SHED Aggregate Derived Exosomes-shuttled miR-222 Promotes the Regenerative Properties of Periodontal Ligament Stem Cells

Background: Periodontal ligament stem cells (PDLSCs) aggregate is still limited in clinical application for lack of angiogenesis. This study aimed to investigate the effects and underlying mechanism of exosomes derived from stem cells from human exfoliated deciduous teeth (SHED) aggregate (SA-Exo) on the aggregate formation and angiogenic properties of PDLSCs. Methods: SA-Exo were isolated by ultracentrifugation. The effect of SA-Exo on the aggregate formation and angiogenic differentiation of PDLSCs were evaluated by investigating extracellular matrix (ECM) deposition and tube formation assay. MicroRNA (miRNA) sequencing was employed to screen different miRNA expression. The effect of targeting miRNA on ECM deposition and angiogenesis of PDLSCs aggregate was investigated after overexpression and inhibition of miRNA. Periodontal bone defect rat models were established to evaluate the effect of the PDLSCs aggregate and SA-Exo combination on periodontal bone regeneration. Results: SA-Exo could signicantly enhance the ECM deposition and angiogenic ability of PDLSCs. The expression of ECM-associated proteins (COL-I, integrinβ1, and bronectin), angiogenesis-related proteins (PDGF, ANG, TGFβRII), and related pathway (p-SMAD1/5 and p-SMAD2/3) were upregulated in PDLSCs aggregate with SA-Exo. Mechanistically, miR-222 was found relatively abundant in SA-Exo, which promoted ECM deposition and angiogenesis of PDLSCs. In vivo experiment further validated that combinational use of PDLSCs aggregate and SA-Exo promote more bone formation and neovascularization in rat’s periodontal bone defect. Conclusions: SA-Exo-shuttled miR-222 contributes to PDLSCs aggregate engineering by promoting aggregate formation and angiogenesis, which might through activate the TGF-β/SMAD signaling pathway.


Background
Periodontal diseases comprise a wide variety of chronic in ammatory conditions which affect the supporting tissues of the teeth including the gingiva, alveolar bone and periodontal ligament, which could result in tooth loss and contribute to systemic in ammation [1]. Traditional treatment strategies for periodontal bone defects include guided bone regeneration (GBR), autologous or allogeneic bone transplantation, and biomaterials combined with genes/growth factors [2]. Recently, periodontal ligament stem cells (PDLSCs)-based cell sheet or aggregate therapy has emerged as a promising strategy in periodontal regeneration [3]. However, the therapeutic e cacy in animal and clinical studies are still controversial [4][5][6]. Lack of angiogenesis might be the leading cause of the defective periodontal tissue regeneration mediated by PDLSCs sheet or aggregate [7]. Therefore, improving angiogenesis of PDLSCs aggregate has become an urgent problem to be solved at present.
Exosomes are nanometer-sized extracellular vesicles, which contain a broad spectrum of bioactive molecules including proteins, microRNAs (miRNAs), mRNAs and lipids [8]. As indispensable mediators in cell-cell communication, exosomes have been reported to participate in regulating angiogenesis [9]. Wu et al. documented that stem cells from human exfoliated deciduous teeth (SHED) derived exosomes promoted the angiogenic differentiation of human umbilical vein endothelial cells (HUVEC) and had potential value in alveolar bone regeneration [10]. We also found that SHED aggregate could regenerate plenty of blood vessels in dental pulp after implantation into injured teeth [11]. Additionally, a recent study investigated that exosomes from oral mucosal epithelial cell sheet showed pro-regenerative effects on skin wound healing [12]. These ndings suggested that exosomes derived from SHED aggregate (SA-Exo) also might have an outstanding effect on angiogenesis and tissue regeneration. However, whether the SA-Exo promote angiogenic differentiation of PDLSCs aggregate in periodontal regeneration remain unclear.
In this work, we investigated the e cacy of SA-Exo in angiogenic potential of PDLSCs. We found that SA-Exo promoted angiogenesis and aggregate formation of PDLSCs. Mechanistically, we revealed that miR-222 shuttled by SA-Exo promoted PDLSCs angiogenesis and aggregate formation. In addition, we explored the therapeutic effect of PDLSCs aggregate combined with SA-Exo on periodontal defects in Sprague-Dawley (SD) rats. Our data indicated that the combination of PDLSCs aggregate and SA-Exo could promote periodontal bone regeneration by enhancing angiogenesis, potentially providing a new strategy for repairing periodontal defects.

Cell Isolation and Aggregate Culture
Healthy human impacted third molars or extracted teeth for orthodontic reasons were collected from individuals aged 18-40 years. The standard of healthy periodontal tissues: no bleeding on probing, probing depth≤4mm, alveolar bone loss≤3mm. PDLSCs were isolated and cultured as previously described [13]. Brie y, the periodontal ligament was gently separated from the middle part of the root surface and enzymatically digested with type I collagenase (3mg/ml, Sigma-Aldrich, USA) for 2h at 37℃ (suspended every 15 mins). Cells were then plated in 6-well plates in alpha modi cation of Eagle's medium (a-MEM; Gibco, USA) with 10% fetal bovine serum (FBS, Gibco, USA), 0.292 mg/ml L-glutamine (Invitrogen, USA), 100 units/ml penicillin (Invitrogen) and 100 mg/ml streptomycin (Invitrogen) at 37 °C in 5% CO 2 , cultured for 2 weeks and the medium was changed every 3 days. To further purify the stem cells, single-cell-derived colonies were obtained using the limiting dilution technique.
Normal exfoliated human deciduous incisors were collected from 7-8-year-old children. SHED was isolated and cultivated in accordance with the protocol previously described [14], that is, the pulp was separated from remnant crown and the digested with 3mg/ml type I collagenase (Sigma) and 4 mg/mL dispase (Sigma) for 1h at 37℃. The single suspensions were seeded into 6-well plates with regular cell culture medium mentioned above. SHED used were at passage 2-5, and the same passage were used for each experiment.
PDLSCs aggregates were prepared as described previously [15]. 1-2×10 5 PDLSCs per well were seeded in 6-well-plate with basal medium and changed for aggregate induction medium with SHED aggregatederived exosomes at different concentrations when reaching 100% con uence. The induction medium was refreshed every 3-4 days. The aggregate was harvested after 10 days for the following experiments.

Exosomes Isolation and Characterization
SHED aggregates were prepared as described previously [11]. Then the SHED aggregate or SHED were cultured in the medium with exosome-deleted FBS for 48h. Exosomes-deleted FBS was prepared by ultracentrifuging for 16h at 100000g before use. Finally, the supernatants were collected and several centrifugates were performed for exosomes isolation as described previously [16]. Brie y, the supernatants were centrifuged at 300g for 10 min, 2000g for 10 min, and 16000g for 30min to remove dead cells and cellular debris, respectively. Then, the supernatants were ultra-centrifuged at 150000g for 70min, and additionally washed with PBS for another 70min at 150000g. All procedures were performed at 4℃. The nal pellets were resuspended in sterile PBS and stored at -80℃ for following experiments.
Western blotting was performed to detected exosome markers, including CD63 (ab217345, Abcam, USA), CD81 (sc-166029, Santa Cruz-Biotechnology, USA), CD9 (ab92726, Abcam). Cell extract was considered as control. Transmission electron microscopy (TEM) (Thermo Fisher, USA) was conducted to observe the morphology of exosomes. The exosomes were dropped onto carbon-coated copper grids and kept for 3-4 min, then dried with lter paper and stained with 1%phosphotungstic acid for 10s, nally washed with distilled water for 30s. Nanoparticle tracking analysis (NTA) was utilized to identify the particle size and distribution.

Exosomes Uptake Assay
Enriched exosomes were labeled with PKH67 (Sigma) according to the manufacturer's protocol. PDLSCs were seeded in 24-well plates with climbing, and incubated with the labeled exosomes for 4 and 24h, respectively. Hoechst (33342, Sigma) was used for staining the nuclei. The uptake of exosomes was visualized by confocal uorescence microscope (Nikon, Japan).

Tube Formation Assay
In vitro angiogenesis was determined by tube formation assay in Matrigel. PDLSCs (4×10 4 ) were seeded onto Matrigel (Sigma)-coated 96-well plate and cultured in FBS-free medium in the presence of exosomes at the indicated concentrations (0, 5, 10, 15, 30, and 60mg/ml) for 6h. The images of tube formation were acquired by inverted phase contrast microscope (Leica, Germany). The total tube length, number of junctions and number of nodes were calculated by randomly selecting ve elds per well using ImageJ 1.53c (National Institutes of Health, USA). The same experimental approach was used in PDLSCs after transfection with mimics and inhibitor.
Alkaline phosphatase (ALP) staining and Alizarin Red S (ARS) Staining PDLSCs were seeded in 12-well plate and cultured until the cells reached 80% con uence, then incubated with exosomes at different concentrations (0, 15, 30, and 60mg/ml) for 48h and further induced with osteogenic medium (basal medium supplemented with 50 mg/mL L-ascorbic-2-phosphate (MP Biomedicals, USA), 0.1 mM dexamethasone, and 5 mM b-glycerophosphate (Sigma)) for 7 days or 28 days for ALP staining or ARS staining, respectively. For ALP staining, the cells were rinsed with PBS, then nitro blue tetrazolium (NBT) and 5-bromo-4-chroro-3-indolyl phosphate (BCIP) (Beyotime, China) were added for the staining. 30 min later, staining agent was discarded and cells were rinsed two times with PBS. For ARS staining, the cells were xed with 60% isopropyl alcohol and stained with 2% ARS (Kermel, China). After rinsing with distilled water, stained calcium nodules were identi ed under microscope. Finally, the mineralized nodules were dissolved with hexadecyl pyridinium chloride, and absorbance was quantitatively measured at 570 nm for statistical analysis.

Scanning Electron Microscopy (SEM)
The cell aggregates from each group were washed with PBS three times, xed with 2.5% glutaraldehyde at 4℃, dehydrated and dried in a critical-point dryer. Finally, the specimens were observed under SEM (Hitachi S-4300; EIKO Engineering, Tokyo, Japan).
Small RNA sequencing Small RNA sequencing was performed by BGISEQ-2000 and the sequencing libraries were constructed in BGI online platform (Shenzhen, China). The miRNAs expression with signi cant differences between the SHED-and SHED aggregate-derived exosomes was shown by the heatmap.

Real-time reverse transcriptase-polymerase chain reaction (RT-PCR) analysis
Total RNA was extracted with TRIzol reagent (Invitrogen) and converted to cDNA using Mir-X TM miRNA First-Stand Synthesis Kit (Takara, Japan). Then RT-PCR was conducted with SYBR ® Premix Ex Taq TM II (Takara) using the quantitative PCR System (Bio-Rad, USA). The expression of miR-222 was normalized to that of U6 snRNA. The miR-222 primer was AGCUACAUCUGGCUACUGGGU (RiboBio, China). The forward and reverser primer of U6 were included in the Mir-X TM miRNA First-Stand Synthesis Kit (Takara). Cell transfection miRNA mimics or inhibitors were transfected into PDLSCs using riboFECT TM CP (RiboBio, China) according to the manufacturer's protocol. The miRNA mimics and inhibitor used were from Sangon Biotech (Shanghai, China).

Periodontal Bone Defect Model
The use of Sprague-Dawley (SD) male rats for research was approved by the IRB of FMMU. The surgical procedures were based on the guidelines of the Animal Care Committee of FMMU. The periodontal defect model was created as previously described [15]. A periodontal bone defect of approximately 3×2×1 mm 3 was created at the buccal alveolar bone of left mandibular molars. A total of 15 SD rats (8 weeks old) from the FMMU Animal Center were randomly divided into 3 groups: (1) a control group without cell and β-tricalcium phosphate (β-TCP) (provided by University of Extremadura, Spain) implantation (control, n=5); (2) a group treated with PDLSCs aggregate wrapping β-TCP (PDLSCA, n=5); and (3) a group treated with combination of PDLSCs aggregate and SA-Exo, which wrapping β-TCP (PDLSCA+SA-Exo, n=5). After 6 weeks, the rats' mandible samples were harvested and xed with 4% paraformaldehyde for 48h.

Micro-CT
The mandible samples were scanned using a Micron X-ray 3D Imaging System (YXLON, Germany) with 10mm resolution. Three-dimensional (3D) images were reconstructed and analyzed by the VG Studio 3.4 (VG, Germany). The ratio of new bone volume to tissue volume (BV/TV) was calculated.

Histological Analysis
After micro-CT analysis, the mandibles were decalci ed with 17% ethylenediaminetetraacetic acid (EDTA) (MP Biomedicals) for 1 month and embedded in para n. Para n sections (3um thick) were stained using hematoxylin and eosin (HE) as described [18]. Photographs were taken using a microscope (OLYMPUS, Japan). The percentage of new bone area in the total area was evaluated quantitatively from 3 randomly-selected sections by ImageJ 1.53c.

Masson's Trichrome Staining
The para n sections were stained with Masson's trichrome staining (Baso Diagnostic Inc., China) according to the manufacturer's instructions. Photographs were taken using a microscope (OLYMPUS).

Immuno uorescence Staining
The sections were depara nized and rehydrated, then subjected to antigen retrieval in boiled sodium citrate buffer solution (pH 6.0) for 10 minutes and cooled to room temperature. Slides were permeabilized with 1% TritonX-100 for 10 minutes and then blocked for 1h in blocking buffer (goat serum). After blocking, samples were incubated with primary antibody overnight at 4°C and then incubated with the related uorescence secondary antibodies for 1 hour at room temperature. Hoechst (MP Biomedicals) was used for counterstaining the nuclei. The immuno uorescent images were obtained using confocal uorescence microscope (Nikon). Antibodies against bronectin (ab2413, Abcam) and CD31 (ab212712, Abcam) were used in this study.

Statistical Analysis
All results were presented as the mean ± standard deviation (SD) of at least three independently experiments. Two-group comparisons were analyzed by Student's t tests. Comparisons among three or four groups were evaluated by one-way ANOVA followed by LSD post hoc test. P value less than 0.05 was re ected to specify statistical signi cance.

SA-Exo promoted the angiogenic and osteogenic differentiation of PDLSCs
Firstly, SHED and PDLSCs were isolated and characterized (Supplementary Figure S1). SA-Exo were isolated, which exhibited a cup-shaped morphology and had a bilayer membrane structure by TEM analysis (Fig. 1a). Western blotting analysis (Fig. 1b) demonstrated that SA-Exo expressed the exosomespeci c markers CD9, CD63, CD81. In addition, the diameters of the exosomes ranged mainly from 90 to 200nm was revealed by NTA. (Fig. 1c). To investigate whether the exosomes could be internalized into PDLSCs, SA-Exo labeled with PKH67 were incubated with PDLSCs for 4 and 24h, respectively. Confocal uorescence microscopy analysis showed that PKH67-labeled exosomes (green) were gradually internalized by the PDLSCs from 4 to 24h (Fig. 1d).
The effects of SA-Exo at different concentrations (0, 5, 10, 15, 30, 60 µg/ml) on PDLSCs angiogenesis were assessed by tube formation assay (Fig. 2a). In comparison with the control group (0µg/ml), the total length, number of junctions and number of nodes were signi cantly increased in high concentrations of exosomes group (10, 15, 30, 60 µg/ml) (Fig. 2b-d). To further investigate the dose effects of SA-Exo on PDLSCs osteogenesis, PDLSCs were cultured at different concentrations (0, 15, 30, 60 µg/ml) of SA-Exo for 48h. After 7days of osteogenic induction, ALP staining showed that 60µg/ml of exosomes obviously promoted ALP expression (Fig. 2e). Alizarin red staining on 28 days indicated that 30 and 60µg/ml of SA-Exo markedly enhanced the mineralization ability of PDLSCs (Fig. 2f, g). Thus, 60µg/ml of SA-Exo can signi cantly promote angiogenic and osteogenic differentiation of PDLSCs, and this dose was selected in the following experiments.

SA-Exo enhanced ECM deposition and angiogenesis of PDLSCs aggregate
After 10 days of culture with SA-Exo through the whole stage, PDLSCs formed complete aggregate that could be detached from the edge of the dishes. Cells in PDLSCs aggregate treated with SA-Exo (PDLSCA + SA-Exo group) exhibited more and better organization compared to those in PDLSCs aggregate (PDLSCA group) (Fig. 3a, rst and second row). SEM was utilized to examinate the micro-structure of aggregate and showed a higher density network in PDLSCA + SA-Exo group (Fig. 3a). Moreover, PDLSCs aggregate combined with SA-Exo showed much more blue-stained collagen bers compared to PDLSCs aggregate by Masson's trichrome staining (Fig. 3a). And then more expression of bronectin, protein that mainly found in ECM, was observed in PDLSCA + SA-Exo group by immuno uorescence analysis (Fig. 3a). In addition, the expression of ECM-related proteins ( bronectin, integrinβ1, and COL-I) were also showed upregulated in PDLSCA + SA-Exo group compared to PDLSCA group by western blot (Fig. 3b). Meanwhile, we detected the expression of angiogenesis associated proteins and related pathways. The expression of PDGF, ANG, and TGFβRII and SMAD singling (phosphorylated SMAD2/3 and SMAD1/5) were upregulated after SA-Exo treatment (Fig. 3c). These outcomes suggested that SA-Exo can promote ECM deposition and angiogenesis of PDLSCs aggregate.

MiR-222 contributed to SA-Exo mediated angiogenesis of PDLSCs
Exosomes mediate biological activity by transforming cargos to the recipient cells. Accumulative evidences have demonstrated that exosomal miRNAs were important in regulation angiogenesis [9,19,20]. Therefore, to gain insights into how the angiogenesis of PDLSCs were promoted by SA-Exo, we analyzed the miRNAs expression pro les of SA-Exo. SHED-derived exosomes (S-Exo) were considered as control. The heat map of miRNA expression indicated that miR-222 was markedly upregulated in SA-Exo (Fig. 4a). The RT-PCR analysis also demonstrated that the expression of miR-222 in SA-Exo was signi cantly higher than that in S-Exo (Fig. 4b). Subsequently, in order to investigate whether miR-222 promote angiogenesis of PDLSCs, the tube formation assay was carried out after upregulation or downregulation of miR-222 (Fig. 4c). The expression of miR-222 in PDLSCs or SA-Exo were signi cantly overexpressed by pretreating PDLSCs or SHED aggregate with mimics, in contrast, miR-222 was suppressed by using the miR-222 inhibitor (Supplementary Figure S2). The total length, number of junctions and number of nodes were signi cantly increased in the miR-222 mimics group compared to the NC group. (Fig. 4d-f). In contrast, suppression results were found in the miR-222 inhibitor group ( Fig. 4g-i). These data demonstrated that miR-222 contributed to SA-Exo mediated angiogenic differentiation of PDLSCs.

MiR-222 contributed to SA-Exo mediated ECM deposition and angiogenesis of PDLSCs aggregate
To further investigate whether miR-222 promote the ECM deposition and angiogenesis of PDLSCs aggregate, histological examination and associated protein expression were detected. When upregulating of miR-222 in PDLSCs aggregate, the more and better organization of cells in aggregate was observed by microscope and a higher density network was showed under SEM compared to the miR-222 mimics NC group ( Fig. 5a and b, rst and second row). Masson's trichrome staining also showed more collagen deposition and immuno uorescence analysis showed an upregulation of ECM-related protein bronectin ( Fig. 5c and d, rst and second row). In contrast, opposite results were found after miR-222 was inhibited (Fig. 5a-d, third and fourth row). Additionally, the expression of ECM-and angiogenesisassociated protein were upregulated in the miR-222 mimics group and downregulated in the miR-222 inhibitor group (Fig. 5e-f). It has high light that TGF-β has a crucial role in the formation of ECM and angiogenesis [21]. Thus, we detected the expression level of TGFβRII and related signaling (p-SMAD2/3 and p-SMAD1/5) by using western blot and found that overexpression of miR-222 activated the TGF-β signaling by upregulating p-SMAD2/3 and p-SMAD1/5 in PDLSCs aggregate compared to the NC group, and then these proteins were downregulated after miR-222 inhibition (Fig. 5f). Collectively, these ndings suggested that SA-Exo shuttled miR-222 might activated TGF-β/SMAD signaling to promote ECM deposition and angiogenesis of PDLSCs aggregate.
SA-Exo promoted PDLSCs aggregate to repair the periodontal bone defect in animal model In order to con rm the therapeutic effect of SA-Exo on PDLSCs aggregate, we carried out in vivo experiments of periodontal bone defect in SD rat. As shown by micro-CT, obvious bone regeneration was showed in PDLSCA + SA-Exo group when compared with the blank control and PDLSCA group (Fig. 6a, b).
Meanwhile, an increased new bone formation in PDLSCA + SA-Exo group was observed by HE staining (Fig. 6c, d). Then Masson's trichrome staining showed more collagen deposition in the SA-Exo treated group (Fig. 6e). Additionally, the expression of angiogenesis-related protein CD31 was upregulated obviously in PDLSCA + SA-Exo group compared to the blank control and PDLSCA group (Fig. 6f). Therefore, the results indicated that SA-Exo promoted the PDLSCs aggregate to repair the periodontal bone defect by upregulating angiogenesis.

Discussion
Stem cell-based tissue engineering is recognized as a prospective approach for periodontal regeneration.
Recently, cell sheet/aggregate technology served as an important strategy has been extensively used in periodontal regeneration, which could deliver high-density stem cells and preserve abundant ECM thereby mimicking natural microenvironments to improve cell survival and function [5,18,22]. However, the therapeutic effect of PDLSCs aggregate is still controversial because lack of vascular and mineral tissue formation in tissue regeneration [7]. In the present study, our results demonstrated that SA-Exo promoted ECM deposition and angiogenesis of PDLSCs aggregate by shuttling miR-222, which might activate TGFβ/SMAD signaling. Compared with PDLSCs aggregate, the combinational use of PDLSCs aggregate and SA-Exo enhanced bone repair in periodontal bone defect rat models, which might provide an alternative option for optimizing the therapeutic effect of PDLSCs aggregate-based periodontal regeneration.
Periodontal regeneration is a complex process, and angiogenesis considered as a fundamental element plays an important role in tissue regeneration due to neovascularization provides su cient oxygen, nutrients and as well as discharges the metabolite [23]. Our previous study has reported that SHED aggregate regenerate plenty of blood vessels in dental pulp after implantation into immature permanent teeth, which suggested that the aggregate had powerful angiogenic ability [11]. Additionally, other than directly participating in tissue regeneration, reports also suggest that the extracellular vesicles (EVs) contribute to the regenerative effect of stem cells [24]. Exosomes are important components of EVs and abundant in ECM, which participate in matrix organization and regulation of cells within it [25]. Previous study has indicated that exosomes derived from clinical-grade oral mucosal epithelial cell sheets showed pro-regenerative effects on skin wound healing [12]. Here, we found that exosomes derived from SHED aggregates would promote PDLSCs angiogenesis. Furthermore, the combinational use of PDLSCs aggregate and SA-Exo would signi cantly promote ECM deposition and angiogenesis.
Exosomes are important mediators of intercellular communication by shuttling a wide range of functional cargos. It has reported that exosomes contain multiple bioactive molecules involved in angiogenesis, including protein like vascular endothelial growth factor [26], broblast growth factor [27], PDGF [28], as well as miRNAs like miR-126, miR-26 [9]. In this study, we investigated the exact exosomal cargos that have a positive effect on angiogenesis of PDLSCs. We analyzed the miRNA expression pro le of SA-Exo and found that miR-222 was markedly upregulated in SA-Exo compared to S-Exo. It has been demonstrated that miR-222 plays different roles in different cells and can positively and negatively modulate angiogenesis. Previous researches have indicated that miR-222 plays anti-angiogenic effect in endothelial cells, while has effects of pro-proliferation and pro-migration in vascular smooth muscle cells (VSMCs) [29]. In contrast, the high level of the miR-222 in many tumor cells as well as in many in ammation-based diseases does not t with the view and exhibits promoting angiogenesis [30,31]. Ribeiro-Rodrigues et al. reported that exosomes released by ischemic cardiomyocytes were enriched in miR-222 and promote angiogenesis. Simultaneously, miR-222 overexpression induced an increase in tubulation and sprouting by HUVECs [31]. Here, we observed PDLSCs overexpressed miR-222 exhibited increase of angiogenic differentiation, while decrease of this was showed when miR-222 inhibition. These ndings imply that exosome-mediated miR-222 plays an important role in the angiogenesis. However, the exact target gene on angiogenesis should be investigated in future studies.
To the best of our knowledge, there are no publication that de nitively show the stimulatory effects of miR-222 on ECM production. However, it has reported that miR-222 is closely related to atherosclerosis and considered as a biomarker for early diagnosis [32]. miR-222 promoted VSMCs proliferation and migration, which secrete diverse cytokines in atherosclerotic lesions, including interferon-γ, macrophage inhibitory factor, and TGF-β, which triggering the ECM remodeling or synthesis/deposition [33]. Its high light that TGF-β executes the production of matrix molecules and represses matrix degradation by stimulation of protease inhibitor synthesis and inhibition of proteases synthesis [21]. TGF-β binding to the TβRII results in phosphorylating and activating activin receptor-like kinase 5 (ALK5), then phosphorylate SMAD2/3 to increase ECM deposition [34]. In addition, high concentration of TGF-β was able to activate p-SMAD2/3 and p-SMAD1/5 induce a pro-angiogenic state [34,35]. Here, we found that miR-222 upregulated ECM associated proteins in PDLSCs aggregate and downregulated with inhibition. Moreover, both p-SMAD2/3 and p-SMAD1/5 was upregulated in PDLSCs aggregate combined with SA-Exo via shuttling miR-222, which can be downregulated by miR-222 inhibition. Collectively, these results might suggest that miR-222 mediated ECM deposition and angiogenic differentiation in PDLSCs might through TGF-β/SMAD signaling pathway.
We further evaluate the function of PDLSCs aggregate and SA-Exo combined implantation in periodontal defect models. Consistent with the in vitro results, micro-CT and histological analyses revealed more new bone formation after treatment with SA-Exo than that in the without exosomes group in the rat periodontal bone defects. Angiogenesis is an essential step during bone regeneration, as restoration of blood ow provides nutrients and renewable autologous cells to heal the defect. In vivo, we found that the addition of the SA-Exo resulted in generating more new blood vessels by immuno uorescence staining for CD31 which is common marker of neovascularization. Thus, our data suggested that combination of PDLSCs aggregate and SA-Exo could be considered as a potential strategy for periodontal bone regeneration. However, there were some limitations in our study. Periodontal defects are in ammationmediated microenvironments and more complex than the defect that we established in this study. To better evaluate the therapeutic effect of SA-Exo on PDLSCs aggregate regeneration in periodontitisinduced alveolar bone defects, the immunoregulatory role should be further investigated in large animal periodontitis models.

Conclusions
Our study rstly showed that exosomes derived from SHED aggregate promote PDLSCs aggregate formation and angiogenesis, which contribute to periodontal bone defects reparation. Moreover, the exosomes transfer miR-222 to PDLSCs and might activate TGF-β/SMAD pathways to increase ECM deposition and angiogenesis. Our study provides a new potential strategy to improve the clinical therapy of PDLSCs aggregate in periodontal regenerative medicine. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.   analysis of the calcium deposition in PDLSCs. *P < 0.05, ** P < 0.01, *** P < 0.001 represent signi cant differences compared with the matched control group.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.