First of all, we fabricated PMVs through centrifuging, sonicating and extruding the derived platelet membrane ghosts. As displayed in Fig. 2a, transmission electron microscopy (TEM) images indicated we have fabricated well-defined spherical vesicle-like structure successfully. Then, as analyzed by dynamic light scattering (DLS), hydrodynamic sizes of PMVs were 100 ~ 200 nm approximately (Fig. 2b). The zeta-potential values of PMVs were around − 80 mV (Fig. 2c), indicating that the vesicle solution is a stable dispersion system. Furthermore, to achieve the effect of cross-linking collagen and promoting remineralization, it is important to make sure the protein on the platelet membrane is still retained after the entire preparation process. SDS-PAGE protein analysis by Coomassie blue staining was completed to run whole proteins of platelets and PMVs. In Fig. 2d, PMVs had a substantial reservation of proteins from platelets. To evaluate the long-term stability of PMVs, we placed PMVs in aqueous solution and the results showed that the particle size of PMVs remained at about 200 nm during the 30 days (Fig. 2e). In a word, the above results verified that the nanoscale PMVs, with the membrane protein retained were fabricated successfully. Besides, to achieve the purpose of continuously acting on demineralized dentin or entering into dentinal tubules, PMVs have to present a certain binding ability with demineralized dentin. ATR-FTIR spectra of the EDTA-etched dentin before and after PMVs coating and after deionized water washing were displayed in Fig. 2f, and ATR-FTIR spectra of the intact dentin was exhibited in Figure S1. Clearly, after coated by PMVs, the peaks of PO43− and amide (I, П, Ш, NH-) groups were significantly enhanced owing to cytomembrane is mainly composed of lipids (especially phospholipid) and proteins. After deionized water washing, above-mentioned characteristic peaks were only decreased slightly, indicating that majority of PMVs still reserved on the demineralized dentin. It can be seen from the results that PMVs exhibited favorable binding ability to demineralized dentin. This satisfactory binding ability lays the foundation for the subsequent role of PMVs in promoting cross-linking and inducing biomineralization.
Before biological applications, the toxicological evaluation of the PMVs was necessary. We chose human dental pulp cells (HDPCs) and human periodontal ligament cells (PDLCs), two species of cells that are closely related to teeth for our experiments. Based on the fact that the destroyed cells could be stained with red-fluorescence and normal cells could be stained with green-fluorescence, the live cells could be distinguished from dead cells. As shown in Fig. 3a and 3b, most of the HDPCs and PDLCs were alive, and even when the cells were co-cultured with 4 times the concentration of PMVs in the treatment solution in subsequent experiments, the cells did not show a distinct trend of death. The cell viability detected by CCK-8 lit remained higher than 90% after 24 h, suggesting that no or little toxic effect on cell proliferation was induced by PMVs (Fig. 3c and 3d). The above results confirmed that the cytotoxicity of PMVs to living cells was negligible.
Since the template function of collagen is very important for mineralization, effective protection of collagen is necessary. In order to explore that PMVs can effectively cross-linking with naked collagen, type I collagen degradation assay was used to verify the pro-crosslinking effect of PMVs at first. The HYP detection kit was used in this experiment. HYP is a collagen-specific amino acid, so the HYP content of a sample can reflect its collagen content. The principle of acid hydrolysis method for detecting HYP is that strong acid and high temperature hydrolyzes collagen and other proteins to generate free amino acids. HYP was then given a color response and its concentration can be determined by measuring its absorbance. Briefly, the pro-crosslinking effect of PMV on collagen, which can effectively inhibit collagen degradation, could be determined by measuring the HYP concentration in the supernatant of collagenase-degraded collagen. As displayed in Figure S2, the HYP concentration in the PMVs-treated group was lower than that in the DDW-treated group, which can be concluded that due to the cross-linking effect of PMVs on collagen, the amount of enzymatic degradation of collagen is reduced. On the other hand, we observed the morphological changes of PMVs-treated group and control group after being immersed in artificial saliva for 3 days. As shown in Fig. 4a, in the control group, a small amount of scattered new mineral deposits can be seen between the exposed collagen fibers, while no obvious minerals are found on the collagen surface (white arrow). This is consistent with previous reports in the literature that a large number of collagen surfaces have an inhibitory effect on mineral formation [33]. In the PMVs-treated group (Fig. 4b), new minerals with a certain thickness were formed on the collagen surface (white arrow), indicating that PMVs effectively combined with naked collagen to induce remineralization of the collagen surface. On the other hand, due to the lack of mineral support within the fibers, the collagen fibers in the control group shrank and collapsed after dehydration, showing a flattened shape (blue arrow). However, the collagen fibers in the PMVs-treated group maintained their original shape after dehydration (blue arrow), and presented discontinuous areas arranged periodically. This is due to the formation of regularly arranged minerals within the collagen fibers. In a word, because collagen is the important template in dentin remineralization, and PMVs can effectively cross-link with collagen in this experiment, so this provides the foundation for our subsequent remineralization of dentinal tubules.
Further, we verified that the surface protein fibrin and vesicle structure of PMVs can be used to induce dentin biomineralization. The intermediate product PLM (membrane fragmentation, no vesicle structure) was set as a separate treatment group (group b). With the intention of characterizing the final remineralization on the dentin surface and within the dentinal tubules, we observed the surface and longitudinal morphologies through SEM. After treated with EDTA, a network of fibrous tissue with periodic horizontal stripes could be observed on the surface of dentin (Figure S3a) and in the dentinal tubules (Figure S3b). This is due to the exposure of dentin collagen fibers in the dentin surface after demineralization. Also, studies have shown that EDTA does not affect the structure of collagen [34]. After 1 week of biomimetic mineralization treatment, the samples treated with deionized water did not experience spontaneous remineralization in artificial saliva (Fig. 5a1). To be specific, the dentinal tubules were still exposed, and no obvious mineralization sedimentation sealed the dentinal tubules and on the dentin surface. After immersed in artificial saliva for 3 weeks, the dentin surface of the samples treated with deionized water has only a very small amount of neogenic mineral deposited (Fig. 5a2). In contrast, after treated with PLM and with PMVs, the exposed collagen was covered with regenerated minerals after 1 week (Fig. 5b1, c1). After the samples were continuously immersed in artificial saliva for 3 weeks, it could be seen that the new minerals increased significantly and a large number of needle-like crystals appeared in samples treated with PLM (Fig. 5b2) and PMVs (Fig. 5c2). It is worth noting that the amount of nascent minerals in the samples treated with PLM (Fig. 5b1, b2) was thinner than that treated with PMVs (Fig. 5c1, c2).
From longitudinal sections of the samples, remineralization within the dentinal tubules was consistent with the neogenic mineral that was seen on the surface. No obvious mineral deposits were seen in the dentinal tubules of the deionized water-treated samples (Fig. 6a1-a4). It was discovered that irregular particles had grown in the dentinal tubules of PLM-treated samples soaked in artificial saliva for 1 week (Fig. 6b1, b2). By 3 weeks, the needle-like crystals in the tubules became denser (Fig. 6b3, b4, Fig. 7). Compared with the PLM-treated group, more and denser needle-like minerals were formed in the dentinal tubules of the PMVs-treated samples. Nearly the entire dentinal tubules were filled with nascent minerals in the samples of the PMVs-treated group that were mineralized for 3 weeks. Distinctively, it can be discovered from Fig. 6c1 that the mineralization depth inside the dentinal tubules has exceeded 60 µm beneath the surface. This outcome come up to the deepest depth beneath the surface of dentin reported so far, which was come up to the depth as Chen Li et al. using amyloid aggregates (lyso-PEG) for deep mineralization of dentin, and was 3–6 times the depth of previous literature reported [35–40]. Compared with clinically used adhesive and fluorinated paints, this deep sealing by PMVs can effectively prevent external stimuli from damaging the pulp through the dentinal tubules [41]. In general, we speculated that the mineralizing effect of PMVs can be attributed to the synergistic effect of membrane’s composition and vesicle structure. The specific proteins on the membrane could cross-link with collagenous fibers and further regulate the concentration of Ca2+ and PO43−. Besides, the phospholipids from PLM can recruit Ca2+ to facilitate the deposition of hydroxyapatite. Moreover, the vesicle structure provides an effective three-dimensional space for mineral deposition, which promotes the efficient progress of mineralization.
It can be seen from the aforementioned SEM images that the mineralization effect of the PLM and PMVs groups were significantly better than that of the control group. Therefore, we performed EDS detection of nascent minerals on demineralized dentin treated with PLM and PMVs. As shown in Figure S4, after immersion in artificial saliva for 1 or 3 weeks, the calcium-to-phosphorus ratio of nascent minerals formed in PLM-treated samples was higher than that in PMVs-treated samples. This means that the mineralization products of the PMVs-treated group are closer to the calcium-to-phosphorus ratio of natural dentin of 1.67.
After the mineralization cycle, XRD was used to characterize the mineral phase and degree of orientation of the nascent minerals. As shown in Fig. 8, in healthy dentin samples, samples after EDTA demineralization, and samples after remineralization in each group, obvious characteristic peaks of hydroxyapatite (002), (211), (112) were observed, which proves that the main component of new minerals is hydroxyapatite. It is consistent with the main components of natural dentin minerals.
In brief, we brought forward a strategy to mimic MVs for biomimetic mineralization of dentin with PMVs, which are prepared from abundant platelets. It is proposed to utilize the related proteins on the surface of PMVs to promote the cross-linking of dentin collagen and provide a template to induce deep biomimetic mineralization in dentinal tubules, which could protect the dental pulp in the event of tooth defect. Our study preliminarily explored the biomineralization effect of PMVs, and further research is needed on the role of PMVs in protecting dentin collagen from enzymatic degradation.