*2.8. Minimum Inhibitory Concentration (MIC)*

The MIC value of cefazolin used against Gram-positive *S. aureus* ATCC 25923 was assessed by using the standard broth macrodilution method as described previously [29].

Cefazolin solution was prepared by reconstituting 1 g vial cefazolin with 10 mL sterile aCSF to give a resultant concentration of 100 mg/mL and further diluted to obtain a stock solution of 0.01 mg/mL (10 μg/mL). A series of 2-fold higher concentrations of cefazolin than the final dilution range as well as control was prepared in duplicate as follows: 10, 5.0, 2.5, 1.25, 0.625, 0.3125, 0.1560, 0.0780, 0.0390 and 0.000 μg/mL. In addition, uninoculated wells of antibiotic-free broth were also prepared to ensure the sterility of the cefazolin solution and aCSF. All the wells were then incubated at 35 ◦C ± 2 ◦C for 18–20 h. The MIC value was determined as the lowest dilution of cefazolin that completely inhibited the visible growth of the test organism which is evidenced by the absence of turbidity in the well. The bacterial inoculum was prepared according to a standard broth culture method (European Committee for Antimicrobial Susceptibility Testing of the European Society of Clinical and Infectious, 2000). In brief, the bacterial colonies were taken with a sterile loop and transferred to 5 mL of sterile nutrient broth (NB). The broth was then incubated in a shaking incubator at 37 ◦C until the visible turbidity (4–6 h) was observed. The density of the suspension of bacterial culture was diluted with NB to give a turbidity equivalent to the 0.5 McFarland standard (approximately 1.5 × <sup>10</sup><sup>8</sup> cfu/mL). This was performed using visual inspection such that the appearance of black lines was compared when observed through the bacterial inoculum and McFarland standard suspension. This inoculum was then transferred to a tube, which was further adjusted with NB to achieve a final organism density of 1.5 × <sup>10</sup><sup>6</sup> cfu/mL.

#### *2.9. Antibacterial Activity*

The antibacterial activity of the samples was evaluated by using the agar disk-diffusion method or Kirby–Bauer test. The *S. aureus* ATCC 25923 (1.5 × 106 cfu/mL) were inoculated on Mueller–Hinton agar plates. Specimens (10 mm × 10 mm) were then gently placed down to ensure even contact to the agar surface. Cefazolin loaded ORC/PCL bilayered composites contained two distinct surfaces, namely rough or porous surface (R) and smooth or dense surface (S), resulting from the processing. Both surfaces of the composite were independently evaluated whereas only one surface of the cefazolin loaded PCL was tested due to indistinguishable appearance between the two sides. All the plates were then incubated at 37 ◦C for 24 h, in which the samples were transferred to fresh *S. aureus* agar plates every 24 h. This was performed repeatedly until the inhibition (clear) zones were not observed. The size of the clear space around the sample indicated the antibacterial activity of the sample. All dimensional measurement was carried out by using a vernier caliper and used to calculate the inhibition zone values using the following formula:

$$\mathbf{H} = (\mathbf{D} - \mathbf{d})/2,\tag{1}$$

where H is the inhibition zone (mm);

D is the clear space diameter around the sample (mm);

d is the specimen diameter (mm).

#### *2.10. Statistical Analysis*

The data were described as the mean ± standard deviation values. Data were analyzed by using statistical analysis software (GraphPad Prism version 6, GraphPad Software, San Diego, CA, USA). One-way analysis of variance (ANOVA), followed by Tukey's multiple comparisons test, was employed to determine the significant between groups or samples. A *p*-value < 0.05 was considered statistically significant.

#### **3. Results**

#### *3.1. Physical and Mechanical Properties*

3.1.1. Microstructure, Bulk Density and Thickness

Figure 1 shows the microstructure of cefazolin loaded samples. Regardless of cefazolin concentrations, all cefazolin loaded PCL generally displayed an open microstructure

resulting from the exchange of solvent and water in the fabrication process (Figure 1a,c,e,g). However, the amount of dense area surrounding pores tended to increase with increasing cefazolin content in the PCL solution. Upon examining higher magnification images (Figure 1b,d,f,h), only PCL\_25 showed a bicontinuous-like structure having a pore size less than 5 μm while PCL\_50, PCL\_75 and PCL\_100 showed cellular structure having larger pore size of about 10–20 μm. In contrast, cefazolin loaded ORC/PCL bilayered composites exhibited a bilayer structure comprising a nonporous layer and a composite layer. In the case of the nonporous layer side, all formulations similarly displayed dense solid structure of PCL similar to those of cefazolin loaded PCL, but no pores were observed in this case (Figure 1i,j,m,n,q,r,u,v). In the case of the composite layer side, different microstructures were obtained depending on the cefazolin concentrations. At a low cefazolin concentration of 25 mg/mL, the composite layer side consisted of knitted ORC fabric mostly embedded within the continuous PCL matrix with exposed fabric in some areas (Figure 1k). Increasing the cefazolin concentration to 50 mg/mL resulted in a dense PCL layer without any exposed ORC fabric resembling the dense structure of the nonporous layer side (Figure 1o). Further increasing the cefazolin concentration to 75 and 100 mg/mL led to the coating of PCL on top of the ORC fabric since the knitted ORC morphology was still evidenced, but with the thick coating of the PCL layer on top (Figure 1s,w). At high magnification, PCL matrix was observed to be dense for P20\_25 and P20\_50 (Figure 1l,p) while some pores were observed in the PCL matrix close to the ORC fibers for P20\_75 and P20\_100 (Figure 1t,x).

Figure 2 shows the bulk density and thickness of cefazolin loaded and drug unloaded samples. The bulk density and thickness of all formulations of cefazolin loaded ORC/PCL bilayered composite did not differ significantly. In comparison to their corresponding drug unloaded samples (P20 or PCL), bulk density of both cefazolin loaded ORC/PCL bilayered composite and cefazolin loaded PCL samples was greater, but the thickness was lower. However, the significant difference (*p* < 0.05) in bulk density between cefazolin loaded and drug unloaded samples was seen for P20\_25 and all cefazolin loaded PCL except PCL\_100. In contrast, thickness of all cefazolin loaded ORC/PCL bilayered composites was significantly lower than that of P20 while only that of PCL75 was significantly different.

### 3.1.2. Tensile Properties

Figure 3 shows tensile properties of drug loaded and unloaded samples. Tensile modulus of cefazolin loaded ORC/PCL bilayered composite increased with increasing cefazolin concentration, which reached statistical significance when cefazolin concentration was 50 mg/mL or above (Figure 3a). However, this was not observed in cefazolin loaded PCL, in which the tensile moduli of all samples were seen to be in the same range and did not differ significantly from that of PCL (Figure 3b). Generally, tensile strengths of both cefazolin loaded ORC/PCL bilayered composite and cefazolin loaded PCL were found to be significantly lower compared to unloaded samples (*p* < 0.05). Tensile strength of cefazolin loaded PCL samples tended to decrease when cefazolin concentration was increased while those of cefazolin loaded ORC/PCL bilayered composite remained relatively unchanged (Figure 3c,d). All cefazolin loaded ORC/PCL bilayered composites exhibited similar tensile strain at break regardless of the concentration of cefazolin, but all were significantly lower than that of drug unloaded composite (Figure 3e). In contrast to cefazolin loaded ORC/PCL bilayered composite, tensile strain at break of cefazolin loaded PCL samples tended to decrease with increasing cefazolin concentration, but only reached statistical significance at the highest concentration of cefazolin (*p* < 0.05) (Figure 3f). Comparing between cefazolin loaded ORC/PCL bilayered composite and cefazolin loaded PCL at a similar cefazolin concentration, the composites had greater tensile modulus than those of PCL, but tensile strengths were in similar ranges. Tensile strain at break of cefazolin loaded PCL was much greater than those of cefazolin loaded ORC/PCL bilayered composite.

**Figure 1.** SEM images showing the microstructure of fabricated cefazolin loaded samples: (**a**) PCL\_25 at 50×; (**b**) PCL\_25 at 1000×; (**c**) PCL\_50 at 50×; (**d**) PCL\_50 at 1000×; (**e**) PCL\_75 at 50×; (**f**) PCL\_75 at 1000×; (**g**) PCL\_100 at 50×; (**h**) PCL\_100 at 1000×; (**i**) P20\_25 smooth side at 50×; (**j**) P20\_25 smooth side at 1000×; (**k**) P20\_25 rough side at 50×; (**l**) P20\_25 rough side at 1000×; (**m**) P20\_50 smooth side at 50×; (**n**) P20\_50 smooth side at 1000×; (**o**) P20\_50 rough side at 50×; (**p**) P20\_50 rough side at 1000×; (**q**) P20\_75 smooth side at 50×; (**r**) P20\_75 smooth side at 1000×; (**s**) P20\_75 rough side at 50×; (**t**) P20\_75 rough side at 1000×; (**u**) P20\_100 smooth side at 50×; (**v**) P20\_100 smooth side at 1000×; (**w**) P20\_100 rough side at 50×; (**x**) P20\_100 rough side at 1000×. Magnification 50× (bar = 500 μm) and magnification 1000× (bar = 10 μm).

**Figure 2.** Influence of cefazolin concentration in the ORC/PCL bilayered composite and PCL on density and thickness of fabricated samples: (**a**) Density of cefazolin loaded ORC/PCL bilayered composite; (**b**) thickness of cefazolin loaded ORC/PCL bilayered composite; (**c**) density of cefazolin loaded PCL; (**d**) thickness of cefazolin loaded PCL. Data are expressed as mean ± standard deviation (SD). Significance between cefazolin loaded samples versus control (P20 or PCL) is indicated as \* (*p* < 0.05).

#### *3.2. Total Cefazolin Content*

The measured total cefazolin loading contents in fabricated cefazolin loaded ORC/PCL bilayered composite and cefazolin loaded PCL are shown in Table 2. Increasing cefazolin concentration in the solution significantly increased the uptake of cefazolin in both types of fabricated samples. For a similar cefazolin concentration employed, the loading content of cefazolin in the ORC/PCL bilayered composite was greater than that of cefazolin loaded PCL.

**Table 2.** Total cefazolin contents loaded in the samples. Data are expressed as mean ± standard deviation (SD).


**Figure 3.** Influence of cefazolin concentration in the ORC/PCL bilayered composite and PCL on tensile properties: (**a**) Tensile modulus of cefazolin loaded ORC/PCL bilayered composite; (**b**) tensile modulus of cefazolin loaded PCL; (**c**) tensile strength of cefazolin loaded ORC/PCL bilayered composite; (**d**) tensile strength of cefazolin loaded PCL; (**e**) tensile strain at break of cefazolin loaded ORC/PCL bilayered composite; (**f**) tensile strain at break of cefazolin loaded PCL. Data are expressed as mean ± standard deviation (SD). Statistical significance between cefazolin loaded samples versus control (P20 or PCL) is indicated as \* (*p* < 0.05).

### *3.3. In Vitro Cefazolin Release*

Figure 4 shows the daily concentration of eluted cefazolin (μg/mL) from cefazolin loaded samples for up to 30 days. All samples displayed similar release profiles including a sharp initial burst release of the highest concentration of cefazolin on the first day, followed by a significantly declined release thereafter. While cefazolin was gradually eluted from cefazolin loaded PCL samples for approximately 15 days (Figure 4a), almost all cefazolin was rapidly eluted from cefazolin loaded ORC/PCL bilayered composite during the 4–10 days depending on the cefazolin loading (Figure 4b). Figure 5 shows the cumulative percentage of released cefazolin as a function of time for cefazolin loaded samples over 30 days. The burst and delayed release were observed for cefazolin loaded PCL samples (Figure 5a), where the cumulative percentage of released cefazolin was found to reach approximately 50% in 1 day. In contrast, cefazolin loaded ORC/PCL bilayered composite exhibited a greater burst release, in which most of cefazolin was released in

1 day and followed by constant small release thereafter (Figure 5b). No differences in daily released content or cumulative release among different formulations for both cefazolin loaded PCL and cefazolin loaded ORC/PCL composite were observed.

**Figure 4.** Cefazoline release profile of cefazolin loaded samples in aCSF solution: (**a**) Cefazolin loaded PCL; (**b**) cefazolin loaded ORC/PCL bilayered composites. Data are expressed as mean ± standard deviation (SD).

**Figure 5.** Cumulative cefazoline release profile of cefazolin loaded samples in aCSF solution: (**a**) Cefazolin loaded PCL; (**b**) cefazolin loaded ORC/PCL bilayered composites. Data are expressed as mean ± standard deviation (SD).

#### *3.4. Minimum Inhibitory Concentration (MIC)*

Table 3 shows a series of half-decreasing concentrations of cefazolin employed against *S. aureus* ATCC 50923. Based on this result, the MIC value, the lowest concentration of cefazolin that inhibited the bacterial growth, was 0.3150 μg/mL (Well 5).

 


**Table 3.** Minimum inhibitory concentration (MIC) value of cefazolin against *S. aureus* ATCC 50923.

#### *3.5. Antibacterial Activity*

Figure 6 shows the antimicrobial activity profiles of cefazolin loaded ORC/PCL bilayered composite and cefazolin loaded PCL, which were assessed by using Kirby–Bauer methods for up to 7 days. Cefazolin loaded ORC/PCL bilayered composite showed an antibacterial activity against *S. aureus* for up to 4 days while cefazolin loaded PCL showed a slightly longer activity for up to 5 days. The greatest inhibition zone was observed on day 1, which was relatively similar in both cefazolin loaded ORC/PCL bilayered composite and cefazolin loaded PCL regardless of the concentration of cefazolin. It can be noted that the inhibition zone of cefazolin loaded ORC/PCL bilayered composite tended to display a greater value compared to that of the cefazolin loaded PCL at the same concentration of cefazolin. The inhibition zone of the smooth side of the cefazolin loaded ORC/PCL bilayered composite and cefazolin loaded PCL tended to increase with increasing cefazolin content and the longest inhibition durations (4 and 5 days, respectively) were similarly found at 100 mg/mL (Figure 6b,c). Contrarily, the inhibition zone and antibacterial duration of the rough side of cefazolin loaded ORC/PCL bilayered composite did not correspond with the concentration of cefazolin loading (Figure 6a). In a similar tested period, P20\_75 tended to show the greatest inhibition zone while P20\_100 showed the second smallest inhibition zone. In the case of antibacterial duration, P20\_50 and P20\_75 displayed a longer period than P20\_50 and P20\_100. A drug unloaded composite (P20) also exhibited an antimicrobial activity on day 1 which decreased thereafter, in which the rough side (Figure 6a) exhibited a greater inhibition zone (6.1 ± 0.1 mm) compared to the smooth side (4.1 ± 0.2 mm) (Figure 6b). However, the inhibition zones of P20 on both sides were still smaller than those of cefazolin loaded ORC/PCL bilayered composite. In contrast, the drug unloaded PCL did not show any antibacterial activity.

**Figure 6.** Antibacterial activity of cefazolin loaded samples against *S. aureus* as indicated by inhibition zone: (**a**) Rough side of cefazolin loaded ORC/PCL bilayered composite; (**b**) smooth side of cefazolin loaded ORC/PCL bilayered composite; (**c**) cefazolin loaded PCL. Bar = 10 mm.

#### **4. Discussion**

The microstructure of cefazolin loaded PCL resembled the microstructure of porous polymers that were fabricated by a nonsolvent induced phase inversion process wherein the delayed demixing took place when the NMP exchanged into water and precipitation occurred in the PCL solution [30,31]. The microstructure was also observed to be influenced by the cefazolin content which led to an increase in observed viscosity of the PCL solution similar to the increase in polymer content or additives. This caused the insufficiency solvent and nonsolvent exchange to form pores during phase separation and solidification which tended to decrease the porosity of the samples [30,31]. Pore morphology was also affected by cefazolin content. It was previously reported that several pore structures could be obtained by an immersion precipitation process including unconnected latex, nodules, bicontinuous structures, cellular structures or macrovoids depending on the parameters employed in the process, including, but not limit to, solvent and nonsolvent used, polymer solution composition and casting conditions [31,32]. In this study, a bicontinuous-like structure containing interconnected pores was attained at the lowest cefazolin concentration (PCL\_25) while cellular structure resulted when using a greater cefazolin concentration.

All cefazolin loaded ORC/PCL bilayered composites retained a bilayered structure which consisted of a composite layer and nonporous PCL layer similarly to those of drug unloaded ORC/PCL bilayered composite, P20 [3]. However, the microstructure of the composite side resembled that of P20 only at the lowest cefazolin loading (P20\_25) wherein the viscosity of PCL solution was still comparable and could infiltrate the ORC fabric readily, but not fully cover the surface since the PCL content was low. Using the immediate viscosity solution of P20\_50, a fraction of the solution would still be able to infiltrate and fill the gaps among the fibers of ORC and the surplus part that could not infiltrate would spread and coat on the surface instead which resulted in the smooth coating on the surface. Using high viscosity solution of P20\_75 and P20\_100, limited infiltration was achieved and the majority of the solution would coat on the ORC while preserving the ORC morphology underneath. The coating layer of P20\_100 was observed to be thicker than that of P20\_75. These changes in solution viscosity and microstructure with increasing cefazolin content were in the same fashion as increasing PCL concentration in fabricating ORC/PCL composite as reported previously [3]. Unlike cefazolin loaded PCL, no effect of such an increase in viscosity was noted on the dense PCL side and all formulations similarly displayed nonporous structure. This was thought to be partly due to the solvent absorption by ORC fabric which concentrated the PCL solution and the easier transport of the solvent though the composite side. These could depress the exchange of solvent and nonsolvent at the dense PCL side, resulting in demixing without pore generation.

Cefazolin loaded ORC/PCL bilayered composite exhibited greater densities, but lower thickness, than drug unloaded composite. These changes after cefazolin loading were also observed in cefazolin loaded PCL. These could all be related to the increase in solution viscosity which decreased the swelling ability of the samples during solvent–water exchange in the solidification step and the decrease in coating layer due to the difficulty to achieve the spreading during the infiltration and recoating process. Despite this, the thickness (0.65–0.69 mm) of the composites was still in the same range as those observed in human dura (0.3–0.8 mm) [33], while densities (0.63–0.67 g/cm3) were lower than that of human dura (1.03 g/cm3) [34]. These results suggested that cefazolin loaded composites could still be employed and handled similarly to the drug unloaded ORC/PCL bilayered composite and natural dura mater.

Apart from physical properties, changes in tensile properties were also observed after cefazolin loading. Cefazolin loaded PCL was weaker and less ductile after cefazolin loading since the tensile strength and strain at break decreased with increasing cefazolin concentration, while the tensile modulus was relatively unchanged (Figure 3b,d,f). This was consistent with the previous findings for other types of drug loaded PCL [35–38]. In contrast, tensile strength of cefazolin loaded ORC/PCL bilayered composite was unaffected, but the increase in tensile modulus and the decrease in strain at break with increasing cefazolin concentration were observed instead (Figure 3a,e). Therefore, the underlying mechanism might be different. The changes in tensile properties observed were anticipated to be due to the effect of incorporated cefazolin that could restrict both the deformation of ORC fabric as well as the polymer chain movement of the PCL matrix. In the case of cefazolin loaded PCL, the restriction of PCL chain movement imposed by cefazolin obviously increased the brittleness and decreased the strength of the sample. In the case of ORC/PCL bilayered composite, the reinforcing effect of ORC fabric was relatively greater and might have offset the effect of cefazolin on strength reduction due to PCL chain movement restriction resulting in overall unchanged tensile strength. The restraint of ORC deformation imposed by cefazolin also amplified the stiffness of ORC and gave rise to the increase in tensile modulus. Despite the changes in the tensile properties of all cefazolin loaded ORC/PCL bilayered composites, these tensile values were still within the same range of human dura mater as reported previously, including 11.2–171.5 MPa, 1.3–27.1 MPa and 16.0–49.7% for tensile modulus, tensile strength and strain at break, respectively [34,39,40].

The release profiles of cefazolin loaded PCL were biphasic, comprising the burst release of approximately half of the total cefazolin content and followed by a gradual release for 15 days which was similar to typical burst biphasic release kinetics of drug loaded polymer matrix wherein the initial burst release phase is followed by a power-law phase resulting from Fickian diffusion processes or "anomalous" processes which encompass both diffusion of drugs and swelling of polymers [41,42]. In contrast, the release profile of cefazolin loaded ORC/PCL bilayered composite was found to be mainly burst release where most of the cefazolin was eluted in this phase and followed by relatively constant release thereafter for up to 4–10 days. This is monophasic release kinetics comprising a zero-order release phase that displays a constant rate of drug release via non-Fickian diffusion [42]. The difference in these release profiles between two types of cefazolin loaded samples could be related to the dissimilarity of the composition and microstructure as observed. The cefazolin loaded ORC/PCL bilayered composite contained two layers comprising a composite layer which in situ produced porous channels by relatively faster ORC resorption and the relatively slower degradation of the nonporous PCL layer. These microstructures would aid the diffusion of aCSF solution into the samples, mainly through the composite layer, and caused a rapid dissolution and release of cefazolin resulting in the burst release as observed. Since three different mechanisms of drug release from the polymer matrix were reported, including release of surface loaded drug, diffusion and degradation of the carrier [43,44], the release of antibiotics from dense PCL layer in ORC/PCL bilayered composite could also occur, but contributed less than that of the composite layer. In the case of cefazolin loaded PCL, the diffusion pathway of aCSF into the PCL matrix to dissolve and lead to the release of cefazolin would be more difficult and drug would be slowly released [45]. The initial rapid release of cefazolin loaded PCL might be due to the release of surface loaded cefazolin and part of the drugs which were close to the inner surface of PCL. A subsequent gradual release would be caused by the diffusion and swelling or degradation of PCL. Moreover, both release phases of cefazolin loaded PCL would also be amplified by the open porous microstructure which could act as an additional route for solution diffusion and drug release. Compared to other cefazolin loaded carriers [20–26], cefazolin loaded PCL fabricated in this study exhibited similar biphasic release kinetics to those reported previously, especially when using PCL as a carrier. The rate of elution tended to be faster than that of dense electrospun fiber [21], but comparable to that of PCL sponge which contained microscopic pores [20]. In contrast, ORC/PCL bilayered composite displayed much faster release of cefazolin than cefazolin loaded carriers reported previously and also exhibited different release kinetics. It should be noted that the limitation of using of a UV–Vis spectrophotometer for cefazolin measurement is that it might not specifically detect the cefazolin compared to chromatographic or mass spectrometric methods.

Concerning the MIC of cefazolin against *S. aureus* which was determined to be 0.3150 μg/mL, the duration of the release content of cefazolin from cefazolin loaded ORC/PCL bilayered composite which was still closed to MIC value was approximately 3 days for all samples, except that of P20\_100, which was 10 days. This correlated quite well with the results of antibacterial activity which was found to last for 3–4 days. In the case of cefazolin loaded PCL, the duration for which the released cefazolin concentration was still in MIC range was about 14 days which differed from the antibacterial activity durations of 4–5 days. This could be due to the dissimilarity in the solution saturated condition in the release study compared to the moistened agar in antibacterial activity testing which would affect the cefazolin release, in particular from PCL matrix. Interestingly, it was also observed that a drug unloaded ORC/PCL composite (P20) also exhibited antibacterial activity. This was probably due to the known antibacterial property of ORC that was reported to produce an acidic environment which was not suitable for microorganisms to survive [46,47]. The antibacterial activity of ORC in the composites can be further confirmed by observation that a greater inhibition zone was observed from the rough side of the composite where the knitted ORC fabric was presented compared to that of the smooth PCL side. However, no clear difference in the inhibition zone of the cefazolin loaded ORC/PCL composite between the rough side and smooth side was observed. This was possibly because the antibacterial activity of released cefazolin was much greater than that of ORC and dominated the activity. In addition, the greater transport ability of solution through the rough side might not be advantageous over the smooth side in this contact situation compared to full immersion in the release test.

The burst effect can be regarded as a negative consequence when long-term controlled release applications are needed, but a rapid release or high initial rates of delivery may be the optimal mechanism in certain situations [48]. It should be noted that the purpose of antimicrobial prophylaxis is to achieve drug levels in serum and tissue that surpass the MIC of the bacterial organisms likely to be faced during operation and prophylaxis after wound close is unessential [49–51]. Therefore, the burst release of cefazolin loaded ORC/PCL bilayered composite, which could be a problem in some applications which require a sustained release of drug for a long period of time, might not be relevant for an antibacterial dural substitute, which requires a localized release of high concentration of antibiotic prophylaxis to ensure that any bacterial infections (if any) are totally and immediately eliminated at the site of implantation. Additionally, the burst release of a high concentration of cefazolin observed in the present study could be an advantage as this could reduce the risk of antibiotic resistance.
