Augmentation of Bone Regeneration by Depletion of Stress-Induced Senescent Cells Using Catechin and Senolytics

Despite advances in bone regenerative medicine, the relationship between stress-induced premature senescence (SIPS) in cells and bone regeneration remains largely unknown. Herein, we demonstrated that the implantation of a lipopolysaccharide (LPS) sustained-release gelatin sponge (LS-G) increases the number of SIPS cells and that the elimination of these cells promotes bone formation in critical-sized bone defects in the rat calvaria. Histological (hematoxylin–eosin and SA-β-gal) and immunohistological (p16 and p21 for analyzing cellular senescence and 4-HNE for oxidation) staining was used to identify SIPS cells and elucidate the underlying mechanism. Bone formation in defects were analyzed using microcomputed tomography, one and four weeks after surgery. Parallel to LS-G implantation, local epigallocatechin gallate (EGCG) administration, and systemic senolytic (dasatinib and quercetin: D+Q) administration were used to eliminate SIPS cells. After LS-G implantation, SA-β-gal-, p16-, and p21-positive cells (SIPS cells) accumulated in the defects. However, treatment with LS-G+EGCG and LS-G+D+Q resulted in lower numbers of SIPS cells than that with LS-G in the defects, resulting in an augmentation of newly formed bone. We demonstrated that SIPS cells induced by sustained stimulation by LPS may play a deleterious role in bone formation. Controlling these cell numbers is a promising strategy to increase bone regeneration.


Introduction
Bone defects attributed to trauma, tumors, or inflammation are still a challenge in dentistry, orthopedics, and plastic surgery [1,2]. Several techniques using autogenous bone, biomaterials,

Characteristics of Sponges
LS-Gs containing 12.42 EU/mg of LPS were prepared as reported previously [17]. Our previous study revealed that LS-G could retain LPS in similar bone defects for at least three weeks. Both LS-G and LS-G+EGCG possessed irregular macropores (Figure 2A,B). Both LS-G and LS-G+EGCG showed no evidence of direct cytotoxic effects for ≤3 days in vitro ( Figure 2C-E).

Hematoxylin-Eosin Staining for Bone Defects
To evaluate host reactions after surgery with or without the sponges, bone defects were stained with hematoxylin-eosin (H-E) ( Figure 3). Several leucocytes in the defects were identified in the defect treated with LS-G and LS-G+D+Q one week after surgery, but not in the defects without LS-G ( Figure 3B). Fewer leucocytes were detected in the defects treated with LS-G+EGCG.

Characteristics of Sponges
LS-Gs containing 12.42 EU/mg of LPS were prepared as reported previously [17]. Our previous study revealed that LS-G could retain LPS in similar bone defects for at least three weeks. Both LS-G and LS-G+EGCG possessed irregular macropores (Figure 2A,B). Both LS-G and LS-G+EGCG showed no evidence of direct cytotoxic effects for ≤3 days in vitro ( Figure 2C-E).

Hematoxylin-Eosin Staining for Bone Defects
To evaluate host reactions after surgery with or without the sponges, bone defects were stained with hematoxylin-eosin (H-E) ( Figure 3). Several leucocytes in the defects were identified in the defect treated with LS-G and LS-G+D+Q one week after surgery, but not in the defects without LS-G ( Figure 3B). Fewer leucocytes were detected in the defects treated with LS-G+EGCG. chemically modified with EGCG. (C-E) Evaluation of cytotoxicity of LS-G and LS-G+EGCG in vitro. The osteoblastic cell line UMR106 was treated with or without the sponges for three days. Control, without the sponges. (C) Live and dead cell staining. Green, live cells; red, dead cells. (D) Apoptosis analysis. Green, apoptotic cells; blue, DAPI (4′,6-diamidino-2-phenylindole) staining representing nuclei. (E) Cell counting kit (CCK-8) colorimetric assay. Data are expressed as mean ± standard deviation (SD), n = 4. N.S., not significant.

Increase in SIPS Cell after LS-G Implantation
Although a universal marker solely expressed in senescent cells has not been identified, SA-βgal, p16, and p21 are widely used biomarkers to determine cellular senescence [11,19]. To identify senescent cells in bone defects, histological staining with SA-β-gal and immunofluorescence staining with p16 and p21 antibodies were performed one ( Figure 4) and four weeks ( Figure 5) after surgery on the defects. The staining levels of SA-β-gal, p16, and/or p21 in defects treated with LS-G remained strong for up to four weeks ( Figure 4; Figure 5) after surgery. Both EGCG and D+Q apparently attenuated the staining levels of p16 and p21 in the defects for up to four weeks. Our results suggest

Increase in SIPS Cell after LS-G Implantation
Although a universal marker solely expressed in senescent cells has not been identified, SA-β-gal, p16, and p21 are widely used biomarkers to determine cellular senescence [11,19]. To identify senescent cells in bone defects, histological staining with SA-β-gal and immunofluorescence staining with p16 and p21 antibodies were performed one ( Figure 4) and four weeks ( Figure 5) after surgery on the defects. The staining levels of SA-β-gal, p16, and/or p21 in defects treated with LS-G remained strong for up to four weeks ( Figure 4; Figure 5) after surgery. Both EGCG and D+Q apparently attenuated the staining levels of p16 and p21 in the defects for up to four weeks. Our results suggest that the implantation of LS-G induced SIPS cells, while local and systemic administration of EGCG and D+Q, respectively, successfully reduced SIPS in bone defects.

Histomorphometric Analysis of Newly Formed Bone
To confirm whether the elimination of senescent cells from bone defects altered bone regeneration, we histomorphometrically analyzed bone defects using microcomputed tomography (µCT) and H-E staining ( Figure 3; Figure 6). The use of both EGCG and D+Q markedly elevated the radiopacity of bone defects ( Figure 6A,B). H-E staining showed that the high radiopacity was the newly formed bone ( Figure 3B,C).

Histomorphometric Analysis of Newly Formed Bone
To confirm whether the elimination of senescent cells from bone defects altered bone regeneration, we histomorphometrically analyzed bone defects using microcomputed tomography (μCT) and H-E staining (Figure 3; Figure 6). The use of both EGCG and D+Q markedly elevated the radiopacity of bone defects ( Figure 6A,B). H-E staining showed that the high radiopacity was the newly formed bone ( Figure 3B,C).

Oxidation in the Bone Defects
Oxidative stress plays a significant role in DNA damage, potentially initiating cellular senescence [9,21]. To confirm the detailed mechanisms underlying the reduction of senescent cells by EGCG and D+Q, we performed immunofluorescence staining of the defects using antibodies for 4-HNE (one and four weeks after surgery (Figure 7)), which is a biomarker for identifying oxidation in tissue [22]. The defects treated with LS-G alone and LS-G+D+Q showed strong 4-HNE staining one week after implantation (Figure 7). Administration of EGCG significantly reduced the staining levels of 4-HNE, but administration of D+Q did not. Approximately four weeks after surgery, the staining level of 4-HNE was similar in all the defects.

Oxidation in the Bone Defects
Oxidative stress plays a significant role in DNA damage, potentially initiating cellular senescence [9,21]. To confirm the detailed mechanisms underlying the reduction of senescent cells by EGCG and D+Q, we performed immunofluorescence staining of the defects using antibodies for 4-HNE (one and four weeks after surgery (Figure 7)), which is a biomarker for identifying oxidation in tissue [22]. The defects treated with LS-G alone and LS-G+D+Q showed strong 4-HNE staining one week after implantation (Figure 7). Administration of EGCG significantly reduced the staining levels of 4-HNE, but administration of D+Q did not. Approximately four weeks after surgery, the staining level of 4-HNE was similar in all the defects.

Discussion
In this study, we demonstrated that implantation of LS-G induced SIPS cells in critical-sized bone defects in rat calvaria. The use of EGCG locally or D+Q systemically attenuated the development of SIPS cells in the defects. This decrease in SIPS cells effectively enhanced bone formation by LS-G for ≤4 weeks.
Thus far, few studies have reported that gelatin potentially induces inflammatory reactions in host tissues [23], while the early inflammatory reaction is augmented after implantation of hydrogel containing a reagent-grade gelatin [24]. Zhao et al. showed that proinflammatory cytokine production increased in the presence of both gelatin and NF-κB activation [23]. This discrepancy might be due to the difference in the bonding between LPS and gelatin. Recently, we identified that reagent-grade gelatin contains a small amount of LPS [17]. We fabricated vacuum-heated gelatin sponges (LS-G) [17] because dehydrothermal treatment using vacuum heating is known to enhance ester bonding between carboxyl and hydroxyl groups of molecules [25]. The vacuum heating technique has successfully enabled LS-Gs to sustain the release of LPS, thereby inducing more cellular senescence in defects for three weeks than that by gelatin sponge lacking ester bonding between LPS and gelatin [17]. The amount of LPS used in the LS-G in defects (28.98 pg per defect) is far lower than the LD50 (3 mg/kg) [26]. LS-G contains 345-17,253 times less LPS than other doses applied previously in vivo [27][28][29]. To confirm the reproducibility of our previous study, we used same the LS-Gs, the material that could result in numerous SIPS cells in defects for four weeks. The results support the evidence that the use of dehydrothermal treatments causing physical cross-linking between microbial components and materials is a usable technique to fabricate experimental models of cellular senescence and chronic inflammation.
Two different techniques to eliminate SIPS cells ( Figure 1B) were adopted in the present study because: (1) EGCG isolated from green tea is a well-known polyphenol, which has anti-inflammatory [30] and anti-oxidant properties [31]. It has also been reported to suppress cellular senescence in vitro [32,33]. Recent studies showed that the chemical modification of gelatin with EGCG elicits greater pharmacological effect and bone formation than a simple mix of EGCG and gelatin [34,35]. Therefore, we hypothesized that the local administration of EGCG would also suppress SIPS cells in vivo. (2) Oral administration of D+Q directly successfully reduces the number of senescent cells in vivo [20]. Moreover, various studies have validated the function of D+Q as a senolytic in the fields of longevity, arteriosclerosis, obesity-related metabolic disfunction, and bone loss, among others [18,20,36,37]. As

Discussion
In this study, we demonstrated that implantation of LS-G induced SIPS cells in critical-sized bone defects in rat calvaria. The use of EGCG locally or D+Q systemically attenuated the development of SIPS cells in the defects. This decrease in SIPS cells effectively enhanced bone formation by LS-G for ≤4 weeks.
Thus far, few studies have reported that gelatin potentially induces inflammatory reactions in host tissues [23], while the early inflammatory reaction is augmented after implantation of hydrogel containing a reagent-grade gelatin [24]. Zhao et al. showed that proinflammatory cytokine production increased in the presence of both gelatin and NF-κB activation [23]. This discrepancy might be due to the difference in the bonding between LPS and gelatin. Recently, we identified that reagent-grade gelatin contains a small amount of LPS [17]. We fabricated vacuum-heated gelatin sponges (LS-G) [17] because dehydrothermal treatment using vacuum heating is known to enhance ester bonding between carboxyl and hydroxyl groups of molecules [25]. The vacuum heating technique has successfully enabled LS-Gs to sustain the release of LPS, thereby inducing more cellular senescence in defects for three weeks than that by gelatin sponge lacking ester bonding between LPS and gelatin [17]. The amount of LPS used in the LS-G in defects (28.98 pg per defect) is far lower than the LD 50 (3 mg/kg) [26]. LS-G contains 345-17,253 times less LPS than other doses applied previously in vivo [27][28][29]. To confirm the reproducibility of our previous study, we used same the LS-Gs, the material that could result in numerous SIPS cells in defects for four weeks. The results support the evidence that the use of dehydrothermal treatments causing physical cross-linking between microbial components and materials is a usable technique to fabricate experimental models of cellular senescence and chronic inflammation.
Two different techniques to eliminate SIPS cells ( Figure 1B) were adopted in the present study because: (1) EGCG isolated from green tea is a well-known polyphenol, which has anti-inflammatory [30] and anti-oxidant properties [31]. It has also been reported to suppress cellular senescence in vitro [32,33]. Recent studies showed that the chemical modification of gelatin with EGCG elicits greater pharmacological effect and bone formation than a simple mix of EGCG and gelatin [34,35]. Therefore, we hypothesized that the local administration of EGCG would also suppress SIPS cells in vivo.
(2) Oral administration of D+Q directly successfully reduces the number of senescent cells in vivo [20]. Moreover, various studies have validated the function of D+Q as a senolytic in the fields of longevity, arteriosclerosis, obesity-related metabolic disfunction, and bone loss, among others [18,20,36,37]. As can be seen in Figures 4 and 5, the two techniques attenuated the generation of SIPS cells induced by LS-G implantation up to four weeks.
SA-β-gal is a common biomarker for detecting cellular senescence [11], but its reliability in bone biology is still controversial [38]. Senescent cells and hyperfunctional macrophages are known to be positive for SA-β-gal staining [39], as β-gal staining reflects hyperfunctional lysosomes [40]. Therefore, we confirmed the presence of SIPS with SA-β-gal staining and the well-known senescence biomarkers, p16 and p21. The staining levels of p16 or p21 were approximately equal to those of SA-β-gal in the defects.
LPS causes inflammation and generates reactive oxygen species, resulting in oxidative stress [41]. Oxidative stress promotes cellular senescence [4]. In our in vivo study, early inflammatory reactions augmented by LS-G were distinguishable by the presence of leucocytes in the sponges one week after implantation; however, this effect was attenuated by Week 4. Coincident with this inflammatory reaction, the staining levels of 4-HNE increased at one week but not at four weeks. Meanwhile, although LS-G significantly increased SIPS cells in the defects after one week, these SIPS remained in the defects for four weeks. Additionally, we determined that the used sponges caused negligible cytotoxic effects on the osteoblastic cell line UMR106 in vitro ( Figure 2C-E). These results suggest that, although the inflammatory reaction that caused oxidation may be temporally and partially associated with the induction of SIPS cells at early stage, its effect was limited. Direct feeble and lasting stimulation by residual LPS of LS-G is likely to cause cellular senescence.
The staining levels of 4-HNE in the defects indicated that the EGCG and D+Q treatments caused different host reactions: EGCG reduced oxidation after one week, but D+Q did not (Figure 7). Although excessive oxidation is known to attenuate osteogenesis [42], there was no strong attenuation of bone formation in defects treated with LS-G+D+Q ( Figure 6). Both LS-G+EGCG and LS-G+D+Q could induce superior bone regeneration compared with LS-G alone. Simultaneously, both treatments successfully reduced the staining levels of SA-β-gal, p16, and p21 after sponge implantation. These results suggest that SIPS cells rather than direct oxidation stress are likely to play a deleterious role in bone regeneration in defects.
In the present study, we could partially verify the relation between SIPS cells induced by LPS stimulation and bone regeneration using a rat calvarium model. However, our study had several limitations. Details regarding the type of senescent cells that are present in or around the sponges are still unknown. It is possible that the species and age of animals as well as the implantation sites alter these reactions. Moreover, we could not determine the exact role of the SASP from induced SIPS cells in the bone formation and remodeling processes. Additional experiments would therefore be essential to deepen the understanding of the mechanisms underlying the relationship between LPS-induced SIPS cells and bone regeneration in vivo. However, our report does provide insights for the advancement of bone regenerative medicine.

Characterization of the Sponges
Reagent-grade type A gelatin from pig skin (Cat. No. G2500) containing LPS was purchased from Sigma-Aldrich (St. Louis, MO, USA). LS-G and LS-G+EGCG were prepared as described before [17]. To produce LS-G, 100 mg gelatin were dissolved in 10 mL MilliQ water at 70 • C. The resulting solution was frozen and freeze-dried using DC800 (Yamato Co., Ltd., Tokyo, Japan) in ϕ 5 mm silicon tubes. The prepared gelatin sponges were then vacuum-heated at 150 • C for 24 h and a gauge pressure of −0.1 MPa to promote physical cross-linking using ETTAS AVO-250NS (AS ONE, Osaka, Japan). LPS level in sponges was evaluated utilizing a ToxinSensor Chromogenic LAL Endotoxin Assay Kit (L00350, GenScript Biotech Inc., Piscataway, NJ, USA) according to the manufacturer's instructions, as in our previous study [17]. We previously demonstrated that LS-G could retain LPS for at least three weeks in a similar bone defect model to the one used in this study [17]. For LS-G+EGCG, an aqueous synthesis method was applied before lyophilization and vacuum heating. Briefly, 100 mg gelatin, 0.07 mg EGCG, 69.2 mg 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium chloride (DMT-MM), and 27.5 µL N-methylmorpholine (NMM) were dissolved and mixed in 5 mL MilliQ water at 23 • C. The aforementioned EGCG dose was selected based on our previous results of superior bone formation (pharmacological effect) [43]. After dialysis to remove the residual reagents using Spectra/Por7 MWCO 1000 (Spectrum Labs, CA, USA) in water in the dark, the resulting solution was diluted to 10 mL with MilliQ water, frozen, lyophilized, and vacuum-heated. The sponges were vacuum-heated at 150 • C for 24 h. The morphology of the sponges was analyzed using a stereomicroscope (SZX12, Olympus Inc., Tokyo, Japan) and field-emission scanning electron microscope (FE-SEM, S-4800; Hitachi, Tokyo, Japan).

Cytotoxic Effect of the Sponges
Rat osteoblastic cell line UMR106 (CRL-1661, American Type Culture Collection, Manassas, VA, USA) was seeded in 48-well plates at a density of 1 × 10 4 /well and cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and 1% antibiotics. One day after cell seeding, the cells were treated with or without 0.5 mg of sponges in each well for three days. The wells without sponges were considered negative control. The cytotoxic effect of the sponges was estimated using LIVE/DEAD Viability/Cytotoxicity Kit (Molecular Probes, Eugene, OR, USA), CellEvent Caspase-3/7 Green Detection Reagent (Thermo Fisher Scientific, Waltham, MA, USA) with DAPI Fluoromount-G (4 ,6-diamidino-2-phenylindole; SouthernBiotech, Birmingham, AL, USA), and a cell counting kit (CCK-8) colorimetric assay (Dojindo Molecular Technologies, Kumamoto, Japan) according to the manufacturer's instructions.

Animal Experiments
As senescent cells accumulate with age [44], eight-week-old Sprague-Dawley rats were used to minimize the misidentification of age-related senescent cells as SIPS cells. The animal experiments in this study were conducted with the permission of and in accordance with the guidelines approved by the local ethics committee of Osaka Dental University (Approval No. 19-04002, 27 May 2019). The male Sprague-Dawley rats were obtained from SHIMIZU Laboratory Supplies Co. (Kyoto, Japan). The rats were housed in a ventilated room with a 12 h light/dark cycle during the experiments. At the center of their calvaria, 9-mm-sized bone defects, commonly recognized as the critical size defect, were created using a trephine bar as reported previously [45]. During surgery, the rats were anesthetized with an intraperitoneal injection of a mixture of medetomidine hydrochloride (0.15 mg/kg; Domitor; Zenoaq, Fukushima, Japan), midazolam (2 mg/kg; Midazolam Sandoz, Sandoz KK, Yamagata, Japan), and butorphanol tartrate (2.5 mg/kg; Vetorphale, Meiji Seika Parma Co., Ltd., Tokyo, Japan). They were divided into four groups with four rats per group ( Figure 1A): Group 1 (control): no implant; Group 2 (control): implantation of LS-G; Group 3: implantation of LS-G+EGCG; and Group 4 (LS-G+D+Q): oral administration of D+Q and implantation of LS-G. One rat treated with LS-G+D+Q for four weeks unexpectedly died at three weeks during the experimental period. D and Q were diluted at a ratio of 1:10 in 10% PEG400 in water (Cat. No. 25322-68-3, FUJIFILM Wako Pure Chemical Co., Osaka, Japan) and orally administered by gavage once a week at 5 (D) and 50 mg/kg (Q). Ten sponges (one sponge:

Histomorphometric Analysis
At one and four weeks after surgery, the rats were euthanized using isoflurane. The calvaria was retrieved and fixed with 4% paraformaldehyde in 0.1 M phosphate buffer for histological and morphometric analyses described below. Bone formation in bone defects was quantified using µCT analysis (SMX-130CT; Shimadzu, Kyoto, Japan or inspeXio SMX-225CT; Shimadzu) and histological staining. Each sample was scanned at 55 kV and 90 µA using a voxel size of 30 × 30 × 30 µm 3 . The 3D morphology of the calvaria was reconstructed using TRI/3D bone software (RATOC Systems Engineering Co., Ltd., Tokyo, Japan). To estimate bone formation and quality in bone defects, the following parameters were quantified: bone volume (BV)/total volume (TV); bone mineral content (BMC)/TV; and BMC/BV. Cylindrical phantoms containing hydroxyapatite (200-1550 mg/cm 3 ) were used to calculate the BMC representing calcified bone tissue.

Histological and Immunofluorescent Staining
After scanning calvaria with µCT, 4 µm-thick, non-decalcified sections were prepared from each sample using the Kawamoto method [46] and a cryotome (Leica CM3050S; Leica Biosystems Inc., Buffalo Grove, IL, USA) for histological and immunohistological observation. The sections were stained with H-E to estimate the presence of newly formed bone or leucocytes (inflammatory response) using HS All-in-one Fluorescence Microscope (BZ-9000, Keyence, Osaka, Japan). Quantitative calculations of newly formed bone were carried out using Photoshop CC2019 (Adobe Systems Inc., San Jose, CA, USA) and NIH ImageJ version 1.52a (Bethesda, MD, USA). The percentage of the newly formed bone in the defect was calculated as follows: (newly formed bone area/total tissue area in defect) × 100.
To evaluate SIPS, sections were stained with SA-β-gal using Senescence Detection Kit (Cat.

Statistical Analysis
Statistical significance of mean values was assessed using GraphPad Prism 8 (GraphPad Software Inc., San Diego, CA, USA) and one-way analysis of variance, followed by Tukey-Kramer test to determine significance.

Conclusions
In this study, we demonstrated that the reduction in the number of SIPS cells using two different techniques enhanced bone regeneration. Our results suggest that SIPS cells are at least partially associated with the modulation of bone regeneration. These data possibly provide basic knowledge for the development of promising new biomaterials and bone regeneration therapies.

Funding:
The study was in part financially supported by funding from a Grant-in Aid for Scientific Research (16K11667 and 18H02986) from the Japan Society for the Promotion of Science.