Tetraethyl orthosilicate (Si(OC₂H₅)₄; Sigma-Aldrich, St. Louis, MO, United States) and calcium nitrate (Ca(NO₃)₂_4H₂O; Showa, Tokyo, Japan) were used as precursors for dicalcium silicate, respectively, and the molar ratio of Si/Ca used in this study was 2:3. A detailed description of the powder’s fabrication using sol-gel methods has been reported (Chen et al., 2009). After sintering at 800°C for 2 h, the granules were then ball-milled for 12 h in ethyl alcohol using a centrifugal ball mill (Retsch S 100, Hann, Germany) and then dried in an oven at 60°C. Bi₂O₃ (Sigma-Aldrich) was added to the ground powder at 20 wt% using a conditioning mixer (ARE-250; Thinky, Tokyo, Japan). To prepare RDSC specimens, the powder was hand-mixed with distilled water in a liquid-to-powder (L/P) ratio of 0.4 ml/g.

After mixing according to the manufacturer’s instructions, the cements were placed in a cylindrical Teflon mold to form the cylindrical specimen with a dimension of 14 mm (diameter)* 2 mm (height); the specimens were stored in an incubator at 100% relative humidity and 37°C for 1 day to set. Then, the specimens were transported into a 60°C oven to dry.

To investigate the phase composition, the specimens were ground to fine powders and then characterized with an x-ray diffractometer (XPert Pro, PANalytical, Netherlands). Fourier transform infrared spectroscopy (FTIR5700, Thermo, United States) was used to analyze the powders. Scanning electron microscopy (Zeiss SIGMA, Zeiss, Germany) was used to characterize the microstructure of the various specimens.

All teeth were collected from the School and Hospital of Stomatology, Wuhan University with informed consent. With the permission of the Institutional Ethical Board of Wuhan University [no. 2017 (1)], we collected and used the human samples based on the guidelines of the National Institutes of Health. Forty-five intact human third molars with open apex extracted for orthodontic reasons were obtained from healthy young patients (15 to 20 years old). The surgery procedure was described previously (Pedano et al., 2020; Téclès et al., 2008). In brief, The teeth were collected immediately after the surgery and transported in centrifuge tubes which contained 10 ml of α-minimum essential medium (α-MEM) supplemented with 1% penicillin-streptomycin and 0.75 μg/ml amphotericin B. After removing remained gingiva, periodontal ligament, and the dental sac with sterilized instruments, teeth were then rinsed with phosphate buffer saline (PBS). A class-I cavity was prepared using a 1.0 mm-diameter carbide bur at high speed with copious cooling. The diameter of the cavity in the enamel layer was approximately 4 mm. Then, the bur was kept working until the penetration to the pulp cavity. Finally, a puncture point was prepared in diameter of approximately 1.0 mm. Afterwards, the cavity was irrigated with sterile saline. Sterile cotton pellets were applied with light pressure for drying. Three commercial pulp capping agents were prepared according to the manufacturer’s instructions and applied at a thickness of approximately 2 mm while no pulp capping agent was applied in the exposed site of control group (Figure 1A). The teeth were divided into 5 groups: 1) RDSC, 2) WMTA, 3) Biodentine, 4) iRoot BP Plus, 5) a negative control (n = 3 for each group). Finally, restoration was completed by glass ionomer and resin. All the samples were then taken into the laboratory and rinsed with sterile PBS for 5 times. The tooth culture medium was α-MEM containing 1% penicillin-streptomycin, 0.75 μg/ml amphotericin B, 10% fetal bovine serum and 2 mM b-glycerophosphate. Then, the samples were hung in the 12-well culture plates and immersed in 4 ml tooth culture medium which was refresh every day. After culture for 1, 7 or 14 days, teeth were rinsed by ddH₂O. Pulp capping materials were exposed. Two independent observers who were blinds to the materials evaluated the discoloration of teeth. If the results of two independent observers were different. The teeth would be evaluated by third veteran observer and the evaluation of the third evaluator should prevail. All observers passed the consistency check before the study. Then, the occlusal surfaces and the cervical part of the teeth were photographed.

After culturing, 15 teeth in 7 days groups were immersed in 4% paraformaldehyde (PFA) for 2 days for fixation. The teeth were then decalcified using 10% EDTA and embedded in paraffin blocks. Subsequently, the teeth were sectioned into slices at a thickness of 5 μm, followed by staining of hematoxylin and eosin solution (Servicebio, WuHan, China) and modified Masson’s Trichrome stain kit (maxim, Fu Zhou, China) for the identification of collagen.

The use of animal model was approved by the Animal Ethics Committee in Wuhan University (protocol00279848). Three female minipigs at the age of 36 months with intact permanent dentitions were used. Each of them was raised in a single cell and fed twice a day with ad libitum access to water.

Before the experiment, the dentition of each miniature pig was scanned using computed tomography to ensure that its teeth were healthy and fully developed. Premolars and molars on the right side were scheduled for 7 days observation group while the ones on the left were for 70 days. Specifically, 70-days group was operated on the first day. After 63 days, 7-days group was scheduled. Then, the animals were sacrificed on the 70th day. Teeth were divided into four material groups: for day 7, 1) RDSC (n = 6, 5 premolars and 1 M), 2) WMTA (n = 6, 5 premolars and 1 M), 3) Biodentine (n = 7, 5 premolars and 2 M), 4) iRoot BP Plus (n = 6, 4 premolars and 2 M); for day 70, 1) RDSC (n = 7, 5 premolars and 2 M), 2) WMTA (n = 5, 4 premolars and 1 M), 3) Biodentine (n = 8, 6 premolars and 2 M), 4) iRoot BP Plus (n = 6, 4 premolars and 2 M), as detailed in Table 1.

General anesthesia was conducted intramuscularly using Zoletil 50 (0.1–0.15 mg/kg; Virbac, Fort Worth, TX, United States). Under local anesthesia by 2% lidocaine, class-I cavities were prepared on the occlusal surfaces. The diameter of the cavities and exposure site was 3.0 and 1.0 mm, respectively. After rinsing with sterile saline, sterile wet cotton pellets were applied gently to control the hemorrhage. Pulp-capping agents were prepared according to the manufacturers’ instructions. 2 mm thick pulp-capping agents were applied, followed by a layer of glass ionomer cement (ShangChi, Changshu, China). Finally, the cavities were restored with a universal and light-curable restorative composites (Herculite Precis, Kerr, CA, United States) (Figure 1B). Soft diet was then provided for the animals.

After euthanasia, the mandible and maxilla of the miniature pigs were separated, and photographs were taken before 4% PFA fixation. The teeth were then removed individually and immersed in the 4% PFA at 4°C for 2 days.

Teeth were scanned using a μCT scanner (Skyscan 1,176, Bruker Micro-CT, Belgium) with the following parameters: 82 kV/277 μA X-ray source, 0.1 cuprum filter, pixel size of 18 μm, averaging frame of 1, rotation step of 0.5, random movement of 50 and 360°rotation around the vertical axis. The raw data were reconstructed using NRecon software (Bruker Micro-CT).

For quantitative analysis, we used Dataviewer software and CTAn software (Bruker Micro-CT) to measure the thickness of newly formed mineralized tissue (from exposure side to material side) using the full-axial slices dataset. The maximum thickness of newly formed mineralized tissue was recorded. The continuity of mineralized tissue was classified as either “complete”, “partial” or “none” in accordance with both micro-CT and corresponding histological section.

The histological procedures were applied as described in the ex-vivo human tooth culture model. The histological responses were evaluated using digital pathology slide scanner (panoramic midi, 3DHISTECH, Budapest, Hungary).

The occurrence of tooth discoloration was statistically compared using a Chi-square test of independence. One-way analysis of variance statistical analysis was used to evaluate the significance of the differences between the mean values. In all cases, the results were considered statistically significant at a p-value less than 0.05.

For all specimens, the broad IR absorption band corresponding to O-H stretching vibration extended over a wide wave-number range around 3435 cm⁻¹. Two sharp peaks at 1,628 and 1,473 cm⁻¹ are detected as O-H bending vibration band. Two peaks at 931 and 1,001 cm⁻¹ are ascribed to Si-O stretching vibration (Figure 2A).

The major diffraction peaks at 2θ 27°and 33°were attributed to the bismuth oxide phase (filled circle). For RDSC, two peaks at 2θ 32°were ascribed to the dicalcium silicate phase (filled diamond). For WMTA, Biodentine and iRoot BP plus, two peaks at 2θ 32°and one peak at 2θ 34°were attributed to the tricalcium silicate phase (opened diamond). For WMTA, Biodentine, peak at 2θ 29°were attributed to calcium carbonate phase (filled inverted triangle). For iRoot BP plus, two peaks at 2θ 28°and 31°were ascribed to zirconium oxide (open square) (Figure 2B).

The SEM micrographs of the four cements are shown in Figure 3. All cement specimens had an appearance with entangled particles and exhibited several micropores. The micropores with a near spherical shape were uniformly and separately distributed. However, it seems that WMTA and Biodentine had more micropores than RDSC and iRoot BP plus (Figures 3A–D). The EDX analysis shows that all specimens contain calcium and silicon (Figures 3E–H). In RDSC and WMTA, bismuth was detected due to the introduction of bismuth oxide as radiopacifer (Figures 3E,F). In Biodentine and iRoot BP, zirconium was detected due to the introduction of zirconium oxide as radiopacifer (Figures 3G,H). Aluminum was detected in WMTA (Figure 3F).

In ex-vivo human tooth culture model, RDSC shows no obvious discoloration on day 1 (Figure 4A) and slight discoloration on day 7 (Figures 4E,I). For in-vivo miniature pig model, RDSC shows discoloration on day 7 and it becomes more visible on day 70 (Figures 4M,Q). WMTA exhibites an obvious discoloration in both models, especially on day 70 (Figures 4B,F,J,N,R). No obvious discoloration can be seen at the cervical part of all ex-vivo groups. Biodentine and iRoot BP Plus merely show discoloration except for one tooth in 70 days group (Figures 4C,D,G,H,K,L,O,P,S,T). The occurrences of discoloration for different materials using in vivo miniature pig model were presented in Table 1. At day 7, the discoloration occurrence for WMTA is 100%, which is significantly higher than that of RDSC (33.3%) (p < 0.05), iRoot BP Plus 0% (p < 0.05) and Biodentine (p < 0.05). For 70 days groups, WMTA also show a significantly higher occurrence rate 80% than that of RDSC (57.1%), iRoot BP Plus (16.7%) and Biodentine (12.5%) (p < 0.05).

The histological micrographs of pulps after 7 days capping with RDSC, WMTA, Biodentine, iRoot BP Plus and negative control (no capping agent) are shown in Figure 5. In RDSC group, mineral foci are observed underneath the exposed area (arrows in Figure 5A) which would lead to the future formation of a mineralized bridge covering/healing the exposed region. Some of them contained sequestered cells showing an osteodentin aspect. Masson’s staining showed blue-stained area around the mineralized foci which represented the accumulation of collagen fibers. The other pulp-capping agents (Figures 5B–D) also show similar results as RDSC groups. Among them, Biodentine group generates the minimum size of mineral foci (Figure 5C). The control group with no pulp-capping materials exhibits a normal pulp morphology, no necrosis occurred (Figure 5E).

Three-dimensional μCT reconstructions of teeth capped with different materials for seven or 70 days are presented in Figure 3. RDSC can be distinctly distinguished from the dentin and high-intensive area is noted underneath the RDSC on day 7 (the arrow in Figure 6A). Similar results are found in WMTA and Biodentine group (arrows in Figures 6B,C). Histological micrographs identify that they are just accumulation of cells and collagen fibers or residual dentin (Figure 7). In contrast to other three materials, Biodentine is not distinct in the μCT photographs (Figure 6C). No high-intensive area is found in the iRoot BP Plus group on day 7. All teeth capped for 70 days exhibite radiopaque layer underneath the materials (arrows in Figures 6E–I). The radiopaque layer is confirmed to be reparative dentin by histological assay (the arrow in Figure 6J). The thickness of the reparative dentin bridge is measured, and RDSC appears to induce the thickest dentin bridge formation (353 μm) among all pulp-capping agents tested in which MTA is 205 mm; iRoot BP Plus is 194 mm; and Biodentine is 124 mm (p < 0.05) (Figure 4K). All teeth in 70 days group are identified as “complete” and confirmed to be reparative dentin (Table 2). Therefore, the exposure sites are all completely sealed.

The histological micrographs of pulps after 7 and 70 days capping with RDSC, WMTA, Biodentine, iRoot BP Plus and negative control (no capping agent) were shown in Figure 8. The result of 7 days group is in consistent with that of ex-vivo human tooth model that showed mineralized foci formation (arrows in Figures 8A–D). Moreover, the sizes and amounts of mineralized foci are bigger and larger than those in ex-vivo human tooth model. In RDSC and WMTA groups, accumulation of collagen fibers is noted (# in Figures 8A,B). Intact and complete reparative dentin bridge formation which possessed tubular structures were noted underneath the exposure site (*) in RDSC groups and the layer next to the pulp is blue-stained and the distant part is red-stained (Figure8E). The reparative dentin bridge facing the material side is disorganized with pores inside. However, the opposite part of the reparative dentin bridge is well-organized with tubular structures (Figure 6E). Similar results could be observed for other cements (Figures 8F–H). Mineral foci are also found close to the dentin bridges (arrows) in the RDSC, Biodentine, iRoot BP Plus groups (Figures 8E,G,H).