structure and function of preweaning dairy calves fed a whole milk powder or a milk replacer high in fat

The composition of milk replacer (MR) for calves greatly differs from that of bovine whole milk, which may affect gastrointestinal development of young calves. In this light, the objective of the current study was to compare gastrointestinal tract structure and function in response to feeding liquid diets having a same mac-ronutrient profile (e.g., fat, lactose, protein) in calves in the first month of life. Eighteen male Holstein calves (46.6 ± 5.12 kg; 1.4 ± 0.50 d of age at arrival; mean ± standard deviation) were housed individually. Upon arrival, calves were blocked based on age and arrival day, and, within a block, calves were randomly assigned to either a whole milk powder (WP; 26% fat, DM basis, n = 9) or a MR high in fat (25% fat, n = 9) fed 3.0 L 3 times daily (9 L total per day) at 135 g/L through teat buckets. On d 21, gut permeability was assessed with indigestible permeability markers [chromium (Cr)-EDTA, lactulose, and d-mannitol]. On d 32 after arrival, calves were slaughtered. The weight of the total forestomach without contents was greater in WP-fed calves. Furthermore, duodenum and ileum weights were similar between treatment groups, but jejunum and total small intestine weights were greater in WP-fed calves. The surface area of the duodenum and ileum did not differ between treatment groups, but the surface area of the proximal jejunum was greater in calves fed WP. Urinary lactulose and Cr-EDTA recoveries were greater in calves fed WP in the first 6 h post marker administration. Tight junction protein gene expression in the proximal jejunum or ileum did not differ between treatments. The free fatty acid and phospholipid fatty acid profiles in the proximal jejunum and ileum differed between treatments and generally reflected the fatty acid profile of each liquid diet. Feeding WP or MR altered gut permeability


INTRODUCTION
Approximately half of dairy calves in the United States are fed exclusively milk replacer (MR) or a combination of MR and whole milk (WM) in the preweaning phase (Urie et al., 2018).Typically, MR contains less fat (16-20% vs. ~30% DM basis) and more lactose (40-50% vs. ~35%) than bovine WM, whereas protein inclusion is rather comparable (Gilbert et al., 2016;Echeverry-Munera et al., 2021;Wilms et al., 2022).Formulating MR with a higher fat content (≥23%) reduced the number of therapeutic interventions (Berends et al., 2020), resulted in lower fecal scores (Amado et al., 2019), and increased intestinal and total gastrointestinal tract (GIT) weights (Welboren et al., 2021) compared with calves fed MR with low fat.Differences in macronutrient inclusion, as well as differences in minerals, vitamins, and bioactive components between MR and WM may affect liquid feed digestibility and calf physiology.
Dietary lipids and proteins from WM are a source energy, but also fatty acids and AA providing important structural and metabolic functions to newborn animals (Delplanque et al., 2015;Grote et al., 2016;Cao et al., 2019).In contrast, MR have a substantially different composition based on the type of raw materials used, which are from either dairy, vegetable, or animal origin (Branco Pardal et al., 1995;Hill et al., 2007Hill et al., , 2011;;Esselburn et al., 2013).When including proteins from dairy origin, the AA profile remains very similar to that of WM, however, the inclusion level of skim milk determines the casein-to-whey ratio that drives gastric emptying dynamics and postprandial nutrient signaling.When considering the fat fraction, MR have a different fatty acid, triglyceride structure, and globule size and structure than WM, which may negatively affect the absorption and digestion of dietary fat (Lien, 1994;Agostoni et al., 1995).Several studies have shown that supplementing selected fatty acids in MR to better mimic the fatty acid profile of milk fat positively affected development and health in calves (Górka et al., 2011a,b;Hill et al., 2011).For instance, supplementation of MR with butyric acid, medium-chain fatty acids, and α-linolenic acid (ALA) resulted in enhanced growth, improved feed efficiency, and decreased the number of days with diarrhea (Hill et al., 2007(Hill et al., , 2011;;Garcia et al., 2015).Supplementation of butyric acid in MR also increased rumen papillae length and width (Górka et al., 2009(Górka et al., , 2011a,b),b).
Meanwhile, little is known about how feeding WM versus MR affects the development of the GIT and intestinal permeability.In intestinal tissues, changes in permeability could be related to changes in mRNA expression of tight junction genes (Suzuki and Hara, 2010), although this does not necessarily represent these molecules at the protein level.The fatty acid profile of the liquid feed has also been associated with changes in intestinal permeability and tissue composition in multiple species (Spector and Yorek, 1985;Jenkins and Kramer, 1990;Ibarguren et al., 2014).and omega-6 (n-6) fatty acids are thought to alter intestinal permeability due to their roles as anti-and pro-inflammatory precursors, respectively (Usami et al., 2001).Thus, it is plausible that composition differences between WM and MR mediate changes in gut permeability by facilitating changes in gene expression and tissue composition of GIT, which in turn alters intestinal permeability in calves.
In this light, this study hypothesized that feeding a WM powder would alter the composition of GIT, promote the development of GIT, and decrease intestinal permeability in calves, as compared with a MR high in fat.Therefore, the objective of this study was to evaluate how 2 liquid diets with similar macronutrient profile in terms of fat, protein, and lactose would affect GIT structure and function in the first month of life in calves.

MATERIALS AND METHODS
This study was conducted between March and June 2019 at the Calf Research Facility of Trouw Nutrition Research and Development (Sint Anthonis, the Netherlands).All procedures described in this article comply with the Dutch Law on Experimental Animals and Directive 2010/63/EC and were accordingly approved by the Animal Care and Use Committee of Utrecht University (CCD no.AVD2040020173425).

Animals and Experimental Design
This complete randomized block design study is part of a larger experiment including 68 male Holstein-Friesian calves (n = 17 calves per treatment groups) investigating the effects of macronutrient profile in MR and WM on postprandial responses, digestibility, and gut permeability.A subset of 18 calves (46.6 ± 1.21 kg BW; 1.4 ± 0.50 d of age at arrival; mean ± SD) collected from 7 dairy farms within 14 km of the research facility were used to measure GIT development and composition.On the farm of origin, a standardized protocol for colostrum management was followed during the first 24 h after birth.A first meal of 3.0 L was offered within the first 3 h after birth, followed by 2 feedings of 2.0 L within 24 h after birth.Colostrum was analyzed using a portable multi-test analyzer (DVM Rapid Test II, Vetlab) and required a value of 22% Brix or greater, indicating an immunoglobin content of 50 mg/L or greater.Successful completion of this protocol was assessed by determining IgG concentrations in the blood upon arrival at the research facility and within 48 to 72 h after birth using a portable multi-test analyzer (DVM Rapid Test II, Vetlab).After completion of the colostrum protocol, calves received 2.5 L of a commercial MR at a concentration of 135 g/L (Sprayfo Excellent, Trouw Nutrition) twice daily until arrival at the research facility.Upon arrival, calves were assigned to blocks based on arrival day and birth to minimize age differences within a same block.The larger experiment included 4 experimental liquid diets that were randomly assigned within blocks by a person that was not involved in scoring and sampling of calves.Nine calves out of a total of 17 per treatment group and belonging to the same blocks were then used for tissue sampling.Dietary treatments included either a WM powder (WP, 26% fat, 24.5% protein, 38% lactose, DM basis, 21.6 MJ/kg) or MR high in fat (25% fat, 22.5% protein, 38.6% lactose, 21.3 MJ/kg), offered 3 times daily at 0600, 1200, and 1800 h in 3.0 L meals (135 g/L; 13.5% solids) via teat buckets until d 32 after arrival.Treatments were blinded to animal caretakers by randomly assigning a letter (A or B) to each treatment.Ingredients and nutrient composition of the dietary treatments of the WP (WMP 26%, Arla Foods amba) and MR (Sloten B.V., Trouw Nutrition) treatments are presented in Wilms et al. (2022).In brief, the ingredients of the MR treatment included 50.0%(% product basis) skim milk powder, 25.0% of a blend of palm and coconut oils (2:1), 12.4% of whey powder, 6.8% of whey protein concentrate, 2% of hydrolyzed wheat protein, 2.0% of maltodextrin, and 1.8% of premix.The premix contained vitamins, minerals, probiotics and prebiotics, stabilizers, and AA, as well as flavoring and stability agents.No solid feed was offered to control nutrient intake, although calves may have consumed some bedding material (flax straw) as indicated by its presence in the forestomach observed postmortem.Water was available ad libitum in plain buckets throughout the experiment.Health status was visually monitored daily by caretakers and a standard veterinary protocol was followed in the event of illness.In case of diarrhea coupled with either dehydration or anorexia, calves received oral rehydration solutions either once or twice daily in between milk meals depending on the diarrhea severity for a maximum of 3 d.In case the calf would not recover within 2 d, the animal would receive Dofatrim i.m. (Dopharma Research B.V.; 1 mL/15 kg BW) for 3 d and Novem 20 s.c.(Boehringer Ingelheim Vetmedica GmbH; 1 mL/40 kg BW) if fever was also present.For respiratory diseases, calves would receive Florkem i.m. (Ceva Animal Health; 1 mL/15 kg BW) for 2 d and Novem 20 s.c.(1 mL/40 kg) for 1 d.In case of prolonged health disorders, a veterinarian would be consulted to provide appropriate medical care.

Housing
Calves were housed indoors in individual pens (1.22 m × 2.13 m) separated by galvanized bar fences and equipped with rubber-slatted floor in the front area (50% of the total pen area) and a laying area in the back, which contained a mattress covered with flax straw.During total urine collection, calves were tethered to the front of the pen and an elevated plateau covered with rubber was placed at the front of the pen to elevate the animals and facilitate urine collection.The temperature in the calf pen was maintained at a minimum of 12°C and a maximum of 28°C.Relative humidity was maintained between 60 and 85%.Calves were exposed to daylight and artificial light from 0530 to 2130 h.

Measurements
Milk and water intakes were recorded daily by weighing the refusals throughout the study period.Fecal scoring was performed over the first 21 d after arrival through visual assessment of photos of feces taken daily after the morning meal.The scoring was performed by only one examiner using a 3-level scoring system as follows: normal feces (score 0), wet feces (score 1), watery feces (score 2).Body weight was determined on the day of arrival and weekly thereafter by weighing the calves with a custom scale (W2000; Welvaarts Weegsystemen).Intestinal permeability was determined by measuring the recovery of indigestible markers in urine.The volume of each marker solution was adjusted according to each calf's BW to obtain an individual dose of chro-mium (Cr)-EDTA (0.1 g/kg of BW; Masterlab), lactulose (0.2 g/kg of BW; Sigma-Aldrich), and d-mannitol (0.12 g/kg of BW; Sigma-Aldrich).Due to their large molecular weight, lactulose (342 g/mol) and Cr-EDTA (344 g/mol) are considered as markers for paracellular permeability, whereas d-mannitol (182 g/mol) is considered a marker of transcellular permeability although it can also be absorbed paracellularly.The marker solutions (Cr-EDTA solution and lactulose-d-mannitol solution) were orally pulse dosed using 100-mL syringes (BD Plastipak, Merkala) to calves at 0600 h on d 21 after arrival instead of the morning milk meal.After marker administration, total quantitative 6-and 24-h urine collection was performed using urine collection bags secured with medical glue, as described in Wilms et al. (2020), to assess marker recovery in urine.In addition, a spot urine sample was taken using plastic bags on d 20 (one day before quantitative urine collection) as a background measurement to ensure absence of these molecules in urine.Samples were transported in boxes with cooling aids and stored at −18°C.

Chemical Analysis of Feed and Urine Samples
Milk replacer and WP samples were analyzed for DM, crude ash, crude fat, CP, macrominerals, and carbohydrates by MasterLab (Boxmeer, the Netherlands) as described in Wilms et al. (2022).Quantification of peroxide value, lipid oxidation, AA, and fatty acid profile of treatments was performed by Mérieux Nutri-Sciences (Chicago, IL).Peroxide value was determined by titrating iodine liberated from potassium iodide with sodium thiosulphate solution.To determine lipid peroxidation, a thiobarbituric acid test was performed according to Kirk and Sawyer (1991).Fatty acid profile of treatments was determined by GC using an internal standard method (Table 1).The AA profiles of WP and MR presented in Table 2 were analyzed by ultraperformance liquid chromatography according to Camara et al. (2020).
Measurement of Cr in urine samples was processed at the University of Nottingham (Nottingham, UK).Urinary Cr was determined using inductively coupled plasma mass spectrometry (ICP-MS, Thermo Fisher XSeriesII, Thermo Fisher Scientific).Measurements of lactulose and d-mannitol in urine were processed at MasterLab (Boxmeer, the Netherlands).For the extraction of d-mannitol and lactulose, the urine samples were diluted with d-mannitol and labeled Lactulose (13-c) as internal standards.The solution extracts were then analyzed using a Phenomenex Luna 3-µm SUGAR 100A 150 × 4.6 mm column (Phenomenex) on a Thermo TSQ Quantis liquid chromatography tandem mass spectrometer (heated electrospray ionization source) with a Vanquish pump, oven, and autosampler (ThermoFisher Scientific).A mixture of 80% acetonitrile and 20% water with 1 mM formate was used as the elution buffer.

Post-Mortem Analysis
Calves were euthanized on d 32 after arrival (33.0 ± 0.43 d of age, mean ± SD) and 3 h after the morning meal by injection of 6 mL of tetracaine (T61; MSD Animal Health Nederland) into the jugular vein.After the calves reached a surgical level of sedation, exsanguination was performed.The esophagus and rectum were tied off with zip ties to prevent loss of contents when the GIT was removed and placed on a surgical table.Each segment of GIT was weighed empty, and the length measured as described in Pyo et al. (2020).Samples of the small intestine were collected as follows: duodenum samples were taken 7.5 cm distal to the pyloric sphincter, proximal jejunum samples were taken 100 cm distal to the duodenum sampling site, and distal jejunum samples were taken 30 cm proximal to the collateral branch of the cranial mesenteric artery.Ileum samples were collected 30 cm proximal to the ileo-cecal junction, and colon samples were collected 30 cm distal to the ileo-cecal junction.The tissue samples were cut longitudinally, washed with phosphate-buffered saline, and cross sections were fixed in 10% buffered formalin solution (4% formaldehyde; Fisher Scientific) or placed in RNALater (Invitrogen).Samples for RNA were incubated at room temperature for 24 h and then stored at −80°C until analysis was performed.

Total Lipid Analysis of Small Intestine Tissue
Snap frozen samples of jejunum tissue were ground under liquid nitrogen with a mortar and pestle and weighed 0.05 g for analysis.Samples were placed in 1 mL of 0.1 M potassium chloride in a glass tube.A homogenizer (Bio-Gen Pro200; ProScientific) was used 3 times at the highest speed for 3 s to grind the sample matrix, which was then added to another glass tube containing 50 µL of 17:0 free fatty acid standard (1 mg/mL).A 2 mL 2:1 chloroform-methanol mixture was added into the original tube and then transferred to the second tube.This process was repeated twice to ensure complete transfer of the sample between tubes.The tubes were vortexed for 5 to 10 s or until mixed, and then the tubes were flushed with nitrogen gas for 5 to 10 s.The samples were left overnight at 4°C.Samples were centrifuged at 465 × g for 10 min at 21°C.Using a 22.9-cm Pasteur pipette, the lower chloroform layer was transferred to a leak-proof 15-mL tube, which was washed with Miele ProCare Lab 10 AP (Miele).The tube was completely dried under a gentle stream of nitrogen.The drying apparatus was cleaned with hexane before and after use.Two milliliters of 0.5 M potassium hydroxide in methanol was added to each tube.Both tubes were saponified in an oven at 100°C for 1 h.The tubes were checked every 10 min to ensure that no solvent evaporated and that the amount of solvent remained equal across all tubes.The tubes were cooled in a fume hood at room temperature for 10 min.To stop methylation, 2 mL of double-distilled water was added to each tube and vortexed.The tubes were then centrifuged at 465 × g for 10 min at 21°C to separate the phases.Using a 14.6-cm Pasteur pipette, the top hexane layer was extracted into a clean GC vial and dried with nitrogen.For GC analysis, samples were reconstituted in 250 µL of hexane and placed in a glass insert.The insert was returned to the GC vial, tightly capped, and run with 30 µL of MA6 as standard.

Phospholipid Analysis of Small Intestine Tissue
Lipids including phospholipids (PL) were extracted from the samples by the method of Bligh and Dyer (1959) in the presence of the internal C17: 0 -phosphatidylcholine (PC) standard.Sample processing was performed as previously described (Bligh and Dyer, 1959;Holub et al., 2011).Sample extracts were dried under nitrogen after addition of butylated hydroxytoluene.The extracts were resolubilized in 200 µL of dichloromethane.Separation of the extracts was performed by solid phase extraction by spotting 50 µL of the lipid extracts containing internal standards onto a silica gel 60 thin layer chromatography plate (Merck 5721-7).Fifty microliters of standard solution was spotted on a thin-layer chromatography plate, which was developed in mobile phase heptane: isopropyl ether: acetic acid 60:40:3 vol/vol/v (Holub et al., 2011).After the solvent had moved 80% up the plate, the plate was removed from the thin-layer chromatography tank and dried in a fume hood for 5 min.The plate was sprayed with 8-anilino-1-napthalene sulfonic acid in methanol to identify the bands of interest based on retention factor by comparison with appropriate standards using long wave UV.A single-edged razor was used to scrape off bands of interest into screw top 16 × 125 mm glass test tubes.Fatty acid methyl esters were prepared with boron trichloride in methanol, and the methylation tubes were heated at 95°C for 50 min on a hot plate.The fatty and methyl esters were then analyzed using an Agilent 7890B gas-liquid chromatograph with a 60-m DB −23 capillary column (0.32-mm inner diameter; Morrison and Smith, 1964).

Small Intestine Histomorphology
Samples of duodenum, proximal jejunum, and ileum were dehydrated overnight by incubation in ethanol.Once samples were successfully dehydrated, they were embedded in paraffin wax, sectioned (5 µm) with a rotary microtome, and stained with hematoxylin and eosin (Gezondheidsdienst voor Dieren).Slides were viewed with a Leica ICC50W microscope at 40× magnification connected to the Leica Airlab app (Leica Microsystems) for imaging.ImageJ software (ImageJ 1.46r, National Institute of Health) was used to determine villus height and width, crypt depth and width, and muscularis layer thickness.The villus height was measured from the fully attached epithelial cell to the villus-crypt junction and the villus width was measured perpendicularly at mid-villus height.Crypt depth was measured from the villus-crypt junction to the bottom of the crypt, and crypt width was measured at the opening between the villi at the villus-crypt interface (Wongdee et al., 2016).Muscularis layer thickness was measured from the apical side below the crypt to the serosal side.Two to 3 measurements were taken from 3 to 5 images per slide (10 measurements/small intestinal region).All measurements were averaged per sample and analyzed as one observation per dependent variable.

Ki-67 Immunohistochemistry of Small Intestine Tissues
Paraffin-embedded duodenum, proximal jejunum, and ileum tissues were also analyzed for expression of Ki-67 antigen, a marker of proliferation (Hall et al., 1990), by immunohistochemistry (University of Guelph, Guelph, Canada).An automated Leica microtome was used to section the tissue.Slides were then deparaffinized, rehydrated, and treated with 3% hydrogen peroxide to quench endogenous peroxidase activity.The slides were then placed in the Dako PT Link (Dako Auostainer, Agilent Technologies) for heat-induced epitope-retrieval at high pH, cooled, and placed on the staining instrument.A universal nonserum blocker was applied for 10 min.Sections were then incubated with a mouse antihuman Ki-67 monoclonal antibody (dilution 1:50; clone MIB −1, Agilent Technologies) followed by an antimouse/antirabbit polymer (EnVision Flex HRP, Agilent Technologies) as detection and Nova Red Chromogen (Vector Laboratories).Duplicate sections from one calf of each of the 2 treatment groups were subjected to the same immunohistochemistry procedure as negative reagent controls, with substitution of antibody diluent alone for the primary antibody.Images were captured using a Leica ICC50W microscope at 100× magnification connected to the Leica Airlab app (Leica Microsystems).Five images were taken per tissue section per calf.The fraction of cells stained positive for Ki-67 antigen to total cells was quantified using bioanalysis software QuPath (Bankhead et al., 2017).The mucosal area of the crypts of each image was selected, and cells stained positive for Ki-67 antigen within the selected area were detected using positive cell detection.A subset of 2 animals and all 3 tissues were used to optimize the background radius (16 pixels), sigma (1.8 pixels), and threshold (1.35) settings to allow accurate detection of positive and negative cells with the right color intensity.For duodenum, proximal jejunum, and ileum, CV was 8.9, 7.1, and 14.8%, respectively.

Quantitative Real-Time Polymerase Chain Reaction
Frozen tissues from the proximal jejunum and ileum were grounded with liquid nitrogen and a mortar and pestle.The RNA from tissues was extracted using the Pure Link RNA Mini Kit (Invitrogen) according to the manufacturer's protocol.Quantity and integrity number of RNA were measured by UV-visible spectroscopy (Nanodrop One Microvolume, Thermo Fisher Scientific) and electrophoresis (TapeStation, Agilent Technologies).The average RNA integrity number was 6.92 and 7.11 for the proximal jejunum and ileum samples, respectively, and the minimum threshold was 5.6.Then, 1 µg of RNA was used to produce cDNA (High-Capacity cDNA Reverse Transcription Kit, Applied Biosystems), which was analyzed for the expression of genes encoding several tight junction proteins and cytokines by performing quantitative real-time PCR (qRT-PCR).The primer pair sequences used are listed in Table 3.The geometric mean of all 3 reference genes (β-actin, RPL19, GAPDH) was used as reference in both proximal jejunum and ileum.NormFinder identified these 3 genes to be the top 3 that were stable across block and treatment (Andersen et al., 2004).Primer efficiencies were tested using 10× serial diluted pool sample and were determined using a standard curve and ranged from 92 to 110%.The qRT-PCR was performed using a StepOnePlus Real-Time PCR System (Applied Biosystems) from the Genomics Facility at the University of Guelph (Guelph, Canada).To each well, 5 µL of cDNA, 10 µL of DNA polymerase-containing supermix (SsoAdvanced Universal Inhibitor-Tolerant SYBR Green, Bio-Rad), 0.8 µL of 5 µM forward and reverse primer mix, and 4.2 µL of nuclease-free H 2 O were added.Polymerase activation was 3 min at 98°C, followed by 40 cycles of 2-step qPCR (10 s at 98°C for denaturation and 30 s at 60°C combined annealing/ extension).A melting curve was then generated at 95°C for 15 s, 60°C for 1 min, and 95°C for 15 s to confirm the specificity of the PCR amplicon.Expression of the target gene relative to the target gene was calculated according to Pfaffl (2001), using the average of the MR group as a control.

Calculation and Statistical Analysis
Based on the outcome of Pyo et al. (2020), investigating villous growth in calves, a standard deviation of 20.8 µm was assumed for villus length.It was considered that the minimal meaningful difference would be 90 µm.Therefore, for a power (1-β) of 80%, and a significance level of 0.05, the smallest meaningful sample size to detect relevant differences was 8 calves per treatment group.Thus, 9 calves per treatment were enrolled in the current study.Villi height and crypt depth were used to calculate the villi to crypt ratio.Mucosal surface area index (SA) was calculated (as shown in Equation 1) from the ratio of mucosal toserosal amplification, as described in Kisielinski et al. (2002), where M = Mucosal SA index; a = villus width (µm); b = villus height (µm); and c = crypt width (µm).
Continuous variables (e.g., blood IgG, growth, gut weights and lengths) were analyzed in a mixed model using the PROC MIXED procedure in SAS (version 9.4M6; SAS Institute, 2018).Pearson correlation values were analyzed using PROC CORR (data not shown).
All the data and individual histology measurements were screened for normality of distribution using the UNIVARIATE procedure of SAS (the Shapiro-Wilk test) before analysis.The experimental unit was the calf, and the model included the fixed effect of treatment and block as random.In the case of BW and ADG, the arrival BW was included as a baseline covariate.All values reported are presented as least squares means with the standard error of the means.Significance was declared at P ≤ 0.05, and the trend threshold was set at 0.05 < P ≤ 0.10.
Mellors et al.: GASTROINTESTINAL STRUCTURE AND FUNCTION composition and fatty acid composition of the phospholipid fraction in 2 intestinal regions (jejunum and ileum) of calves fed WP or HF were performed using the web-based metabolomics data processing tool MetaboAnalyst 5.0 (Pang et al., 2021; see http: / / www .metaboanalyst.cafor detailed methodology).Data were transformed using the generalized log-transformation and then Pareto scaled (mean divided by the square root of the standard deviation of each variable) to correct for heteroscedasticity, reduce skewness, and mask effects.Principal component analysis (PCA) and volcano plots were applied to obtain an overview of tissue fatty acid composition and phospholipid fraction according to treatment (WP and MR) in 2 intestinal regions.Based on the FC threshold of 2 on the x-axis and a false discovery rate (adjusted P-value) threshold of 0.05 on the y-axis, volcano plots were constructed to identify the significant fatty acids between treatments in 2 intestinal regions of calves.

General Parameters and Gut Permeability
Serum IgG concentrations measured at arrival did not differ between treatments and was 21.4 ± 0.26 g/L (LSM ± SEM) for WP and 19.3 ± 0.23 g/L for MR.The dietary treatments did not affect ADG, which was 663 ± 70.0 g/d (LSM ± SEM) for WP and 633.6 ± 56.7 g/d for MR over the first 21 d (P = 0.60).For MR intakes, data were not processed because of infinite likelihood as most calves consumed their entire milk allowance.The proportion of calves that experienced diarrhea (defined as a score of 2) at least once during the first 21 d was similar between WP (7 calves) and MR (6 calves).In addition, diarrhea duration defined as the number of days in which a calf had a fecal score of 2 over the first 21 d did not differ across groups and was 4.7 ± 0.84 d (LSM ± SEM) for WP and 4.1 ± 1.49 d (LSM ± SEM) for MR (P = 0.67).One calf from the WP treatment received oral rehydration solutions for diarrhea at 10 d of age, as well as an antibiotic treatment at 20 d of age over a 3-d period for lameness.In addition, another WP calf received medication for lung disorder at 19 and 21 d of age.In contrast, no HF calves received medical treatments.Urinary recovery of lactulose (P = 0.05) and Cr-EDTA (P = 0.02) between 0 and 6 h were greater in calves fed WP than in those fed MR (Figure 1).In addition, there was a trend for a higher Cr-EDTA recovery in WP-fed calves between 0 and 24 h urine collection (P = 0.07).In contrast, no differences were detected in d-mannitol recovery.

Gastrointestinal Tract Weights and Lengths
Weights and lengths of the GIT are presented in Table 4. Body weight measured on the dissection day did not differ between treatments.Omasum, reticulum, and rumen were heavier (P < 0.05) in WP-fed calves  than in MR-fed calves.This led to a greater forestomach weight in WP-fed calves (P < 0.01).In addition, whole small intestine and jejunum weights were greater (P < 0.01) in WP-fed calves.No differences were observed in the weights of the liver and kidney, whereas the weight of the spleen was greater (P = 0.01) and the pancreas was lighter (P = 0.05) in MR than in WP-fed calves.
Finally, the cecum tended to be longer (P = 0.06), and the colon tended to be longer (P = 0.08) in WP-fed calves.This led to a trend for a longer whole large intestine in WP-fed calves (P = 0.07).When expressing GIT weights as percentage of BW, omasum, rumen, and reticulum percentage were greater in WP-fed calves (P ≤ 0.05; Supplemental Material S1, https: / / doi .org/ 10 .6084/m9 .figshare.21824487;Mellors et al., 2023).This led to a greater whole forestomach weight as percentage of BW in WP-fed calves (P < 0.01).Similarly, jejunum expressed as percentage BW was greater in WP-fed calves (P < 0.01), leading to a greater percentage of whole small intestine in WP-fed calves (P < 0.01).In addition, jejunum weight was greater and ileum weight was lower when expressed as percentage of small intestine weight in WP than MR-fed calves (P ≤ 0.05).

Intestinal Histomorphology, Immunohistochemistry, and Gene Expression
In the duodenum, the crypt width was greater in MR-fed calves (P = 0.02; Table 5).In the proximal jejunum, the villus width (P < 0.01) and SA index (P = 0.04) was greater in WP-fed calves.In the ileum, the crypt depth tended to be greater in WP-fed calves (P = 0.10) and the ratio of villus height to crypt dept tended to be greater in MR-fed calves (P = 0.08).The muscularis thickness of the ileum tended to be greater in MR-fed calves (P = 0.09).No differences were observed in the detection of Ki-67 antigen as well as in the tight junction genes and nutrient transporters in either the proximal jejunum or ileum presented in Figure 2.

Multivariate Analyses
The results of the PCA are shown in Figure 3.This analysis showed a clear distinction between the 2 dietary treatments based on tissue fatty acid composition (Figure 3A) and fatty acid composition of phospholipid composition data (Figure 3B).As shown in Figures 3A and B, the first 2 principal components accounted for 57 and 67% of the total variance between treatments, respectively.Volcano plots were used to show the differences in fatty acid composition between treatment groups in the jejunum and ileum using foldchange analyses and volcano plots adjusted for false discovery rate (Figure 4).Seven of the free fatty acids measured in the proximal jejunum tissue (Figure 4A) and 7 free fatty acids in the ileum (Figure 4B) differed between treatments according to the volcano plots.In the proximal jejunum and the ileum of MR-fed calves, the proportions of C15:0, C14:1, C17: 1cis -10, C18: 3n -3, C18: 4n -3, and C22: 5n -3 were lower, whereas the proportions of C12:0 was higher than in WP-fed calves.When considering the phospholipid fatty acid composition of both tissues, 6 of the phospholipid fatty acids measured in the proximal jejunum (Figure 4C) and 3 in the ileum (Figure 4D) differed between treatments according to the volcano plots.In the proximal jejunum of calves fed MR, the proportions of C18: 3n -3, C22: 5n -3, C16:1, C20: 2n -6, C20: 5n -3, and C20: 4n -3 were lower.In the ileum of calves fed MR, the proportions of C18: 3n -3, C20: 5n -3, and C22: 5n -3 were lower than in WP-fed calves.

DISCUSSION
The current study hypothesized that the WP treatment would enhance GIT development, alter the composition of intestinal tissues, and decrease intestinal permeability compared with calves fed MR.Results showed that WP-fed calves had increased GIT mass and increased SA index in the proximal jejunum, but also had greater gut permeability.The results of the present study suggest that for a similar macronutrient composition in terms of fat, protein, and lactose, differences in micronutrients can affect GIT growth and function.In addition, free fatty acid and phospholipid fatty acid profiles in jejunum and ileum tissues contained greater proportions of n-3 PUFA in WP-fed calves and greater proportions of n-6 PUFA in MR-fed calves, resulting in a greater ratio of n-6 to n-3 PUFA in MR-fed calves.
Most studies comparing WM and MR differed in feeding rate, as well as in macronutrients in terms of lactose, fat, and protein inclusion (Niwińska, 2005;Górka et al., 2011b), making it difficult to draw conclusions.A study from Lee et al. (2009) showed that calves fed WM would have a greater BW at weaning (d 49) and postweaning (d 70) than calves fed a MR with a similar macronutrient profile and fed at the same feeding level and percentage of solids.In the current study, both treatments were offered at the same feeding rate and the MR contained a similar profile in fat, lactose, and protein than WP.However, in a study including the same liquid diets as in the present study, Wilms et al. (2022) reported a lower whey protein nitrogen index for WP (1.9 mg/g) than MR (6.9 mg/g) due to the high heat applied either during the pasteurization or during the evaporation process.This may have resulted in lower protein digestibility of the WP treatment.Similarly, thiobarbituric acid, which is an indicator of lipid oxidation, was greater in the WP (0.883 mg/kg) than the MR treatment (0.203 mg/kg).In contrast, peroxide values which gives an indication of rancidity in unsaturated oils did not differ with 10.5 and 11.9 mEq/kg for WP and MR, respectively.Although the WP treatment aimed to represent bovine WM, major differences exist between dehydrated and fresh WM.These include the fat globule structure and size, as the processing steps involve homogenization that disrupts the fat globule membranes.The high heat applied during processing may also inactivate bioactive compounds present in WM.This may explain why no growth differences between WP and MR were detected in Wilms et al. (2022) and in the current study, unlike the study from Lee et al. (2009).In that study, the enhanced growth for calves fed WM may also be related to the presence of vegetable proteins (12% DM of wheat protein concentrate and 5.5% DM of soy protein) and low skim milk inclusion in the MR diet.Amino acids such as leucine have been shown to promote intestinal development, whereas others such as phenylalanine inhibited intestinal development in milk-fed Holstein calves (Cao et al., 2019).The AA profile of the 2 liquid diets were similar due to the use of proteins from dairy origin in the MR treatment and may therefore not be the major driver of the observed differences.However, it cannot be ruled out that differences in the casein-to-whey ratio or in protein quality could affect GIT development.The largest treatment differences were present in the composition of the fat fraction, whose fatty acid profiles largely differed between milk fat and the blend of palm and coconut oils in MR.Several studies suggested that supplementation of butyric acid, medium-chain fatty acids, ALA, lauric acid, and myristic acid in MR improves performance and reduces the incidence of diarrhea in calves (Hill et al., 2007(Hill et al., , 2011;;Esselburn et al., 2013;Garcia et al., Mellors et al.: GASTROINTESTINAL STRUCTURE AND FUNCTION

2015)
. Therefore, it is plausible that differences in the fatty acid profile between MR and the WP could explain part of the observed differences in GIT mass.
As for fatty acids, butyrate has been extensively studied as a dietary supplement for MR and is commonly administered in the salt form sodium butyrate.Butyrate inclusion in MR enhances GIT development when administered to calves either in MR or starter feed (Guilloteau et al., 2009;Niwińska et al., 2017).In the present study, butyrate content in the WP and MR treatments were 3.24 and 0.07% of total fatty acid, respectively, and may have contributed to increased forestomach mass in the WP group.This is consistent with studies showing that butyrate supplementation in MR increased forestomach weight and rumen papillae length and width (Górka et al., 2009(Górka et al., , 2011a,b),b).This suggests that butyrate may have had a proliferative effect in the rumen in the present study.However, the lack of treatment effect on the Ki-67 antigen analyzed in duodenum, proximal jejunum, and ileum does not support this mechanism in intestinal tissues.A possible hypothesis would be that butyrate enhances the development of the GIT by increasing GLP-2-mediated intestinal proliferation (Górka et al., 2011b;Penner et al., 2011); however, no GLP-2 receptors were detected in the rumen of mature cattle (Taylor-Edwards et al., 2010).Thus, the difference between treatments in butyric acid intake may have contributed to the increase in the foregut mass of WP-fed calves, although it cannot be ruled out that other micronutrients in WP may also have played a role.
The higher SA index in the proximal jejunum of WPfed calves in this experiment, as well as greater mass in the proximal jejunum and the small intestine, indicate greater potential for nutrient absorption.The trend for a greater villus height to crypt depth ratio was only in the ileum of MR-fed calves and could indicate increased intestinal proliferation (Jeurissen et al., 2002).However, detection of Ki-67 antigen did not differ in any of the small intestinal segments, suggesting that changes in microstructure in the proximal jejunum may not be due to changes in proliferation.When examined individually, villus height and crypt depth did not differ between WP and MR.The histological measurements in this study suggest that the type of liquid diet affected both gut mass and epithelial microstructure differently.However, the overall effect of these changes on nutrient absorption is unknown.The greater small intestinal weight in WP-fed calves could also be related to a lower nutrient availability consequent to the low whey protein nitrogen index, as feeding a low-quality skim milk powder has been associated with greater intestinal mucosa thickness in rats (Hillman et al., 2019).However, in the present study, no differences in muscularis thickness were detected in the duodenum and proximal jejunum, whereas higher values were observed in MR-fed calves in the ileum.
The AA profile and other protein characteristics may also play a role in neonatal development of the GIT.The protein source in WP was mainly casein (82%) with a smaller fraction of whey proteins (18%) resulting in a casein-to-whey ratio of 4.6, whereas the MR used in this study contained a ratio of 2.2.However, the influence of the casein-to-whey ratio in MR including dairy proteins on digestion and GIT development of dairy calves is not known (Terosky et al., 1997;Lammers et al., 1998).In piglets, both glutamine and arginine supplementation improved intestinal morphology, such as absorption area, and in weaned pigs, glutamine supplementation reduced the severity of diarrhea (Wu  , 1996;Jiang et al., 2009).In calves, histological measurements from the duodenum and jejunum showed that villus height, width, and SA index were enhanced in calves fed WM (20% arrival BW as volume) when supplemented with arginine and glutamine (van Keulen et al., 2020).However, in the current study arginine and glutamine content were comparable in WP and MR, making it unlikely that the changes in GIT growth and microstructure were influenced by these AA.Ahangarani et al. (2020) reported that glutamic acid supplementation in MR did not alter permeability.Although AA profiles were similar in both treatments, the bioavailability of some AA may have been affected in WP due to the high heat applied during process-  ing.Considering existing literature, the possible effect of the AA profile on GIT development in the present study is unlikely but cannot be excluded.
Recent studies have shown that many dietary factors such as weaning transition (Wood et al., 2015), osmolality in MR (Wilms et al., 2019), fat inclusion in MR (Amado et al., 2019;Welboren et al., 2021) can affect intestinal permeability in calves when assessed by oral administration of ingestible permeability markers.In the current study, calves fed WP had increased urinary recovery of Cr-EDTA and lactulose in the first 6 h after marker administration, suggesting a greater paracellu- lar permeability in WP than MR-fed calves.It should be noted that because markers were orally pulse dosed instead of the morning milk meal, the marker solutions likely transited through the rumen and may therefore not solely represent intestinal permeability.Indeed, the GIT has regional differences in permeability and these variations are hypothesized to be determined by nutrient absorption (Zhang et al., 2013), endoocytoxic potential (Nejdfors et al., 2000), and microbial fermentation (Penner et al., 2014).Penner et al. (2014) determined that the jejunum of ruminants boasts the highest transcellular permeability of the intestinal region as indicated by mannitol transport, and the omasum is the region of greatest large molecule permeability, as measured by inulin transport.After the omasum, the rumen is the second most prominent site of paracellular permeability, followed by the jejunum.Thus, in calves who are functionally monogastric and with an underdeveloped rumen, it can be hypothesized that the jejunum is the major site of paracellular permeability.Nevertheless, it cannot be excluded that the greater lactulose and Cr-EDTA urinary recovery in WP-fed calves could be related to a greater ruminal and intestinal absorptive surface.
Increased intestinal permeability has previously been associated with a loss of epithelial barrier integrity from GIT (Klein et al., 2008;Araujo et al., 2015;Wood et al., 2015), but it is important to note that greater permeability is not necessarily associated with diarrhea or indicative of a damaged epithelium.A decrease in intestinal permeability has been associated with an increase in jejunal and colonic occludin protein expression (Huang et al., 2015).However, in the current study, despite differences in gut permeability assessed by indigestible markers recovery in urine, no differences in tight junction gene expression and inflammatory cytokines were detected.However, the gene expression results do not necessarily represent these molecules at the functional protein level.The results of the present study indicate that feeding WP and MR alters intestinal permeability without affecting gene expression of tight junction proteins or cytokines, although this observation is only valid for the target genes evaluated in this study.Age differences may also have influenced the results because gut permeability was assessed on d 21 and dissection was performed on d 29 after arrival at the facility.Unlike metabolic diseases such as subacute rumen acidosis, for which there is a known threshold for subclinical and clinical manifestations (AlZahal et al., 2007), there is no such threshold for intestinal permeability.It would be beneficial to develop a threshold for calves fed fresh WM to quantify what different degrees of permeability mean to calves in terms of health and performance.
Differences in fatty acid profiles of diets are known to alter intestinal composition and function in piglets (Christon et al., 1989;Daveloose et al., 1993).These studies examined membrane fluidity and composition; however, the results of Khajuria et al. (2002) suggest that intestinal permeability, membrane fluidity, and histomorphology of intestinal tissues are related.In the present study, the phospholipid fatty acid profile of jejunal tissue and urinary recovery of indigestible markers were correlated (data not shown), although there was a week elapse between permeability measurement and dissection.Polyunsaturated fatty acids from the n-3 and n-6 series have been investigated for their ability to influence permeability due to their anti-inflammatory and pro-inflammatory functions and their incorporation into cell membranes (Sardesai, 1992;Alexander, 1998;Calder, 2010).Wilms et al. (2022) showed that calves fed ad libitum levels of the same MR as in the present study had increased serum amyloid A (SAA) concentrations compared with calves fed the WP treatment.Elevated SAA may be indicative of metabolic inflammation, although this was not correlated with adverse health outcomes in calves fed MR in that study.Furthermore, according to the PCA plots, the lipid profiles were also separated by the treatments in terms of free fatty acids and phospholipids fatty acid profiles.On the volcano plots, the differences in fatty acid composition of gut tissues in calves fed MR relative to WP were highlighted by lower levels of C18: 3n -3, C22: 5n -3, C20: 5n -3, and C15:0.The composition of free fatty acids and phospholipid fatty acid profile in jejunal and ileal tissues largely reflected the fat fraction of the liquid diets, as indicated by higher levels of eicosapentaenoic acid (EPA), a lower n-6 to n-3 ratio, and less LA in the diet of the WP-fed calves.Although previous in vitro studies found that EPA improved tight junction functionality and consequently reduced intestinal permeability (Usami et al., 2001;Hossain and Hirata, 2008), this was not observed in the present study, which could be due to the damaged protein fraction in the WP treatment.In conclusion, the fatty acid profiles of the fat fraction of WP and MR altered the composition of free fatty acids and phospholipid fatty acid profile in small intestinal tissue and may play a role in altering GIT function.
This study was the first to investigate the effects of WP or MR with a same feeding rate and similar macronutrients in terms of fat, lactose, and protein on GIT development in calves.However, although colostrum management at birth was standardized, this study included a small number of calves originating from several dairy farms and further work is needed on a larger sample size.Future studies looking at the development of GIT and intestinal permeability may also benefit from including fresh WM to obtain biological references and to establish numerical boundaries for intestinal permeability and health in calves.The results of this study suggest that differences in fat composition, and possibly protein composition, affect the GIT morphology, intestinal permeability, and fatty acid tissue composition in preweaning calves.

CONCLUSIONS
The results of the current study demonstrated that the whole milk powder treatment increased the mass of the forestomach and the small intestine, when compared with a milk replacer having a similar macronutrient profile in terms of fat, lactose, and protein.
Additionally, the surface area index of the proximal jejunum and the intestinal permeability were greater in calves fed whole milk powder; however, no differences were observed in tight junction protein gene expression between treatments.Dietary fatty acid composition of the liquid diets was largely reflected in the free fatty acid tissue content and phospholipid fatty acid profiles of the jejunum and ileum tissues.Future studies should investigate which elements of the nutrient profiles of the liquid diet play the most important role in gut development and function to optimize milk replacer formulation.

Figure 1 .
Figure 1.Urinary recovery of d-mannitol, lactulose, and chromium (Cr)-EDTA after an oral dose of the markers in calves fed a whole milk powder (WP; 26% fat; n = 9), or a milk replacer (MR) with high fat (HF; 25% fat; n = 9).Urine was collected on d 21 after arrival over 0-6 h and 6-24 h collection periods.

Figure 2 .
Figure 2. Gene expression of tight junction and inflammatory genes in the proximal jejunum (A) and ileum (B) of 33-d-old calves fed a whole milk powder (WP; 26% fat; n = 9), or a milk replacer (MR) with high fat (HF; 25% fat; n = 9).Gene expression is normalized to 3 reference genes (β-actin, GADPH, RNA-pol) and expressed relative to the average of the WP group.No differences were detected between the two treatments.

Figure 3 .
Figure 3. Principal component analysis (PCA) plot of tissue (A) free fatty acid composition and (B) fatty acid profile of the phospholipid fraction in the intestine (proximal jejunum and ileum) of 33-d-old calves fed whole milk powder (26% fat; WP; n = 9) or a milk replacer high in fat (25% fat; MR; n = 9).

Figure 4 .
Figure 4. Volcano plot of tissue free fatty acid composition in the (A) proximal jejunum and the (B) ileum tissues and fatty acid profile of the phospholipid fraction in the (C) proximal jejunum and the (D) ileum tissues of 33-d-old calves fed a whole milk powder (WP; 26% fat; n = 9) or a milk replacer high in fat (MR; 25% fat; n = 9).Each circle represents a single fatty acid.The colored circles represent the fatty acids above the threshold, with red representing higher and blue representing lower levels in the MR compared with the WP treatment.Important fatty acids identified with fold-change (FC) threshold (x-axis) 2 and false discovery rate (FDR) threshold (y-axis) 0.05.Note both fold changes and P-values are log-transformed and Pareto scaled.

Figure 4 (
Figure 4 (Continued).Volcano plot of tissue free fatty acid composition in the (A) proximal jejunum and the (B) ileum tissues and fatty acid profile of the phospholipid fraction in the (C) proximal jejunum and the (D) ileum tissues of 33-d-old calves fed a whole milk powder (WP; 26% fat; n = 9) or a milk replacer high in fat (MR; 25% fat; n = 9).Each circle represents a single fatty acid.The colored circles represent the fatty acids above the threshold, with red representing higher and blue representing lower levels in the MR compared with the WP treatment.Important fatty acids identified with fold-change (FC) threshold (x-axis) 2 and false discovery rate (FDR) threshold (y-axis) 0.05.Note both fold changes and P-values are log-transformed and Pareto scaled.

Table 1 .
Fatty acid profile of the fat fraction of a whole milk powder (WP) and a milk replacer (MR) fed 3.0 L, 3 times daily to calves 7Sum of PUFA from 18 to 22 carbons in length.

Table 2 .
Amino acid profiles of a whole milk powder (WP) and a milk replacer (MR) fed 3.0 L, 3 times daily to calves

Table 3 .
Mellors et al.:GASTROINTESTINAL STRUCTURE AND FUNCTION Primer pair sequences of genes analyzed in proximal jejunum and ileum tissue collected in 33-d-old calves fed a whole milk powder or a milk replacer

Table 4 .
Mellors et al.:GASTROINTESTINAL STRUCTURE AND FUNCTION The effect of feeding a whole milk powder (WP) or a milk replacer (MR) on gastrointestinal lengths and weights in 33-d-old calves fed 3.0 L, 3 times daily 1Treatments included a whole milk powder (26% fat; 21.6 MJ/kg; WP; n = 9), and a milk replacer with high fat (25% fat; 21.3 MJ/kg; MR; n = 9).Treatment concentration was 135.0 g/L to reflect the solid percentage of bovine whole milk. 2

Table 5 .
Mellors et al.:GASTROINTESTINAL STRUCTURE AND FUNCTION Histomorphometric measurements of villi, crypts, and muscularis layer thickness within the duodenum, proximal jejunum, and ileum of 33-d-old calves fed a whole milk powder (WP) or a milk replacer (MR) fed 3.0 L, 3 times daily 2Treatments included a whole milk powder (26% fat; 21.6 MJ/kg; n = 9), and a milk replacer with high fat (25% fat; 21.3 MJ/kg; n = 9).Treatment concentration was 135.0 g/L to reflect the solid percentage of bovine whole milk.

Table 6 .
Mellors et al.:GASTROINTESTINAL STRUCTURE AND FUNCTION Free fatty acid composition of proximal jejunum tissue of 33-d-old calves fed a whole milk powder (WP) or a milk replacer (MR) fed 3.0 L, 3 times daily 1Expressed in % total fatty acids, unless specified otherwise.

Table 7 .
Phospholipid fatty acid composition of proximal jejunum tissue of 33-d-old calves fed a whole milk powder (WP) or a milk replacer (MR) with the same fat inclusion 3.0 L, 3 times daily Mellors et al.: GASTROINTESTINAL STRUCTURE AND FUNCTION