Two strains of Lactobacillus spp. (L. johnsonii x-1d-2 and L. mucosae x-4w-1) were obtained from Livestock Research Institute, Council of Agriculture, Executive Yuan (Tainan, Taiwan). L. johnsonii x-1d-2 and L. mucosae x-4w-1 were originally isolated from feces of Landrace×Duroc crossbred piglets at 1 d and 4 wks, respectively. These strains have been previously identified by 16S rDNA sequence analysis and found to be tolerant to pH 3.0 and 0.2% (wt/vol) bile salts in the preliminary tests (M. J. Chen, unpubl. data). Before the experiments, each strain was inoculated (1.0%, vol/vol) and subcultured twice in lactobacilli de Man, Rogosa and Sharpe (MRS) broth (BD Difco, Sparks, MD, USA) at 37°C for 16 h.

Soybean meal (Central Union Oil Corp., Taichung, Taiwan), molasses (Taiwan Sugar Corp., Tainan, Taiwan) and sodium acetate (Echo Chemical Co., Ltd., Miaoli, Taiwan) were used as the main variables influencing the growth of lactobacilli in this study. A Box-Behnken design (BBD) (Box and Behnken, 1960) for these three variables with different coded levels (low, middle and high) was used to determine the optimized conditions for production of L. johnsonii x-1d-2 and L. mucosae x-4w-1. The growth of lactobacilli strains was measured in duplicate under different levels of the three variables after incubation at 37°C for 16 h. Seventeen combinations and the average values of the observed responses are shown in Table 1. The responses, as linear, two-factor interaction (2FI), quadratic and cubic models with the variables, were tested for adequacy and fitness using model analysis and lack-of-fit test. The model analysis compared the validities of various models for the response according to their F-values and a model with p-values (p>F) below 0.05 was regarded as significant. The highest-order significant polynomial regression model was selected. The lack-of-fit test, measuring the fitness of the model obtained, compared the residual and pure errors at replicated design points. A model with no significant lack-of-fit was selected. Based on BBD, the polynomial model with undetermined coefficients for predicting the optimal condition is shown as follows:

where Y is the predicted response, X₁, X₂, and X₃ are the independent variables, β₀ is the intercept, β₁, β₂, and β₃ represent linear coefficients, β₁₂, β₁₃, and β₂₃ represent interactive coefficients, β₁₁, β₂₂, and β₃₃ represent quadratic coefficients. The experimental results were analyzed using Matlab software (version 5.3, The Math Works Inc., Natick, MA, USA). Response surface plots were drawn to represent the interaction effects of the variables in the experiments. To search the optimal ratio of culture medium, an optimization program consisting of a multi-start SQP technique was coded to search for the global optimum. The program generates a series of uniformly distributed random points, with each point representing an initial design of the ratio of medial ingredients. Then SQP is applied to find the optimum based on each initial design. If the probability of the optimum being the global one exceeds a preset value (99.99% in this study), the global optimum is considered found. Otherwise, the next random initial design is generated and the SQP re-executed. Additionally, verification experiments were also conducted in triplicate under the optimal condition of the variables to confirm the accuracy of the model in this study. The cell numbers of lactobacilli were determined by pour plating (1.0 mL) of diluted cells on lactobacilli MRS agar (BD Difco, USA). Viable colonies were counted after 24 h of incubation at 37°C.

L. johnsonii x-1d-2 and L. mucosae x-4w-1 were grown at 37°C for 16 h in the estimated optimal medium (5.5% (wt/vol) soybean meal, 1.0% (wt/vol) molasses and 1.0% (wt/vol) sodium acetate), respectively. Biomasses from optimal medium cultures were centrifuged at 3,000×g for 10 min and freeze-dried under vacuum at 6.7 Pa for 24 h in a freeze-drier (Model 4.5, Labconco Crop., Kansas, MO, USA). The freeze-dried cultures of lactobacilli, which corresponded to ca. 10⁸ CFU/g, were mixed with feed administered to animals during the feeding trial.

A total of 24 Landrace×Duroc crossbred piglets (12 males and 12 females) weaned at 21±3 d of age with an average weight of 6.77±0.37 kg, obtained from Livestock Research Institute, Council of Agriculture, Executive Yuan (Tainan, Taiwan), were used in the experiments. The piglets were randomly distributed into 4 groups (6 piglets/group): i) control group (piglets fed basal diet); ii) J group (piglets fed basal diet supplemented with freeze-dried L. johnsonii x-1d-2, 10⁸ CFU/g per piglet per d); iii) M group (piglets fed basal diet supplemented with freeze-dried L. mucosae x-4w-1, 10⁸ CFU/g per piglet per d); and iv) JM group (piglets fed basal diet supplemented with freeze-dried L. johnsonii x-1d-2, 5×10⁷ CFU/g and L. mucosae x-4w-1, 5×10⁷ CFU/g per piglet per d). The basal diet consisted of yellow corn (67.7%, wt/wt), soybean meal (19.0%, wt/wt), fish meal (5.0%, wt/wt), skim milk powder (2.0% wt/wt), whey power (2.0%, wt/wt), dicalcium phosphate (1.6%, wt/wt), soybean oil (1.0%, wt/wt), calcium carbonate (0.8% wt/wt), salt (0.5%, wt/wt), mineral premix (0.15%, wt/wt), vitamin premix (0.1%, wt/wt), choline (0.1%, wt/wt) and copper sulphate (0.04%, wt/wt). Each group was separately housed in an individual cage and randomly fed basal diet or diet supplemented with freeze-dried probiotic strains for 21 d at ambient room temperatures (22 to 27°C) and 50 to 60% relative humidity. Body weight gain (BWG), feed intake (FI), gain/feed ratio, and lactobacilli and E. coli counts in the feces were measured for each piglet after different periods of the feeding trial, respectively. Animal experiment followed the Guide for the Care and Use of Laboratory Animals in Taiwan.

The fresh feces were taken from the rectum of weaned piglets with a sterile cotton swab. Approximately 1.0 g of fecal sample was serially diluted in 0.85% (wt/vol) saline and then pour plated (1.0 mL) on Lactobacillus selective Agar Base (Acumedia, Baltimore, MD, USA) and CHROMagar ECC (CHROMagar Microbiology, Paris, France) for lactobacilli and E. coli, respectively. The numbers of viable colonies on triplicate plates were counted after incubation at 37°C for 48 h.

All replicate experiments were executed independently. The data were analyzed with the general linear model procedures of SAS statistical software (SAS Institute, Cary, NC, USA). Significant differences (p<0.05) between treatment means were determined using Tukey’s pair-wise comparison test (except RSM study).

In the preliminary experiment, several nitrogen sources (brewer’s yeast, sugar yeast, soybean meal and whey protein) and carbon sources (glucose, lactose, sucrose, and molasses) were tested. We found that soybean meal and molasses were the most economical and effective ingredients for the growth of L. johnsonii x-1d-2 and L. mucosae x-4w-1 compared with other nitrogen and carbon sources (data not shown). Moreover, the addition of sodium acetate into the medium also promoted the growth of the two probiotic strains (data not shown). Considering the material cost and the growth of lactobacilli, soybean meal, molasses and sodium acetate were used for components to develop a low-cost culture medium for mass production of L. johnsonii x-1d-2 and L. mucosae x-4w-1 in this study.

The results for the growth of L. johnsonii x-1d-2 and L. mucosae x-4w-1 were used to build the response surface model according to the BBD presented in Table 1. The analysis of variance for the BBD in Table 2 showed that the quadratic and 2FI models, suggested by model analysis and lack-of-fit test, were well suited to represent the experimental data of L. johnsonii x-1d-2 and L. mucosae x-4w-1 with determination coefficients (R²) of 0.92 and 0.94, respectively. The polynomial regression models obtained for L. johnsonii x-1d-2 and L. mucosae x-4w-1 were as follows:

To depict the optimization results, a 3-D response surface plot was generated by fixing two of the three variables as shown in Figure 1. The SQP was applied to search the optimal ratio of growth medium for maximum cultivation of L. johnsonii x-1d-2 and L. mucosae x-4w-1. The greatest population of L. johnsonii x-1d-2 (8.90 log CFU/mL) and L. mucosae x-4w-1 (9.30 log CFU/mL) was attained with the optimal points X₁ (soybean meal) = 5.5% (wt/vol), X₂ (molasses) = 1.0% (wt/vol) and X₃ (sodium acetate) = 1.0% (wt/vol). The optimal medium ratio for maximum production of L. johnsonii x-1d-2 and L. mucosae x-4w-1 derived from the SQP scheme was further verified by an additional experiment. In this study, the experimental values were very close to the predicted values (Table 3). Additionally, the bacterial numbers of the two probiotic strains cultured in the optimal medium were similar to those grew in MRS broth after incubation for 16 h (data not shown). These results indicated that the growth of L. johnsonii x-1d-2 and L. mucosae x-4w-1 can be effectively increased using the optimal culture medium. Similar studies on the application of RSM reported that optimization of medium composition improved the productivity of probiotic L. fermentum and L. salivarius (Lim et al., 2007; Gao et al., 2009). However, to the best of our knowledge, this study is the first report on the optimization of growth medium for cultivation of L. johnsonii and L. mucosae using RSM and SQP.

Soybean meal and molasses are by-products of soybean industry and sugar manufacturing process, respectively. Soybean meal is an important source of protein widely used in farm animal diets (Cromwell et al., 1999). Its mean value of protein is above 45.0% (wt/wt). Molasses contains significant amounts of sucrose, glucose and fructose with several essential nutrients such as vitamins, minerals and other organic compounds (Makkar and Cameotra, 2002). Recent studies have shown that inexpensive soybean and molasses can be used as primary nitrogen and carbon sources to replace the costly yeast extract and other conventional sugars for the production of some microorganisms, including Bacillus and Lactobacillus spp. (Saha, 2006; Tari et al., 2006). Besides, sodium acetate, a basic component of commercial MRS medium, is known as an energy source as well as selective agent for lactobacilli (Stiles et al., 2002). After calculation, the cost of the optimal medium (ca. USD $0.3/L) was 96% lower than the commercial MRS medium (ca. USD $8/L). Once developing a low-cost culture medium, we massed production of two probiotic strains, L. johnsonii x-1d-2 and L. mucosae x-4w-1, and made the probiotic powders by freeze drying for the following animal study.

The effect of probiotic feeding on the growth performance of weaned piglets is presented in Table 4. All piglets kept healthy and no deaths occurred during the feeding trials. The initial body weight (BW) of piglets showed no difference (p>0.05) among all groups. After 21-d probiotic feeding, the final BW was significantly higher (p<0.05) in piglets fed basal diet supplemented with freeze-dried probiotic cultures (L. johnsonii x-1d-2 (J) group, L. mucosae x-4w-1 (M) group and mixed-strain (JM) group) than in piglets fed basal diet (control group). The BWG values were significantly increased (p<0.05) for feeding with L. mucosae x-4w-1 (M group), followed by JM, J, and control groups. Higher FI was observed in piglets of J, M, and JM groups than in control group. Addotionally, the probiotic treated groups showed a significantly higher (p< 0.05) gain/feed ratio than the control group. The piglets fed with L. mucosae x-4w-1 had the best feed efficiency. These results showed that addition of L. johnsonii x-1d-2 and L. mucosae x-4w-1 to the basal diet improved the growth performance of weaned piglets.

Similar studies have used Lactobacillus spp., including L. reuteri (Chang et al., 2001), L. casei subsp. casei (Guerra et al., 2007), L. fermentum (Yu et al., 2008) and L. sobrius (Konstantinov et al., 2008) to improve the growth performance of piglets. In this study, L. johnsonii x-1d-2 and L. mucosae x-4w-1, originally isolated from piglet feces, are believed to be able to colonize and survive in the intestine to exert the desirable effects. Van Tassell and Miller (2011) reported that the probiotic properties of lactobacilli are mainly from adhesion to the gastrointestinal tract, competition exclusion against pathogens and production of antimicrobial compounds. The features of L. johnsonii x-1d-2 and L. mucosae x-4w-1 might provide a beneficial effect on the maintenance of intestinal microflora balance in weaned piglets.

The data on viable counts of lactobacilli and E. coli in the feces of weaned piglets during 21-d feeding are shown in Table 5. The viable counts of lactobacilli in the feces of probiotic (J, M, and JM) groups increased as the feeding period was extended, whereas lactobacilli counts remain constant in non-probiotic (control) group during feeding. After a 21-d feeding period, piglet feces from probiotic groups contained significantly higher (p<0.05) lactobacilli counts compared with control group. The greatest lactobacilli numbers were observed in M group, followed by JM and J groups, with a bacterial population of 8.25, 8.13, and 7.76 log CFU/g, respectively. On the other hand, during the 21-d feeding, a reduction of 0.56, 0.23, and 0.47 log CFU/g in E. coli counts was observed in J, M, and JM groups, respectively. At the end of feeding, the numbers of E. coli were significantly lower (p<0.05) in probiotic groups than in control group. This study showed that probiotic feeding (L. johnsonii x-1d-2 and L. mucosae x-4w-1) could decrease the E. coli population and increase the lactobacilli population in the intestinal tract of weaned piglets.

The results suggested that the improvement of weaned piglet growth performance might likely be the effect of L. johnsonii x-1d-2 and L. mucosae x-4w-1 on the intestinal microbial ecosystem. It is worth noting that weaned piglets fed with L. mucosae x-4w-1 showed the best growth performance and greatest number of lactobacilli among the probiotic treatments. Furthermore, indigenous lactobacilli have been implicated in the suppression of intestinal colonization potential of E. coli (Blomberg et al., 1993). Being the most common bacterial microflora of the gut, E. coli is one of major reason causing piglet diarrhea, which could affect the intestinal barrier and cause disturbed integrity (Liu et al., 2008). This indicates that high concentration of lactobacilli and reduced population of E. coli in the intestinal tract might decrease the incidence and score of diarrhea.

The adhesion of probiotic lactobacilli to intestinal mucus makes them colonize the gut efficiently and settle down to a stable population, that may provide health benefit and avoid infectious disease for piglet (Li et al., 2008). In addition, a number of studies have identified various cell-surface proteins such as mucus-binding proteins, adhesion-like proteins, fimbrial subunits, elongation factor, and heat shock GroEL protein in L. johnsonii and L. mucosae. These proteins were stated to be involved in bacterial adhesion to intestinal mucosa, protection against pathogen colonization, and immune system stimulation (Roos et al., 2000; Granato et al., 2004; Pridmore et al., 2004; Bergonzelli et al., 2006; Watanabe et al., 2010). The possible mechanisms of improvement of piglet growth performance in this study may be related to the adhesive, antimicrobial and immunomodulatory activities of L. johnsonii x-1d-2 and L. mucosae x-4w-1. Therefore, we suggest that these two probiotic strains have high potential to be used as feed additives for weaned piglets.