Multi-omic profiles of hepatic metabolism in TPN-fed preterm pigs administered new generation lipid emulsions[S]

We aimed to characterize the lipidomic, metabolomic, and transcriptomic profiles in preterm piglets administered enteral (ENT) formula or three parenteral lipid emulsions [parenteral nutrition (PN)], Intralipid (IL), Omegaven (OV), or SMOFlipid (SL), for 14 days. Piglets in all parenteral lipid groups showed differential organ growth versus ENT piglets; whole body growth rate was lowest in IL piglets, yet there were no differences in either energy expenditure or 13C-palmitate oxidation. Plasma homeostatic model assessment of insulin resistance demonstrated insulin resistance in IL, but not OV or SL, compared with ENT. The fatty acid and acyl-CoA content of the liver, muscle, brain, and plasma fatty acids reflected the composition of the dietary lipids administered. Free carnitine and acylcarnitine (ACT) levels were markedly reduced in the PN groups compared with ENT piglets. Genes associated with oxidative stress and inflammation were increased, whereas those associated with alternative pathways of fatty acid oxidation were decreased in all PN groups. Our results show that new generation lipid emulsions directly enrich tissue fatty acids, especially in the brain, and lead to improved growth and insulin sensitivity compared with a soybean lipid emulsion. In all total PN groups, carnitine levels are limiting to the formation of ACTs and gene expression reflects the stress of excess lipid on liver function.

In NAFLD, free fatty acids can induce reactive oxygen species causing mitochondrial dysfunction, including reduced fatty acid -oxidation (19). This reduced -oxidation capacity can trigger a compensatory increase in activity of hepatic peroxisomal and microsomal fatty acid oxidizing enzymes, like cytochrome P450 2E1 (CYP2E1), that generate large amounts of hydrogen peroxide (20). A key protective function of vitamin E is its strong antioxidant activity that confers defense against reactive oxygen species-mediated lipid peroxidation (21). In addition, there is evidence suggesting that natural vitamin E (RRR--tocopherol and - and -tocopherol) can act as ligand for the pregnane X receptor and activate xenobiotic metabolic pathways (22). Treatment with the pregnane X receptor agonist, rifampicin, has been shown to be beneficial in the treatment of pruritus of cholestasis (23). The effect of new generation lipid emulsions on the vitamin E status in preterm infants and enrichment in tissue, such as the liver, is poorly understood.
In our previous report, we established that new generation lipid emulsions containing fish oil and vitamin E lead to decreased indices of liver damage and lipid accumulation (7). In this report, we have taken a multi-omics approach to examine the effect of those lipid emulsions on whole body metabolism, fatty acid utilization, and lipidomic and metabolomic profiles in the liver and muscle tissue in our PN-fed piglet model. Our second aim was to characterize the impact of these different lipid emulsions on hepatic gene networks using a genome-wide transcriptomic analysis.

Animals and surgery
The study protocol was approved by the Animal Care and Use Committee of Baylor College of Medicine and was conducted in accordance with the Guide for the Care and Use of Laboratory Animals (Division of Receipt and Referral/National Institutes of Health, Bethesda, MD). Sows were obtained from the Texas Department of Criminal Justice (Huntsville, TX) and housed in the Children's Nutrition Research Center. Our surgical protocol has been previously described (7). Briefly, domestic crossbred piglets were delivered 7 days preterm on gestation day 108 by caesarian section. Piglets were administered injections of iron dextran (100 mg/ml), ampicillin (125 mg/ml), and buprenorphine (0.3 mg/ml) prior to surgery. Piglets received either ENT nutrition or three forms of TPN. Saline or TPN was administered via surgically implanted jugular and umbilical arterial catheters. Enterally fed piglets were implanted with an orogastric feeding tube and TPN groups received a sham puncture.

Nutritional support and study design
TPN consisted of a solution containing a complete nutrient mixture of amino acids, glucose, electrolytes, vitamins, and trace minerals, administered via jugular catheter, and a parenteral lipid emulsion, which was infused separately via an umbilical arterial catheter. Piglets in the TPN groups randomly received one of the following lipid emulsions: 100% soybean oil [20% Intralipid (IL)], 100% fish oil [Omegaven (OV)], or a mixture of 30% soybean oil, 30% MCTs, 25% olive oil, and 15% fish oil [SMOFlipid (SL)]; all three lipid emulsions were provided by Fresenius Kabi (Bad Homburg, Germany). ENT piglets were fed a milk-based nutritional deficiencies, promote growth, and to achieve normal maturation. However, PN can lead to hepatic metabolic complications, such as cholestasis, within 2 weeks of PN administration (1), and 90% of infants who receive PN for greater than 3 months develop PN-associated liver disease (PNALD) (2). Limiting the lipid load administered to premature infants has been shown to be effective in preventing the onset of cholestasis (3). However, infants require sufficient caloric intake to achieve appropriate growth and neurodevelopment (4), so the option to limit the PN lipid to prevent PNALD is not ideal. For these reasons, it is important to understand the process of PNALD development in this neonatal population.
The lipid fraction of PN solutions has been a target of research to understand the development of PNALD. Soybean-based lipid emulsions are the prevalent lipid emulsion used in the US and are approved by the Food and Drug Administration for clinical use in neonates, but have been associated with PNALD (5). Total PN (TPN) solutions that contain fish oil-based emulsions have been associated with prevention of cholestasis and reversal of cholestasis in some limited studies (6,7), but the basis for these observations is not clear. Some research suggests that phytosterols, a component of soy lipid emulsions, can suppress bile acid synthesis and impact bile acid efflux target genes in the liver through inhibition of the nuclear hormone receptor, farnesoid X receptor. A recent report by El Kasmi et al. (8) suggests that phytosterols taken up by Kupffer cells mediate an inflammation-induced injury to hepatocytes. However, our recent study in PN-fed piglets showed that a lipid emulsion blend of soy, medium-chain triglycerides (MCTs), olive oil, and fish oil, which contains less than one-half the total phytosterol content as soy-lipid emulsions [207 mg/dl vs. 439 mg/dl (9)], also protects against PNALD (7).
Pure fish oil and fish oil blend emulsions, unlike soybean oil-only emulsions, contain considerable amounts of longchain omega-3 PUFAs. The administration of fish oil high in omega-3 fatty acids, DHA, and EPA, can suppress both insulin-and carbohydrate-mediated triglyceride synthesis through repression of sterol regulatory element-binding protein (SREBP) (10) and carbohydrate responsive elementbinding protein (11), respectively. Omega-3 fatty acids are also capable of activating the nuclear hormone receptor, PPAR, to upregulate downstream target genes of fat oxidation in both mitochondria and peroxisomes (12,13). In addition, omega-3 fatty acids promote the formation of anti-inflammatory prostaglandins, eicosanoids, and resolvins (14). It is likely that improved steatohepatitis and glucose sensitivity observed in adults administered omega-3 treatments are due to a combination of these targeted effects (15,16). However, there is limited evidence in preterm infant studies describing how these markedly different fatty acids present in new generation parenteral lipid emulsions alter tissue fatty acid profiles and metabolic function, especially in the liver.
As with omega-3 fatty acids, clinical trials in adults have shown that vitamin E protects against nonalcoholic fatty liver disease (NAFLD) (17) and chronic hepatitis C (18).

Vitamin E analysis
Vitamin E (-tocopherol and -tocopherol) in plasma and liver was analyzed as described previously (28,29). In brief, liver tissue was homogenized in two volumes of ethanol with an Ultra-Turrax homogenizer on ice. Aliquots of the homogenates corresponding to 200 mg liver were saponified in a mixture of ethanol, methanol, ascorbic acid (20% w/v), and KOH-water (1:1 w/v) at 80°C for 30 min, subsequently cooled, and extracted twice with 5 ml of heptane. The HPLC column for determination of tocopherol consisted of a 4.0 × 125 mm Perkin-Elmer HS-5-Silica column (Perkin-Elmer GmbH, Überlingen, Germany). The mobile phase consisted of heptane containing 2-propanol (3.0 ml/l) and degassed with helium. The flow rate was 3.0 ml/min. A comparison of retention time and peak areas with Merck (Damstadt, Germany) external standards was used to obtain the identification and quantification of the tocopherol.

RNA sequencing
RNA was extracted from liver tissue using TruSeq RNA sample preparation kit v2 (Illumina) and sequenced using Illumina RNA Hi-seq 2000. Sequencing data was analyzed using a series of software tools, which included Bowtie, tophat, and the cufflinks package [cufflinks, cuffcompare, cuffmerge, and cuffdiff (30)]. These generated lists of differentially expressed and regulated genes and transcripts. Piglet genome Sus scrofa 10.2.68 from Ensembl (http:// www.ensembl.org/Sus_scrofa/Info/Index) was used as a reference. Cutoff values for significance were: fold change greater than ±1.5, P < 0.05, and q < 0.1. Analyzed data was subjected to Ingenuity Pathway Analysis (IPA) (http://www.ingenuity.com).

Metabolomics analysis
Acyl-CoAs (ACs), acylcarnitines (ACTs), and organic acids (OAs) were analyzed by targeted MS. For preparation of tissue samples for AC and OA content, tissues were homogenized in 50% aqueous acetonitrile containing 0.3% formic acid. OAs were analyzed by Trace Ultra GC coupled to a Trace DSQ mass spectrometer using a stable isotope dilution method described previously (31,32). For analysis of long-chain ACs, chloroacetaldehyde was added to the homogenate and samples were separated by HPLC and then detection of individual ACs was done fluorometrically (33). For ACT analysis, samples were homogenized in water and analyzed by flow injection MS/MS, as described previously (34). Mean absolute concentrations of each metabolite were then transformed using base package in R (http://www. R-project.org). Clustering heatmaps were created using the heatmap.2 function in gplot (35) in conjunction with Mkmisc (36) and Rcolor (37).

Real-time PCR
Real-time quantitative (q)PCR was performed on frozen liver samples. The cDNA was generated from RNA extracted from 100 to 150 mg of frozen liver tissue, as described previously (7). Realtime qPCR was performed with SYBR Green chemistry (Applied Biosystems) on a Bio-Rad CFX96. Primers were designed using software from NCBI Primer Blast. The primers for porcine Aox1, Cyp2e1, Cyp1a2, Fmo5, interleukin (Il)-1, Il-6, Tnf, and Il-8 (supplemental Table S1) were designed using the predicted porcine sequence available on Ensemble Genome Browser [Gene identification (ID): ENSSSCG00000012871]. Amplification efficiency was controlled by the use of an internal control (GAPDH or actin). Relative quantification of target mRNA expression was calculated and normalized to GAPDH or actin expression. All reactions were performed under the following thermal cycling conditions: 10 min at 95°C followed by 40 cycles of 95°C for 15 s and 60°C for 60 s. The 2 CT method was used to compare gene formula (Litter Life; Merrick, Middletown, WI) at 240 ml/kg in eight feeds per day (24). Postsurgery, TPN was started at 5 ml/(kg·h) and gradually increased to 10 ml/(kg·h). ENT piglets also received TPN with IL following surgery, but were started on ENT feeds on day of life 2 and gradually weaned off TPN. On day 7, all TPN and enterally fed piglets received full amounts of nutrition per kilogram body weight: fluid, 240 ml; energy, 195 kcal; carbohydrate, 25 g; protein, 14 g; and lipid, 5 g, as described (7).
Piglets were weighed every other day and blood samples were drawn during surgery (day 0) and at day 14. Immediately after the last blood sample was taken on day 14, the animals were anesthetized with isoflurane and euthanized with injection of Beuthanasia (pentobarbital sodium, phenytoin sodium). Organs were isolated and weighed. Liver tissue samples were frozen in liquid nitrogen and stored at 80°C until analysis. Liver samples were also fixed in OCT for histopathology.

Sample preparation and analysis
Blood samples were collected in Na 2 EDTA tubes and centrifuged for 10 min at 2,000 g in 4°C. Samples were flash-frozen in liquid nitrogen and stored at 80°C until analysis. Glucose and insulin were analyzed as previously described (24). As piglets were on continuous TPN infusion, they received a constant glucose infusion rate. Utilizing this, we performed homeostatic model as-

Indirect calorimetry and [1-13 C]palmitate fatty acid oxidation
For measurements of respiratory exchange ratio (RER), heat production, and 13 CO 2 , piglets were placed in air-tight calorimetry chambers (Columbus Instruments) and measurements were performed as described previously (26). Specifically for 13 CO 2 analysis, on day 11, piglets were placed in a calorimetry chamber and administered [1-13 C]palmitate. Prior to infusion, arterial and breath samples were taken to determine background enrichment of 13 C. After background sampling, a primed continuous 4 h coinfusion of [1-13 C]palmitate (15 mol/(kg·h)) with respective lipid treatment was administered. Expired breath samples from the calorimetry chamber were collected every 30 min for analysis of 13 CO 2 enrichment during the 4 h infusion. Whole-body [1-13 C] palmitate oxidation was calculated using standard steady state equations (26).

Fatty acid analysis
Total fatty acids were isolated from plasma and tissues using a modified method from Folch (27). Briefly, 20-150 mg tissue samples were homogenized in 1-2 ml PBS. Plasma samples (200 l) were diluted with 500 l PBS. Samples were extracted from plasma and tissues (500 l) by the addition of chloroform-methanol (2:1 v/v). The samples were then incubated for 10 min on ice, and then spun at 2,500 rpm for 10 min. The top layer was discarded and the remaining sample was dried under nitrogen gas. Methanolic NaOH (0.5 ml, 0.05 M) was added to the dried sample and incubated for 3 min at 100°C, followed by addition of 0.5 ml BF 3 -methanol. The samples were heated for an additional 1 min at 100°C, and then 1 ml hexane and 6.5 ml saturated NaCl were added. Samples were vortexed and spun at 1,700 rpm for 4 min. The upper hexane phase was used for analysis. Quantification of fatty acid methyl esters was performed on a gas chromatograph (HP5890 Series II, Hewlett Packard) equipped with a SP-10 capillary column (Supelco) attached to a mass spectrometer (HP-5971, Hewlett Packard). 13 C-palmitate were measured (Fig. 2). RER of the OVtreated piglets was significantly (P < 0.05) greater than ENT, IL, and SL. Overall, the RER was high in all groups (>90%). Heat production, as an indicator of metabolic activity, was not different between any groups. During steady-state, the rate of 13 C-palmitate oxidation to 13 CO 2 did not differ between groups, although there was a slight trend for elevated 13 CO 2 in the OV group. Distribution of fatty acids from the diets was examined as well as the tissue enrichment of those fatty acids in the liver, brain, and muscle (supplemental Table S2, Table 2,  Table 3, and supplemental Table S3, respectively). The overall percent fatty acid composition differed significantly among the four treatment groups. ENT and IL diets lacked PUFAs, EPA, and DHA, whereas OV was highly enriched with 19.9 and 17.3%, respectively, in those fatty acids. Percent enrichment of palmitic acid (16:0) was greatest in the ENT diet. The percent enrichment of linoleic acid (LA, 18:2n-6) was greatest in the IL emulsion (36%) with very limited enrichment in both OV and SL with 3.9 and 11.8%, respectively. The content of MCT in SL diet was markedly higher than that of ENT, IL, or OV. Caprylic acid (8:0) and capric acid (10:0) comprised 31.0 and 12.7% of SL lipid emulsion, respectively. The pattern of fatty acid enrichment in both the liver ( Table 2) and muscle (supplemental Table S3) tissues mirror the dietary (supplemental Table S2) fatty acid patterns, except for the medium-chain fatty acids. The brain tissue (Table 3) showed significant (P < 0.05) enrichment of DHA in the OV and SL groups compared with ENT and TPN. The arachidonic acid (AA) concentration in the ENT groups was moderately, but significantly, higher (P < 0.05) compared with the OV and SL groups. The distribution of fatty acids in the brain mirrored results seen in the retina (data not shown), with a trend to resemble the concentration of lipids found in the diets.
Within the muscle, there were fewer differences in fatty acid enrichment among the ENT and TPN treatments. LA was significantly (P < 0.05) different between ENT, IL, and OV (11.21 ± 0.75%, 26.30 ± 1.51%, and 1.80 ± 0.13%, respectively). DHA, like in the liver, was significantly (P < 0.05) expression levels between samples, which were analyzed to determine the fold induction of mRNA expression.

Statistics
Statistical analyses were performed using SPSS 16 software (Armonk, NY). Differences among the four groups were first analyzed using one-way ANOVA, and post hoc analysis was done using Tukey's test, as described in the figure legends. For results that were not of normal distribution, the nonparametric Kruskal-Wallis ANOVA on Ranks followed by pairwise comparisons using Dunn's test was used to determine significance. P < 0.05 was considered significant. Results are presented as mean ± SEM.

RESULTS
Following 14 days of ENT or TPN administration, body weight and tissue weights differed in piglets between experimental treatments ( Table 1). Body and liver weights were previously published, but included in this analysis for continuity and comparison with other tissue weights (7). Growth rates [g/(kg·h)] were significantly decreased (P < 0.05) in the IL group versus other treatments. Jejunum and ileum weights were significantly lower, whereas liver and spleen weights were higher in the TPN-administered groups compared with ENT (P < 0.05). The pancreas weight was lower (P < 0.05) in the OV and SL groups compared with ENT, but IL did not differ from ENT. The heart tissue weight was significantly higher (P < 0.05) in the TPN versus ENT piglets.
Serum lipid and glucose values were analyzed as shown in Fig. 1. Serum cholesterol and triglyceride levels were significantly lower in all three TPN groups compared with ENT. Insulin was significantly elevated (P < 0.05) in the IL piglets compared with ENT piglets. OV insulin was moderately elevated, but nonsignificant, whereas insulin levels in ENT and SL piglets were not different. Calculated HOMA-IR ratios show a similar pattern as insulin with significant elevation in the IL group compared with ENT.
To assess whether change in body weight was contingent upon the metabolic activity and lipid content of the emulsions, energy expenditure and fat oxidation using (P < 0.05) elevated in OV group by 6.62 ± 0.72%. EPA plasma content was elevated in the SL group, but did not achieve statistical significance. We further examined the different fatty acid compositions of the lipid emulsions by calculating the ratio of n-6:n-3 fatty acids and LA:DHA (Fig. 3). From day 0 to day 14, the ENT and IL groups had a significant (P < 0.05) increase in plasma proportion of n-6 fatty acids, whereas OV and SL had a significant (P < 0.05) decrease in n-6:n-3 ratio. Between treatment groups at day 14, the n-6:n-3 IL, OV, and SL ratios were significantly (P < 0.05) different than ENT. OV treatment resulted in the lowest n-6:n-3 ratio and was significantly (P < 0.05) different from IL and SL. The SL group's n-6:n-3 ratio was also significantly (P < 0.05) lower than the IL group.
The LA:DHA ratio in the ENT group was markedly elevated at 14 days with a value >30. Interestingly, the OV group maintained the same ratio at 14 days, as compared with 0 days. The concentrations of sphingosine and ceramide were also assessed in the diet, plasma, and liver (supplemental Fig. S1). Ceramide was markedly elevated in the IL group compared with the OV or SL group. However, liver and plasma ceramide levels did not differ significantly between any of the TPN groups, although there was a trend toward lower plasma ceramide in the SL group. Dietary sphingosine was lower in the SL group compared with the IL and OV groups, and plasma sphingosine was elevated in all three TPN groups compared with ENT. Plasma sphingosine was highest in the IL group (P < 0.05). However, liver sphingosine levels did not differ between the ENT and TPN groups.
As there is increasing interest in the functional impact of vitamin E in TPN administration to prevent the onset of PNALD, we measured the concentration of the main biologically active form of vitamin E, -tocopherol, and its various stereoisomers as they were distributed in the plasma and liver following 14 days of TPN administration ( Table 5). Total plasma -tocopherol was more than 2-fold (P < 0.05) lower in the IL treatment group versus ENT. The SL group was also significantly (P < 0.05) lower in -tocopherol with 3.12 ± 0.39 g/g compared with an ENT concentration of 4.61 ± 0.24 g/g. The dominant stereoisomer form of -tocopherol in IL was RRR--tocopherol, which comprised 95.7% of the total -tocopherol in the plasma; this was significantly higher than the other groups, i.e., 24.3% (ENT), 22.2% (OV), and 30.4% (SL) of the total -tocopherol in plasma ( Table 5). The plasma 2S forms of -tocopherol's contribution to total -tocopherol content in the OV (26%) and the SL (20.8%) treatment groups was markedly higher compared with that of ENT (12.4%) and IL (0.53%). In the liver, there was a significantly (P < 0.05) higher concentration of -tocopherol in the OV and SL groups compared with ENT for total -tocopherol. -Tocopherol was measured in the liver of the piglets in the IL group, but was not detectable in the liver of the piglets in the other groups. The RRR--tocopherol contributed with 96% of the total -tocopherol in the IL group, whereas other groups had around 20% of RRR--tocopherol. In absolute concentrations, both the OV and SL groups had high RRR-tocopherol enrichment at 40.8 ± 0.36 g/g and 28.9 ± higher in OV (10.67 ± 0.84%) compared with ENT (0.31 ± 0.02%). SL had the fewest differences in fatty acid profile compared with the ENT group, but did have 8-fold higher percent DHA enrichment (P < 0.05) than ENT. The SL group had significantly lower (50%) LA compared with IL (13.82 ± 0.84% vs. 26.30 ± 1.51%).
Plasma fatty acid concentrations were markedly different after the 14 day administration of the various lipid emulsions (supplemental Tables S4, S5, and Table 4). The most pronounced changes relative to ENT nutrition occurred in the OV group. After 14 days, the increase in LA was significantly (P < 0.05) less in OV versus ENT piglets (5.42 ± 0.62% and 21.42 ± 0.64%). In the ENT and IL groups, there was a significant decrease in DHA (2.45 ± 0.14% and 0.79 ± 0.08%, respectively), whereas DHA was markedly increased in the OV and SL groups (6.56 ± 1.02% and 2.67 ± 0.24%, respectively). The EPA proportion did not change after 14 days in the ENT and IL groups, but was significantly MCT C10:2 in the liver and C6 and C6-DC/C8-OH in the muscle.
OA concentrations show little change between the ENT and TPN treatments (supplemental Fig. S2) in liver and muscle tissues. Lactate is the most abundant intermediate in all four groups in both liver and muscle. No significant differences in the citric acid cycle intermediates were observed, except for higher liver citrate in the ENT group compared with all TPN groups.
To gain insight into the effects of TPN treatment on gene expression in the liver, RNA sequencing was performed (Fig. 6). The number of genes that significantly differed from the ENT group among the three TPN treatments were compared and visualized in a Venn diagram. The total number of genes that differed from ENT was 115, 101, and 15 in IL, OV, and SL, respectively. Overall, 11 overlapping genes were altered in all three TPN groups versus the ENT groups, which included Aox1, Dak, Derl3, Hamp, Id1, Lcn2, Mfsda2a, S100g, Serpine1, Tmem86b, and Ubd. Canonical unbiased pathway analysis of the genes altered by IL treatment compared with ENT was performed. Of the 10 top pathways, oxidative degradation, oxidative stress, immune response, and lipid metabolism were recurrent. The nicotine degradation pathway that is reliant upon CYP pathway genes and other oxidation genes was the most significant pathway. Other pathways associated with nuclear hormone receptor activity, liver X receptor/retinoid X receptor activation, inflammatory response, and acute phase response 1.10 g/g, respectively; yet, this was a nonsignificant enrichment compared with ENT. The proportion of 2S-tocopherol differed in the liver, and was significantly lowest for the IL group (1.6%) and higher for the ENT (22.5%), OV (35%) and SL (42%) groups. Figure 4 shows the heatmaps for ACs in liver and muscle. The most abundant ACs in all four groups were the very small-chain ACs. Based on mean comparison by ANOVA, in liver, the IL group had significantly more linoleoyl-CoA and C20:2-CoA compared with the other three groups. OV had more myristoyl-, C20:5-, C22:6-, C22:5-, and C16:3-CoA. SL had a significantly higher concentration of octanoyl-, decanoyl-, and C11:1-CoA. Like liver, muscle tissue also had an abundance of very small-chain ACs, but no significant differences in the relative concentrations of different ACs were observed. Also, relative to the liver, fewer ACs were detected.
Free carnitine in the plasma, liver, and muscle was significantly reduced in TPN groups compared with the ENT group (Fig. 5). Within the liver, 18 of the 66 ACTs were significantly (P < 0.05) higher in the ENT-treated group compared with the TPN groups. However, only 3 of the 66 ACTs in the muscle were significantly elevated in the ENT treatment versus TPN. The most abundant ACTs in all four groups in liver and muscle were the small-chain ACTs. OV treatment significantly (P < 0.05) increased the ACT, C12:1-OH/C10:1-DC, compared with ENT, IL, or SL in the liver tissue only. SL treatment led to a significant increase in the significantly different in any treatment following qPCR. Additional genes not directly seen in the RNA seq results were also examined that are associated with nuclear hormone receptors that regulate lipid metabolism. Whereas most genes failed to show significance, Srebp1c was were also altered by IL administration. Target genes that were enriched in multiple pathways from IPA analysis were confirmed using qPCR (Fig. 7). The expression of most of the genes was consistent with results obtained from the RNA seq; however, liver X receptor (Nr1h3) was not  of TPN is linked to the complications that culminate in the development of PNALD. Recent attention has focused on how soybean oil-based emulsions (e.g., IL) may contribute to PNALD and we recently reported that new generation lipid emulsions (e.g., OV and SL) can prevent PNALD in preterm piglets (7). Beyond the impact on hepatic metabolism, lipid nutrition makes a major contribution to the whole-body energy needs, fatty acid requirements, and immune and metabolic function in developing neonates. In this study, we performed comprehensive analysis of lipidomics, metabolomics, and transcriptomics profiling of tissues in PN-fed preterm piglets given three parenteral lipid emulsions with different lipid constituents, including n-6 versus n-3 PUFA, vitamin E, and sphingolipids. Lipid emulsions are administered to premature infants to achieve optimal growth by providing a balanced high-energy parenteral nutrient intake that minimizes hyperglycemia. In this study using premature piglets to model metabolism of premature infants, we found that IL produced the lowest growth rates, whereas the growth of OV and SL was similar to ENT piglets. Interestingly, the lower growth rates of IL piglets occurred despite a differential increase in tissue mass of liver, heart, and spleen, and decrease in intestinal growth in all PN groups relative to ENT. Moreover, the presence of hepatomegaly in all PN groups occurred despite the prevention of steatosis and cholestasis in the OV and SL versus IL piglets (7). Our previous study also showed that TPN-induced hepatomegaly was due to increased protein and DNA, but not lipid content (24). The differential increase in these vital organs among all PN groups, but lower body growth rates in IL versus OV and SL piglets, points to a limitation in lean/skeletal muscle significantly (P < 0.05) decreased in all TPN groups compared with ENT.
We measured the expression of inflammatory cytokines to test whether there was a lipid treatment-specific effect on cytokine expression (Fig. 8). At the transcript level, Il-6 was the only cytokine that differed among treatment groups. In the IL group, Il-6 expression is 2-fold greater (P < 0.05) than ENT.

DISCUSSION
TPN is necessary as a life-saving intervention in many premature and gastrointestinal tract-compromised infants. Considerable evidence suggests that the lipid component 2.69 ± 0.14 Behenic acid (22:0) 0.14 ± 0.10 0.14 ± 0.05 a 0.11 ± 0.02 0.00 ± 0.10 Erucic acid (22:1n-9) 0.22 ± 0.05 0.09 ± 0.14 0.13 ± 0.08 0.14 ± 0.36 Adrenic acid (22:4n-6) 0. 40   versus IL and ENT piglets. LC-PUFAs are necessary for normal development of the central nervous system and possibly also other organ systems (45). Formation of DHA from -linolenic acid (ALA) is limited and highly variable, and DHA may be considered conditionally essential during early development (46). This is especially important for preterm infants who are born at a stage of gestation before the normal placental transfer and deposition of DHA in fetal tissues are completed. This loss of fetal DHA transfer due to preterm birth may be difficult to restore by current nutritional practices and result in a "DHA gap of prematurity" (6,9). The challenge of restoring this DHA gap in premature infants is further complicated when ENT feedings are delayed and lipid intake is exclusively via parenteral lipid emulsions. In contrast, the capacity of the piglets to synthesize AA appears to remain intact as the percentage of AA in the ENT, IL, and SL diets are at low concentrations, yet both the liver and muscle have substantial AA contents. In piglets, synthesis of DHA from ALA is greatly improved with a low LA:ALA ratio (43). High LA concentration competes with ALA for desaturases to form LC-PUFAs. The higher AA in the ENT and IL groups may in part be due to the high LA content suppressing overall DHA synthesis, as well as leading to the observed elevation in AA. This study does show that there is some functional capacity for LC-PUFA synthesis from precursor LA, or release from body stores is sufficient at 2 weeks of age in premature piglets. Also, our data suggest that premature piglet tissues have the capacity to readily take up preformed DHA and EPA given parenterally, especially DHA uptake in the brain. Ceramide concentrations have been implicated in the development of obesity and NAFLD (47,48). It was therefore interesting to see the high concentration of ceramide in the IL emulsion and the lack of ceramide in OV. Very low levels of ceramide were found in SL. However, the concentration of liver and plasma ceramide levels did not appear to differ between the treatment groups. The downstream metabolite of ceramide, sphingosine, did not show any difference in content in the liver either. It would therefore growth, which we reported previously (24). A major concern of PN-fed premature infants is the growth and neurodevelopment of the brain. The high LA content of IL has been linked to increased risk for inflammation and potential neurodevelopmental complications, whereas AA and DHA are major constituents of brain lipid composition and diets low in AA and DHA are considered a potential risk for impaired brain development in premature infants (38,39). In light of these considerations, we were surprised that brain mass among all four treatment groups did not differ, and especially given that emulsions OV and SL are enriched in DHA, the long-chain PUFA (LC-PUFA) considered essential for normal perinatal neurodevelopment. Our data does not exclude the possibility that differences in brain function and neurodevelopment may still be affected by the fatty acid composition of the lipid emulsions. These are piglets that did receive a processed cow's milk formula, not natural sow's milk, which might also account for the lack of difference between the ENT and TPN groups.
An important metabolic finding was that TPN administration with IL led to hyperinsulinemia and insulin resistance compared with ENT piglets, confirming our previous work (24), but this phenotype was diminished with OV and SL. We previously showed that insulin resistance with IL was due to impaired insulin signaling and was associated with increased hepatic inflammation, and thus it was notable in the current study that we found that hepatic Il-6 expression was highest in IL versus OV and SL piglets (40). The explanation for lower hepatic inflammation could be due to the anti-inflammatory and anti-oxidant effects associated with higher liver enrichment of n-3 PUFAs, DHA, and EPA, as well as -tocopherol in the OV and SL versus IL piglets (24,41,42).
The tissue and plasma fatty acid profiles generally reflected the pattern of fatty acids present in the lipid emulsions consistent with prior piglet studies (43,44). In contrast to these previous reports, we found significant enrichment of DHA and EPA in brain tissue in OV and SL    not seem that the excess ceramide in IL was shunted toward this metabolic pathway. The exact metabolic pathway of excess ceramide is not clear at this point. The presence of ceramides in the IL emulsion does represent a unique finding that does warrant further study for the potentially negative role it could play in the development of PNALD. Vitamin E was initially added to the new generation lipid emulsions, as manufacturers were concerned about the degree of lipid peroxidation that would occur to the LC-PUFAs, such as EPA and DHA. For this reason, both OV and SL have much higher concentrations of vitamin E compared with IL. The vitamin E concentration in IL was insufficient to maintain plasma levels similar to that seen in piglets on the ENT diet. In accordance with our previous study (32), the IL group had a considerable hepatic concentration of -tocopherol, whereas -tocopherol was not detectable in the liver of other groups, or in plasma. The contribution of -tocopherol and the stereoisomer composition of -tocopherol in plasma and liver tissue in the IL versus OV and SL probably reflected the difference in the chemical forms of vitamin E in the emulsions. The low plasma -tocopherol level of IL piglets is in agreement with clinical observations on the status of serum vitamin E in patients receiving IL compared with healthy controls (49). Vitamin E (-tocopherol) is a key antioxidant for cellular function, so an increase in circulating -tocopherol and incorporation into cellular membranes may play a role in the benefit derived from both the OV and SL groups in our study. The concentration of -tocopherol per mole percent total unsaturated fatty acids was three times higher in OV and 2.3 times higher in SL compared with IL. At the same lipid dose of 5 g/(kg·d), we recently showed prevention of cholestasis in preterm piglets by supplementing IL with vitamin E to a concentration equivalent to that present in OV (32). However, a recent publication in term piglets administered IL supplemented with vitamin E at 10 g/(kg·d) failed to show any effect on treatment (50). In this study, there were no changes in oxidative markers in piglets as well, even with an elevated serum vitamin E concentration in the supplemented compared with control group. At this point the relevance of lower serum vitamin E concentration in IL piglets is still unclear.
Metabolomic profiling of lipid intermediates in fatty acid metabolic pathways in muscle and liver revealed some prevalent trends. The ACs within the liver reflected the composition of lipid emulsion fatty acids, but the muscle ACs were not different among treatments. OV diets are preferentially enriched in the LC-PUFA CoA conjugates and SL livers are enriched in MCTs. The high CoA conjugates of MCTs in the liver occurred even though no free MCTs were present in the liver, suggesting that MCTs are very rapidly shunted toward metabolic pathways in these piglets. Of interest is that OV livers contained elevated myristoyl-CoA. Myristic acid has been shown to increase the activity of -6-desaturase and to increase accumulation of DHA in tissues (51)(52)(53). At this point, we have no direct data to suggest myristic acid activation of -6-desaturase is driving the synthesis of DHA in preterm piglets, but this observation warrants further inquiry in our preterm model. The acetyl-CoA concentration was high in both liver and muscle, yet Genes associated with pathways altered in the comparison of ENT versus IL groups were subjected to qRT-PCR analysis for comparison of ENT to IL-, OV-, and SL-treated TPN groups. Liver samples from piglets treated with ENT or TPN emulsions for 14 days were used. Values are mean ± SEM; n = 7-12 piglets per group. *P < 0.05 versus ENT activation of oxidative stress pathways and inflammation were consistent with an insulin-resistant phenotype (55) and PNALD development (56). High LA present in IL is also consistent with the increase in oxidative stress and inflammation (57). The hepatic triglyceride accumulation observed in these piglets (7) led to increased fatty acid fluxmediated mitochondrial respiratory chain dysfunction resulting in increased oxidative stress (19). To compensate for elevated oxidative stress, an increase in Sod2 in steatohepatitis is known to occur (58) and was increased in our piglet model. Interestingly, genes upregulated in nonalcoholic steatohepatitis in various animal models, Aox1 and Cyp2e1, were downregulated in the TPN groups (59,60). The latter is in accordance with Fisher et al. (61), who showed liver biopsies of patients with increasing severity of steatohepatitis had decreasing expression of Cyp2e1 and suggest an effect of duration of disease. Alternatively, liver immaturity in premature infants on TPN may lead to a transcriptomic profile similar to more severe steatohepatitis. Nevertheless, TPN clearly suppresses genes of alternative pathways of fat oxidation. It is also interesting that the expression of Srebp1c is also reduced, as this is a key regulator of lipogenesis. Taken together, there is an overall dysregulation of fat metabolism-associated genes in the TPN-treated piglets.
In conclusion, our results provide valuable lipodomic, metabolomic, and transcriptomic profiles of preterm piglets fed TPN with lipid emulsions containing different fatty acid compositions. We show that the soybean oil-based emulsion (IL) that is high in the ratio of n-6:n-3 fatty acids results in lower body growth, lower -tocopherol status, the content of acetylcarnitine in the liver and muscle was significantly higher in the ENT group compared with the TPN groups. The free carnitine in the plasma, liver, and muscle was also significantly lower in all TPN versus ENT groups. Additionally, in the liver, multiple ACTs were elevated in the ENT group compared with TPN groups, reinforcing the evidence of depleted carnitine pools in the latter. Carnitine is a key cotransporter for efficient uptake of fatty acids into the mitochondria for -oxidation. Our results confirm earlier reports that the decreased circulating carnitine levels in TPN-fed infants reflect low carnitine stores and inadequate capacity for carnitine biosynthesis (54). We measured the carnitine content of the diet of ENT piglets (53.95 nmol/g diet) and found that the concentration in the plasma of these piglets was higher than what could be achieved from the diet alone. This suggests that the endogenous synthesis of carnitine is activated by ENT feeding and that signaling is lost during TPN. Interestingly, despite the diminished tissue carnitine pools in the TPN versus ENT groups, there were no differences observed in either the whole-body palmitate oxidation rate or the RER. The RER was relatively high in all groups, including the ENT group, indicative of greater carbohydrate utilization. Similarly, there were no major changes in citric acid cycle intermediates among TPN groups, suggesting no overt changes in energy metabolism through this pathway in response to different lipid emulsion composition.
The hepatic transcriptomic profiles between piglets on ENT compared with IL emulsion revealed changes in pathways associated with oxidative stress response, inflammation, and alternative fatty acid oxidation pathways. The increased insulin resistance, and increased hepatic fat and inflammation compared with piglets fed enterally or TPN with other new generation lipid emulsions. The reduced growth in IL versus OV or SL groups was not reflected in diminished growth of specific organs measured, suggesting an overall suppression of lean skeletal muscle growth. As predicted, we found that the types of fatty acids infused parenterally were reflected in the tissue fatty acids, even in the brain, but we found no evidence that TPN per se or lipid emulsion fatty acid composition affected brain growth in the preterm piglet within the 2 week postnatal period. We also confirmed that TPN causes depletion of free carnitine and lowering of tissue ACT levels, but this did not translate into reduced whole-body fatty acid oxidation. Finally, hepatic transcriptomic profiling by RNA seq confirmed a phenotype in IL piglets marked by pro-inflammatory and oxidative stress genes reflective of poor hepatic metabolic function compared with new generation emulsions.