More potential risk factors for implant in patients with type 2 diabetes were detected by proteomics in addition to hyperglycemia.

Background: It is commonly accepted glycemic control can decrease the negative effects of implant in type 2 diabetes mellitus (T2DM) patients. Whether the remaining pathological changes besides hyperglycemia caused by T2DM will affect implant-bone integration during the healing period has remained unclear. This study aims to determine whether other risk factors besides hyperglycemia lead to failed osseointegration in T2DM patients during healing period. Methods: First, we compared the success rate between T2DM patients and non-T2DM patients during the healing period at our center. Bone marrow mesenchymal stem cells (BMSCs) were cultured from all enrolled subjects. Then, proteomics was used to detect differentially expressed proteins (DEPs) among the DM failure (DM-F), DM success (DM-S) and control (Con) groups. Additionally, the relationship between expression of the nine target DEPs and glucose concentration in media was investigated. Results: Significantly higher failure rates in T2DM patients were found. Fifty-two DEPs were found in DM-F group compared with DM-S group. Seventy-three DEPs were found in DM-F group compared with the Con group. Forty-three DEPs were found in DM-S group compared with Con group. Four target DEPs was influenced by glucose, while the other five expressed the same in different glucose concentration media. DEPs in DM-F group may affect the biological function and regulatory potential of BMSCs through gene ontology annotation and functional enrichment analysis. Conclusions: DEPs in DM-F group can be potential risk factors and intervention targets for dental implant in T2DM patients. More potential risk factors affect implant-bone integration besides hyperglycemia. including apolipoprotein (APOE), fibulin-1 (FBLN1), glutamine synthetase (GLUL), integrin alpha-10 (ITGA10), matrix metalloproteinase 2 (MMP2), transgelin (TAGLN), insulin-like growth factor-binding protein 2 (IGFBP2), leptin receptor (LEPR), atrial natriuretic peptide receptor 3 (NPR3). The functions of the nine proteins were reported in association with cell proliferation, migration and osteogenesis 11 − 19 . Western blot results showed that all DEPs had the same expression tendency as the proteomics results

other stem cells. BMSCs play a substantial role in the process of osseointegration around implants.
BMSCs migrate to the sites of titanium implants and differentiate into osteoblasts 5 . However, the proliferation, differentiation, migration and mineralization ability of BMSCs are impaired by highglucose media 6 − 7 . Thus, glycemic control is an important intervention to reduce the adverse effects of implant in clinic. Some scholars have even considered that satisfactory glycemic control could eliminate the adverse effect on implant success rate in T2DM patients 8 . Unfortunately, few studies of the relationship between glycemic control and implant success in T2DM patients have been conducted. Thus, the question arises: does glycemic control reduce or eliminate the adverse effect of T2DM on implant success? In other words, are there factors other than blood glucose that affect implant-bone integration in T2DM patients?
To answer these questions, we carried out a clinical study to determine the implant success rate in the healing period of well-controlled T2DM patients and non-T2DM patients at our center and used proteomics to detect protein expression profiles in diabetics with failed osseointegration, T2DM patients with successful osseointegration, and non-T2DM patients with successful osseointegration.
The differentially expressed proteins (DEPs) were then identified. The relationship between the expression of partial DEPs and glucose concentration in culture media were studied. Gene ontology (GO) annotation and functional enrichment analysis were conducted to determine the impaired biological functions of BMSCs derived from T2DM patients with failed osseointegration.

Subject enrollment and ethics statement
The present study was approved by the Ethics Committee of Beijing Stomatological Hospital, School of Stomatology, Capital Medical University (Approval No.: CMUSH-IRB-KJ-PJ-2018-08) with informed patient consent. All subjects were recruited by one surgeon in the department of Dental Implant Center. All patients with controlled T2DM (HbA1c < 8%) and ready for implant surgery in the past three years were enrolled as study subjects (T2DM group). Subjects in the non-T2DM group were enrolled at a ratio of 2:1 compared to the T2DM group (non-T2DM group), and all basic information was matched to avoid the influence of other elements, including age, sex, general health condition, DM type, implant system, and implant position. Implant sites were left to heal for at least three months after tooth extraction. Subjects with implants inserted with guided bone regeneration were excluded from the study. Patients with contraindications for implant surgery, such as cardiovascular disorders, renal diseases or uncontrolled periodontitis were excluded. The STROBE (Strengthening the Reporting of Observational studies in Epidemiology) guidelines were followed to ensure the rigor of our study (Appendix Table 1). Implant placement and bone chip harvest and cell culture All implant surgeries in the present study were performed by one surgeon. Implants were inserted with the method as previously described 9 . Bone chips were harvested during implant socket preparation and were reserved in phosphate-buffered saline (PBS, Gibco, USA) with antibiotics (10,000 U/ml penicillin, 10 mg/ml streptomycin, Gibco, USA). After being centrifuged at 1100 rpm, the bone chips were resuspended in mesenchymal stem cell medium (MSCM, ScienCell, USA), seeded in dishes and then cultured in a humidified, 5% CO2 cell incubator at 37 °C without movement for seven days. A 70-µm pore-size strainer (Falcon, BD Labware, USA) was used to obtain single-cell suspensions, which were then cultured in MSCM. The medium was replaced every three days. The fourth-passage BSCMs were used for subsequent experiments.

Clinical assessments
The healing periods for the upper and lower jaw were six months and three months, respectively.
Before implant loading, failed implants, defined as implants with clinical mobility and peri-implant radiolucency, were recorded to determine the failure rate and to perform whole-cell proteomic analysis.
Grouping for whole-cell proteomic analysis 6 The isolated BMSCs were divided into three groups: BMSCs derived from the T2DM patients with failed implant (DM-F group), BMSCs derived from T2DM patients with successful implant (DM-S group), and BMSCs derived from the non-T2DM patients with successful implant (Con group). Five subjects from the DM-F group were selected randomly for proteomic analysis with each from the DM-S and Con groups, respectively, were also selected for proteomics and were matched to the five subjects selected from the DM-F group according to all the basic information mentioned above.
Whole-cell quantitative proteomic analysis Tandem mass tagging (TMT)-based proteomic analysis was performed as previously described 10 .  Table 2. Table 2 Top 10 of DEPs in DM-F, DM-S and Con group.   Table 2.
To confirm the reliability of proteomic results, nine proteins were chosen from Table 2 Table 2, except for MMP2 (Fig. 1A).
The relationship between the expression of target DEPs and glucose concentration To determine whether the expression of DEPs was associated with glucose concentration, the fourthpassage BMSCs from the Con group were cultured in media with different glucose concentrations, including normal glucose concentration (1000 mg/L, NG group), high glucose concentration (4500 mg/L, HG(+) group), and ultra-high glucose concentration (9000 mg/L, HG(++) group) (DMEM, HyClone, USA) for seven days. The expression of GLUL is decreased by high glucose concentration, while the expression of LEPR, MMP2, and NPR3 are increased with ultra-high glucose concentration.

GO annotation of DEPs among the DM-F, DM-S, and Con groups
To characterize the function of the DEPs in the three groups, a bioinformatic analysis based on GO which are categorized as biological process, molecular function and cellular component was carried out.
In the biological process category, DEPs among the three groups mostly showed enrichment in cellular processes, single-organism processes, and biological regulation, reaching more than 35% ( Fig. 2A). In the molecular function category, approximately 70% of DEPs showed enrichment in binding and catalytic activity in all three groups (Fig. 2B). The cellular component enrichment revealed that over 65% of the DEPs were cell-, organelle-and membrane-associated proteins Regarding the biological process category, the upregulated DM-S/Con DEPs showed enrichment in plasma lipoprotein particle remodeling, macromolecular complex remodeling, and protein-lipid complex remodeling. Regarding molecular function, the DEPs showed enrichment in sulfur compound binding, heparin binding, and glycosaminoglycan binding, among other terms. Regarding cellular component, DEPs showed enrichment in the plasma membrane, cell periphery, and membrane ( Fig. 3C, left). Moreover, the downregulated DM-S/Con DEPs showed enrichment in the organonitrogen compound biosynthetic process, mitotic cell cycle phase transition, and cell cycle phase transition, among other terms. Regarding molecular function, DEPs showed enrichment in Ras GTPase binding, small GTPase binding, and transferase activity, among other terms (Fig. 3C, right).

Discussion
Diabetes mellitus is recognized as a risk factor for implant surgery. Despite this risk, with the continuous improvement of implants, especially implant surface technology, the implant success rate has increased, including in T2DM patients 20 . These results mean that most scholars believe that patients with well-controlled T2DM do not need to be excluded from implant therapy. Hyperglycemia is the main characteristic of diabetes 21  the clinical results showed significant differences between well-controlled T2DM patients and non-T2DM patients in implant failure rate (10.77% vs 0.75%). This result indicates that implantation failure rate in T2DM patients during the healing period was still higher than that in non-T2DM patients, even for patients with satisfactory glycemic control. Thus, we propose a hypothesis that hyperglycemia is not the sole risk factor for implant surgery in T2DM patients. Other risk factors for osseointegration during the healing period must be explored in further studies.
Proteomics is often used to find biomarkers for early diagnosis of tumors and target drugs 24 − 25 . In the present study, proteomic results showed a certain number of proteins to be differently expressed in the three groups. The results indicated that failure in the DM group was not an accident and that DEPs in the DM-F group may cause failed osseointegration during the healing period. Therefore, could hyperglycemia cause the abnormal expression of these DEPs?
To explore the relationship between hyperglycemia and DEP expression, BMSCs from the Con group were cultured in different glucose concentration media. Nine proteins were selected from DEPs, which related to cell proliferation, migration and osteogenesis. Western blot results suggested that the expression of GLUL, LEPR, MMP2 and NPR3 were changed with the glucose concentration, whereas the expression of other five DEPs was unrelated. This result indicated that a high-glucose microenvironment was not the only reason for DEPs to arise, even if it was the major reason. Noteworthy enrichment in downregulated proteins that might inhibit bone-implant integration was found. Regarding biological processes, the downregulated DEPs principally showed enrichment in osteoblast proliferation. Thus, the decreased expression of proteins, which regulate osteoblast proliferation in diabetics, might consequently damage the ability of bone formation. Integrin binding, extracellular matrix and extracellular matrix organization were also major enrichment terms revealed by the GO function. Integrin is a receptor that is associated with the ligands in the extracellular matrix and plays an indispensable role in cell migration 26 . In the bone repair process, BMSCs first migrate to the injury site to participate in bone formation. The decreased integrin binding ability might also be a reason for weak bone formation in T2DM patients. The extracellular matrix is a cell microenvironment, and thus, alterations in the extracellular matrix could also impair bone formation, which must be further investigated.

Conclusions
The clinical portion of the present study indicated that the success rate in well-controlled T2DM patients was still lower than for non-T2DM patients, meaning that T2DM is still a risk factor for implant therapy. The molecular changes of BMSCs in T2DM patients contribute to failed osseointegration, which arose not only due to hyperglycemia. Glycemic control cannot eliminate the negative effect completely, even though it helps increase the success rate of implantation in T2DM patients to some extent. DEPs in DM-F can be indicators or intervention targets for implant therapy in T2DM patients.
All results remind us that it is time to reappraise the position of hyperglycemia in implantation failure with T2DM patients, and more potential risk factors at the molecular level should be paid more attention in the future.

Availability of data and materials
The datasets used or analyzed during the current study are available from the corresponding author on reasonable request. and lower in lower in DM-F than Con; GLUL expressed lower in DM-F than Con; IGFBP2 expressed higher in DM-F than Con; ITGA10 expressed lower in DM-F than DM-S; LEPR expressed higher in DM-S than Con; MMP2 expressed higher in DM-F than DM-S and higher in DM-S than Con; TAGLN expressed higher in DM-F than DM-S; NPR3 expressed higher in DM-S than Con. All expression tendencies were as same as the results in Table 2