Ten-week-old young male Wistar rats (Japan SLC, Inc., Shizuoka, Japan; n = 162) were used to examine bone regeneration in a state of relatively active bone remodeling without the effects of estrogen. All experiments were conducted following Hokkaido University’s guidelines for animal care and research use (approval #18-0076). The rats were allocated to three groups and treated with vehicle (control group, n = 54), β-tricalcium phosphate (βTCP group, n = 54; TERUFILL®, Olympus Terumo Biomaterials, Inc., Tokyo, Japan), or a combination of 60% βTCP and 40% PPL (βTCP + PPL group, n = 54). Pullulan was purchased from Hayashibara Co., Ltd., Okayama, Japan, and its concentration was adjusted as described below. The rats were anesthetized with a mixture of 0.3 mg/kg of medetomidine, 4.0 mg/kg of midazolam, and 5.0 mg/kg of butorphanol. After partially shaving the hairs of the tibiae, a round 2.0-mm-diameter defect was created in the arterial center of the tibia, specifically the region approximately 1 cm below the epiphyseal articular surface, using a round bur (023#8, D+Z, Inc., Germany, 2.0 mm diameter) (Figures 1A, B). The defect site was irrigated with sterilized physiological saline during drilling to prevent overheating. Once the defect was drilled, the vehicle, 2 mg of βTCP, or a combination of βTCP and PPL (n = 54 per group) were inserted into the defects using a plugger or syringe. The defects were covered with tibial skins by suturing to avoid material loss.

PPL was prepared as described in a previous study (Yoshida et al., 2013) with some modifications. Briefly, 50 g of pullulan (Hayashibara Co., Ltd.) was dissolved in 3,750 ml of an aqueous solution containing 41.5 g of sodium hydroxide and stirred overnight at 20°C. Next, 32.5 g of phosphoryl chloride was added to the solution dropwise and stirred at 0°C overnight. Sodium phosphate and sodium chloride were then eliminated by dialysis to extract an aqueous solution of phosphorylated pullulan, which was then spray-dried and sterilized with 25 kGy of gamma/X-rays (average molecular weight: 1,300,000).

For grafting, βTCP (0.15–0.5 mm original size, TERUFILL®) was ground in a mortar into a fine powder (100–250 μm particles). Next, 2 mg of crushed βTCP granules were grafted into the bone defects using a plugger. For the βTCP + PPL grafts, 50 mg of PPL was dissolved in 250 μl of 2% CaCl₂, after which 2 mg of a mixture of 60% crushed βTCP granules and 40% PPL with 2% CaCl₂ was injected into the defects using a syringe.

Rats from the control, βTCP, and βTCP + PPL groups were anesthetized with an intraperitoneal injection of the excessive mixture of medetomidine, midazolam butorphanol for bodyweight determination at 1, 2, and 4 weeks post-surgery (n = 54 per group). Further, thirty-six rats per group were perfused with 4% paraformaldehyde diluted in 0.1 M cacodylate buffer (pH 7.4) through the left cardiac ventricle. After perfusion, the tibiae were extracted and immediately immersed in the same solution for 24 h at 4°C. The tibiae were then cut at the assumed proximal and distal lines 2 mm distant from the walls of the bone defect (inside the dotted line in Figure 1C) and washed in 0.1 M cacodylate buffer (pH 7.4). For histochemical examination, some of the tibiae (n = 18 per group) were decalcified with 10% EDTA-2Na for 2 months for paraffin embedding, while the remaining specimens without decalcification (n = 18 for each group) were processed for epoxy resin embedding for TEM and electron microprobe analyses. For TEM observation, the samples were post-fixed with 1% osmium tetraoxide in 0.1 M cacodylate buffer for 8 h at 4°C, dehydrated in an acetone solution gradient, and finally embedded in epoxy resin (Epon 812, Taab, Berkshire, United Kingdom) (Hasegawa, 2018; Hasegawa et al., 2022). Semi-thin sections without decalcification were subjected to von Kossa staining. Ultra-thin sections obtained from the undecalcified specimens were examined via TEM (Hitachi H-7100, H-7800, Hitachi Co., Tokyo, Japan) at 80 kV. Tibiae from other anesthetized rats without paraformaldehyde fixation were extracted for reverse transcription quantitative PCR (RT-qPCR) analysis (n = 18 for each group).

Micro-CT images were obtained from the region of interest (ROI) surrounding the cortical endosteum facing the bone cavity and 1.0 mm proximally and distally from the edges of the cavity using a micro-CT unit (tube voltage 90 kV, tube current 88 μA, FOV 10 mm, voxel size 20 μm, CosmoScan FX, Rigaku Corporation, Tokyo, Japan) (Figure 1B). Image reconstruction was conducted using the CT analyzer software (CosmoScan Viewer, Rigaku Corporation, Tokyo, Japan) as described by Bouxsein et al. (2010).

For immunohistochemistry and enzyme histochemistry, 5 μm-thick paraffin sections were prepared. Dewaxed sections were treated with 0.1% hydrogen peroxidase for 20 min to inhibit endogenous peroxidase and pre-incubated with 1% bovine serum albumin in phosphate-buffered saline (pH, 7.4, BSA-PBS) for 30 min at room temperature. Antisera against ALP (Oda et al., 1999), osteopontin (LSL Co., Tokyo, Japan) or PHOSPHO1 (Cusabio Technology LLC., Houston, TX) diluted at 1:300, 1:3000, and 1:100 with 1% BSA-PBS, respectively, were applied to the sections overnight at 4°C. The sections were then incubated with horseradish peroxidase (HRP)-conjugated anti-rabbit IgG (Chemicon International Inc., Temecula, CA). The immunoreactions were visualized with diaminobenzidine as a substrate before observation under a light microscope (Eclipse E800, Nikon, Ltd., Tokyo). For tartrate-resistant acid phosphatase (TRAP) enzyme histochemistry, the sections were deparaffinized and immersed in 50 ml aqueous solution of 5 mg naphthol AS-BI phosphate (Sigma-Aldrich, St Louis, MO), 25 mg red violet LB salt (Sigma-Aldrich), and 100 mM L(+) tartaric acid (0.76 g; Nakarai Tesque, Kyoto, Japan) diluted in a 0.1 M sodium acetate buffer (pH 5.4) for 15 min at 37°C. Finally, these sections were faintly counterstained with methyl green.

An assumed box-shaped area of 1,000 µm (width) × 1,500 µm (length) located on the bone defect was used to assess the ALP-reactive area/TV, PHOSPHO1-positive area/TV, and the number of TRAP-positive osteoclasts. The images of the ALP-reactive area and PHOSPHO1-positive area/TV were measured using Image-Pro Plus 6.2 (Media Cybernetics, Bethesda MD) and divided by the index of the assumed boxed areas. TRAP-positive cells with more than two nuclei were considered multinucleated osteoclasts.

Semi-thin sections of epoxy resin-embedded specimens were cut with an ultramicrotome (Sorvall MT-5000; Ivan Sorvall, Inc., Norwalk, CT) and incubated with the silver nitrate aqueous solution until a dark brown/black staining of the bone tissue was discernible under a light microscope (Hasegawa et al., 2023). The stained sections were counterstained with toluidine blue and observed under a Nikon Eclipse E800 microscope (Nikon Instruments Inc., Tokyo, Japan).

Undecalcifed specimens from the control, βTCP, and βTCP + PPL groups at 1, 2, and 4 weeks (n = 6 each) were subjected to calcium (Ca) and phosphorus (P) elemental mapping using an electron probe microanalyzer (EPMA; JXA-8530F, JEOL Co., Ltd., Japan). Undecalcified specimens at all experimental stages were embedded in epoxy resin and trimmed until the central regions were reached, in parallel to the sagittal plane, including the grafted defects. The specimens were then polished and sputter-coated with carbon. For each sample, 256 × 192 pixels mapping was conducted for higher magnified images. The accelerating voltage and beam current were set to 15 kV and 2 × 10⁻⁸ A, respectively, and the integrating time was 0.050 s per pixel (Yamamoto et al., 2016).

The MC3T3-E1 osteoblastic cell line (RIKEN BRC through the National Bio-Resource Project of the MEXT/AMED, Japan) was cultivated in a medium containing 10% fetal calf serum, 50 μl/ml of ascorbic acid, and 10 mM glycerol phosphate in MEMα (FUJIFILM Wako Pure Chemical, Inc., Osaka, Japan). Additionally, 50 μg/μl of PPL was added to the culture medium. PPL was dissolved in the cultured medium. Total RNA was extracted from the treated MC3T3-E1 cells after 3, 6, 9, and 12 days for RT-qPCR analysis.

To evaluate the expression profiles of tissue nonspecific alkaline phosphatase (Alp), Trap, Osteopontin, and Phospho1, total RNA was extracted from the specimens, including the surrounding bone (2 mm proximally and distally) (n = 6 for each). Furthermore, to determine the expression of Alp, Osteopontin, Osteocalcin, Integrin αv, and Dmp-1 in the MC3T3-E1 cells cultivated with PPL-containing medium, total RNA was extracted from these cells at 3, 6, 9, and 12 days post-cultivation. The specimens were homogenized in 10 mL TRIzol reagent (Life Technologies Co., Carlsbad, CA) per 1 g tissue to extract total RNAs using a Multi-beads Shocker homogenizer (Yasui-Kikai, Osaka, Japan). The mixture was centrifuged at 15,000 rpm for 5 min at 4°C to remove small debris. The supernatant was then transferred to a new tube, 2 mL chloroform was added, and the mixture was vortexed for 15 s. The lysate was transferred to a new tube and incubated for 5 min at room temperature. After phase separation, the aqueous phase containing the RNA was transferred to a fresh tube, and RNA was precipitated by adding 5 mL isopropyl alcohol per 10 mL TRIzol reagent. After 10 min of incubation at room temperature, the mixture was centrifuged at 15,000 rpm for 60 min at 4°C. The RNA pellets were washed with 1 mL of 75% ethanol and briefly air-dried, then dissolved in 30 μL DEPC-treated water. First-strand cDNA was synthesized from 2 μg of total RNA using the SuperScript VILO cDNA Synthesis Kit (Life Technologies). RT-qPCR assays were performed using Taqman probes (Applied Biosystems) to quantify the expression of Alp (Rn01516028_m1), Trap (Rn00569608_m1), Osteopontin (Rn00563571_m1), and Phospho1 (Rn01496967_m1) in rat specimens and Alp (Mm00475834_m1), Integrin αV (Mm01339539_m1), Osteopontin (Mm00436767_m1), Osteocalcin (Mm03413826_m1), and Dmp-1 (Mm01208363_m1) in cultured mouse MC3T3-E1 cells. The Ct values for these genes were detected using a StepOne Real-Time PCR System (Applied Biosystems) and normalized to Gapdh (Rn01775763_g1 or Mm99999915_g1) expression using the 2^−ΔΔCT method.

All statistical analyses were performed using the BellCurve for Excel software (Social Survey Research Information Co., Ltd., Tokyo, Japan). For animal experiments, significant differences between the control and βTCP groups or the βTCP + PPL groups at 1, 2, and 4 weeks were determined using one-way ANOVA followed by Dunnett’s multiple comparison test. For comparisons between the control and PPL groups at 3, 6, 9, and 12 days in the cell culture experiments, the Student’s t-test was performed. All values are presented as mean ± standard deviation. p-values < 0.05 were considered statistically significant.

Micro-CT analysis revealed rapid bone formation in the control group at 1 week; however, the regenerated bone was not resorbed until 4 weeks (Figures 2A–C). In addition, a gathering of fine trabeculae was observed in the bone defects at week 1 using H-E staining (Figure 2D). The number of trabeculae decreased during the second week (Figure 2E), and only a few trabeculae were left in the original bone marrow of the defect at 4 weeks (Figure 2F). In the βTCP group, radiopaque granules identical to βTCP were identified from the first week to fourth weeks under micro-CT observation (Figures 2G–I). H-E staining highlighted βTCP granules throughout the experimental periods; however, the newly-formed trabeculae thickened progressively (Figures 2J–L). Similar to the βTCP group, the βTCP + PPL group exhibited radiopaque granules (Figures 2M–O). However, many fine trabeculae were identified in the βTCP + PPL group through H-E staining compared to those in the βTCP group from weeks 1–4 (Figures 2P–R).

At higher magnification, the control specimen exhibited fine immature trabeculae at 1 week (Figure 3A). The trabeculae were surrounded by bone marrow cells at 2 weeks (Figure 3B) and subsequently fragmented into small pieces until week 4 (Figure 3C). In the βTCP group, many fibroblast-like cells occupied the βTCP granules interspace at week 1 (Figure 3D). New bone tissue associated with the βTCP granules was observed at the second week (Figure 3E), and this tissue surrounded the βTCP by week 4 (Figure 3F). Similar to the βTCP group, many fibroblast-like cells filled the interspace of the βTCP granules and PPL in the βTCP + PPL group at week 1 (Figure 3G). During week 2, new bone was associated with the βTCP granules and PPL (Figure 3H). The regenerated bone included many granules of βTCP and irregularly shaped PPL at 4 weeks (Figure 3I).

In the control group, many ALP-positive osteoblastic cells, i.e., osteoblasts and preosteoblasts, covered the bone surface, and PHOSPHO1-positive osteoblasts identical to matrix vesicle-synthesizing mature osteoblasts were lined up in a row on the new bone from week 1 to week 4 (Figures 4A–C, 5A–C). In the βTCP group, ALP-positive osteoblastic cells and PHOSPHO1-positive osteoblasts were observed on the newly-formed bone and βTCP granules at week 1 (Figures 4D, 5D). Given that the βTCP granules were embedded in the regenerated bone matrix, PHOSPH1-positive osteoblasts were predominantly found on the new bone until week 4 (Figures 5E, F). ALP-positive osteoblastic cells were located on a broader range of bone surfaces compared to PHOSPHO1-positive osteoblasts (Figures 4E, F). In the βTCP + PPL group, ALP-positive osteoblastic cells and PHOSPHO1-positive osteoblasts were observed not only on the newly-formed bone and βTCP granules but also directly on the PPL (Figures 4G–I, 5G–I). Although βTCP granules were embedded in the regenerated bone, ALP-positive osteoblastic cells and PHOSPHO1-positive osteoblasts were localized on the PPL at week 4 (Figures 4I, 5I). There were no significant differences in PHOSPHO1-positive area/TV in the control, βTCP, and βTCP + PPL groups during the experiments (Figure 5J), whereas the ALP-positive area/TV in the βTCP and βTCP + PPL groups at week 4 had higher indices compared to that of the control group (Figure 4J). RT-qPCR detected no significant differences in Alp or Phospho1 expression in the control, βTCP, and βTCP + PPL groups during the experiments (Figures 4K, 5K).

Similar to the Alp and Phospho1 expression, the gene expression of Osteopontin were detected at week 1 but decreased progressively until week 4 (Figure 6J). There was no statistical difference in the expressions of Osteopontin at weeks 1, 2, and 4 among any of the experimental groups. In the control group, the immunoreactivity of osteopontin, known to have a high affinity to crystalline calcium phosphates (Amizuka et al., 2012), was detected on the superficial layer of newly-formed bone (Figure 6A). Later, the immunoreactivity was restricted to the cement lines (i.e., the boundary between new and old bones) (Figures 6B, C). In the βTCP group, osteopontin was found in the inner region of the grafted βTCP granules (Figure 6D) and consistently detected in the granules embedded in the regenerated bone matrix at weeks 2 and 4 (Figures 6E, F). In the βTCP + PPL group, osteopontin was detected not only on the surface of the newly-formed bone at week 1 but also on the PPL, which persisted until week 4 (Figures 6G–I).

We next examined the localization of TRAP-reactive osteoclasts (Figure 7). Initially, TRAP-reactive osteoclasts appeared early at week 1, increased their numbers at week 2, and then decreased at week 4 in all three groups (Figures 7A–C, G–I, M–O, S). At a higher magnification, the control group exhibited many TRAP-reactive osteoclasts on the trabeculae during the first and second weeks; however, as the trabeculae decreased, fewer of these osteoclasts were also observed at week 4 (Figures 7D–F). In the βTCP group, many TRAP-reactive osteoclasts were located on the βTCP granules from week 1 and found predominantly localized on the βTCP rather than on the newly-formed bone at week 2 (Figures 7J, K). Fewer osteoclasts were seen on the thick trabeculae of the βTCP group (Figure 7L). In the βTCP + PPL group, TRAP-reactive osteoclasts were majorly observed on the newly-formed trabeculae, while only a few were located on the PPL from weeks 1–4 (Figures 7P–R). Although no significant differences in the index of osteoclast number were observed in the experimental groups at weeks 1 and 2, there was a trend towards a higher number of osteoclasts in the βTCP and βTCP + PPL groups compared to the control group at weeks 4 (Figure 7S). RT-qPCR detected no significant differences in the expression of Trap at weeks 1, 2, or 4 among any of the experimental groups (Figure 7T).

Next, we examined whether PPL could affect the expression of genes associated with bone synthesis in the osteoblastic cell line MC3T3-E1. mRNAs encoding Alp, Integrin αv, Osteopontin, Osteocalcin, and Dmp-1 were significantly upregulated during the experiment (Figure 8). However, there was no significant difference observed in the expression of these genes between the control and PPL-treated MC3T3-E1 cells except the expression of Osteocalcin and Dmp-1 at 6 days.

Finally, we attempted to demonstrate whether calcium phosphate crystals could be directly deposited onto PPL. We have employed EPMA to detect the intensity of signaling showing calcium (Ca) and phosphorus (P) in PPL. The Ca and P indices were examined on the lines crossing the PPL and surrounding regenerated bones, as illustrated in Figures 9A–C. The Ca index was not detectable at week 1 (Figures 9D, G). Moreover, a high Ca index was observed in the regenerated bone associated with PPL but neither in the PPL itself nor in the soft tissues at week 2 (Figures 9E, H). The Ca index was further increased at week 4 compared to week 2 (Figures 9F, I). Interestingly, P was not detected at week 1 (Figures 9J, M). P indices were high in the newly-formed bone, but unlike the Ca index, there was no difference in the P indices between weeks 2 and 4 (Figures 9K, L, N, O). Therefore, the Ca composition index appeared to increase only in the regenerated bone associated with PPL but not in the PPL itself.

EPMA analysis showed that calcification in the PPL was difficult to observe after 1 week, whereas, after 4 weeks, the PPL was almost embedded in the newly formed bone that had been remodeled. As shown in Figure 9, the PPL-including specimens after 2 weeks, rather than those after 1 and 4 weeks, appeared to be adequate for demonstrating deposition of calcium phosphate crystals onto the PPL. Thus, calcification in the regenerated bone associated with the PPL at week 2 post-surgery was examined via von Kossa staining and TEM analysis (Figure 10). At a higher resolution, needle-like calcifications extending into the PPL were observed (Figure 10B), and osteoblasts that migrated inside of the PPL exhibited globular calcified materials in their vicinity (Figure 10C). TEM observed identified many calcified globular structures identical to calcified nodules and linear calcified structures characteristic of calcified collagen fibrils in the superficial layer of the PPL (Figures 10D, E). At a higher magnification, the calcified globular and linear structures were found to be composed of assembled fine needle-like crystals (Figures 10F, G).