Controlled biodegradation of magnesium alloy in physiological environment by metal organic framework nanocomposite coatings

Magnesium-based implants (MBIs) have recently attracted great attention in bone regeneration due to elastic modulus similar to bone. Nevertheless, the degradation rate and hydrogen release of MBIs in the body have to be tackled for practical applications. In the present study, we present a metal–organic framework (MOF) nanoplates to reduce the degradation rate of AZ91 magnesium alloy. Zeolitic imidazolate frameworks (ZIF-8) with a specific surface area of 1789 m2 g−1 were prepared by solvothermal methods, and after dispersion in a chitosan solution (10% w/w), the suspension was electrospun on the surface of AZ91 alloy. Studying the degradation rate in simulated body fluid (SBF) by electrochemical analysis including potentiodynamic polarization and electrochemical impedance spectroscopy reveals that the degradation rate of the surface-modified implants decreases by ~ 80% as compared with the unmodified specimens. The reduced alkalization of the physiological environment and hydrogen release due to the implant degradation are shown. In vitro studies by fibroblasts and MG63 osteosarcoma cells exhibit improved cell adhesion and viability. The mechanisms behind the improved degradation resistance and enhanced bioactivity are presented and discussed. Surface modification of MBIs by MOF-chitosan coatings is a promising strategy to control the biodegradation of magnesium implants for bone regeneration.

Synthesis of metal-organic frameworks. ZIF-8 MOF particles were synthesized by solvothermal methods. Zn(NO 3 ) 2 and 2-methylimidazole (MIm) with a Zn 2+ /MIm molar ratio of 1:8 were dissolved in methanol by dropwise adding and gentle stirring. After mixing for 60 min, the solution was transferred in a Teflonlined autoclave and treated for 24 h at 120 °C. The precipitate was separated by centrifugation and washed three times with fresh methanol to remove any unreacted reagents. The resulting product was vacuum dried to attain ZIF-8 MOFs.
Surface modification of magnesium alloy. The AZ91 slab was cut into small cubes with dimensions 1 mm × 1 mm × 1 mm by electrical discharge machining. Before electrospinning, a surface treatment including grinding with a SiC emery paper (mesh #180) followed by chemical etching in 1 M HNO 3 for 30 s was www.nature.com/scientificreports/ performed. As shown elsewhere 33 , this pre-treatment created a rougher surface and improved the adhesion of electrospun fibers to the magnesium surface. For electrospinning, at first, an aqueous solution of chitosan was prepared by dissolving the polysaccharide (4 wt.%) in 4 mL acetic acid solution (50%) with a pH = 5.1. The mixture was stirred for 24 h at 250 rpm, and then filtered to remove any residuals. To facilitate electrospinning, PEO with a mass ratio of 1:4 to chitosan was added to the polymer solution and stirred overnight. For the preparation of composite suspensions containing ZIF-8 MOFs, the nanoparticles (10 wt.%) were added to the polymer solution, stirred for 2 h, and sonicated for 45 min. A single nozzle electrospinning apparatus (ES1000, FNM, Iran) was used to deposit chitosan and MOF/chitosan composite films on the surface of the magnesium specimens. The processing parameters were determined based on trial and error efforts. The applied voltage was 21 kV. A flow rate of 1 mL/h was utilized. The distance of the syringe to the aluminum rotating drum (2500 rpm) was 10 cm. To attain films with the same thickness of 30 µm, the processing parameters were kept the same for both chitosan and composite coatings.

Materials characterizations.
Scanning electron microscopy (SEM, Philips, XL30) was used to study the size and morphology of MOF particles. For the phase analysis, X-ray diffraction (XRD) was carried out by a Philips PW 1730 diffractometer (Netherlands) which uses Cu-kα radiation (0.178897 nm,40 kV, 30 mA) in 25 °C. Brunauer-Emmett-Teller (A Belsorp mimi BET, Japan) was employed for the surface area analysis. The size and morphology of the fibers, as well as the thickness of the electrospun films on the surface of the magnesium alloy, were studied by SEM. The biodegradation products after immersion in SBF were also analyzed by SEM and XRD. Besides, electron dispersive spectroscopy (EDS) was used for elemental analysis. To determine the effect of surface modification on the hydrophilicity, the sessile drop technique was employed according to ASTM D7334 standard. A small droplet of deionized water was placed on the surface of the samples by micropipette and photographed by a camera. The water contact angle was determined by the ImageJ software.
Degradation studies. All biodegradation evaluations were carried out in SBF at 37 °C. Potentiodynamic polarization test was carried out by a potentiostat (302N, Metrohm Autolab, Netherlands). A conventional threeelectrode configuration including a working electrode (AZ91 alloy), a saturated calomel electrode (control), and a platinum counter electrode was used. Potentiodynamic polarization curves were recorded by changing the voltage in the range of − 2.5 to 0 mV with a rate of 20 mV/s 34 . To establish dynamic equilibrium conditions, the specimens were immersed in SBF for 60 min before testing. To determine the corrosion current (i corr ), the Tafel slopes were determined in the vicinity of the corrosion potential (E cor ) 35 and used in the Stern-Geary relation 36 : βa and βc are the anodic and cathodic Tafel slopes, and Rp is the polarisation resistance that was obtained from the electrochemical impedance spectroscopy (EIS) technique. EIS was carried out at a 1 V DC potential (vs. the standard electrode). A sinusoidal potential amplitude (10 mV) in the frequency range of 0.1-105,000 Hz was applied. The acquired EIS data were fitted to an equivalent circuit by employing the ZView software (Scribner Associates Inc., USA).
To determine the rate of hydrogen evolution during biodegradation, cubic specimens with a volume of 1 cm 3 were inserted in a transparent plastic syringe containing 60 mL SBF. The container was incubated at 37 °C for 15 days. The amount of hydrogen volume was then determined by measuring the volume change according to the procedure explained in Ref. 37 . Briefly, the volume of the released hydrogen was estimated from the below reaction: For the evolution of 1 mL H 2 , we have: Statistical analysis was carried out by GraphPad Prism with considering *p < 0.05 as significant. Data are expressed as mean ± standard deviation (SD).
In vitro studies. L929 fibroblast cells were obtained from the National Cell Bank of Iran (Pasteur Institute of Iran, Tehran) All the specimens were sterilized by 70% ethanol for 30 min in a 24 well-plate and dried under a laminar flow biosafety cabinet (KG-A100, Kimia Gene Co., Iran). The specimens with a surface area of 1 × 1 cm 2 were then placed in a 24-multi well plate and 5 × 10 4 cells were seeded on each sample. The control was the well plate without the AZ91 sample. After incubation for selected times, 20 μL of 1 mg/mL MTT (3-[4,5-dimethythiazol-2-yl]-2,5-diphenyl-tetrazolium bromide) solution was added to each well and incubated for 4 h at 37 °C. To dissolve the insoluble purple formazan crystals, the medium was replaced by (DMSO). The absorbance in each well was then recorded at 540 nm through an ELISA reader 38 .
For evaluation of cellular adhesion, human osteoblast-like MG-63 cells (the National Cell Bank of Iran, Pasteur Institute of Iran, Tehran) were used. The cells with a density of 5 × 10 4 were incubated on the specimens in a 5% CO 2 humidified incubator at 37 °C. After 3 days, the non-adherent cells were washed out by PBS rinsing. The attached cells were then fixed by a fixative solution containing glutaraldehyde and stained in an alcoholic solution for 5 min and dried. Changes in the pH of the medium were measured by a pH meter (pH7110, Inolab, Germany).

Results
Characteristics of the metal-organic frameworks. Different analytical techniques were employed to determine the morphology, porosity, and structure of the synthesized ZIF-8 MOFs. Table 2 summarizes the results. Figure 1 shows the characteristics of the ZIF-8 MOFs. The XRD pattern indicates well-defined crystal planes with slightly broadened peaks (Fig. 1a). In agreement with the simulated pattern of the ZIF-8 structure 39 , the particles are crystallized in a hexagonal structure 40 . The SEM image indicates that the particles have thin plate-like morphology with a lateral size of 70 to 80 nm and thickness of about 22 nm, yielding an aspect ratio of 0.3 (Fig. 1b). BET analysis at 77 K determines a type I adsorption isotherm with a small hysteresis loop (Fig. 1c). Analysis of the pore size distribution (Fig. 1d) reveals the nanoporous structure of MOF with the characteristics reported in Table 2. As seen, the processed MOF nanoplates have a very high specific surface area (1789 m 2 g −1 ) and large pore volume (1.28 cm 3 g −1 ). FTIR spectroscopy exhibits characteristic peaks at 993 cm −1 , 1142 cm −1 , and 1421 cm −1 , which are associated with the C-N stretching vibration 41 . Two small peaks at 3130 cm −1 and 2927 cm −1 are ascribed to aromatic C-H stretching vibration and aliphatic C-H stretching vibration of the imidazole ring and the methyl group, respectively 42 . The peak at 1583 cm −1 is attributed to C=N stretching vibration. Other peaks that appeared below 1250 cm −1 correspond to the in-plane and out-of-plane bending of the imidazole ring 43 . Zn-N stretching vibration is detected at 422 cm −1 . The peaks in the range of 1350 to 900 cm −1 could be attributed to in-plane bending of the ring while the peaks at 754 and 688 cm −1 are associated with aromatic sp 2 C-H bending 41 . Table 2. BET data for ZIF-8 MOFs prepared by solvothermal method.

Particle size (nm)
Specific surface area (m 2 /g) Average pore diameter (nm) Total pore volume (cm 3 /g) Biodegradation of magnesium alloy. To study the biodegradation of AZ91 alloy with and without surface modification, the electrochemical responses of the Specimens by triplicate of AZ, AZC, and AZCZ samples in SBF were measured and then the Tafel diagram was drawn for each of the samples and the values obtained were averaged and reported in Table 3. Also, the degradation rate of samples was calculated using the mpy formula. The Tafel plots of the examined specimens are shown in Fig. 3a. The extracted data from the polarization curves are summarized in Table 3. The results indicate that the magnesium alloy exhibits a relatively high corrosion current (i corr = 34.2 µA cm -2 ) in SBF, which is comparable with other studies such as Mena-Morcillo and Velleva 36 , Zheng et al. 45 , and Razavi et al. 46 . Surface modification by the chitosan film significantly reduces the degradation rate and corrosion current to about half (i corr = 16.6 µA cm −2 ). The potential (E corr ) also slightly www.nature.com/scientificreports/ increases. Interestingly, the incorporation of ZIF-8 MOFs in chitosan further decreases the degradation rate and corrosion current by ⁓ 65% without changing E corr . As compared to unmodified magnesium alloy, i corr drops to 6.5 µA cm −2 , which is only about 20% of that of AZ91. Herein, it is important to mention that the electrochemical results on Mg alloys highly vary with the chemical composition, the type of the electrolyte, the surface condition, and the testing condition [47][48][49] . Therefore, it is difficult to compare directly our results with other studies. Employing a dense and bioinert layer on the magnesium surface yields lower degradation rates but this method suffers from good cell attachment and bioactivity. Porous and active layers such as fibrous membranes provide a better platform for bone integration although the degradation rate is higher. The amount of hydrogen evolution measured at different time intervals (up to 15 days) after immersion in SBF is shown in Fig. 3b. The amount of gas evolution increases with increasing the immersion time for all   Table 4). The rate of gas evolution exhibits a declining trend with time, indicating the effect of degradation products on the corrosion rate. It seems that the products of degradation act as a physical barrier to prevent rapid degradation in PE. It is known that magnesium hydroxide (Mg (OH) 2 ) and magnesium oxide (MgO) are commonly formed during degradation and form a passive layer on the metal surface 50 . To study the degradation products, SEM and XRD were employed (Fig. 4). SEM studies show the formation of rod-shaped or flower-like precipitates on the surface. The XRD pattern of unmodified Mg alloy determines that basically magnesium hydroxide and magnesium oxide phases are formed. Small amounts of calcium carbonate and silicon oxide are also detected. The degradation products of the chitosan film are mostly hydroxyapatite. On the MOF-containing film, more complex phases are formed, owing to the presence of Zn ions. Besides, small amounts of Na 3 MgC 2 O 6 and Mg (HCO 3 ) (OH) 2 (H 2 O) are detected. Anyway, the results of hydrogen evolution determine that the surface modification  Electrochemical impedance spectroscopy. To explore the mechanism of the biodegradation retardant effect of chitosan and MOF nanoplates, electrochemical impedance spectroscopy (EIS) in SBF at 37 °C was employed. The resulting Nyquist and Bode plots are shown in Fig. 5. In the Nyquist plots, two capacitive loops in both high and low-frequency regions are observed (Fig. 5a). In concurrence, the phase angle Bode plots exhibit two-time constants (Fig. 5b). After immersing the specimens in the PE solution, the biodegradation products are formed on the surface and two capacitive loops (at high-and low-frequency regions) develop due to changes in the charge transport, surface chemistry, and mass transfer 51 . The high-frequency region of the EIS relayed on the electrolyte penetration process including water uptake and the electrolyte interference. The diameter of the loops is directly proportional to the surface film resistance for mass and charge transport 52,53 . The unmodified magnesium alloy exhibits relatively a small loop that indicates a rapid biodegradation rate. The surface modification by chitosan film increases the diameter of the circle with double humps. The behavior of the specimen can be described based on capacitive behavior originated from the degradation of the magnesium alloy in the medium and inductive behavior resulted from continuous adsorption and desorption chemicals elements or ions and formation of corrosion products 37 . The incorporation of ZIF-8 MOFs further enlarges the circles with little effect on the inductive behavior. Therefore, the chitosan-MOF film has further increased the degradation resistance of the magnesium substrate.
To get a better insight into the mechanism of biodegradation, the equivalent circuit for the Nyquist plot was constructed (Fig. 5c,d). The values obtained by data fitting are reported in Table 5. As so the low-frequency region  Table 5. www.nature.com/scientificreports/ for the bare and chitosan-MOF specimens is too much noisy, an equivalent circuit with a one-time constant was used to fit the EIS plots. However, a circuit with two-time constants was used to fit the chitosan-modified AZ91 sample. R s is representative of the solution resistance which is in series with the unit of the oxide layer system. For the bare electrode, R1 and R 2 in parallel with constant phase elements (CPE 1 and CPE 2 ) are the resistance of naturally formed oxide film on the substrate and charge transfer resistance of the faradaic process on the metal surface, respectively 52,53 . For the chitosan and chitosan-MOF specimens, the resistances (R 1 and R 2 ) and constant phase elements (CPE 1 and CPE 2 ) represent the resistance of the porous outer layer and the inner coating/substrate interface, respectively. This part of the circuits was not quantitatively evaluated for the bare and chitosan-MOF coated AZ91 electrodes as the corresponding error was relatively high. Noted that the value of fitting errors should be less than 1% 54,55 . Besides, CPE which models the behavior of a double layer (as an imperfect capacitor) was considered instead of a capacitive element because of the surface inhomogeneity and possible diffusional factors affecting the shape of the semi-circles 56 . Table 5 indicates that the R 2 value for the chitosan-MOF specimen (1256 Ω) is much higher than that of the chitosan-modified AZ91 (651.9 Ω) and the bare sample (291.3 Ω). This finding affirms corrosion resistance enhancement of AZ91 through surface modification by the chitosan-MOF film via inserting a physical barrier against diffusion/penetration of the corrosive ions from the solution toward the metal surface. For the chitosan-modified AZ91 specimen, the value of R2 (651.9 Ω) is far higher than that of R 1 (10.6 Ω). This observation could indicate the vital role of the magnesium/ coating interface in the biodegradation process.
In vitro biocompatibility and cell adhesion. To assay the in vitro biocompatibility of the specimens, L929 cells were used for MTT assay (Fig. 6a). The unmodified AZ91 alloy exhibits poor biocompatibility with cell viability of less than 50% even after a short incubation time (24 h). A longer incubation time further decreases the cell viability, indicating the cytotoxic effect of Mg degradation, as reported in the literature 57 . The magnesium alloy modified with chitosan film exhibits higher cell viability (> 90%) at the early stage but reduced to about 70% at prolonger times. Moderate biocompatibility is attained for the chitosan-MOF film without a major effect of the incubation time (within the examined period). Nevertheless, the cell viability is at least double of the unmodified Mg alloy. The attachment of MG63 cells on the surface of the specimens incubated for 72 h is shown in Fig. 6c-e. No live cells are observable on the surface of unmodified Mg alloy due to the rapid degradation and formation of corrosion products. In contrast, the cells have adhered to the fibrous chitosan film but the formation of filopodia and spreading is not noticeable. The chitosan-MOF film results in the formation of relatively better cell adhesion and spreading.

Discussion
In this study, ZIF-8 nanoplates were employed and incorporated in the chitosan nanofiber to modify the biodegradation rate of AZ91 Mg alloy in PE. It was shown that the MOF particles decorated the surface of the nanofibers and slightly decreased their diameters. The reduced sizes can be attributed to the increase in the size of the Taylor cone and decrease in the jet velocity due to a change in the viscosity of the solution 58,59 . Since ZIF MOFs are hydrophilic 39 , the particles tend to be pushed toward the outer layer and collected on the surface. Meanwhile, no remarkable agglomeration was noticed.
The results indicated that the chitosan coating containing MOFs exhibited better protection against biodegradation in PE. The advantage of using ZIF-8 is ascribed to its anti-corrosion properties 60,61 that can reduce the electrochemical potential and current density of AZ91 alloy 24,62 as well as its high stability in physiological environment and drug loading capacity [24][25][26][27][28] . Although our coating materials seem to be less protective as compared to bioinert and dense protective layers 23,63-65 , employing fibrous membranes provide a better platform for cell adhesion and bone integration.
Electrochemical studies revealed that the coatings were mainly operated as a physical barrier. EIS studies determined that the impedance of the surface-modified specimens was higher than that of the bare metal. It seemed that the coating layer reduced the ion diffusion and thus decreased the corrosion rate 37 . The Bode plot determined low-frequency and high-frequency resistor zones and a capacitive behavior in the mid-frequency range due to the maximum phase changes (θ max ) vs. log(f) 66 . Based on the shape and values of θ at high frequencies, it appeared that the electrochemical response of the surface-modified samples had a higher tendency toward capacitive behavior. At low frequencies, due to absorption and reduction of corrosion products such as Mg (OH) and Mg (OH) 2 at the interface between the electrolyte solution and the metal surface, an induction loop appears that is equivalent in shape to the inductor and offers resistance 67 . The Nyquist plots and the capacitive loops also indicated the existence of a distinct layer on the surface of the specimens. The diameter of the capacitive ring indicated the higher polarization resistance of the chitosan-MOF composite specimen 68 . Here, the role of the precipitates on the corrosion rate should also be considered, It was shown that corrosion of magnesium in PE was accompanied by the formation of precipitates on the surface (Fig. 4). Degradation and dissolution of magnesium increased the pH of the medium that promoted phosphate formation due to saturating of SBF at highly alkaline conditions 50 . The rise of the pH is attributed to the biodegradation of the metal surface and the release of hydroxide ions in the cell culture medium 33,50 as well as the formation of hydrogen bubbles 10 . The surface modification reduced the amount of pH raise as the coating layer acted as a barrier against direct contact of the medium and the metal surface. Not only the fast release of Mg in the culture medium affects the cell viability due to the alkalization, but also the degradation of MOFs may influence the cell response.
The cell viability assay indicated improved biocompatibility of chitosan-modified specimens over 72 h. Measurements of the pH value of the culture medium (Fig. 6b) indicated that for the bare metal, the pH exceeded beyond the physiological level and gained a high value (8.6) that cells can tolerate. Hook et al. 24 showed that at concentrations beyond 30 µg mg −1 , ZIF-8 exhibited cytotoxicity due to the release of Zn2 + in mitochondrial ROS products. This negative effect suppresses the cell cycle in the G2/M phase, which is irreversible and finally caused DNA damage and ultimately activates cellular apoptosis pathways 34 . We used a low amount of ZIF-8 in our mats; therefore, the effect of Zn ions should be marginal. It was also shown that cell adhesion on the surface of the fibrous coatings was improved. Cells usually interact better with surfaces that have more pores and roughness 69 . Hydrophilicity and mechanical durability of the surface also affect the interaction of cells with the material at the interface 30,57 . For instance, it has been shown that MG63 osteoblast-like cells significantly proliferate at rough surfaces such as acid-etched or plasma 69,70 . Hydrophilic surfaces have much more integration and adhesion with osteoblastic cells with implants compared to hydrophobic or less hydrophilic surfaces 57 .

Conclusions
To control the biodegradation rate of AZ91 magnesium alloy in physiological environments, composite films of ZIF-8 MOF/chitosan were deposited on the surface by electrospinning. Plate-like MOF particles with lateral sizes of 70-80 nm with an aspect ratio of 0.3 and a specific surface area of 1789 m 2 g −1 were prepared by solvothermal methods. Continuous, uniform, and bead-free fibers with an average diameter of 30 µm were deposited on the surface of the magnesium alloy. The incorporation of ZIF-8 particles in the chitosan coating did not significantly change the size and uniformity of the fibers, but the particles tended to decorate the surface of the polymer. In vitro evaluations of biodegradation in PE by electrochemical methods indicated that the corrosion current of AZ91 alloy decreased by ⁓ 50% after surface modification by the fibrous chitosan coating. The incorporation of MOF particles further reduced the corrosion current by ⁓ 65%. The rate of hydrogen evolution was also reduced significantly. EIS determined that the polymer and composite films operated as a physical barrier and increased the charge transfer resistance while changing the capacitive behavior of the biodegradation process. The degradation products were changed from magnesium compounds for the bare metal to hydroxyapatite precipitates with more complex Ca-Zn-Si compounds. In vitro cytotoxicity examinations also revealed that surface modifications improved the biocompatibility due to the reduced degradation rate of Mg in the physiological medium. In contrast to the bare metal, cell adhesion with filopodia formation and cell spreading were also improved. Therefore, the MOF-chitosan composite coating showed promising potential for the surface modification of magnesium alloys to limit their biodegradation in vitro. In vivo examinations, however, are required to evaluate the biological response of the surface-modified BMIs to reveal their potential applications for bone regeneration.