The experiment was divided into 2 phases. The first phase refers to a gestational feeding trial conducted from December 2021 to March 2022. The second phase refers to the postnatal development of calves from birth to slaughter (March 2022–August 2023). Animal trials were conducted the Ontario Beef Research Centre (University of Guelph, Guelph, ON) and the slaughters at University of Guelph Meat Laboratory in accordance with Canadian Council of Animal Care guidelines and approved by the University of Guelph (Guelph, ON) Animal Care Council (Animal Utilization Protocol #4730).

A total of 30 Angus-Simmental cross bred multiparous cows averaging 705 ± 12 kg of body weight (BW) and 12 ± 0.8 mm of subcutaneous fat thickness were used. All cows were from the Ontario Beef Research Centre (University of Guelph, Guelph, ON), and underwent to a fixed-time artificial insemination using 8 sires with similar genetic merit). Only cows pregnant with male calves were used to avoid confounding effects with between the dietary treatments and calf sex. From insemination to 180 d of gestation, all cows were managed as a single herd and fed the same diet. Cows were housed in pens (5.9 × 13.4 m²; n = 6 cows/pen) and all animals within a pen had access to 3 Insentec feed bunks (Hokofarm Group, Marknesse, Netherlands). At 180 d of gestation cows were randomly assigned to one of 2 experimental treatments: CONTROL, where cows were fed a basal diet with 4.1 KIU of vitamin A/kg; and VITA, where cows were fed the same basal diet enriched with pure vitamin A to 12.2 KIU of vitamin A/kg (Table 1). Vitamin A was sourced from DSM (ROVIMIX^®) and its inclusion rate was chosen based on the pre-existing level of vitamin A in the control diet and supplementing 3 times the amount of vitamin A in the supplemented group ensuring a substantial disparity between the 2 groups, as well as to compensate the losses due to microbial metabolism as suggested by Rode et al. (1990). Rations were formulated according to NASEM (2016) and cows were fed at 100% maintenance requirements for pregnant mature beef cows to avoid confounding effects between feed intake and vitamin A supplementation. Cattle were weighed every 28 d, and the BW was used to adjust the feed intake, dry matter intake (DMI) allowance was 1.7% of BW. The feed intake data was cleaned using the IntakeInspectR package, version 2.1.4 Innes et al., (2023). Animals from both groups were fed the experimental diets from 180 d of gestation until parturition. After parturition, the cow-calf pair was managed as a single herd and kept under the same dietary and environmental conditions until weaning.

At 10 d of age, calves were submitted to a muscle biopsy sampling, and tissue was collected from the right side of Longissimus muscle located between the 12^th and 13^th ribs. The biopsy site was initially shaved and sanitized with 4% chlorohexidine and 70%-ethanol alcohol. Following the sanitization, the biopsy site was anesthetized with 2% lidocaine HCl (6 mL). A 4 cm incision was then created using a sterile blade 10 scalpel across the skin and subcutaneous fat (when present). Next, muscle tissue was excised by using an 8-mm biopsy punch and rinsed with sterile phosphate-buffered saline solution. Visible connective (epimysium) and subcutaneous fat tissue (when present) were trimmed off, and the muscle sample was immediately snap-frozen, powdered in liquid nitrogen, and then stored at −80°C until analysis of gene expression and protein abundance of intramuscular adipogenesis markers.

Total RNA was extracted from 0.1 g of tissue using Trizol^® (InvitrogenTM, Thermo Fisher Scientific^®, Oregon, USA) following the manufacture’s recommendations. The total RNA was quantified with a NanoDrop spectrophotometer (Thermo Fisher Scientific^®, Waltham, MA, USA), ensuring an optimal 260/280 ratio between 1.8 and 2.0, and the integrity was assessed in a 1% agarose gel. The RNA samples were reverse transcribed into cDNA using the GoScriptTM Reverse Transcription System Kit (Promega, Madison, WI, USA). The primers (Table 2) for amplification of target and endogenous genes were designed using PrimerQuest Software (PrimerQuest–design qPCR assays | IDT [idtdna.com]) with sequences obtained from GenBank (GenBank Overview [nih.gov]). Real-time quantitative PCR was performed in the thermal cycler QuantStudio 3 (Applied Biosystems, Foster City, CA, USA) using the SYBR Green detection method (Applied Biosystems, Foster City, CA, USA) and GoTaq^® qPCR Master Mix Kit (Promega, Madison, WI, USA). Results are expressed relative to GAPDH and were calculated according to the methods described by Pffafl (2001).

Total protein was extracted from 0.1 g of tissue in 1 mL of lysis buffer (10 mM of Tris HCl [pH 7.6], 150 mM of NaCl, 1% of Triton X-100, 0.5% sodium deoxycholate, 1% sodium dodecyl sulfate [SDS], and 1% of protease inhibitor cocktail mammalian cells and tissues [Sigma-Aldrich^®]), and the lysate was sonicated. Total protein content was estimated using the Bradford protein assay (Bio-Rad, Hercules, CA, USA) and stored at −80°C. The proteins were separated using a 10% SDS-PAGE gel loaded with 40 μg of protein per sample, transferred to a 0.45 μm Nitrocellulose membrane (Bio-Rad, Hercules, CA, USA) using a semi-dry Trans-Blot Turbo Transfer System (Bio-Rad, Hercules, CA, USA), and blocked for 1 h at room temperature with 2% bovine serum albumin (BSA, Sigma Aldrich^®) in 1x Tris-Buffered Saline (TBS1x; 50mM Tris-HCL, pH 7.5; 150mM NaCl; Sigma Aldrich^®). Subsequently, the membranes were incubated for 12 h at 4°C with the primary antibodies (Table 3) diluted in blocking solution. After 12 h of incubation, membranes were washed 3 times for 15 min with Tris Buffered Saline and 0.1% Tween^® (TBSt) and incubated with the secondary antibody (Table 3) diluted in blocking solution for 1 h at room temperature. Membranes were then washed with TBSt, revealed by ECL Plus Western Blotting Detection System (GE HealthCare, Buckinghamshire, Ul) and the images generated by ChemiDoc XRS+ System (Bio-Rad, Hercules, CA, USA). The α-tubulin was used as a housekeeping protein to normalize the abundance of the target proteins.

At birth the offspring from both treatments were weighed, submitted to a castration procedure within 24 h after birth and kept with the dams under the same environmental and nutritional conditions until weaning (216 ± 0.9 d old). The birth and weaning weights were used to calculate the average daily gain (ADG) of the calves during the suckling phase, BW gain from birth to weaning divided by days of age at weaning. During the suckling phase, only cows had access to feed via the Insentec bins while calves only consumed their dam’s milk. After weaning, the steers were managed as a single herd and fed the same growing diet composed of straw (16.5%), hay (31.9%), corn silage (9.9%), haylage (38.5%), and mineral mixture (3.2%) for 100 d. The weaning weight and the initial BW at the finishing phase were used to calculate the ADG during the growing phase. At the end of the growing phase, steers were moved to finishing barn of Ontario Beef Research Centre (University of Guelph, Guelph, ON) and were fed the same finishing diet composed of corn silage (14%), high moisture corn (56%), corn grain (20%), soybean meal (8%), and mineral mixture (2%) for 90 d prior to slaughter (479 ± 1.6 d of age). Steers were housed in pens (5.9 × 13.4 m²; n = 6 steers/pen) with access to 3 Insentec feed bunks within their pen allowing monitoring of their individual DMI. The feed intake data was cleaned using the IntakeInspectR package, version 2.1.4 (Innes et al., 2024).

Carcass ultrasound measurements of the offspring were taken at approximately 280, 330, 380, 410, 440, and 480 d of age. The measurements of the ribeye area and subcutaneous fat thickness were taken at Longissimus dorsi muscle between the 12^th and 13^th rib. A second measurement of subcutaneous fat was taken at the rump. Carcass subcutaneous fat thickness resulted from the average of measurements taken at Longissimus dorsi muscle and at the rump. Intramuscular fat percentage was also measured by ultrasound using a minimum of 4 independent images collected across the 11^th to 13^th ribs at a lateral position the animal’s midline at a point three-fourths of the distance from the medial end of the Longissimus dorsi muscle. All images were collected by using an ExaGo ultrasound (IMV Imaging, Bellshil, Scotland) equipped with a linear probe with 18 cm of length. Images were analyzed by using the Centralized Ultrasound Processing Lab (CUP LAB, Ames, IA, US) software (2023 version).

At the end of the finishing phase, steers were slaughtered at the Canadian Food Inspection Agency (CFIA) inspected Meat Science Laboratory of the University of Guelph. Due to the capacity of the meat science laboratory, the animals were not slaughtered on the same day. However, the number of animals per treatment was balanced across slaughtering dates to avoid confounding effects with the experimental treatments. Steers were humanely handled and slaughtered (via captive bolt stunning, followed by exsanguination) using commercial industry standards under the inspection of CFIA agent. At the end of the slaughtering process, each carcass was split, and each side of the carcasses was chilled intact at 4°C. After 48-h postmortem chill period, the longissimus muscle area (LMA) was measured at the 12^th rib on the left side of each carcass. Longissimus muscle areas were traced on transparencies and measured later with a planimeter. Next, longissimus thoracis rib portions anterior to the 12^th and 13^th rib interface was collected and had their position tracked until subsampling into steaks. The Longissimus thoracis samples were then cut into 2.5-cm thick steaks beginning at the posterior end (12^th/13^th rib interface) being the first steak (closest to the 12^th/13^th rib interface) used for instrumental tenderness assessment. Steaks were vacuum packaged and aged for 14 d at 4°C, then frozen prior to instrumental tenderness analysis.

Steaks were thawed before being cooked to an internal temperature of 72 °C on a clamshell Garland Grill (Ed-30B: Garland Commercial Ranges LTD, Mississauga, Canada) set to a surface temperature of 105°C. Following cooking, samples were cooled in a refrigerator to an internal temperature of approximately 4°C. Eight 1.25-cm diameter cores running parallel to the muscle fibers were removed from each steak. Each core was sheared perpendicularly to the muscle fibers with a Warner–Bratzler blade using a TA-XT Plus Texture Analyzer (Texture Technologies Corp., Scarsdale, USA) at a crosshead speed of 3.3 mm/s. The average peak force was recorded for each steak.

The data measured were analyzed according to the following base linear mixed model: Yikl=μ+Di+Sk+b1iBWikl+b2iSFTikl+eikl[Eq. 1]where Yikl is the observed value; μ is the intercept; Di is the fixed effect of the i^th level of Maternal dietary treatment, with i = 1, 2; Sk is the random effect of the k^th Sire, with k = 1… 8, assuming S∼N(0,Iσs2), where I represents the identity matrix; b1 is the partial regression coefficient for the covariate initial body weight (iBW) of the cow, where iBWikl represents the iBW of the l^th cow; b2 is the partial regression coefficient for the covariate initial subcutaneous fat thickness (iSFT) of the cow, where iSFTikl represents the iSFT of the l^th cow; and eikl is the random error associated with yikl, assuming e∼N(0,Iσe2).

For traits measured at finishing and slaughter, an additional covariate was included in Equation [1] to account for the time the calf spent at these periods, respectively. For traits measured multiple times across time in the calves, the following mixed model was used: Yijkl=μ+Di+Tj+(D*T)ij+Sk+b1iBWikl+b2iSFTikl+b3wAgeikl+eijkl[Eq. 2]Where all variables are as described in Eq. [1], except for Tj that is the fixed effect of the j^th level of time-point of measurement, with j = 1…6; (D*T)ij is the fixed effect interaction between the i^th level of Maternal dietary treatment and the j^th level of time-point of measurement; b3 is the partial regression coefficient for the covariate age of the progeny of the cow at weaning (wAge), where wAgeikl represents the wAge of the progeny of the l^th cow; and eijkl is the random error associated with yijkl, assuming e∼N(0,SPσe2), where SP represents a spatial power covariance structure, such that correlations amongst repeated records where a function of the age of the calves at each time point. Prior to final analyses, other covariance structures, such as compound symmetry, unstructured, and identity, were evaluated, but they were not used in the final analyses for showing greater Akaike information criterion values than SP.

Prior to final analyses, residuals were evaluated for assumptions of the models used. For each analysis, data points were removed one at a time if assumptions were not met, such as absolute Studentized residuals greater than 3 and significant (P < 0.01) Shapiro-Wilk’s test for normality.

Expected means were generated from the final models and separated using Tukey’s test. All analyses were performed in R (R version 4.3.2; R Core Team, 2023), and differences were declared when P ≤ 0.05.

The cows from both treatments showed similar initial BW (P = 0.28). Despite the observed trend of increased DMI by the cows at the control group (P = 0.06) no differences were observed for BW gain (P = 0.91) and final BW (P = 0.91; Table 4). Moreover, vitamin A supplementation during late gestation did not influence calves’ birth weight (P = 0.62), weaning weight (P = 0.91) and ADG from birth to weaning (P = 0.41; Table 5). Calves’ performance at the backgrounding phase was not influenced by maternal treatments, final BW (P = 0.64), BW gain (P = 0.21), and ADG (P = 0.21) (Table 5). Furthermore, at the finishing phase, maternal supplementation with vitamin A at late gestation did not alter the offspring’s final BW (P = 0.36), BW gain (P = 0.95), ADG (P = 0.94), and DMI (P = 0.74; Table 5).

A greater mRNA expression of retinoic acid receptor β (RARβ) was observed in the skeletal muscle of the calves from the vitamin A group (P = 0.05; Figure 1), whereas the mRNA expression of retinoic acid receptor γ (RARγ; P = 0.78) and the protein abundance of retinoid X receptor (P = 0.63) were similar between treatments (Figure 1).

The mRNA expression (P = 0.89) and protein abundance (P = 0.57) of platelet-derived growth factor receptor alpha (PDGFRα) did not change between treatments (Figure 2). Similarly, the mRNA expression of ZFP423 (P = 0.73) in the skeletal muscle of the calves was not affected by vitamin A supplementation of the cows at late gestation (Figure 3). However, a greater protein abundance of delta like non-canonical notch ligand 1 (DLK1; P < 0.01) was observed in the skeletal muscle of calves from vitamin A–supplemented cows (Figure 3). Regarding the peroxisome proliferator activated receptor (PPARγ), despite the lack of differences between treatments in mRNA expression (P = 0.95; Figure 4), a greater protein abundance of PPARγ was observed in calves born to dams supplemented with vitamin A at late gestation (P = 0.03; Figure 4).

Vitamin A supplementation at late gestation allowed an increase in intramuscular fat percentage in the offspring throughout all the evaluated stages of their post-natal life (Figure 5; P ≤ 0.05). No effects of vitamin A supplementation at late gestation were observed in the subcutaneous fat thickness (P > 0.05; Figure 6) or on LMA (P > 0.05; Figure 7) throughout the offspring’s life.

There were no effects of maternal vitamin A supplementation at late gestation on hot carcass weight (HCW; P = 0.83), and carcass dressing percentage (P = 0.40; Table 6). Similarly, there was no effect of vitamin A supplementation at late gestation of Kidney Pelvic Heart (KPH) fat in kilograms (P = 0.72) or percentage (P = 0.69) in the carcass of the offspring (Table 6). The carcass final temperature (P = 0.56) and pH_24h (P = 0.89) did not differ between calves born to dams fed the control diet or vitamin A–enriched diet (Table 6). However, beef from calves born to dams fed vitamin A–enriched diet tended to show greater Warner-Bratzler shear force values compared to beef from calves born to dams from the control group (P = 0.07; Table 6).