Micromorphological effect of calcium phosphate coating on compatibility of magnesium alloy with osteoblast

Abstract Octacalcium phosphate (OCP) and hydroxyapatite (HAp) coatings were developed to control the degradation speed and to improve the biocompatibility of biodegradable magnesium alloys. Osteoblast MG-63 was cultured directly on OCP- and HAp-coated Mg-3Al-1Zn (wt%, AZ31) alloy (OCP- and HAp-AZ31) to evaluate cell compatibility. Cell proliferation was remarkably improved with OCP and HAp coatings which reduced the corrosion and prevented the H2O2 generation on Mg alloy substrate. OCP-AZ31 showed sparse distribution of living cell colonies and dead cells. HAp-AZ31 showed dense and homogeneous distribution of living cells, with dead cells localized over and around corrosion pits, some of which were formed underneath the coating. These results demonstrated that cells were dead due to changes in the local environment, and it is necessary to evaluate the local biocompatibility of magnesium alloys. Cell density on HAp-AZ31 was higher than that on OCP-AZ31 although there was not a significant difference in the amount of Mg ions released in medium between OCP- and HAp-AZ31. The outer layer of OCP and HAp coatings consisted of plate-like crystal with a thickness of around 0.1 μm and rod-like crystals with a diameter of around 0.1 μm, respectively, which grew from a continuous inner layer. Osteoblasts formed focal contacts on the tips of plate-like OCP and rod-like HAp crystals, with heights of 2–5 μm. The spacing between OCP tips of 0.8–1.1 μm was wider than that between HAp tips of 0.2–0.3 μm. These results demonstrated that cell proliferation depended on the micromorphology of the coatings which governed spacing of focal contacts. Consequently, HAp coating is suitable for improving cell compatibility and bone-forming ability of the Mg alloy.

We previously developed HAp and octacalcium phosphate (OCP) coatings for Mg/Mg alloys by a novel single-step chemical solution deposition method. [32][33][34][35][36][37][38] The type of Ca-P coating can be controlled with the pH of the treatment solution. [35,36] OCP and HAp coatings have two-layer structures: the inner layer is continuous and consists of nanocrystals, with a thickness of 1 to 5 μm depending on the treatment period; the inner layer of the OCP coating contains nanopores, with that of the HAp coating is microscopically dense. [35] Both types of Ca-P coatings exhibit protective effect in cell-culture medium and soft tissue of mouse. Both coatings suppressed subcutaneous inflammation and foreign body reaction in mouse by reducing the corrosion of substrate. [39] The protectiveness of OCP coating was slightly lower than that of HAp coating. [35,36] The difference in corrosion resistance is expected to affect cell behavior on OCP-and HAp-coated Mg/Mg alloys.
The outer layer consisted of highly crystalline platelike OCP or rod-like HAp crystals, growing from the inner layer along the [002] axis of OCP and HAp crystals. Therefore, the edge of (001) or (010) OCP planes or the tip of (001) HAp plane is preferentially exposed to the surface. The length of OCP plates and HAp rods varied from about 0.5 to 5 μm with increasing treatment period. The width and thickness of OCP plates were about 1 μm and <30 nm, respectively. The diameter of HAp rods was about 150 nm for pure Mg treated at 363 K for 2 h. [35] As cell behavior depends on the surrounding topography [31,40], the crystal shape and size of HAp particles can affect the proliferation behavior of osteoblast-like cells. [30] Thus, the difference in morphology of OCP and HAp coatings could cause variances in cell adhesion and growth properties.
In this study, cell proliferation and cell adhesion behavior of OCP-and HAp-coated Mg alloy was examined using osteoblast-like MG-63 cells and AZ31 alloy.

OCP and HAp coating of magnesium alloy
AZ31 disks of 15 mm diameter and 1 mm thickness were cut from extruded rods (Osaka Fuji Co., Amagasaki, Japan). The composition of the AZ31 rod is shown in Table 1. The surface of disks was ground with SiC papers (Buehler, IL, USA) up to #1200 and rinsed ultrasonically in acetone. Mechanically ground AZ31 disks were named Mpol-AZ31.
Coating treatment solutions were prepared with 500 mmol l −1 ethylenediaminetetraacetic acid (EDTA) calcium disodium salt hydrate (C 10 H 12 CaN 2 Na 2 O 8 , Ca-EDTA) solution, 500 mmol l −1 potassium dihydrogenphosphate (KH 2 PO 4 ) solution, and 1 mol l −1 sodium hydroxide (NaOH) solution. The same volumes of the Ca-EDTA and KH 2 PO 4 solutions were mixed and the pH was adjusted to 6.1 or 8.9 with the NaOH solution. Mpol-AZ31 disks were immersed in the treatment solutions at 90°C for 2 h. The pH of the solutions did not change after the treatment. OCP and HAp coatings were formed at pH 6.1 and 8.9, respectively. OCP-and HAp-coated AZ31 specimens were named OCP-and HAp-AZ31, respectively. The crystal structure was analyzed by X-ray diffraction (XRD) (RINT Ultima III, Rigaku, Tokyo, Japan). The surface and cross-sectional morphology of the coatings was observed by scanning electron microscope (SEM; FEI Quanta FEG250, OR, USA and Miniscope TM3000, Hitachi, Tokyo, Japan). Cross-section specimens were prepared by scraping off the OCP and HAp coatings with a cutter.

Measurement of surface pH and peroxide production
Surface pH of OCP-and HAp-AZ31 was measured using phenol red pH test paper (PR, pH 6.6-8.2, Advantec, Tokyo, Japan). A piece of pH test paper was placed on the surface immediately after a tiny amount of pure water was dropped on the surface. The surface pH was decided depending on the color of the pH test paper.
The peroxide production capacity of Mpol, OCP-and HAp-AZ31 surfaces was examined using produced on specimen surfaces was semi-quantified according to the graduation of blue color on the test paper. A piece of test paper was moistened with 10 μl of pure water, and then lightly pressed down on the sample surface for 10 s. The color of the test paper was observed.

Cell proliferation analysis by fluorescent staining
Human osteosarcoma cells MG-63 (cell no. RCB1890, RIKEN BioResource Center, Tsukuba, Japan) were used for cell proliferation analysis. Mpol-, OCP-and HAp-AZ31 disks were immersed in acetone for 10 s for sterilization and then dried in air. Then, the disks were placed in a 24-well plate (well diameter is about 16 mm). MG-63 cells were seeded into 24-well plates at a density of approximately 5 × 10 3 cells per well, in minimum essential medium (MEM) supplemented with 10 vol.% fetal bovine serum (FBS), 100 U ml −1 penicillin, and 100 μg ml -1 streptomycin (Invitrogen, CA, USA). After incubating at 37°C for 24 h in a 5% CO 2 incubator, disks seeded with cells were each transferred into a φ15 cm dish and 50 ml of culture medium was added. Thereafter, medium was changed every 24 h. On days 2 and 6 of culture, the cytoplasm of living cells and the nuclei of dead cells were stained with calcein and propidium iodide (PI), respectively. The nuclei of living and dead cells were also stained with Hoechst33342. The cell culture procedure is shown in Figure 1(a).
Nine (3 × 3) images surveying the disk surface were taken by a fluorescence microscope (DMIL, Leica Microsystems, Wetzlar, Germany) with 1.25 × objective lens, following which they were merged to exhibit the entire surface of each disk. The surface of HAp-AZ31 after six days of culture was observed with an optical microscope and SEM.

Cell proliferation assay
The proliferation of MG-63 cells was measured using a luminescent cell viability assay kit (CellTiter-Glo™, Promega, WI, USA). Culturing of MG-63 cells on Mpol-, OCP-and HAp-AZ31 disks was performed in the same procedure as that for cell staining: except that medium-replacement was not performed after transferring the disks with cells to a φ15 cm dish at day 1 of culture as shown in Figure 1(b). At days 1 and 6 of culture, medium was removed from each well or dish, and cells were washed with 1 ml of phosphate buffered saline (PBS). Then, 0.5-1 ml of trypsin-EDTA (0.05%) (Gibco™, Thermo Fisher Scientific Inc., MA, USA) was added, and the 12-well plates or dishes were placed at 37°C for 5 min. After cells on the disks were detached, 0.5-1 ml of room temperature medium was added and cells were resuspended. Subsequently, the entire cell suspension was transferred to a new 24-well plate and incubated for 4.5 h.
CellTiter-Glo reagent was diluted twofold and 250 μl of the diluted reagent was added to each well after removing the medium. The 24-well plate was gently shaken for 2 min and then left standing at room temperature for 10 min. Then, 200 μl of the sample solution was transferred to a 96-well white solid plate, and luminescence was measured with micro-plate reader (MTP-880Lab, Corona, Ibaraki, Japan).
Magnesium ions in the used medium collected at days 1 and 6 were quantified by a colorimetric method using Xylidyl blue-I. [41,42]  diameter, growing from the inner layer. The spacing between HAp crystal tips was 0.2-0.3 μm. HAp rods formed were denser than OCP plates. At day 2, HAp-AZ31 exhibited a more homogeneous distribution of living cells than OCP-AZ31, on which living cells were sparsely distributed. At day 6, cells had almost completely covered the HAp-AZ31 surface, while cells on OCP-AZ31 surfaces formed homogeneously distributed colonies. Dead cells on HAp-AZ31 were obviously localized, while dead cells on OCP-AZ31 were almost homogeneously distributed amongst the living cells at days 2 and 6.

Cell adhesion morphology on OCP-and HAp-AZ31
OCP-and HAp-AZ31 showed differences in corrosion morphology. OCP-AZ31 exhibited homogeneously distributed micro-pits with white corrosion products as shown in Figure 4(e). HAp-AZ31 exhibited localized distribution of dead cell regions, some of which associated with visible filiform and round corrosion pits ( Figure  4(f)).
To examine the relationship between corrosion and cell-death on HAp-AZ31, magnified fluorescence and optical microscopic images, and a SEM image of an area containing dead-cell regions were examined: these are

Immunostaining of cells
MG-63 cells were cultured on OCP-and HAp-AZ31 using the same procedure described for the cell viability assay shown in Figure 1(c). At day 1 of culture, focal contacts and actin filaments were stained with paxillin and phalloidin, respectively, according to the protocol of the manufacturer. After medium was removed, 0.5 ml of 4 w/v% paraformaldehyde in PBS was added to each well, and cells were washed three times with 0.5 ml of PBS. Then, cells were incubated in 0.5 ml of PBS containing 1 w/v% bovine serum albumin (BSA) and 1 w/v% TritonX-100 on ice for 5 min, followed by washing with PBS. Subsequently, cells were incubated in 0.5 ml of 1 w/v% BSA in PBS on ice for 30 min. After the supernatant was removed, 50 μl of 1 μg ml −1 anti-paxillin antibody solution (Rabbit, H-114 sc-5574, Santa Cruz Biotechnology, Inc., TX, USA) was added to each well and incubated for 1 h at room temperature, away from light. After incubation, cells were washed with PBS, and 50 μl of secondary antibody solution (1 μg ml −1 Alexa Fluor 555 goat anti-rabbit IgG (H+L)) and 4 U ml −1 Alexa Fluor 488 phalloidin was added to each well and incubated for 1 h at room temperature, away from light. Afterwards, cells were washed with PBS, and 50 μl of 1 μg ml −1 Hoechst PBS solution was added and incubated for 15 min, following which they were washed with PBS. Antifade reagents was dropped on the surface of samples and a cover glass was placed on top for observation.

Statistical analysis
For cell proliferation tests, the results (cell number and Mg 2+ ion concentration of the medium) were compared using a one-way analysis of variance (ANOVA) with Tukey-Kramer post hoc test for each incubation period. Statistical significance was set at p value less than 0.05. Figure 2 shows XRD patterns of AZ31 disks treated at pH 6.1 and 8.9. Diffraction peaks mainly from (002) and (004) planes of OCP and HAp were observed on pH 6.1and pH 8.9-treated disks, respectively. This result agrees with the previous report. [36] The formation of OCP and HAp coatings on AZ31 disks was confirmed.  Mpol-AZ31 surfaces. Cells proliferated about by 30 times on HAp-AZ31 from day 1 to day 6, whereas cell growth on OCP-and Mpol-AZ31 surfaces was remarkably suppressed.

Morphology of OCP and HAp coating
Magnesium ion concentrations of media collected at days 1 and 6 are shown in Figure 6(c) and (d). Since the whole medium was refreshed at day 1, the value at day 6 is the accumulation of Mg 2+ ions released from days 2 to 6. OCP and HAp coatings significantly suppressed Mg 2+ ion release from the AZ31 substrate at day 1. The suppression effect of the HAp coating was more pronounced than that of the OCP coating. The release rate was reduced after day 1. OCP-and HAp-AZ31 specimens showed slight Mg 2+ ion release, and Mpol-AZ31 showed significant Mg 2+ ion release, at day 6.
Cell density was plotted as a function of the Mg 2+ ion concentration, as shown in Figure 6(e) and (f). At day 1, there was a certain correlation between cell density and Mg 2+ ion concentration, as the cell density decreased with a corresponding increase in Mg 2+ ion concentration. At day 6, cell density on OCP-AZ31 became drastically lower than that on HAp-AZ31, although there was no statistically significant difference in Mg 2+ ion concentration. These results suggest that the initial corrosion of the AZ31 substrate slightly affected the initial cell survival and/or cell adhesion. The subsequent corrosion with lower corrosion rate did not dominate the cell growth.
shown in Figure 5. Obvious dead cell regions are encircled with dotted line on the fluorescence and optical microscopy images. Scratches in the upper side of the disk were formed with a pair of tweezers, which was used to remove the disk from the cell-culture plate. The largest dead-cell region on the right side of the disk corresponds to the largest filiform corrosion. Relatively small dead-cell regions sometimes corresponded to visible micro-pits with a diameter of around 100 μm, but did not always correspond to visible corrosion pits. Even with a scanning electron microscope, traces of visible corrosion were not observed on the surface of some dead-cell regions.

Cell proliferation on OCP-and HAp-AZ31
Cell density was obtained by counting cells removed from OCP-, HAp-and Mpol-AZ31 and polystyrene dishes (PS) at days 1 and 6, which are shown in Figure  6(a) and (b). The obtained cell densities might be slightly lower than the actual values because some cells presumably remained attached to the specimen surfaces. At day 1, there was no statistically significant difference in cell density between specimens; however, OCP-and HAp-AZ31 showed slightly higher cell densities than Mpol-AZ31, and HAp-AZ31 had a slightly higher cell density than OCP-AZ31. At day 6, HAp-AZ31 had significantly higher cell density than OCP-and from medium, which according to a previous study, consisted of calcium phosphate and Mg(OH) 2 . [39] Under the coatings, OCP-AZ31 locally showed a dark gray layer of corrosion product of Mg(OH) 2 , with a thickness of 2-4 μm (Figure 7(g) and (j)). HAp-AZ31 had a trace layer of corrosion product (Figure 7(h)). Mpol-AZ31 had a layer of corrosion product with a thickness of 10-50 μm (Figure 7(i) and (l)). The smaller  of rod-like HAp crystals on HAp-AZ31, and actin filaments ran along the longitudinal axis of the cell body. The density of focal contacts appeared to be lower on OCP-AZ31 than that on HAp-AZ31, which is presumably attributed to the wider spacing between crystal tips on OCP-AZ31 than those on HAp-AZ31: 0.8-1.1 μm and 0.2-0.3 μm, respectively. The morphologies of focal contacts and actin filaments were also observed at day 3, with similar results being obtained.
After plate-like OCP and rod-like HAp crystals were polished off surfaces with diamond paste, the morphologies of focal contacts and actin filaments of MG-63 cells were observed. The remaining inner continuous layer was confirmed by XRD measurement (Figure  9(a) and (b)) and SEM observation (Figure 9(c) and (d)). Figure 9(e) and (f) show fluorescence microscopy images of focal contacts, actin filaments, nuclei, and a merged image of a cell cultured for one day on OCP-and HAp-AZ31, without plate-like or rod-like crystals. In both cases, focal contacts were homogeneously distributed in the cell body, while actin filaments preferentially ran along the longitudinal axis of the cell body. It was demonstrated that the localization of focal contacts disappeared after removing plate-like crystals on OCP-AZ31.
formation of corrosion products underneath the HAp coating than those formed underneath the OCP coating corresponds to the lower Mg 2+ ion concentration of HAp-AZ31 compared with OCP-AZ31 (Figure 6(c) and (d)).  Figure 8(e) and (f) show surface SEM images of as-prepared OCP-and HAp-AZ31, with the same magnification as that of the fluorescence images. The shape and density of focal contacts formed on OCP-AZ31 were similar to those of tips of plate-like OCP crystals, and actin filaments preferentially distributed on focal contacts. Dot-shaped focal contacts were distributed as densely and homogeneously as tips capability of MC3T3-E1 cells [44], H 2 O 2 generation on the specimens was examined. Figure 11 shows optical images of a piece of wet H 2 O 2 test paper placed on various specimens. The test papers for Mpol-AZ31 and pure Mg exhibited an existence of about 0.5 and 2.0 mg l −1 of H 2 O 2 , respectively. In contrast, the test papers for OCPand HAp-AZ31 was not colored. No generation of H 2 O 2 in MgCl 2 solution and Mg(OH) 2 suspension indicates that Mg 2+ ions did not cause coloring of the H 2 O 2 test paper. It was revealed that H 2 O 2 was generated from AZ31 and pure Mg surfaces and that OCP and HAp coatings suppressed H 2 O 2 generation. Figure 10 shows optical images of a piece of pH paper placed on Mpol-, OCP-and HAp-AZ31. The pH of OCP and HAp coatings was around 7.4 and greater than 8.2, respectively. The OCP coating had a lower pH than the HAp coating.

Surface chemical properties of OCP and HAp coatings
Because it was reported that MgO and ZnO powders generated reactive oxygen species and exhibited antibacterial activity against Gram-positive bacteria [43] and that galvanically coupled Mg-Ti particles generated reactive oxygen species which exhibited a killing The relatively alkaline property of the HAp coating ( Figure 10) seems to have no effect on cell adhesion or proliferation. Prevention of H 2 O 2 generation with OCP and HAp coatings ( Figure 11) allowed discussion of the morphological effect of the coatings on cell behavior. The H 2 O 2 generation on Mpol-AZ31 presumably played an important role in the cell compatibility. However, further investigation is necessary to elucidate the effect of surface pH and H 2 O 2 generation on cells on Mg alloy surfaces.
Cells around obvious corrosion pits were selectively dead on HAp-AZ31 surfaces (Figure 4). Corrosion of Mg alloy generates mainly Mg 2+ and OHions and H 2 gas, and a part of Mg 2+ and OHions deposited as Mg(OH) 2 . [39] Seuss et al. [45] reported that a pH increase of medium reduced the proliferation of Hela cells. Formation of H 2 gas bubbles and Mg(OH) 2 gel around cells might separate the cell bodies from medium and prevent the cell metabolism. The cell death was thus attributed to a pH increase and H 2 and Mg(OH) 2 generation caused by corrosion. In contrast, cells were also dead in the regions without microscopically visible

Adhesion and proliferation of MG-63 cells on OCP-and HAp-AZ31
OCP and HAp coatings improved MG-63 cell proliferation on AZ31 alloy by reducing the corrosion of the substrate alloy (Figures 4 and 6). The improvement of cell proliferation with the HAp coating was macroscopically more pronounced than that with the OCP coating. The higher proliferation on the HAp coating is attributed to the smaller release of Mg 2+ ions, and the denser packing of crystal tips for HAp-AZ31 than those for OCP-AZ31 (Figures 6 and 8) as mentioned in Section 4.2. However, such a difference in cell proliferation would disappear for a long period of in vivo implantation time: only a slight difference was observed between OCP-and HAp-AZ31 in the thickness of the fibrous tissue layer formed due to foreign body reaction (Supplemental Figure S2). This is attributed to the disappearance of OCP plate-like crystals in 16 weeks of implantation (Supplemental Figure  S3). [39]  permeate through the coating layer, leading to cell death. Further investigation is necessary to identify the dominant factor of local cell death.
corrosion pits (Figures 3 and 4). In this situation, local corrosion underneath OCP and HAp coatings (Figure 7) could occur and H 2 gas and Mg 2+ and OHions could  not observed after a long period of in vivo implantation. Both OCP-and HAp-AZ31 showed similar large round pits in subcutaneous tissue of mouse after 16 weeks of implantation . [39] This was attributed to the low diffusivity in soft tissue compared with that in bulk medium. The presence of immune cells like macrophages might also influence the corrosion morphology of Ca-P-coated Mg alloy.
Corrosion on HAp-AZ31 was remarkably localized, compared with that on OCP-AZ31 (Figure 4(e) and (f)), and caused the localized cell death on HAp-AZ31 ( Figure  4(c) and (d)). The corrosion morphology was attributed to the microstructure of the inner continuous layer of the coatings: OCP and HAp coatings showed nanopores and microscopically dense structure, respectively. [35] On the other hand, this difference in corrosion morphology was  c-plane possibly enhanced focal contact formation on the tips of rod-like HAp crystals and ridges of plate-like OCP crystals.

Conclusions
Cell adhesion and proliferation behavior were examined on OCP and HAp coatings formed on AZ31 using human osteosarcoma MG-63 cells. HAp and OCP coatings improved cell viability by preventing corrosion of the substrate AZ31. The surface pH of the HAp coating was slightly higher than that of the OCP coating, and both coatings suppressed H 2 O 2 generation; however, further investigation is necessary to understand the effect of such surface chemical properties on cell behavior. HAp coating remarkably improved cell adhesion and proliferation, and showed a dense and homogeneous distribution of living cells, while dead cells were localized on/ over corrosion pits of the substrate. OCP coating showed a similar initial cell adhesion density on the HAp coating; however, cell elongation and subsequent cell proliferation were reduced, leading to sparsely distributed living and dead cells. Cells formed the focal contacts on the tips of rod-like HAp and plate-like OCP crystals. The focal contacts on HAp coating with narrower tip spacing showed higher density than those on OCP coating. Consequently, the shape, orientation and density of Ca-P crystals in the coatings have the greatest influence on the spacing of the focal contacts that governs cell adhesion, elongation and proliferation; a higher density of focal contacts leads to increased cell proliferation. In addition, the local cell death on HAp-AZ31 indicates the importance of entire surface observation for the evaluation of cell compatibility.