Synergistic effect of cerium and structure directing agent on drug release behavior and kinetics

This work was mainly aimed at the study of the dual effect of cerium and structure directing agent template concentration, on vancomycin release profiles and kinetics from Ce-substituted mesoporous bioactive glasses (MBGs). MBG based on (20-x) B2O3 - 55 SiO2 – 20 CaO – 5 P2O5 – x Ce2O (x = 0, 1 and 3 mol %) was synthesized by the evaporation-induced self-assembly process using two molar ratios (0.01 and 0.02 molar ratio) of nonionic block copolymer Pluronic® 123 (P123) surfactant. The TGA-DTA, FTIR, and textural features analyses were carried out for the glasses. Moreover, the in vitro bone-forming activity and degradation analysis were tested using simulated body fluid (SBF). The drug loading capacity, drug release profile, and kinetics (using different kinetic models such as first order, Higuchi, Hixson-Crowell, and Baker-Lonsdale models) were determined using vancomycin as a drug model. The results showed that the isotherms of all MBGs fit with type IV isotherms, and the surface area of MBGs synthesized by 0.02 M template was higher than that prepared by 0.01 M, where it ranged from 174.05 m2.g−1 to 256.73 m2.g−1. The pore size diameter was decreased as cerium content increased in all MBGs (decreased from 5.44 to 3.54 nm). Moreover, the MBGs induced the formation of a bone-like apatite layer, and their biodegradation properties can be tailored by controlling glass composition. Furthermore, Ce-free MBGs showed the lowest drug adsorption and the highest drug release percentage. The drug release kinetic was best fitted with Higuchi and Baker-Lonsdale models which denoted that the mechanism of drug release from MBGs was a diffusion release from spherical particles. In conclusion, vancomycin release was controlled by the glass composition. Meanwhile, the MBGs synthesized in this study are proposed to be applied for bone regeneration, bone cancer treatment, and reducing the bacterial activity around the tumor site. Graphical abstract Graphical abstract The surface area and pore size of NBG is tailored by controlling structure directing agent template concentration. Cerium is added to the glass as anticancer oxide. The drug release kinetics can be regulated by cerium and structure directing agent template. The surface area and pore size of NBG is tailored by controlling structure directing agent template concentration. Cerium is added to the glass as anticancer oxide. The drug release kinetics can be regulated by cerium and structure directing agent template.


Introduction
Due to the shortage of traditional drug delivery, there have been many attempts to find new materials to solve these disadvantages. Nevertheless, traditional drug delivery systems are usually not sufficiently effective due to the need for a high concentration of drugs to reach high enough concentrations in diseased tissues. Therefore, drug delivery systems are a superior alternative method to avoid this shortage. However, the application of biocompatible materials as drug delivery systems is very useful in this approach. Mesoporous bioactive glasses are superior biocompatible materials [1][2][3][4]. Its bioactive nature of it accompanied by its ability to deliver a drug locally is an outstanding perspective for bone therapy purposes.
Mesoporous bioactive glasses are new sol-gel derived materials that have attracted special interest in the last years. Because they have better bioactivity, high surface area and porosity, large mesoporous volume, and their ordered mesoporous network, as well as, their capability to host active agents and drugs. In this strategy, the incorporation of a structure-directing agent is essential for obtaining mesoporous structures. The essential feature of structuredirecting agents is the presence of chemically bonded hydrophobic and hydrophilic components that makes phase segregation on the nanoscale. Where these molecules selforganize into micelles. Micelles link the hydrolyzed oxide precursors through the hydrophilic component and selfassembly to form an ordered mesophase [5].
The substitution of bioactive glasses with therapeutic ions, such as Zn, Mn, Cu, Li, Fe, Ag, and Ce ions, to treat specific diseases has produced beneficial materials in different biomedical applications. Among other therapeutic elements cerium, it is a rare earth element and it has found several applications such as fuel additives and catalysts. Recently, it was used in biomedical applications, and its evaluation in such a field is still underestimation. Cerium has shown promising antimicrobial activities [6][7][8], it was demonstrated that its antimicrobial activity came from an ability to dissociate the outer membrane of bacterial cells from the cytoplasmic membrane [9]. Therefore, ceriumdoped bioactive glasses have been potentially applied in the treatment of periodontal pockets and hypersensitive teeth [10]. Moreover, cerium showed in vivo low toxicity for a rat (1000 mg/kg of body weight) [11]. Besides its antibacterial properties, cerium oxide nanoparticles have also been shown to protect nerve cells [12], and they used recently, to sweep reactive oxygen species accumulated in neuron cells, so they decrease the death of such cells [13].
In addition, the incorporation of boron in bioactive glass compositions improves their properties as bone fillers. Where, it was reported that it can improve bone growth [14,15]. Furthermore, boron compounds conveyed a great antiinflammatory effect with minimum side effects [16]. Besides, boron-containing compounds can reduce the reactive oxygen species (ROS) [16,17]. Accordingly, the addition of boron and cerium to the bioactive glass is proposed to achieve the antioxidant and antibacterial activities of the glass.
There have been numerous previous works that studied cerium-containing mesoporous bioactive glass (MBG). Salinas, et al. studied an effect of substitution of 3.5% Ce 2 O 3 , 3.5% Ga 2 O 3 or 7.0% ZnO on the textural properties of mesoporous bioactive glass (MBG) based on 80 SiO 2 -15 CaO -5 P 2 O 5 (mol. %) system [18]. However, they did not evaluate such glasses as a drug delivery system and they did not study the effect of the addition of these ions to the drug release profiles and kinetics. But, they evaluated them in a subsequent study as a delivery system for curcumin, and they showed that the amount of curcumin released was affected by the amounts of added cerium and gallium [19]. Varini, et al. substituted the same previous glass system with 1.2, 3.6, and 5.3 mol % of CeO 2 and fabricated hydrogel beads based on alginate and these glasses and studied their antioxidant properties [20]. Similarly, Nicolini, et al. studied investigated the bioactivity and enzymatic-like mimetic activities of Cesubstituted MBG [21]. These studies did not determine these MBGs as a drug delivery system. Atkinson, et al. studied the physical properties and biocompatibility of Ce-containing MBGs based on a 70 SiO 2 -(26-x) CaO -4 P 2 O 5 -x CeO 2 (x = 0, 1 and 5 mol %) system [22]. Likewise, Kurtuldu, et al. studied the bioactivity, biocompatibility, and antibacterial activity of Ce-and Ga-containing MBGs [23]. Zhang, et al. compared the doping of rare earth ions (cerium was one of them) in MBG from the view of in vitro bioactivity, and doxorubicin drug release [24]. Kurtuldu, et al. evaluated the antibacterial and anti-inflammatory activities of Ce-containing MBG. Herein, Ce ions were incorporated into the glass with two routes; by direct addition of Ce ions during the glass synthesis, or by the template ion-exchange method. The results showed that the glass prepared by the first method showed higher cytotoxicity toward preosteoblast cells than that synthesized by the second approach [25].
However, most previous works did not study deeply the effect of cerium amount together with template concentration on the drug release profile and kinetics of MBGs. We previously studied the in vitro ciprofloxacin drug release behavior of Ce-doped nanobioactive glass based on 60 SiO 2 -(10-x) B 2 O 3 -25 CaO -5 P 2 O 5 -xCeO 2 (x = 0 and 5 mol %) system in oxidative stress (H 2 O 2 ), and the study concluded that when Ce-containing nanobioactive glass used as a drug delivery system, the nature of the surrounding medium affected the drug release behavior from such glass particles [8].
This work was aimed at the study of the change of cerium content and template concentration on drug loading efficiency and release kinetics using different kinetic models, such as first order, Higuchi, Hixson-Crowell, Baker-Lonsdale, and Korsmeyer-Peppas models, to study the mechanism of drug release from MBGs to tailor the glass properties to be used as a successful drug delivery system.

Characterization of mesoporous bioactive glass
The textural features of different glass samples were determined by nitrogen adsorption porosimetry by using Belsorp max BEL, Japan INC. Before the adsorption measurement, the samples were degassed under vacuum for 24 h at 120 o C. The surface area was measured by applying Brunauer-Emmett-Teller (BET) method [26]. The pore size distribution was obtained by Barret-Joyner-Halenda (BJH) method [27]. Thermogravimetric analysis (TGA) and differential thermal analysis (DTA) were carried out in a Pyris Diamond TG/DTA thermal analyzer using an airflow of 200 ml.min −1 and heating from 35 to 1000 o C at 10 o C min −1 . The infrared (IR) absorption spectra of bioactive glass powder samples were analyzed at room temperature in the wave number range of 4000-400 cm −1 using Fourier transform infrared (JASCO FT/IR-4600). The prepared samples each of 2 mg were mixed with 200 mg KBr in an agate mortar and pressed into a pellet. For each sample, the FTIR spectrum was normalized with a blank KBr pellet.

Non-cellular in vitro bioactivity test
The bone-forming activity of MBGs in vitro will be tested by the samples in simulated body fluid (SBF) to monitor the formation of HCA on the surface of MBGs over time.

Drug delivery experiment
A definite volume of glass samples (20 mg) was loaded with vancomycin HCl. Where they will be immersed in a definite amount of vancomycin drug (1 mg/ml) solution for 2 days under static conditions. The drug solutions will be collected, and their absorbance was measured by UV-visible spectroscopy at a wavelength of 282 nm. The concentration of non-adsorbed drug were determined by plotting that absorbance on the standard curve. For drug release evaluation, each sample was immersed in 10 ml of phosphate buffer saline (PBS) (with ion concentrations 137 mM NaCl, 2.7 mM KCl, 8 mM Na 2 HPO 4 , and 2 mM KH 2 PO 4 ), pH 7.4, in the static condition at 37°C for a period. At each predetermined time (1, 3, 6, 12 h, 1, 2, 4, 7, 11, and 15 d) 1.5 ml of immersed solution was collected and replaced by 1.5 ml of a fresh one. The collected solutions were kept at -20°C up till measurement. The concentrations of released drug in the collected solutions were determined by measuring the absorbance by a UV-visible spectrophotometer at wavelength 282 nm. The concentrations were calculated by comparison with a standard curve.

Drug release kinetic
The results of concentrations of released vancomycin were fitted in different kinetic equations, such as first order, Higuchi, Hixson-Crowell, and Baker-Lonsdale models, to study the mechanism of vancomycin release from the samples. The following equations represent different kinetic models: First order: it is the relation between the log cumulative percentage of released drug and time [28]. This model has been used to describe the absorption and elimination of drugs.
Higuchi model: it is the relation between the cumulative percentage of released drug and the square root of time [29]. This model is useful for studying the release of watersoluble and poorly soluble drugs from a variety of matrices, including solids and semi-solids.
Hixson-Crowell model [30]: this expression applies to pharmaceutical dosage forms such as tablets, where the dissolution occurs in planes that are parallel to the drug surface if the tablet dimensions diminish proportionally, in such a manner that the initial geometrical form keeps constant all the time.
Baker-Lonsdale model [14]: this model was developed by Baker and Lonsdale (1974) from the Higuchi model and described the drug release from spherical matrices according to the equation: Korsmeyer-Peppas model is represented by the following equation: To find out the mechanism of drug release, the first 60% of drug release data were fitted in such a model. Where Q t is the amount of drug released in time t, Q 0 is the initial amount of drug in the sample, K 1 , K H , K HC , K BL, and K KP are rate constants calculated from first order, Higuchi, Hixson-Crowell, Baker-Lonsdale, and Korsmeyer-Peppas models, respectively, and n is the kinetic exponent. Peppas [15] used this "n" value to characterize the drug release mechanism. For the case of cylindrical tablets, 0.45 = n corresponds to a Fickian diffusion mechanism, 0.45 < n < 0.89 attributed to non-Fickian transport, n = 0.89 to Case II (relaxational) transport, and n > 0.89 to super case II transport. The kinetic exponent, n, can be determined by plotting of log cumulative percentage of drug release versus log time.
To determine which model is suitable for drug release kinetics, the regression coefficients, R 2 , were calculated by regression analysis.

Statistical analysis
The sample was tested three times (n = 3) in this paper and the experimental data were presented as means ± standard deviation (SD). Student t test was used to test for significant differences between the means of data sets. An alpha level of 0.05 was considered the minimum level for statistical significance.

Results and discussion
3.1 DTA-TGA Figure 1a and b shows the TGA curves and DTA traces, respectively, of selected samples (G0-B, G1-B, and G3-B dry gels prepared by 0.02 M template). The TGA curves showed that there were two main stages of weight loss. The first stage of weight loss occurred between about 25°C and 200°C, which can be assigned to the evaporation of water molecules, organics of precursors, and P123 template burnout. As mentioned in a previous study, the weight loss of P123 was recorded between 150°C and 250°C [31]. The second stage of weight loss was detected between 300°C and 450°C, which was attributed to nitrate loss and remains of organic components. Regarding the DTA analysis of dry gels (Fig. 1b), there were two obvious exothermic peaks. The first exothermic peaks were detected at 213°C, 207°C, and 203°C, and the second ones were recorded at 354°C, 340°C, and 320°C for G0-B, G1-B, and G3-B, respectively. These two peaks can be revealed as combustion peaks of P123 template burnout. Then again, such combustion peaks were decreased as Ce content increased, which can be explained by the disruption effect of Ce on the glass network. Thoroughly, as will be explained in the next paragraphs, cerium oxide increased the number of non-bridging oxygens (NBOs) causing the generation of more voids inside the glass framework, such voids were facilitated a burning off a template from dry gel leaving behind pores in the final glass material.
3.2 Surface area and pore size Figure 2a and b shows the N 2 adsorption isotherms of substituted cerium mesoporous glasses prepared by two molar ratios of P123 template (0.01 M and 0.02 M). From the curves, it could determine the type of isotherms, where, they were detected as type IV isotherms. Moreover, the figures showed that all glass samples demonstrated a uni-modal pore size distribution centered between 3.5 and 5.5 nm. Then again, the observation of the adsorption isotherms showed that as Ce content in glass increased the adsorption capacity of the samples decreased. Figure 2c and d, and Table 2 represent surface area (m 2 .g −1 ), pore size (nm), and total pore volume (cm 3 .g −1 ) of different cerium substituted MBGs prepared by two molar ratios (0.01 M and 0.02 M) of P123 template. For glasses prepared by 0.01 M template, it can be noticed that surface area was decreased as Ce content increased, where it was recorded for G0-A, G1-A, and G3-A samples as 241.36, 223.5, and 174.05 m 2 .g −1 , respectively. On contrary, in the case of using 0.02 M, the surface area was increased as the Ce percentage increased. Where, it was 229.23, 225.11, and 256.73 m 2 .g −1 for G0-B, G1-B, and G3-B, respectively. On the other hand, the pore size diameter was decreased from 5.44 nm for G0-A and G0-B samples to 4.39 nm and 3.54 nm for G3-A and G3-B samples, respectively. Interestingly, the pore size diameter decreased as the template molar ratio increased for analogous samples.
The isotherms of all MBGs showed that they were fit type IV isotherms. This type of isotherm is characteristic of mesoporous materials. Therefore, the shape of isotherms of glasses under investigation confirmed their mesoporous texture. On the other hand, the decrease of the adsorption capacity of the samples by the increase of Ce content was attributed to the presence of disordered domains, which reduces the mesopore volume [32]. Furthermore, a unimodal pore size distribution observed in the figures indicated the homogeneity of pore size diameter.  sample were attributed to asymmetric stretching and bending of Si-O-Si, respectively. The former peak was shifted to higher frequencies as Ce wt. % increase in the glass, and it became 1058 and 1065 cm −1 for G1-B and G3-B, respectively. This can be explained by the role of Ce in the glass network, it likely strengthened such Si-O bond. Moreover, the peak shoulder appeared at 941 cm −1 became and stronger in G3-B glass than in G0-B and G1-B glasses. This band was believed to be a result of either Si-O-M or nonbridging oxygen (NBO) [33], where M might be Ca, B, or Ce. Furthermore, the weak and broad peak centered at 1457 cm −1 for G0-B glass may be ascribed to B-O stretching vibration mode [34], this peak was shifted to a lower frequency (1417 cm −1 ) for G3-B glass due to a decrease of wt. % of boron in the glass composition. In addition, the peak noted near 799 cm −1 was attributed to O-Si-O bending vibration mode in SiO 4 4− , which shifted to a higher frequency (804 cm −1 ) as Ce content increased. Despite this peak, it can be proposed that Ce disrupted the silicate network, which can be confirmed by the previously mentioned NBO vibration mode that appeared at 941 cm −1 . The band noted around 569 cm −1 was assigned to P-O bending mod.  Table 2 Surface area (m 2 .g −1 ), pore size (nm), and total pore volume (cm 3

In vitro degradation and bioactivity test
The key to the bioactivity of a material is its ability to form an apatite layer on its surface which is more viable to bond with the living cells. Determination of this property can be simply performed by immersing the sample in SBF and following up the formation of a new bone-like apatite layer. Figure 4 represents SEM and EDX analysis of G0- Fig. 4 a-c are G0-B, G1-B, and G3-B glass samples, respectively, before immersion in SBF. d-f are G0-B, G1-B, and G3-B glass samples, respectively, after immersion in SBF for 15 d B, G1-B, and G3-B glass samples before and after immersion in SBF for 15 days. the figure showed that there was no big change between the surface of G0-B glass sample before and after immersion in SBF, except tiny spheres that observed on glass surface after soaking in SBF, as well as, the intensity of P was slightly increased in EDX analysis. For G1-B glass, larger spheres were observed on the glass surface after immersion in SBF, and the intensity of P in EDX analysis was increased than that observed in G0-B glass. Furthermore, the surface of G3-B glass was completely covered by more mature spherical forms, in addition, the intensity of P and Ca in EDX analysis was increased and they became higher than the intensity of Si. Moreover, Ce was detected in EDX analysis for glass-contained Ce before and after immersion in SBF. Ca/P ratios measured from EDX analysis for G0-B, G1-B, and G3-B were 2.16, 1.37, and 1.46, and consequently, the ratio of G3-B was closer to the Ca/P ratio (1.67) of stoichiometric hydroxyapatite than G0-B and G1-B samples. And so, the addition of Ce to the glass enhanced the glass's bioactivity. These findings are reversing the results found in previous research. Where, it was found that the addition of Ce to a melt-derived glass reduced the in vitro bioactivity [35,36] as a result of the formation of less soluble CePO 4 crystals on the glass surface. This can be explained by the higher surface area of sol gel-derived glass than melt-derived glass, and so higher reactivity in SBF to form a new bone-like apatite layer. On the other hand, the substitution of boron with cerium made a disruption of the silicate network which increased the number of NBOs (as confirmed by IR spectra) throughout the silicate framework. Such NBOs can be easily hydrated by water when immersed in the simulated fluid to form silanol (Si-OH) groups as an initial step to form hydroxyapatite crystals. The change of pH of different MBGs samples was explored by following pH values of SBF at predetermined times. Figure 5a and b shows pH variations with the time for SBF immersion medium. It can be observed from the figure that the pH change outlines for both group samples (glass made by 0.01 M and 0.02 M template) were nearly similar. The pH values increased from the initial time to the third day of immersion. This was as attributed to a rapid ion exchange between Ca 2+ and H + or H 3 O + in SBF solution which caused an increase of hydroxyl groups in the solution, and so, a rising in fluid pH. This exchange resulted in the breaking of Si-O-Si glass network bonds and formed a silica-rich layer composed mainly of SiOH (silanol) groups on the glass surfaces. This newly formed layer possessed the affinity to attract Ca 2+ and PO 4 3− from the surrounding solution and subsequently formed bonelike apatite crystals [37]. This increase in pH was followed by a decrease in its value between 3 d and 10 d which was attributed to the formation of the apatite layer on the glass surface.
The change of concentrations of cerium, calcium, and phosphate ions in SBF was also examined as an indication of the formation of hydroxyapatite on the glass particle surfaces. Figure 5b-d shows the cumulative concentration of calcium and phosphate ions in SBF solution incubated G0-A, G1-A, G3-A, G0-B, G1-B, and G3-B samples. Moreover, a linear fitting of the cumulative concentrations with the square root of time was carried out to calculate the ions' release rates. The released Ce ion concentrations in the solution cannot be measured because they were too low to be measured (<1 ppm). It can be noted from the figure that all ion concentrations were nearly linearly increased with time. But, they became relatively low in the later stage of incubation. The relatively fast ion release rate in the initial time can be allocated to the ion dissolution from the glass surfaces into the incubating medium, while, the relatively slow release was assigned to a precipitation of the bone-like apatite layer on their surfaces [37]. The concentration of Ca 2+ ions in SBF contained G0-A was significantly (p < 0.05) higher than G1-A and G3-A glasses. Whereas, the concentration of G3-B was significantly (p < 0.05)lower than G0-B and G1-B. The computed Ca 2+ ion concentration rate of G0-A, G1-A, and G3-A was 146.6 ppm.d −0.5 , 179.9 ppm.d −0.5 , and 127.9 ppm.d −0.5 , respectively, and 209.8 ppm.d −0.5 , 199.4, ppm.d −0.5 , and 156.1 ppm.d −0.5 for G0-B, G1-B, and G3-B, respectively (Table 3). However, the template concentration presented a significant effect on the release rate of calcium ions, where the release rates were increased by the increase of the template concentration. This can be explained by a higher surface area, and so a higher number of NBOs generated the glass network by Ce addition. The release outlines of phosphate ions and rates from different glass samples were similar. The concentrations were relatively high during the first 2 d, followed by relatively low concentrations till the end of incubation time.     of soaking in buffer solution, while, this percentage was released from Ce-contained samples (G1-A, G3-A, G1-B, and G3-B) after 4 days of incubation. The fast release was followed by a nearly steady state release throughout the rest of the incubation period. Nevertheless, glass-free Ce (G0-A and G0-B samples) released a higher percentage of the drug than glass-contained Ce regardless of the molar ratio of the template used in MBG synthesis. Where, G0-A and G0-B released about 56 % and 64 %, respectively, at the end of soaking time. While, G1-A, G3-A, G1-B, and G3-B samples released around 34 %, 26 %, 28 %, and 27 %, respectively, at the end of immersion time. This sustained release is appropriate to provide the infected tissue with an effective drug dose in the fast-release stage to destroy the bacteria, followed by a slow-release stage for a long time to prevent bacterial re-infection [38].

Drug delivery
To analyze the in vitro release kinetics of vancomycin drug, different release data were fitted in different kinetic models (first order, Higuchi, Hixson-Crowell, Baker-Lonsdale, and Korsmeyer-Peppas), and the correlation  coefficient, R 2 , was used as an indication of data fitting to know the mechanism of drug release from these formulations. The drug is usually adsorbed into the mesopores of MBG either in physical way or by forming bonding between it and the inner glass pore channel. The dimensions of the vancomycin molecule are 2.2 × 3.2 nm (diagonal~3.9 nm) [39] which able the drug to exist inside the pore channels. In addition, the vancomycin molecule includes highly electronegative ions, such as OHwhich can form hydrogen bonding with the Si-OH and P-OH groups at the surface of pore wall of MBG [40]. The results showed that glasscontained cerium adsorbed more amounts of the drug than that Ce-free ones. As mentioned before, the existence of cerium in the glass structure caused more disruption of the glass silicate framework and increase the number of NBOs, and so increased the number Si-OH groups in the aqueous solution [41]. This was more obvious in the case of using a 0.02 M template in MBG synthesis due to the higher surface area of these glasses. Moreover, Ce-free MBGs showed drug release with two-step kinetics, where the fast release of vancomycin molecules was due to a discharge of physically adsorbed drug molecules, while, the slow release was because of a withdrawal of drug molecules placed inside the mesopores. While Ce-substituted MBGs showed one-step drug release kinetics. That was attributed to the formation of more hydrogen bondings between the drug and the inside pore surface. Thus, vancomycin release was controlled by the glass composition, and cerium can be tailored to the drug loading and release kinetics of MBGs, which was advantageous for reducing the bacterial activity around the tumor site.

Conclusion
In this work, multifunctional Ce-substituted mesoporous bioactive glasses (MBGs) were prepared for antibacterial and anticancer application. The dual effect of cerium and structure directing agent template concentration on vancomycin release profiles and kinetics were evaluated. The isotherms of all MBGs showed that they fit type IV isotherms, and the surface area of MBGs synthesized by 0.02 M template was higher than that prepared by 0.01 M. in addition the pore size diameter was decreased as cerium content increased in all MBGs. Moreover, the biodegradation properties of MBGs can be tailored by controlling glass composition. Thus, vancomycin adsorption and release were controlled by the glass composition, which was more obvious in glasses characterized by high surface area MBGs prepared by 0.02 M template. Thus, cerium can be tailored to the drug loading and release kinetics of MBGs. And hence, MBGs synthesized in this study are proposed to be applied for bone regeneration, bone cancer treatment, and reducing the bacterial activity around the tumor site.