The experiment conducted herein was approved by the Institutional Animal Care and Use Committee of the University of Tennessee (protocol no. 2655-0219). All experimental procedures were performed in accordance with the animal ethics approval and regulations. The study was carried out in compliance with the ARRIVE guidelines.
Experimental design
A total of thirty-two (1- to 2-week-old) bull calves were obtained from a commercial operation and raised following industry standard recommendations. Calves [body weight (BW) = 121 ± 2.2 kg; 12 ± 1 weeks of age; mean ± SD] were housed in individual pens in climate-controlled rooms (19.8 ± 0.8 °C constant ambient temperature) 3 day prior to the study at the East Tennessee Research and Education Center—Johnson Animal Research and Teaching Unit at the University of Tennessee-Knoxville14,22. Each room accommodated 8 pens so that the study was conducted in 2 cohorts of 16 calves each. Calves were housed at either thermoneutral (TN; constant 19.5 °C ambient temperature) or heat stress (HS; diurnal maximal ambient temperature of 37.8 °C) for 7 days. Diurnal HS climate resulted in 12 h/day of heat stress from day 1 through 722 (Supplementary Figures). Commercial milk replacer was fed in bottles to each individual animal once daily at 0500 h at 340 g following industry recommendations23. Water was offered ad libitum four times daily at 0500, 1200, 1700, and 2000 h. Calves did not show signs of health issues prior and during the course of the study.
Treatments
Calves were randomly assigned to 1 of 4 treatment groups (n = 8 calves/treatment). Treatments were (1) TN conditions fed ad libitum starter (TN), (2) HS conditions fed ad libitum starter (HS), 3) HS supplemented with 3.0 g/calf/day of AO postbiotic in milk replacer (HSP; Biozyme, Inc., St. Joseph, MO), and TN with ~ 8% restriction of starter consumption (TNR). The postbiotic was mixed in each bottle thoroughly with milk replacer to ensure consumption and post-ruminal delivery. The postbiotic was administered 13 days prior to imposing heat stress11 and continue until the end of the study on day 7. Full consumption of milk replacer was confirmed on each individual calf prior and during allocation to temperature-controlled rooms. Calf starter was offered four times daily at 0500, 1200, 1700, and 2000 h to allow for 5–10% refusal (i.e. ad libitum) in the TN, HS, and HSP treatments. The feed intake restriction imposed in TNR calves was based on feed intake data reported on heat-stressed growing dairy cattle13. Body weights were measured prior to the administration of the postbiotic and used as covariate in the statistical analysis (mean group body weight was 98, 102, 101, and 101 kg/animal for HS, HSP, TN, and TNR, respectively).
Thermal load assessment
The temperature and relative humidity in the rooms were monitored on day 1 to 7 every 10 min using HOBO U23 Pro v2 (Onset Computer Corp., Bourne, MA; accuracy ± 0.21 °C and 2.5% relative humidity) as previously used14,24. Each calf’s thermal response was evaluated for four times daily at 0700, 1100, 1500, and 1900 h using rectal temperature (RT; GLA M700 digital thermometer; accuracy ± 0.1 °C), skin temperatures (ST) at a clean shaven 10 cm × 10 cm patch on the rump at ~ 15 cm in distance (FLIR imaging gun; accuracy ± 1.5 °C), and respiration rates by counting flank movements for 15 s and reported as breaths/min. Additional RT data collected on HS and HSP calves was obtained every 60 min from 0700 to 2000 h. Mean body temperatures (MBT) were calculated using RT and ST in the following equation25: MBT = (RT × 0.70) + (ST × 0.30).
Performance measurements
Body weight was measured on day 1 and 7, and consumption of water, milk replacer, and starter was recorded daily on day 1 through 7. All calves consumed the totality of milk replacer offered and feed intake was calculated by adding the amount consumed of milk replacer and starter on a DM basis. Feed to gain ratio was calculated as kg of total intake on DM basis/kg of BW gain. Total energetic efficiency was calculated as the gross energy gain/metabolizable energy intake26, and partial energetic efficiency was calculated as gross energy gain/the difference between metabolizable energy intake and net energy maintenance27. Samples were taken of the milk replacer and the pelleted starter to analyze nutrient contents (Supplementary Tables). Samples of rectum content collected on day 3, 4, 5, 6, and 7 were used to determine water content in feces.
Analysis of plasma proteins and metabolites
Individual blood samples were collected at 0700 h daily on day 1 to 7 by jugular venipuncture in sodium heparin tubes and separated for plasma collection at 1200×g for 10 min at 4 °C within 30 min and stored at − 80 °C. Plasma acute phase proteins were analyzed using enzyme-linked immunosorbent assays (bovine haptoglobin: Immunology Consultants Laboratory, Inc., Portland, Oregon; multispecies serum amyloid A : Tridelta Development, Maynooth, County Kildare, Ireland18) on day 1, 3, 5, and 7. Bovine Lipocalin-2 was detected according to manufacturer protocol (MyBioSource, Inc., CA; Catalog N MBS018977) on day 1, 3, 5, and 7. The biochemical technique is based on Lipocalin-2 antibody-Lipocalin-2 antigen interactions (immunosorbency) and a colorimetric detection system to detect Lipocalin-2 antigen targets in samples. Bovine zonulin was detected using enzyme-linked immunosorbent assay kit according to manufacturer protocol (Haptoglobin Precursor; Antibodies-online Inc., PA; Catalog No. ABIN992457) on day 1, 3, 5, and 7. Plasma glucose, urea-N, and NEFA concentrations were determined on day 1 through 7. Glucose and urea-N concentrations were determined using commercially available enzymatic assays (Sigma-Aldrich, St. Louis, MO). Plasma NEFA concentrations were determined using commercial assay kit (Wako Diagnostics, Mountain View, CA). Plasma l-lactate concentrations were determined using a commercial assay kit (BioAssay Systems, EnzyChrom (ECLC-100), # CA09A28) on day 1, 3, 5, and 7. Concentration of metabolites were determined using a microplate spectrophotometer (BioTek Synergy H1 Multi-Mode Reader; Winooski, VT). Intra-assay and inter-assay coefficients of variation showed a range of 1.0 to 16.7%.
Analysis of blood gases
Whole blood samples collected on day 1 and 7 were used to conduct blood gas analysis using i-STAT analyzer according to protocol provided by manufacture (Abbott Point of Care Inc., Princeton, NJ; Supplementary Tables).
Intestinal permeability
Small intestine permeability was assessed by adding lactulose (0.50 g/kg BW) and mannitol (0.10 g/kg BW) in milk replacer at 0500 h on day 7 (Sigma-Aldrich, St. Louis, MO28). Blood samples collected at 0700 h were used to harvest plasma then stored at − 80 °C until analysis. Plasma was submitted to a commercial lab to determine concentrations of the synthetic sugars using high performance liquid-chromatography coupled with mass spectrometry (University of North Texas, Denton, TX29).
Amino acid absorption
An exploratory analysis was conducted on 12 calves randomly selected (n = 4/treatment). Two catheters were placed into ipsilateral jugulars on day 614. On day 7, a sterile stable isotope-labeled AA mixture (0.10 g of 13C-labeled AA, 4.4 mg of 13C-labeled L-Met, and 5.7 mg of 13C-labeled L-His-HCl-H2O dissolved in 120 mL of saline) was infused into one of the catheters over 8 h at a constant rate of 1.0 mL/min using medical peristaltic pumps (Plum XL IV; Abott-Lifecare, San Antonio, TX). Twelve blood samples (5 mL) per calf were taken over the entire infusion period from the other jugular catheter into Na-EDTA tubes. Plasma was collected at 1200 × g for 10 min at 4 °C within 60 min and stored at − 80 °C. Plasma samples were prepared for ion ratio mass spectrometry analysis to determine the intestinal entry rate of individual AA as previously described30.
Amino acid model descriptions and parameter estimation
Briefly, a 4-pool dynamic model was constructed and used to estimate plasma amino acid (AA) entry rates and AA turnover rates between the fast and slow pool as previously described30. The fast pool represents blood, interstitial, and cytoplasmic free AA, which was assumed as 14.9% of BW. The slow pool represents protein-bound AA and was calculated using the assumption that body protein is 18.8% of BW31. The AA composition of the fast and slow pools were set based on plasma and muscle AA concentrations as previously reported32. Model predictions of isotope ratios in the fast pool were fitted to the observed plasma AA isotope ratios for each AA within the infusion by maximizing a log-likelihood function using the Nelder-Mead optimization algorithm33. To adjust AA intake among individual animals, AA relative bioavailability was calculated by dividing plasma AA entry rate by AA intake. All modeling work was conducted in R (version 3.5.134).
Statistical analysis
Data were analyzed using a mixed model in SAS version 9.4 (SAS Institute Inc., Cary, NC). Data were analyzed for homoscedasticity and normality of residuals. Body weight data collected prior to the beginning of the postbiotic feeding were included as a covariate adjustment in the model. Best-fit models were determined using backwards manual selection, specifically taking low Akaike information criterion (AIC) into consideration. All models included the overall mean, the fixed effect of treatment, the fixed effect of replica, the random effect of calf, the covariate effects, and the random error. A repeated measure was included in the model for non-random and consecutive measurements taken over time (h or day). Repeated measures procedure was used to determine overall differences related to treatments and time and treatment by time interactions. Covariate analysis was included in the model if statistically significant (P ≤ 0.05). Thermoregulatory responses related to changes in ambient temperature and time were characterized using treatment replica, time effects, and all interactions in the model. Significant differences were declared at P ≤ 0.05, and trends were declared at 0.05 < P ≤ 0.10. All results are reported as least squares means or slopes ± standard error of the mean. Models to characterize thermoregulatory responses with treatments were also tested to determine if ambient and rectal temperatures captured the information in both variables. Rectal temperature was characterized using treatment and ambient temperature regression effects. Respiration rate was characterized using treatment, ambient temperature, and rectal temperature regression effects.
