Mg alloy surface immobilised with caerin peptides acquires enhanced antibacterial ability and improved corrosion resistance

1. Genecology Research Centre, University of the Sunshine Coast, Maroochydore DC, QLD 4558, Australia. 2. The First Affiliated Hospital/School of Clinical Medicine of Guangdong Pharmaceutical University, Guangzhou 510080, China 3. Institute of Industrial Science, Department of Mechanical and Biofunctional Systems, the University of Tokyo, 4-6-1, Komaba, Meguro, Tokyo 153-8505, Japan. 4. Centre for Microscopy and Microanalysis, University of Queensland, St. Lucia QLD 4072, Australia 5. Cancer Research Institute, First People's Hospital of Foshan, Foshan, Guangdong 528000, China 6. School of Mechanical, Materials, Mechatronic and Biomedical Engineering, Faculty of Engineering and Information Sciences, University of Wollongong, Northfields Ave, Wollongong NSW 2522, Australia.


Introduction
Fractures cause a considerable public health burden worldwide and are usually caused by accidents and/or related to osteoporosis.It is estimated that 1 in 2 women over 50 years of age will suffer one fracture, as will 1 in 5 men [1].In United Kingdom, the overall annual fracture incidence is 3.6%, and age-standardised lifetime fracture prevalence is estimated to be 38.2%[2].After trauma or surgical intervention, often it is essential to fix bone fragments together to immobilise the fragments and permit healing.Traditional bone fixation materials, such as titanium alloy and stainless steel, remain widely used.However, they are significantly different from human bone tissues, with differences in elastic modules and mechanical strengths [3,4].These materials also result in stress shielding and local pH increases due to the release of metal ions or hydrogen [5], which may lead to inflammation and subsequent bone loss and failure of the metal implants [6].Besides personal suffering, this produces substantial socio-economic losses.Therefore, fixation materials with better biological compatibility are urgently needed.
Magnesium (Mg) is the lightest metal (ranging from 1.74 to 2.0 g/cm 3 ), weighing 33% and 77% less than aluminium and steel, respectively [7].Mg shows similarities to human bones with respect to strength and Young's moduli [8].These characteristics greatly reduce the risk of stress shielding effects in the human body, making it an ideal candidate for human bone fracture internal fixation.In addition, Mg is a necessary component of the human metabolism, making it a more biocompatible than other fixation materials.Roughly 60% of Mg is stored in the bone as skeletal Mg, one third of which resides on cortical bone either on the surface of hydroxyapatite or in the hydration shell around the crystal [9,10].Most bone Mg is likely deposited as an integral component of the apatite crystal and be reabsorbed by bones when released.Intracellularly, Mg is vital to numerous physiological functions, including providing energy in the cells as fundamental component of ATP, involvement in the synthesis of lipid, protein, and nucleic acid as a cofactor of enzymes, and stabilising cell membranes.
It also antagonises calcium [11] and functions as a signal transmitter [12].
Another complication for internal implanting is periprosthetic infections (PPI).An implanted material surface may accumulate bacteria, leading to bacterial colonisation and the formation of bacteria biofilm [14, 33, and 34].To manage PPI, treatments including antimicrobial treatment, implant removal and surgical revision may be conducted, however each of these increase the patient's physical stress and economic costs [35,36].Many studies have focused on the development of novel medical biomaterials with antibacterial activities, including metal implants with bacterial clearance properties.
Antimicrobial peptides (AMPs) are generally short and positively charged peptides, usually between 12 and 50 amino acids, and commonly referred to as host defence peptides (HDPs).AMPs have potent antibacterial, antiviral, and antifungal activity and are ubiquitous among multicellular eukaryotes, with most plant and animal species expressing dozens of distinct AMP genes in epithelial tissues and in response to infection [37].Therefore, AMPs become a potential candidate for the PPI treatment of internal implant.With a controlled release of the AMP, a three-layer coating of a broadspectrum AMP (HHC-36, KRWWKWWRR-OH) on TiO2 nanotube was confirmed to be highly effective against both Staphylococcus aureus and Pseudomonas aeruginosa bacteria, without observable cytotoxicity to osteoblast-like cells (MG-63) [38].A gentamicin-loaded mesoporous bioactive glass was developed for a controlled antibiotic delivery system to tackle orthopedic PPIs, which prevented biofilm formation by S. aureus and S. epidermidis, with better biocompatibility with human bone marrow stromal cells (HBMSCs) than non-mesoporous biological glass (NBG) [39].
Zn/Ag co-implanted titanium by plasma immersion ion implantation also showed excellent osteogenic activity and antibacterial ability and has potential applications in orthopedic and dental implants [40].
However, few studies have focused on improving the antibacterial effect of Mg and its alloys by employing AMPs.
As typical HDPs, peptides extracted from amphibians displayed the ability to kill bacteria by targeting the bacterial membrane [41,42].More than 200 HDPs have been identified from epidermal secretions of Australian frogs and toads [43].Research has shown that Australian frog-derived caerin peptides were potent membrane-active peptides, which can inhibit the formation of nitric oxide [44].

Mg alloy preparation
Two types of Mg alloy materials were used: i) cold extruded Mg AZ31 and ii) fully annealed Mg AZ31 under a no interference atmosphere (vacuum or inert gas).The two materials were manufactured into a thin plate with a dimeter of 4.4 mm and thickness of 2 mm (Fig. 1A).The samples were grinded by 400 grit silicon carbide paper for 1-3 min to remove the original oxide layer and polished by 800-2400 grit silicon carbide paper for 2-5 min to improve the sample surface qualities and obtain homogeneous roughness.All the samples were rotated 90° to ensure that the next step of grinding and polishing removed the scratches generated prior.Before experimentation, all samples were ultrasonically cleaned in 70% ethanol for 5 mins to remove any surface residues.All the samples were ultrasonically rinsed in distilled water for 3 mins.Cleaned alloys were placed under UV radiation for 30 mins on each side for sterilisation [54].Sterilised AZ31 samples were used as non-immobilised controls.To stabilise the surface prior to coating, a group of AZ31 samples were subjected to a fully annealed heat treatment to eliminate the original textures and microstructures.They were heated to 330-350 ℃ in an inert gas atmosphere (Argon), kept for 3-5 hrs, and cooled in furnace.

Chemical click reaction and peptide immobilisation
Since peptides cannot covalently bond with metal directly, polyurethane was layered over the surface using a chemical click reaction.Briefly, polyurethane was thoroughly dissolved in chloroform solution at a ratio of 3% (W/V).The two ultrasonic cleaned Mg-based alloys AZ31 samples were immersed into polyurethane/chloroform solution for 60 min with intermittently stirring.The two Mg-alloy materials uniformly immobilised with polyurethane were taken out of the solution and dried in a watch glass for 16 hrs to form a uniform, tight and stable polyurethane coating layer on the metal surface.
The immobilised metal samples were treated in a plasma reactor (MiniFlecto®, Plasma Technology GmbH, Herrenberg, Germany) with oxygen plasma at 2.45 GHz for 1 min and placed in the atmospheric environment for 15 min to promote the formation of surface peroxy groups and hydroxyl groups.The metal was transferred back to the plasma reactor, the vacuum was adjusted to 26.7 and slowly passed in the acrylic vapor to 66.7 Pa.The metal was taken out after one min of reaction, washed with an ultrasonic cleaner for 10 min, transferred to a mixed aqueous solution (pH 5.0) of 1.25 mg/ml N-hydroxysuccinimide and 5 mg/ml 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide and shaken and stirred at 4°C for 20 hrs.The process of chemical click-reaction is shown in Fig. 1B.
The synthesised F1 and F3 were dissolved in a 0.1 M sodium phosphate solution to prepare a 5 mM homogeneous solution.The surface-activated polyurethane-coated Mg alloy sample was immersed in peptide solution at 4 °C for 20 hrs and ultrasonically washed twice in ultrapure water for 10 min.The preparation was completed by drying F1/F3 immobilised Mg alloys under a fume hood, which were subjected to the following bioassay (Fig. 1C).

Coverage of peptides
Normally, the amount of immobilised peptide can be calculated by the equilibrium adsorption capacity [55].In this study, peptides might be adsorbed on the PU surface via weak interaction, e.g., Van der Waals attraction.However, we expected the amount of peptides adsorbed on the PU surface would be relatively low, since the samples were ultrasonically washed twice in ultrapure water for 10 min post peptide immobilisation (see Section 2.4), which should significantly remove those peptides physically adsorbed on the PU surface.Thus, we did not measure the physical adsorption capacity, but used the coverage of peptides to evaluate the amount of the immobilised peptides.
Post the peptide immobilisation step, we measured the concentration of peptides left in the solution for immobilisation and MilliQ water used to ultrasonically wash the samples.We considered the difference in total peptide corresponding to the amount of peptide linked to the coating layer.The concentration of peptides in the MilliQ water used for the second wash cannot be detected using the current BCA assay (Pierce™ BCA Protein Assay Kit, Catalogue number: 23227, Thermoscientific), thus only the peptides in the water for the first wash was included.Thus, we have added the following equation to estimate the coverage of different peptides.The calculation showed that F3 had higher coverage values for 1E and AZ31, while the coverage on AP was similar with respect to F1 or F3.
where, mc is the mass of peptide covalently bonded to the coating layer (mg), mt is the mass of total peptide used in peptide immobilisation step (mg), ml and mw represent the masses of peptide left in the tubes for peptide immobilisation and ultrasonic wash (mg), respectively.
where, C is the coverage of the peptide (nmol/mm 2 ), MM is the molar mass of the peptide used in immobilization (molar mass of peptide F1 is 3030.68 g/mol and molar of peptide F3 is 2593.19g/mol) and S is the surface area of the Mg alloy specimen (mm 2 ).

Bonding force test
Bonding force between the PU coating and the substrate mainly depends on the surface morphology (surface roughness and surface asperity) and bonding area.It can be tested qualitatively by following methods: X-cut tape test, Cross-cut test and Pull off test for adhesion.In this study, the Cross-cut test was employed with certain scale according to the international standard [56,57] and related literature [58].All the sample's surfaces before and after the test were mapped by Leica DFC 550 Fluorescence/Light Microscope.

In vitro corrosion test
Corrosion samples (the original two Mg alloys AZ31 in different treatment states and two Mg alloys immobilised with peptides F1 and F3 respectively) are uniformly made into small-sized Mg pieces with a diameter of 4.4 mm and a thickness of 2 mm.The dimensional tolerance is 0.002 mm.Surface cleaning of the original two Mg alloys AZ31 with two different treatment states and two Mg alloy samples that respectively form F1 and F3 polypeptide coatings through cleaning with pure water and in an ultrasonic cleaner for 3-5 min to remove surface residues and adhesives.All the peptideimmobilised AZ31 samples were immersed in DMEM and kept in a 37°C thermostat incubator for 120 hrs.DMEM was selected to simulate the physiological condition.During the in vitro corrosion process the two parameters were used to assess the corrosion resistance of the peptide-immobilised AZ31 samples, including the weight variation of the sample and pH value of the DMEM medium.

pH value
During the in vitro corrosion test, the pH value of corrosion solution (DMEM culture media) was tested at 0, 0.5, 1, 2, 4, 8, 16, 32 etc. hrs by the Cyberscan pH510 Bench Meter.Prior to pH value testing the meter was calibrated and after the test the detector was cleaned by pure water.The corrosion of Mg alloys usually take place as follows: Mg alloy reacts with water to produce like ionic magnesium Mg and the hydroxide in the solution increases, causing the pH value to rise as shown in Eqns.(3) and (4).

Weight change
At designated time points along the incubation, the samples were taken out from their tubes, washed with deionised water and dried in a biosafety fume hood at room temperature.Then, they were weighed by an electronic balance (Shimadzu electronic balance AUW220D) and returned to the respective tubes.
Variations in sample weight were recorded and analysed accordingly.

Surface characterisation
After the 120-hr in vitro corrosion test, all samples were washed in phosphate buffer saline (PBS buffer solution) for 3-5 min and dried in a biosafety fume hood at the room temperature.Due to poor conductivity of the peptide-immobilised two group samples, all samples were carbon-sprayed by Q150T turbomolecular pumped coater before observation.After carbon spray, all samples including peptide-immobilised and no peptide-immobilised samples, were fixed to the sample stages with double-layer conductive tapes.FEI SCIOS SEM/FIB Dual Beam system was employed to scan all peptide-immobilised samples and no peptide-immobilised samples to assess morphology, distribution and microstructure.
The elemental composition of caerin peptide-coating was analysed using energy dispersive X-ray spectroscopy (EDS) coupled with scanning electron microscope JEOL 6010 SEM.Carbon is the most important component of peptides and Mg oxide is the main product during in vitro corrosion test.By analysing the carbon component and Mg oxide we can understand the peptide evolution on the samples' surface and the quantity of Mg corroded during the test.Therefore, the EDS analysis primarily focussed on the variation of these two components.

Bacteriostatic test
After

Coverage of peptides
This study was to develop an approach to chemically immobilise caerin peptides onto PU coating magnesium alloy.Thus, as shown in Fig. 1B, there should be a single layer of peptides immobilised on the surface via covalent bonding, which could enable multiple contacts with bacteria, potentially contributing to the prolonged antibacterial effect.Thickness of the polyurethane on the magnesium alloy surface was measured.Average thickness of polyurethane coating is about 135 µm.
The results of coverage of peptides have been added with the data recorded in Table S1.For peptide F3, extruded AZ31 has the highest coverage, which is about 5.65 nmol/mm2, followed by annealed AZ31 with a peptide coverage of 5.57nmol/mm2.Pure Mg has the lowest coverage for peptide F3, which is 5.49 nmol/mm2.For peptide F1, the coverage order is as: pure Mg (5.49 nmol/mm2), annealed AZ31 (5.47 nmol/mm2) and extruded AZ31 (5.14 nmol/mm2).Overall, annealed AZ31 has the better coverage for caerin peptides F1 and F3.

Bonding force test
Bonding force test results are shown in Fig. 2. Original surface of Mg specimen is shown in Fig. 2A.
Fig. 2B shows the surface of PU coated Mg alloy.As shown in Fig. 2 C, after the test, 5 % -10 % small flakes of the coating were detached at intersections, which fell in a reasonably rate of bonding force [57].

In vitro corrosion test
The microstructure of two types of Mg alloys used in coating were compared (Figs.3A and 3B).
Annealed AZ31 has equiaxed grain with average grain size of 15.85 µm, there are few annealed twins distributed inside grain or trans-grains (Fig. 3A).While cold extruded AZ31 has typcial cold deformed microstructure with average grain size of 9.57 µm and many cold-extruded twining texture which distributed along the related deformation axis, and inside or trans-grains (Fig. 3B).The methods used in the current study, including the changes of weight and pH value before and after immersion test, are direct and effective approaches to check the outcomes of corrosion reaction.Indeed, electrochemical techniques is also an important approach to study anticorrosion performance, which will be an ideal approach in the future study for corrosion resistance of coating of PU immbolised caerin petides on Mg alloys.

pH value
As shown in Figs.3C and 3D, in the 120 hrs in vitro corrosion tests the pH of the pure metal AZ31 soared to 8.91.The pH value was the highest among all samples during the entire corrosion process, indicating that the corrosion rate was the fastest.The pH value of the annealed AZ31 sample immobilised with F1 polypeptide coating was slightly higher than that of the AZ31 immobilised with F3 polypeptide coating within the first 48 hrs.At 48 hrs, due to slight bacterial contamination, the pH value of the F1 immobilised sample was lower than the F3 immobilised sample, however this does not mean that the corrosion rate is slower in the later period.It was clear that the annealed AZ31 sample immobilised with F3 coating had the slowest corrosion rate and the annealed AZ31 sample immobilised with F1 polypeptide coating exhibited a corrosion rate that is considerably lower than that of bare metal annealed AZ31 sample.Both peptide coatings significantly inhibit the release of metal ions into the DMEM solution, thereby inhibiting the progress of corrosion and significantly improving the corrosion resistance of the material.

Weight loss
Weight loss is another indicator of corrosion.The quality analysis of the samples during the corrosion process showed that within the first 12 hrs all three samples had degraded significantly and the quality of the non-immobilised annealed AZ31 samples decreased significantly after 12 hrs (Fig. 4).The

Surface morphology and EDS analysis
After 120 hrs in vitro corrosion tests large cracks appeared on the surface of the annealed AZ31 sample (Fig. 5A), but the annealed AZ31 immobilised with the two peptides did not show any similar degradation (Figs.5B and C).There were no significant differences of their surface morphologies before and after the in vitro corrosion tests, suggesting that the corrosion process of the two peptide-immobilised materials was relatively slow and less severe.It appears that the coating prevents the contact between the Mg and the corrosion media and therefore effectively hinders the corrosion reactions of Mg samples.polypeptide coating.Annealing AZ31 produces the largest amount of MgO (Fig. 5D).These results indicate that bare annealing AZ31 corrodes fastest, F1 immobilised-annealed AZ31 followed by F3 peptide immobilised-annealed AZ31 samples.At the same time, the carbon content is significantly higher for peptide F1-immobilised AZ31 than F3, indicating that the F1 peptide protects against degradation more than F3.
FTIR spectra is also an effective approach to track the evolution of components and the linkage of the peptide to the coating layer.Though not employed in current research, it will be employed in future study to further demonstrate the coating of the peptide on Mg alloys.

Bacteriostatic test
The results of bacteriostatic tests were shown in Fig. 6.The diameter of blank disks is 7.47 mm, while the diameter of original AZ31 samples is 4.4 mm.For the purpose of analysis the results were divided into two groups for analyses according to the different sample diameters.The bacteriostatic ring was introduced to analyse the difference of the different peptide-immobilised Mg samples and positive control (Tazocin).The effective area of the bacteriostatic ring can be calculated by the following equation.

Ae=At-As
where, Ae is the effective area of the bacteriostatic ring and At and As are the total area and sample original area, respectively.
At and As can be obtain as At=prt 2 (7) where, rs is the original diameter of the sample and rt is the total diameter including the effective area of bacteriostatic ring and the sample original area.clear antibacterial effect within 24 hrs, similar to Tazocin, and the antibacterial effect disappears after 24 hrs; (v) the antibacterial effect of polypeptide F3 is significantly better than F1 polypeptide (Fig. S2).Both AZ31 samples immobilised with F1 peptide displayed an antibacterial effect within 24 hrs, however the antibacterial effect disappeared after 24 hrs.In contrast, F3 immobilised-annealed AZ31 samples presented excellent antibacterial behaviours within 100 hrs (Fig. S2).Antibacterial effect is significantly better than those of cold-drawn AZ31 samples immobilised with F1 peptide coating.Two Mg alloy AZ31 samples immobilised with F3 peptide showed a visible and lasting antibacterial effect within 100 hrs.The bacteriostatic effect showed a decreasing trend.At the same time, the antibacterial effect of the annealed AZ31 immobilised with F3 peptide is significantly better than that of the cold drawn AZ31 immobilised with F3 polypeptide coating.Since the diameter of the Mg alloy immobilised with peptide coating is very different from the diameter of Tazocin-immobilised drug-sensitive tablets there is no way to directly compare the antibacterial effect of Tazocin, hence the effective antibacterial ring area was compared.As shown in Figs. 4 and S2, the annealed AZ31 sample immobilised with F3 polypeptide coating has a significantly higher bacteriostatic effect at 24 hrs than Tazocin, and the effect is also obvious after 100 hrs.In addition, the cold-drawn AZ31 metal sheet immobilised with F3 showed a lasting and significant bacteriostatic effect within 100 hrs.

Advantages of the immobilisation method
Fast degradation of Mg in vivo significantly limits its application in medical industry.Innovations to improve the corrosion resistance of Mg has been a consistent area of interest.Many coating materials have been employed for Mg-based materials to improve their corrosion resistance.However, very few coating methods achieved ideal outcome with excellent bioactive properties adhere tightly to the Mg surface, and are environmentally friendly and biocompatible within human body [59][60][61][62].
Compared with other AMPs, caerin peptides have broad spectrum of antibacterial properties which make caerin peptides potential candidates as coating molecules.However, it is very difficult to directly apply the peptides to form a stable layer outside the Mg-based alloy.Although Mg might chelate with certain side chains (e.g., His and Lys) of caerin peptides to generate a loose layer containing peptides, this layer is likely to be deformed in physiological environment and may also lead to damage on metal surface [63].In this study we developed a coating method using click chemistry reaction to keep metal surface intact.
Click chemistry encompasses a group of powerful, simple linking reactions.They are highly selective and specific, produce high yields, require minimal purification and are versatile in joining diverse structures without requiring protection steps [64].Click reactions are bio-orthogonal; neither the reactants nor their products interact with functionalized biomolecules.The reactions are usually under mild or nontoxic conditions, such as at room temperature and in water [65].As one of major classifications of click reactions, carbon-carbon multiple bonds include epoxidations, aziridinations, dihydroxylations, sulfenyl halide additions, nitrosyl halide additions, and certain Michael additions [66,67].
Research demonstrates that most coating methods applied for Mg surface modification involve the immersion of Mg substrates in an aqueous environment [68][69][70][71].Immersion is often fettered by the interference of corrosion effects on the coating formation, resulting in low coating affinity/low bonding strength on the Mg surface.To avoid the effect of concomitant corrosion reaction polyurethane was chosen as a coating agent due to its lack of reactivity with Mg.This formed a stable layer that protected Mg alloy from corrosion when conducting later steps in chemical click reaction.
Surface modification of polyurethane was conducted by chemical click-response to connect with caerin peptides, which activate polyurethane through the formation of surface peroxy and hydroxyl (Fig. 1B).The bio-orthogonality of click reaction caused functional groups to form on the surface of polyurethane, which will assure that peptide will be chemically and firmly adhered on the polyurethane and form a complete coat of caerin peptide covering AZ31 sample' surfaces.The mixed aqueous NHS solution (pH 5.0) of 1.25 mg/ml N-hydroxysuccinimide and 5 mg/ml 1-ethyl-3-(3dimethylaminopropyl) carbodiimide (EDC) did not show any bacterial resistance (Fig. S1A).This reduced interference from the environmental medium and assured that the analysis of antibacterial effect for caerin peptide would proceed much more easily and accurately.
The reaction between EDC/NHS and peptides has been reported and widely used, especially in peptide synthesis and cross-linker chemistry.Three of the pioneer studies [72, 73 and 74] can support the formation of covalent bonds.As mentioned in Section 2.4, the samples were ultrasonically washed twice in ultrapure water for 10 mins post peptide immobilisation, which would remove peptides largely if there were no covalent bonds involved.In addition, the comparison of antibacterial effects between samples with PU coating only (Fig. S2), post EDC/NHS reaction (Fig. S1A) and post peptide immobilisation (Fig. 6D) also suggested that there was peptides covalently bonding to EDC/NHS.
In addition, the procedure of coating is under optimisation to improve the bonding force by, for example, increasing contact areas between the PU coating and Mg surface, and testing PU constructed with monomers of varied polarity and functional groups.We will report the optimization in the coming study.

Effect of microstructures and membrane mechanisms on antibacterial behaviour and corrosion resistance
Caerin peptides were first isolated from Australian frogs [75][76][77].Research demonstrated that caerin peptides show significant antibacterial effect on a wide spectrum of gram positive and gram-negative bacterial strands [53].Though many factors can contribute to the corrosion resistance of specimen, corrosion resistance of this study mainly focused on the PU immobilized peptide coating.
Thickness of PU, F1 and F3 are important factors which obviously affect corrosion resistance of the coating, will be further investigated with the optimized the preparation procedure of specimen.In this study, we just investigated the possibility of employing these peptides as a coating-materials for Mg alloys.
Two properties of these peptides were investigated.Firstly, their capacity to increase the corrosion resistance of Mg alloy was assessed.Previous studies showed that certain types of PU coating can provide protection on Mg alloys from corrosion [78], yet did not present any effect of bacterial resistance.The acidic metabolites secreted/excreted by bacteria (depending on the strains) can cause further corrosion once these bacteria adhere to the surface [79].Thus, a higher level and relatively long term of antibacterial activity possessed by the surface could also improve corrosion resistance.This was suggested in Fig. 6 that large cracks appeared on the surface of the annealed AZ31 sample without peptide immobilization, but not for the annealed AZ31 immobilised with F1 or F3; in addition, F1 and F3 immobilisation showed significantly different weight percentages of C and MgO on surface, which indicated the peptides affected corrosion resistance.Secondly, the longevity of their anti-bacterial activity, which would prevent bacterial adherence to the Mg alloy and prevent bacterial infection from unsterilised wounds, was observed.When immobilised on metal AZ31 with same condition, peptide F3 and F1 exhibited different increases in corrosion resistance and antibacterial effect, with F3 inhibiting the growth of MRSA more comprehensively than F1(Fig.S1A).When immersed in DMEM medium, F3 immobilised AZ31 displayed superior corrosion resistance than the peptide F1 immobilised alloy.Between 12 hrs and 48 hrs of exposure, annealed AZ31 immobilised with peptide F3 showed a lower variation of PH value than annealed AZ31 immobilised by peptide F1 (Fig. S1B).The change of pH value is related to variation in the concentrations of H + and Mg 2+ in the medium, both of which are indicative of Mg corrosion.Minimal variation in the pH value indicates deceleration in the rate of corrosion.
Mass loss is a direct measure of the corrosion velocity.From 24 hrs to 100 hrs the mass of nonimmobilised annealed AZ31 decreased much quicker than the annealed AZ31 with F1 and F3 peptide coatings.The mass of annealed AZ31 with F3 peptide coating was almost unchanged while annealed AZ31 with F1 peptide coating showed a slightly more mass loss (Fig. 4C).The corrosion speed of AZ31 with F3 coating is less than that of AZ31 with F1 coating (Fig. 4C).EDS analyses were conducted to investigate the metal surface variation after corrosion tests.As two typical components of Mg corrosion test, C and Mg oxide were analysed (Fig. S1C).They were chosen because carbon is the major element of peptides and MgO is a corrosion product of Mg. Results showed that AZ31 immobilised by peptide F3 has the less C and MgO than those of AZ31 immobilised with F1.This indicates that in comparison to peptide F1 immobilised AZ31, peptide F3-immobilised AZ31 superior corrosion resistance during this in vitro corrosion tests.
The in vitro antibacterial test showed that non-immobilised annealed AZ31 did not show any bacterial resistance, while Mg alloy immobilised with F1 and F3 both showed antibacterial activity against MRSA (Fig. 6C).Peptide F1 immobilised annealed AZ31 exhibited an obvious antibacterial effect only within 24 hrs while annealed AZ31 with F3 coating showed considerably longer lasting bacterial resistance (up to 100 hrs).Moreover, peptide F3 immobilised AZ31 alloys with annealed or cold extruded treatment showed obviously superior bacterial resistance on MRSA compared to that of peptide F3 alone (Fig. S1D).Furthermore, annealed AZ31 with F3 coating showed a more pronounced and prolonged antibacterial effect compared to antibiotic Tazocin (Fig. S1D).Tazocin only showed bacterial resistance with 24 hrs in the current experimental setting.Though its bacterial resistance is lower than Tazocin, F1 immobilised extruded AZ31 also showed a 100 hrs antibacterial effect.All F3 immobilised samples displayed antibacterial effects over 100 hrs, while F1 immobilised AZ31 did not show these effects longer than 24 hrs.
Caerin peptides inhibit bacteria growth by membrane targeting mechanisms (MTMs) [80].A common structural feature of AMPs is the clustering of hydrophobic and cationic residues on opposite faces of the peptide a-helix, which renders these molecules amphipathic [81][82][83].The net positively charged regions bind to the negatively charged head bacterial membranes [84].Mechanisms of action have been suggested as transmembrane pore (pore) and non-pore models [85], including the Barrelstave model, Toroidal pore model, Carpet model and Aggregate model [86,87] as shown in Fig. 7. Fig. 7. Mechanisms of action for antibacterial HDPs and the pore forming mechanisms [80].
Peptide length is an important factor which has a significant effect on transmembrane mechanisms.If peptides are greater than 20 amino acid residues, α-helices of sufficient length can be formed to individually span a lipid membrane [88,89].At sufficiently high peptide concentrations, transmembrane helices may destroy the stability of bacteria membrane, or alternatively generate barrel-stave' [90,91] or toroidal pores [92].These pores act as some nonselective channels through which ions, toxins and metabolites can freely flow.As a result of the preventing maintenance of homeostasis, the microbe may eventually be killed.On the other hand, peptides with less than 20 residues cannot span through a membrane individually, but are more likely to act through the "carpet" mechanism [88, 90, and 91].This mechanism involves shorter peptides accumulating on the surface of membrane until a critical threshold forms and they solubilize or lyse the membrane like a detergent.
Besides the length, factors that determine which mechanism transpire include peptide-induced membrane thinning [93,94] and hydrophobic mismatch [95,96].All mechanisms may depend on the peptide concentration relative to threshold values (e.g.alamethicin and magainin [97]).Wong et al [98] pointed out that structure and activity of caerin 1.1 interacting with membranes is similar to other antimicrobial peptides using a carpet-like mechanism.This involves individual peptides aggregating in a helical manner and orienting themselves parallel to the membrane in a sheet-like arrangement described elsewhere [99].While Huang et al [100] found that antibacterial peptides magainin extracted from the skin of the amphibian have 23 residues and kill bacteria in a toroidal (or wormhole) model by which permeabilise the cytoplasmic membrane, they do not lyse eukaryotic cells.Peptide F1 is the modified caerin 1.1, and may have a similar antibacterial mechanism to that of caerin 1.1.Meanwhile, we observed that caerin peptide F3 has a different sequence of residues from that of F1.F3 has 25 residues while F1 have 29 residues.Though both belong to longer peptides, but they have different length of residues, thus may act as different mechanisms of killing bacteria.From the previous report, if the concentration of longer peptides exceeds the threshold value they may destabilize the bacteria membrane or span the membrane by generating a barrel-stave, toroidal pores or other models.
Interestingly, when peptide F3 was immobilised on metal AZ31, the immobilised material showed bacterial resistance for about 100 hrs, yet did not show this property when not immobilised on material (Fig. 3B).We suspect that a stable microstructure is formed when the peptide is immobilised on the metal surface which maintains the correct secondary microstructure of the peptides.However, it is still unclear by which mechanism peptide F3 kills bacteria.The relationship between the residues and membrane mechanism is also unknown.Therefore, further investigation is required to reach any substantive conclusion.
Conventional materials may allow the bacteria to form a bacteria biofilm which will may eventually lead to PPIs and result in a failure of an implant.F3 peptide-immobilised AZ31 has the potential to prevent bacteria (MRSA) growth when used as implanted hard tissues or internal fracture fixing materials.This specific property may avoid the adhesion and bacterial colonization.Therefore, F peptide-immobilised AZ31 material may become an ideal implant material which exhibits both degradation decelerating and bactericidal properties.Annealed AZ31 anneals at about 330-350°C, resulting in full recrystallization and a homogenous microstructure lacking any residual stress (Fig. 8B).In comparison, cold extruded AZ31 samples suffer an axial tensile force and circumferential compressive stress [101,102] (Fig. 8A).In this study we did not consider the effect of hydrostatic pressure.Stresses of cold extrude AZ31 sample can be obtained by the following Equation (8).
only showed a 24 hrs antibacterial effect (Fig. 6).Within 24 hrs cold extruded AZ31 immobilised by peptide F3 showed less antibacterial effect, but its antibacterial effect lasted for 100 hrs.

Conclusions
In this study, a method to immobilise host defence peptide caerin 1.9 (F3) and modified caerin 1.1 (F1) on PU immobilised AZ31 (cold extrusion and fully annealing) was developed.We found that relative to un-immobilised metal samples, the immobilisation of both peptides on PU coated AZ31 samples showed observable improvements in corrosion resistance.Moreover, F3 immobilised on AZ31 exhibited superior corrosion resistance than F1-samples.Meanwhile, both peptide-immobilised AZ31 samples displayed improved antibacterial effects, with F3 immobilisation exhibiting a longer-lasting and more notable bacterial resistance than the widely used clinical antibiotics Tazocin.The distinct antibacterial behaviours may be caused by different antibacterial mechanisms.
On the other hand, different original treatments of samples showed significant influence on corrosion resistance and antibacterial behaviours of peptide immobilised samples.Due to complicated original stress state of the cold-extruded sample, the resulted residual stresses can easily form some surface defects (like H2 gas pockets), which can lead to a break of local coating area and accelerate the coating's failure.Conversely, more compact and uniform coating can be formed on annealed samples due to their homogeneous stresses and microstructures.All the advantages can significantly improve these samples' corrosion resistances.Also, different original treatments on samples can lead to different coverages of peptides (F1 or F3) on metal substrates, further affect their antibacterial abilities.
Another interesting observation is that excellent and long-lasting antibacterial behaviours of samples may presumably resulted from the microstructure formed by the peptide and the metal substrate.This will require further investigation.
Taken together, the current results suggested that F1 and F3 immobilised on PU-coating Mg alloys may become a promising implant material in hard tissues both inside the human body and in open wound areas.

Fig. 1 .
Fig. 1.Mg alloy fabrication and peptide immobilisation procedure.(A) Schematic of Mg samples and all size parameters; (B) Surface modification of polyurethane coating Mg alloy by chemical clickreaction using O2 plasma; (C) Mg alloy immobilised by polyurethane and immobilisation of F1 or F3 on surface.
the surface cleaning of the original two Mg alloys AZ31 with two different treatments the two Mg alloy samples were immobilised by F1 and F3 polypeptides through being cleaned by distilled water in an ultrasonic cleaner for 3-5 min to remove impurities and adhesives from the surface.The cleaned samples were put into a bacteriostatic petri dish to conduct a 100-hr bacteriostatic test on drugresistant S. aureus in a 37°C temperature incubator.The methicillin-resistant S. aureus (MRSA, GDM1.1263) were cultured to a logarithmic phase and adjust the suspension concentration of MH (Mueller-Hinton) medium to 2.0 × 10 5 CFU/ml.A sterile cotton swab was used to dip the bacteria solution and squeeze the tube wall several times to remove the excess.The swab was used to smear the entire M-H drug-sensitive agar plate (Guangzhou Yuanming Bio Company).Aliquots of 30 µg of F1 and F3 peptides were added drug-sensitive papers (OXOID, UK) and the papers were pasted on M-H agar plates.The plates were inverted and incubated at 37°C overnight.A volume of 30 µg piperacillin sodium and tazobactam sodium with a weight ratio of 8:1 (Tazocin, Haikou Qili Pharmaceutical Co., Ltd, Haikou) and blank drug-sensitive tablets (BASD, Thermo Fisher Scientific, Shanghai) and two original AZ31 Mg alloys were used as controls.A Vernier calliper was used to measure the size of the zone of inhibition.

Fig. 2 .
Fig. 2. Surfaces of different specimens: (A) Mg alloy only; (B) Mg alloy with PU coating, and (C) PU coated Mg alloy after bonding force test

Fig. 4 .
Fig. 4. Analysis of weight loss of different samples over 120 hrs in vitro corrosion tests: (A) comparison of F3 immobilised extruded AZ31 and fully annealed AZ31; (B) F1 immobilised extruded AZ31 and fully annealed AZ31; (C) three different surface conditions of annealed AZ31 samples; (D) three different surface conditions of extruded AZ31 samples.

Fig. 5 .
Fig. 5. SEM mapping of different samples after 120-hr vitro corrosion test.(A) The surface of the metal non-immobilised with any peptides; (B) the annealed Mg alloy AZ31 immobilised with polypeptide F coating; (C) annealed AZ31 immobilised with peptide F3; (D) EDS analysis of C and MgO on the surface at 120 hrs.

4. 3
Effect of different original treatments on corrosion resistance and coating of antibacterial peptides on Mg alloys AZ31 with different treatment showed different behaviours in in vitro corrosion and bacteriostatic tests.The annealing or extrusion result in different surface stresses and surface microstructures, which might lead to different strengths of adhesion with PU, and consequently cause the outside of PU layer for O2 plasma and EDC/NHS reaction differently.This could eventually impact the amount of peptide that could be covalently immobilized onto the surface.Just as showed in Section 3.1, coverage of F1 or F3 is obviously different between extruded AZ31 and annealed AZ31; compared to the extruded AZ31, annealed AZ31 has the less coverage of F3, but more coverage of F1.The different coverages of peptides on Mg alloys may lead to their different antibacterial abilities.Also, different corrosion resistance can lead to different lasting time of antibacterial behaviour.Thus, brief comparisons of effects of annealing and extrusion on corrosion resistance were necessary.

Fig. 8
Fig. 8 Stress states of different samples and their coatings.(A) Stress state of cold extruded AZ31 before and after peptide coating.(B) Stress state of annealed AZ31 before and after peptide coating.(C)Schematic of H2 pocket formation and broking process, from original coating and Mg substrate, to a formed local bulging area, to corrosion reaction in the bulging areas and forming of H2 pocket, and final broken of H2 pocket.SEM maps of cold extruded AZ31 immobilised by peptides F1 or F3, H2 gas pockets appears (D) and a broken H2 gas pocket (E).

Figs. 8A
Figs. 8A and 8B show stress state of two samples with extruded and annealed treatment.