All experimental procedures were reviewed and approved by the Sixth Medical Center of the Chinese PLA General Hospital Animal Ethics Board (Beijing, China).

A modified intraspinal compression spring was designed according to the anatomy measurements of the male Sprague Dawley (SD) rat lumbar spine (Fig. 1). The modified intraspinal compression spring can provide persistent compressive stress to the lumbar facet joints. The spring was manufactured by the Yongxing Kangli Hardware Factory (Wenzhou, China) with the following dimensions: i) External diameter, 2 mm; ii) length, 2 mm; iii) wire diameter, 0.4 mm; and iv) 2 mm accessory length. The string was made of 316L stainless steel, which can offer 50 g in force when stretched to 4 mm. The force of the spring was set according to the biomechanics of the facet joint in humans. Generally, the bilateral facet joints can bear ~20% of the body weight (15). Therefore, for a rat weighing ~250 g, the application of 50 g in pressure can simulate the facet stress that is typically observed in humans (15).

A total of 40 healthy male SD rats (aged 6-8 weeks; weight, 255-320 g) were provided by the Experimental Animal Center of the Sixth Medical Center of Chinese PLA General Hospital [Animal production license number: SCXK (Beijing) 2015-0012]. Animals were kept in cages with the environmental temperature, maintained at a temperature of 22-26˚C and a relative humidity at 40-80% under a 12 h light-dark cycle, and allowed free access to food and water. Routine cleaning and disinfection were performed daily. For each rat, one of the L4/5 and L5/6 spinal fragments was selected randomly using a random number Table marked as Compression level for persistent compressive injury. The other level of the same animal was indicated as Control level treated with sham operation. All rats were anesthetized by an intraperitoneal injection of ketamine (5 mg/100 g), xylazine (0.5 mg/100 g) and acepromazine (0.1 mg/100 g). Following skin preparation and disinfection, a posterior midline incision was performed, where the multifidus muscle at the L4-L6 spinous process was resected to expose the L4-L6 spinal processes. The modified intraspinal compression springs were implanted into the defined compression level by attachment to the spinous processes (Fig. 2). The control level in the same animal underwent the same incision and exposure as that of compression level without implantation of the compression spring. Subsequently, the muscle was sutured and the skin was then closed at both compression and control levels. The animals were treated with subcutaneous injections of antibiotics (penicillin, 80,000 Units) once per day per rat for 3 days. Following surgery, the rats were allowed to recover and were subsequently examined on days 7, 14, 28, 42 and 56 as described below.

On days 7, 14, 28, 42 and 56 following surgery, 8 rats were randomly euthanized by an intraperitoneal overdose injection of pentobarbital sodium (200 mg/kg). Anteroposterior (AP) and lateral X-ray examinations of the rats were performed to verify the spring location prior to lumbar vertebral segment collection. The L4-L6 spinal sample was harvested en bloc after being sacrificed via cervical dislocation. The pathological defects in the L4/5 and L5/6 lumbar facet joints were evaluated by macroscopic observation. A macroscopic tissue observation scoring system proposed by Pelletier et al (16) was used, with a score of 0-4 corresponding to the following conditions: i) 0, intact articular surface with normal calamine blue color; ii) 1, rough articular surface with a gray color; iii) 2, defected superficial cartilage; iv) 3, defected deep cartilage with ulceration; and v) 4, exfoliated cartilage with exposed subchondral bone.

Following macroscopic observation, the L4-L6 segments were fixed in 4% formaldehyde for 48 h at room temperature, followed by decalcification in EDTA solution for 4-8 weeks. The decalcified facet joints were paraffin-embedded and subsequently sliced into 4-6 µm thick sections in the transverse plane. Serial facet joint sections of the experimental and control levels were deparaffinized using xylene followed by a descending series of ethanol at room temperature. The sections were stained with toluidine blue at 20˚C for 10 min. The results were detected using a light microscope (Olympus Corporation) at a magnification of x100, and three fields of view were scored for articular cartilage degeneration according to the Osteoarthritis Research Society International (OARSI) score (ranging from 0 to 24 points) (17).

The aforementioned deparaffinized facet joint sections were incubated with rabbit-anti rat antigen monoclonal antibodies targeting interleukin (IL)-1β (1:100; cat. no. ab9722; Abcam) and tumor necrosis factor (TNF)-α (1:100; cat. no. ab6671; Abcam) at 4˚C overnight. Subsequently, the sections were further incubated with a biotin-labeled goat-anti rabbit secondary antibody (1:200; cat. no. 074-1506; KPL, Inc.) for 60 min at room temperature. The staining was made visible using 3,3'-diaminobenzidine, where cells exhibiting positive staining are brown in color. Staining intensity and the percentage of positive cells were evaluated from three random fields for each section under a light microscope (Olympus Corporation; magnification, x100). A computer interface (Image-Pro Plus; version 6.0; Media Cybernetics Inc.) was used for image analysis. The relative expression of the targeted molecules was calculated by multiplying the staining intensity score with the positive cell score. Staining intensity was scored from 0 to 3, representing no staining, light, moderate and strong staining, respectively. Positive cells (the ratio of brown cells to total cells) were scored from 1 to 4 as follows: i) <10%; ii) 11-50%; iii) 51-75%; and iv) >75%, respectively.

Continuous variables are presented as the mean ± SD. Comparisons made for non-parametric data, between the control and treated groups at different time points were performed using Wilcoxon signed rank sum tests with Bonferroni's adjustments and parametric data comparisons were performed using the paired Student's t-test. All statistical analyses were performed using the SPSS 17.0 software (SPSS, Inc.). P<0.05 was considered to indicate a statistically significant difference.

AP and lateral X-ray examinations were performed on all rats at the indicated days prior to sacrifice. All springs were found to be in the correct position without any dislocation (Fig. 3). The facet joint articular cartilage in the control group exhibited a bright calamine blue color without cracks or defects at 7, 14, 28, 42 and 56 days following surgery. Compared with the control group, time-dependent pathological changes were observed in the experimental group. In addition, joint synovial membranes were slightly swollen at 7 and 14 days following surgery. Marked joint swelling and a gray, dim articular cartilage surface was observed at 28 days, whereas at 42 and 56 days following surgery, a rough cartilage surface and minor cracks were observed, respectively (data not shown).

The macroscopic scores of the facet joint cartilage at each measured day following surgery are presented in Fig. 4. Comparisons between the control and the compression groups were performed at the indicated days. Statistically significant differences were observed on days 14, 28, 42 and 56 following surgery (P<0.05 for day 14 and P<0.01 for all other cases). However, significant differences were not observed at 7 days following surgery (P>0.05). The results suggest that after 7 days, the macroscopic scores of the facet joint cartilage in the compression levels were significantly higher compared with those in the control levels throughout the testing period (P<0.05).

Toluidine blue staining of the articular cartilage indicated smooth and normal surfaces without cracks, erosions or ulcerations in the control level. A smooth cartilage surface with normal cartilage thickness and regular chondrocyte arrangement were observed in control level at all time points selected (7 and 56 days following surgery) as shown in Fig. 5A and B. There was almost no difference between the two time points either in quantity or in the size of chondrocytes. Compared with the control level, the morphology of articular cartilage of the compression level was altered over time. At 7 days following surgery, the facet joint space narrowed (Fig. 5C), whereas at 14 days the cartilage surface became rough, with lighter staining on the surface (Fig. 5D). At 28 days, the number of chondrocytes in the surface layer were reduced, with a rougher cartilage surface and reduced thickness (Fig. 5E). At 42 days, the number of chondrocytes was further reduced and a further reduction in the thickness of the cartilage surface was observed (Fig. 5F). At 56 days following surgery, distinct roughness was identified on the articular surface as well as a decrease in thickness (Fig. 5G).

Articular cartilage degeneration in the control and compression groups was evaluated according to the OARSI scoring system. The OARSI scoring results are presented in Fig. 6. Comparison between the control and compression groups was performed at the indicated days. Significant differences were observed at all the observed time points (P<0.05 on day 14 and P<0.01 for all other cases) apart from day 7 (P>0.05). These results demonstrated that facet joint articular cartilage degeneration in the compression level was significantly worse compared with that in the control level at all the observed time points, except for the first measured postoperative time point on day 7.

The immunohistochemical staining observations of IL-1β and TNF-α levels in the control group were almost the same in the indicated time points. Mild IL-1β and TNF-α staining was observed in the cartilage matrix (Figs. 7A and B as well as 8A and B). In the cartilage of the facet joint of the compression group, the percentage of IL-1β and TNF-α-positive chondrocytes gradually increased from 7 to 56 days following surgery (Figs. 7C-G and 8C-G). At 56 day after surgery, a high number of IL-1β and TNF-α-positive chondrocytes was observed throughout the cartilage. Semi-quantitative analysis of immunohistochemical staining showed that expression of both IL-1β and TNF-α in the cartilage were significantly increased in the compression group in comparison with the control group at every time point (Tables I and II; P<0.01).