Formation of hydroxyapatite on titanium implants in vivo precedes bone-formation during healing

The bone material interface has been an area of intense study over many decades, where studies of the healing process ranging from simple mineral deposition in vitro to actual healing in vivo have given important clues to the importance of calcium minerals in the bone/implant interface. Here, the authors use a combination of in vitro cell culture methods and in vivo implantation to study how the role of the spontaneously formed hydroxyapatite layer on Ti-implants for the in vivo-healing into the bone tissue of rat tibia. Initial experiments were made in reduced systems by incubation of TiO2 in cell culture medium and analysis by time of flight secondary ion mass spectrometry (ToF-SIMS) and energy-dispersive x-ray spectroscopy followed by subsequent exposure of human embryological stem cells analyzed by von Kossa staining and environmental scanning electron microsopy. In vivo studies of the bone–material interface was analyzed by ToF-SIMS depth profiling using both C60 þ ions as well as a gas cluster ion source beam, Ar1500 þ as sputter source. The low ion yield of the Ar1500 þ for inorganics allowed the inorganic/organic interface of the implant to be studied avoiding the erosion of the inorganic materials caused by the conventional C60 þ beam. VC 2017 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). https:// doi.org/10.1116/1.4993986


I. INTRODUCTION
The bone-material interface of metal implants can be metaphorically described as an archaeological excavation site, where studies can deliver important clues to the mechanism of bone healing at material surfaces.][3][4][5][6][7][8][9][10][11][12][13][14][15] However, studies of the established material/bone interface cannot give definite answers to the time-dependence of the subprocesses of bone formation.The only possible way to obtain this information is to make time-lapse studies of bone healing at implants.The simplest system for such studies is to follow the mineral deposition onto titanium oxide from simulated body fluids, 16 or to analyze the adsorption of proteins and cells from blood onto titanium. 17More complex studies can be made by analyzing bone healing at implants after various time periods of healing. 13ecently, we found that hydroxyapatite (HA) is formed spontaneously during incubation of Ti-implants with cell culture medium or blood. 18The apatite formed was low crystalline carbonated hydroxyapatite as detected by x-ray photoelectron spectroscopy after incubation with cell culture medium [Dulbecco's modified Eagle's Medium (DMEM)] or low CaCO 3 , as detected as CaO þ by Time of flight secondary ion mass spectrometry (ToF-SIMS) after incubation in whole blood, suggesting that blood decreased the rate of HA-formation. 18The HA layer on titanium is generally thought of as being formed by passive precipitation, but recent data on recovered implants show that Ti-surface is coated with Ca, indicating a binding between Ti and Ca as an early event in the healing process. 19he aim of the present study was to elucidate the role of the spontaneously formed HA layer on Ti-implants for the in vivo-healing into the bone tissue of rat tibia.The bone-material interface was analyzed by depth profiling using ToF-SIMS with both C 60 þ ions as well as a gas cluster ion source beam (GCIB) using Ar 1500 þ .Experiments were also made in a reduced systems, incubation of TiO 2 in cell culture medium, analyzed by ToF-SIMS and energy-dispersive xray spectroscopy (EDX), and subsequent exposure of human embryological stem cells (hESCs) analyzed by von Kossa staining of hydroxyapatite and environmental scanning electron microsopy (ESEM).

II. EXPERIMENT A. Sample preparation
Commercial granules of TiO 2 (>99.9%pure, product 204757, lot MKBH6783V, Sigma-Aldrich, Stockholm, Sweden) were incubated in cell culture medium (DMEM, Gibco, Grand Island, NY.Product No. 11995-065) at 37 C in humidified atmosphere for 24 h, rinsed in saline and distilled water, and dry sterilized at 160 C for 2 h.The grain size was determined to 2 lm by light microscopy of a water suspension.Sterilized dental implants were used as received from Elos Medical AB (Timmersdala, Sweden) and were implanted into rats as described in Sec.II G.

B. Human mesenchymal stem cell culture
The human embryonic stem cell (hESC) lines used in this study were SA167MFG-hESC and AS034.1MFG-hESC at passage 12 and 44, respectively, derived and characterized in a previous study. 20Cell culture was performed at the Department of Clinical Chemistry and Transfusion Medicine, Institute of Biomedicine, University of Gothenburg.Note that the stem cells used adhere to plastic surfaces and can be cultured in polystyrene dishes.

D. Exposure of stem cells to TiO 2 preincubated with DMEM
Undifferentiated hESCs were cultured on regular tissue culture plastic without predifferentiation stages such as embryoid body formation.
Cell exposure was performed by adding the HA-coated TiO 2 in concentrations of 5 or 0.5 mg/ml into the culture medium for 24 h.

E. Fixation and Von Kossa staining
Mineral production was studied using von Kossa staining performed by washing the cells in phosphate-buffered saline followed by fixation in glutaraldehyde solution (25% in H 2 O Sigma-Aldrich diluted 1:10) for 2 h.A solution of AgNO 3 (2% w/v: Sigma-Aldrich) was added, and the plates were kept in dark for 10 min.The plates were then rinsed three times with distilled H 2 O before being exposed to bright light for 15 min.After washing with distilled H 2 O, samples were quickly dehydrated adding 100% EtOH prior to microscopic inspection for mineralization as described earlier. 21Cells, fixed with glutaraldehyde and dehydrated in ethanol, were also analyzed by ESEM under low vacuum conditions.
F. Cell viability hESCs were seeded onto a 24 well plate at density of 10 000 cells/well.Cells were incubated in growth medium with or without the presence of metal oxides for 24 h to allow for attachment.The attached cells were considered viable and floating cells nonviable.

G. Animal surgery
All animal work was approved by the Gothenburg animal experiment ethical committee.The surgery was performed as described in a previous study, 22 shortly as follows.Male Sprague Dawley rats 200 g (Charles River, Holland) were used.The animals were anesthetized with Isofluran (Baxter Medical CO, Kista, Sweden).A hole was drilled (d ¼ 1 mm) in the tibial facies lateralis.Sterile dental Ti-implants (Elos Medtech, Timmersdala, Sweden) were implanted intramedullary.The surface chemistry was analyzed by EDX and found to be C 5.6%, O 13.2%, and Ti 94.5%.The animals were kept at the Experimental Biomedicine facility and were fed commercial pellets and water ad libitum.The animals were harvested after 24 or 72 h or 5 weeks by separating the heart from the main arteries, and the tibia bones were dissected.Four animals (eight implants) were used for ToF-SIMS analysis, and four animals were used for the EDX analysis.

H. Environmental scanning electron microscopy and EDX
The rat tibiae were immersed into absolute ethanol on dry ice.The bone tissue was substituted with alcohol for one week at À80 C, warmed to room temperature, and cut with a diamond saw, using absolute ethanol as lubricating liquid and rinsed carefully in absolute ethanol.The samples were left to dry at room temperature.An FEI Quanta 200 FEG ESEM operating at an accelerating voltage of 20 kV was used for imaging and chemical analysis.All images were acquired in the backscattered electron imaging mode at a pressure of 1 Torr in the low vacuum region in order to avoid charging effects.Energy dispersive x-ray (EDX) data was recorded using an Oxford EDX detector and spectra were evaluated with the INCA software.Results were expressed as mass % of all detected elements or atom % for calculation of Ca/P ratio.

I. ToF-SIMS
ToF-SIMS analysis was performed with a TOF.SIMS 5 instrument (ION-TOF GmbH, M€ unster, Germany) using a Bi 3 þ cluster ion gun as the primary ion source.Multiple (n ¼ 5) regions ranging from 100 Â 100 to 500 Â 500 lm were analyzed using a pulsed primary ion beam (Bi 3 þ , 0.34 pA at 25 keV) with a focus of approximately 2 lm and a mass resolution of M/DM ¼ 5000 fwhm at m/z 500.All spectra were acquired and processed with the SURFACE LAB software (version 6.4,ION-TOF GmbH, M€ unster, Germany), and the ion intensities used for calculations were normalized to the total ion dose of each measurement.Depth profile analysis was performed using the pulsed Bi 3 þ gun (0.4 pA) while sputtering was carried out with a 10 keV C 60 þ beam with a current of 0.6 nA.The maximum ion dose density of Bi 3 þ was kept at 2 Â 10 11 ions cm À2 and therefore below that of the static limit, i.e., 1 Â 10 13 ions cm À2 , to make sure that the experiment was ended before the primary ion beam had considerably damaged the surface of the sample.The ion dose for C 60 þ ranged from 4 Â 10 14 to 6 Â 10 14 ions cm À2 .Low energy electrons were used for charge compensation during analysis.Profiles were also obtained using an argon gas cluster ion beam (GCIB) source, using 20 keV Ar 1500 ions at a current of 0.364 nA.The ion dose for Ar 1500 ranged from 4 Â 10 14 to 7 Â 10 14 ions cm À2 .

III. RESULTS AND DISCUSSION
The chemical composition of the samples was analyzed by EDX and ToF-SIMS before (data not shown) and after incubation in DMEM.The EDX data from the analysis of untreated Characterization of the TiO 2 samples, control and preincubated in DMEM by ToF-SIMS, is shown in Fig. 1.The control showed clear signals from Ti þ and TiO þ (range not shown) but no HA related peaks as can be seen in Fig. 1(a).ToF-SIMS was however able to detect several ion species from HA such as Ca 2 PO 4 þ ,Ca 3 PO 5 þ , and Ca 5 PO 7 on the DMEM incubated sample as indicated in the spectrum in Fig. 1(b).
The HA-coated TiO 2 powder was then added to hESC cultures.The cells proliferated during their exposure to the TiO 2 grains.The loss of adhering cells was less than 10% of the cell layer surface.Images of the exposed cell cultures are shown in Fig. 2. Figure 2(a) shows a low magnification ESEM image of the cell layer and polystyrene surfaces without cells.On the polystyrene surfaces, single TiO 2 grains or small aggregates of grains are seen.On the cell layer, the TiO 2 grains have been taken up by the cells, as judged by light microscopy and excreted into extracellular areas between the cells.These aggregates are shown at higher magnification in Fig. 2(b).Figure 2(c) shows an untreated cell culture control.Figure 2(d) shows von Kossa stain image of the cultured hESCs exposed to DMEM-preincubated TiO 2 .The cell layer contains silver precipitate indicating the presence of mineral.As shown by Bonewald et al., 23 the von Kossa staining is not specific for HA, but can also stain "dystrophic mineralization of unknown chemical composition."Since the added TiO 2 grains are coated with low levels of HA, they may be stained by the von Kossa.The aggregated TiO 2 particles shown in Figs.2(a) and 2(b) are also seen by light microscopy [Fig.2(e)].The aggregate pattern of TiO 2 grains in this image may be compared to the von Kossa stained culture [Fig.2(d)].From this experiment, we conclude that the cells, apparently, add mineral to the aggregates during the process of uptake and excretion.
Ti-implants were operated into rat tibia and allowed to heal for 24-72 h.The tissue layer formed during the first 3 days of healing was analyzed with depth profiling ToF-SIMS as shown in Figs.3(a)-3(c).Figures 3(a) and 3(b) show data generated by conventional C60 þ sputtering and Fig. 3(c) data generated by sputter using a gas cluster ion source using Ar 1500 þ .The C60 þ based profiles show a similar behavior with an initial layer consisting of cells and organic matter, represented by the phosphatidylcholine headgroup, C 5 H 15 PNO 4 þ , stemming from phosphatidylcholine and phosphatidylethanolamine molecules, followed by a HA layer, represented by signals from P 2 OH þ and Ca 2 PO 4 þ .The CaO þ signal, likely stemming from CaCO 3 , peaks at approximately 0.8 Â 10 14 ions/cm 2 and is detected immediately before the titanium signal.The signal eventually decays before the Ti þ reaches its maximum.Very little difference can be found between the 48 and 72 h samples with regards to the organic and HA layers while the CaO layer possible appears to be thicker for the 48 h implanted surface.Depth profiling using argon GCIB produces a somewhat different result as seen in Fig. 3(c), where the argon is unable to erode the CaO away fully and CaO reached maximum at 3 Â 10 14 ions/cm 2 but never fully sputters away as for C 60 þ .The Ar 1500 þ beam is clearly not as efficient as C 60 þ in sputtering the inorganic CaO layer.][26] The elemental composition of the cut bone was C 27.However, the Ca levels are higher than the corresponding values of untreated bone. 27The implant related bone thus seems to be identical to the compact cortical bone after 5 weeks of healing and the compact bone is affected by the healing in of the implant, which is in accord with previous findings during healing of MgO into bone. 27The 2D morphology of the implant-related bone shows an archipelague structure of the bone, interdigitated with areas of bone marrow.
9][30] Callus bone appears in the wound after 96 h, apparently originating and extending from the endosteum and periosteum.The presented results show that the titanium implant is coated with a stratified layer containing CaO þ on the titanium surface, HA in an intermediate stratum, and phosphatidylcholine from cell membranes were found in an outer stratum, as seen in Fig. 3.This layering has been confirmed by previous studies, showing crystals of HA beneath the cells. 5,8The inner stratum of Ca on the Ti surface has been described previously after analysis of the bone-implant interface of healed-in implants by atom probe tomography. 19This analytical method can only detect basic elements and does not provide molecular information.The result of the present study, made by analysis with ToF-SIMS, suggests that the inner layer contains CaO, probably representing a Calcium mineral such as CaCO 3 or Ca(HCO 3 ) 2 transformed to CaO by heat-generated sublimation of CO 2 in the ion beam.The formation of a CaO-layer on Ti-implants after 1 week of healing, as analyzed by ToF-SIMS, has been reported previously. 31After this time of healing, the bone FIG. 2. Human embryonic stem cells grown for 24 h with or without the presence of 0.5 mg/ml of TiO 2 grains, with a diameter of 2 lm, preincubated in DMEM.After 24 h, the cells were fixed with 2.5% glutaraldehyde.marrow cavity is filled with callus bone and it is not certain whether bone-mineral or Ca-mineral is detected.Also, the ion gun used in the previous study did not allow for detection of the HA-derived fragments 32 as could be seen here, and that are now routinely observed using cluster ion ToF-SIMS. 33ere, we also see a clear difference in sputter ability between the conventional C 60 þ source and the GCIB argon source.Even though the information about the different layers are similar between the different sources, the argon GCIB quickly erodes the organic material from the implant while the inorganic CaO layer remains virtually unaffected on the Ti surface as can be seen in Fig. 3(c).This is in contrast to C60 þ that slowly erodes also the CaO layer from the Ti surface.This is to be expected since GCIB sources are well known or their high sputter yield for organic materials compared with inorganic materials making them highly suitable for probing inorganic/organic implant surfaces. 26,34he results of several previous studies indicates the existence of an organic stratum at the surface of a titanium implant. 3,5,8This finding can still be a valid one and there is no contradiction between the data presented here and previously published findings.As shown in Fig. 4, the titanium surface is coated with approximately equally sized areas of bone and bone marrow after 5 weeks of healing.It is undisputable that a titanium surface in contact with blood is rapidly covered with plasma proteins and platelets, 17 while the CaO layer is formed at a lower rate, 18 which means that it forms by diffusion of Ca 2þ through the pre-existing organic layer.The results of a previous study indicates that the formation of the HA-layer was slower in blood than in the cell culture medium, whereas the CaCO 3 layer was formed at equal rates in both liquids, which can be explained by inhibitory binding of osteopontin to the HA-layer.Osteopontin is known to regulate the formation of HA, 35 and has been detected by immunocytochemistry at the boneimplant interface. 36Osteopontin is not believed to affect the differentiation of stem cells. 37Thus, it was important to perform the present study of the process of HA-coating of Ti-implants in vivo.The presented data indicate that HA is formed in vivo on top of a layer of CaCO 3.This can be seen in both the C 60 þ as well as Argon 1500 þ depth profiles where HA is a clearly separated layer just below the organic components from cells and membranes, and CaO þ can be seen tightly connected to the Ti implant itself.Hence the implant is first coated by CaCO 3 , then phosphate before the HA layer, which is then covered by cells and organic material.The argon GCIB shows the most accurate result here due to its low inorganic sputter yield hence allowing us to probe the still intact inorganic CaO/Ti interface.The production of HA by stem cells exposed to HAcoated TiO 2 , as can be seen in Fig. 2, can be regarded as a stress reaction and is not necessarily a sign of stem cell differentiation, since differentiation of embryonic stem cells to osteoblasts is expected to take 7 days rather than 24 h. 38owever, stem cells can be a significant source of HA at the bone-implant interface during bone healing, and HA itself seems to promote the differentiation of stem cells. 39

IV. SUMMARY AND CONCLUSIONS
We have demonstrated the advantage of using argon GCIB for analyzing in vivo implants where now the organic/ inorganic interface can be probed much more accurately.The results suggest that the established view of bone healing, that bone progenitor cells react with the TiO 2 surface, must be reevaluated since the formation of hydroxyapatite on the implant surface occurs more rapidly than the induction of bone formation.The actual sequence of events is that recruited stem cells are reacting to their contact with hydroxyapatite at the Ti-surface.This insight is important for a proper understanding of how implants heal into bone.
FIG.2.Human embryonic stem cells grown for 24 h with or without the presence of 0.5 mg/ml of TiO 2 grains, with a diameter of 2 lm, preincubated in DMEM.After 24 h, the cells were fixed with 2.5% glutaraldehyde.(a) ESEM image of the surface of the culture dish showing cell-covered areas with large aggregates of TiO 2 grains and polystyrene areas with single grains or small clusters of TiO 2 grains.Bar ¼ 2 mm.(b) ESEM image of the cell layer of cultured hESCs with large aggregates of TiO 2 grains in extracellular spaces, single grains or small clusters of TiO 2 grains.Bar ¼ 200 lm.(c) ESEM image of the cell layer of hESCs cultured in DMEM for 24 h and fixed in glutaraldehyde.Bar ¼ 50 lm.(d) Von Kossa staining of the cell layer of hESCs cultured in the presence of TiO 2 grains.The Von Kossa-positive staining shows large aggregates of mineral in extracellular spaces.(e) Light microscopy of the cell layer of hESCs cultured in the presence of TiO 2 grains.

FIG. 3 .
FIG. 3. (a) Depth profiling of an in vivo titanium implant after 48 h of healing using C 60 þ .Depth profile analysis was performed using the pulsed Bi 3 þ gun to analyze an area of 156 Â 156 lm while sputtering at 400 Â 400 lm was carried out with a C 60 þ beam with a current of 0.6 nA.Ti is shown in blue, CaO þ in orange, P 2 OH þ in brown, Ca 2 PO 4 þ in green, and C 5 H 15 PNO 4 þ in red.Data are shown as normalized to maximum.Total fluence was 4 Â 10 14 ions/cm 2 for the C 60 þ ions.(b) Depth profiling of an in vivo titanium implant after 72 h of healing using C 60 þ .Depth profile analysis was performed using the pulsed Bi 3 þ gun to analyze an area of 216 Â 216 lm while sputtering at 300 Â 300 lm was carried out with a C 60 þ beam with a current of 0.6 nA.Ti is shown in blue, CaO þ in orange, P 2 OH þ in brown, Ca 2 PO4 þ in green, and C 5 H 15 PNO 4 þ in red.Data are shown as normalized to maximum.Total fluence was 4.0 Â 10 14 ions/cm 2 for the C 60 þ ions.(c) Depth profiling of an in vivo titanium implant after 72 h days of healing using argon GCIB.Depth profile analysis was performed using the pulsed Bi 3þ gun to analyze an area of 70 Â 70 lm while sputtering at 500 Â 500 lm was carried out with a Ar 1500 þ beam with a current of 1.3 nA.Ti þ is shown in blue, CaO þ in orange, P 2 OH þ in brown, Ca 2 PO 4 þ in green, and C 5 H 15 PNO 4 þ in red.Data are shown as normalized to maximum.Total fluence shown 1 Â 10 15 ions/cm 2 for the argon ions.