Forty entire male crossbreed pigs (26.2±1.2 kg body weight [BW]) from a commercial line hybrid (Landrace × Large White) were used. Pigs were allotted randomly to one of five dietary treatments in a randomized complete block design based on the initial BW of pigs, with eight replicate pens and one animal per pen as experimental unit.

At the beginning of the experiment, the animals were weighed and housed in a masonry facility with a ceramic roof equipped with side curtains and a skylight system. Pens (2.7 m²) had masonry and metal fence and were equipped with feeders and suspended nipple drinkers. All the pens had metal chains and plastic bottles hanging from the pen wall.

All diets were corn and soybean meal-based with industrial aminoacids (AA), minerals, and vitamins. Diets were formulated to meet nutritional requirements according to Rostagno et al. (2017), except for CP content in negative control (Table 1), and offered as mash. Dietary treatments were as follows: NC - negative control (2 and 1% reduction of CP in grower-finisher phases, respectively, no protease supplementation); NC150 - NC + 150 mg protease kg⁻¹ diet; NC300 - NC + 300 mg protease kg⁻¹ diet; PC - positive control (no CP reduction and protease supplementation); and PC300 - PC + 300 mg protease kg⁻¹ diet. The doses of the enzyme tested were based on recommendations of the company.

The experiment was divided into four experimental phases: grower I: entire male pigs, from 25 to 50 kg (d 0 to 26); grower II: entire male pigs, from 50 to 70 kg (d 26 to 44); finisher I: entire male pigs, from 70 to 100 kg (d 44 to 69); and finisher II: immunocastrated male pigs, from 100 to 120 kg (d 69 to 87). Animals were administered two doses of anti-GnRF (Vivax^®): grower II (d 40) and finisher I (d 65).

Exogenous protease supplemented to diets (alkaline protease EC Code: 3.4.21.14) was obtained from Bacillus licheniformis and had a catalytic activity of 200,000 units (U) g⁻¹, in which U is the amount of protease that releases 1 μg of tyrosine min⁻¹ from casein at 40 °C and pH 10.5, with pH and temperature from 6.0 to 12.0 (optimum 11.0) and from 30 to 65 °C (ideal 55 °C), respectively.

Animals had free access to diet and water throughout the experiment. Offered diets and leftovers were recorded daily (UL-50 digital scale, DIGI-TRON, Curitiba, Brazil) throughout the experiment to determine the average daily feed intake (ADFI, g d ⁻¹). Pigs were weighed at the beginning and at the end of each phase using a two-bar digital scale (ULB-3000, IWM bivolt, Curitiba, Brazil). Body weight and feed intake were monitored considering four different phases. Initial BW (IBW, kg), final BW (FBW, kg), ADG (g d⁻¹), and gain to feed ratio (G:F, g:g) were determined.

Feces were sampled to determine the ATTD of nutrients and the apparent digestible energy at the end of the grower and finisher II phases. Before feces sampling, the marker acid insoluble ash (AIA, celite^TM) was added to diets (10 g kg⁻¹ diet) and then mixed for 10 min in a y-type mixer, as previously reported by Sakomura and Rostagno (2016). Pigs were fed diets containing the marker for 3 d and then subjected to partial feces sampling for 1 d, as previously reported by Kavanagh et al. (2001).

Feed intake was recorded daily. Feces were sampled for 12 h right after defecation to avoid contamination. Feces samples were pooled, placed in plastic bags, and stored at −20 °C. Then, the samples (140 g) were oven-dried (SF-325 NM, Tecnalbrand, Piracicaba, SP, Brazil) in duplicates at 55 °C for 72 h (Silva and Queiroz, 2009). After drying, the samples were ground in a ball-type mill (Solab, model SL-38; São Paulo, Brazil).

Dry matter (DM), ash, and CP were determined in samples as previously described by Silva and Queiroz (2009). Acid insoluble ash was analyzed as previously described by Van Keulen and Young (1977). Gross energy (GE) was determined using a bomb calorimeter (IKA^®, model C200, Wilmington, NC, USA). The coefficients of total tract apparent digestibility (CTTAD) of DM, OM, CP, and GE, and digestible nutrients and energy were calculated as previously reported by Sakomura and Rostagno (2016).

All animals were subjected to blood sampling (16:00 h) at the end of the grower and finisher II phases. Animals fasted for 8 h (08:00 to 16:00 h), and then blood samples (≅ 10 mL) were withdrawn from the anterior cranial vena cava using 1.20 × 40 mm needles as previously described (Moreno et al., 1997). Blood samples were transferred to three different sterile glass tubes containing heparin (for creatine kinase and calcium analyses), potassium fluoride (for glucose analysis), or no anticoagulant (for total protein, urea, and albumin analyses) and then placed on ice (4 °C). Afterward, blood samples were sent to the laboratory and centrifuged at 3,000 g for 10 min (analog centrifuge 80-2B, Centrilab, São Paulo, Brazil). A 3 mL aliquot of the supernatant was placed in eppendorf tubes and stored at −5 °C.

The following variables were assessed: urea (enzymatic-colorimetric method, Cat. 427), glucose (enzymatic-colorimetric method, Cat. 434), albumin (bromocresol green colorimetric method, Cat. 419), total protein (TP, colorimetric-biuret method, Cat. 418), creatine kinase (UV kinetic method, Cat. 458), and globulins (TP:albumin). All blood analyses were performed within 15 d after sapling using a spectrophotometer (Bel SPECTRO S05, Bel Engineering, Monza, Italy) and commercial kits (Gold Analisa Diagnóstica, Belo Horizonte, Brazil).

At the end of the trial (d 88), pigs were fasted for 11 h and transported (within 4 h) to a commercial abattoir (Medianeira, PR, Brazil), where they rested for a further 8-h period before slaughter. Slaughter was performed according to the National Council for the Control of Animal Experimentation (normative resolution no. 37 of February 15, 2018). All pigs were stunned by electronarcosis (200 volts for 5 s), followed by exsanguination, scalding, and evisceration. Then, the carcasses were divided into halves that were weighted.

All carcass and meat traits analyses were performed and calculated as described by Bridi and Silva (2006). Carcass quantitative traits were measured at the last rib (6 cm from the cutting line) using a Hennessy GP4/BP4 swine carcass typing pistol (Hennessy Garding Systems, Auckland, NZ). Backfat thickness (BFT), loin depth (LD), loin eye muscle area (LEA), carcass weight, muscularity index (kg carcass⁻¹), and lean meat were assessed.

Carcasses were stored at −2 °C for 4 h. Samples (15 cm) of each loin were cut in the last rib area (insertion of the last thoracic to the first lumbar vertebra) for further analyses. Then, the following measurements were taken from the right carcass half: carcass length (straight line from the forward edge of the atlas to the forward edge of the aitch bone) using a measuring tape and the BFT using a pachymeter (6 in, 150 mm MTX, Tools World, São Paulo, SP, Brazil).

Longissimus thoracis temperature was measured after 4 and 24 h post mortem using a digital skewer-type thermometer (Tp101 Xt-1234, São Paulo, Brazil), and pH was measured after 24 h post mortem using a portable digital pHmeter (AK103, Akso Instrumentos de Medição, São Leopoldo, RS, Brazil).

For qualitative analyses, 2.5-cm subsamples from the 15-cm samples were taken to determine intramuscular fat (marbling) and losses (dripping, thawing, and cooking). Longissimus thoracis color was measured 24 h after slaughter, and six luminosity measurements were taken on the muscle surface using a Minolta CR-400 colorimeter device (Konica Minolta's, São Paulo, Brazil). Results were expressed using the CIELAB color system with 8 mm aperture, area illumination (Illuminant C D65), and 0° viewing angle. Color parameters were measured as L^* (luminosity), a^* (red-green component), and b^* (yellow-blue component) and expressed in the CIELAB color system. Then, the saturation (chroma or purity) and the tint (color or hue) in l. thoracis were calculated.

Color and marbling in l. thoracis were measured using Pork Quality Standards Score Table with a six-point color scale (1 = pale pinkish-gray to white; 6 = dark purplish-red) and a 10-point marbling scale (1 = devoid of marbling and 10 = abundant), respectively (NPPC, 1999).

Loin eye muscle area was determined in 2.5-cm samples of l. thoracis using a scanner printer (Officejet 4500 Desktop - G510a, HP, São Paulo, Brazil). A black box was used to block the lighting and obtain the scanned image and an object with a known area. Then, readings and calculation of LEA were performed using a software (ImageJ 1.53e - Java).

The cooked samples of the l. thoracis were used to determine shear force (g.cm⁻²). Then, six cylindrically shaped subsamples (1.2-cm diameter) were taken longitudinally in the direction of the muscle fibers in each sample. Shear force was performed in the meat lab (Medianeira, PR, Brazil) using a texture meter (Stable Micro Systems TA-XT/plus, United Kingdom) equipped with a Warner-Bratzler shear force probe and software (Texture Expert Exponent – Stable Micro Systems, Vienna Court, United Kingdom).

Statistical analyzes were performed using the general linear model procedure of SAS (Statistical Analysis System, University Edition). A standardized residuals analysis was performed before variance (ANOVA) or covariance (ANCOVA) analysis. Outliers were set at residuals ≥ 3. The normality of variable errors was assessed using the Shapiro-Wilk test.

Pen was considered the experimental unit for growth performance data. The following general model was used:

in which Y_ijk = average observation of the dependent variable in each plot measured in the i-th treatment, in the j-th block, and in the k-th replicate; μ = overall average; T_i = fixed effect of treatment (i = 1, 2, 3, 4, and 5); b_j = random effect of block (j = 1 and 2); β = regression coefficient of Y over X; X_ijk = average observation of the covariate (initial BW) in each plot, measured in the i-th treatment, in the j-th block, and in the k-th replicate; X¯… = overall average for the covariate X; ε_ijk = random error of the plot associated with each Y_ijk observation. Other variables were analyzed using the model mentioned above without the covariate effect.

Treatment effects on the dependent variables were analyzed via ANCOVA or ANOVA. Whenever significant (P≤0.05), differences among treatments were compared using multiple contrast test (non-orthogonal contrasts): NC vs NC150, NC vs NC300, NC vs PC, PC vs PC300, and NC + NC150 + NC300 (NC group) vs PC + PC300 (PC group). Data were presented as averages with their pooled SEM.

A treatment effect (P≤0.05) was observed only in grower II pigs (Table 2). Pigs fed NC diet showed (P<0.001) greater ADFI than those fed NC300 or PC diets and lower ADFI than pigs fed NC150 diet. In addition, pigs fed PC diet had lower ADFI (P<0.001) than those fed PC300 diet. Greater ADG (P = 0.039) and G:F (P = 0.001) were observed in pigs fed NC diet compared with pigs fed NC300 or NC150 diets and NC300 diet, respectively. Pigs fed PC diet showed better G:F (P = 0.001) than pigs fed PC300 diet. No treatment effect (P>0.05) was observed for growth performance variables in grower and finisher I phases.

In grower II phase, lower (P≤0.05) CTTAD of DM and OM, and DDM, DOM, and digestible protein (DP) were observed in pigs fed NC diet compared with those fed NC150 or NC300 diets (Table 3). In addition, pigs fed NC diet showed a decrease of 17.6 and 3.6% in DP (P<0.001) compared with those fed PC or NC150 diets, respectively. Positive control treatment increased DP compared with the NC treatment.

In the finisher II phase, pigs fed the NC diet showed lower DDM, DOM, and CTTAD of OM and GE (P≤0.05) than pigs fed NC150 or NC300 diets (Table 3). In addition, pigs fed PC-based diet showed increase of 17.5% in DP (P<0.001) compared with those fed NC diet. Pigs from PC group had greater DP (P<0.001) than pigs from NC group.

No effect (P>0.05) of dietary treatments on blood parameters was observed in grower II phase (Table 4). However, pigs fed PC diet showed an increase of 27.6% in albumin concentration (P = 0.05) compared with pigs fed PC300 diet in the finisher II phase. Additionally, pigs on PC treatment showed greater TP and globulin concentrations (P≤0.05) than those on NC treatment.

Pigs fed NC or PC300-based diets showed greater (P = 0.035) luminosity in the l. thoracis muscle than those pigs on PC diet (Table 5). Furthermore, a greater luminosity was observed in the l. thoracis of pigs from NC group compared with those from the PC group. In addition, a greater color score (P = 0.021) was observed in the l. thoracis muscle in pigs fed PC-based diet compared with pigs fed PC300 diet.

The animals remained healthy throughout the experimental period. Isolated exogenous protease supplementation has been previously reported to be beneficial on growth performance (Zuo et al., 2015; Tactacan et al., 2016; Upadhaya et al., 2016). However, in the present study, the supplemented protease enzyme alone in diets was not effective in improving pig performance. Animals fed NC diets did not show lower performance attributed to positive protein balance when dietary CP was reduced and, hence, N intake and usage are greater because N excretion is reduced, and more proteins are required to support performance (Maestá et al., 2008; Terzis et al., 2010; Monteiro et al., 2018).

Dietary nutritional factors and protein components can affect protease action and impair growth performance responses (Zuo et al., 2015). The lack of response could also be attributed to a fully developed gastrointestinal tract in grower-finisher pigs, which provides a greater capacity to use dietary nutrients efficiently (Nguyen et al., 2019). However, Chen et al. (2017) reported positive effects of protease supplementation on the growth performance of grower pigs due to the nutritional quality of the tested protein source. The authors mentioned above also reported greater feed intake in pigs fed NC diet, which is explained by their intent to meet nutritional requirements when a diet with poor or low-quality ingredients is provided.

Although protease enzymes can improve nutrient digestibility in swine diets, the lack of consistent improvement in growth performance can be attributed to a combination of factors related to diet composition (e.g., formulation, ingredients, protein type) and interaction with other ingredients. Thus, the aforementioned factors can influence the use of nutrients to promote animal growth, as well as the isolated form in which the enzyme is found in the diet. Consequently, we hypothesized that if protease supplementation in diets increases protein breakdown, this may exceed the ability of the digestive system to assimilate and utilize the resulting amino acids. This imbalance could lead to inefficient use of nutrients and, therefore, to no substantial improvement in growth performance.

In the present study, animals supplemented with protease showed greater ATTD of nutrients, as previously reported by Upadhaya et al. (2016). This enzyme can improve the ATTD of nutrients by increasing the usage rate of nutritional compounds and support the digestive system (Nguyen et al., 2019; Park et al., 2020). Indeed, protease degrades nutrients that are resistant to endogenous digestive enzymes in growing-finishing pigs (O'Doherty and Forde, 1999; O'Shea et al., 2014) or neutralize antinutritional factors such as enzyme inhibitors to improve nutrient utilization (Zuo et al., 2015).

Nguyen et al. (2018) observed a lack of effect on ATTD compared with our findings, which could be explained by the inclusion of phytase in the diet. Indeed, diets containing other enzymes may affect the response to protease supplementation (Lee et al., 2018). This hypothesis agrees with Sultan et al. (2010), who reported that when protease is supplied alone, the ileal digestibility of protein increases compared with an enzyme blend supplementation.

Upadhaya et al. (2016) reported greater CTTAD of DM and GE in animals fed a diet containing protease, which agrees with what we observed in the present study, in which NC + protease treatments improved CTTAD of DM and GE compared with the NC diet. However, results on the benefits of protease supplementation are conflicting. A greater ATTD of nutrients is not always followed by an improved growth performance (O'Shea et al., 2014; Nguyen et al., 2019; Perez-Palencia et al., 2021), which corroborates the results we observed. This effect is supported by a previous study (Liu et al., 2013), in which greater starch and CP digestibility were observed, but not an improved growth performance.

Variations in the available substrates contribute to the response of a specific protease (Acamovic, 2001), as well as to the dietary dose (Perez-Palencia et al., 2021) and protease type (Lee et al., 2018; Torres-Pitarch et al., 2019). Hence, these variations can result in an inconsistent action of protease on the growth performance. Indeed, it may affect the growth rate in pigs (Nguyen et al., 2019; Perez-Palencia et al., 2021). The greater ATTD of nutrients we observed did not influence protein availability for growth and/or muscle deposition, which agrees with Selle et al. (2006) and Lei et al. (2017).

Little information is found in the literature regarding studies assessing protease effects on pig blood parameters (Nguyen et al., 2019). As blood is widespread throughout the animal system, any disorder can alter the blood profile (Kohn et al., 2005). However, the pigs and environmental conditions were normal in our study, and therefore, glucose, urea, and creatine kinase concentrations were not altered by alkaline protease supplementation neither by CP content. Tactacan et al. (2016), and Zaworska-Zakrzewska et al. (2022) also observed no changes in blood glucose concentrations in pigs fed diets containing protease. Results indicated no kidney damage or muscle injuries due to treatments, as suggested by creatine kinase concentrations (Kaneko et al., 2008). The blood parameter concentrations we observed in growing-finishing pigs were within the reference range for pigs (Moreno et al., 1997; Klem et al., 2010).

Total protein, albumin, and globulins were analyzed to verify whether changes in CP dietary content and protease supplementation would cause metabolic disorders, liver diseases, and protein losses (Messer, 1995; Kaneko et al., 2008). We observed that animals from NC group showed lower TP concentration, which would negatively affect albumin and globulin production in the liver (Fischer et al., 2000). However, this TP reduction was reflected only in a lower globulin concentration in pigs fed NC diet. This result is related to the fact that changes in blood constituents regarding diet formulation and protease supplementation may not be easily detected under less challenging pig rearing conditions (Min et al., 2019). This case is supported by previous studies (Tactacan et al., 2016; Zaworska-Zakrzewska et al., 2022), in which no changes in TP concentrations, albumin, and globulin in pigs were observed.

Wang et al. (2020) reported greater serum albumin and a trend to greater TP concentrations in young pigs fed diets containing greater CP concentration (3% more). In the present study, when diets had less CP (1%), TP and globulins concentration were reduced in pigs from NC group, suggesting a dietary deficiency, as previously reported by Fischer et al. (2000) and Wang et al. (2020). Indeed, lower dietary CP reduces blood protein concentrations (Diaz González and Silva, 2008).

According to Rotter et al. (1994) and Chen et al. (2008), there is a negative correlation between albumin and globulin concentration, i.e., an increase in globulin concentration due to immune functions inhibits albumin synthesis in the liver as a compensatory mechanism to maintain constant TP concentration. This correlation supports the reduced albumin concentration in pigs fed PC300 diet. However, in cases of liver dysfunction, albumin concentration is lower than that of globulin concentration (Diaz González and Silva, 2008). This effect was observed in pigs from PC group, in which albumin concentrations reflected a change in globulin concentration.

Diaz González and Silva (2008) reported that reduced albumin and urea concentrations suggest a protein deficiency. Although we observed no changes in urea concentration in pigs as previously observed by Tactacan et al. (2016) and Zaworska-Zakrzewska et al. (2022), urea concentration was analyzed as a sensitive and immediate indicator of protein intake. Urea concentration is also a marker of renal function (Kaneko et al., 2008; Upadhaya et al., 2016; Nguyen et al., 2019). Different results from our study were observed by Upadhaya et al. (2016), Nguyen et al. (2018), and Nguyen et al. (2019), who reported that dietary supplementation of protease increased the concentration of blood urea in pigs, differing from the findings evidenced by Zuo et al. (2015).

In the present study, the dietary alkaline protease supplementation did not alter the quantitative and composition traits of carcass and meat in pigs. This result agrees with previous studies (Choe et al., 2017; Min et al., 2019; Perez-Palencia et al., 2021) and is attributed to the lack of response in the growth performance of animals (Perez-Palencia et al., 2021). However, we observed a change in L^* and color score of the l. thoracis muscle in pigs fed PC vs PC300 diets. Regarding L^*, the average values previously reported for pigs ranged from 49.0 to 51.3 (Brewer et al., 2001; Bertol, 2019). Thus, pigs fed PC diet showed lower L^* than pigs fed PC300 diet, resulting in darker meat. Color may vary within the same muscle in a small evaluation space because myoglobin concentration is variable (Lawrie and Ledward, 2006). It explains why color score did not differ among treatments as observed for L^*.

In addition, the pH after slaughter could explain the differences among dietary treatments in L^* because variation in pH_24h is also related to changes in meat color and water retention capacity (Lawrie and Ledward, 2006). However, no differences were observed in water loss from the l. thoracis muscle. With regard to L^*, not only the animals fed protease, but all animals showed an effect for this variable. Hence, more studies are needed to assess the effects of protease on l. thoracis color in pigs.

The effectiveness of dietary proteases was also associated to how the protease is supplied (Ghazi et al., 2003) and whether the enzyme is incubated with protein source or only supplemented to the diet with a post-intake activation (Pan et al., 2016). Altogether, protease supplementation had low effects on diets for growing-finishing pigs, suggesting that greater dosages of the enzyme may be necessary (Perez-Palencia et al., 2021). The increased ATTD may not always benefit other variables. Hence, the beneficial effects of protease in diets for pigs may be directly influenced by dietary, metabolic, and physiological factors (Zuo et al., 2015; Tactacan et al., 2016; Upadhaya et al., 2016; Lee et al., 2018; Perez-Palencia et al., 2021).