The metabolites were produced from five strains of L. plantarum, namely, TL1, RG11 RG14, RS5 and RI11. The L. plantarum strains were isolated from Malaysian food (Foo et al. 2003) and kept at −20°C in Man Rogosa Sharpe broth (MRS; Merck, Darmstadt, Germany) containing 20% (v/v) glycerol. The stock culture was revived twice in MRS broth and incubated anaerobically at 30°C for 48 h. The liquid metabolites were prepared according to the method as described by Foo et al. (2003). Metabolites were collected by separating the producer cell by centrifugation at 12,000 rpm for 15 min at 4°C and stored at 4°C until used. The combinations were mixed by equal volume from each strain before being fed to piglets in their diets.

This experiment was carried out at the Commercial Research Unit in a pig farm, Tanjung Sepat, Selangor, Malaysia. The protocol used in this experiment complied with the guidelines by the ethical committee of University Putra Malaysia. A total of 120 postweaning crossbred male piglets (Large White × Landrace × Duroc) of 20 litters from third to fourth parity of sows, at 26 days of age with an average initial body weight (BW) of 6.32 ± 0.14 kg were used in this study. The piglets were kept in raised slatted floor pens (1.2 × 1.6 m) at temperature ranging from 26°C to 32°C and the humidity was 86 ± 4%. Heater was provided from 20:00 to 08:00 daily to keep the piglets warm. Each pen consisted of four piglets and six replications were used per treatment. Water and feed were offered ad libitum throughout the experimental period. The piglets were randomly assigned into one of five treatments. The feed compositions of dietary experiment are shown in Table 1. The basal control diet was formulated to meet the nutrient requirement of piglets according to the NRC (1998). Five treatments were: (1) −ve control (free antibiotic); (2) + ve control (0.03% antibiotic of chlortetracycline); (3) Com 1 (0.3% metabolite of TL1, RG11 and RI11 strains); (4) Com 2 (0.3% metabolite of TL1, RG14 and RS5 strains); (5) Com 3 (0.3% metabolite of RG11, RG14 and RI11 strains). The experiment was investigated for five consecutive weeks. The BW and feed intake were measured weekly. At the end of experimental period, three piglets with similar BW from each treatment were sacrificed for intestinal tissue and faeces samples. Table 1

Composition of piglet basal diet

The diarrhoea score of each piglet was recorded on days 3, 5, 10, 12, 17 and 24. Diarrhoea was assessed visually from the scale 0 to 3, based on the consistency of the faeces forms according to the following scores: 0 = pellet faeces, 1 = semi-pellet faeces, 2 = soft faeces and 3 = watery faeces. This observation was followed the method that described by Loh et al. (2002).

Faecal samples were collected directly from the rectum of the scarified piglets at slaughter and the pH was measured immediately after the collection. Faecal samples were then stored at –20°C until further analyses for microfloral counts and SCFA.

Ten percent (w/v) of faecal sample was diluted in sterile peptone water as prepared by 25.5 g/L and homogenized to stand at room temperature for 1 h. The samples were diluted (10% v/v) for further tenfold series with peptone water. The dilution was vortexed and spread using a sterile glass spreader. Enumerations of bacteria were performed on MRS agar for LAB and incubated in anaerobic jar at 30°C for 48 h. ENT enumeration were performed on eosin methylene blue (EMB; Merck, Darmstadt, Germany) agar and incubated an anaerobic condition at 37°C for 24 h. Number of colony forming units (CFU) are expressed as log₁₀ CFU per g. Faecal SCFA contents were determined using gas chromatography as described by Thanh et al. (2009).

The samples from the middle of duodenum, jejunum and ileum of small intestine of each segment were sectioned 6–8 cm, flushed and kept in neutral buffered formalin solution until further morphometric analysis. The samples were then cut into two parts from each segment for cross and length section of intestine surface. Each segment was fixed in 10% formalin. Intestinal samples were excised and dehydrated for 16 h in a tissue processing machine (Leica ASP 3000, Japan) and then embedded in paraffin wax (Leica EG 1160, Japan). Each section was cut 4 μm thick (Leica RM 2155, Japan) and fixed on to the glass slides, heated at 57°C until samples were dried. The samples were stained with haematoxylin and eosin, mounted with coverslips. The villus height was measured as the distance between the crypt mouth and the tip of villi. The crypt depth was measured as the distance between the basement membrane and the mouth of crypt as using microscopes and read by PC Life Science Olympus software (Olympus Soft Imaging Solutions). This procedure was based on the method described by Bejo (1990).

Data were analysed using one-way analysis of variance. The effects of diet treatments on different animals and faecal parameters were tested using the General Linear Model procedure of SAS 1998 (SAS Inst., Inc., Cary, NC). Initial BW was used as a covariate in analyses of piglet performances. As for faecal consistency analyses, the pen was used as main effect of sampling day in the model. Duncan’s multiple range test system was used to compare the significant difference of treatments at P < 0.05. The data were presented as the mean ± SEM.

Initial BW was not significantly different (P > 0.05) among the treatment groups. However, the final BW for the −ve control was significantly lower (P < 0.05) than Com 2 and the + ve control (Table 2). No significant differences (P > 0.05) were found between the + ve control group and metabolite treatment groups. The average daily gain (ADG) for the −ve control group was significantly poorer (P < 0.05) than other treatments except for the Com 3. The average daily feed intake was not significantly different (P > 0.05) among the treatments during the 5 weeks of the experiment trial. The FCR was the highest (P < 0.05) in the −ve control group and no significant difference (P > 0.05) was found between the + ve control and metabolite treatment groups. Table 2

Growth performance and diarrhoea score of piglets fed with different dietary treatments

The piglets fed with Com 2 and Com 3 treatments demonstrated significant lower (P < 0.05) diarrhoea scores compared to the −ve control and Com 1 groups. However, there were no significant differences (P > 0.05) between Com 2, Com 3 and the + ve control group. No diarrhoea incidence of piglets was observed after 24 days of posweaning (Fig. 1). Fig. 1

Diarrhoea scores in piglets fed with different dietary treatments

The Com 2 group had significantly lower (P < 0.05) faecal pH than the −ve control group. No significant differences (P > 0.05) were found between metabolite combination groups (Table 3). Faecal LAB counts of metabolite treatment groups were significantly higher (P < 0.05) than those without metabolite combinations (Table 3). No significant differences (P > 0.05) were found for LAB counts in Com 1, Com 2 and Com 3 metabolite treatment groups. However, the Com 2 treatment showed a significant lower (P < 0.05) faecal ENT counts when compared to the −ve control, Com 1 and Com 3 treatment groups. Table 3

Faecal pH, LAB, ENT population and SCFA concentration in piglets fed with different dietary treatments

The main composition of faecal SCFA was acetic acid and there was no significant difference (P > 0.05) among the treatment groups in the current experiment (Table 3). The concentration of propionic acid in Com 2 group was significantly higher (P < 0.05) than those presented in −ve and + ve control groups. In addition, the propionic acid in Com 1 group was also significantly higher (P < 0.05) than those obtained in the −ve control group. Butyric acid in Com 2 treatment was the highest (P < 0.05) among all the treatment groups. Isovaleric acid in Com 1 treatment was significantly higher (P > 0.05) as compared to control group. There was no significant difference (P > 0.05) for isobutyric and valeric acids in the piglet faeces among treatment groups. Total faecal SCFA in the Com 2 treatment was significantly higher (P < 0.05) than those of the −ve control group but did not present any significant differences (P > 0.05) between the antibiotic group and metabolite treatment groups.

The duodenal villus height was significantly higher (P < 0.05) in the + ve control and Com 2 treatment groups compared to the −ve control group. However, no significant difference (P > 0.05) between treatment groups was observed in the jejunum and ileum villus height and for the intestinal crypt depth in the intestine of piglets (Table 4). Table 4

Villus height and crypt depth in small intestine of piglets fed with different dietary treatments