Processing and characterization of aligned electrospun gelatin/polycaprolactone nanofiber mats incorporating borate glass (13-93B3) microparticles

Aligned biodegradable fibers incorporating bioactive glass particles are being highly investigated for tissue engineering applications. In this study, 5, 7 and 10 wt% melt-derived 1393B3 borate glass (BG) microparticles (average size: 3.15 µm) were incorporated in 83 wt% polycaprolactone (PCL) and 17 wt% gelatin (GEL) (83PCL/17GEL) solutions to produce aligned electrospun composite nanofiber mats. Addition of 5 wt% BG particles significantly increased the alignment of the nanofibers. However, further incorporation of BG particles led to reduced degree of alignment, likely due to an increase of viscosity. Mechanical tests indicated a tensile modulus and tensile strength of approximately 51 MPa and 3.4 MPa, respectively, for 5 wt% addition of 1393B3 BG microparticles, values considered suitable for soft tissue engineering applications. However, with the increasing amount of 1393B3 BG, the nanofiber mats became brittle. Contact angle was reduced after the addition of 5 wt% of 1393B3 BG particles from ∼45° to ∼39°. Cell culture studies with normal human dermal fibroblast (NHDF) cells indicated that 5 wt% 1393B3 BG incorporated nanofiber mats were cytocompatible whereas higher doping with 1393B3 BGs reduced biocompatibility. Overall, 5 wt% 1393B3 BG doped PCL/GEL nanofiber mats were aligned with high biocompatibility exhibiting desirable mechanical properties for soft tissue engineering, which indicates their potential for applications requiring aligned nanofibers, such as peripheral neural regeneration.


Introduction
Electrospinning was first introduced by Formhals in 1934 and the technique has been finding a wide range of applications, specially in the last decades [1]. It is a commonly practiced inexpensive and simple approach to produce fibrous scaffolds for tissue engineering [2][3][4]. The technique involves the application of a high intensity electric field (5-15 kV) between a negatively charged fiber collector and a needle tip connected to a syringe pump [4][5][6][7]. Usually, an electrically charged polymeric solution is employed, which forms a 'Taylor cone' at the needle tip. The electric field overcomes the surface tension at the 'Taylor cone' which leads to the ejection of a polymer jet. The solvent evaporates and the polymer jet stretches and accelerates forming a fiber on the collector [4]. Chloroform, hexaflouro-2-propanol, dichloromethane, and dimethylformamide are the common solvents used, however due to their toxicity such solvents have started to be replaced by so-called benign solvents, such as acetone, formic acid and acetic acid, specially for electrospinning biopolymers [8,9]. Electrospinning process is affected by factors including polymer solution properties as well as environmental and process parameters [6,9]. Many factors affect the diameter of the fibers and distribution of fiber diameters. For example, research indicated that increase of polymer content, which ultimately increases viscosity, leads to increase of fiber diameter [10]. Humidity also affects fiber diameter as this changes the rate of solvent evaporation [11]. High humidity was also indicated to reduce mechanical properties due to inadequate mechanical bonding between fibers [12]. Moreover, solely use of acetic acid as the solvent phase led to large fiber size distribution, whereas the addition of formic acid narrowed down the distribution of fiber diameter. This was due to the lower ionic conductivity of acetic acid compared to formic acid [13].
Poly(caprolactone) (PCL) is a semi-crystalline, non-toxic, biocompatible and degradable polymer. PCL has a low degradation rate in vivo, therefore it is used when longer healing time is required [3]. Stability of PCL enables structural support for cell growth and proliferation in vivo [10]. Gelatin (GEL) is a biocompatible and biodegradable natural polymer which leads to cell adhesion and proliferation. However, it has poor mechanical properties. Therefore, by combining PCL and GEL, both desired physical and mechanical properties can be achieved for tissue engineering applications [14,15].
Since their discovery in the late 1960s, bioactive glasses have attracted research attention for biomedical applications [16][17][18]. The most recent member of the bioactive glass family is borate glass (BG) [19]. BGs with certain compositions are osteoconductive, bioactive, and degradable [20]. Melt-derived 13-93B3 BG (54B 2 O 3 -22CaO-8K 2 O-8MgO-6Na 2 O-2% P 2 O 5 in mol %) is the most investigated BG to date [21][22][23]. As well as for hard tissue applications, 13-93B3 BGs have started to be considered for soft tissue engineering applications with very promising results, for example in skin wound healing [24,25]. For instance a commercial product based on 13-93B3 BG fibers has been shown to be successful in healing diabetic wounds which are usually difficult to heal [25]. In many studies, usually BGs in particulate form are incorporated in a polymeric matrix, and the composite solution is electrospun into nanofiber mats [26][27][28]. Indeed, the incorporation of BG nano and microparticles in polymers for electrospinning has found considerable interest in recent years, as reviewed recently [29].
Aligned fibers are usually produced by using a rotating mandrel for various tissue engineering applications [4]. The degree of alignment gradually increases with the increase of rotation speed of the mandrel [30]. Research has indicated that aligned fibers usually lead to increase of cell migration and proliferation for wound closure in comparison to random fibers [31,32]. Nerve regeneration is another promising application of aligned electrospun fibers [33]. Aligned electrospun fibers can act as neural guidance conduits for regeneration of damaged peripheral neurons. This enables diffusion of neurotrophic factors and axonal growth [34]. PCL nanofibers can enhance differentiation of stem cells into neurons promoting neural outgrowth [35]. Several studies indicated that incorporation of GEL in PCL nanofibers can further improve neural growth [4,36].
The ability of BGs to stimulate neural regeneration has been investigated in a few studies [37]. Marquardt et al [27] investigated the survival of embryonic chick dorsal root ganglia (∼80% glia, ∼20% neurons and fibroblasts) on aligned 13-93B3/fibrin microfibers. According to the study, the addition of 13-93B3 BG enhanced % cell viability. Moreover, neural extensions were observed, suggesting the utility of 13-93B3 BG for applications involving neural tissue engineering. For potential neural regeneration applications, Gupta et al [38] added 50 wt% of 13-93B3 particles in PCL to produce films. Additionally, 13-93B3 particles were doped with 1 wt% of various ions including cerium, gallium, iron, zinc and iodine, and films were produced incorporating doped 13-93B3. 13-93B3 particles with 1 wt% of gallium, iron or zinc ion doping improved neurite outgrowth from dorsal root ganglia. The focus of the mentioned study was to observe the effect of doped 13-93B3 for neurite outgrowth, but alignment of fibers is critical to mimic peripheral tissue and stimulate its regeneration. In this study, for the first time, different concentrations of 13-93B3 BG microparticles are incorporated in PCL/GEL solution to prepare aligned electrospun nanofiber mats, and the fabrication and characterization of such novel fibers are reported.

Experimental section
Synthesis of 13-93B3 bioactive glasses: 13-93B3 BG microparticles were produced by conventional meltquenching technique [9]. Na, K and Ca were introduced as carbonates, B as boron oxide (B 2 O 3 ) and P as ammonium dihydrogen phosphate (H 6 NO 4 P). The carbonate and oxide powders were dried before weighting. Then, the mixed powders were first decarbonated before being molten twice at 1050 • C followed by quenching by immersion of the bottom of the platinum crucible in water. Table 1 shows the nominal composition of the produced BG.
Ball milling and sieving: 13-93B3 BG granules were crushed using a Jaw Crusher (Retsch, Haan, Germany) and planetary ball milled (Retsch Haan, PM100) for 30 min. The microparticles were sieved with a mesh size of 65 µm sieve (Retsch Haan, Germany) for 1 h. After mixing the solution overnight, the solution was fed at 0.3 ml h −1 through a 23 G needle. Nanofibrous mats were prepared by using commercial electrospinning equipment (EC-CLI, IME Technologies Netherlands). Nitrogen flux was set at 8 ml min −1 , the temperature and relative humidity were kept as 23 • C and 40%, respectively. The target was covered with an aluminum foil. Then, the electrospinning process was conducted by keeping the distance between the target and the needle fixed at 11 cm. The applied voltage was 14 kV at the needle and −2 kV at the rotating mandrel (1500 rpm). After electrospinning, the aluminum foil covered with the nanofiber mats was carefully removed from the collector for analysis. Typical images of the produced fiber mats are shown in figure S3.

Characterization studies
Scanning electron microscopy (SEM) was used to analyze the morphology of the 13-93B3 BG particles, and the nanofiber mats (Auriga SEM instrument, Zeiss, Oberkochen, Germany). Then using Image-J software, the particle sizes and nanofiber sizes were determined for each study group. Fiji software was used to determine the degree of orientation of the nanofibers. The alignment degree was measured by calculating the % of fibers oriented between +10 • and −10 • along the main direction.
Energy dispersive spectroscopy (Oxford Instruments, Abingdon, UK) was used to assess the distribution of 13-93B3 BG microparticles in the nanofiber mats.
Inductively coupled plasma (ICP) optical emission spectroscopy was used to obtain the composition of the 13-93B3 BG batches. 13-93B3 BG was first dissolved for more than 24 h in HF solution then HNO 3 was added. In case of B analysis, 13-93B3 BG powder was first dissolved in Na 2 CO 3 and then the total product was further dissolved in HNO 3 . Then, measurement for wt% of each component was taken by using ICP-OES analyzer Spectro Genesis from Ametek.
Structural analysis was performed by Fourier transform infrared spectroscopy (FTIR; IRAffinity-IS, Shimadzu) in attenuated total reflectance mode, using a wavenumber range of 4000-400 cm −1 with a resolution of 4 cm −1 with 32 spectral scans for the 13-93B3 BG particles and the nanofiber mats.
X-ray diffraction (XRD) patterns of 13-93B3 BGs and nanofiber mats were obtained by using X-ray diffractometer (Miniflex, 600, Rigaku) equipped with Cu Kα radiation in the 2θ range of 10-80 • both for BGs and fiber mats. A step size of 0.02 • and dwell time of 1 • per minute were used.
Static contact angle was measured using a drop shape analyzer DSA30 (Kruss GmbH) to determine wettability of fiber mats. A snapshot of the water droplet was taken after 5 s to determine the water contact angle for 5 samples.
For mechanical testing, samples were cut into rectangular shapes with 4 cm length and 5 mm width. Uniaxial tensile strength was measured using a universal testing machine (5960 Dual Column Tabletop Testing System, Instron®, Darmstadt, Germany) at room temperature. Crosshead speed of 10 mm min −1 using a 50 N load cell was used to carry out measurements with ten replicates for each study group. Then stress-strain curves were drawn, and tensile modulus and tensile strength were determined for all study groups.
Degradation studies were carried out in phosphate buffer saline (PBS) solution for 120 h at 37 • C using scaffold holders (CellCrownTM 24, Scaffdex, Sigma Aldrich, Germany). pH measurements were taken each day for three replicates.
Cell culture studies were carried out with primary normal human dermal fibroblasts (NHDF) on PCL/GEL, 5BG/PCL/GEL, 7BG/PCL/GEL, and 10BG/PCL/GEL nanofiber mats. NHDF cells (Promocell, Germany) were obtained from Translation Research Center of Friedrich-Alexander-University Erlangen-Nuremberg. The cells were cultured in DMEM supplemented with 10% (FBS) and 1% penicillin/streptomycin solutions in 75 cm 2 cell culture flasks. The nanofiber mats were fixed on CellCrownTM-24 inserts in 24 well-plates and sterilized by ultraviolet light irradiation for 30 min on both sides. 50 000 cells in 100 µl were drop seeded on the scaffolds. The cells were absorbed by the film for around 30 min. Then, 900 µl of total growth medium was added onto the fiber mats by covering the film in each well and incubated at 37 • C in a humidified CO 2 incubator. After culturing the NHDF cells on the fiber mats for 48 h, 5 vol % WST-8 reagent (CCK-8 kit, Sigma-Aldrich) in cell culture medium was added and incubated for three hours at 37 • C. After this, absorbance of the obtained dye was measured at 450 nm using a spectrophotometric plate reader (PHOmo, anthos Mikrosysteme GmbH, Germany).

Statistics
Quantitative data are reported as mean value ± standard error from at least three independent experiments. Statistical differences between groups were analyzed using the two-way analysis of variance statistical test, with Tukey's pairwise post hoc test. Statistical significance is represented as # p < 0.05 (in comparison to the control group).

Results and discussion
In this study, first the produced 13-93B3 BG powder was characterized before production of the fibers. Firstly, the chemical composition of the granules was determined by ICP-optical emission spectroscopy. This is shown in table 3. Table 3 demonstrates that the actual composition of the BG granules was very similar to the nominal composition of 13-93B3 BG (table 1). After crushing, ball milling and sieving the granules into microparticles, the morphology of the 13-93B3 BG particles was examined by using SEM. Then, the size of the particles was measured in the Image-J software for over 55 particles from up to five SEM images. Figure 1 shows the (a) morphology and (a) size distribution of the 13-93B3 BG particles.
As shown in figure 1, 13-93B3 BG particles were mainly micron-sized, with an average size of ∼3. 15 µm. The highest frequency of the particle size was between 1.25 and 2.5 µm. Figures 2(a) and (b) show the XRD pattern and the FTIR spectrum of 13-93B3 BG particles, respectively.
In figure 2(a), the XRD pattern demonstrates the amorphous structure of 13-93B3 BG [16] confirming that the production method enabled obtaining BGs without crystallization. In figure 2(b), the resonance at around 720 cm −1 was due to the bending mode of the BO 3 group. The resonance at 1370 cm −1 was for B-O stretching mode of BO 3 group and the resonance at 970 cm −1 was assigned to the B-O stretching mode of the BO 4 group [39,40]. After obtaining phase pure 13-93B3 BGs, the BG particles were incorporated in PCL/GEL solution in 5, 7 and 10 wt% and the electrospinning process was applied. Figure 3 shows SEM images of the electrospun fibers with different loadings of 13-93B3 BG particles. Then, the fiber size was measured by using the Image-J software for approximately 55 nanofibers from several SEM images.
According to figure 3, the produced fibers were bead free. With the increase of 13-93BG BG particle content to 10 wt%, the fiber diameter drastically increased by 700%. This rise is related to increase of solution concentration which ultimately increases viscosity with the increase of wt% of 13-93B3 BG particles [41]. Viscosity is the dominant property determining the fiber diameter [42]. If the viscosity is too high or too low, electrospinning cannot take place [43,44]. The increase of viscosity increases fiber diameter [45,46]. The drastic increase of fiber diameter with the increase of 13-93B3 BG loading to 7 wt% is indicated in figure 4. Additionally, fiber size distribution also increased with the increase of 13-93B3 BG content. Moreover, the nanofibers had homogeneous distribution of 13-93B3 BG particles within the nanofiber mats, which is demonstrated in figures S1 and S2.
Jet elongation may direct the orientation of the polymer chains along the fiber axis [1,47]. In this work, rotating collectors were used to obtained aligned fibers [48]. By using SEM images, Fiji software was utilized to determine the degree of alignment of the fibers. Figure 5(a) shows the orientation of the fibers for different compositions and their % alignment is illustrated in figure 5(b).
According to figures 5(a) and (b), the alignment of the fibers increased by 25% with incorporation of 5 wt% 13-93B3 BG particles into the fibers for the same processing conditions. This is probably because of increase of conductivity with the incorporation of 13-93B3 BG [49][50][51][52]. Similarly, Hu et al [50] obtained aligned fibers after addition of multi-walled carbon nanotubes into polymers (due to increase of electrical conductivity). However, in this study, the % alignment decreased with the increase of wt% of 13-93B3 BG particles in the solution, which is probably because of increase of viscosity, which in turn reduced   the jet elongation. It is reported also in the literature that high viscosity and low electrical conductivity reduce alignment of fibers [53]. Figure 6 shows the FTIR spectra and XRD patterns of the fibers.
In figure 6(a), FTIR spectra demonstrate peaks for the fibers. PCL peaks found at 1727, 1240 and 1169 cm −1 are related to carbonyl stretching, asymmetric and symmetric stretching of C-O-C bonds, respectively [54]. The peak at 1360 cm −1 is also related to stretching vibration of C-O-C [55]. GEL peaks were at 1651 cm −1 and 1514 cm −1 for amide I and N-H deformation of amide II, respectively [8,36,56]. 13-93B3 BG peaks are not detectable probably due to their low amount in the fibers [8,56]. Figure 6(b) represents XRD patterns for PCL/GEL fibers as well as 13-93B3 BG incorporated fibers. The two sharp peaks at 2θ = 21.3 • and 23.6 • are belonging to PCL. Again, 13-93B3 BGs showed no peaks due to its low amount [57]. Despite this, from EDX analysis in figures S1 and S2, incorporation of 13-93B3 BG particles in the nanofiber mats can be clearly observed. Figure 7 shows the mechanical test results of the fiber mats.
Mechanical properties of the scaffolds should be suitable to support cells therefore it is important to match the mechanical properties of the native tissue to enable cell growth and proliferation [10]. Neural tissue has an ultimate tensile strength and % strain of approximately 12 MPa and 40%, respectively [58]. In this study, the mechanical properties of 5BG/PCL/GEL fiber mats were found to be suitable for neural regeneration applications. Indeed, incorporation of 5 wt% 1393B3 BG in the PCL/GEL fibers led to an increase of the tensile modulus and tensile strength of the fiber mats. In fact, the tensile strength increased 8.5 times compared to the control group. The mechanical properties of fiber mats are usually enhanced when fibers are aligned [59]. As can be seen from SEM images in figure 3(b), fiber alignment and homogeneous distribution of BG particles in 5 wt% 1393B3 BG PCL/GEL fibers led to improved mechanical performance. The results indicate that tensile modulus increased gradually with the addition of 1393B3 BG up to 7 wt%. However, the nanofiber mats became gradually more brittle with increasing 1393B3 BG wt.%. SEM images also indicate a more pronounced random distribution of fibers with increasing 1393B3 BG wt.%, which possibly reduced the mechanical properties of the fiber mats. Moreover, due to high BG particle loading, agglomeration might have occurred which may lead to stress concentration points and reduced mechanical performance [60]. Due to their relatively higher tensile strength and modulus, fibers containing 5 wt% 1393B3 BG were found to be the optimum in terms of fiber mat mechanical performance. The modulus of 1393B3 BG incorporated fibers was in the range of 50-110 MPa and the tensile strength was between 1.2 and 2 MPa. These values are suitable for soft tissue engineering applications [10]. Figure 8 shows the results of contact angle measurements of the fiber mats.
Cell behavior is strongly affected by surface wettability and the optimal water contact angle is reported to be between 40 • and 70 • . Both too high  hydrophilicity and hydrophobicity are not favorable for cell adhesion [61]. Contact angle measurements indicate that contact angle was reduced gradually with the increase of wt% 13-93B3 BG. A similar trend was observed by Tamjid et al [62] for electrospun bioactive silicate glass (45SiO 2 -24.5Na 2 O-24.5CaO-6P 2 O 5 )/PCL fibers. These values were found to be in the lower range of ideal contact angle values for cell adhesion and growth. Figure 9 shows (a) the %weight loss of the samples, (b) the results of pH measurements and (c) the absorbance (O.D.) measurement from the WST-8 assay on the fiber mats, respectively. Figure 9(a) indicates that samples did not show degradation within a period of 5 d. Figure 9(b) illustrates that all samples had a pH between 7.1 and 7.4 which is the physiological pH of tissues, therefore no toxic affect is expected from the scaffolds due to pH.  studies with NHDF cells on the fiber mats. According to the cell culture studies, there was no significant difference of % cell viability between the control, PCL/GEL fiber mats and 5BG/PCL/GEL fiber mats. However, further incorporation of 13-93B3 BG into the PCL/GEL fibers significantly reduced the absorbance values for NHDF cells. This result indicates that higher concentrations of 13-93B3 BG were relatively toxic. This result may be due to excessive hydrophilicity of the samples as BG increases, as shown in figure 8. The result may be also related to toxic levels of boron released when the content of BG increases. The concentration of released B (and other ions) must be investigated by ICP in the future. Although 5BG/PCL/GEL fibers are found to be cytocompatible in this preliminary assessment, cell culture studies  must be carried out with neural cells to be able to determine the applicability of these fibers for neural regeneration applications.
Overall, the results indicated that the 5BG/PCL/GEL composition was suitable to produce aligned fibers with high biocompatibility. In the future, lower wt% of BG particles can be incorporated in fiber mats to investigate their influence on fiber alignment. Moreover, it would be beneficial to measure the viscosity of the initial solution and its conductivity to fully understand the influence of 13-93B3 BG particles on nanofiber alignment in depth. In the future, these fiber mats may be studied for neural regeneration applications considering their suitable mechanical properties and cytocompatibility.

Conclusion
In this study, 13-93B3 BG particles were prepared and incorporated in PCL/GEL-based solution to produce aligned electrospun fiber mats. With incorporation of 5 wt% BG in the PCL/GEL solution, the fiber alignment significantly increased. This is likely related to the increase of conductivity of the polymeric solution which improves fiber alignment. However, further incorporation of 13-93B3 BG particles reduced the alignment and increased fiber diameter, which was probably due to the increase of viscosity. Moreover, the addition of 5 wt% BG improved the mechanical properties, which is possibly related to the alignment of the fibers. All loadings of 13-93B3 BG in PCL/GEL fibers significantly increased their hydrophilicity. Additionally, PCL/GEL fiber mats containing 5 wt% 13-93B3 BG were cytocompatible whereas higher loading of 13-93B3 BG particles reduced biocompatibility significantly. Therefore, optimized fiber mats were chosen with 5 wt% of 13-93B3 BG particle incorporation. In future, lower loadings of 13-93B3 BG particles may be considered to examine their effect of (nano)fiber alignment. The results of this work demonstrate that the produced PCL/GEL fiber mats incorporating 5 wt% 13-93B3 BG are promising candidates for applications requiring aligned fibers, such as neural regeneration.

Data availability statement
The data cannot be made publicly available upon publication because no suitable repository exists for hosting data in this field of study. The data that support the findings of this study are available upon request from the authors.