Effect of rumen-protected choline on dairy cows' metabolism, immunity, lactation performance, and vaginal discharge microbiome

Rumen-protected choline ( RPC ) promotes benefits in milk production, immunity, and health in dairy cows by optimizing lipid metabolism during transition period management and early lactation. However, the RPC success in dairy cows depends on choline bioavailability which is affected by the type of protection used in rumen-protected choline. Therefore, our objectives were to determine the effects of a novel RPC on DMI, markers of metabolism and immunity, and lactation performance. Dry Holstein (n = 48) cows at 245 ± 3 d of gestation were blocked by parity and assigned to Control or RPC treatment within each block. Cows enrolled in the RPC treatment received 15 g/d of CholiGEM TM (Kemin Industries, Cavriago RE, Italy) from 21 d prepartum and 30 g/d of CholiGEM TM from calving to 21 d postpartum. During the transition period, DMI was measured daily, and blood was sampled weekly for energy-related metabolites [e.g., β-hydroxybutyrate ( BHB ), glucose, and nonesterified fatty acids ( NEFA )] and immune function markers [e.g., haptoglobin ( Hp ) and lipopolysaccharide-binding protein (LPB )]. Vaginal discharge samples were collected at the calving and 7 d postpartum ( DPP ) and stored in microcentrifuge tubes at −80°C until 16S rRNA sequencing. The main responses of body condition score, body weight, DMI, milk yield, milk components, and immune function markers were analyzed using the GLIMMIX procedure of SAS with the effect of treatment, time, parity, and relevant covariates added to the models. The relative abundance of micro-biome α-diversity was evaluated by 3 indexes (Chao1, Shannon, and Simpson) and β-diversity by principal coordinate analysis and PERMANOVA. There were no differences in DMI in the pre-and postpartum periods.


INTRODUCTION
Recent scientific evidence from 2 meta-analyses (Humer, Bruggeman andZebeli, 2019, Arshad et al., 2020) helped to consolidate the message that rumen-protected choline (RPC) supplemented during the transition period improves pre-and postpartum dry matter intake (DMI), milk yield, milk composition, and metabolism of dairy cows.It is essential to note, though, that the success of RPC feeding is dependent on factors such as the timing of feeding both pre-and postpartum Effect of rumen-protected choline on dairy cows' metabolism, immunity, lactation performance, and vaginal discharge microbiome (Lima et al., 2012, Arshad et al., 2020, Bollatti et al., 2020a), as well as the bioavailability of choline, which is affected by the type of encapsulation used (Humer, Bruggeman and Zebeli, 2019).Although RPC is often linked to benefits in performance and metabolism, its impact on cow health has been less consistent (Lima et al., 2012, Bollatti et al., 2020b).Research suggests that RPC tends to decrease the incidence of retained fetal membranes (RFM) and mastitis (Lima et al., 2012) but it does not impact hypocalcemia, metritis, displaced abomasum, and ketosis (Lima et al., 2012, Zhou et al., 2016, Bollatti et al., 2020b).Albeit unclear why discrepancies in health outcomes are present in the scientific literature, the conundrum might be an artifact of the nature of studies evaluating nutritional interventions.In general, feed trials are designed to accurately deliver the nutritional intervention and measure feed intake over time to test hypotheses related to performance.However, it might lead to limited statistical power to identify a reduction in the incidence of health disorders that are not highly prevalent and expected to change dramatically (Pirestani and Agkakhani, 2017, Zenobi et al., 2018a, Bollatti et al., 2020b).
A common approach to assess the potential benefits of RPC for cow health is the relationship between the immune system and the risk of disease development.Some studies suggest that feeding RPC may decrease inflammation indicators, including lower rectal temperatures, reduced plasma concentrations of haptoglobin prepartum and fibrinogen postpartum, and reduced levels of tumor necrosis factor-α, among other factors (Zenobi et al., 2020).Choline deficiency has been associated with disrupted gut integrity in rats (Takahashi, Mizunuma and Kishino, 1982), and compromised gut integrity in cattle has been linked to systemic inflammation that can be assessed by an increased presence of lipopolysaccharide-binding protein (LBP) (Kvidera et al., 2017a, Dickson et al., 2019, Koch et al., 2019).The endometrium also relies on choline as a building block to cell integrity.At parturition, dairy cows face a concomitant sharp decrease in immune function at the uterine environment and breach of physical barriers such as the cervix and endometrium, which allows for rapid colonization of the uterus by bacteria that are ubiquitous in dairy cows' environment (Elliott et al., 1968, Kehrli and Goff, 1989, Cai et al., 1994, Sheldon and Dobson, 2004).A case has been made that compromised gut integrity is a major source of chronic endotoxemia and a common denominator derailing adaptation to lactation in dairy cows (Kvidera et al., 2017a, Kvidera et al., 2017b, Dickson et al., 2019, Koch et al., 2019).However, the aggregation of immunological, physical, and environmental challenges suffered by the uterus in parturition makes this organ another strong candidate to have an altered microbiome that can trigger systemic endotoxemia, increased pro-inflammatory cytokines, reduced feed intake, impaired metabolism, and lactation performance.
In addition, choline has been shown to play a role in the gut microbiome and immune system (Harris et al., 2017).Studies also suggest that choline supplementation can influence the composition and function of the gut microbiome, increasing the abundance of beneficial bacteria, such as Lactobacillus (Zimmerman, Gyawali and Ibrahim, 2017), or decreasing the abundance of potentially harmful bacteria, such as Escherichia coli (Consoli et al., 2022, Ferreri et al., 2022).These changes in the gut microbiome occur due to the part choline plays a role in the metabolism of certain types of bacteria that contain choline TMA-lyase (Zeisel, Wishnok and Blusztajn, 1983).However, the impact of RPC on the uterine microbiome in promoting the growth of beneficial bacteria and enhancing local immune function has not been explored.
Choline is also utilized as a nutrient by rumen microbes (Sharma and Erdman, 1989, Arce-Cordero et al., 2021, Arce-Cordero et al., 2022), reducing the amount available for gut absorption.Studies have shown that the type of encapsulation to protect choline from ruminal degradation is crucial to increase plasma choline concentrations metabolites and achieving the overall benefits provided by choline (Humer, Bruggeman andZebeli, 2019, France et al., 2022).
While RPC benefits for dairy cows during the transition period have been widely investigated, studies using a RPC impact on lactation performance and metabolism for the specific formulation used in the current study are still needed.We hypothesized that RPC improves lactation performance, energy metabolism, and immunity markers.The objectives were to determine the effects of feeding RPC on pre-and postpartum DMI, body condition score (BCS), body weight (BW), milk yield and composition, energy-related metabolites [e.g., β-hydroxybutyrate (BHB), glucose and nonesterified fatty acids (NEFA)], immune function markers [e.g., haptoglobin (Hp) and LBP], disease incidence, and vaginal discharge microbiome as a proxy for uterine health.

Power analysis
Power analyses were performed to calculate sample sizes using G Power 3 (Universität Düsseldorf, Germany).The sample size was calculated to allow sufficient experimental units to detect an increase in milk production of 1 kg/day as previously detected (1) with α = 0.05 and β (the probability of a type II error) of 0.20 in a one-tailed test.Under these assumptions, 20 experimental units per treatment were deemed necessary.Twenty percent additional cows were enrolled in each treatment to prevent attrition throughout the study.Thus, a total of 24 cows per treatment were enrolled in the study.

Housing, diets, and experimental design
All procedures for this study were approved by the Institutional Animal Care and Use Committee of the University of California, Davis, under protocol #22440.The experiment was conducted at the Teaching and Research Facilities at the University of California, Davis.The study followed a complete randomized block with dry Holstein cows at 245 ± 3 d of gestation being blocked by parity (Primiparous vs. Multiparous) and within each block assigned to Control or RPC treatment (Figure 1).In the previous 305-d lactation, no significant differences (P = 0.85) were observed between treatments for multiparous cows (Control = 29648.3± 769.9 vs. RPC = 29446.7 ± 769.9).
Cows enrolled in the RPC treatment received 15 g/d of CholiGEM TM (Kemin Industries, Cavriago RE, Italy), an encapsulated choline chloride product with 60% choline chloride concentration and approximately 90% bioavailability, from 21 d prepartum, and 30 g/d of CholiGEM TM from calving to 21 d postpartum (DPP) (Figure 1).During the morning feed, the RPC product was top-dressed onto the total mixed ration (TMR) at each cow's specific feeding gate.To ensure the daily amount designated, the RPC product was offered when the cow had access to the TMR and visually consumed the top layer on which RPC was top-dressed.
A total of 53 cows were initially enrolled in the study, but 5 cows were removed due to various reasons (e.g., displaced abomasum during the feeding period, calving after less than one week in the prepartum diet, leg nerve injury, manager's decision due to changes in behavior).Forty-eight cows were included in the final population used for all statistical analyses.Ingredients and nutrients of diets in pre-and postpartum are shown in Table 1.The average number of days supplemented with RPC during the prepartum in the RPC treatment was 19.0 ± 0.7 d for multiparous and 17.0 ± 0.7 d for primiparous.

DMI, BCS, and BW
Cows in both treatments were assigned to an individual feeding gate (American Calan Inc., Northwood, NH) for a one-week adaptation period before initiating intake measurements, which were conducted from 21 d prepartum to 21 DPP.Each cow was fed twice daily (6 a.m. and 6 p.m.) after enrollment and feed was pushed forward in each Calan gate every 4 h.Lactating cows received the same diet to meet or exceed the nutrient requirements of a lactating Holstein cow, producing 45 kg/day of milk with 3.5% fat and 3.2% true protein when DMI is 25 kg/day (NRC, 2001).Cows were fed 110% of the expected consumption, and individual feed intake was measured by weighing the amount of feed offered and refused daily.From 22 to 150 DIM, cows were moved to a free-stall barn equipped with headlock stations.
Total mixed rations, forages, and grain supplements were sampled monthly and evaluated for dry matter (DM) content (55°C for 48 h).Daily DMI was calculated based on feed intake corrected for DM contents of feed offered and refused.Monthly composites of dried dietary ingredients and orts were analyzed for chemical composition (Dairyland Laboratories, Inc., Arcadia, US).Diet energy density was estimated at 3x maintenance intake (NRC, 2001) based on chemical analyses of individual ingredients.

Milk yield and composition
Cows were milked twice daily at 4 a.m. and 4 p.m. in an 8-single parlor, and milk yield was recorded automatically (DeLaval, Tumba, SE) from 0 to 150 DIM.Milk samples were collected bi-weekly in 2 consecutive milkings from 0 to 150 DIM to evaluate fat, protein, lactose, and somatic cell count (SCC).Samples from 2 consecutive milkings were collected for each cow into tubes containing 2-bromo-2-nitro-1,3-propanediol preservative, and a composite sample representative of the day was submitted for composition analyses.The SCC was transformed to SCS for statistical analysis according to the following formula: SCS = Log 10 (SCC/12.5)/Log 10 (2).
The final concentration of milk was calculated using the daily milk yield and the respective bi-monthly milk composition from each sampling.Milk yields were corrected for 3.5% fat (FCM) and energy (ECM) up to 150 DIM and calculated according to (Council, 2001, NRC, 2001)

Blood sampling
Blood was sampled weekly from 21 d pre-to 21 DPP to evaluate energy-related metabolites (glucose, NEFA, and BHB) and characterize immune function markers (LBP and Hp).Samples were collected by puncture of coccygeal vessels into evacuated tubes containing K2 EDTA (Vacutainer, Becton Dickinson, Franklin Lakes, USA), following the IACUC-31 procedure policy for blood collection volumes.Blood sampling was performed after the morning milking before cows gained access to fresh feed.After collection, samples were placed on ice until returned to the laboratory, where the plasma was separated by centrifugation (2,000 × g, 15 min, 4°C) and stored at −80°C until assayed.All samples, including the standards, were tested in duplicate.

Energy-related metabolites analysis
Glucose, NEFA, and BHB were analyzed in an automated clinical chemistry analyzer (RX Daytona, Randox Laboratories Ltd., Crumlin, UK) using reagents supplied by the manufacturer.Intra-and inter-assay coefficients of variation were 1.8 and 3.9% for glucose, 0.7 and 1.8% for NEFA, and 1.0 and 1.4% for BHB.Cows with a BHB concentration greater than or equal to 1.2 mmol/L were classified as having subclinical ketosis.

Immune function markers analysis
The optical density values of samples and standards were measured on an automatic microplate reader (Spectramax ® ABS 96; Molecular Devices, San Jose, USA) at 450 nm.The concentration of LBP was assessed using an ELISA commercial kit with cross-reactivity to bovine LBP (Hycult Biotech Inc., Uden, The Netherlands) according to the manufacturer's instructions.
Plasma Hp concentration was determined using a colorimetric assay via quantification of the haptoglobin/ hemoglobin complex by the estimation of differences in peroxidase activity (Makimura and Suzuki, 1982).Assays were performed in 16 × 100 mm borosilicate tubes; briefly, 5 µL of sample, deionized water (blank), or standard were added to 7.5 mL of a solution containing 0.6 g/L of o-Dianisidine, 13.8 g/L of sodium phosphate monobasic, and 0.5 g/L EDTA (pH = 4.1).Immediately, 25 µL of a solution containing 0.3 g/L bovine hemoglobin was added to each tube, followed by water-bath incubation at 37°C for 45 min.After, 100 µL of a freshly prepared 156 mM hydrogen peroxidase solution was added to each tube, and samples were incubated at room temperature for 60 min.Then, 200 µL of each tube was transferred to a 96-well plate and the optical density read at 450 nm in a SpectraMax ABS 96-well absorbance microplate reader (Molecular Devices LLC, San Jose, USA).Haptoglobin concentration was calculated using standard curves generated by serial dilutions of a sample of known concentration determined by a commercially available ELISA kit as previously described (Cooke and Arthington, 2013).Intra and inter-assay coefficients of variation were 5.2 and 7.3%, respectively.Standard curves were generated by serial dilution of a known concentration Hp stock sample as previously validated (Machado et al., 2020).

Disease incidence
Data from farm routinely diagnosed disorders such as RFM, metritis, milk fever, mastitis, and displaced of abomasum were collected from calving to 150 DIM.We evaluated vaginal discharge at calving 3, 7, and 14 DIM using the Metricheck device (Simcro, Hamilton, NZ).Discharge was scored using a 1 to 5 scale adapted (Williams et al., 2005), where 1 = clear mucus; 2 = mucus with flecks of pus; 3 = discharge containing ≤50% mucopurulent material; 4 = discharge containing >50% purulent material; and 5 = watery, reddish, or brownish color fetid discharge.Rectal temperature was assessed daily from 0 to 7 DIM and on alternate days thereafter until 14 DIM.Metritis was defined for cows with vaginal discharge score 5 at 3, 7, or 14 DIM associated or not with rectal temperature ≥39.5°C.Purulent vaginal discharge was defined for cows with score ≥3 at 30 DIM.

Vaginal discharge sampling
Vaginal discharge (VD) samples were collected at calving and 7 DIM to evaluate the microbiome as a proxy for the effect of RPC on uterine immune competence.Metricheck device (Simcro Tech Ltd., Hamilton, NZ) were used to collect the samples.Briefly, after cleaning the vulva with a paper towel and alcohol solution (70%), the instrument was rinsed with chlorhexidine solution (0.05%) followed by alcohol solution (70%) and introduced into the vagina and positioned near the cervix to scoop the sample from the cow.Each VD sample was stored in a 2 mL sterile microcentrifuge tube immediately placed in ice, and later stored at −80°C for DNA extraction.

DNA extraction, library preparation, and sequencing
DNA extraction was carried out using Mag-Bind® Universal Pathogen 96 Kit (Omega Bio-Tek, Norcross, GA) in accordance with manufacturer instructions.Library preparation and sequencing were conducted similarly to those described previously (Kozich et al., 2013).Amplification was performed through polymerase chain reaction (PCR) in a Bio-Rad C1000 TouchTM Thermal Cycler (BIO-RAD, Hercules, USA).The V4 region of the 16S rDNA gene was amplified using the Earth Microbiome Project barcoded (forward: GTGYCAGCMGCCGCGGTAA and reverse: GGAC-TACNVGGGTWTCTAAT) bacterial primers through an initial 95°C denaturation for 5 min, followed by 30 cycles of 30 s at 95°C, 30 s at 55°C, 1 min at 72°C, and 5 min for final elongation at 72°C.Primers and small DNA fragments were removed using a 1% low melting agarose gel extraction kit (National Diagnostics, Atlanta, GA, USA).Purification and normalization of amplicons were performed using magnetic beads Mag-Bind ® TotalPure NGS (Omega Bio-Tek, Norcross, GA) in accordance with manufacturer instructions.Total DNA concentration was measured using Qubit® fluorometric quantification kit (Thermo Fisher Scientific Inc., Carlsbad, USA) considering that the concentration of pure double-stranded DNA with an A 260 of 1.0 was 50 µg/ mL.Adapters were added to the amplicons, and a DNA library was prepared by equally pooling them together; qualitative real-time PCR was used for quality check.A total of 40 samples were sequenced using an Illumina MiSeq 2500 platform.Sequecing data were deposited in the Sequence Read Archive of the National Center for Biotechnology Information under the Bio-Project accession number PRJNA980301.

Bioinformatics and statistical analysis
Data were analyzed by the MIXED or GLIMMIX procedures of SAS (version 9.4, SAS/STAT, SAS Institute Inc., Cary, NC).Continuous data were tested for the distribution of the residuals using Shapiro-Wilk and homogeneity of variance by plotting residuals against predicted values after fitting the statistical models.
The statistical model included the fixed effects of treatment (Control vs. RPC), parity, DMI at week before enrollment and BW at enrollment as covariates, the interactions between treatment, treatment and day, treatment and parity, and the random effect of cow within the treatment.
For the bioinformatics analyses, a metadata was prepared considering the same grouping arraignment cows were in for regular statistical analyses.Upstream analysis of the sequenced amplicons was performed in R. Sequences were denoised using the DADA2 pipeline (Callahan et al., 2016), in which demultiplexed fastq files were inspected, filtered, and trimmed based on their quality scores and error rates.Chimeras were removed, and an ASV table was created.Taxonomy was assigned using the 16S rRNA SILVA v138 database (Pruesse, Peplies and Glockner, 2012) with the phyloseq package (McMurdie and Holmes, 2013).Total taxa were then split into taxonomy levels, and the relative abundance of the ASVs within each taxonomy level was calculated using the phyloseq package.Two microbiome RPC samples had uncountable reads, likely due to library preparation, and were discarded from the remaining analyses.Downstream analysis was performed by testing microbial community differences within (α-diversity) and between (β-diversity) dietary treatments (Control vs. RPC).Alpha-diversity indexes [(total sequences, chimeras, unused sequences, Chao 1, Shannon, Simpson, and Rarity (low and rare abundant taxa)] were calculated using the microbiome and vegan packages (Shetty et al., 2017, Oksanen et al., 2020).Data were normalized using Center-Log Ratio (CLR) transformation (Gloor et al., 2017, Weiss et al., 2017, Quinn et al., 2019, Quinn and Erb, 2021) for the generation of Non-Metric Multidimensional Scaling (NMDS) for graphical visualization of β-diversity differences.Two NMDS plots were constructed to dissect the vaginal discharge microbial community differences between Control and RPC primiparous and multiparous cows.Graphs were generated using the ggplot2, dplyr, hrbrthemes, viridis, ggsci, and RColorBrewer packages.Permutational multivariate analyses of variance [PER-MANOVA; (Anderson, 2008)] were performed to test the bacterial community's dispersion differences with the respective data sets from NMDS.A final linear mixed model similar to previously described was used to assess the microbial differences when considering the interaction between treatment treatment and day relative to calving.Microbial taxa significant at P ≤ 0.05 or that tended to be different at 0.05 < P ≤ 0.10 are displayed.

DMI, BW, BCS, and lactation performance up to 21 DIM
Supplementation of RPC during the transition period had no effect on DMI pre-(P = 0.13; Table 2) and postpartum (P = 0.30, Table 2).Supplementation of RPC had no effect on lactose (P = 0.96, Table 2) during the first 21 DIM.However, cows fed RPC increased ECM (P = 0.05), 3.5% FCM (P = 0.004), milk fat percent (P < 0.001), and fat yield (P < 0.001) compared with the cows in the Control (Table 2).While there were changes over time in ECM, 3.5% FCM, fat, protein, and lactose (P < 0.001), but there were no interactions between time and treatment (P > 0.05) for the variables assessed.There were effects of treatment (P < 0.001), parity (P < 0.001), and an interaction treatment by parity (P < 0.001) for protein percentage with primiparous RPC having lower protein percentage than primiparous Control cows.There were effects of treatment (P < 0.001), parity (P < 0.001), and an interaction treatment by parity (P < 0.001) for lactose percentage with primiparous RPC having higher lactose percentage than primiparous Control cows.
At enrollment (week −1) and during postpartum wk 1 and 2, there was no significant difference in average BW between the treatments, as illustrated in Figure 2A (P > 0.05).However, for multiparous cows, the weight in the Control was higher compared with the RPC treatment (P = 0.05, Figure 2A).For primiparous cows, no significant BW differences were observed across the weeks (P > 0.05, Figure 2B).Likewise, the average BCS did not differ between Control and RPC treatments at enrollment (week −1), postpartum wk 1, 2, and 3 for multiparous (P > 0.05, Figure 2C) and primiparous (P > 0.05, Figure 2D).

Lactation performance
Although milk yield was unaffected by RPC up to 150 DIM between treatments (P > 0.10) when milk was adjusted for 3.5% fat and energy, multiparous and primiparous cows fed RPC had greater ECM, 3.5% FCM, milk fat percent, and milk fat yield (P < 0.001, Table 2), with the interaction between treatment and parity (P < 0.001, Table 2).Over time, there were noticeable changes in ECM, 3.5% FCM, fat, protein, and lactose yields (P < 0.001), but there were no interactions between treatment and time (P > 0.05).There were no effects on protein (P = 0.70) and lactose (P = 0.81) yields (Table 2), but time and parity were significant (P < 0.001).There were effects of treatment (P < 0.001), parity (P < 0.001), and an interaction treatment by parity (P < 0.001) for protein percentage with primiparous RPC having lower protein percentage than primiparous Control cows.There were effects of treatment (P < 0.001), parity (P < 0.001), and an interaction treatment by parity (P < 0.001) for lactose percentage with primiparous RPC having higher lactose percentage than primiparous Control cows.

Energy-related metabolites and disease incidence
There were no effects of RPC on glucose (P = 0.54) or NEFA (P = 0.69), regardless of parity (Figure 3A and 3B).No interaction was found between the treatment and week for NEFA in multiparous (P = 0.55, Figure 3C).However, an interaction between the treatment and week was detected in primiparous (P = 0.03, Figure 3D), with plasma concentration of NEFA increasing up to 2 weeks postpartum in cows fed RPC, while in Control cows after week one postpartum NEFA decreased.
Even though there were no overall differences in BHB between treatments for multiparous (Figure 4A) and primiparous (Figure 4B), multiparous cows in the RPC treatment tended to have a lower incidence of subclinical ketosis cases by 3 weeks postpartum (12.7 ± 3.3%  There was an effect of time (P < 0.001) for ECM, 3.5% FCM, fat, protein, and lactose but no interaction between time and treatment were found (P > 0.05).

Immune function markers
There were no differences between treatments on Hp (P = 0.52) and LBP plasma concentration (P = 0.57), regardless of parity.There was an effect of week relative to calving on plasma concentration of Hp (P = 0.005) in both primiparous (P < 0.01, Figure 5A) and mul-tiparous cows (P < 0.01, Figure 5B).Albeit there were no effects of the week in relation to calving on LBP in multiparous cows (P = 0.52, Figure 5C), in primiparous cows, the plasma concentration of LBP decreased from week one to 2 postpartum independently of the treatment (P = 0.04, Figure 5D).

Vaginal discharge microbiome (VDM)
Vaginal discharge samples (n = 90) were analyzed from 45 cows enrolled in Control (n = 23) and RPC (n = 22) at calving and 7 d later.Three cows were not included in the analysis because their samples were not collected due to an occurrence of RFM.A total of 5,519 distinct microbial taxa were identified in the vaginal discharge samples, belonging to 15 different phyla or- ganized into 176 distinct genera.Supplementation of RPC did not affect the number of sequencings obtained per sample (P = 0.81, Figure 6A).The Chao1 (P = 0.03, Figure 6B), Shannon (P < 0.001, Figure 6C), and Simpson indexes (P = 0.03, Figure 6D) were lower at calving in the RPC treatment than in the Control, but similar on d 7. Phylum differences between treatments were observed (Figure 7).A similar distribution between treatments was observed at calving, with Firmicutes, Bacteroidota, and Proteobacteria as the most prevalent phyla (Figure 7).By contrast, at 7 DPP, differences emerged between the Control and RPC treatments (Figure 7) in the top 6 phyla identified (Firmicutes, Bacteroidota, Proteobacteria, Actinobacteriota, Verrucomicrobiota, and Fusobacteriota).Differences in VDM between cows fed RPC and Control cows were further confirmed independent of parity through NMDS analysis and PERMANOVA (P < 0.01, Figure 8A and B).However, only primiparous cows showed an interaction between treatment and day (P < 0.01, Figure 8B).Cows in the RPC treatment exhibited an altered VDM at the phylum and genus levels at 7 DPP, showing, for example, a lower relative abundance of Fusobacteriota (P = 0.03, Figure 9A) and Fusobacterium (P = 0.02, Figure 9B), that is the family and genus, respectively, of the common Fusobacterium necrophorum, a common pathogen associated with metritis development metritis (Jeon et al., 2015, Galvao, Bicalho andJeon, 2019).

DISCUSSION
Our study identified that RPC supplemented from 21 d pre-to 21 d postpartum had positive effects on lactation performance, health, and VDM structure.Choline is an essential nutrient for dairy cows, playing a crucial role in various physiological functions, such as lipid metabolism, milk production, and liver metabolism (Caprarulo et al., 2020).Albeit choline-rich feedstuffs have various levels of bioavailability, choline is rapidly degraded by the rumen microbiota (Sharma and Erdman, 1989) and needs to be supplemented as a stable and accessible source (Humer, Bruggeman and Zebeli, 2019).Indeed, inconsistent results on lactation performance and health (Lima et al., 2012, Zhou et al., 2016, Bollatti et al., 2020b) have been at least in part attributed to bioavailability in 2 recent meta-analysis (Humer, Bruggeman andZebeli, 2019, Arshad et al., 2020) Rumen-protect choline has been shown to improve milk performance by promoting better efficient nutrient utilization and energy balance in the cow (Zenobi et al., 2018a, Bollatti et al., 2020a, Mecionyte et al., 2022).Essentially, choline's contribution to phosphatidylcholine synthesis and its role in transporting very low-density lipoproteins from the liver (Xu et al., 2006) can improve liver function and fat metabolism, further benefiting the cow's nutrient utilization.A study assessing the benefits of RPC on lactation performance up to 105 DIM observed improvements in yields of milk, fat, lactose, true protein, ECM, and 3.5% FCM without increases in DMI or measures of lipid mobilization (Zenobi et al., 2018a, Bollatti et al., 2020a, Mecionyte et al., 2022).Similarly, albeit we did not observe changes in DMI and BCS by 21 DPP, multiparous and primiparous cows supplemented with RPC displayed increased yields in ECM, 3.5% FCM, and milk fat, which persisted until 150 DIM, suggesting RPC's longterm benefits in nutrient utilization and feed efficiency.
Although there is a scarcity of information in the scientific literature on the long-term positive impact of lactation performance, the well-characterized effects of RPC on liver function and energy balance can also be surmised as potential contributors to these responses observed in the mid-lactation (Piepenbrink and Overton, 2003, Zenobi et al., 2018a, Zenobi et al., 2018b).Studies also have suggested that RPC might benefit cows in late gestation when concentrations of choline metabolites are lowest and mammary cell accretion remains active (Zhou et al., 2016, Bollatti et al., 2020b).Our findings emphasize RPC's potential role in enhancing the mammary gland function needed to support lactation beyond the advantages of prepartum caloric intake (Zenobi et al., 2018a).
The effect of RPC on increased milk fat up 150 DIM in both multiparous (0.14 kg) and primiparous (0.06 kg) may be related to better lipid metabolism being redirected to the mammary gland (Chandler and White, 2017).A previous study shows that RPC increases mammary transcription of fatty acid desaturase 1 (FADS1) and longevity assurance gene 2 (LASS2), which play a role in the synthesis of unsaturated fatty acids and milk fat globule membranes, both essential for milk fat synthesis (Lashkari et al., 2020).Albeit specific functions of LASS2 in the context of longevity and aging are not fully understood, alterations in ceramide metabolism and signaling may contribute to the aging process and age-related diseases (Wegner et al., 2016).
The response to RPC supplementation may also vary between parity due to the physiological differences and nutrient requirements.Multiparous and primiparous cows take advantage of choline through different mechanisms.Multiparous cows often have better-developed mammary glands, a more established metabolic adaptation to the lactation (Oikonomou et al., 2008), and generally produce more milk.However, they also have a greater negative energy balance (Goselink et al., 2013).These factors lead to a greater nutrient demand  (Ingvartsen and Moyes, 2013).The additional choline provided by RPC supplementation suggests that multiparous cows fulfill these higher requirements more efficiently, probably by improving lipid metabolism during the transition period, resulting in an increased 3.5% FCM and milk fat yield.On the other hand, primiparous cows still have the requirement of growth postpartum (Coffey, Hickey and Brotherstone, 2006) (Kim et al., 2017).Chao 1, Shannon, and Simpson indexes showed lower richness and diversity in the RPC at calving than the Control.Data represent the mean ± SD.Data are statistically different when P < 0.05.and might utilize the RPC supplementation for nutrient partitioning, tissue acquisition (Moretti et al., 2020) and support the S-adenosylmethionine (SAM) cycles (Glier, Green and Devlin, 2014), thereby contributing for greater 3.5% FCM and milk fat yield observed in our study.Nevertheless, research on RPC supplementation in primiparous dairy cows is less extensive compared with multiparous cows (Zenobi et al., 2018a).Further research may be required to fully understand the extent of these benefits and the underlying mechanisms for primiparous cows.
Any changes in plasma glucose and NEFA levels during the transition period reinforce the greater lactation performance in cows fed RPC, differing from other studies (Zhou et al., 2016).The improved efficiency could be due to better liver function might lead to better overall metabolic function and nutrient utilization (Zom et al., 2011, Goselink et al., 2013).A recent study that investigated the effects of supplementing RPC between 14 d before and 21 d after parturition on liver function in healthy periparturient dairy cows supports these findings (Du et al., 2023).This study, based on liver transcriptome analysis at 21 d postpartum, indicated that RPC upregulated the expression of genes related to glucose and lipid metabolism (FGF21, SLCO1B3, SLC13A5, FBP2, and CYP26A1), reducing the body energy consumption (Du et al., 2023).Aligning with the higher insulin levels observed in cows supplemented with RPC (Zhou et al., 2016), our findings suggest that RPC may help to maintain proper  glucose levels during the transition period needed to support the regulation of glucose/glucogen synthesis of carbohydrate precursos by FBP2 (Bakshi et al., 2018) and insulin sensitivity through SLCO1B3/OATP1B3 (Meyer Zu Schwabedissen et al., 2014, Ren et al., 2017).
During an inflammatory event, elevated levels of various inflammatory markers such as Hp and LBP have been used to evaluate the presence and severity of systemic inflammation (Kvidera et al., 2017a, Dickson et al., 2019, Koch et al., 2019).Haptoglobin is an acute-phase protein produced by the liver that binds to free hemoglobin, preventing tissue damage and oxidative stress, while LBP is a soluble acute-phase protein that binds to lipopolysaccharides, playing a critical role in the innate immune response to bacterial infections (Ceciliani et al., 2012).Haptoglobin quickly increases and has been used in cattle as an acute phase protein and immune function marker (Zhou et al., 2016, Bollatti et al., 2020b).However, the impact of RPC on plasma LPB concentration as a maker of systemic inflammation has not been explored.In our study, neither primiparous nor multiparous exhibited the RPC effect in Hp or LBP.The adequate nutritional management and appropriate choline levels might also minimize metabolic stress and its related support to proper gut integrity (Kvidera et al., 2017a, Dickson et al., 2019, Koch et al., 2019).Indeed, direct effects of choline on gut integrity have been reported in monogastric species (Takahashi, Mizunuma and Kishino, 1982) and deserve further exploration in ruminants.
Most of the studies evaluating the effect of RPC during the transition period were unable to identify a reduction in the risk for metritis and mastitis but indicated benefits for ketosis (Ardalan, Rezayazdi and Dehghan-Banadaky, 2010, Lima et al., 2012, Bollatti et al., 2020b).Similarly, our study did not observe effects on mastitis and metritis between treatments.However, despite no changes in postpartum BHB levels, RPC tended to reduce the incidence of subclinical ketosis.According to (Du et al., 2023), dairy cows fed RPC during the transition period showed an increase in fibroblast growth factor 21 (FGF21) gene expression, which plays a regulatory role in ketogenesis and glucose metabolism (Degirolamo, Sabba and Moschetta, 2016).Considering the early lactation stage's challenges, such as milk production increases with energy intake lower than required combined with fat mobilization, RPC supplementation could be potential pathways may be through FGF21 production to support lipid metabolism to increase ECM and 3.5% FCM.RPC supplementation may act through FGF21 production to bolster lipid metabolism, leading to elevated ECM and 3.5% FCM levels.
One of the central findings of our study revolves the effects of RPC supplementation on VDM in promoting the changes in bacteria profile as a proxy for uterine health.Previous research indicates that choline supplementation may have an impact on the composition and functionality of the gut microbiome (Zimmerman, Gyawali and Ibrahim, 2017, Consoli et al., 2022, Ferreri et al., 2022).Similarly, the current study echoes previous findings but introduces an additional dimension.Reduced richness and diversity at calving, as indicated by the decreased Chao 1, Shannon, and Simpson indexes, suggest that the VDM environment in cows supplemented with RPC is more homogeneous within cows of this treatment.Accounting that Simpson index considers both microbial evenness and richness (Kim et al., 2017), less within-treatment variation and more dominance of some microbes could indicate a mechanism by which RPC could promote a better reproductive tract microbiome and potentially uterine health of dairy cows during this critical time when cows are more susceptible to infections.
Indeed, we could also observe a between-treatment variation (β-diversity) effect caused by the supplementation of RPC in the VDM of cows.The NMDS analysis at the phylum level in multiparous and primiparous show a reduced relative abundance of Fusobacterium at 7 d postpartum, one important pathogen associated with the development of metritis (Jeon et al., 2015, Galvao, Bicalho andJeon, 2019).These findings suggest that RPC could support the eubiosis of the VDM to create an environment less favorable for the growth of Fusobacterium, which in turn could contribute to reducing the odds of metritis development, but further research is needed to explore this theory.
In addition to these observations, recent studies have thrown light on the antimicrobial capabilities of micelles of quaternary ammonium based on choline.Specifically, these choline-based micelles had activity against Gramnegative bacteria, including antibiotic-resistant strains, impacting bacterial biofilm and motility of E. coli and P. aeruginosa (Consoli et al., 2022, Ferreri et al., 2022).According to previous studies (Ly et al., 2004, Malek et al., 2011), quaternary ammonium and hydroxyl treatments of choline can disrupt bacterial cell membranes by establishing electrostatic and hydrogen bond interactions to specific transporters found on the surface of Gram-negative bacteria.Integrating these findings with our research, it is reasonable to surmise that RPC supplementation could act as a substrate for the development of immune active compounds and potential aid for uterine health.However, it is important to underscore that these results, while encouraging, serve as a foundation.Further studies are needed to ascertain the RPC's role in uterine health.
Marques et al.: RUMEN-PROTECTED CHOLINE IN DAIRY COWS

Figure 1 .
Figure 1.After one week of adaptation period in the Calan gates, cows were enrolled at −21d before the due date, blocked by parity treatment prepartum (primiparous vs. multiparous), and assigned randomly to the Control or RPC (9 g/d of choline chloride from 21 d prepartum and 18 g/d of choline chloride from calving to 21 DPP) treatments when dry matter intake (DMI) was measured.Body weight (BW), body condition score (BCS), and blood samples were evaluated at enrollment and weekly from calving to 21 postpartum.Vaginal discharge (VD) was collected at calving and 7 d postpartum.Milk yield was recorded daily, and milk composition monthly from 0 to 150 DIM.
Marques et al.: RUMEN-PROTECTED CHOLINE IN DAIRY COWS Marques et al.: RUMEN-PROTECTED CHOLINE IN DAIRY COWS Marques et al.: RUMEN-PROTECTED CHOLINE IN DAIRY COWS Marques et al.: RUMEN-PROTECTED CHOLINE IN DAIRY COWS in the last 21 d of gestation.Postpartum = measurements in the first 21 d postpartum. 2 Control = 0 g/d choline ion from 21 d prepartum to 21 d postpartum.3 RPC = 9 g/d choline chloride as rumen-protected choline chloride (CholiGEM TM , 60% choline chloride; Kemin Industries, Cavriago RE, Italy) from 21 d prepartum and 18g/d from calving to 21 d postpartum.

Figure 2 .
Figure 2. Body weight (BW) and body condition score (BCS) of multiparous (A, C) and primiparous (B, D) in Control and RPC (9 g/d of choline chloride from 21 d prepartum and 18 g/d of choline chloride from calving to 21 d postpartum) treatments.

Figure 3 .
Figure 3. Plasma concentration of glucose and NEFA in multiparous (A, C) and primiparous (B, D) in Control (n = 24), and RPC (n = 24, 9 g/d of choline chloride from 21 d prepartum and 18 g/d of choline chloride from calving to 21 d postpartum) treatments.

Figure 4 .
Figure 4. Plasma concentration of BHB in multiparous (A) and primiparous (B) in Control (n = 24), and RPC (n = 24, 9 g/d of choline chloride from 21 d prepartum and 18 g/d of choline chloride from calving to 21 d postpartum) treatments.

Figure 5 .
Figure 5. Plasma concentration of haptoglobin and LBP in multiparous (A, C) and primiparous (B, D) in Control (n = 24), and RPC (n = 24, 9 g/d of choline chloride 21 d from prepartum and 18 g/d of choline chloride from calving to 21 d postpartum) treatments.

Figure 6 .
Figure 6.Diversity of uterine microbiota collected on calving (d 0) and 7 d later (d 7) from cows enrolled in Control (n = 23), and RPC (n = 22, 9 g/d of choline chloride from 21 d prepartum and 18 g/d of choline chloride from calving to 21 d postpartum) treatments.(A) The number of reads.(B) Chao1 index estimates the total number of microbial in a sample based on the number of singleton and doubleton taxa.(C, D) Shannon and Simpson indexes consider the microbial richness and evenness.Shannon index puts greater weight microbial richness, while the Simpson diversity index puts greater weight on microbial evenness(Kim et al., 2017).Chao 1, Shannon, and Simpson indexes showed lower richness and diversity in the RPC at calving than the Control.Data represent the mean ± SD.Data are statistically different when P < 0.05.

Figure 7 .
Figure 7. Descriptive analysis of the vaginal discharge microbiome composition at Phylum at calving (d 0) and 7 d later (d 7) from cows enrolled in Control (n = 23), and RPC (n = 22, 9 g/d of choline chloride from 21 d prepartum and 18 g/d of choline chloride from calving to 21 d postpartum) treatments.Data of 16S rRNA sequencing show differences at Phylum on d 7 between treatments, suggesting RPC may modify the vaginal discharge microbiome and maybe lead to uterus healthy.
Figure 8. Non-metric multidimensional scaling (NMDS) and permutational multivariate ANOVA (PERMANOVA) of the vaginal discharge microbiome in multiparous (A) and primiparous (B) enrolled in Control (n = 23) and RPC (n = 22, 9 g/d of choline chloride from 21 d prepartum and 18 g/d of choline chloride from calving to 21 d postpartum) treatments.Significance was declared at P ≤ 0.05.

Figure 9 .
Figure 9. Differential abundance analyses from the vaginal discharge microbiome collected on calving (d 0) and 7 d later (d 7).Samples were from cows enrolled in Control (n = 23), and RPC (n = 22, 9 g/d of choline chloride from 21 d prepartum and 18 g/d of choline chloride from calving to 21 d postpartum).In panel A, the mean relative abundance of Fusobacteriota (A) at calving 7 d post-calving is reported, while in panel B the mean relative Fusobacterium (B) at calving 7 d post-calving is reported.
Marques et al.: RUMEN-PROTECTED CHOLINE IN DAIRY COWS

Table 2 .
Marques et al.: RUMEN-PROTECTED CHOLINE IN DAIRY COWSPredicted productive responses and energy status (average ± SEM) to supplemental rumen-protect choline (RPC) in the transition period