Celebrex (CAS: 169590-42-5), chitosan (degree of deacetylation: 95%; molecular weight: 20 w; CAS: 9012-76-4), and β-glycerophosphate (β-GP) (CAS: 154804-51-0) were all purchased from Sigma-Aldrich (St. Louis, United States).

The 2, 3, and 5 g chitosan powder were, respectively, weighed and dissolved with 0.1 mol dilute hydrochloric acid. A total of 5.6 g of β-glycerophosphate sodium (β-GP) powder was weighed and dissolved in 10 ml ultrapure water with the assistance of magnetic stirring. A 56% β-GP solution was obtained and sterilized by using a filter membrane (22 μm) and stored in the refrigerator at 4°C for further experiment. A total of 1 g of Celecoxib powder was weighed and dissolved in 50 ml ultrapure water with the assistance of magnetic stirring. 1 ml of 56% β-GP solution was, respectively, added to 9 ml of 2%, 3%, and 5% chitosan solutions with slow stirring. According to the previous relevant literature, a total of 1–1.5 ml of 2% celecoxib solution was added to 9 ml of 2% chitosan solution with constant stirring (Choi et al., 2020).

After freeze-drying, the chitosan hydrogel samples with different concentrations were fixed on the specimen platform. The samples were covered with gold produced by the sputter coater, and its morphology was observed by using a scanning electron microscope (SEM). Different parts of the hydrogels were randomly selected for further analyses of celecoxib and the hydrogel. Therefore, the hydrogels were analyzed by using an energy dispersive spectrometer (EDS) to observe the distribution of celecoxib in chitosan hydrogels. After coating gold on the surface of the sample, samples were placed into the sample chamber of the scanning electron microscope, and its morphology was observed with 15 kV accelerating voltage. The qualitative and semi-quantitative analyses of samples were conducted by using an X-ray energy spectrum analyzer.

According to the former study, a total of 3 ml TICHC was placed into sample tubes. Then, the tubes were placed in 37°C incubators to analyze temperature-sensitive characteristics of the hydrogel using the tube inversion method.

Samples of 5 ml liquid 2% TICHC were, respectively, extracted by using a 10-ml syringe. The 10-ml syringe was used to inject hydrogel with the appropriate force to simulate the injection process in vivo.

In order to evaluate the cytocompatibility of TICHC samples, the cytotoxicity was analyzed by using the cell counting kit (CCK; Sigma-Aldrich Co., Ltd., United States) and the Hoechst staining experiment (Sigma-Aldrich Co., Ltd., United States), respectively. The different extraction concentrations of hydrogel samples were divided into three groups including the control group, 50% extraction group, and 25% extraction group. The cultured L929 cells were dissociated to obtain a single cell after adding trypsin. After 1, 3, and 5 days of culture, the fresh medium and CCK reagent were added and incubated for 30 min at 37°C after the removal of the old medium. Then, the measurement of absorbance was conducted at 450 nm. After adding the hydrogel solution for 5 days, 10 ug/ml Hoechst 33342 staining solution was added to the plate at 37°C for 15 min. The cell cultured glass was fixed with 4% paraformaldehyde for 15 min and then washed with PBS solution three times to be observed under a fluorescence microscope (Leica DMI4000B, Germany).

Eighteen female New Zealand rabbits (6 months old, approximately 2.0–2.5 kg) were provided from the animal center of Binzhou Medical University (Yantai, China) and were used to analyze the biocompatibility and degradability of hydrogel in vivo. Eighteen rabbits were divided into 3 groups, 6 in each group. They were the control group, the hydrogel group, and the drug-loaded hydrogel group, respectively. About 3 ml of normal saline, chitosan hydrogel, or drug-loaded hydrogel was injected subcutaneously in the front of the thigh of each New Zealand white rabbit, depending on its group. After injection, 2 rabbits in each group were selected and the skin of the right thigh was cut at 2 weeks, 1 month, and 2 months after injection. The histocompatibility and degradation of the hydrogel samples in vivo were observed.

This study was approved by the Affiliated Hospital of Qingdao University Ethics Committee. The animal experiment and feed were in accordance with the guidelines of the animal center of the Binzhou Medical University, Yantai, Shandong. The fifteen New Zealand rabbits (5–6 months, 2.0–2.5 kg) in the in vivo study were obtained from the animal center of Binzhou Medical University. They were divided into three groups based on the different times including 2 weeks, 1 month, and 2 months. Five rabbits were selected to conduct the X-ray and MRI at different times. All rabbits were housed in single cages under controlled conditions (17–23°С, and 30–70% air humidity with appreciative air circulations) and provided with adequate feed and water. All rabbits fasted for 12 h and were forbidden water for 6 h before anesthesia. Each rabbit was fixed into the anesthesia box to be anesthetized by urethane through the marginal ear vein and the penicillin was injected before the operation. The rabbit was placed on the small operation table in the left lateral position. After the removal of the fur from the dorsal surface, the anterolateral intervertebral disc 2–6 were, respectively, exposed through an anterolateral approach by blunt dissection of muscles. The IDD model of rabbits was obtained by using a 20G needle puncture at L3/4, L4/5, and L5/6, respectively. Each disc was punctured five times by using a needle and continuously aspirated with a 10-ml syringe for 30 s. The annulus fibrosus was punctured to establish the local defect, and the nucleus pulposus was partially sucked by the needle. In order to avoid individual differences, the different lumbar segment of the rabbit was conducted with different interventions on the same rabbit. A total of four groups were established including a control group (L2/3), a degeneration group (L3/4), a chitosan hydrogel treatment group (L4/5), and a chitosan hydrogel–loaded with celecoxib treatment group (L5/6). The 1-ml injection with an 18G needle was used to inject the hydrogel into different intervertebral discs. After the injection of hydrogel, the incision was sutured and the rabbits were placed on the operation table at a constant temperature to wait for recovery. The rabbits were injected with penicillin for 3 consecutive days after surgery.

The X-ray can show the height changes of each segment to indirectly prove the instability. After 2 weeks, 1 month, and 2 months following the establishment of the IDD model, five rabbits were randomly selected for X-ray and magnetic resonance imaging (MRI) assessments at each time point. The X-ray imaging was performed by an X-ray system (Siemens, German), and the MRI tests were conducted using a 3.0-T MRI system (Siemens, German).

The X-ray and MRI image analysis and measurements were analyzed by two independent radiologists. The Bradner disc index (BDI) was used to evaluate the intervertebral height changes. The modified Thompson classification was used to evaluate the disc degeneration changes by using the T2-weighted MRI images with grades I to IV.

The air embolism after anesthesia was used to euthanize the rabbits after 2 months. The spine samples were fixed by using formalin (10%) for 1–3 days, and the different intervertebral discs were, respectively, cut off from the spine. We used EDTA (10%) to decalcify the surrounding bone of intervertebral discs for 1–2 months. We selected the coronal median section of the intervertebral disc for the histological examination. After obtaining the wax-embedded intervertebral disc samples, we, respectively, stained the samples with H&E. The histological classification was conducted based on the observation by using light microscopy.

The data are shown as the mean ± standard deviation (SD). All the experiments were conducted at least three times independently. The data of results were analyzed by SPSS 23.0 statistical software. The statistical significance of the differences between the groups was calculated by using the one-way ANOVA and Tukey’s post-hoc test or nonparametric test. p < 0.05 is considered to be statistically significant.

The different concentrations of chitosan hydrogel in this study were the liquid state with different viscosity at room temperature. The appearance of chitosan hydrogel loaded with celecoxib was a milk-white liquid at room temperature. For this reason, it can be injected by using a different-sized injector. As shown in Figures 2A,B, the thermosensitive chitosan liquid hydrogels with/without celecoxib were able to turn solid at 37°C in 8–15 min. Hence, it was suitable for the repair of intervertebral disc defects through injection in a narrow surgical space. The thermosensitive chitosan liquid hydrogel was able to convert to a solid state at a local defect under the normal body temperature in a short time. As shown in Figure 3, the different concentrations of hydrogel showed the porous structure with different densities (higher the concentration, higher the density) by scanning electron microscopy. After comparing the different concentrations of chitosan hydrogel, the 2% hydrogel was more suitable for injection and better porous density. As shown in Figure 4, the F (red) and S (green) elements were uniformly distributed within the chitosan hydrogel by EDS, and it was indirect proof that the celecoxib was successfully loaded in the chitosan hydrogel.

In order to evaluate its liquid–solid transformation ability, we used the tube inversion method and different time measurement methods to analyze the liquid–solid conversion time of different hydrogels. As shown in Figure 2B, the results showed that the liquid–solid conversion time of 2% chitosan hydrogel with/without celecoxib was, respectively, 12.20 ± 0.84 min and 12.80 ± 1.48 min. So, the celecoxib was not able to significantly affect the liquid–solid conversion time of hydrogel. The shortest liquid–solid conversion time was 8.40 ± 1.14 min (5% chitosan hydrogel). The results indicated that the higher concentration chitosan hydrogel had a shorter liquid–solid conversion time. (p < 0.05).

The cytotoxicity of hydrogel was analyzed by using L929 cells. As shown in Figures 5A,B, the chitosan hydrogel with/without celecoxib had no significant effect on the proliferation of L929 cells than the control group at 1, 3, and 5 days of incubation. Moreover, the high concentration extracts of chitosan hydrogel loaded with celecoxib slightly inhibited the proliferation of L929 cells, displaying 79.56 ± 3.55%, 71.35 ± 1.39%, and 70.56 ± 6.10% cell viability for the hydrogel solution with the prolongation of culture time. Overall, the hydrogel with/without celecoxib solutions showed low cellular toxicity.

A live/dead staining assay of L929 cells was conducted using Hoechst 33342 to further confirm the cellular toxicity of hydrogels. Hoechst 33342 is a blue fluorescent dye with low toxicity, which is able to penetrate the cell membrane of the cell. The fluorescence microscopy was used to observe the proliferation of L929 cells after dyeing. As shown in Figure 5C, the results showed that the cell numbers of the three samples had no significant difference. Compared with the control group, there was no significant reduction of cell numbers in the hydrogel with/without celecoxib simples. The results indicated that the hydrogel loaded with celecoxib has satisfying biocompatibility and is a suitable injectable biomaterial for biomedical fields.

There was no significant infection in 18 rabbits after local hydrogel injection. In 2 weeks, there was no normal saline residue under the skin of the rabbits in the control group. No significant infection was found in the subcutaneous fascia and surrounding muscles. So, it was not necessary for further observation of the rest rabbits in the control group. As shown in Figure 6, there was obvious hydrogel residue under the skin of the rabbits of the other two groups for 2 weeks. Moreover, there was no sign of infection with the obvious distinction between hydrogel residue and surrounding tissue. With the extension of observation time, the hydrogel residues were gradually degraded under the skin, and there was no obvious hydrogel residue under the skin in 2 months. The results indicated that the injectable thermosensitive celecoxib-loaded chitosan hydrogel was able to be used as non-toxic material with good biocompatibility and histocompatibility for biomedical application.

In a total of two New Zealand white rabbits, death occurred in the process of anesthesia and post-operation. There was no obvious sign of infection, and all rabbits were in good condition after surgery. The postoperative X-ray and MRI showed the success of the establishment of the intervertebral disc degeneration model by needle puncture. Therefore, the composite hydrogel showed good sealing properties of the AF defect by injecting the exogenous hydrogel.

Postoperative X-rays can visually display the intervertebral disc height. The postoperative intervertebral disc height changes were judged using the Bradner disc index (BDI). In addition, the intervertebral disc degeneration can be evaluated by the observation of sagittal MRI T2-weighted image. At 2 weeks, 1 month, and 2 months after the initial surgery, the X-ray imaging showed that a relatively normal BDI was maintained in the hydrogel treatment groups than in the degeneration group (Figures 7A,B) (p < 0.05). Moreover, the degeneration group had a weaker MRI signal than the hydrogel treatment groups after the operation (Figures 7C,D). With the extension of time, the difference became more obvious. However, the intervertebral height and the degeneration grade of IVD still gradually decreased in hydrogel groups than in the degeneration group.

The H&E staining (Figure 8A) and safranin O-fast green staining (Figure 8B) were able to show the chromatin in the cell nucleus, cytoplasmic ribosome, cytoplasm, and the extracellular matrix. Normal nucleus pulposus tissue after staining can clearly show the NP cells and onion-shaped AF. As shown in Figure 8, the staining showed an obvious decrease in NP cells and height of IVD with a local defect in the degeneration group. However, there still was NP cell residue in the center of IVD without an obvious local defect in the hydrogel loaded with the celecoxib group. The histological grade was 9.6 ± 0.9 in the degeneration group and 7.0 ± 1.0 in the hydrogel loaded with the celecoxib group (Figure 8C). In the hydrogel loaded with the celecoxib group, there was no significant difference in histological grade compared with the hydrogel group, but it was worse than that in the control group. In addition, the number of NP cells can be observed by staining. Compared with the degeneration group, there were more NP cells residual in hydrogel groups which is beneficial for delaying the progress of IVD.