Rumen-protected choline and methionine during the periparturient period affect choline metabolites, amino acids, and hepatic expression of genes associated with one-carbon and lipid metabolism

Feeding supplemental choline and Met during the periparturient period can have positive effects on cow performance; however, the mechanisms by which these nutrients affect performance and metabolism are unclear. The objective of this experiment was to determine if providing rumen-protected choline, rumen-protected Met, or both during the periparturient period modifies the choline metabolitic profile of plasma and milk, plasma AA, and hepatic mRNA expression of genes associated with choline, Met, and lipid metabolism. Cows (25 primiparous, 29 multiparous) were blocked by expected calving date and parity and randomly assigned to 1 of 4 treatments: control (no rumen-protected choline or rumen-protected Met); CHO (13 g/d choline ion); MET (9 g/d DL-methionine prepartum; 13.5 g/d DL-methionine, postpartum); or CHO + MET. Treatments were applied daily as a top dress from ~21 d prepartum through 35 d in milk (DIM). On the day of treatment enrollment (d −19 ± 2 relative to calving), blood samples were collected for covariate measurements. At 7 and 14 DIM, samples of blood and milk were collected for analysis of choline metabolites, including 16 species of phosphatidylcholine (PC) and 4 species of ly-sophosphatidylcholine (LPC). Blood was also analyzed for AA concentrations. Liver samples collected from multiparous cows on the day of treatment enrollment and at 7 DIM were used for gene expression analysis. There was no consistent effect of CHO or MET on milk or plasma free choline, betaine, sphingomyelin, or glycerophosphocholine. However, CHO increased milk secretion of total LPC irrespective of MET for multiparous cows and in absence of MET for primiparous cows. Furthermore,


INTRODUCTION
The choline requirement for dairy cattle has not been defined (NRC, 2001), although it is considered to be an essential nutrient for other mammals, including humans (Zeisel and da Costa, 2009). Choline deficiency impairs an animal's ability to export lipids from the Rumen-protected choline and methionine during the periparturient period affect choline metabolites, amino acids, and hepatic expression of genes associated with one-carbon and lipid metabolism liver because it is required for synthesis of phosphatidylcholine (PC), the major phospholipid that comprises very-low-density lipoproteins (Yao and Vance, 1988). Phosphatidylcholine is synthesized through one of 2 pathways: (1) the cytidine diphosphate (CDP)-choline pathway using choline derived from the diet; and (2) the phosphatidylethanolamine N-methyltransferase (PEMT) pathway whereby a series of 3 methylation reactions occur to convert phosphatidylethanolamine to PC (Caudill, 2010). Because dietary choline is rapidly degraded in the rumen (Atkins et al., 1988), it is likely that in ruminants the majority of PC is synthesized via the PEMT pathway. Feeding rumen-protected choline (RPC) could lead to more PC synthesis via the CDPcholine pathway. DeLong et al. (1999) suggested that PC species that contained PUFA were derived from PEMT origin in rats. Thus, the profile of individual PC species in blood or milk that result when RPC is supplemented in periparturient cows could be indicative of a change in PC source.
Methionine is an AA that is used not only for protein synthesis, but also for synthesis of S-adenosylmethionine (SAM), one of the most important methyl donors in the body (Chiang et al., 1996). S-adenosylmethionine is required for de novo synthesis of PC via the PEMT pathway (Li and Vance, 2008), which is why Met has been investigated for its potential role as a lipotrope. Additionally, SAM is also implicated in the regulation of gene expression via histone methylation reactions (Mentch and Locasale, 2016).
Choline and Met metabolism are integrated through their participation in one-carbon metabolism. Choline indirectly serves as a methyl donor in the one-carbon metabolic pathway via betaine, the product of choline oxidation. Betaine is required for the regeneration of Met from homocysteine (Hcy) via the betainehomocysteine S-methyltransferase (BHMT) pathway (Martinov et al., 2010). Alternatively, Hcy can also be converted to Met via the methyltetrahydrofolate pathway by methionine synthase, an enzyme that requires vitamin B 12 to function (Preynat et al., 2009). Methionine also has the potential to indirectly influence PC synthesis via the PEMT pathway through SAM.
In an effort to derive nutritional strategies that improve cow health and performance during the periparturient period, attention has been given to nutrients involved in one-carbon metabolism such as choline and Met (Zhou et al., 2016b;Duplessis et al., 2017;Zenobi et al., 2018;Zang et al., 2019) due to their functions associated with lipid transport and metabolism, immunity, and gene expression. Previous investigations have shown that supplementing many of these nutrients during periods of negative energy balance that occur around the time of calving can modify hepatic expres-sion of genes associated with choline, Met, and lipid metabolism (Goselink et al., 2013;Zhou et al., 2017a;Coleman et al., 2019a,b) as well as the plasma AA profile (Zhou et al., 2017b;Zang et al., 2019). However, only the study by Zhou et al. (2017a,b) investigated specific effects associated with supplementation of choline, Met, or both during the periparturient period. Recent attention has also been given to the effects of methyl donors on the blood choline metabolite profile in lactating cows (Wang et al., 2021) in an effort to elucidate mechanisms by which these nutrients affect metabolism and performance of postpartum cows. We hypothesized that choline and Met supplementation would modulate methyl group metabolism via (1) the CDP-choline pathway using choline derived from the diet; and (2) the PEMT pathway.
The objective of this study was to determine if providing supplemental choline, Met, or both during the periparturient period modifies plasma AA concentrations, the choline metabolic profile of plasma and milk, as well as hepatic expression of genes associated with choline, Met, and lipid metabolism.

MATERIALS AND METHODS
All procedures that utilized animals were approved by the University of Maryland, College Park, Institutional Animal Care and Use Committee (Protocol #857801-6).

Animals and Study Design
Between March and December of 2017, 25 primiparous and 29 multiparous Holstein cows from the Central Maryland Research and Education Center (Clarksville, MD) were utilized in a randomized block design experiment with a 2 × 2 factorial treatment structure. The 2 factors were 0 or 60 g/d RPC (ReaShure, Balchem Corporation) and 0 or 12 g/d RPM (Smartamine M; Adisseo USA, Inc.) prepartum and 18 g/d RPM postpartum. Thus, the 4 resulting treatments included: (1) no RPC or RPM (control; CON); (2) 60 g/d RPC (CHO); (3) 12 g/d RPM prepartum and 18 g/d RPM postpartum (MET); and (4) a combination of CHO and MET treatments (CHO + MET; 60 g/d RPC + 12 g/d RPM prepartum and 18 g/d RPM postpartum). Although not reported in our previous paper on production responses (Potts et al., 2020), the number of animals included in the study were based on a power analysis to provide an 80% chance of detecting a 2 kg/d difference in milk production across the main effects of CHO and MET within parity at P < 0.05. Before the start of the experiment, cows were blocked by age (primiparous vs. multiparous) and expected calving 4561 date and randomly assigned to treatment. Treatments were applied daily as a top dress from 21 d before expected calving through 35 DIM. Because treatments were administered daily by project personnel, treatment assignments were not blinded. Additional details regarding basal pre-and postpartum diets and animal care and housing were described by Potts et al. (2020).

Sample Collection and Analysis
Blood was collected via coccygeal venipuncture before feeding on approximately −21, 7, and 14 d relative to calving. Because some cows calved early or late, prepartum samples were actually collected on −19 ± 2 d relative to calving. Samples were collected into one 10-mL evacuated tube containing potassium EDTA and one 10-mL evacuated tube containing sodium heparin. Tubes were immediately placed on ice and then centrifuged at 2,000 × g for 15 min at 4°C. Plasma aliquots were stored at −80°C until analysis.
Choline metabolite concentrations of plasma and milk (7 and 14 d postpartum) were determined by hydrophilic interaction liquid chromatography tandem MS (Artegoitia et al., 2014). Metabolites examined included betaine, free choline, glycerophosphocholine (GPC), sphingomyelin (SM), 16 species of PC, and 4 species of lysophosphatidylcholine (LPC). Amino acid concentrations of plasma were determined using the methods described by Sunny and Bequette (2010). Briefly, plasma samples (50 µL) were deproteinized with 500 µL of cold acetonitrile and evaporated to dryness under N 2 . Amino acids and organic acids were then converted to their respective t-butyldimethylsilyl derivatives by heating at 90°C for 1 h. Metabolites were separated by GC (HP-5ms, 30 m × 0.25 mm × 0.25 µm, Agilent Technologies) before selected ion monitoring of specific ion fragments with MS under electron ionization.
Approximately 150 mg of liver tissue was collected from each animal via percutaneous liver biopsies at approximately −21 d (actual prepartum samples were obtained on −19 ± 2 d) and 7 d relative to calving using a 14-gauge biopsy needle (14-gauge × 15 cm; Tru-Cut, Merit Medical). Further details regarding the liver biopsy procedure are described by Potts et al. (2020). Tissue was immediately snap frozen in liquid nitrogen and stored at −80°C. Because 2 multiparous cows in block 6 and 1 multiparous cow in block 3 were removed early from the experiment due to health problems (Potts et al., 2020), gene expression analysis was completed on liver samples obtained from the 5 completed blocks of multiparous cows (n = 20, 5 cows per treatment).
Hepatic mRNA levels of genes related to choline and Met metabolism (BHMT; PEMT; phosphate cytidylyltransferase 1A, PCYT1A; glutathione synthetase, GSS; 5-methyltetrahydrofolate-homocysteine methyltransferase, MTR) and lipid metabolism (carnitine palmitoyltransferase 1A; diacylglycerol O-acyltransferase 1, DGAT1; 3-hydroxy-3-methylglutaryl-CoA synthase 2, HMGCS2; microsomal triglyceride transfer protein, MTTP; peroxisome proliferator activated receptor α, PPARα) were determined via real-time quantitative PCR. Primers were designed (Table 1) using the National Center for Biotechnology Information primer BLAST Software (2012). Primers needed to (1) span an intron, (2) target a region as close to the 3′ end of the sequence as possible, (3) amplify all splice variants, (4) have an annealing temperature of 58 to 60°C, (5) have a guanine: cytosine ratio of 40 to 60%, (6) be between 18 and 30 nucleotides in length, (6) generate a PCR product that was 100 to 250 nucleotides in length, and (7) specifically target the gene of interest. Primers (25 nmol DNA oligos; standard desalting purification) were obtained from Integrated DNA Technologies and reconstituted in ultrapure water. Amplification efficiencies were determined for each primer pair by performing reverse-transcription quantitative PCR (QuantiTect SYBR Green PCR Kit, Qiagen Inc.) using 2-fold serial dilutions of pooled cDNA (1 µg). Efficiency was calculated using the equation [(10 (−1/slope) ) − 1], where slope is equal to the slope of the regression of the cycle threshold value on the log 10 (copy number). Efficiencies ranged from 0.96 to 1.08 for all primers tested. Primer specificity was verified by dissociation curve analysis, agarose gel electrophoresis, and sequencing of PCR products.
Liver total RNA was extracted using the RNeasy Lipid Tissue Mini Kit with on-column DNase digestion (Qiagen Inc.). Approximately 20 to 30 mg of liver tissue was weighed and kept frozen in liquid nitrogen until homogenization in 0.5 mL of QIAzol Lysis Reagent (Qiagen Inc.). The remainder of the RNA extraction protocol was carried out according to manufacturer instructions. After extraction, RNA was stored at −80°C. The concentration of RNA in each sample was determined using a commercially available kit (Quant-iT RiboGreen RNA Assay Kit, Catalog #R11490, ThermoFisher Scientific) and 1 µg of RNA was used for cDNA synthesis (QuantiTect Reverse Transcription Kit; Qiagen Inc.). For the reverse transcription reactions, a reaction of a pool of total RNA without reverse transcriptase was conducted as a control for genomic DNA contamination. Complementary DNA was not diluted before PCR analysis and was stored at −20°C.
The PCR reactions were carried out using a commercially available kit according to the manufacturer instructions (QuantiTect SYBR Green PCR Kit; Qiagen Inc.). The PCR reactions were performed in 96-well plates (VWR International, LLC) in a CFX Connect Real-Time PCR Detection System (Bio-Rad Labora- tories, Inc.) for 40 cycles using the following program: 95°C for 15 min (activation), 94°C for 15s (denaturation), 60°C for 30s (annealing), and 72°C for 30s (extension). Dissociation curve analysis ensured amplification of a single PCR product and absence from the no reverse-transcription and water controls. Data were normalized to phosphoglycerate kinase 1 and analyzed by the 2 −ΔΔCt method.

Statistical Analysis
Because production responses to treatment appeared to be different between primi-and multiparous cows (Potts et al., 2020), AA and choline metabolitic data were analyzed for primi-and multiparous cows separately. Postpartum blood AA and choline metabolite concentrations were analyzed using a repeated-measures mixed model (SAS Institute, version 9.4) that included the random effect of cow nested within block and fixed effects of wk relative to calving (1 or 2), the main effects of CHO and MET, and all 2-and 3-way interactions. Prepartum metabolite concentration, determined from samples collected in the morning before first treatment application (−19 ± 2 d relative to calving), was also included in the model as a covariate. Week relative to calving served as the repeated factor and a total of 8 covariance structures were tested, and the one that resulted in the lowest Akaike information criterion was selected for each variable. Milk choline metabolite yields were analyzed using a similar model that did not include a prepartum covariate measurement.
Hepatic mRNA level fold-changes (relative to postpartum CON) were analyzed in a mixed model that included the main effects of CHO and MET, their interaction, and the random effect of cow. Prepartum gene expression fold-change (relative to postpartum CON) was included in the model as a covariate. Thus, postpartum gene expression results are expressed as covariate-adjusted fold-changes relative to CON. Cook's distance was used to detect potential outliers using a threshold of 4 divided by number of observations (4/n). If a cow was identified as a potential outlier for more than 4 of the 10 genes of interest, she was removed from the analysis. Based on this method, results from 2 cows (one from CON and one from CHO) were considered outliers and removed from the final analysis. Statistical significance was declared at P ≤ 0.05, and tendencies were considered at P < 0.10.

Blood Choline Metabolites
Multiparous Cows. There were no effects of CHO or MET on plasma betaine, free choline, GPC, or total PC concentrations (Table 6). However, CHO decreased plasma SM concentrations (P = 0.04) and MET tended to decrease total LPC concentrations (P = 0.09). In line with the latter observation, MET reduced plasma concentrations of LPC 18:0 (P = 0.05; Table 5). In addition, feeding CHO increased plasma concentrations of LPC 16:0 and LPC 18:1, but this effect was dependent upon MET (CHO × MET: P = 0.05 and P = 0.02, respectively). Plasma total PC concentrations were unaffected and individual PC species were largely unaffected by CHO or MET (Table 7). However, CHO increased concentration of PC 18:0/18:1 during wk 1 postpartum (CHO × wk: P = 0.05) and tended to increase plasma concentrations of PC 16:0/20:4 when fed without MET (CHO × MET: P = 0.09). Furthermore, feeding CHO and MET together tended to reduce PC 16:0/18:2 concentrations (CHO × MET: P = 0.09).

Blood Amino Acids
Multiparous Cows. Feeding MET increased plasma Met concentrations (P < 0.01) but reduced Ser concentrations during wk 2 postpartum (MET × wk: P = 0.03; Table 10). Feeding MET without CHO  increased Phe concentrations and tended to increase Ile, Leu, and total branched-chain AA concentrations (CHO × MET: P = 0.04, P = 0.06, P = 0.09, and P = 0.08, respecitvely). There were few effects of CHO on plasma AA, but feeding CHO without MET decreased glutamate concentrations during wk 1 but not wk 2 postpartum (CHO × MET × wk: P = 0.03). Primiparous Cows. Results for plasma AA in primiparous cows are shown in Table 11. Feeding MET increased plasma Met concentrations (P < 0.01) but tended to reduce Gly concentrations (P = 0.08). Feeding MET reduced Ser concentrations (P = 0.01), but this effect tended to be intensified when CHO was fed with MET (CHO × MET: P = 0.08). The effect of CHO on plasma AA were minimal, although CHO increased Gln and tended to increase Gly concentrations during wk 1 postpartum (CHO × wk: P = 0.04 and P = 0.07, respectively).

Hepatic Gene Expression
Relative mRNA levels of hepatic genes associated with choline, Met, and fatty acid metabolism for a subset of multiparous cows (n = 20) is shown in Table 12.
Of the one-carbon metabolism genes examined, only BHMT and PCYT1A were affected by CHO or MET. Feeding MET reduced mRNA levels of PCYT1A (P < 0.01). Additionally, CHO tended to increase mRNA levels of BHMT and PCYT1A when fed without MET (Figure 1, CHO × MET: P = 0.09 and P = 0.10, respectively). The mRNA levels of HMGCS2 and PPARα tended to be reduced by CHO (P = 0.10 and P = 0.08). Feeding MET tended to reduce mRNA expression of DGAT1 (P = 0.08).

DISCUSSION
Our results suggest that supplemental choline and Met both seem to affect one-carbon metabolic pathways during the periparturient period but that responses are inconsistent between primi-and multiparous cows and appear to be dependent on Met availability. Although we did not examine metabolite flux through pathways associated with choline and Met metabolism, changes in metabolite secretion in milk and concentrations in plasma are indicative of shifts in these pathways. Alterations in hepatic expression of genes associated with one-carbon and lipid metabolism in multiparous cows further support this idea. However, the tendency for interactive effects on hepatic gene expression suggest that supplemental choline may play a role in promoting the CDP-choline and BHMT pathways but that the response is dependent on Met availability. Interactive

Choline Metabolites
Baseline milk secretion and plasma concentrations of free choline, betaine, SM, total LPC, and total PC were comparable between our study and that of previous studies (Artegoitia et al., 2014;de Veth et al., 2016;Zenobi et al., 2018;Wang et al., 2021).  2016) indicated an increase in milk free choline and betaine secretion when RPC was fed to lactating cows. In our study, CHO decreased milk secretion of betaine during the first wk postpartum in multiparous cows, which coincides with our observation for the tendency of CHO to increase hepatic BHMT expression for multiparous cows when fed without MET. This increased use of betaine by the liver likely reduced betaine availability to the mammary gland. In addition to putative increases in dietary choline oxidation to betaine, the overall increase in milk secretion of LPC and several LPC species for cows fed CHO also suggest that CHO increased choline metabolite production via the CDP-choline or PEMT pathways.
Feeding CHO tended to increase milk secretion of free choline and betaine for primiparous cows in our study, which is contrary to our observations for multiparous cows but similar to the findings reported by de Veth et al. (2016). This increase in free choline could suggest an adequate choline supply and increased ability to support the BHMT pathway for Met synthesis. However, in their study of choline and Met kinetics in lactating goats, Emmanuel and Kennelly (1984) indicated that very little choline is directed toward Met synthesis, which would explain why there was no effect of CHO on plasma Met concentrations in our study. Varied responses to CHO between primi-and multiparous cows for milk secretion of choline metabolites in our study are likely due, in part, to differences in physiological state and production level, which presumably affect the demand for methyl donors. Younger cows seem to have lower choline and Met requirements than older cows (Potts et al., 2020;Swartz et al., 2022) and methyl donor demands are presumably greater for high-producing, early-lactation cows (Pinotti et al., 2002). Therefore, it is possible that the methyl donor  balance of the primiparous cows in our study was more similar to that of the mid-lactation cows studied by de Veth et al., (2016), which may help explain the agreement of responses to CHO between the primiparous, but not multiparous, cows in our study and the cows in the study conducted by de Veth et al. (2016). Although CHO did not affect plasma betaine or free choline concentrations in primiparous cows, MET tended to increase plasma free choline concentrations, which suggests either a preferential increase in the use of Met as a methyl donor over choline or that Met was used to synthesize additional choline (Emmanuel and Kennelly, 1984). In contrast, Zhou et al. (2016a) reported no effect of MET on plasma free choline concentrations in multiparous, periparturient cows, which agrees with our observations for multiparous cows. Plasma concentration and milk secretion of SM was unaffected by treatment in primiparous cows, which agrees with data reported by de Veth et al. (2016). Milk secretion of SM was also unaffected by treatment in multiparous cows. However, the decrease in plasma SM concentrations for multiparous cows fed CHO was somewhat surprising, given that previous work by Zenobi et al. (2018) indicated an increase in plasma SM when dry cows were fed RPC and de Veth et al. (2016) showed no effect of postruminal choline supply on milk or plasma SM. Sphingomyelin is produced from PC and is involved with cell signaling associated with inflammation (Nixon, 2009;Avota et al., 2019), so a decrease in SM could suggest a reduced inflammatory response. This presents a mechanism by which choline may exert favorable effects with regard to minimizing inflammation during the periparturient period. Sphingomyelin has also been shown to be inversely associated with insulin sensitivity in monogastrics (Chang et al., 2019;Li et al., 2011), which would suggest greater insulin sensitivity for multiparous cows fed CHO in our study. However, additional research is needed to determine what, if any, effects dietary choline has on insulin sensitivity in dairy cows.

Lysophosphatidylcholine and Phosphatidylcholine
Work presented by de Veth et al. (2016) indicated no change in total LPC secretion in milk when supplemental choline was delivered via abomasal infusion or as RPC. The increased milk secretion of LPC in our study combined with the trend for reduced betaine secretion, is supportive of increased use of choline for PC synthesis. Because GPC secretion was also unaffected, CHO may have diverted additional PC toward LPC production, rather than to GPC. Our observations for primiparous cows were similar, although this effect seemed to depend on Met availability. This was largely   as well as individual LPC species in response to feeding protected choline sources (Zenobi et al., 2018;Wang et al., 2021). Although we did not observe an increase in plasma total LPC concentrations in response to CHO, our results did indicate increases in concentrations of LPC 16:0 and 18:1 when CHO was fed without MET for multiparous cows. The fact that MET negated this increase seems indicative of a shift in priority when sufficient Met was available. The tendency for MET to decrease total plasma LPC concentrations in multiparous cows further supports this idea. Assuming that Met supply was sufficient for primiparous cows in our study, this observation would coincide with the lack of effect of CHO or MET on plasma total LPC and LPC 16:0, 18:1, and 18:0 in primiparous cows. The observation for plasma LPC 18:1 concentration aligns with our observation for milk secretion of LPC 18:1 in multiparous cows, in which MET minimized the CHO-induced increase. Research in nonruminants has indicated that LPC has immunomodulatory effects (Liu et al., 2020), which provides a mechanism through which dietary choline or Met could interact with the immune system. Because MET seemed to mute the CHO-induced increases of these LPC species, it is possible that MET may have provided an alternative mechanism for activating the immune system, such as through glutathione (Martinov et al., 2010).
In addition to its effects on the immune system, LPC is also associated with energy metabolism. Both LPC 16:0 and LPC 18:1 are shown to be reduced during weight loss in obese humans (Heimerl et al., 2014). The elevation of these LPC species in multiparous cows fed CHO without MET could indicate that these cows were mobilizing less body fat. This idea is supported by the numerically greater body condition score, greater body weight, and lower plasma fatty acids observed for the multiparous cows fed this treatment during the postpartum period (Potts et al., 2020). The increase in LPC secretion in milk and concentration in plasma for cows fed CHO suggests an increase in PC conversion to LPC, which is supportive of a sufficient PC supply.
Overall, these findings demonstrate that supplemental choline can affect LPC synthesis, which has the potential to modulate the immune system as well as energy metabolism, but that these changes depend on the Met availability to the animal. Despite a lack of treatment effect on total PC secretion in milk and concentration in plasma, the changes in secretion of several PC species for multiparous cows suggests that both CHO and MET affected PC synthesis pathways. Research in rodents indicated that SFAand UFA-PC species are typically derived from the CDP-choline pathway, whereas PUFA-PC species are derived from the PEMT pathway (DeLong et al., 1999). Although these findings can provide valuable insight for interpreting our results, caution should be exercised when extrapolating these findings to ruminant animals, which typically have a relatively limited access to di-   etary choline in the small intestine. The fact that CHO and MET both enhanced secretion of several PUFA-PC species suggests that both promoted the PEMT pathway in multiparous cows. However, CHO only increased secretion of these PC species during wk 2, suggesting that choline can modulate the PEMT pathway but that this adaptation takes time. By contrast, the apparent increase in the PEMT pathway was not time-dependent for MET, as it increased PUFA-PC secretion, and likely flux through the PEMT pathway, irrespective of week. The PEMT pathway requires SAM to provide 3 methyl groups to produce PC from phosphatidylethanolamine (Cole et al., 2012). It is possible that there was insuf-ficient Met available to support the PEMT pathway via SAM during wk 1 for cows fed CHO. It appears that feeding MET helped to overcome this challenge irrespective of CHO supplementation. Together, these data suggest that, while both CHO and MET have the ability to support PEMT pathway activities, MET may act earlier to enhance availability of PEMT-derived PC-species through the provision of additional SAM. The general lack of treatment effect on the secretion of individual PC species for primiparous cows suggests minimal change to the apparent balance between the PEMT and CDP-choline pathways (DeLong et al., 1999). This finding agrees with the hypothesis that the primiparous cows in our study had a more positive choline balance and is supported by the increase in free choline and betaine secreted into milk when CHO was fed. If the primiparous cows had achieved a positive choline balance, it is likely that PC supply was already sufficient before CHO or MET supplementation and that there was little need to alter the pathways by which PC needs were being met.
There was minimal effect of CHO or MET on the plasma concentrations of individual PC species in multiparous cows, suggesting minimal shift in PC origin (DeLong et al., 1999). However, in primiparous cows, MET reduced plasma concentrations of several species derived from both PEMT and CDP-choline pathways during wk 2, suggesting that MET decreased PC needs during this time. These results are somewhat contradictory to those presented by Zenobi et al. (2018), who showed an increase in plasma concentrations of several PC species of CDP-choline pathway origin when RPC was fed to feed-restricted dry cows. However, previous research in rats suggested that dietary Met alters plasma concentrations of various PC species, specifically by reducing those that contain linoleic acid and PC 16:0/20:4 (Sugiyama et al., 1997), which would suggest a decrease in PC synthesized via the PEMT pathway. Indeed, we also observed a reduction in plasma concentration of several PEMT-PC species in primiparous cows fed MET during wk 2 postpartum, including PC 16:0/20:4. However, the biological significance of this change is unclear.
Research in humans suggests that the PC and LPC profiles differ among the lipoprotein fractions (Wiesner et al., 2009). Thus, the treatment-induced changes in plasma PC and LPC profiles for cows in our study could reflect alterations in the lipoprotein profile. However, we did not measure plasma lipoprotein fractions in our study. Although the measurement of lipoprotein fractions in bovine plasma is more complex than for humans (Grummer et al., 1986), additional research to examine lipoprotein responses to dietary choline and Met supplementation is needed. Potts et al.: PERIPARTURIENT RUMEN-PROTECTED CHOLINE AND METHIONINE Figure 1. Differences in postpartum hepatic mRNA expression of the genes BHMT and PCYT1A that are associated with Met and choline metabolism, respectively, in multiparous cows (n = 20) fed the control (CON) diet or the CON diet plus rumen-protected choline (RPC; 60 g/d; CHO), rumen-protected Met (RPM; 12 g/d prepartum, 18 g/d postpartum; MET), or both (CHO + MET) from samples taken ~7 d after calving. Data are expressed as covariate-adjusted fold-changes relative to the CON treatment. For both genes, MET × CHO tended to be significant (P = 0.09 and P = 0.10, respectively). For PCYT1A, the effect of MET was significant (P < 0.01).

Amino Acids
All changes in plasma AA concentrations due to treatment occurred without any changes in protein intake, as DMI was not affected by treatment (Potts et al., 2020).
Of the nonessential AA, only plasma Ser, glutamate, Gln, and Gly were affected by CHO or MET. Contrary to these observations, previous work with periparturient dairy cows by Zhou et al. (2017b) indicated no effect of Met or choline on plasma Ser, glutamate, Gln, or Gly concentrations. Furthermore, they observed that Met increased plasma Ala, aspartate, Asn, Pro, and glutamate, which is also contrary to our results. It is curious that MET decreased plasma Ser for both primi-and multiparous cows in our study, given that Emmanuel and Kennelly (1984) indicated that Met can be used to synthesize Ser to a limited extent in lactating goats and that Ser is required for the first step toward glutathione production from Hcy, as well as for regeneration of 5, 10-methyl-tetrahydrofolate from tetrahydrofolate in the methyltetrahydrofolate pathway (Selhub, 1999). However, because Ser is a nonessential AA and the rate-limiting substrate for glutathione production is Cys (Lu, 2013), it is unlikely that this reduction is indicative of changes in glutathione production. For the same reasons, the decrease in glutamate concentration during wk 1 postpartum for multiparous cows fed CHO without MET are also probably not related to alterations in glutathione production. Instead, these observations are more likely indicative of increased protein synthesis. This aligns with the absence of a treatment effect on hepatic GSS expression and the increased milk protein content we observed for multiparous cows fed MET in our study (Potts et al., 2020).
Consistent with previous work (Blum et al., 1999;Zhou et al., 2017b), MET increased plasma Met for both groups of cows, indicating that the Met fed as RPM was absorbed. However, effects of supplemental Met on plasma concentrations of other essential AA in previous studies has been variable. Previous research reported by Zhou et al. (2017b) indicated that Met increased plasma concentrations of several other essential AA in periparturient cows, including Arg, Lys, and Trp. Contrary to these findings, the essential AA affected by MET for multiparous cows in our study were Ile, Leu, Met, and Phe. Previous work (Blum et al., 1999;Preynat et al., 2009) showed that RPM decreased plasma concentrations of the branched-chain AA Leu, Ile, and Val. Interestingly, we observed the opposite, where MET tended to increase concentrations of the branched-chain AA Ile and Leu. Although changes in branched-chain AA concentrations can be indicative of changes in DMI, intake was similar across treatments in our study (Potts et al., 2020). Alternatively, this response could be explained by alterations in muscle AA metabolism because it occurred in absence of a change in DMI. A recent examination of muscle tissue from periparturient dairy cows fed RPM that demonstrated the ability of MET to affect key pathways related to AA and protein metabolism (Thanh et al., 2022) supports this idea. Leucine has been long been shown to be an important mediator of protein metabolism (Stipanuk, 2007). More recently, both Leu and Ile have been shown to stimulate protein synthesis in bovine mammary cells via mammalian target of rapamycin (Appuhamy et al., 2012). Thus, the increase in plasma Ile and Leu concentrations for multiparous cows in our study could help explain the increased milk protein concentration observed for multiparous cows fed MET (Potts et al., 2020).

Liver Gene Expression
The PCYT1A gene encodes phosphate cytidylyltransferase 1 choline α, which is the enzyme that catalyzes the rate-limiting step of the CDP-choline pathway (Li and Vance, 2008). Our observations lend support to the hypothesis that supplemental choline increases PC synthesis via the CDP-choline pathway. This change coincided with the increase in milk secretion of several CDP-choline-derived PC species by CHO and is suggestive of the ability for supplemental choline to increase PC synthesis via this pathway in periparturient cows. Because this response seemed to be dependent on MET supplementation, it is possible that inclusion of MET redirected supplemental choline for another purpose. Zhou et al., (2017a) also observed an increase in PCYT1A mRNA expression for periparturient cows fed choline, although they indicated that supplemental Met also increased PCYT1A mRNA expression. Contrary to our results, previous work by Zhou et al. (2017a) indicated that supplemental choline did not affect expression of BHMT for periparturient cows. However, CHO tended to increase BHMT expression when fed without MET, suggesting an increase in Met recycling from Hcy via the BHMT pathway for these cows. This response seemed to be dependent on MET supplementation, which could be indicative of a reduced need to synthesize Met from Hcy via this pathway when Met supply is sufficient. In support of this hypothesis, recent in vitro work by Zhang et al. (2016) showed a reduction in BHMT mRNA expression with increasing concentrations of Met, indicating that when Met is in ample supply, less Met is regenerated from Hcy through the BHMT pathway.
Lack of an effect of CHO or MET on mRNA expression of PEMT, GSS, or MTR suggests that the effects of these nutrients on these pathways do not occur at the transcription level. This observation for PEMT expression aligns with the lack of effect of CHO or MET on the plasma concentrations PEMT-derived PC species in multiparous cows. However, this is contrary to previous work which indicated an increase in hepatic mRNA expression of PEMT in periparturient cows fed supplemental Met (Preynat et al., 2010;Osorio et al., 2014;Zhou et al., 2017a). It is possible that effects of CHO or MET do not occur at the transcription level for MTR or GSS. This idea is supported by the work of Zhou et al. (2017a), who reported a decrease in MTR activity, but not MTR mRNA expression, for periparturient cows fed supplemental choline or Met. It is also possible that, with only 5 cows per treatment, we did not have sufficient statistical power in our study to detect changes in mRNA expression of these genes.
The mRNA levels of HMGCS2 and PPARα tended to be reduced by CHO (Table 12), which would suggest reductions in ketogenesis and hepatic fatty acid oxidation by cows fed choline. However, CHO did not affect postpartum plasma BHBA or nonesterified fatty acid concentrations for multiparous cows (Potts et al., 2020). Previous research did not show changes in HMGCS2 (Morrison et al., 2018) or PPARα (Goselink et al., 2013;Morrison et al., 2018) mRNA expression when choline was fed to periparturient cows. Feeding MET tended to reduce mRNA expression of DGAT1 (Table 12), suggesting a reduction in liver triglyceride (TG) synthesis, although there was no effect of MET on liver TG content (Potts et al., 2020). The lack of a CHO effect on genes associated with TG synthesis observed in our study and by others (Goselink et al., 2013) suggests that choline has little effect on liver TG synthesis during the periparturient period. Our observations for hepatic gene expression agree with the lack of treatment effect on energy balance, plasma fatty acid concentrations, and liver TG content for multiparous cows in this study (Potts et al., 2020). However, it is worth noting that the multiparous cows in our study did not experience severe negative energy balance during the postpartum period (−3.1 Mcal/d) and that liver TG accumulation was relatively mild (4.9% DM), which may have affected gene expression responses.
Feeding CHO or MET did not affect mRNA levels of MTTP in this study. Microsomal triglyceride transfer protein is an enzyme involved in packaging TG into very low density lipoprotein in the liver (Bernabucci et al., 2004). An increase in MTTP mRNA expression could indicate an increase in the rate of very low density lipoprotein formation, which is the proposed route by which choline could reduce hepatic TG concentration in dairy cows (Cooke et al., 2007). Previous research has indicated variability in the effect of choline supple-mentation on MTTP expression, where some studies showed an increase (Goselink et al., 2013;Caprarulo et al., 2020) and others (Morrison et al., 2018) showed no effect. Our observations are consistent with those of Morrison et al. (2018) and suggest a limited effect of CHO on TG export for the cows in the current study.

CONCLUSIONS
With the exception of milk secretion of 3 LPC species, effects of choline and Met on apparent choline metabolism were not consistent for primi-and multiparous cows. Although there were subtle changes in the PC profile of milk and plasma in response to supplemental choline or Met, substantial, consistent changes did not suggest definitive shifts in the balance of PC synthesis via the PEMT or CDP-choline pathways. Supplemental choline alone tended to increase hepatic expression of genes associated with choline metabolism in multiparous cows, which suggests a probable role in promoting the CDP-choline and BHMT pathways. However, when supplemental Met was combined with supplemental choline, this response seemed to be negated, which suggests a dependency on Met availability to the cow. This is further supported by the differences in responses between primi-and multiparous cows, who likely differ in their choline and Met requirements. Additional work is required to explore how choline and Met affect onecarbon metabolism during the periparturient period and how these responses may differ with choline and methionine availability to the cow.