above 1.90) were used for subsequent cDNA synthesis by reverse transcriptase. Primers were designed for Pemt through the Primer-BLAST program on the NCBI website and ordered from Sigma-Aldrich, Oakville, Ontario, Canada (5′-CCCAGCTTTGTGGCGGCTGT-3′). Expression was determined by SYBR Green qPCR master mix and an Applied Biosystems 7500 real-time PCR system (Streetsville, Ontario, Canada) with software version 1.2.3 using 25 l as the reaction volume with an initial incubation at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s, and 60°C for 1 min. A dissociation curve was obtained by increasing the temperature from 60°C to 95°C at 1°C/min. The threshold cycle number (Ct) was determined and normalized to the Ct observed for Gapdh, a housekeeping reference gene during pregnancy (33). The expression of target genes was determined using the 2 -Ct method and the values were expressed as the mean expression relative to baseline.

Enzyme activity of PEMT
A liver PEMT activity assay was performed as described by Ridgway and Vance (34). Briefly, to assess activity, we used 50 g of protein homogenized in Tris-HCl per liter (pH 9.2) and 5 mm DTT buffer per liter (Sigma-Aldrich). Samples were incubated with 200 mol S-adenosyl-L-methionine per liter containing 0.5 Ci S-adenosyl-L-[methyl- 3 H]methionine (55.70 Ci/mmol) and 0.4 nmol exogenous phosphatidyl dimethylethanolamine per liter (Avanti Polar Lipids, Alabaster, AL). The reaction was carried out for 60 min at 37°C and then stopped by adding ice-cold CHCL 3 :methanol:1 N HCL (100:50:1, v/v). Samples were vortexed, washed with 0.1 M KCL in 50% methanol and centrifuged, and the organic phase was collected. Samples were then dried under nitrogen gas and resuspended in chloroform. Samples were then applied to and separated on a Silica Gel GTLC plate (Analtech Inc., Newark, DE) using CHCL 3 :methanol:acetic acid:water (50:30:5:2, v/v/v/v). The PC band was identified and collected using a reference standard (Sigma-Aldrich) and radioactivity counted by liquid scintillation (Beckman LS6500; Beckman Coulter, Mississauga, Ontario, Canada). PEMT activity was then calculated as PC formed (picomoles per milligram per hour).

Fatty acid composition determinations
The fatty acid composition of total lipids for the rodent diet and total lipids and various lipid classes of plasma and liver were determined. Internal standards used for quantitation of fatty acids were as follows: total lipids [docosatrienoic acid (22: . Fatty acids were extracted by using 2:1 (v/v) chloroform:methanol (35) and 50 g/ml butyl-hydroxytoluene. The organic phase containing the lipids was isolated by the addition of sodium phosphate buffer followed by centrifugation. Fatty acid methyl esters were then prepared from the total lipid extracts of the rodent diet and tissues. In addition, total lipid extracts of plasma and liver were aliquoted for neutral and polar lipid separation techniques by thin-layer chromatography.
The fatty acyl groups in total lipid and lipid class extracts were transesterified to fatty acid methyl esters using 14% BF 3 in methanol with hexane for 1 h at 95°C. Additional hexane and a water wash were used to collect the fatty acid methyl esters for analysis by gas chromatography with flame ionization detection. Briefly, a fast gas chromatography protocol was employed (37) using a Varian 3900 gas chromatograph equipped with a DB-FFAP 15 m × 0.10 mm injected dose × 0.10 m film thickness nitroterephthalic acid-modified polyethylene glycol capillary column (J&W Scientific/Agilent Technologies, Mississauga, Ontario, Canada), with hydrogen as the carrier gas and temperature ramping design to maximize peak resolution (38). Individual fatty acid peaks were identified by comparison to a reference mixture of fatty acids (GLC-462; Nu-Chek Prep Inc.), which was also used to calibrate fatty acid response factors on the gas chromatograph. Fatty acid data was expressed as concentrations and relative weight percentages for proper interpretation (3).

Semi-targeted lipidomic analyses
After initial fatty acid composition data was analyzed, ad hoc semi-targeted lipidomic analyses were performed on plasma (PC) and liver (PC and PE) samples for day 15 and day 20 of pregnancy    Lipids were separated with a high-performance liquid chromatograph (Agilent 120 series with binary pump; Agilent Technologies, Santa Clara CA) equipped with a Phenomenex Kinetex C18 column with dimensions of 100 × 2.1 mm × 2.6 m (Phenomenex, Torrance, CA). Mobile phase A consisted of water/ acetonitrile/methanol (2/1/1, v/v/v) + 0.05% formic acid and mobile phase B consisted of tetrahydrofuran/acetonitrile/methanol (2/1/1, v/v/v) + 0.05% formic acid using a gradient of 40% to 85% B over the course of 11 min with a flow rate of 0.4 ml/min.
The mass spectrometer (API4000 triple-quadrupole; AB SCIEX, Concord, Ontario, Canada) was operated in positive electrospray ionization mode with multiple reaction monitoring. Analytespecific precursor to product ion mass transitions were chosen for each lipid. For PC species, a precursor of M+H with a product ion of m/z 184.1 (representing the PC head group) was used. To identify PE species, neutral loss scanning for m/z 141.02 (representing the PE head group) was used. Each PC or PE was then uniquely identified by the precursor to product ion mass transitions and chromatographic retention times. The presence of DHA acyl chains in PC and PE were confirmed in positive MS/MS mode by looking for lysoPC DHA and lysoPE DHA in the product ion spectra. PC data was expressed semi-quantitatively and PE data as relative response ratios.

Transcriptomic microarrays
Similarly, ad hoc transcriptomic analyses were performed on liver samples for day 15 and day 20 of pregnancy only (n = 4 for each group) after initial quantitative (q)PCR data was analyzed. RNA was extracted from livers using Trizol reagent, as described previously (39). Purity of extracted RNA was quantified using the 260/280 ratio on a Nanodrop 2000c prior to shipment to the University of Guelph (Guelph, Ontario, Canada). At Guelph, the integrity of the RNA was confirmed with an Agilent BioAnalyzer (Agilent) prior to hybridization to Affymetrix Rat Gene 2.1 ST array strips (Affymetrix, Fremont, CA), as described previously (40). Briefly, total RNA was used to synthesize cDNA, followed by cRNA synthesis. Second cycle cDNA was then synthesized, fragmented, labeled with biotin, and hybridized to microarrays. Strips were washed, stained, and scanned on the Affymetrix Gene Atlas platform.
Microarray data were corrected for background noise, quantile normalized using RMA, and log transformed (JMP Genomics version 5; SAS, Cary, NC). Principal component analysis was used to assess within and between group variability. ANOVA was used to identify differentially expressed genes with a false discovery rate (FDR) of 0.05 to control for multiple testing. A gene expression list including all genes with P < 0.05 (i.e., not adjusted for multiple testing) was also examined to obtain a more global perspective of the biological pathways effecting DHA metabolism during the late phase of pregnancy. Hierarchical clustering analysis confirmed the two conditions clustered together and did not overlap. FunNet was used to identify biological pathways that were differentially regulated in our gene expression list (41) and characterized into biological themes using pathways established from the Kyoto Encyclopedia of Genes and Genomes (KEGG).

Statistics
The one-way ANOVA test was used to determine differences in mRNA, protein, and fatty acids at different stages of pregnancy in rat models with Tukey post hoc testing to compare individual means after a significant F-value. For lipidomic data, unpaired Values are expressed as mean ± SD, n = 6 for each group. Fatty acids with different superscripts are significantly different by Tukey's post hoc test (P < 0.05) following significant F-value by one-way ANOVA. SFAs, saturated fatty acids.
t-tests were used to compare day 15 and day 20 of pregnancy. Statistical significance was assumed with P < 0.5. Statistical methods used for transcriptomics are described above. Data are presented as mean ± SD.
PEMT expression and activity were increased during pregnancy (Fig. 3). For mRNA, expression was increased at 15 days of pregnancy, as compared with all other time points. Differences in protein levels were subtle with measures at 20 days of pregnancy being higher than at baseline and 7 days of postpartum, but similar to levels at 15 days of pregnancy. Activity levels resembled the mRNA response with a significant increase at 15 days of pregnancy only.
There was also evidence of increased expression of some of the genes involved in PUFA biosynthesis. Protein levels of FADS2 were 45% higher at 20 days of pregnancy, as compared with baseline rats, and returned to baseline levels on 7 days postpartum (Fig. 4A). FADS2 levels at 15 days of pregnancy were intermediate between baseline and 20 days of pregnancy. In contrast, protein levels of ELOVL5 were decreased 30% between post conception days 15 and 20 and returned to baseline levels on day 28 (Fig. 4C). No effects of pregnancy were observed in protein levels of ELOVL2, MFP2, or FADS1.
Pregnancy and postpartum had various effects on the fatty acid composition of the various lipid fractions in plasma and liver. In general, plasma total fatty acid concentrations were increased by 140% at 20 days of pregnancy, relative to baseline (P < 0.05 by Tukey's post hoc after significant F-value by one-way ANOVA), while liver total fatty acids did not differ (data not shown). The increases in plasma total fatty acids at 20 days of pregnancy were also observed in the plasma PC and TAG, but not in the PE fraction (Tables 1-3). DHA concentrations in plasma total lipids increased even more dramatically, particularly at the later stages of pregnancy, with a 1.85-fold increase at Values are expressed as mean ± SD, n = 6 for each group. Fatty acids with different superscripts are significantly different by Tukey's post hoc test (P < 0.05) following significant F-value by one-way ANOVA. SFAs, saturated fatty acids. 15 days and a 4.75-fold increase at 20 days of pregnancy, relative to baseline (Fig. 5A). While DHA concentrations in plasma TAG and PE also increased at 20 days of pregnancy, their contribution to total DHA in plasma was relatively minor. Therefore, changes in DHA concentrations in the PC fraction, especially during late pregnancy, were largely responsible for the changes observed in the total lipid fraction.
The percentage of DHA in total fatty acids tended to be increased in plasma lipid fractions (Tables 1-3). In plasma PC, the percentages of DHA and 22:5n-6 were increased and the percentages of EPA and ARA were decreased at 20 days of pregnancy (Table 1). This pattern was also evident in plasma PE and TAG, but only the increases in DHA and decreases in EPA were significant in PE (Table 2), and ARA was not decreased in TAG (Table 3). In addition, the percentages of total saturates increased and total monounsaturates decreased in both of the plasma phospholipid fractions at 20 days of pregnancy, but this tended to be reversed in plasma TAG, as the percentage of total saturates was the lowest at 20 days of pregnancy.
In liver, despite the lack of change in liver total fatty acids, DHA concentrations and relative percentages tended to be slightly increased during pregnancy. Again this increase in total DHA concentrations was mainly due to increases in DHA in the PC fraction where concentrations were approximately 50% higher at 20 days of pregnancy as compared with baseline and 7 days postpartum (Fig.  5B), while there was no difference in DHA concentrations in PE and DHA concentrations in liver TAG decreased from baseline through pregnancy and the postpartum (Fig. 5B).
Some of the patterns observed in percentage changes in plasma were also evident in liver, as 22-carbon PUFA tended to be increased and 20-carbon PUFA decreased at 20 days of pregnancy in PC and PE. However, shifts in total saturates and monounsaturates were not consistent across liver lipid fractions and were not consistent with the changes observed in plasma, although a shift toward increased 16:0 and decreased 18:0 at 20 days of pregnancy appeared to be consistent in the phospholipid fractions (Tables 4-6). Interestingly, liver TAG levels of DHA were not increased at 20 days of pregnancy and appeared to be transitioning toward the significantly lower levels observed at 7 days postpartum ( Table 6). DHA concentrations in the cholesteryl ester and nonesterified fatty acid fractions in plasma and PS and PI in liver were below levels observed in the PE and TAG fractions in these tissues (data not shown).

Lipidomics
Based on the fatty acid analyses, semi-targeted analyses of plasma and liver PC were completed to determine the acyl species of lipids responsible for the large shift in DHA concentration between 15 days and 20 days of pregnancy. Values are expressed as mean ± SD, n = 6 for each group. Fatty acids with different superscripts are significantly different by Tukey's post hoc test (P < 0.05) following significant F-value by one-way ANOVA. SFAs, saturated fatty acids.

Transcriptomics
Using a FDR of 5%, 76 transcripts were differentially expressed, of which 57 had unique Gene IDs (supplemental Table S1). KEGG pathway analysis indicated that at 20 days of pregnancy, C21-steroid, cysteine and phenylalanine metabolism, and phenylalanine, tyrosine, tryptophan, and alkaloid biosynthesis pathways were upregulated. Downregulated pathways at 20 days of pregnancy included biosynthesis of steroids, metabolism of cytochrome P450, butanoate, pyruvate, propanoate, and linoleic acid, synthesis and degradation of ketone bodies, and degradation of valine, leucine, isoleucine, lysine -hexachlorocyclohexane, and benzoate. When expanding the list of differentially expressed genes by relaxing our statistical criteria to P < 0.01 without a FDR, 712 transcripts having unique Gene IDs were identified. A total of 413 genes were upregulated, which included choline phosphotransferase 1 (Chpt1) and diacylglycerol O-acyltransferase 1 (Dgat1), and 219 genes were downregulated that included phosphate cytidylyltransferase 2 ethanolamine (Pcyt2), and acyl-CoA synthetase long-chain family member 1 (Acsl1).

DISCUSSION
In the present study, changes in maternal plasma and liver DHA status in various lipid classes throughout pregnancy were examined in a rat model using fatty acid and lipidomic analyses. This is the most comprehensive report of the fatty acid composition of the lipid classes of plasma and liver throughout pregnancy to date and allowed us to perform more targeted lipidomic assessment to determine that 16:0/DHA PC is the specific lipid responsible for most of the DHA increases in liver and plasma during pregnancy. Based on a preliminary hypothesis that PEMT was responsible for the mobilization of hepatic DHA into the PC of lipoproteins, mRNA, protein, and enzyme activity were assessed. Dietary intake was also monitored and targeted protein assays were used to examine pathways related to DHA biosynthesis. In addition, transcriptomic screening was also used to identify potential gene products that could be involved in the large increase in DHA that was observed in circulating plasma phospholipids. We demonstrated that hepatic PEMT expression and activity is increased and appears to be involved in supporting the mobilization of maternal DHA to meet fetal demand by selectively increasing plasma 16:0/DHA PC between 15 and 20 days of pregnancy. We also report increased protein concentrations of FADS2 during pregnancy for the first time, which confirms a previous observation by mRNA (42). This observation and shifts in the n-3 PUFA pool in the liver and plasma toward higher amounts of DHA and fewer n-3 PUFA precursors suggest that increased DHA biosynthesis during pregnancy through upregulated FADS2 supports the production of 16:0/DHA PC. In our model, we showed that DHA increases in maternal plasma during pregnancy, which has been observed previously in humans (5)(6)(7)(8) and rats (14,42). In rats, the increase in plasma DHA occurs primarily in late pregnancy, as we and others (42) found an increase in DHA in plasma at 20 days of pregnancy, but not earlier (12-15 days of pregnancy). While DHA increased in various lipid classes in plasma, the increase in DHA in plasma phospholipids accounted for 82% of the increase in DHA in all plasma lipids from 15 to 20 days of pregnancy. Ad hoc lipidomic analyses identified two DHA-containing PCs, with 18:0/ DHA PC increasing by 69% and 16:0/DHA PC increasing by 288%. While an increase in DHA in plasma phospholipids has been documented (8,11), most of the interest in the supply of DHA for maternal-fetal transfer of DHA has been focused on plasma NEFA and TAG (7,11). This appears to be based on placental transfer studies and documented maternal lipidemia (2,7,11); however, the importance of DHA accumulation in plasma phospholipids has been documented (43). The focus on NEFA and TAG has also led to a focus on maternal adipose stores as a mobilizable source of DHA during pregnancy (2,14), despite the fact that these lipid pools tend to have relatively low concentrations of DHA (42,44,45). Our acyl species analysis indicating the specific increase in plasma 16:0/ DHA PC suggests considerable hepatic involvement of DHA mobilization and a mechanism where DHA enrichment of PC in the lipoprotein monolayer may be a delivery mechanism. The placenta can bind lipoproteins (46) and placental lipases can release NEFA from the TAG of lipoproteins by lipoprotein lipase (16,19). However, the placenta is known to have considerable endothelial lipase activity, which has a considerable phospholipase activity that would generate NEFA and lysoPC (17,18,20). It has also been proposed that docosahexaenoyl lysoPC could be a carrier for placental transport of DHA (47) based on studies of brain uptake of DHA (48,49), although it has also been proposed that DHA NEFA is the main source for brain uptake (50). Regardless, 16:0/DHA PC has the potential to serve as a source of DHA for placental transport.
Hepatic involvement has been proposed previously, with the liver either repackaging circulating NEFA into TAG (7,11) and/or biosynthesis of DHA from precursor omega-3 PUFA released from adipose, such as EPA and 18:3n-3 (42). In the present study, it appears that hepatic TAG DHA could be getting mobilized to support DHA incorporation into PC. It has been demonstrated that estrogen is linked to increased DHA biosynthesis (12), and that there is a shift in human plasma toward 22-carbon PUFA at the expense of 20-carbon PUFA during pregnancy (6). These patterns were observed presently at day 20 of pregnancy, as FADS2 protein levels were increased and the percentages of DHA and 22:5n-6 were increased and the percentages of EPA and ARA were decreased in plasma PC. We did not Values are expressed as mean ± SD, n = 6 for each group. Fatty acids with different superscripts are significantly different by Tukey's post hoc test (P < 0.05) following significant F-value by one-way ANOVA. SFAs, saturated fatty acids.
observe changes in the protein concentrations of FADS1, ELOVL2, and MFP2, while ELOVL5 concentrations were actually lower at 20 days, as compared with 15 days of pregnancy. These observations are consistent with previous research in this area examining the effect of sex hormones on these enzymes, with differences in mRNA expression occurring in several enzymes, but protein concentrations being restricted to an increase in FADS2 and a slight decrease in ELOVL5 (25,32).
The lack of a FADS1 response seems to agree with the observations that EPA was decreased at day 20, while 18:3n-3 was relatively unchanged during pregnancy in various plasma and liver lipid pools. The increase in the FADS2 protein and higher DHA and 22:5n-6 levels suggest an adaptation during pregnancy to increase the ability of FADS2 to engage in what is considered a second round of desaturation. This is generally considered a 6 desaturation of a 24-carbon PUFA that is then peroxisomally -oxidized back to a 22-carbon PUFA; however, it is possible that the upregulated FADS2 could be catalyzing a 4 desaturation reaction (51). This latter possibility could explain the lack of upregulation of ELOVL2 and ELOVL5 that should produce 24-carbon PUFA. However, a consistent shift away from 20-carbon PUFA toward 22-carbon PUFA was observed without support from increased ELOVL2 and ELOVL5 expression. This confirms previous work indicating that gene expression of ELOVL2 and ELOVL5 may not reflect PUFA synthesis. Interestingly, ELOVL6 is consistently upregulated at the mRNA and protein level in response to female sex hormones. Further examination of PUFA elongation during pregnancy should consider a comprehensive assessment of the elongase isoforms and examine substrate specificity using activity assays.
Interestingly, the increase in DHA was much greater in circulating plasma as compared with the increase in DHA in the liver, while relatively modest changes in ARA in plasma appeared to be accompanied by decreased liver ARA. This was also reflected in the lipidomic analysis, as the 288% increase in 16:0/DHA PC and 69% increase in 18:0/DHA PC in plasma were accompanied by a 156% increase in 16:0/DHA PC and no changes in 18:0/DHA PC or DHA species in PE in liver. The shifts in ARA-containing phospholipid species were interesting to contrast, as 38:4 PC did not change in plasma, but both 38:4 PC and 38:4 PE decreased in liver. These results appear to suggest that both ARA and DHA are being mobilized from liver stores to the plasma, but additional mechanisms, such as a shift in PUFA biosynthesis toward 22-carbon, as described above, are occurring to provide additional DHA, while large hepatic ARA stores are mobilized instead. The responses of other PC species in plasma and liver and PE in liver provided support of hepatic mobilization of PE toward PE to support increased PC plasma content with pregnancyinduced hyperlipidemia. Values are expressed as mean ± SD, n = 6 for each group. Fatty acids with different superscripts are significantly different by Tukey's post hoc test (P < 0.05) following significant F-value by one-way ANOVA. SFAs, saturated fatty acids.
The observation of a specific increase in 16:0/DHA PC in plasma during pregnancy supports the hypothesis of an active role of PEMT in mobilizing DHA. PEMT is increased during pregnancy (31) and estrogen influences PEMT expression and activity (52). Potential estrogen response element motifs have been identified on the PEMT gene promoter region and treating hepatic cell cultures with estrogen at doses mimicking concentrations in humans (0-100 nmol/l) significantly upregulates mRNA expression and enzyme activity of PEMT (28). While the increased PEMT activity during pregnancy appears to be an evolutionary mechanism to meet the high demand for choline to support anabolic pathways (31), plasma PC content of DHA is also directly correlated with PEMT activity (30). Conversion of PE to PC may serve as a mechanism to mobilize hepatic DHA for fetal transport. PE has a higher content of DHA, which resides predominantly on the inner bilayer of membranes, as compared with PC (25)(26)(27). PEMT can then Values are expressed as mean ± SD, n = 6 for each group. Fatty acids with different superscripts are significantly different by Tukey's post hoc test (P < 0.05) following significant F-value by one-way ANOVA. SFAs, saturated fatty acids. convert DHA-rich PE to PC that can then be transferred to plasma as the principal phospholipid component of lipoproteins. PC derived from PEMT has been shown, in humans, to partition toward the fetus in a tracer study using choline labeled with deuterium (31). It has also been shown that PEMT has a high selectivity for converting 16:0/DHA PE to 16:0/DHA PC, but that the fatty acyls on the newly synthesized PC are remodeled to lower DHA content within 6-12 h (53). This remodeling could result in increased DHA hepatic pools of NEFAs and/or acyl-CoAs and be available for incorporation into other lipid classes. However, in our study, DHA enrichment was the largest in 16:0/ DHA, suggesting that DHA remains or is selectively reincorporated in this lipid during pregnancy. Unfortunately, comparing the hepatic transcriptomes between 15 and 20 days of pregnancy was not successful in identifying changes in the transcriptome that could directly explain the increase in plasma 16:0/DHA PC. This is likely due to the fact that numerous metabolic changes occur during pregnancy, which may consequently mask subtler changes in the transcriptome. In addition, our analysis was limited to a comparison between 15 days and 20 days of pregnancy. We cannot exclude that changes in mRNA expression related to 16:0/DHA PC synthesis may have occurred in the period between baseline and 15 days of pregnancy. When a less conservative statistical approach was used (P < 0.01 without FDR), there was some evidence supporting increased PC synthesis (choline phosphotransferase 1 was upregulated) and decreased PE synthesis (phosphate cytidylyltransferase 2 ethanolamine was downregulated).
We have provided evidence that the increased DHA in plasma that is observed during pregnancy is largely driven by a specific increase in 16:0/DHA PC. It also appears that the plasma increase in 16:0/DHA PC is at least partially mediated by the conversion of DHA-rich PE to PC through increased expression and activity of PEMT in the liver. Other metabolic pathways, such as DHA biosynthesis, also appear to support the increase in DHA in plasma, but hepatic transcriptomics did not identify changes in genes coding for enzymes involved in fatty acid transport and incorporation into complex lipids. This may be due to our decision to compare 15 days of pregnancy with 20 days of pregnancy rather than comparisons to baseline, as changes in mRNA expression may occur before the biochemical phenomenon. Further analyses at additional time points between baseline and 15 days of pregnancy are warranted to investigate changes in the expression of genes related to hepatic fatty acid and lipid metabolism. Given the large increase in 16:0/DHA PC in plasma, it also appears that future studies examining maternal hepatic adaptations during pregnancy should consider the role of other maternal tissues as a potential source of DHA. In addition, manipulations of dietary levels of DHA and other n-3 PUFAs during pregnancy could reveal additional maternal adaptations. A comprehensive understanding of maternal adaptations is required to understand the resiliency in maternal-fetal DHA transport in order to establish dietary DHA requirements during pregnancy.