Introduction

Diabetes mellitus (DM) affected 463 million people worldwide in 2019, and approximately USD 760 billion in global health expenditure was spent on treating DM [1]. DM comprises metabolic disorders characterized by hyperglycemia related to inadequate insulin production or improper response of the cells to insulin, and the major complications include poor wound healing, cardiovascular diseases, retinopathy, and nephropathy [2]. As hyperglycemia causes the shift in subgingival microbiota and the accumulation of advanced glycation end-products (AGEs) activates receptor for AGE (RAGE) to increase proinflammatory cytokine secretion in the periodontium [3,4,5], periodontal destruction deteriorates, wound healing is retarded, and the outcome of periodontal regeneration is impaired [6, 7]. The American Diabetes Association defined periodontal disease as a common comorbidity of DM in 2018 [8].

Stem cell therapy has been proposed to regenerate bony defects in compromised environments [9]. Adipose-derived stem cells (ADSCs), isolated by liposuction, can be collected in large quantities with low donor site morbidity and a high yield of stem cells and have shown potential as a promising source of stem cells [10]. Previous evidence supports the capability of ADSCs to regenerate skin and osseous defects and treat wounds with inferior healing potential [11,12,13].

One critical aspect of stem cell therapy is to maintain stemness. While signals from the microenvironment play important roles in regulating the differentiation capability of stem cells [14], the establishment of self-assembled cell spheroids to mimic native cellular morphology, cell–cell contact, and cell–extracellular matrix interactions in a three-dimensional (3D) microenvironment has been proven to retain the properties of ADSCs [15]. Gelatin-based hydrogels appeared to be suitable vehicles to deliver ADSC spheroids (ADsp) in vivo because they are biocompatible and biodegradable and can rapidly absorb biological fluid to occupy defects. However, the cross-linking process, frequently achieved by UV irradiation or chemical reagents, exhibits certain cytotoxicity and might influence the viability of encapsulating cells [16, 17]. A recent study reported that microbial transglutaminase (mTG)-crosslinked gelatin hydrogels showed low cytotoxicity, preserved the stemness of ADsp, and exhibited excellent injectability and tissue response [18].

The hypothesis of this study was that ADsp embedded in mTG (ADsp-mTG) promoted the healing of diabetic periodontal wounds. Experimental gingival mucosal wounds and calvarial osseous defects of rats were created to represent the mucosal and osseous compartments of the periodontium and craniofacial defects. The outcome could serve as a reference for ADSC application in the diabetic periodontal wound and craniofacial defect regeneration.

Methods

Information on the manufacturers of materials, formulations of media, mucosal wound healing indices, the conditions of real-time PCR, the sequence information of primers and probes, and protocols of immunohistochemical staining is listed in the Appendix.

Ethical statements

The harvest of ADSCs from humans was conducted in accordance with the Declaration of Helsinki in 1975, as revised in 2013 [19], and approved by the Institutional Review Board of National Taiwan University Hospital (NTUH) under protocol no. 201303038RINB. The animal study was conducted in accordance with Basel Declaration in 2010 and was approved by the Animal Care and Use Committee of National Taiwan University under protocol no. 20190195.

Characterization of ADSCs

Isolation of ADSCs

ADSCs were isolated from the subcutaneous fat tissue of four non-smoking, non-diabetic female patients (aged 32–57 years, BMI 21.0–26.6) planned to receive lipoaspiration for aesthetic plastic surgery at NTUH as previously described [20], and all of them signed informed consent prior to the harvest procedure. Following lipoaspiration, subcutaneous adipose tissue was obtained, extensively washed, and red blood cells were lysed. The remaining adipose tissue was digested with collagenase, followed by filtration and centrifugation to obtain the stromal vascular fraction (SVF). Cells in the SVF were subsequently incubated at 37 °C with 5% CO2 under 99% humidity and cultured in a growth medium. The medium was renewed every 2–3 days and passaged until 90% confluence was reached, and cells at passage 3 were used for all subsequent assessments. According to the results from our previous investigation, > 99% of cells were positive for stem cell surface markers including CD44, CD73, CD90, and CD166, and showed adipogenic and osteogenic potential [20].

Expression of pluripotent markers

The expression levels of pluripotency-associated transcription factors, including SRY-box transcription factor-2 (Sox2), octamer-binding transcription factor-4 (Oct4), core-binding factor alpha-1 (Cbfa1), and Nanog homeobox, in ADSCs were analyzed. A human foreskin fibroblast cell line (Hs68) was used as the negative control, and an induced bone marrow mesenchymal stem cell line (iBMSC) was used as the positive control. All examined cells were seeded on 6-well dishes at 1 × 105 cells/well and incubated with a growth medium for 24 h. RNA was isolated using an RNA isolation kit and reversely transcribed to cDNA using a cDNA synthesis kit to run real-time PCR, and the TaqMan probes of GAPDH (housekeeping gene) and pluripotent markers were used. The level of gene expression was calculated using the comparative CT method according to the level of GAPDH and was further normalized to the expression level of Hs68. All experiments were performed in triplicate.

Trilineage differentiation

To evaluate the capability of adipogenic, osteogenic, and chondrogenic differentiation, cells (Hs68, iBMSCs, and ADSCs) were cultured in adipogenic, osteogenic, and chondrogenic induction medium. After 2 weeks, the cell monolayer was fixed in 4% paraformaldehyde and stained with Oil Red to observe lipid droplets for adipogenesis, stained with Alizarin Red to observe mineralized matrix apposition for osteogenesis, and stained with Alcian Blue to observe cartilage-specific proteoglycans for chondrogenesis.

Preparation and characterization of ADsp-mTG

Preparation of ADsp-mTG

To form ADSC spheroids (ADsp), the agarose microwell plates were immersed in 75% ethanol for 1 h and transferred to PBS before use. Each agarose microwell plate was placed into a 12-well plate with growth medium, and the entrapped bubbles were removed by centrifugation. ADSCs were seeded at 1 × 105 cells/well, centrifuged to ensure that ADSCs were captured in the microwells, and then incubated for 3 days at 37 °C in 5% CO2 with 1 mL growth medium/plate. After ADsp were formed, ADsp were detached and passed through a cell strainer to collect uniform ADsp. Gelatin was dissolved in growth medium at 70ºC, followed by sterile filtration through 0.22-µm filters. A 1 mL/well gelatin/microbial transglutaminase solution (6% gelatin, 60 U g−1 microbial transglutaminase) mixed with ADsp was loaded to achieve a final density of 1.2 × 106 cells/100 μL (approximately 3072 ADsp/100 μL).

Behaviors of ADsp in mTG

The morphology and dynamic changes of ADsp in the medium and mTG were observed under a light microscope on days 1–4. The area and Feret diameters of ADsp were measured using ImageJ. The trilineage differentiation of ADsp in mTG was assessed by loading ADsp-mTG into 6-well plates, which were incubated with the appropriate induction medium for 14 days and stained as described in “Trilineage differentiation” section.

Preclinical validation

The induction of diabetes and randomization

Male Sprague–Dawley rats (weighing 250 g) were purchased from BioLASCO Taiwan Co. Ltd. (Taipei, Taiwan) and were housed in a climate-controlled room (21 °C) on a 12:12-h light/dark cycle with free access to food and water. Several of them received a one-time intraperitoneal injection of streptozotocin (STZ) (65 mg/kg) diluted in citrate buffer (0.05 M, pH 4.5) to induce diabetes. Fasting blood glucose (FBG) was checked using a glucose meter, and glycated hemoglobin (HbA1c) was checked using an HbA1c analyzer before and after 3 weeks of induction [7]. FBG > 250 mg/dL was considered diabetic. Rats without STZ injection (ND rats) served as controls, and FBG and HbA1c were checked at the time of wound/defect creation.

The mucosal wound model

The design of the model is illustrated in Fig. 1A. Twenty-four rats with successful diabetes induction (DB) and 16 ND rats were utilized. The surgical procedures were covered by general anesthesia using an intraperitoneal injection of 0.1 mL/kg zolazepam–tiletamine and 0.5 mL/kg xylazine. In DB rats, bilateral 1 × 5 mm full-thickness gingival wounds on the palatal gingival margin of the maxillary first and second molars were surgically created, and the wounds were randomly assigned to the following treatment: untreated (DB-CL), ADsp alone (DB-AS), or ADsp-mTG (DB-AT). The randomization was determined by a simple draw by a blinded administrator and was minorly adjusted to ensure that the contralateral site received a different treatment assignment, and in each wound, 10 µL hydrogel was applied to ensure complete coverage. In ND rats, gingival wounds were created unilaterally without any further treatment (ND-CL). Animals were euthanized using carbon dioxide on days 7 and 28 (n = 8/treatment/time point), and the maxillae containing the examined wounds were harvested upon euthanasia.

Fig. 1
figure 1

Designs of animal models. A The mucosal wound model. B The osseous defect model

Full size image

Gross observation of mucosal wounds

The photographs of mucosal wounds were taken using a digital camera on days 4, 7, 14, and 28. The healing status was assessed using the clinical healing index (CHI) and early wound healing score (EHS) [21, 22].

Histologic and immunohistochemical assessments of mucosal wounds

All harvested specimens were fixed in 10% buffered formalin for 3 days and then decalcified with 12.5% EDTA (pH 7.4) for 4 weeks, embedded in paraffin, and cut into 5-mm-thick sections. Masson’s trichrome staining was performed to outline the dynamics of wound healing. Histologic healing was divided into the following categories: no epithelization with a sloughing surface; full epithelization but with an epithelial cleft and reduced thickness; and full epithelization with normal consistency and contour. The signs of keratinization and cell proliferation were further evaluated by immunohistochemical staining for cytokeratin-10 (K-10) and proliferating cell nuclear antigen (PCNA). Cells expressing PCNA in the mucosal wounds were quantified in three randomly selected areas per slide at 400 × magnification, and the results were presented as the ratio of PCNA-positive cells to the total number of cells in the investigated area. All stained slices were observed under a light microscope, and images were acquired using a digital camera equipped with an imaging system.

The osseous defect model

The design of model is illustrated in Fig. 1B. A total of 56 animals, including 40 DB rats and 16 ND rats, were utilized in this part. The surgical procedures were covered by general anesthesia as described in “The mucosal wound model” section. Following a midline incision and blunt dissection of underlying tissue, the calvarium was exposed. Bilateral 5-mm-diameter cylinder calvarial defects were surgically created using a dental trephine bur. The defects were randomly assigned to the following treatments: untreated (DB-CL), ADsp alone (DB-AS), ADsp-mTG (DB-AT), pretreated ADsp alone (DB-PAS), or pretreated ADsp-mTG (DB-PAT). The defects were minorly adjusted to ensure that the contralateral site received a different treatment assignment. As a previous investigation reported that alkaline phosphatase activity, an early sign of osteogenesis, was significantly elevated by incubating ADSCs in the osteoinductive environment for 5–7 days [23], to direct the differentiation of ADSC toward osteogenic lineage, ADSC pretreatment was performed by incubating ADSCs in the osteogenic induction medium for 10 days before forming spheroids. The osteogenic differentiation level of the pretreated ADSCs was examined by the Alizarin Red assay described in “Trilineage differentiation” section and was quantified using an ELISA reader at a wavelength of 405 nm. In each defect, 20 µL hydrogel was applied to ensure complete coverage. In ND rats, the defects were created unilaterally without any further treatment (ND-CL). Wounds were closed by reflex wound clips. Animals were euthanized on days 7 and 28 (n = 8/treatment/time point), and the calvarias containing osseous defects were harvested upon euthanasia.

Micro-CT assessments of the osseous defects

The harvested calvarias were fixed in 10% buffered formalin for 3 days and then examined by a micro-CT scanner with a voxel size of 18 μm. The entire osseous defects were selected as the region of interest (ROI). The micromorphometric bone parameters within the ROI, including relative bone volume (BV/TV), trabecular thickness (Tb. Th), and trabecular number (Tb.N), were calculated using a micro-CT image analysis software.

Histologic and immunohistochemical assessments of the mucosal wounds

After micro-CT imaging, the specimens were processed as described in “Histologic and immunohistochemical assessments of mucosal wounds” section and were stained with hematoxylin and eosin for descriptive histology. Histomorphometric analysis was performed using ImageJ, and new bone (NB) bridging, defined as the percentage of linear bone formation to the entire defect length, and defect fill, defined as the percentage of NB area to the entire defect area, were assessed.

The signs of osteogenic differentiation and cell proliferation in the defects were further evaluated by immunohistochemical staining of bone sialoprotein (BSP) and PCNA. Cells expressing PCNA in the osseous defects were quantified in three randomly selected areas, two at the borders and one at the center of the defect, at × 400 magnification. The assessment protocol and data presentation were the same as described in “Histologic and immunohistochemical assessments of mucosal wounds” section.

Statistical analysis

Statistical analysis was performed using statistical software. Unpaired t test was used to compare the concentration of Alizarin Red between the cultures of unpretreated and the cultures of pretreated ADSCs in vitro. Paired t-test was used to compare the levels of AC sugar and HbA1c before and after STZ injection. In DB animals, one-way ANOVA and Tukey’s post hoc tests were used to compare the differences among the treatment assignments. For CHI and EHS, a nonparametric Kruskal–Wallis test was used. The data are expressed as the means ± standard deviation, with a p value of less than 0.05 considered statistically significant.

Results

Characterization of ADSCs

The expression levels of Nanog, Oct-4, and Cbfa1 were significantly higher in ADSCs than in Hs68 (Fig. 2A). After an induction period of 14 days, there was no sign of adipogenesis, osteogenesis, or chondrogenesis in the cultures of Hs68. In the adipoinductive cultures, lipid droplets slightly accumulated in iBMSCs but obviously accumulated in ADSCs (Fig. 2B). Mineralized nodules and cartilage-specific proteoglycans were deposited in osteoinductive and chondroinductive cultures of iBMSCs and ADSCs.

Fig. 2
figure 2

In vitro characterization of ADSCs and ADsp-mTG. A Pluripotent gene expression in ADSCs after 24 h of seeding. B The trilineage differentiation of ADSCs by Oil Red O (adipogenesis), Alizarin Red (osteogenesis), and Alcian Blue (chondrogenesis) at day 14. Scale bar: 100 µm. C The spreading of ADSCs from spheroids embedded in mTG. D The trilineage differentiation of ADsp in mTG at day 14. Scale bar: 100 µm. There is no downstream processing or averaging to adjust the resolution of the microscopic images. (Significant difference compared to Hs68: *p < 0.05, **p < 0.01, ***p < 0.001)

Full size image

Characterization of ADsp-mTG

ADsp successfully formed with an area and Feret diameter of 0.0051 ± 0.0010 mm2 and 0.093 ± 0.010 mm, remaining spheroid in shape for 24 h, with an area and Feret diameter of 0.0073 ± 0.0025 mm2 and 0.119 ± 0.029 mm without mTG and 0.0059 ± 0.0023 mm2 and 0.105 ± 0.024 mm with mTG (Fig. 2C). Without the support of mTG, ADsp spread widely at 48 h. Conversely, when ADsp was encapsulated in mTG, the shapes were maintained for 96 h (Fig. 2C; lower panel), with an area and Feret diameter of 0.0051 ± 0.0015 mm2 and 0.099 ± 0.015 mm at 48 h and 0.0055 ± 0.0017 mm2 and 0.103 ± 0.016 mm at 96 h. Furthermore, successful adipo-, osteo-, and chondro-induction were noted in ADsp-mTG at day 14 (Fig. 2D).

The induction of diabetes in rats

In DB rats, FBG and HbA1c were 93.58 ± 13.34 mg/dL and 5.18 ± 0.58% before STZ injection and increased significantly to 419.27 ± 87.83 mg/dL and 7.53 ± 1.31% after 3 weeks of STZ injection (p < 0.001 for both; Fig. 3A, B). In ND rats, FBG and HbA1c were 113.20 ± 10.12 mg/dL and 5.37 ± 0.62% at the time of wound/defect creation.

Fig. 3
figure 3

Gross observation of diabetic mucosal wound healing within 28 days. A Fasting plasma glucose (FBG) of ND and DB rats. B Fasting plasma glucose (FBG) of ND and DB rats. C Representative photographs of mucosal wounds. Scale bar: 2 mm. D mCHI (range 0–4). E mEHS (range 0–10). (In (A, B), significant difference compared to the same animals before diabetic induction: ***p < 0.001; in (D, E), significant difference between DB-CL and DB-AS group: *p < 0.05, ***p < 0.001; significant difference between DB-CL and DB-AT group: ##p < 0.01)

Full size image

The effect of ADsp-mTG on diabetic oral mucosal wounds

Gross observation

In ND-CL group, wounds were occupied with newly formed tissue at day 4 and were apparently invisible from day 7. Wound dehiscence was noted until day 14 in DB-CL group and until day 7 in DB-AS group. In DB-AT group, wounds were covered with reddish tissue at day 4 and became invisible from day 14 (Fig. 3C).

ND-CL group showed higher scores in CHI and EHS than all groups in DB rats at days 7–14 (Fig. 3D, E). Among DB rats, compared with DB-CL group, CHI in DB-AS and DB-AT groups was greater, specifically at days 4 and 28 (Fig. 3D), and EHS in DB-AS and DB-AT groups was also greater at days 4–14 (Fig. 3E).

Histologic and immunohistochemical assessments

At day 7, wound healing was almost complete in the ND-CL group (Fig. 4A). In DB rats, inflammatory cell infiltration along with necrotic tissue adjacent to the tooth was noted. Compared with DB-CL group, the area occupied by necrotic tissue was smaller, and collagen bundle formation beneath the epithelium apparently increased in DB-AS and DB-AT groups. At day 28, the necrotic tissue disappeared, and epithelization was complete in all groups. However, the cleft of the epithelium and sequestrum, with low levels of inflammation, was occasionally seen in DB-CL, DB-AS, and DB-AT groups.

Fig. 4
figure 4

Histologic and immunohistochemical assessments of diabetic mucosal wound healing at days 7 and 28. A Representative images of Masson’s trichrome staining. Magnification: 40 × . Scale bar: 250 µm. B Representative images of immunohistochemical staining for cytokeratin-10 (K-10). Magnification: 100 × . Scale bar: 100 µm. C The ratio of proliferating cells (PCNA( +) cells) in the mucosal wound. D The histologic healing condition was categorized into three levels: incomplete epithelization (level 1), complete epithelization with epithelial cleft or reduced connective tissue thickness (level 2), and complete epithelization with normal consistency (level 3). Dashed lines indicate the edge of mucosal wounds. There is no downstream processing or averaging to adjust the resolution of the microscopic images. (Among groups with diabetes, significant difference compared to DB-CL: *p < 0.05, **p < 0.01, ***p < 0.001)

Full size image

At day 7, K-10, a marker of mucosal keratinization, was not expressed in DB-CL group but was expressed at the border of the wound in DB-AS group and was scattered toward the tooth surface in the epithelium of DB-AT group (Fig. 4B). At day 28, K-10 was evenly distributed in the outer epithelium but was not expressed at the sulcular epithelium or the bottom of the epithelial cleft, regardless of the treatment group (Fig. 4B). Compared with DB-CL group, the ratio of proliferating cells significantly increased in DB-AS and DB-AT groups at day 7 and slightly increased in DB-AT group at day 28 (Fig. 4C).

The categorized histologic healing condition is shown in Fig. 4D. At day 7, epithelization was seen in most specimens of ND-CL group but wound dehiscence was generally noted in DB rats. Among DB rats, epithelization was more commonly noted in DB-AT group. At day 28, although wound dehiscence without coverage of the epithelium was still noted in one specimen from DB-CL group, complete epithelization with adequate lamina propria formation was seen in > 50% of specimens in all groups.

The effect of ADSC-mTG on diabetic osseous defects

The osteogenic differentiation level of pretreated ADSCs in vitro

Compared with the cultures of unpretreated ADSCs, mineralized matrix was evidently deposited, and the concentration of Alizarin Red was significantly greater, in the cultures of pretreated ADSCs (p < 0.01; Fig. 5A, B).

Fig. 5
figure 5

Osteogenic differentiation of ADSCs after 10 days pretreatment in vitro and the micro-CT assessments of diabetic osseous defect healing at day 28. A Mineralized matrix deposition by Alizarin Red. Scale bar: 50 µm. B The concentration of Alizarin Red. C Representative slices from the images. Scale bar: 1 mm D Percentage bone volume (BV/TV) Tb.N. F Tb.Th. Dashed lines indicate the borders of osseous defects. (In (B), significant difference compared to the unpretreated ADSCs: **p < 0.01; in (D-F), among groups with diabetes, significant difference compared to DB-CL: *p < 0.05, **p < 0.01, ***p < 0.001)

Full size image

Micro-CT assessments

NB in the defects was not visualized at day 7 in all groups (data not shown). At day 28, NB was noted on the defect borders in all groups (Fig. 5C). Compared with the ND-CL group, BV/TV and Tb.N were inferior in DB-CL group (Fig. 5D, E). Among DB rats, BV/TV was significantly greater in DB-PAT group than in DB-CL, DB-AS, and DB-AT groups (Fig. 5D); Tb.N was significantly greater than that in DB-CL and DB-AS groups (Fig. 5E); and Tb.Th was significantly greater than that in DB-AS group (Fig. 5F).

Histologic and immunohistochemical assessments

At day 7, the defects were mainly occupied with nonmineralized granulation tissue in all groups (Fig. 6A). BSP, an early osteogenic marker, was lightly deposited at the defect peripheries in DB-CL, DB-AS, DB-AT, and DB-PAS groups and was widely spread in DB-PAT group (Fig. 6B, and see Appendix Figure for the images of DB-AS and DB-PAS groups). Compared with DB-CL group, the ratio of proliferating cells was significantly greater in DB-PAT group at days 7 and 28 (Fig. 6C). At day 28, NB formation from the peripheries toward the center of the defect was notable in all groups (Fig. 6A). Although the closure of the defect did not take place in any of the specimens, NB formation was more prominent in DB-PAT groups among all groups. BSP was deposited within NB, with heavier deposition in DB-PAT group (Fig. 6B). Compared with DB-CL group, defect filling was significantly greater in DB-AT and DB-PAT groups, and NB bridging was significantly greater in DB-PAT group (Fig. 6D).

Fig. 6
figure 6

Histologic and immunohistochemical assessments of diabetic osseous defect healing at days 7 and 28. A Representative images of hematoxylin and eosin staining. Magnification: 40 × . Scale bar: 1 mm. B Representative images of immunohistochemical staining for bone sialoprotein (BSP). Magnification: 100 × . Scale bar: 100 µm. C The ratio of proliferating cells (PCNA( +) cells) in the osseous defect at days 7 and 28. D Histomorphometric analysis of the defect fill and new bone (NB) bridging at day 28. Dashed lines indicate the borders of osseous defects. There is no downstream processing or averaging to adjust the resolution of the microscopic images. (Among groups with diabetes, significant difference compared to DB-CL: *p < 0.05, **p < 0.01, ***p < 0.001)

Full size image

Discussion

The present study demonstrated the stemness of ADSCs based on the pluripotent gene expression and trilineage differentiation capability (Fig. 2A-B). According to a previous study, encapsulating ADSC spheroids (ADsp) in mTG provided favorable injectability for delivering ADsp [18]. This configuration also maintained the 3D configuration to mimic the in vivo microenvironment and the differentiation properties of ADSCs (Fig. 2C, D), thereby increasing the regenerative potential for therapeutic use [15, 18]. These results supported ADsp-mTG as a good candidate for promoting the healing of mucosal wounds and osseous defects.

A previous study reported that keratinocyte and fibroblast migration and proliferation were impaired in diabetic animals [24], and similar outcomes were noted in the mucosal wounds of DB-CL group in the present study (Fig. 4B, C). ADSCs have been recommended to treat diabetic wounds according to their multilineage differentiation, chemotaxis promotion, extracellular matrix synthesis, and anti-inflammation capabilities [25]. While ADsp-mTG maintained the 3D configuration of cellular aggregates and stemness and improved the handling property and kept ADSCs in place, ADsp-mTG delivery further facilitated wound healing, and keratinization was noted in diabetic mucosal wounds (Figs. 3 and 4).

The retardation of wound repair in diabetic animals was also noted in the osseous defects (Figs. 5 and 6). However, applications of ADsp (i.e., DB-AS and DB-AT groups) showed limited effects (Figs. 5 and 6D), potentially due to the effects of diabetes on the activities of stem cells. Previous studies reported that stem cells from the diabetic environment presented inferior osteogenic and angiogenic potential but were apparently more adipogenic [26, 27]. Stem cells incubated with AGEs or in a high-glucose environment also exhibited reduced osteogenic ability [28, 29]. Due to the unfavorable osteogenic environment and lack of osteogenic signals in diabetes, pretreatment of ADSCs to direct differentiation toward osteogenesis may be required for treating diabetic osseous defects. In DB-PAT group, BSP, an osteogenic marker, was highly expressed at the early stage, whereby such distinction was not seen in DB-CL and DB-AT groups (Fig. 6B), and NB formation was significantly greater than that in DB-CL group at the later stage (Fig. 6). It is noteworthy that NB formation and defect fill in DB-PAT group were even greater than those in ND animals (Figs. 5 and 6). Taken together, pretreated ADsp-mTG recovered the osteogenic impairment influenced by diabetes and could be a viable application for treating diabetic osseous defects.

This study has limitations. First, STZ destroyed pancreatic islet β-cells and induced mostly type I DM in this study [30], but type II DM accounts for approximately 90% of clinical DM [1]. While FBG and HbA1c were significantly elevated following STZ injection, hyperglycemia and AGE disturbance of DB animals still caused comorbidities similar to clinical DM. Second, because the creation of a critical-sized periodontal defect in rats similar to that in humans is impractical, we assessed the mucosal and bony compartments separately by using oral mucosal wounds and calvarial osseous defects. However, as the calvarium develops differently from jaw bones [31] and the etiology of defect generation is different from periodontal defects, the results from this study should be interpreted carefully. Third, without protection from wound dressings or adhesives, mTG may be washed out rapidly in mucosal wounds to attenuate the influence of ADSCs. Furthermore, the osseous defect fill in this study was obviously lower than that in other relevant animal studies [32,33,34], and the potential reason was the lack of the osteoconductive framework provided by the scaffold. The combination of ADsp with osteoconductive scaffolds, such as 3D-printed hydroxyapatite-based scaffolds [35], may be beneficial to promote the efficiency of ADsp-mTG for osseous defect regeneration. Further investigations in more clinically relevant conditions, with adequate dressing/adhesive materials and osteoconductive scaffolds, are still necessary.

Conclusion

ADsp-mTG showed trilineage differentiation capability. ADsp-mTG accelerated diabetic oral mucosal wound healing, and osteogenically pretreated ADsp-mTG promoted diabetic calvarial osseous defect regeneration, validating the concept of ADsp-mTG application for facilitating the healing of diabetic periodontal wounds and craniofacial defects.