Prepartum dietary energy intake alters hepatic expression of genes related to peroxisome proliferator-activated receptor and inflammation in peripartal dairy cows

We determined the effect of prepartum plane of energy intake on liver function and metabolism pre-and postpartum by combining in vivo and in vitro data with mRNA expression data. A subset of multiparous prepartal Holsteins (n = 18) from a previously conducted experiment consumed 1 of 3 amounts of dietary energy intake, relative to their requirements. A diet formulated to allow consumption of ≥150% of net energy requirements during the far-off dry period and the close-up dry period was fed for ad libitum intake (150E) or in restricted amounts so that cows consumed 80% of requirements for energy (80E). A second diet was formulated to include wheat straw (26.1% of dry matter) to limit energy intake to 100% of NRC (2001) requirements for energy when fed ad libitum during the far-off period (100E). In the close-up period, 100E was fed the 150E diet for ad libitum intake. Expression of mRNA for genes related to fatty acid oxidation ( PPARA , CPT1A , ACOX1 ) was greater for 100E cows than 150E cows on d 14 postpartum. These expression patterns were related to in vitro data for conversion of palmitate to CO 2 , acid-soluble products, and esterified products by liver slices. Abundance of mRNA for PC displayed a sharp peak for all groups on d 1 postpartum, but serum glucose did not reflect this peak. The mRNA expression of SREBF1 was greater for 150E and 100E cows prepartum compared with 80E, and was positively related to rate of palmitate esterification postpartum. Expression of NR1H3 ( LXRA ) mRNA was greater for 100E cows on d 14 postpartum compared with 150E cows, which corresponded to expression of PPARA . An inflammatory response occurred in the liver around the time of parturition for 150E cows, as expression of


INTRODUCTION
Overfeeding cows during the dry period predisposes them to postpartum metabolic disorders such as fatty liver and ketosis (Murondoti et al., 2004;Overton and Waldron, 2004;Janovick et al., 2011).Dann et al. (2006) reported that plane of nutrition during the far-off dry period (4 to >8 wk prepartum) altered metabolic adaptation to lactation more than did nutrition during the close-up dry period (last 3 wk before parturition).Microarray profiling of a subset of cows in that study indicated that temporal changes in gene expression patterns were affected by energy intake in the dry period (Loor et al., 2005(Loor et al., , 2006)).Loor et al. (2005) observed that the mRNA for the acute-phase protein serum amyloid A (SAA) was upregulated, suggesting that an inflammation-associated response was activated in the liver during the periparturient period.Abundant evidence links inflammation with peripartal disorders and diseases (Bradford et al., 2015;Horst et al., 2021).
Peroxisome proliferator-activated receptors (PPAR) are nuclear receptors that influence the transcription of genes encoding CPT1A and ACOX1, the rate-limiting enzymes of mitochondrial and peroxisomal β-oxidation, respectively (Desvergne and Wahli, 1999;Desvergne et al., 2006;Feige et al., 2006).Mice that lack the ability to express PPARα, the most abundant PPAR expressed in liver, develop fatty liver because they are unable to upregulate genes involved in fatty acid oxidation pathways (Reddy, 2001;Reddy and Sambasiva Rao, 2006).
As reviewed by Feige et al. (2006) and Tontonoz and Spiegelman (2008), PPARGC1A is a cofactor needed to assemble activating molecular complexes such as those involved with activation of PPAR.Expression of PPARGC1A is increased in response to CREB.When complexed with HNF4, CREB and PPARGC1A upregulate important gluconeogenic genes including phosphoenolpyruvate carboxylase, glucose-6-phosphatase, and pyruvate carboxylase (PC; Desvergne et al., 2006).
Lipogenesis also is influenced by PPAR and other transcription factors.The liver X receptors (LXR; NR1H3) function to protect cells from over-accumulation of cholesterol (Desvergne et al., 2006).In rodent models, LXRα increases expression of sterol regulatory binding element binding proteins (SREBP) in liver, which mediates insulin response and decreases triacylglycerol (TAG) accumulation in the liver (Desvergne et al., 2006).When SREBP are active, oxidation genes regulated by PPARα are downregulated to allow balance between fatty acid oxidation and lipogenesis according to the energy status of the body (Feige et al., 2006).In mice, interference with expression of SREBP helps to prevent the development of fatty liver; however, it does not prevent insulin resistance symptoms (Yahagi et al., 2002).Because SREBP competes with PPARGC1A, which is required for HNF4 activation, expression of gluconeogenic genes is affected in rodent models of obesity (Yamamoto et al., 2004).
The PPAR are also linked to the inflammatory response (Chinetti et al., 2000).Predominantly expressed in liver, PPARα decreases the inflammatory response by inhibiting production of IL-6 and C-reactive protein via interference with IL-1 signaling through NFKB1 (Feige et al., 2006).Lipid accumulation in hepatocytes of obese mice demonstrating a type II diabetes phenotype induced subacute inflammation mediated via activation of NFKB1 (Cai et al., 2005).In these mice, hepatic production of IL-6, IL-1β, and TNF-α was increased.Similar to PPARα, PPARγ also has anti-inflammatory effects, reducing the production of TNF-α, IL-6, and IL-1β by downregulation of the NFKB1 pathway (Feige et al., 2006).
Directly, and through its interaction with other nuclear receptors, PPARα lies at the center of fatty acid oxidation and esterification, lipogenesis, gluconeogenesis, and inflammation.Such interactions of nuclear factors such as PPARα, SREBP, HNF4, and NR1H3 likely are affected by diet in dairy cows and affect downstream genes involved in fatty acid oxidation or esterification, gluconeogenesis, lipogenesis, and inflammation.These interactions may contribute to develop-ment of fatty liver and ketosis during the periparturient period.How these genes interact in response to dietary energy intake prepartum and how this may influence the development of fatty liver in dairy cows postpartum deserves further research.
Therefore, our hypothesis was that differences in prepartum energy intake would differentially affect expression of PPARA and its downstream targets, as well as other genes and nuclear receptors that control metabolism and inflammation.The objective of this study was to determine the effect of prepartum plane of energy intake on mRNA expression patterns for genes linked to PPAR in the liver during the transition period for multiparous cows.Previously collected in vivo and in vitro data were combined with measures of relative mRNA expression to discern how prepartum energy intake might influence gene expression in ways that could contribute to development of fatty liver and ketosis.

Animal Management
All procedures were conducted under protocols approved by the University of Illinois Institutional Animal Care and Use Committee.Complete detail of diets, experimental design, and animal management can be found in Dann et al. (2006).A subset of 18 cows (n = 6 per treatment) from Dann et al. (2006) was used for quantitative real-time PCR analysis of relative expression patterns of mRNA for 12 genes related to fatty acid oxidation, gluconeogenesis, lipogenesis, or inflammatory processes in the liver.All of these genes could be linked to or interact with the PPAR.This subset of cows represented 3 of the 6 dietary treatments used by Dann et al. (2006).Dietary treatments (Table 1) based on prepartum energy consumption were as follows: (1) cows that consumed >150% of NRC (2001) requirements for energy during the far-off period and were fed for ad libitum intake during the close-up period (150E); (2) cows that consumed 100% of NRC ( 2001) requirements for energy during the far-off period and were fed for ad libitum intake during the close-up period (100E); and (3) cows that were restricted to 80% of NRC (2001) requirements for energy during the far-off period and close-up period (80E).At parturition, all cows were fed a lactation diet through 56 DIM.Prepartum dietary treatments were selected to represent the extremes of over-and underfeeding, as well as the group designed to meet requirements.The number of cows per treatment was determined by a power test to detect a 10% difference in expression of genes with 80% power, using variation from data in previous experiments in our  (Loor et al., 2005(Loor et al., , 2006)).Cows were chosen randomly within each treatment group from cows that had all necessary samples.

Measurements, Sampling, and Analyses
In Vivo and In Vitro Data.Previously collected data described in Dann et al. (2006) were reanalyzed for the same subset of cows used for mRNA quantification.These data included frequency of occurrence of health problems, BCS, milk yield, energy balance, serum glucose, insulin, nonesterified fatty acids (NEFA), and BHB and total lipids, TAG, and glycogen in liver.All health records, weekly pre-and postpartum BCS, and weekly milk yield were used from this subset of cows and were not matched to data points used for mRNA analyses; rather, they were used as descriptors to characterize cows before and after parturition.Time points used for blood metabolites and liver composition were matched to time points used for mRNA analyses.Likewise, previously collected data for fatty acid metabolism by liver slices in an in vitro system described in Litherland et al. (2011) were used.Time points again matched those used for quantitative PCR analyses.
Relative Quantification of mRNA.Total RNA was extracted from liver tissue on d −65, −30, −14, 1, 14, and 28 relative to parturition (all ± 1 d range) as described by Loor et al. (2005).Biopsies occurred at approximately 0700 h after feed refusals had been removed, and before cows received fresh feed.For lactating cows, these biopsies occurred after the morning milking.Total RNA was isolated from samples using TRIzol Reagent (Invitrogen Corp.), and purified using acid-phenol chloroform extraction.Samples were precipitated for 2 h at −20°C using isopropanol, washed with 75% ethanol, and quantified using a spectrophotometer.Integrity of RNA was verified by viewing on a 1% denaturing agarose gel, and RNA was stored at −80°C.
Reverse transcription reactions were done using 100 ng of total RNA as a template with 50 ng of random hexamers, 50 µM oligo(dT) 20 , 10 mM dNTP, 40 U RNasin (Promega), and 200 U of SuperScript III reverse transcriptase (Invitrogen) in a final volume of 20 µL.Upon completion of reactions, each cDNA pool was diluted 1:4 using DNase-and RNase-free water.To make standards, pooled liver RNA from several cows and time points was used to make cDNA as described for experimental samples.Serial dilutions were made from this cDNA stock to create a set of standards for the experiment.
Real-time quantitative PCR reactions were performed in triplicate in 384-well plates using SYBR  Green I universal master mix (Applied Biosystems), 400 nM each of the forward and reverse primers (Integrated DNA Technologies), and 4 µL of diluted cDNA in a final reaction volume of 10 µL per well.Amplification and data collection were carried out using an ABI Prism 7900 HT SDS instrument (Applied Biosystems) with reaction conditions of 2 min at 50°C, 10 min at 95°C, and 40 cycles of 15 s at 95°C and 60 s at 60°C.Standard curves were run for each gene on each plate to determine relative copy number of target cDNA for each gene.Dissociation curves were generated following the last cycle.

Statistical Analyses
Production, metabolite, and liver composition data for this subset of cows from Dann et al. (2006), in vitro liver metabolism data from Litherland et al. (2011), and quantitative PCR data were analyzed as a randomized design using the MIXED procedure of SAS version 9.2 (SAS Institute Inc.) with the following model: where y ijk = an observation from the ith day relative to calving, jth treatment, and kth cow; µ = the grand mean; D i = effect of the ith day; T j = effect of the jth treatment; DT ij = effect of the day by treatment interaction; and C (ij)k = random experimental error from the kth cow nested within the ith day and jth treatment.Cow was the experimental unit.
The REPEATED statement was used for days relative to parturition.The random error term used for all mixed models was cow within treatment and the covariance structure yielding the lowest Akaike's information criterion was used (Littell et al., 1998).Using this methodology, an autoregressive covariance structure best fit the data set.Degrees of freedom were estimated by using the Satterthwaite option in the model statement.Treatment means were separated using the DIFF statement with a Bonferroni correction for multiple comparisons.When interactions of treatment and time were significant, the slice option of the least squares means statement was used to explore significant effects of treatment and time across time and treatments, respectively.Residuals were examined to determine normality and homoscedasticity.Significance was declared when P ≤ 0.05, and tendencies were declared at 0.05 < P ≤ 0.15.
Health data were analyzed with the FREQ procedure in SAS and interpreted using the Fisher's exact test probabilities.To relate measurements taken using quantitative PCR to in vivo and in vitro measurements taken by Dann et al. (2006) and Litherland et al. (2011), the CORR procedure in SAS was used to determine relationships between these variables.Pearson correlations are reported using Fisher's z-transformations for correlation estimates and 95% confidence interval estimates.

In Vivo and In Vitro Variables
Body Condition, Energy Balance, Milk Yield, and Health.Prepartum BCS tended to be greater for 150E and 100E cows compared with 80E cows (P = 0.06; Figure 1A).Whereas 150E and 100E cows gained BCS prepartum, 80E cows maintained BCS prepartum (treatment × week, P = 0.002).As designed, 150E and 100E cows were in positive energy balance prepartum over both the far-off and close-up periods (data not shown), which likely explained these changes.Because of the differences in DMI prepartum, cows in the 150E group experienced a 14% drop in energy balance (as a percentage of their requirement) from the far-off to the close-up period, whereas 100E cows experienced a 30% increase in energy balance.These changes then were followed by a 37% decrease in energy balance for 150E cows and a 43% decrease in energy balance for 100E cows through wk 1 postpartum.In contrast, cows in the 80E group experienced very little change in energy balance from the far-off to the close-up period (0.9%) and an increase in energy balance from the close-up to wk 1 postpartum (19%).
Cows in all treatment groups lost BCS postpartum (week, P = 0.001); however, prepartum dietary treatment effects did not carry over to the postpartum period (P = 0.34), and this change was not different among groups (treatment × week, P = 0.55).Milk yield was not affected by prepartum dietary treatment (P = 0.50; Figure 1B), but increased regardless of group as week of lactation progressed (week, P < 0.001).When milk yield and DMI were considered for these cows, 150E were in more negative energy balance relative to their requirements during the first 3 wk postpartum compared with 100E cows (P = 0.05), whereas 80E cows were not different from either group (P ≥ 0.17).Calculated energy balances were 89.3, 102.1, and 98.1% of requirement for 150E, 100E, and 80E cows, respectively, during the first 3 wk postpartum.
Cows in the 150E group had a greater frequency of treatment for subclinical ketosis compared with 100E or 80E cows (P = 0.03; Table 3).Inflammatory conditions such as mastitis, metritis, and foot and leg problems were not affected by dietary treatments (P ≥ 0.35).In this subset of cows, retention of fetal membranes was not affected by treatment (P = 0.16).
Serum Hormones and Metabolites.Serum insulin was not affected by dietary treatment (P = 0.26) or by the treatment × day interaction (P = 0.72; Figure 2A).From d −14 to d 1 relative to parturition, all groups experienced a decline in serum insulin.Serum glucose was not affected by dietary treatment (P = 0.99; Figure 2B), day (P = 0.60), or their interaction (P = 0.41).
Serum NEFA were unaffected by prepartum diet (P = 0.54; Figure 2C) or by the interaction of dietary treatment × day; however, a strong effect of day was observed, as NEFA spiked on d 1 postpartum (P < 0.001).Serum BHB tended to be affected by treatment (P = 0.07; Figure 2D) and was affected by day (P = < 0.001), as a peak occurred in all groups after parturition.The 150E cows had greater BHB than 100E cows (P = 0.01), and 80E cows tended to have higher BHB compared with 100E cows (P = 0.06).
In Vitro Liver Metabolism and Liver Composition.Rate of palmitate conversion to CO 2 and percentage of total palmitate metabolism converted to CO 2 were affected by a treatment by day interaction (P < 0.001; Figure 3A and 3D).On d −30, rate of palmitate conversion to CO 2 was greater for 100E compared with 80E cows (P = 0.002) and tended to be greater than 150E cows (P = 0.12); however, on d −14 relative to parturition, rate of palmitate conversion to CO 2 was greater for 150E cows compared with either 100E or 80E cows (P ≤ 0.01).On d −14, 150E cows had a greater percent of palmitate converted to CO 2 compared with 100E cows (P = 0.02) and also tended to be greater than 80E cows (P = 0.14).At d 14 postpartum,   3C), with greatest rates of conversion to EP on d 1 relative to parturition; however, no effect of treatment or interaction of day with treatment was observed (P ≥ 0.29).The percentage of palmitate converted to EP was affected by a treatment by day interaction (P = 0.03; Figure 3F).On d −14, 150E cows had a lower percentage of palmitate converted to EP compared with 80E cows (P = 0.03), but on d 1 postpartum, 150E cows had greater percentage of palmitate conversion to EP compared with 80E cows (P = 0.03).
The changes observed for in vitro metabolism of liver slices among dietary treatment groups were not reflected in liver composition (Figure 4A-C; P ≥ 0.69).Day relative to parturition affected liver total lipids, TAG, and glycogen contents as a percentage of wet tissue weight (P ≤ 0.004).The content of lipids and TAG peaked at d 14 postpartum, whereas glycogen content reached a nadir at d 1 postpartum.A tendency for a treatment by day interaction was observed for liver glycogen (P = 0.07).Liver glycogen was greater in 100E cows compared with 80E cows on d −14 (P = 0.01) and tended to be greater than 150E cows (P = 0.14).

mRNA Abundance
Fatty Acid Oxidation.Expression of PPARGC1A was affected by day relative to parturition (P = 0.04; Figure 5A); however, treatment or interaction had no main effect (P ≥ 0.39).Expression of PPARA tended to be affected by a treatment by day interaction (P = 0.12; Figure 5B).On d 14 postpartum, expression of PPARA was higher for 100E cows compared with 150E or 80E cows (P ≤ 0.02).Expression of CPT1A increased as cows moved from prepartum to postpartum (day, P = 0.003; Figure 5C) but was not affected by dietary treatment (P = 0.23).The tendency (P = 0.15) for a treatment × day interaction was attributable to the greater expression (P = 0.02) on d 14 for 100E than for 150E, with 80E being intermediate and not different from the other treatments.Expression of ACOX1 tended to be affected by dietary treatment (P = 0.12; Figure 5D), with 80E cows trending lower compared with 150E and 100E cows.
Gluconeogenesis.Expression of HNF4A was strongly affected by an interaction of treatment and day (P = 0.001; Figure 6A).Cows in the 80E group had lower expression of HNF4A on d −30 compared with either 100E or 150E cows (P ≤ 0.05) and lower expression of HNF4A on d −14 compared with 150E cows (P = 0.03).Postpartum, 100E cows had greater expression of HNF4A on d 14 compared with 150E or 80E cows (P ≤ 0.003), but on d 28 postpartum, expression of HNF4A was greater for 150E cows compared with 80E cows (P = 0.01) and tended to be higher than 100E cows (P = 0.08).Expression of PC was strongly influenced by day relative to parturition (P < 0.0001; Figure 6B), as it spiked on d 1 postpartum, but was not affected by treatment or the interaction of treatment and day.
Lipogenesis.Expression of NR1H3 mRNA was affected by day relative to parturition (P < 0.0001; Figure 7A).A main effect of diet tended to occur (P = 0.08), and the interaction of treatment × day (P = 0.05) indicated that expression for 100E cows was greater than either 150E or 80E cows, or both, at several time points.Expression of SREBF1 was affected by day (P < 0.0001; Figure 7B) and a treatment by day interaction (P = 0.001), with a tendency (P = 0.08) for a treatment effect.Expression of SREBF1 was greater for all groups during the prepartum period compared with the postpartum period, with large differences between 150E or 100E and 80E cows on d −30 relative to parturition (P ≤ 0.001) and between 100E and 80E cows on d −14 (P < 0.001).Expression of SREBF1 also tended to be greater in 100E cows compared with 150E cows on d −14 (P = 0.11).
Inflammation.Expression of NFKB was largely unaffected by treatment or day (P ≥ 0.16; Figure 8A) or by the treatment × day interaction (P = 0.22).Expression of TNF was affected by day (P = 0.01; Figure 8B) but was not affected by treatment or treatment × day.Expression of IL1B was greater for 150E cows compared with 100E or 80E cows (treatment, P = 0.04; Figure 8C).Expression of SAA3 was affected by day (P = 0.004; Figure 8D), as a spike in expression occurred on d 1 postpartum, but was not affected by treatment or treatment × day.

Relationships of In Vivo and In Vitro Variables with mRNA Expression
Across all treatments, significant relationships occurred between previously measured variables and the mRNA expression patterns (Table 4).The expression of ACOX1, NR1H3, and HNF4A mRNA was negatively associated with ASP production and positively associated with TAG esterification rates in liver slices (P ≤ 0.05).On the other hand, CPT1A mRNA expression was positively related to ASP production and negatively related to TAG esterification rates in liver slices (P = 0.04).Expression of NR1H3 mRNA was also negatively associated with BHB postpartum (P = 0.04).Expression of SREBF1 mRNA was positively related to liver glycogen and TAG esterification rates in liver slices (P ≤ 0.05).The expression of NFKB and IL1B mRNA prepartum was negatively associated with ASP production and positively associated with TAG esterification in liver slices postpartum (P ≤ 0.04).Expression of IL1B mRNA on d 1 and d 14 postpartum was negatively related to glucose prepartum and positively related to insulin prepartum (P = 0.02).Expression of SAA3 and TNFA mRNA on d 1 postpartum was positively associated with CO 2 production by liver slices (P = 0.05).Expression of TNFA mRNA was also positively related to ASP production by liver slices (P = 0.05).

DISCUSSION
Characteristics of the treatment groups aligned with the larger experiment (Dann et al., 2006), although many differences were not significant in this smaller subset of data.Cows overfed energy (150E) had greater positive energy balance during the dry period.Cows restricted in intake (80E) were in slight negative energy balance and had lower BCS than 100E or 150E groups.Although cows fed 150E had numerically lower milk yield and numerically lower DMI (approximately 2.0 kg/d) postpartum compared with 100E or 80E cows, 150E cows still were in more negative energy balance postpartum than 100E or 80E cows.Differences in NEFA were not significant among treatments but tended to be greater on d 1 postpartum for 150E cows.
The concentration of BHB tended to be greater for 150E and 80E than for 100E.Greater BHB for 150E cows might be attributable to greater prepartum intakes of grain leading to greater ruminal production of BHB.For 80E cows, the negative energy balance prepartum could have led to hepatic oxidation of NEFA and production of ketone bodies.Although not reaching significance, numerical trends indicated that 150E cows accumulated more lipids and TAG in liver postpartum, which could have influenced liver function and metabolism.
In the in vitro system used, the production of CO 2 by liver slices is an indicator of complete oxidation of fatty acids and ASP is an indicator of incomplete oxidation to ketone bodies (Jesse et al., 1986;Drackley et al., 1991).Metabolism of palmitate by liver slices indicated that 150E cows had greater potential for TAG accumulation in liver.In 150E cows, metabolism switched from a greater proportion of fatty acid metabolism as oxidation and production of ketones prepartum to a greater proportion as esterification postpartum.Greater oxidation prepartum for 150E cows would be unexpected in the presence of similar insulin concentrations, unless some degree of insulin resistance was present.Cows in the 100E group had a balance between fatty acid oxidation, ketone production, and esterification prepartum, and a greater proportion of fatty acid metabolism as oxidation postpartum compared with 150E.Comparable to 100E, 80E cows also had a similar balance preand postpartum.Because NEFA concentration tended to be greater on d 1 postpartum for 150E compared with 80E or 100E cows, the supply of NEFA to the liver would have been greater for 150E cows.Coupled with lower oxidation and greater esterification rate of palmitate by liver slices in the 150E group postpartum, this likely predisposed 150E to greater hepatic lipid accumulation.Mean total lipid and TAG concentrations in liver peaked at d 14 for 100E and 150E cows and at d 28 for 80E cows.
Expression of PPARGC1A was not affected by treatments.Although originally thought to be required for the transcription of PPARA, at least in nonruminants, it is now known that PPARGC1A interacts with several nuclear receptors and other proteins capable of transcriptional activation (i.e., this coactivator can exert control on distinct sets of genes across diverse signaling pathways; Charos et al., 2012).Thus, the lack of change in PPARGC1A, despite changes in expression of PPARA at d −30, −14, and 14, could mean that a specific "signal" was missing (e.g., cAMP), that β-oxidation did not require PPARGC1A to function, or that the amount of PPARGC1A was already sufficient.A similar trend was detected in one of our previous studies comparing transcriptional responses in liver of cows fed a diet similar to 100E relative to cows fed a diet similar to 150E (Khan et al., 2014).Although PPARGC1A is an important cofactor for activation of PPAR, ligands specific to PPAR may be more important activators of their downstream genes (Tontonoz and Spiegelman, 2008).Cows fed 100E had greater expression of PPARA at d −14 and 14 postpartum compared with other groups, whereas at d −14 150E had the greatest expression.Greater expression of PPARA did not correspond to changes in expression of the target genes CPT1A or ACOX1.The only significant time point for differences of CPT1A expression was at d −65, before the start of the study, although a trend was noted for d 14.It is possible that mRNA expression for PPARA was not limiting its protein activity, and that ligand activation of existing PPARA was more important in determining target gene expression.Expression of CPT1A was correlated positively with fatty acid conversion to ASP.In a previous study, peroxisomal β-oxidation, which is another avenue for NEFA oxidation by the liver (Drackley, 1999), was negatively correlated with  hepatic TAG accumulation in the periparturient period and was greater for cows fed fat prepartum (Grum et al., 1996).In the present study, expression of ACOX1, the flux-generating step of peroxisomal β-oxidation, was related negatively to rate of ASP formation.Cows fed 80E, with the lowest mean TAG and lipid concentrations in liver, tended to have lower ACOX1 than cows fed 100E or 150E.Esterification of fatty acids was correlated positively with NR1H3, HNF4A, and SREBF1, and related negatively to CPT1A.Relative expression of mRNA for genes related to fatty acid oxidation and esterification together with in vivo or in vitro data suggest that 100E cows were better equipped metabolically to oxidize NEFA rather than to esterify NEFA, in contrast to 150E cows.
When PPARGC1A and HNF4A are upregulated, this combination influences PPARA expression and positively influences genes in gluconeogenic pathways (Desvergne et al., 2006;Feige et al., 2006).In the present study, expression patterns of PPARGC1A and HNF4A were similar prepartum, and broadly similar to expression of PPARA.Nevertheless, expression of PC as one measure of gluconeogenesis did not correspond to the same pattern of expression as that of PPARGC1A, HNF4A, and PPARA.This lack of association may be a result of the constant gluconeogenesis in ruminants.Expression of PC, a key regulatory enzyme for gluconeogenesis from nonpropionate carbon sources, was sharply increased at d 1 but was not affected by treatments.Flux of NEFA oxidation around parturition may have had some influence on the increase of PC expression on d 1 postpartum.In dairy cows, PC mRNA abundance is affected by physiological state as reflected in the insulin to glucagon ratio (Greenfield et al., 2000;Khan et al., 2014), dietary energy density (Khan et al., 2014), changes in nutritional status (Velez and Donkin, 2005), and increased glucose demand (Bradford and Allen, 2005).It is possible that changes in intake and energy balance as well as insulin around parturition allowed greater supply of acetyl-CoA from NEFA oxidation, which then may have influenced expression of PC.
Even though differences were noted in expression of HNF4A pre-and postpartum, this did not seem to relate directly to in vivo measures on corresponding days.In nonruminants, HNF4A decreases ketogenesis by downregulating key genes such as 3-hydroxy-3-methyl-glutaryl-CoA synthase 2 (HMGCS2), whereas PPAR promotes upregulation (Hegardt, 1999).Lower HNF4 expression prepartum for cows fed 80E might have been related, therefore, to the greater BHB concentrations observed prepartum for those cows.The protein HNF4A has also been reported to be important in VLDL export of TAG from the liver in mice (Watt et al., 2003).
Although expression of NR1H3 and SREBF1 is linked in nonruminants (Edwards et al., 2002a,b;Desvergne et al., 2006), expression of mRNA for these genes was not related in this study, as each had distinct expression patterns.Greater expression of SREBF1 was observed prepartum in both groups of cows fed for ad libitum DMI (150E and 100E), similar to a previous report (Khan et al., 2014), and level of feeding alone likely influenced these changes in mRNA expression prepartum.Expression of SREBF1 was strongly correlated with hepatic lipid concentrations and with hepatic esterification rates in the present study.Greater expression of PPARA on d 14 postpartum in 100E Liver slice in vitro metabolism measurement abbreviations: ASP = acid-soluble product, nmol/(g wet tissue wt × h); ASP, % of total = ASP production as a percent of total palmitate metabolism; CO 2 = CO 2 production, nmol/(g wet tissue wt × h); CO 2 , % of total = CO 2 production as a percent of total palmitate metabolism; TAG = triacylglyceride esterification, nmol/(g wet tissue wt × h); TAG, % of total = TAG esterification as a percent of total palmitate metabolism.
cows might have helped to downregulate SREBF1, as demonstrated in rodents (Yoshikawa et al., 2003), thus reducing esterification of TAG in liver.At least in nonruminant mammals, insulin promotes upregulation of SREBP (Desvergne et al., 2006).Insulin resistance postpartum then might have been responsible for the decreased expression of SREBP postpartum despite the increase in hepatic lipid concentrations.
Studies in mice have demonstrated that NR1H3 interferes with PPARA expression and downstream gene transcription (Ide et al., 2003;Handschin and Meyer, 2005;Feige et al., 2006), acting as a switch between lipid storage and lipid oxidation pathways.This role was not evident in cows in our study, however, as expression of NR1H3 on d 14 was highest in 100E cows as was expression of PPARA, PPARGC1A, CPT1A, and ACOX1.Together with changes observed for mRNA patterns for genes related to fatty acid oxidation, as well as in vivo and in vitro measurements, greater expression of NR1H3 in 100E cows on d 14 postpartum compared with 150E cows may have helped limit hepatic accumulation of lipids and TAG.Also interesting to note was the expression pattern for NR1H3 in the 80E cows.Expression of NR1H3 was relatively stable for 80E cows both pre-and postpartum compared with 100E and 150E cows.This finding is consistent with the lesser degree of TAG and lipid accumulation noted for 80E cows.In nonruminants, fasting promotes greater rates of fatty acid oxidation and, with this change, lower NR1H3 and greater PPARA mRNA expression (Handschin and Meyer, 2005).Perhaps restricting intake in 80E cows did not provide the same physiological signals as fasting.
In the present study, the most striking evidence for an inflammatory response was increased IL1B expression in 150E cows both pre-and postpartum.It seems possible that an inflammatory response occurred both pre-and postpartum.In rodents, Tomita et al. (2006) demonstrated that responsiveness of TNF-α receptors in liver was related to activation of Kupffer cells and development of fatty liver disease.Lipid accumulation in rodent livers has also been linked to oxidative stress and damage of liver tissue, resulting in release of TNF-α from Kupffer cells (Reddy and Sambasiva Rao, 2006).When active, PPARα inhibits IL-6 and acute-phase response activation via IL-1 by interfering with NFKB1 activation (Feige et al., 2006).Similar changes in mRNA abundance for NFKB1 and other proinflammatory factors (e.g., TLR4, MYD88, RELA) around parturition were reported in cows fed a diet similar to 150E compared with a diet similar to 100E (Khan et al., 2015).Furthermore, although 80E cows had a larger spike in TNF expression on d 1 rela-tive to parturition compared with other groups, the elevated expression of IL1B both pre-and postpartum in 150E cows may have prevented increased expression of PPARA and, thus, downstream oxidation genes in these cows postpartum.In mice, infusion of TNF-α reduced expression of PPARA and ACOX1 in liver tissue (Beier et al., 1997).Kim et al. (2007) demonstrated suppression of NR1H3, PPARA, CPT1A, and SREBF1 expression, and increased NFKB1 transcription in liver cells cultured with TNF or IL-1 relative to controls.
Surprising in the context of other results was that the expression of SAA3, although highest for all cows on d 1, was relatively lower in 150E cows.In rodents, activation of the acute-phase response in liver is associated with lower expression of PPARA (Mandard et al., 2004); however, for cows in the present study, PPARA was lowest for 150E cows, which also had lower expression of SAA3 on d 1.The overall change in SAA3 expression in the present study was small compared with changes observed in TNF or IL1B, and perhaps the acute-phase response was active to some degree in all groups of cows at parturition as demonstrated previously (Khan et al., 2015).Also, SAA3 is not as highly expressed in liver as SAA1 and is expressed in greater amounts in mammary tissue (Saremi et al., 2013).Both TNF-α and IL-1β have been associated with a greater production of acute-phase proteins such as haptoglobin and lower production of serum albumin in cultured bovine hepatocytes (Yoshioka et al., 2002).Recombinant bovine TNF-α also increased circulating haptoglobin in lactating dairy cows (Kushibiki et al., 2003).When liver composition data are considered, it is possible that damage to liver tissue had accumulated both pre-and postpartum in 150E cows.In nonruminants, SAA synthesis in the liver is associated with high density lipoprotein (HDL) secretion in the liver (Coetzee et al., 1986;Wei et al., 2008).In cows with fatty liver, the secretion of HDL is impaired (Katoh, 2002); thus, it is possible that lipid-related damage to liver tissue around parturition affected SAA3 expression in the present study.
Similar to results observed in diabetic and obese nonruminants, liver lipid accumulation and overnutrition were related to an inflammatory response around parturition in 150E cows.Changes in expression of IL1B mRNA may have been related to overnutrition and the beginning of liver tissue impairment, as lipids began to accumulate prepartum.Postpartum, the increased expression of IL1B mRNA in cows overfed energy during the entire dry period may have been an indicator of activated inflammatory pathways in Kupffer cells in the liver, possibly related to tissue damage due to lipid accumulation.Furthermore, the expression of PPARA mRNA was relatively lower in the overfed cows on d 14 postpartum, which may have further contributed to activation of inflammation and the tendency for greater rates of esterification and lower rates of NEFA oxidation postpartum.More work is needed to determine whether prepartum overnutrition per se creates an inflammatory response that results in detrimental effects postpartum or whether inflammation occurs as a result of lipid accumulation.
Finally, the cows used in this study were relatively healthy and did not exhibit large amounts of lipid accumulation in their liver postpartum.Because of the absence of severe metabolic health problems, such as clinical ketosis or inflammatory conditions, it seems highly likely that the effects observed were primarily related to diet.

CONCLUSIONS
Prepartum plane of energy intake affected expression patterns of mRNA for several genes related to PPARα during the transition period.Furthermore, many of these mRNA expression changes were related to in vivo and in vitro measurements.Limiting energy intake in the far-off dry period had positive effects on expression patterns for mRNA important to liver function postpartum.These cows had greater expression of PPAR, CPT1A, ACOX1, HNF4A, and NR1H3 and less accumulation of lipids in their liver on d 14 postpartum, which corresponded to liver metabolism measured in vitro.On the other hand, cows that were allowed to overconsume energy for the entire dry period exhibited changes in mRNA expression patterns that likely contributed to greater postpartal accumulation of lipids in their liver.On d 14 postpartum, when the greatest amount of lipids accumulated in the liver for all dietary treatment groups, PPARA, CPT1, ACOX1, HNF4A, and NR1H3 were downregulated in cows fed 150E relative to cows fed to meet requirements.Restricting energy intake over the entire dry period was not detrimental to mRNA expression patterns or measures of metabolism in this study.
Figure 1.Least squares means of weekly median body condition score and weekly milk yield for multiparous cows fed different levels of energy prepartum.Pooled standard error bars are shown.Treatment abbreviations: 150E = cows fed to consume ≥150% of NRC (2001) requirement for energy during the far-off and close-up dry period; 100E = cows fed to meet NRC (2001) requirements for energy during the far-off dry period, then ad libitum to meet or exceed energy requirements during the close-up dry period; and 80E = cows restricted to 80% of NRC (2001) requirement for energy during the far-off and close-up dry period.Far-off diets were fed from dry-off through 26 d before parturition, and close-up diets were fed from 25 d before expected parturition until parturition.One lactation diet was fed to all cows from parturition through 56 DIM.Asterisks indicate differences among treatments at a given week (P < 0.05).(A) Body condition score, prepartum: treatment, P = 0.06; week, P = 0.002; treatment × week, P = 0.002.Postpartum: treatment, P = 0.34; week, P = 0.001; treatment × week, P = 0.55.(B) Milk yield: treatment, P = 0.50; week, P < 0.001; treatment × week, P = 0.25.

Figure 2 .
Figure 2. Least squares means of serum insulin, serum glucose, serum nonesterified fatty acids (NEFA), and plasma BHB in multiparous cows fed different levels of energy prepartum.Pooled standard error bars are shown.Treatment abbreviations: 150E = cows fed to consume ≥150% of NRC (2001) requirement for energy during the far-off and close-up dry period; 100E = cows fed to meet NRC (2001) requirements for energy during the far-off dry period, then ad libitum to meet or exceed energy requirements during the close-up dry period; and 80E = cows restricted to 80% of NRC (2001) requirement for energy during the far-off and close-up dry period.Far-off diets were fed from dry-off through 26 d before parturition, and close-up diets were fed from 25 d before expected parturition until parturition.One lactation diet was fed to all cows from parturition through 56 DIM.(A) Serum insulin: treatment, P = 0.26; day, P = 0.002; treatment × day, P = 0.72.(B) Serum glucose: treatment, P = 0.99; day, P = 0.60; treatment × day, P = 0.41.(C) Serum NEFA: treatment, P = 0.54; day, P < 0.001; treatment × day, P = 0.37.(D) Serum BHB: treatment, P = 0.07; day, P = < 0.001; treatment × day, P = 0.63.

Figure 3 .
Figure 3. Least squares means for in vitro conversion of [1-14 C] palmitate to acid-soluble products, CO 2 , and esterified products, and in vitro conversion of [1-14 C] palmitate to acid-soluble products, CO 2 , and esterified products, expressed as percent of total palmitate metabolism by liver slices from multiparous cows fed different levels of energy prepartum.Pooled standard error bars are shown.Treatment abbreviations: 150E = cows fed to consume ≥150% of NRC (2001) requirement for energy during the far-off and close-up dry period; 100E = cows fed to meet NRC (2001) requirements for energy during the far-off dry period, then ad libitum to meet or exceed energy requirements during the close-up dry period; and 80E = cows restricted to 80% of NRC (2001) requirement for energy during the far-off and close-up dry period.Far-off diets were fed from dry-off through 26 d before parturition, and close-up diets were fed from 25 d before expected parturition until parturition.One lactation diet was fed to all cows from parturition through 56 DIM.Asterisks indicate a difference among treatments at a given day (P < 0.05).(A) Carbon dioxide production: treatment, P = 0.36; day, P = 0.01; treatment × day, P = 0.001.(B) Acid-soluble products: treatment, P = 0.34; day, P = 0.14; treatment × day, P = 0.02.(C) Esterified products: treatment, P = 0.29; day, P < 0.001; treatment × day, P = 0.53.(D) Carbon dioxide production, % of total palmitate metabolism: treatment, P = 0.77; day, P = < 0.001; treatment × day, P = 0.04.(E) Acidsoluble product production, % of total palmitate metabolism: treatment, P = 0.73; day, P = 0.39; treatment × day, P = 0.03.(F) Esterified product production, % of total palmitate metabolism: treatment, P = 0.76; day, P = 0.21; treatment × day, P = 0.03.

Figure 4 .
Figure 4. Least squares means for total lipid, triacylglycerol, and glycogen in liver for multiparous cows fed different levels of energy prepartum.Pooled standard error bars are shown.Treatment abbreviations: 150E = cows fed to consume ≥150% of NRC (2001) requirement for energy during the far-off and close-up dry period; 100E = cows fed to meet NRC (2001) requirements for energy during the far-off dry period, then ad libitum to meet or exceed energy requirements during the close-up dry period; and 80E = cows restricted to 80% of NRC (2001) requirement for energy during the far-off and close-up dry period.Far-off diets were fed from dry-off through 26 d before parturition, and close-up diets were fed from 25 d before expected parturition until parturition.One lactation diet was fed to all cows from parturition through 56 DIM.Asterisks indicate a difference among treatments at a given day (P < 0.05).(A) Liver total lipid: treatment, P = 0.69; day, P = 0.004; treatment × day, P = 0.96.(B) Liver triacylglycerol: treatment, P = 0.72; day, P = 0.003; treatment × day, P = 0.82.(C) Liver glycogen: treatment, P = 0.91; day, P < 0.001; treatment × day, P = 0.07.

Figure 5 .
Figure 5.Effect of prepartum plane of energy intake on mRNA expression for genes related to fatty acid oxidation in liver tissue.Data were collected with real-time quantitative PCR using the standard curve method for quantification of target cDNA in experimental samples.Pooled standard error bars are shown.Treatment abbreviations: 150E = cows fed to consume ≥150% of NRC (2001) requirement for energy during the far-off and close-up dry period; 100E = cows fed to meet NRC (2001) requirements for energy during the far-off dry period, then ad libitum to meet or exceed energy requirements during the close-up dry period; and 80E = cows restricted to 80% of NRC (2001) requirement for energy during the far-off and close-up dry period.Far-off diets were fed from dry-off through 26 d before parturition, and close-up diets were fed from 25 d before expected parturition until parturition.One lactation diet was fed to all cows from parturition through 56 DIM.Asterisks indicate a difference among treatments at a given day (P < 0.05).(A) PPARGC1A, treatment, P = 0.73; day, P = 0.04; treatment × day, P = 0.39.(B) PPARA, treatment, P = 0.48; day, P = 0.003; treatment × day, P = 0.12.(C) CPT1A, treatment, P = 0.23; day, P = 0.003; treatment × day, P = 0.15.(D) ACOX1, treatment, P = 0.12; day, P = 0.10; treatment × day, P = 0.22.

Figure 6 .
Figure 6.Effect of prepartum plane of energy intake on relative mRNA expression for genes related to gluconeogenesis in liver tissue.Data were collected with real-time quantitative PCR using the standard curve method for quantification of target cDNA in experimental samples.Pooled standard error bars are shown.Treatment abbreviations: 150E = cows fed to consume ≥150% of NRC (2001) requirement for energy during the far-off and close-up dry period; 100E = cows fed to meet NRC (2001) requirements for energy during the far-off dry period, then ad libitum to meet or exceed energy requirements during the close-up dry period; and 80E = cows restricted to 80% of NRC (2001) requirement for energy during the far-off and close-up dry period.Far-off diets were fed from dry-off through 26 d before parturition, and close-up diets were fed from 25 d before expected parturition until parturition.One lactation diet was fed to all cows from parturition through 56 DIM.Asterisks indicate a difference among treatments at a given day (P < 0.05).(A) HNF4A, treatment, P = 0.16; day, P = 0.08; treatment × day, P = 0.001.(B) PC, treatment, P = 0.28; day, P < 0.001; treatment × day, P = 0.85.

Figure 7 .
Figure 7. Effect of prepartum plane of energy intake on mRNA expression for genes related to lipogenesis in liver tissue.Data were collected with real-time quantitative PCR using the standard curve method for quantification of target cDNA in experimental samples.Pooled standard error bars are shown.Treatment abbreviations: 150E = cows fed to consume ≥150% of NRC (2001) requirement for energy during the far-off and close-up dry period; 100E = cows fed to meet NRC (2001) requirements for energy during the far-off dry period, then ad libitum to meet or exceed energy requirements during the close-up dry period; and 80E = cows restricted to 80% of NRC (2001) requirement for energy during the far-off and close-up dry period.Far-off diets were fed from dry-off through 26 d before parturition, and close-up diets were fed from 25 d before expected parturition until parturition.One lactation diet was fed to all cows from parturition through 56 DIM.Asterisks indicate a difference among treatments at a given day (P < 0.05).(A) NR1H3, treatment, P = 0.08; day, P < 0.001; treatment × day, P = 0.05.(B) SREBF1, treatment, P = 0.08; day, P < 0.001; treatment × day, P = 0.001.

Figure 8 .
Figure 8.Effect of prepartum plane of energy intake on mRNA expression for genes related to inflammation in liver tissue.Data were collected with real-time quantitative PCR using the standard curve method for quantification of target cDNA in experimental samples.Pooled standard error bars are shown.Treatment abbreviations: 150E = cows fed to consume ≥150% of NRC (2001) requirement for energy during the far-off and close-up dry period; 100E = cows fed to meet NRC (2001) requirements for energy during the far-off dry period, then ad libitum to meet or exceed energy requirements during the close-up dry period; and 80E = cows restricted to 80% of NRC (2001) requirement for energy during the far-off and close-up dry period.Far-off diets were fed from dry-off through 26 d before parturition, and close-up diets were fed from 25 d before expected parturition until parturition.One lactation diet was fed to all cows from parturition through 56 DIM.(A) NFKB1, treatment, P = 0.16; day, P = 0.90; treatment × day, P = 0.22.(B) TNF, treatment, P = 0.38; day, P = 0.009; treatment × day, P = 0.22.(C) IL1B, treatment, P = 0.04; day, P = 0.33; treatment × day, P = 0.28.(D) SAA3, treatment, P = 0.22; day, P < 0.001; treatment × day, P = 0.80.
Janovick et al.: ENERGY INTAKE AND HEPATIC MRNA EXPRESSION

Table 1 .
Experimental diets fed to cows during the dry and lactation period 1

Table 2 .
Primer sets used for the relative quantification of cDNA Gene 1

Table 3 .
Frequency of occurrence of health problems in multiparous cows fed different amounts of energy prepartum