The Animal Research Ethics Committee of the Universidade do Estado de Santa Catarina approved the project (protocol number: 4948210322), following the guidelines of CONCEA/Brasil.

The GPS in cv. Isabel (Vitis labrusca) is composed of 3.63% stems, 50.19% barks, and 46.18% seeds (based on the dry matter). This material was collected in the municipality of Pinheiro Preto (27˚ 03' 02" S; 51˚ 13' 51" W) at an industrial winery. The residue was derived from the pressing process and separation of the solid fraction. Soon after being collected, the GPS was transported for storage in a trench-type silo (3.0 × 0.9 × 6.0 m for width, height, and length, respectively), coated with plastic on the sides and bottom. Once the GPS was placed in the silo, it was compacted with an agricultural tractor and then the silo was sealed with specific tarpaulin for storage of preserved bulky foods (Parcifil^®, Sapiranga, Brazil). The GPS was ensiled for 3 months before starting the experiment, characterized as grape pomace silage (GPS).

The production of grape pomace bran (GPB) required dehydrating the GPS. For logistical reasons, it was necessary to dry the GPS in two batches. The first dehydrated batch was carried out in a closed environment, using gas heaterand fans to heat the environment and force air circulation. The second batch of GPS was dehydrated in an open environment (i.e., in the sun) on plastic sheets. Both forms of drying were carried out at the Universidade do Estado de Santa Catarina (27˚ 06' 17" S 52˚ 36' 51" W) and lasted about 7 days until reaching values below 10% humidity to avoid deterioration during storage. After drying, the batches were mixed and ground using an industrial hammer grinder, using a sieve that provided a particle size of 4 mm.

Diets were formulated to meet nutritional requirements for an estimated average daily gain of 1.5 kg of body weight (BR-CORTE, 2016). By-products (GPS or GPB) were added to the diet based on total replacement of wheat bran and partial replacement of soybean hulls form the Control diet ( Table 1 ). The rations for the three dietary treatments were isonitrogenous and isoenergic.

Twenty-four steers were used, from industrial crosses (½ Charolais x ½ Nellore), to constitute three dietary treatments (eight steers/treatment) in a randomized design to standardize initial body weight between groups, considering each steer as an experimental unit. The animals from each treatment were fed in individual pens: control diet (traditional feed diet), GPS diet (inclusion of 100 g kg⁻¹ in the dry matter (DM) of grape pomace silage in the diet), or GPB diet (inclusion of 100 g kg⁻¹ in the DM of grape pomace bran in the diet). The study lasted 121 days with 21 days of diet adaptation and 100 days of the experimental period. The production aspects were previously published by Molosse et al. [18] .

¹Calcium (Max.) 220.00 g kg⁻¹; Calcium (Min.) 160.00 g kg⁻¹; Phosphorus (Min.) 40.00 g kg⁻¹; Magnesium (Min.) 6.00 g kg⁻¹; Sodium (Min.) 85.00 g kg⁻¹; Sulfur (Min.) 12.00 g kg⁻1; Cobalt (Min.) 20.00 mg kg⁻¹; Copper (Min.) 520.00 mg kg⁻1; Iodine (Min.) 25.00 mg kg⁻¹; Manganese (Min.) 650.00 mg kg⁻¹; Selenium (Min.) 10.00 mg kg⁻¹; Zinc (Min.) 2000.00 mg kg⁻¹; Fluoride (Max.) 400.00 mg kg⁻¹.

After 121 days of experimentation, 24 steers with an average body weight of 414 kg and 13 months of age were transported, 17 hours before slaughter, to a commercial slaughterhouse in the city of Chapecó, SC, Brazil, 25 km from the confinement area. Transport lasted about 1 hour. At the slaughterhouse, the animals remained in the covered waiting room for around 16 hours, with feed restriction and access to drinking water.

After rest period and water diet, the animals were directed to the containment box and stunned with a captive-bolt pistol. The slaughters followed current legislation and underwent veterinary inspection. The hot carcasses were weighed on a digital scale immediately after each carcass was dressed and washed. After 30 minutes of slaughter, pH was measured using a portable digital meter (TESTO^®, 205PH), between the 12^th and 13^th ribs in the left Longissimus thoracis et lumborum (LTL) muscle region. For calibration/adjustment of the measurement system (potentiometer), pH 7.00 and 4.01 commercial buffer solutions were used. A built-in temperature sensor ensured temperature compensation for the pH measurements. After slaughter, the carcasses were refrigerated and kept at 6.0 ± 0.2˚C for 24 hours. After this period, the pH was measured again, and the LTL muscle on the left side was removed and transported by 30 min in a thermal box to the laboratory. Then, we began analyzing the meat.

Six 2-cm thick steaks were cut from the LTL from each sample aseptically with sterile knives in a controlled environment using a Bunsen burner. Of these, one was used immediately to determine the parameters of meat color, water-holding capacity, cooking loss, shear force, and FA profile; and another steak was used to analyze shelf life, representing the beginning of the period (day 1 - moment that corresponds to 24 hours after the slaughter of the animal). The remaining steaks were stored refrigerated (as described below) for analysis at different shelf-life periods. All meat analyzes were carried out at the Animal Nutrition Laboratory and Microbiology Laboratory of the Universidade do Estado de Santa Catarina, UDESC, Chapecó, Brazil.

The steaks were cooked and after 2 minutes of exposure to ambient air, the surface color of the meat was measured using a portable gauging colorimeter (Konica Minolta^®, CR-400) with a D65 illuminant and spectral reflectance included as calibration mode with an 8-mm observation aperture and an 11-mm illuminant/observer. Luminosity (L*), redness (a*), and yellowing (b*) were measured at three randomly selected sample points. The values of the parameters were expressed as the mean of three measurements.

The water-holding capacity was measured using an adaptation of the methodology of [19] . 0.3 g of meat ground (by scraping the knife over the meat) was weighed and placed them on filter paper (15 × 15 cm) between two acrylic plates. A weight of 10 kg was placed on top of the sample for 5 minutes. After pressing, the meat was weighed, and water loss was calculated based on the weight difference. The result was expressed as a percentage.

Cooking loss and shear force analyses were performed using 2-cm steaks. Before cooking the meat, the sample was weighed and covered in aluminum foil. Cooking was carried out on a portable grill (Mondial^® Due Grill Smart), preheated to 170˚C until reaching an internal temperature of 75˚C (measured using a culinary thermometer inserted in the geometric center of the sample during cooking). After cooking, the samples were removed from the aluminum foil and the final weight was obtained after temperature stabilization. according to the methodology adapted from researchers [19] . The percentage of cooking water loss (CL) was calculated as follows: [(weight of raw steak − weight of cooked steak)/weight of raw steak] × 100.

The cooked and chilled samples were cut into subsamples in the fiber’s longitudinal direction. Using a digital caliper, the height and width of the subsamples were measured for later calculation of the area. The shear force was measured on each cuboid perpendicular to the fiber, using a texture analyzer (Stable Micro Systems^®, TA-XT plus) coupled to a 1 mm thick Warner-Bratzler V-shaped cutting blade. The test speed was 3.30 mm s⁻¹ to measure the force required to cross-cut each cuboid, and the values were expressed as N per mm2 of meat. All determinations were performed on two replicates per sample.

The parameters to measure meat quality—microbiological analysis, color and oxidative state was evaluated at 1, 3, 7, and 10 days of shelf life (a proxy for retail exposure), with day 1 considered 24 hours after slaughter (chilled meat). The fatty acid profile was evaluated in meat only on days 1 and 10 of shelf life.

Four steaks (2 cm thick per experimental unit) were placed in polystyrene trays coated with a low-density polyethylene film ( Figure 1 ). The trays were packed, labeled, and randomly placed in a controlled greenhouse, illuminated with white fluorescent light (Blumenau^®, five LED Tube lamps, 9W, 6500 K, 900 lumens) with forced ventilation and a temperature of 4 ± 0.4˚C. Lighting remained on 12 hours a day, simulating retail display conditions. The wrapped trays were changed places daily to minimize differences in the incidence and intensity of light and possible changes in temperature inside the equipment.

Before the other analyses were conducted, a subsample of 10 g of meat was aseptically separated and weighed in sterile plastic bags for food and homogenized with 90 ml of buffered peptone water for 10 min on a horizontal orbital shaker table (LOGEN SCIENTIFIC LS2312). From this dilution (10⁻¹), 1 mL was transferred to tubes containing 9 mL of buffered peptone water, obtaining the other serial dilutions of 10⁻² and 10⁻³, used on days 1, 3, and 7 of analysis. For day 10, dilutions of 10⁻⁴, 10⁻⁵, and 10⁻⁶ were used. Total bacterial counts (total mesophils) were performed on standard count agar (SCA), and lactic acid bacteria counts were performed on de Man, Rogosa, and Sharpe agar (MRS) using the pour plate method. The three serial dilutions performed previously were used, where 1 mL of each dilution was inoculated into sterile Petri dishes. Subsequently, the SCA plates were incubated in a bacteriological oven at 37 ± 1˚C for 24 h, and the MRS plates were placed in an anaerobic chamber (Panasonic^®, MCO-19AIC UV) in a 5% CO₂ atmosphere at 30 ± 1˚C for 72 h. To count Escherichia coli/coliforms and Enterobacteriaceae, 3M Petrifilm™ EC 6414 and EB 6420 plates were used, respectively. Initially, the 10⁻¹ dilution was used, and later (day 10), the 10⁻³ dilution was used. The plates were incubated at 37 ± 1˚C for 24 h. Bacterial counts were expressed as logarithms of colony-forming units per gram of meat (log CFU g⁻¹ meat).

From the central area of the steak, 0.5 grams were removed and homogenized in saline solution to determine the antioxidant/oxidant status. The samples were centrifuged for 10 min at 7500 g. The supernatants were collected and stored in microtubes at –20˚C until analysis.

Meat glutathione S-transferase (GST) was measured based on Habig et al. [20] and was expressed as µmol CDNB min⁻¹ mg⁻¹ of meat protein. Serum lipid peroxidation was measured based on the amount of thiobarbituric acid reactive substances (TBARS) according to literature [21] , and the results were expressed as nmol MDA mL⁻¹. Determination of reactive oxygen species (ROS) in the meat homogenate was based on Halliwell and Gutteridge [22] and expressed as U DCFHmg⁻¹ meat protein. According to Ellman [23] , protein thiols were measured, and the results were expressed as nmol thiols mg⁻¹ of protein.

The method of Bligh and Dyer [24] was modified for extraction of lipids from beef and animal feed samples. 1.5 g was added of bovine muscle samples, 0.5 mL of water, 5 mL of methanol, and 2.5 mL of chloroform into 15-mL polypropylene tubes, and mechanical shaking was performed for 60 min. Then, 2.5 mL of chloroform and Na₂SO₄ 1.5% solution were added to promote a biphasic system. This mixture was shaken for 2 minutes and then centrifuged for 15 minutes at 2000 rpm. Lipids obtained from the chloroform phase were subjected to FA analysis.

FA methylation was performed by a transesterification method proposed by researchers [25] . We added 1 mL of 0.4 M KOH methanolic solution to the extracted lipids in a test tube and shook the mixture in a vortex for 1 min. Samples were kept in a water bath for 10 min at boiling point. Subsequently, the material was cooled at room temperature, and 3 mL of 1 M H2SO4 was added, shaken in a vortex, and maintained in a water bath for 10 min. After cooling, we added 2 mL of hexane and centrifuged at 2000 rpm for 10 min. Finally, hexane with the FA methyl esters (FAMEs) was subjected to chromatography analysis.

For the FAME determination, a gas chromatograph model TRACE 1310 was equipped with a flame ionization detector (Thermo Scientific). One microliter of samples was injected in a splitsplitless injector operated in 1:20 ratio split mode at 250˚C. Hydrogen was used as carrier gas at a constant flow rate of 1.5 mL min⁻¹. FAME separation was carried out using an RT 2560 chromatography column (100 m × 0.25 mm × 0.20 μm thickness film, Restek, USA). The FAME compounds were identified by comparing experimental retention times with those of an authentic standard (FAME Mix-37, Sigma Aldrich, St. Louis, MO). The results were expressed as a percentage of each FA identified in the lipid fraction (Material Supplementary 1), considering the chain size equivalent factor of FAME for FID and conversion factor of ester to the respective acid, according to literature [26] .

All data were analyzed using the ‘MIXED procedure’ of SAS (SAS Inst. Inc., Cary, NC, USA; version 9.4), with the Satterthwaite approximation to determine the denominator degrees of freedom for the test of fixed effects. The carcass weight, yield, meat pH, shear force, water-holding capacity, CL, and intestinal and liver antioxidant variables were tested for fixed effects of treatment using the animal (treatment) as the random effect. All other variables were analyzed as repeated measures and were tested for fixed effects of treatment, day, and treatment × day, using the animal (treatment) as the random effect. The compound symmetry covariance structure was selected according to the lowest Akaike information criterion. Means were separated using the PDIFF method, and all results were reported as LSMEANS and standard errors. Significance was defined when P ≤ 0.05, and tendency when P > 0.05 and ≤ 0.10.

The GPS-fed steers had greater chilled carcass pH than the control animals (P < 0.05, Table 2 ); the animals that received the GPB diet had a pH similar to the steers in the control group. There were no diet differences for other carcass variables (weight, yield, and hot carcass pH) (P > 0.05).

The experimental diets influenced the meat’s chemical composition (P < 0.05). While there were no diet differences for DM content (P > 0.05), LTL from GPB fed steers has a greater crude protein content than LTL from steers fed the control diet (P < 0.05) ( Table 3 ).

There were no diet differences for meat luminosity (L*), redness (a*), and yellowing (b*) over various periods of simulated retail exposure (P > 0.05, Table 3 ). There was also no day effect for luminosity (P = 0.63) or red color (P = 0.13); however, there was a day effect for meat yellowness (P < 0.001) with lower values on day 1 than the other days (3, 7, and 10) for the 3 dietary treatments. There were no diet differences for Warner-Bratzler shear force, CL, and water-holding capacity (P > 0.05).

¹Dietary treatment: control diet (traditional confinement diet), GPS diet (inclusion of 100 g kg⁻¹ of grape pomace silage in the diet, DM basis), and GPB diet (inclusion of 100 g kg⁻¹ of grape pomace bran in the diet, DM basis). ^a-b Within a row, means differ (P ≤ 0.05) or tend to differ (0.05 < P ≤ 0.10).

¹Dietary treatment: control diet (traditional confinement diet), GPS diet (inclusion of 100 g kg⁻¹ of grape pomace silage in the diet, DM basis), and GPB diet (inclusion of 100 g kg⁻¹ of grape pomace bran in the diet, DM basis). ^a-b Within a row, means differ (P ≤ 0.05) or tend to differ (0.05 < P ≤ 0.10).

There were diet effects for oxidative and antioxidant system indicators ( Table 4 ). The TBARS and ROS were lower for GPS and GPB treatments than the control (P < 0.05). There was a treatment x day interaction (P = 0.04) for GST activity in meat; LTL from cattle fed GPS and GPB had greater GST activities on day 1 than LTL from cattle fed the control diet. The GST activity tended (P = 0.07) to be greater in LTL from cattle fed GPS and GPB versus LTL from cattle fed the control diet. There was a day effect (P < 0.01) for TBARS, ROS, and GST for all diets, with a linear and increasing trend over days stored for levels of TBARS and ROS. There was a linear and decreasing trend over days stored for GST activity. There were no treatments, treatment x day, or day effects (P > 0.05) for protein thiols.

¹Dietary treatment: control diet (traditional confinement diet), GPS diet (inclusion of 100 g kg⁻¹ of grape pomace silage in the diet, DM basis), and GPB diet (inclusion of 100 g kg⁻¹ of grape pomace bran in the diet, DM basis). ²Thiobarbituric acid reactive substances (nmol MDA/mL); ³Glutathione S-Transferase (µmol CDNB/min/mg protein); ⁴Protein thiols (µmol SH/mg of protein); ⁵Reactive oxygen species (U DCFH/mg protein); ^a-b Within a row, means differ (P ≤ 0.05) or tend to differ (0.05 < P ≤ 0.10).

The bacterial counts are shown in Table 5 . Diets did not influence total mesophiles or lactic acid bacteria counts at various simulated retail exposure periods (P > 0.05). There were diet effects (P < 0.05), but no treatment x day interactions (P > 0.05) for total coliforms (TC) and enterobacteria (ENTB). The LTL from steers fed GPS had lower TC counts than LTL from steers fed the control diet. The TC counts for LTL from steers fed GPB were similar to counts from LTL fro steer fed GPS or control diets. There was a day effect (P < 0.001) for TC, ENTB, and total mesophile counts in LTL from cattkle fed all diets, with an increasing trend over the days of meat exposed to simulated retail conditions. Likewise, there was a day effect (P < 0.001) for lactic acid concentrations, which increased linearly over time.

Small modulations were observed by the inclusion of different experimental diets on the lipid profile in fresh meat at 24 h post-slaughter ( Table 6 ). The GPS and GPB diets increased cholesterol-lowering fatty acids CLFA (cis-oleic, linoleic and α-linolenic fatty acids) concentrations when the control diet (P = 0.05). The LTL from steers fed GPS had lower concentrations of SFA when compared to LTL for steers fed control and GPB diets (P < 0.05). There was a trend for greater amounts of C20:4 n6 in LTL from steers fed GPS and GPB diets than in steers fed the control diet (P = 0.07; Table 6 ). Steers fed BPG tended to have greater amounts of total omega 6 FA when compared to steers fed the control diet (P = 0.08). Steers fed GPS and GPB diets tended to have greater concentrations of C18:1 n9c than steers fed the control diet (P = 0.10). Total omega 9 FA concentrations were greater in LTL from stees fed GPS than steers fed GPB (P = 0.10). There were no diets effects on the amount of total monounsaturated fatty acids (MUFA) and PUFA and for the atherogenic index (P > 0.05).

¹Dietary treatment: control diet (traditional confinement diet), GPS diet (inclusion of 100 g kg⁻¹ of grape pomace silage in the diet, DM basis), and GPB diet (inclusion of 100 g kg⁻¹ of grape pomace bran in the diet, DM basis). ²Values expressed as log CFU/g meat. ^a-b Within a row, means differ (P ≤ 0.05) or tend to differ (0.05 < P ≤ 0.10).

¹Dietary treatment: control diet (traditional confinement diet), GPS diet (inclusion of 100 g kg⁻¹ of grape pomace silage in the diet, DM basis), and GPB diet (inclusion of 100 g kg⁻¹ of grape pomace bran in the diet, DM basis). ²MCFA: midium-chain fatty acids; ³LCFA: long-chain fatty acids. ^a-b Within a row, means differ (P ≤ 0.05) or tend to differ (0.05 < P ≤ 0.10).