1. Introduction
Numerous applications of natural-origin polymers in the biomedical field (e.g., drug and cell carriers) are focusing the attention of researchers [
1]. Polyhydroxyalkanoates (PHAs) are a large sub-branch of natural polyesters that can be extracted from bacteria or genetically modified plants. The poly(3-hydroxybutyrate) (PHB) homopolymer and the poly(3-hydroxybutyrate-
co-3- hydroxyvalerate) (PHBV) copolymer are the members of the PHAs family with the highest applications. The main problem of PHB concerns to its high brittleness and crystallinity and therefore copolymers incorporating small percentages of 3-hydroxyvalerate (HV) units (i.e., PHBV copolymers) are being commercialized since they can also be easily produced by bacteria (e.g.,
Escherichia coli,
Paracoccus denitrificans,
Ralstonia eutropha) as storage products.
Due to properties such as biocompatibility, biodegradability, non-toxicity, and piezoelectricity, the use of PHBV copolymers in a variety of medical fields including surgical sutures, wound dressings, controlled release, and tissue engineering has been reported [
2,
3,
4,
5,
6]. Chemical and mechanical properties of PHBV copolymers can logically be controlled in function of the HV content. Despite relatively high crystallinity levels can be achieved at various HV ratios, it is obvious that the increase of the comonomer HV content lead to polymers with lower degrees of crystallinity and melting temperatures. In addition, besides exhibiting full degradability in aqueous environments and producing non-toxic by-products, the degradation rate of copolymers can be tuned by varying the HV content [
7,
8,
9].
The thermally induced phase separation (TIPS) technique has extensively been used in non-biomedical fields for fabricating synthetic membranes. Applications in the biomedical sector are also habitual as for example for the development of drug delivery systems. Specifically, the methodology has been employed to prepare microspheres incorporating pharmaceutical and biological agents [
10,
11]. Today, TIPS is a common technique to fabricate porous scaffolds for tissue engineering applications [
12,
13,
14,
15]. This method is based upon thermodynamic demixing of a homogeneous polymer solution into polymer-rich and polymer-lean (solvent-rich) phases [
16]. The solvent in the polymer-lean phase can subsequently be eliminated by extraction, evaporation, or sublimation [
17], leaving behind a highly porous polymer network [
18,
19]. TIPS experimentally allows controlling the final structure of the scaffold in terms of morphology, average pore size and degree of interconnection [
20]. The final structure and pore morphology of the phase-separated polymer matrices are greatly dependent on the combination of the selected polymer and solvent system, the polymer concentration, the phase-separation temperature and the temperature gradient applied to the polymer solution [
21].
Various biodegradable polymers have been considered to fabricate three-dimensional scaffolds through the TIPS technique and investigated for tissue regeneration applications [
22]. In this regard, the application of the phase separation method for scaffolding purposes has been reported for several biodegradable polyesters, especially polylactide (PLA) and poly(lactide-
co-glycolide) (PLGA). Depending on the polymer system and phase separation conditions, these 3D polyester scaffolds can structurally be classified as solid-walled isotropic and anisotropic (like microtubular), fibrous, nanofibrous, and platelet-like architectures [
23,
24,
25,
26,
27]. Moreover, the different types of TIPS techniques—i.e., solid–liquid [
23,
24,
25], liquid–liquid [
26,
27], and crystallization-induced phase separations [
27]—have been used for creating different micro- and nano-structured polymer constructs. Organic solvents with high freezing points like 1,4 dioxane or benzene and others with low freezing points like THF, DMF, and pyridine have successfully been used to fabricate scaffolds by solid–liquid and liquid–liquid phase separations, respectively [
23,
24,
25,
26,
27].
The phase separation procedure has also been applied to different scaffolding materials based on polyhydroxyalkanoates. Thus, the fabrication of nanofibrous and microtubular architectures have mainly been reported for systems based on PHB, poly(3-hydroxybutyrate-
co-3-hydroxyhexanoate) (PHBHx) and poly(3-hydroxybutyrate-
co-4-hydroxybutyrate) (P(3HB-4HB)) [
28,
29]. Nevertheless, the fabrication of phase-separated porous scaffolds made of PHBV copolymers, has received less attention. Furthermore, scarce studies can be found evaluating TIPS-obtained PHBV scaffolds in terms of pore morphology and paying attention to the copolymer properties and the phase separation conditions.
In the present study, the potential of PHBV copolymers and TIPS technique to develop interconnected 3D networks with solid-wall and platelet-like structures has been appraised. We have specifically addressed how altering the quenching temperatures and copolymer characteristics have a considerable effect on the phase separation process and the scaffold properties. The disparities observed in the morphological features and mechanical properties of resulting scaffolds were discussed with respect to thermodynamic and kinetic conditions of phase separation.
1,4-dioxane is used as a solvent for a variety of practical applications and can be found for example at minimum levels in cosmetics and personal care products. Some toxicologic effects of dioxane have been recognized, being consequently a potential health concern that received the attention of FDA. Despite no specific law requirements have been formulated, manufacturers have been encouraged to remove dioxane from technological processes [
30]. Therefore, we have paid also special attention, through biocompatibility tests, to ensure a complete solvent removal in the final processed scaffolds.
2. Materials and Methods
2.1. Materials and Scaffold Preparation
Poly(3-hydroxybutyrate-co-3-hydroxyvalerate)s containing 5 and 12 molar percentages of 3-hydroxyvalerate were purchased from Sigma-Aldrich (St. Louis, MO, USA) and were used for fabrication of polymeric scaffolds. The molecular weights of PHBV(5%HV) and PHBV(12%HV) were 320 kDa and 240 kDa and their corresponding melting points were 165 °C and 161 °C, respectively. 1,4-dioxane was used as solvent and was supplied by Acros Organics (Morris Plains, NJ, USA), with linear formula of C4H8O2, molecular weight of 88.11 g/mol, density of 1.033 g/mL, melting point of 12 °C, and purity of 99.5%. The polymers and the solvent were used without further purification.
For preparing polymeric foams, the corresponding polymer solutions were cooled until phase separation occurred. Subsequently, porous structures were achieved after removing the solvent. Specifically, both PHBV(5%HV) and PHBV(12%HV) copolymers were dissolved at a concentration of 2% (w/v) in 1,4-dioxane by heating and stirring. When the temperature reached about 70 °C, a clear homogenous polymer solution was attained. The solutions (0.4 mL) were poured into a cylinder-shaped glass container with a diameter of 14 mm and height of 40 mm and then sealed. The samples first were cooled spontaneously to the room temperature and then immediately incorporated to the corresponding cooling devices preset on −5 °C or −25 °C. In this process, phase separation occurred during cooling. Note that this process occurred under two different thermal gradients, that is, from room temperature to −5 °C and −25 °C. Afterwards, samples were kept at rest for 24 h at the selected final temperature. Finally, the samples were lyophilized (Gamma 2-16 LSC, Martin Christ, Osterode am Harz, Germany) for 40 h. The resulting porous scaffolds were dried in a vacuum oven at room temperature to reach a constant weight.
2.2. Cloud Point and Cooling Rate Determination and DSC Analysis of Polymer Solutions
The cloud point of polymer solutions was evaluated by visual turbidimetry. In order to predict the location of binodal curve at relatively low concentrations of the PHBV-dioxane phase diagram, the cloud points of 1–10% (w/v) solutions were determined. The solutions were poured into transparent sealed glass containers and then transferred to a refrigerated incubator to reach equilibrium conditions through a controllable slow cooling. The incubator was preset at 35 °C and programmed to be cooled at a rate of 0.033 °C/min (i.e., 1 °C each 30 min). The temperature at which the clear solution became turbid was recognized as the cloud point. At least three independent turbidimetric assays per sample were performed, being the results averaged and the standard deviations obtained.
Cooling rate of the different polymer solutions was determined using a digital thermometer (ESCORT 20 T/C, EIC, Taipei, Taiwan) inserted into the center of the tube containing the respective solutions. The thermometer was connected to the computer and plotted the cooling diagrams (temperature versus time) while the samples were cooling from room temperature to −5 °C or −25 °C. Cooling rates were determined from the slopes of the corresponding curves. In fact, three experiments were performed for each condition and the values of the resulting slopes averaged and taken as the cooling rates to be considered in the subsequent differential scanning calorimetry (DSC) analyses. In this way, cooling rates of 2 °C/min and 6 °C/min were obtained, when solutions were cooled from room temperature to −5 °C and −25 °C, respectively.
In the next step, a differential scanning calorimeter (200 F3, NETZSCH DSC, Selb, Germany) was used to evaluate the crystallization behavior of pure solvent and the polymer solutions. The DSC analyses were carried out by cooling from +40 °C to −50 °C at rates of 2 °C/min and 6 °C/min. The temperature, at which an exothermic peak appeared throughout cooling, was taken into account as the solvent crystallization temperature. The studied samples are summarized in
Table 1, with abbreviations according to the HV content and the cooling rates.
2.3. SEM and DMTA Analyses of Polymer Scaffolds
A scanning electron microscope (SEM) (VEGA II, TESCAN, Brno, Czech Republic) was used to study the porous structure of PHBV scaffolds. The microstructural features were evaluated from the outer surface and the transverse cross-section of the scaffolds. These cross-sections were obtained by soaking the scaffolds in liquid nitrogen for 2 h before to split them in two parts. Prior to microscopy, the samples were sputter-coated with a thin layer of gold by using a Mitec K950 Sputter Coater (Quorum Technologies Ltd., Ashford, UK).
A dynamic mechanical thermal analyzer (DMTA) (TRITEC DMA 2000, DMA-TRITON, Lincolnshire, UK) was used to estimate the viscoelastic behavior of PHBV scaffolds under dynamic loading conditions (ASTM E1640-04). Polymeric scaffolds (2 cm × 0.7 cm × 0.2 cm) were subjected to cyclic tensile strains of 0.008 mm with frequency of 1 Hz, while temperature was increased from −50 °C to 180 °C at a rate of 5 °C/min. The stress response of samples was recorded via in-phase modulus (E′), lag modulus (E˝), and loss tangent (E˝/E′) versus temperature.
2.4. Assays of Cell Adhesion and Proliferation
MDCK cells (with epithelial-like morphology and derived from Madin–Darby Canine Kidney, ATCC) and NRK cells (with epithelial-like morphology and derived from the kidney of the Rattus norvegicus, ATCC) were employed. Both cell lines grow adherently, and were cultured in Dulbecco’s modified Eagle’s medium (DMEM with 4500 mg/L of glucose, 110 mg/L of sodium pyruvate and 2 mM of l-glutamine) supplemented with 10% fetal bovine serum (FBS), 50 U/mL penicillin, 50 mg/mL streptomycin, and l-glutamine 2 mM at 37 °C in a 10% humidified atmosphere of 5% CO2 and 95% air. Culture media were changed every two days. For sub-culture, cell monolayers were rinsed with PBS and detached by incubating them with 0.25% trypsin/EDTA for 2–5 min at 37 °C. The incubation was stopped by resuspending in 5 mL of fresh medium and the cell concentration was determined by counting with Neubauer camera and using 4% trypan blue as dye vital.
HBV5 and HBV12 scaffolds were cut off into pieces of 1 cm × 1 cm. These samples were placed in tissue culture plates of 24-wells and fixed to bottom plate with a small drop of silicone (Silbione® Med Adh 4300 RTV, Bluestar Silicones France SAS, Lyon, France), sterilized by exposed to UV light for 15 min. 100 µL containing 5 × 104 cells/well to assess cell adhesion, and 2 × 104 cells/well for the cell proliferation assay were seeded in each well and incubated for 60 min to allow cell attachment to the material surface. Then, 1 mL of culture medium was added to each well. Quantification of viable cells was performed after 24 h and 7 days to evaluate the cellular adhesion and proliferation, respectively. The control was performed by cell culture on the plate without any material.
The percentage of cells adhered and proliferated was determined through the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay [
31]. After 24 h or 7 days, 50 µL of MTT (3 mg/mL) were added to each well in the plates and incubated for 4 h. After that, samples were washed twice with PBS and the specimens deposited in a new plate. 1 mL of dimethyl sulfoxide (DMSO) was subsequently added and the absorbance was measured at 570 nm in a microplate reader (Biochrom EZ-Read 400, Cambridge, UK) after 15 min of gentle stirring. Three replicas were evaluated and the corresponding values were averaged and graphically represented. The statistical analysis was performed by one-way ANOVA to compare the average values of all groups; Tukey-test was then applied to determine a statistically significant difference between two studied groups. The tests were performed with a confidence level of 95% (
p < 0.05).
Samples were fixed overnight with 2.5% formaldehyde in PBS at 4 °C, and then washed five times with PBS to obtain images showing the morphology of cells coming from adhesion and proliferation assays. Samples were also stained to get fluorescence microscopy images. Specifically, actin was labeled with green-fluorescent Alexa Fluor Atto-488 phalloidin dye, and the nucleus was labeled with DAPI (4′,6-diamidino-2-phenylindole). Then, samples were observed using a confocal laser scanning microscope (LSM 900 Zeiss, Oberkochen, Germany), images were taken with a camera controlled by ZEN 2.6 software (blue edition) (Carl-Zeiss Microscopy GmbH, Jena, Germany).
3. Results
3.1. DSC Testing of Polymer Solutions
DSC results (
Figure 1a,b) revealed that crystallization temperature of 1,4-dioxane (
Tc of solvent) in the polymer solutions was higher than the pure solvent at the two assayed cooling rates (i.e., 2 °C/min and 6 °C/min).
Additionally, it was observed that the crystallization temperature of 1,4-dioxane was higher at both cooling rates when the copolymer was enriched in HV units (i.e., PHBV(12%HV) solutions gave rise to a higher solvent crystallization temperature than PHBV(5%HV) solutions). DSC cooling runs showed large exothermic peaks associated with the crystallization of 1,4-dioxane, and small peaks related to the well-known reversible phase transition of 1,4-dioxane from its monoclinic phase I to the monoclinic phase II [
32].
Neither the large nor the small peaks observed in each cooling trace of the studied polymer solutions could be attributed to a crystallization of the polymer. Note that pure 1,4 dioxane exhibited both mentioned peaks at the same temperature range. According to the values of 1,4 dioxane crystallization temperature in the samples (i.e., −4.3 °C and −4.6 °C for DXN-R2 and DXN-R6; −3.4 °C and −3.9 °C for HBV5-R2 and HBV5-R6; 1.8 °C and −1.5 °C for HBV12-R2 and HBV12-R6), a decrease in Tc of solvent was detected for all the samples without exception, when a higher cooling rate was applied, a feature that was more significant for the copolymer enriched in HV units.
3.2. Cloud Point of Polymer Solutions
Cloud point is a temperature at which a clear polymer solution becomes turbid during cooling because of the liquid–liquid phase separation [
26]. The boundary of the liquid–liquid demixing region in the polymer-solvent phase diagram is usually named binodal curve, but the term “cloud point curve” is more appropriate for polydisperse polymers [
17].
Figure 2 shows the variation of the cloud point as a function of polymer concentration for the binary systems of PHBV(5%HV)-1,4 dioxane and PHBV(12%HV)-1,4 dioxane.
According to the experimental cloud point curves, a higher cloud point was observed as the polymer concentration increased (at the evaluated concentration range). Additionally, the temperature at which the solution became cloudy decreased with the increase in HV content in the copolymer. Specifically, the cloud point of 2% (w/v) solutions of PHBV(5%HV) and PHBV(12%HV), which were used for fabrication of the scaffolds, was 9.3 ± 1.1 °C and 2.7 ± 1.5 °C, respectively. Any trace of cloudy state was not seen in the 1% (w/v) solution of the PHBV(12%HV) before being frozen. Gelation was also observed to occur before to achieve a cloudy state when concentrated solutions (e.g., higher than 2% (w/v)) were slowly cooled, especially for the copolymer with lower HV content.
3.3. Morphology of Porous Scaffolds
Scanning electron micrographs of split cross-sections (
Figure 3) showed that the scaffolds tended to form large pores of around 100 microns with well differentiated walls when underwent phase separation at the higher cooling rate.
It is interesting to note that upon the slower cooling condition, platelet-like structures were mainly distinguished. This structure was also observed to a greater extent in the PHBV(5%HV) copolymer. Some areas representing platelet-like morphology have been indicated by dashed-line circles in
Figure 4. Micrographs demonstrated that scaffolds had a three-dimensional porous structure and that the larger pores were further obtained from the scaffolds derived from the copolymer having the higher HV content. Specifically, PHBV(12%HV) scaffolds prepared at the higher cooling rate appear ideal considering the pore sizes, the homogeneous structure and the reduced platelet-like regions.
The tendency towards forming more large pores and reducing platelet-like morphologies were intensified in the surface image micrographs (
Figure 5). Images showed again that the HBV12-R6 sample was the more uniform one. The presence of small orifices in the pore walls ranging from several to tens of micron in size were discernible in both cross-section and surface micrographs, being considered as interconnectivities of the structure.
3.4. DMTA Analysis of Polymer Scaffolds
DMTA results revealed that mechanical properties of resulting scaffolds were influenced by the selected cooling rate and logically by the HV molar content of the copolymer, even for the small increase from 5 wt % to 12 wt % (
Figure 6).
An increase of the loss modulus (E″) and especially of the storage modulus (E′) was observed when solutions were cooled at the highest rate. Therefore, the E″/E′ ratio (i.e., loss tangent or tan δ) decreased. A similar effect was roughly observed when the HV content was lower. In summary, the locus of storage modulus curve shifted to the higher values and that of the loss tangent curve to lower values either by increase in cooling rate or decrease in HV contents. The dramatic decrease in the modulus at temperatures around 160 °C is associated with the melting point of the polymer, in accordance with supplier’s specifications.
3.5. Biocompatibility Assays for Scaffolds Prepared by TIPS from 1,4-Dioxane
HBV5 and HBV12 samples were evaluated as appropriate scaffolds to support cell adhesion and proliferation. Thus, epithelial-like MDCK and NRK cells were seeded in direct contact with the prepared scaffolds (
Figure 7). Cell adhesion was determined after 24 h as an early event of the cell growth in the scaffolds, while cell proliferation was determined after 7 days to demonstrate that cell growth and colonization were effective in the prepared scaffolds.
Images of fluorescence microscopy gave evidences of the cell adhesion (
Figure 7a,b) and of the formation of a cell monolayer onto all the scaffold samples (
Figure 7c,d). In the cell adhesion assay, cells appeared spread onto the surface of the scaffolds and the porous structure was maintained as evidenced by the dark and deep zones. In the cell proliferation assay, there was a clear increase of the number of cells grown on the surface of the sample. Micrographs showed that MDCK and NRK cells grew normally to contact each other and formed a cell monolayer by clusters and stackings, being drawn to the profile of the pores in the scaffold. Cells had a smaller size after proliferation due to its density increase, and the porous scaffold structure was maintained as deduced from the dark and deep zones. In this way, the prepared scaffolds had a sufficiently large pore size to not restrict the entry of cells into the scaffolds. The sponge-like morphology of these scaffolds (
Figure 5) is compatible with the excellent biocompatibility demonstrated in both cell adhesion and proliferation assays. Results confirmed that the studied scaffolds had a great potential for applications focused on tissue regeneration and remodeling.
Quantitative data of cell adhesion and proliferation are shown in
Figure 8a,b, respectively. The cell viability was determined considering the ratio between the number of cells grown in the scaffold and on the control (well of the culture plate), respectively.
Results indicated that cell adhesion was quantitatively similar in the prepared scaffolds and the control. Only HBV5-R2 and HBV12-R6 samples showed a significant reduction in the number of adhered MDCK cells, with values around 80% of cell viability. However, the samples HBV12-R2 with MDCK cells and HBV5-R6 with NRK cells were not significantly different despite having average values around 80% viability due to the greater dispersion of data as evidenced by their respective standard deviations (
Figure 8a). Regarding cell proliferation, which is a more consistent experiment because it corresponds to a period of 7 days, it was observed that the MDCK cells showed similar growth percentages as the control, while the NRK cells in the HBV5 samples showed a significant growth reduction. However, the measured values are close to 80% of viability (
Figure 8b), which can cause us to consider that differences may be caused by uncontrolled experimental factors. In this sense, it should be indicated that volume of scaffolds should be taken into account instead of surface (1 cm × 1 cm square samples were analyzed) since the scaffold galleries allow cell entry and colonization inside the scaffold. These considerations are supported by the morphological evidence of fluorescence microscopy for both adhesion and proliferation assays (
Figure 7). Therefore, results allow us to indicate that HBV5 and HBV12 scaffolds obtained from the 1,4-dioxane solutions at both 2 and 6 °C/min cooling rates are suitable and biocompatible supports for cell adhesion and proliferation in 3D cultures, and show potential interest for tissue regeneration applications.
5. Conclusions
The principal aim of this work was the study of process-properties relationship for poly(hydroxybutyrate-co-hydroxyvalerate) scaffolds fabricated by thermally induced phase separation. Due to presence of different driving forces for polymer and solvent crystallization and also liquid–liquid demixing, the phase separation process became relatively complicated. Thus, it was hard to consider a distinctive mechanism, being responsible for generating the different microporous structures. Variables corresponding to the fabrication process (i.e., cooling rate applied to the polymer solution) and material selection (i.e., PHBV with different HV molar ratios) strongly affected the phase separation process and led to different microporous structures and mechanical properties of the resulted scaffolds. Strictly speaking, more regions having the morphology associated with crystallization of the polymer were conspicuously detected upon slower cooling. These regions of relatively poor structural continuity were assumed to be the reason of the observed decrease in rigidity of the scaffolds. Besides, a tendency towards forming the typical morphology related to a solid–liquid phase separation was also observed in either higher cooling rate or higher HV content. Eventually, a high degree of crystallinity was recognized as the cause for the higher rigidity of the scaffolds having lower HV content. Finally, in-vitro cytocompatibility studies confirmed that these sponges-like scaffolds were nontoxic toward MDCK and NRK cells, and had a suitable porosity to cell adhesion and growth. Our data demonstrated that these biocompatible scaffolds with interconnected 3D networks are a promising to applications focused on tissue regeneration and remodeling.