2.1. Mussel-Derived Peptide Synthesis and Surface Modification
The mussel-derived peptide with clickable DBCO group was synthesized by solid-phase peptide synthesis strategy through a Mfps-like peptide mimicking method 38, 46. To introduce catecholic amino acid DOPA into the peptide sequence, a commercially available acetonide-protected Fmoc-DOPA (acetone)-OH was used during solid-phase peptide synthesis. To leave enough catechol groups for Zn2+ coordination and display accessible DBCO groups for biomolecular clicking after peptide adhesion on Ti surfaces, hexavalent DOPA units with one amino acid interval and DBCO with a long polyethylene glycol (PEG) chain were used to prepare the mussel-derived peptide Ac-(DOPA)-G-(DOPA)-G-(DOPA)-K[(PEG5)-(Mpa)-(Mal-DBCO)]-(DOPA)-G-(DOPA)-G-(DOPA) ((DOPA)6-PEG5-DBCO) (Figure 1A and S1). In addition, the BMP-2-derived peptide (KIPKASSVPTELSAISTLYL), which was derived from the 73-92 amino acid fragment of the BMP-2 finger epitope, was conjugated with 2-azidoacetic acid and PEG5-carboxyl coupling, respectively (Figure 1B and S1). The obtained azido-capped BMP-2-derived peptide (Azido-KIPKASSVPTELSAISTLYL, (2-Azido)-PEG5-BMP-2) thus could be easily connected with (DOPA)6-PEG5-DBCO-bound surfaces, providing a flexible surface modification strategy. The two synthesized peptides were first purification by high performance liquid chromatography (HPLC, purity > 95%) (Figure S2). Electrospray ionization mass spectrometry (ESI-MS) was used to confirm their molecular structures. The monoisotopic mass [M+2H]2+ of (DOPA)6-PEG5-DBCO and [M+2H]2+ of (2-Azido)-PEG5-BMP-2 were found at 1036.69 Da and 1247.63 Da, which were corresponding to the theoretical molecular weight at 2070.18 Da and 2492.86 Da, respectively (Figure 1C and 1D). These results indicated the successful synthesis of clickable mussel-derived peptide and the complementary azido-capped BMP-2-derived peptide. It is worth mentioning that the azido-capped osteogenic peptide (2-Azido)-PEG5-BMP-2 still show excellent osteogenic activity (Figure S3). Bone marrow mesenchymal stem cells (BM-MSCs) incubated with (2-Azido)-PEG5-BMP-2 peptide could elicit potent alkaline phosphates (ALP) activity and matrix mineralization after 14 days, demonstrating the efficient osteoinductive ability of this molecularly modified BMP-2 peptide. These results indicated the successful preparation of the clickable mussel-derived peptide and azido-capped osteogenic biomolecule.
With the two peptides in hand, we then performed surface coating on Ti-based substrate to prepare a Zn2+ and BMP-2 peptide co-modified surface. TiO2-deposited quartz substrate (noted as TiO2) was used to mimic the surface of medical Ti biomaterials for all in vitro investigations. (DOPA)6-PEG5-DBCO-coated substrates (noted as DBCO) were prepared by incubated the TiO2-deposited quartz in phosphate buffer saline solution (PBS, 0.02 mM, pH=7.2) with the mussel-derived peptide. The catechol residues on peptide-coated substrates were then coordinated with Zn2+ after immersion into zinc acetate (ZnAc2) solution to obtain Zn2+-loaded surface (noted as Zn). Finally, (2-Azido)-PEG5-BMP-2 was conjugated through bioorthogonal click chemistry to prepare a Zn2+ and BMP-2 peptide co-modified surface (noted as Zn/BMP-2). Note that a BMP-2-modified surface without loading Zn2+ (noted as BMP-2) was also prepared as a control. The changes of surface roughness after peptide modification were first checked by atom force microscope (AFM) (Figure 1E). We could observe a clear change after (DOPA)6-PEG5-DBCO peptide modification. Although the Zn2+-modification led to negligible change, the second peptide modification using (2-Azido)-PEG5-BMP-2 resulted in a visibly increased surface roughness. Quantitative analysis confirmed the results (Figure 1F), indicating the efficiency of peptide modification through mussel-like adhesion and subsequent peptide clicking. It could be also found that the surface wettability showed significant improved after Zn2+ or BMP-2 peptide modification (Figure 1G and 1 H), probably due to the hydrophilicity of surface chelated Zn2+ and the amino acid sequence of BMP-2 peptide. The success of Zn2+-modification was confirmed by energy dispersive X-ray spectrometry (EDS) elemental mapping (Figure 1I). Although scanning electron microscopy (SEM) analysis on different surfaces showed no differences after co-modification (Figure S4), EDS elemental mapping revealed a homogeneously distributed Zn element on the surface after treatment with Zn2+. Surface elemental compositions were further determined by X-ray photoelectron spectroscopy (XPS) to further confirm Zn2+ and BMP-2 peptide co-modification (Figure 1J-1L). On the TiO2-deposited surface (TiO2 group), only signal peaks of C, Ti and O elements could be found. While in the groups of Zn and Zn/BMP-2, Zn 2p3/2 and Zn 2p1/2 signal peaks at 1021.75 Da and 1044.85 Da could be found (Figure 1L). Additionally, the N 1s signal (400.13 eV) could be found on the DBCO-TiO2, BMP-2 and Zn/BMP-2 groups. We could also observe a gradual increase in N1s signal when the DBCO-TiO2 surface was further modified with (2-Azido)-PEG5-BMP-2 (i.e., the BMP-2 and Zn/BMP-2 groups) (Figure 1K). For example, the N/Ti atomic ratio increased form 0.052 (the TiO2 group) to 0.686 (the BMP-2 group); the Zn/Ti atomic ratio increased form 0.000 (the TiO2 group) to 0.003 (the BMP-2 group) (Table S1). Quantitative analysis revealed that atom percentages of Zn and N elements on the Zn/BMP-2 surface was 2.57% and 4.89%, respectively, indicating the efficient of Zn2+ ion and BMP-2 peptide co-modification (Figure 1M). The durability of surface modified BMP-2 peptide and the sustained release of Zn2+ ion was further evaluated by incubation of the Zn/BMP-2 substrate in Dulbecco’s modified Eagle’s medium (DMEM, 37 °C) and PBS solution for 2 weeks, respectively. As shown in Figure 1N, the intensity of N 1s signal in XPS showed a slight decrease less than 15 %, indicating long-term stability of the surface modified BMP-2 peptide. In addition, the Zn2+ release was also comparable to previous reported Zn2+-modified surface by sequential sulfonation and magnetron sputtering 25 (Figure 1O). Although there was a burst Zn2+ release on the first day (0.14 ppm), it could slow down in the following days and showed a steady-state release (0.04 ppm) that could last 3-4 weeks. These results jointly indicated the success of Zn/BMP-2 co-modification on TiO2-based surfaces and potential display of long-term bioactivity on the surfaces.
2.2.Surface Cytocompatibility In Vitro
The in vitro biocompatibility of Zn2+ or BMP-2 modified surfaces was investigated by seeding a popular murine macrophage cell line (RAW 264.7 cells) and bone marrow mesenchymal stromal cells (BM-MSCs) on the surfaces. The Live/Dead staining was first carried out and the results show that the Zn2+, BMP-2 peptide, or their co-modified surfaces have slightly reduced dead cells compared to the bare TiO2 surface (Figure 2A). In addition, the adhesion morphology of BM-MSCs on different surfaces was also investigated to check the surface cytocompatibility. SEM images showed that BM-MSCs on the TiO2 surface exhibited less spreading state compared with cells on the Zn, BMP-2 and Zn/BMP-2 surfaces (Figure 2B). Further studies by cytoskeleton staining (FITC-phalloidin/DAPI) revealed that BM-MSCs on TiO2 exhibiting a relatively spherical morphology with almost no filopodia after 1 day of culture. On the contrary, the Zn, BMP-2 and Zn/BMP-2 surfaces exhibited better cell adhesion with polygonal shape and high expression of filamentous F-actin (Figure 2C). Note that, there were no obvious differences in cell adhesion and spreading state between all groups after 4 days of culture. Nevertheless, cell counting Kit-8 (CCK-8) assay was then employed to evaluate the proliferation of BM-MSCs and RAW 264.7 cells. The results showed that the surfaces of Zn, BMP-2 and Zn/BMP-2 afforded better cell viability for both RAW 264.7 and BM-MSCs compared with others. Interestingly, the Zn2+-containing surfaces (i.e., the groups of Zn and Zn/BMP-2) elicited the fastest proliferation of RAW 264.7 cells, while BMP-2 peptide-containing surfaces (i.e., the groups of BMP-2 and Zn/BMP-2) were inclined to enhance BM-MSCs proliferation (Figure 2D). This implied the potential immunoactivity and osteoinductivity for Zn2+ ion and BMP-2 peptide, respectively. The above results, together with lactic dehydrogenase (LDH) assay (Figure 2E), demonstrated that the Zn2+ and BMP-2 peptide co-modified surfaces had positive effect on the growth of both macrophages and BM-MSCs. The reduced cytotoxicity, improved adhesion and proliferation of both the immune cells and multipotential stem cells, would be a prerequisite to create favorable microenvironment for tissue regeneration.
2.3. Macrophage Phenotypic Switching In Vitro
The typical immunomodulation of biomaterials for tissue repair is the switching of pro-inflammatory M1 macrophages to anti-inflammatory M2 macrophages 18, 50, 51. For bone implants, studies have showed that excessive M1 phenotype macrophages cause bones absorption, which was an important factor leading to the failure of prosthesis implantation loosening 52, 53. It has been demonstrated that functional coatings, loaded with cytokines or active ions, could promote macrophage polarization towards an M2 phenotype both in vitro and in vivo and improving the integration of implants into bone tissue 21, 25. In this work, we successfully employed mussel adhesion-mediated ion coordination and molecular clicking strategy to incorporate Zn2+ and BMP-2 peptide onto bone implants. To investigate the effect of our strategy on the regulating of immune microenvironment, the polarization of macrophages on these surfaces was then studied. Macrophage-like cell RAW 264.7 in the resting state (M0) were first stimulated by lipopolysaccharide (LPS), and their phenotypic switching was then evaluated after culture on different surfaces (Figure 3A). As known, macrophages matured (M0) showed a round shaped cells, and these cells stimulated by LPS would develop into the pro-inflammatory M1 phenotype with a pancake-like shaped (Figure 3B). Due to the surface modification, the morphology of macrophages on different surfaces showed different status (Figure 3C). RAW cells on the surfaces without Zn2+ coordination (i.e., the TiO2, DBCO-TiO2 and BMP-2 groups) were predominantly a pancake-like shaped. On the contrary, macrophages could switch their morphology into an elongated shape on the Zn2+ modified groups (i.e., the Zn and Zn/BMP-2 groups). The significant increase of elongated cells on Zn and Zn/BMP-2 surfaces preliminary indicated the polarization of macrophages from pro-inflammatory M1 phenotype to anti-inflammatory M2 phenotype (Figure 3D).
As we known, the M1 and M2 phenotypes have distinguishable makers on the surface and cytokine secretion profiles 5, 13. Thus, the cytokine secretion and the relative expression levels of inflammatory gene markers was determined using enzyme-linked immunosorbent assay (ELISA) and real-time polymerase chain reaction (RT-PCR). Clearly, the secretion of pro-inflammatory cytokine TNF-α in groups of TiO2, DBCO-TiO2 and BMP-2 was significantly higher than that in the Zn2+-modified groups (Zn and Zn/BMP-2) (Figure 3E), indicating the M1 phenotype of macrophages on these surfaces. In contrast, the secretion of anti-inflammatory cytokine IL-10 in the groups of Zn and Zn/BMP-2 was efficiently increased, showing potent inflammation-attenuating effect of Zn2+ ions on macrophage phenotypic switching (Figure 3F). The M1 phenotype surface markers (CD86 and iNOS) and M2 phenotype surface markers (CD206 and Arg-1) in RAW 264.7 macrophages were further labelled by immunofluorescence staining. As shown in Figure 3G-3L, LPS stimulation upregulated the proportion of M1 macrophages (F4/80/CD86+ and F4/80/iNOS+, red) on the surfaces of TiO2, DBCO-TiO2 and BMP-2 groups. In contrast, anti-inflammatory M2 macrophages on the Zn and Zn/BMP-2 surfaces were dominant (CD206+ and Arg-1+, red). Interestingly, the Zn/BMP-2 group possessed the highest percentage of CD206 and Arg-1 positive cells and the lowest percentage of iNOS and CD86 negative cells compared with others. It was also mentioning that the BMP-2 group also showed a slight downregulation of the M1 phenotype markers (iNOS and CD86), and slight upregulation of the M2 phenotype markers compared with the controls (TiO2 and DBCO-TiO2), probably due to the potential of BMP-2 protein in regulating local osteoimmune microenvironment 54.
Quantitative RT-PCR analysis also showed similar results. The gene expression levels showed downregulation for pro-inflammatory TNF-α and upregulation for anti-inflammatory IL-10 in the case of Zn2+ modification, further indicating the potential of Zn2+-modified groups for the switch of macrophage to M2 phenotype (Figure 3M and 3N). The expression levels of surface markers associated with M1 (e.g., CCR7) and M2 (e.g., CD206) phenotypes were further determined. The efficiently suppressed expression of CD206 and significantly increased expression of CCR7 in Zn and Zn/BMP-2 further indicated the positive immunomodulatory function of Zn2+-modified groups for M2 phenotype polarization (Figure 3O and 3P). Note that the above results all reflected a phenomenon that the Zn2+ and BMP-2 peptide co-modified surface elicited the most efficient M2 phenotype polarization. This is probably due to the potential immunomodulatory role of BMP-2 peptide which may motivate the immunoactivity of Zn2+ ions 54. Additionally, we also found that the gene expression levels of osteogenic cytokines secreted from the M2 macrophages (e.g., BMP-2 and VEGF) were also upregulated on the Zn2+-containing surfaces, in which the Zn/BMP-2 surface showed the highest expression osteogenic cytokines (Figure 3Q and 3R). The Zn2+-enhanced osteogenic cytokine secretion, together with the BMP-2 peptide-enhanced immunomodulatory function, jointly revealed the different but overlapping roles of Zn2+ ion and BMP-2 peptide in immunomodulation and osteoinduction. Therefore, their combination (i.e., Zn/BMP-2) may provide more favorable microenvironment and create improved bone repair potentials at bone-to-implant interfaces.
2.4. Immunomodulation-Enhanced Osteogenic Differentiation In Vitro
The immunomodulatory effect of Zn/BMP-2 surface on in vitro osteogenic differentiation was also investigated by using macrophage conditioned medium (MCM) (Figure 4A). The MCMs were prepared by collecting the supernatant of cell culture medium after seeding macrophages on different surfaces for 24 h. The collected MCMs were then used to induce osteogenic differentiation of BM-MSCs. At day 14, the expressions of three osteogenic-related proteins, including alkaline phosphates (ALP), calcium binding proteins and osteopontin (OPN), were determined to evaluate the efficiency of osteogenic differentiation. ALP staining clearly showed that the MCMs derived from Zn2+ or BMP-2 peptide modified surfaces (i.e., the Zn, BMP-2 and Zn/BMP-2 groups) could elicit significantly higher ALP activity compared with the controls (TiO2 and DBCO-TiO2) (Figure 4B). ALP immunofluorescence images further confirmed the results, in which the Zn/BMP-2 derived MCM could induce the highest level of ALP activity (Figure S5). Quantitative analysis of ALP staining revealed that the ALP activity in the Zn/BMP-2 derived MCM were 5.06, 4.22, 2.00, and 1.80-folds higher than that of TiO2, DBCO-TiO2, Zn and BMP-2 groups, respectively (Figure 4E and 4F). In addition to ALP protein level, similar trends were also found in the expressions of other osteogenesis-related proteins (calcium binding proteins and OPN). For example, Alizarin Red S (ARS) staining for calcium binding proteins in the mineralized matrix at day 14 showed that the size and quantity of the mineral nodules in Zn/BMP-2 group was 1.50 and 1.33-fold higher than that produced from the Zn and BMP-2 groups, indicating the most efficient calcium deposition in BM-MSCs cultured with Zn/BMP-2 derived MCM (Figure 4C and 4G). OPN immunofluorescence staining also confirm the most efficient enhancement of osteogenic differentiation of Zn/BMP-2 derived MCM, in which the relative OPN expression could increase 46.10% and 34.11% as compared with the Zn and BMP-2 groups, respectively. Apart from the above osteogenesis-related proteins, we also investigated the expression levels of osteogenesis-related genes, including ALP, runt-related transcription factor 2 (Runx2), type I collagen (Col-1) and OPN (Figure 5I-5L). As expected, the mRNA expressions of these osteogenesis-related genes in BM-MSCs cultured with Zn/BMP-2 derived MCM were all significantly higher than others, confirming the potent ability of Zn/BMP-2 derived MCM to enhance osteogenic differentiation. Overall, these results demonstrated that the combination of immunoactive Zn2+ ion and osteoinductive BMP-2 peptide would be more conducive to the regulation of macrophage phenotypic switch from M1 to M2, and the secretion cytokines would provide an optimum osteoimmunomodulatory microenvironment and lead to immuno-enhanced osteogenesis.
2.5. Macrophage Phenotypic Switching In Vivo
The above in vitro studies have demonstrated that the Zn/BMP-2 co-modified surface possessed excellent properties including improved cytocompatibility to accelerate cell growth, efficient macrophage phenotype regulation from M1 to M2, and potent osteoimmunomodulatory activity to enhance osteogenicity. These virtues guide us to further investigate the immunomodulatory activity and osteogenicity in vivo. A commercially available cortical bone self-tapping Ti-based screw was used as the model implant. The screws were treated the same as that of TiO2-deposited substrates to obtain screws different surface modification (named as DBCO-TiO2, Zn, BMP-2 and Zn/BMP-2, respectively). The untreated Ti screw was also named as TiO2, since the tight and continuous TiO2 layer on it. Screw implantation in the femoral condyles of rats was then performed according to standard surgical protocol. To minimize the destruction of surface layer during implantation, a drill with diameter between the concave and convex thread of screw was used in animal experiment (within the yellow line Figure S6). After 4 days implantation, the rat femoral condyles containing the implanted Ti screws were harvested for histological hematoxylin-eosin (H&E) staining. H&E images showed that screws in the groups of Zn, BMP-2 and Zn/BMP-2 exhibited milder inflammatory response, thinner fibrous layers, more complete bone structures (within the black lines) as compared with that of the TiO2 and DBCO-TiO2 groups (Figure 5A). The thinnest fibrous layer was observed in the Zn-BMP-2 group, while the milder inflammatory response in Zn and Zn-BMP-2 screws were comparable (Figure 5D and 5E). It was worth mentioning that the BMP-2 group showed slightly less improvement on inhibition of fibrosis and inflammatory response, but still indicating the potential immunoactive of the BMP-2 peptide 54. On the contrary, the thick fibrous layers and large-scale infiltration of inflammatory cells around the bare screw will be reminiscent of the reason of most implantation failures. In addition, we further evaluated the phenotypic conversion of macrophages around the screws using immunofluorescence staining. Inflammatory cell infiltration was determined by labelling CD68, and the M1 to M2 phenotype switching was determined by labelling CCR7 and CD206-postive cells, respectively (Figure 5B). Immunofluorescence images showed CD206-postive cells (i.e., M2 phenotypic macrophages) around Zn, BMP-2 and Zn/BMP-2 screws were significantly more than that around the control screws in TiO2 and DBCO-TiO2. Quantitative analysis showed that the proportion of M2 macrophages in Zn, BMP-2 and Zn/BMP-2 groups was about 1.92, 1.50 and 2.25 folds higher than that in the TiO2 group, respectively (Figure 5F). In contrast, screws in the TiO2 and DBCO-TiO2 groups resulted in higher number of CCR7-postive cells (M1 phenotypic macrophages) than others, while screws in the Zn/BMP-2 groups elicit lowest amount M1 macrophages (Figure 5G). Studies on immunohistochemical staining further revealed that, the deposition of anti-inflammatory cytokine IL-10 dramatically increased in the Zn (10.63%), BMP-2 (8.16%) and Zn/BMP-2 (15.61%) groups as compared with the TiO2 control (5.76%). This result, together with the reduced number of infiltrated macrophages and increased conversion of M1 to M2 phenotypes, confirmed the most potent immunomodulatory activity of Zn/BMP-2 in vivo and implied the potential to reverse the injury-caused excessive inflammation and enhance interfacial osteogenesis around the implants.
2.6 Osseointegration In Vivo
As we have verified that the Zn/BMP-2 surface possessed the optimal immunoactivity to tune macrophage M1/M2 polarization in vitro and in vivo and significantly enhanced osteogenesis in vitro, the new bone formation and osseointegration at the screw-to-bone interface was then evaluated after 8 weeks of implantation. The in vivo toxicity reaction of the implanted screws on heart, liver, spleen, lung, and kidney were first evaluated, and no significant tissue toxicity was observed, indicated the cytocompatibility of our strategy (Figure S7). Then, Micro-CT 3D reconstruction and histological analysis of the harvested bone tissues with screws were performed. The reconstructed micro-CT 3D images showed that the Zn/BMP-2 group produced the highest amount of newly formed bone tissue around the screws, while the TiO2 control was only covered with few disconnected bone tissues (Figure 6A). Quantitative analysis further confirmed this result. Screws in the Zn/BMP-2 group had the highest percentage of bone mineral density (BMD) and bone volume to tissue volume (BV/TV) and exhibited the best trabecular structural features of the new bone under the same volume of interest (VOI) (Figure 6B). The value of BV/TV was 1.37 and 1.26-fold higher in the co-modified Zn/BMP-2 group (82.06 ± 1.46%) as compared to screws in Zn (59.77 ± 3.89%) and BMP-2 groups (65.35 ± 3.63%), respectively. In addition, the values of trabecular separation (Tb.Sp), trabecular thickness (Tb.Th), and trabecular number (Tb.N) in the Zn/BMP-2 group were 70.92%, 128 %, and 162% of that in Zn group, 84.28%, 137%, and 143 % of that in BMP-2 group. This result is due to the synergy of immunoactive Zn2+ ion and osteoinductive BMP-2 peptide, while a mono-modification (e.g., the Zn or BMP-2 group) might not provide the most favorable immunomodulatory microenvironment for bone regeneration. Sequential fluorescence labelling was performed by using Calcein (green) and Alizarin Red (red) to mark the newly formed bone, and similar results were obtained (Figure 6C and 6D). We could observe a large area of new bone mineralization on the screw surface of Zn/BMP-2 group (18.80%), while a gradually decreased bone mineralization in the groups of BMP-2 (11.10%), Zn (8.22%), DBCO-TiO2 (4.20%) and TiO2 (3.74%). Quantitative analysis showed that the bone-implant contact ratio (BIC) for Zn/BMP-2 was significantly higher (more than 2-folds in the value) than the other groups (Figure 6E).
Since the stable connections between the implants and surrounding bone tissue are closely related to the clinical outcomes of implants, we then applied a biomechanical pull-out experiment to test the anchorage force of Ti screws in bone tissue. As shown in Figure 6F, the maximal push-out forces in the Zn, BMP-2 and Zn/BMP-2 groups were all significantly improved as compared with the TiO2 control, indicating the excellent stability. For example, the Zn/BMP-2 group showed the highest maximum pull-out force (203.3 ± 14.3 N) that was nearly 2.1-fold higher than that of the TiO2 control (98.6 ± 16.0 N). These results confirmed that the Zn/BMP-2 surface could significantly promote interfacial osteogenesis and enhance osseointegration invivo. Overall, we verified that both the immunomodulatory function and direct osteogenicity were crucial to bone implants, according to the overlapping but distinct stages in tissue regeneration (i.e., immune responses and healing processes). In this work, an osteoinductive and immunomodulatory dual-effect implant could be readily obtained by co-modification of immunoactive Zn2+ ion and osteoinductive BMP-2 peptide through a mussel adhesion-mediated ion coordination and molecular clicking method. The dual-effect implant enabled not only M2 phenotype switching but also direct osteoinductivity, which synergistically create a favorable microenvironment in vivo for bone regeneration (Figure 6G). Therefore, this work may provide new idea and solution for improving the clinical outcomes of bone implants.