DSS solution (3.5%; Sigma Healthcare, Melbourne, Australia) was dissolved in sterile, distilled water and freshly prepared every other day (10). Antibiotic solutions were composed of kanamycin (8 mg/ml), gentamicin (0.7 mg/ml), polymyxin 34,000 U/ml, metronidazole (4.3 mg/ml) and vancomycin (0.9 mg/ml).

A total of 40 adult BALB/c (8-week-old females; 19.20±2 g), 10 C57BL/6 (8-week-old females; 18.8±2 g) and 10 TLR4^−/− mice (8-week-old females; 19±2 g) were purchased from the Animal Experimental Center of Guangdong Academy of Medical Sciences (Guangzhou, China). Animals were maintained under standard conditions, and fed rodent food and water, according to the Guide for the Care and Use of Laboratory Animals (11). The mice were maintained as follows: Temperature, 20–22°C; relative humidity, 55±5%, 12-h light/dark cycle, and ad libitum access to food and water, Mice were fed in cages containing an average of six mice/cage. The protocol was approved by the Review Board of the Institute of Medical Animal Laboratory at the Guangzhou Medical University (Guangzhou, China).

Mice were administered 3.5% DSS in drinking water for 5 days, followed by normal drinking water for 14 days, according to a method described previously by Chen et al (10).

Prior to the experiment, 6-week-old adult mice were depleted from intestinal bacteria by administration of antibiotic solution (100 µl/day/mouse) for 2 weeks. Fresh mice fecal samples were harvested daily and cultured in aerobic and anaerobic conditions. When bacterial growth could not be detected in the culture media, the mice were classified as BD mice.

The mice were randomly divided into 4 groups as follows: i) Group A, normal drinking water; ii) group B, DSS-induced colitis only; iii) group C, DSS-induced colitis in BD mice; and iv) group D, DSS-induced colitis in BD mice treated with E. coli (fed E. coli 1×10⁹ CFU once every other day feeding). The animals received treatments for 14 days. The experiments were carried out using BALB/c mice, and subsequently repeated with C57BL/6 and TLR4^−/− mice.

In the TLR4^−/− experiment, the mice were randomly divided into three groups as follows: i) Control group (Group A1); ii) Group B1, DSS-induced colitis mice in BD C57BL/6 mice treated with E. coli as described previously in Group D; Group C1, DSS-induced colitis mice in BD TLR4^−/− mice treated with E. coli as described previously in Group D.

Animal body weight, stool characteristics and occult blood were recorded and evaluated daily. These parameters were scored by a trained observer blinded to the protocol, in order to calculate disease activity index (DAI), as described in Table I. The mice were sacrificed on day 14 using CO₂, and total colon specimens were collected. The samples were stained with hematoxylin and eosin (H&E), which was used for the pathological score evaluation of the colon tissues. The histological severity of colitis was classified into mucosal damage (D) and extent of disease (E), according to the scoring criteria proposed by Cooper et al (12). The criteria were as follows: D: 0 points, none; 1 point: 1/3 of the crypt near the basement membrane was lost; 2 points, 2/3 of the crypt near the basement membrane was lost; 3 points, all crypts were destroyed, leaving only the surface mucosal epithelium; 4 points: Crypt epithelium was lost. E: 0 points, none; 1 point, focal lesions; 2 points, lesions were present in ~1/3 mucosa; 3 points, lesions were present in 2/3 mucosa; 4 points, lesions were present in all mucosal tissues. The histological score (HS) is equal to the product of D and E, i.e., HS=D × E.

The distal ileum, colon and rectum were removed and cleaned with cold sterile saline. The end of the colon (1 cm from anus) was divided into 0.5 cm sections. The histological specimens were fixed in 10% neutral formalin at room temperature for 12 h, paraffin embedded, sliced and stained with hematoxylin (Mayer) for 5 min and 0.5% eosin for 1–3 min at room temperature (H&E staining). A blinded method was used in order to score the pathological sections according to the severity of colitis, as previously described (13).

Tissue MPO activity was determined using the MPO Peroxidation Assay kit (cat. no. KA1338; Abnova, Taipei, Taiwan), as described previously (14). MPO is an index used to evaluate neutrophil infiltration (15). Colon tissue weights were measured prior to cooling in liquid nitrogen and stored at −80°C. The tissues were homogenized four times (5 sec each) with 10 sec intervals by ultrasonic pulverization (14 kHz) at 4°C, and the supernatant was removed by centrifugation at 14,000 × g for 20 min at 4°C. After 5 min, the absorbance value was measured at 460 nm to determine the MPO activity of each sample.

The application of the standard three-step antibody method was used to detect NF-κB expression levels (16,17). Antibody I [NF-κB p65 (F-6), cat. no. sc-8008; Santa Cruz Biotechnology, Inc., Dallas, TX, USA] was a monoclonal antibody specific for the NF-κΒ p65 B subunit. Antibody II [goat anti-mouse immunoglobulin G (IgG)-B: sc-2039, Santa Cruz Biotechnology, Inc.] was a biotin-conjugated mouse IgG antibody specific for the goat anti-mouse epitope. Antibody I (10 mg/ml) was incubated with the sections for 1 h in a humid chamber at room temperature. Subsequently, Antibody II (2 mg/ml) was incubated with the sections for 1 h in a humid chamber at room temperature. Finally, the sections were incubated with avidin-biotin complex reagent containing horseradish peroxidase (cat. no. 554058; BD Biosciences, Franklin Lakes, NJ, USA) for 30 min at room temperature. NF-κB activation was scored at ×400 magnification using a fluorescence microscope (IX71; Olympus Corporation, Tokyo, Japan) with visualization of red fluorescence in five fields of view and a mean value was calculated. NF-κB was scored according to the percentage of positive cells, as described previously (10): 0 points, 0–1%; 1 point, 2–5%; 2 points, 6–10%; 3 points, 11–25%.

RNA was extracted using MiniBEST Universal RNA Extraction kit (cat. no. 9767; Takara Biotechnology Co., Ltd., Dalian, China), SuperScript™ IV First Chain Synthesis system (cat. no. 18091050; Thermo Fisher Scientific, Inc., Waltham, MA, USA) was used for RT, and TaqMan™ Fast Advanced Master Mix (cat. no. 4444557; Thermo Fisher Scientific, Inc.) was used for PCR. Total RNA was isolated from colon tissues and qPCR was performed as previously described (18,19). Gene expression was normalized to GAPDH using the 2^−ΔΔCq method (20). The primers used were as follows: GAPDH (253 bp), 5′-ACAGCAACAGGGTGGTGGAC-3′ (forward) and 5′-TTTGAGGGTGCAGCGAACTT-3′ (reverse); TLR4 (239 bp), 5′-CCAGAGCCGTTGGTGTATCT-3′ (forward) and 5′-TCAAGGCTTTTCCATCCAAC-3′ (reverse); and NF-κB p65 (251 bp), 5′-GGCAGCACTCCTTATCAACC-3′ (forward) and 5′-GAGGTGTCGTCCCATCGTAG-3′ (reverse). The data were representative of at least three independent experiments.

The results were analyzed using GraphPad statistical software (v.5.0; GraphPad Software, Inc., La Jolla, CA, USA). Experiments were repeated three times. The results were presented as the mean ± standard deviation. One-way analysis of variance followed by Bonferroni's correction was used for comparisons between groups. Kruskal-Wallis test and Steel-Dwass test were performed to compare histological scores among the test groups. P<0.05 was considered to indicate a statistically significant difference.

The body weight of mice decreased following ingestion of DSS. When DSS was ceased after 5 days, the weight of the mice slightly decreased until day 6. In the BALB/c experiment, the non-E. coli BD treatment group mice continued to decrease in weight after DSS treatment was ceased. The body weight of the mice in the E. coli treatment group declined slightly (Fig. 1A). In the experiment using TLR4^−/− mice, the E. coli treatment BD TLR4^−/− group retained the weight decrease. The body weight of the E. coli-treated DSS-induced BD C57BL/6 mice declined slightly, but increased following the cessation of DSS treatment (Fig. 1B).

In BALB/c mice, the survival rate at the end of the experiment was 100% for groups A, B and D. Group C exhibited a survival rate of 60% (Fig. 2A), suggesting that bacterial depletion followed by DSS-induced colitis resulted in increased mortality. In the experiment using C57BL/6 and TLR4^−/− mice, Groups A1 and B1 had a survival rate of 100%, whereas C1 had a survival rate of 60%, suggesting that treatment with E. coli could promote recovery in BD wild-type mice, but not in TLR4^−/− mice. This finding demonstrated that E. coli may have promoted recovery through the TLR4 pathway (Fig. 2B).

DAI was used to assess the severity of colitis in the DSS-induced colitis mice. All mice with DSS-induced colitis underwent fecal occult blood tests from day 3 of DSS ingestion. In BALB/c mice, Groups B and D had significantly lower DAI scores compared with Group C (P<0.05; Fig. 3A). In the TLR4^−/− experiment, the DAI score of groups A1 and B1 declined slowly from day 10 of the experiment. However, the clinical score of the TLR4^−/− group exhibited a continuous increase (Fig. 3B).

In BALB/c mice with DSS-induced colitis (Group B), the colon was shorter than that noted in the mice of group A, and showed extensive mucosal and glandular defects with crypt destruction and a large amount of inflammatory cell infiltration (Fig. 4A and B). In Group D, the colonic mucosa defects were partially repaired, compared with group C (Fig. 4C and D). Concomitantly, the crypt damage decreased and inflammatory cell infiltration and depth of inflammation were reduced. There was a significant reduction in tissue damage in group D compared with group C (P=0.03; Fig. 4E).

In the TLR4^−/− experiments (Fig. 5), no signs of inflammatory response were noted in Group A1 (Fig. 5A). Group C1 indicated extensive mucosal and glandular defects, crypt destruction and inflammatory cell infiltration (Fig. 5C). In the E. coli-treated BD C57BL/6 mice with DSS-induced colitis (Group B1), crypt damage decreased, and inflammatory cell infiltration and depth of inflammation were reduced, thus indicating that the colonic mucosal defects were partially repaired compared with in group C1 (Fig. 5B). The differences noted between TLR4^−/− and wild type C57BL/6 group were significant (P<0.05; Fig. 5D).

In wild-type BALB/c mice, the intestinal mucosal MPO activity in Group C was significantly reduced compared with that in Groups B and D (P<0.05). Notably, the intestinal mucosal MPO activity in the E. coli-treated BD TLR4^−/− group was significantly reduced compared with that noted in the E. coli-treated BD C57BL/6 group (P<0.05; Fig. 6).

In BALB/c mice, activated NF-κB was mainly localized in the lamina propria, as demonstrated by brown nuclear staining under high magnification (data not shown). Groups B and D had higher NF-κB expression, whereas group C had a significantly lower expression level of NF-κB, compared with these groups (P<0.05). In addition, the results indicated that the TLR4^−/− group exhibited less NF-κB expression compared with the E. coli-treated BD C57BL/6 group (P<0.05; Fig. 7), suggesting that TLR4 knockout decreased NF-κB cell activation by E. coli.

In control and DSS-induced BD BALB/c mice, the mRNA expression levels of TLR4 and NF-κB was relatively low. Following administration of E. coli, TLR4 and NF-κB mRNA expression levels increased significantly (P<0.05; Fig. 8A).

In the experiment using TLR4^−/− mice, Group C1 exhibited significantly decreased mRNA levels of TLR4 and NF-κB compared with group B1 (P<0.05; Fig. 8B).