Extrusion 3D-Printing and Characterization of Poly(Caprolactone Fumarate) for Bone Regeneration Applications

Abstract

Polycaprolactone fumarate (PCLF) is a cross-linkable PCL derivative extensively considered for tissue engineering applications. Although injection molding has been widely used to develop PCLF scaffolds, platforms developed using such technique lack precise control on architecture, design, and porosity required to ensure adequate cellular and tissue responses. In particular, the scaffolds should provide a suitable surface for cell attachment and proliferation, and facilitate cell-cell communication and nutrient flow. 3D printing technologies have led to new architype for biomaterial development with micro-architecture mimicking native tissue. Here, we developed a method for 3D printing of PCLF structures using the extrusion printing technique. The crosslinking property of PCLF enabled the unique post-processing of 3D printed scaffolds resulting in highly porous and flexible PCLF scaffolds with compressive properties imitating natural features of cancellous bone. Generated scaffolds supported excellent attachment and proliferation of mesenchymal stem cells (MSC). The high porosity of PCLF scaffolds facilitated vascularized membrane formation demonstrable with the stringency of the ex ovo chicken chorioallantoic membrane (CAM) implantation. Furthermore, upon implantation to rat calvarium defects, PCLF scaffolds enabled an exceptional new bone formation with a bone mineral density of newly formed bone mirroring native bone tissue. These studies suggest that the 3D-printed highly porous PCLF scaffolds may serve as a suitable biomaterial platform to significantly expand the utility of the PCLF biomaterial for various tissue engineering applications.

1.0. Introduction

In the US alone, millions of bone grafting procedures are performed each year, making bone the second most common transplanted tissue worldwide [1]. Bone exhibits a vast regenerative and self-repair capacity, unlike other tissues. However, this capacity can be limited in critical-sized bone defects (in which large defects fail to heal despite stabilization and surgery) due to improper vascularization or other systemic diseases that alter bone healing [2]. In these instances, bone tissue engineering (BTE) aims to restore biological, mechanical, and structural support of native bone tissue. 3D printing is vital in BTE and regenerative medicine to mimic complex and intricate bone structures and cellular interactions [3]. 3D-printed scaffolds that promote cellular proliferation, migration, and differentiation are thus essential for effective healing and recovery of functional bone tissue [4].

Macro- and micro-porosity are important properties of 3D-printed scaffolds and have been shown to promote angiogenesis, vascular development, and bone growth [2,5,6]. Research has also shown that porous architecture influences bone cellular growth, morphology, and interactions [2,5,7] as well as the scaffolds’ osteoconductive and osteoinductive properties [7]. Especially, porous struts support greater bone tissue formation and early vascular ingrowth and ongrowth [2], which promotes a suitable environment for cell proliferation, differentiation, and activity [2].

In the past decade, various advances on 3D printing technology, including extrusion-based bioprinting, have contributed to the fabrication of tissue-specific polymer scaffolds that are biodegradable and biocompatible. These polymers can be either natural, such as gelatin or chitosan, or synthetic, including polylactide (PLA), polycaprolactone (PCL), and polycaprolactone fumarate (PCLF). PCL has been widely implemented in BTE due to its biocompatibility, biodegradability, and Food and Drug Administration (FDA) approval as an implantable and injectable biomaterial [8]. To modulate the properties of PCL, various chemical modifications have been made, including the addition of fumarate segments to form PCLF [9]. Adding these fumarate groups renders PCLF crosslinkable [10] with tunable mechanical strength and biological response for various biomedical applications [8]. PCLF polymers have been developed and extensively studied for BTE and as nerve conduits [11–14]. Importantly, PCLF nerve conduits have significantly improved peripheral nerve regeneration and recovery by supporting the addition of bioactive factors that enhance angiogenesis and functional tissue repair [11,12]. Despite the extensive research of PCLF polymers for biomedical applications, 3D-printed PCLF scaffolds remain to be established and characterized.

In this study, we have developed a method for 3D-printing PCLF scaffolds using PCL as the supporting material via extrusion-based 3D printing. To our knowledge, this is the first report describing a method to 3D print PCLF. Moreover, through unique post processing, scaffolds with significantly porous struts have been developed. The scaffolds were extensively characterized, including in-vitro cell attachment and proliferation, vascular membrane formation using the ex ovo chicken chorioallantoic membrane (CAM) model [15], and in vivo calvarium bone formation [16]. The high macro-micro porosity of PCLF scaffolds resulted facilitated higher cell proliferation, better vascularized membrane formation, and accelerated bone formation.

2. Methods

2.1. Synthesis and characterization of PCLF

PCLF was synthesized by the esterification reaction of PCL diol with fumaryl chloride, as reported earlier [17]. Briefly, PCL diol of nominal molecular weight of 2000 g mol⁻¹ was dried overnight in a vacuum oven at 50 °C. Methylene chloride was dried and distilled over calcium hydride before dissolving the polymer for the reaction. Fumaryl chloride was purified by distillation at 161 °C. Potassium carbonate (K₂CO₃) powder was dried at 100 °C.

PCL diol, fumaryl chloride, and K₂CO₃ were measured in a molar ratio of 1:0.95:1.2 for the reaction. PCL diol (50 g) was dissolved in methylene chloride (300 ml) and placed in a three-neck flask containing K₂CO₃ (4.2 g). The mixture was stirred well with a magnetic stirrer to make a slurry. Fumaryl chloride (2.5 ml) was added dropwise to the slurry under nitrogen. The reaction mixture was maintained at 50 °C under reflux using a condenser for 12 h.

For purification, the mixture was cooled down, transferred to centrifuge tubes, and spun down at 4000 rpm for 15 min to eliminate the unreacted K₂CO₃. The supernatant was then added dropwise to petroleum ether to precipitate the polymer, which was eventually vacuum-dried to obtain the purified PCLF. The polymer was stored at −20 °C for further use.

Synthesized PCLF was characterized with Fourier transform infrared (FTIR) spectroscopy in attenuated total reflection mode. PCL and PCLF samples were scanned from 3500 to 750 cm⁻¹ to identify and confirm the modification of PCL to PCLF.

2.2. Preparation of PCLF ink

To prepare the PCLF ink for extrusion printing, synthesized PCLF and commercial PCL (average molecular weight of 80,000 g mol⁻¹, Sigma Aldrich, WI) were blended at two weight ratios of 75:25 and 70:30 (PCLF:PCL) in methylene chloride. Photoinitiator bisacylphosphinoxide (BAPO, Ciba Specialty Chemicals, Tarrytown, NY) dissolved in methylene chloride was added to the polymer solution at 0.5 wt% and stirred for 1 h to enable the proper mixing under protection from light. The whole mixture was left overnight in the dark to remove the solvent. The solid polymer block was prepared as small pellets for printing.

2.3. Thermal and rheological characterization of printable ink

Differential scanning calorimetry (DSC, TA Instruments) was used to evaluate the melting temperature (T_m) of PCLF and the blends. The temperature of loaded polymer specimens was scanned from 0 to 200 °C. with a constant rate of 20 °C min⁻¹.

Frequency-sweep test was performed for the PCLF-PCL ink at 75 °C (printing temperature) and for PCLF at 75 °C and 40 °C (melting temperature of PCLF). To perform the test, polymer pellets were placed on the lower plate of the rheometer (DHR-1 discovery hybrid rheometer, TA instruments, DE) and heated to the desired temperature. Frequency sweeps from 1 to 100 Hz was performed to measure the storage modulus (G’), loss modulus (G”), and complex viscosity.

2.4. Extrusion printing and post-processing of 3D printed scaffolds

The extrusion printing and post-processing steps are schematically presented in Figure 1. The printing parameters for PCLF-PCL ink were optimized for the nozzles with sizes 400 μm and 200 μm to print the multiple layers with minimal artifacts on the original design. A thermoplastic printhead fitted to a BIO-X bioprinter (CELLINK, MA) was used for extrusion printing. The polymer pellets were loaded into the stainless-steel extruder and heated to 75 °C. After maintaining the temperature for 15 min, the polymer melt was extruded to print the desired structures. The extrusion pressure and printing rate were set to 200 KPa and 5 mm/s for 400 μm nozzle and 220 KPa and 3 mm/s for 200 μm nozzle. The printed structure was immediately UV crosslinked for 20 min in a UV curing chamber. These scaffolds are labeled as PCLF-PCL for the rest of the study.

Figure 1.

Schematic representation of extrusion printing of PCLF-PCL scaffold and its post-processing to achieve highly porous PCLF scaffolds.

To eliminate the PCL component from the PCLF-PCL scaffold, it was further post-processed with methylene chloride and acetone solvents. The PCLF-PCL scaffolds were first treated with methylene chloride for 30 min. The swelled scaffolds were then transferred to an acetone bath for 24 h to completely eliminate PCL and sol fraction of crosslinked PCLF. The scaffolds were placed inside a UV curing chamber for 15 min post-curing and thoroughly washed with deionized (DI) water. These scaffolds are referred to as the PCLF scaffolds throughout this article.

2.5. Characterization of 3D printed scaffolds

2.5.1. Parameters of post-processing

The swelling ratio and physical weight loss of the PCLF-PCL scaffolds during post-processing were determined after saturating the scaffolds with methylene chloride solvent. The initial dry weight (W_di) of the PCLF-PCL scaffold before saturation, immediately after saturation (W_w), and final dry weight after saturation (W_df) were recorded. The swelling ratio was calculated as (W_w-W_di)/W_di. The percentage of physical weight loss was determined as ((W_di-W_df)/W_di) X 100.

2.5.2. In vitro degradation rates

The PCLF-PCL and PCLF scaffolds were immersed in phosphate-buffered saline (PBS) or 1N NaOH solution and incubated at 37 °C on an orbital shaker. The remaining weight of the scaffolds at different time points was determined after thorough washing and complete drying.

2.5.3. Surface morphology

Scanning electron microscopy (SEM) was used to characterize the surface morphology, macro-pores, and struts size. The images were captured at 2000X (surface morphology), 50X (strut size), and 35X (macropores). The dried scaffolds were sputter-coated with gold-palladium and imaged using SEM (S-4700, Hitachi Instruments, Tokyo, Japan) at a voltage of 5 kV. Five different areas of three different scaffolds were imaged for the quantitative analysis of strut and pore sizes.

The micropores on the surface of struts were further analyzed using an atomic force microscope (AFM). The samples were fixed on the aluminum stubs and scanned using a Nanoscope IV PicoForce Multimode AFM machine (Bruker, Santa Barbara, CA). For the quantitative analysis of surface nanoroughness, three samples from each group were scanned at three different locations. The root mean square (Rq) of roughness and maximum peak height were recorded.

2.5.4. Mechanical characterization of the scaffolds

The compressive mechanical properties of PCLF-PCL and PCLF scaffolds were determined using MTS 858 Bionix II testing machine (MTS Systems Corporation, Eden Prairie, MN). The specimens for testing were prepared as cylinders with an average diameter-to-height ratio of 1:2. For comparative analysis, the specimens for testing were prepared by injection molding and 3D printing. For injection molding, the PCLF-PCL and PCLF polymer melts were directly cast into cylindrical molds. For quasi-static compressive testing, the load was measured using an MTS load cell with 100 lb capacity, with test speed set to 3 mm/min, and data was collected at 256 Hz. For each specimen, the force and displacement data from the load frame were used to calculate the stress-strain values. Compressive strength was calculated as maximum stress before the specimens yielded, and the compressive modulus was calculated as the slope of the linear region of stress-strain curves.

2.5.5. Protein adsorption to the scaffolds

The adsorption of proteins to the scaffolds from the cell-culture medium was determined for 3D printed PCLF-PCL and PCLF scaffolds. The scaffolds were immersed for 2h at 37 °C in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS). After washing with PBS thrice to remove unabsorbed proteins, the scaffolds were treated with 1% sodium dodecyl sulfate (Bio-Rad Laboratories, Hercules, CA) to wash out the adsorbed proteins. The concentration of proteins was determined using a Micro BCA protein assay kit (Pierce, Rockford, IL) following the kit instructions. The concentration was normalized to the total surface area of the scaffolds for quantitative analysis.

2.6. Mesenchymal stem cell response to the scaffolds

For in vitro studies, the scaffolds were immersed in ethanol for 8 h. This was followed by immersion in DI water for another 8 h for thorough washing. The sterilized scaffolds were dried and stored under sterile conditions. Sprague Dawley (SD) rat bone marrow-derived mesenchymal stem cells (rMSC) (Fisher Healthcare, MA) were passaged 3-4 times in a low glucose-DMEM complete medium for the study.

2.6.1. MTS assay

For MTS assay, rMSC were plated on a 24-well plate at 20,000 cells/well for 24 h to allow the complete attachment to the plate. Sterilized scaffolds (Ø 5 mm, height 2 mm) were then carefully placed on the top of cells, and fresh media was added. The plates were returned to the incubator to perform the MTS assay at predetermined time points. The goal here was to evaluate the cytotoxicity of leach out components (if any) of the scaffolds to rMSC. To compensate for background readings from the scaffolds, certain wells without cells loaded with scaffolds were prepared simultaneously. On days 1, 3, and 5, the MTS assay (CellTiter 96, Promega, Madison, WI) was performed following the manufacturer’s instructions. The results are presented are presented as percent change in viable cells compared to TCPS control.

2.6.2. Cell attachment and proliferation on the scaffolds

2.6.2.1. Cell seeding

The scaffolds were incubated with complete DMEM for 15 min and attached to the bottom of a 24-well-plate using silicon-based high vacuum grease (Dow Corning, Midland, MI). This grease was previously autoclaved and stored in a sterile environment before use. For seeding, 50,000 rMSC in 300 μl of the medium, enough to barely cover the scaffolds, was added to the top of the scaffolds and incubated for 3 h. Another 700 μl culture medium was then added to the wells and subsequently cultured for 14 days with medium change every 2-3 days.

2.6.2.2. Cell attachment and proliferation on the scaffolds

The attachment and proliferation of cells on the scaffolds were observed by using a rhodamine-phalloidin stain (RP, Cytoskeleton Inc, Denver, CO). Phalloidin binds specifically to the actin filaments of the cells. For staining, cells on the scaffolds were fixed with 4% paraformaldehyde solution and permeabilized with 0.5% Triton X-100 solution. The stain solution was added to the permeabilized cells and incubated for 30 min in the dark at room conditions. After thorough washing with PBS, the cellular nuclei were counter-stained with 4’,6-diamidino-2-phenylindole (DAPI, Thermo Fisher Scientific, Waltham, MA). RP-stained rMSC were imaged using a laser scanning confocal microscope at 20X magnification (Carl Zeiss, Germany).

2.6.2.3. Quantitative assessment of cell proliferation on the scaffolds

CyQUANT cell proliferation assay kit (Thermo Fisher Scientific, Waltham, MA) was used for the fluorescence-based quantitative assessment of cell proliferation on the scaffolds. At pre-defined time points, scaffold-cell constructs were washed once with PBS. The scaffolds were then carefully transferred to a new well and washed once with PBS. The cells attached to the scaffolds were detached and released by trypsinization. The complete release was ensured by repeated pipetting. The cell suspension was transferred to a microcentrifuge tube and centrifuged at 2000 rpm for 10 min at 4 °C. The supernatant was removed without disturbing the cell pellet. The cells were resuspended in PBS and centrifuged again at the same conditions. The cell pellets were then kept frozen at −80 °C before all the specimens, on different days, were ready for the assay. The actual number of cells on the scaffolds was determined by generating a standard curve with different cell numbers according to the kit instructions.

2.7. Vascularized membrane formation on the scaffolds

The chicken chorioallantoic membrane (CAM) model was used to study the integration of porous 3D printed scaffolds with vasculogenic membranes. For ex ovo model of CAM, the eggshells of fertilized chicken eggs (Hoover’s Hatchery, Rudd, IA) were cracked open, and the embryos were transferred to plastic weighing boats. After 5 days of incubation in a humidified incubator, the highly vascularized CAM was ready for scaffold implantation. The implantation site on CAM was randomly selected, mostly away from the developing embryo. Smaller blood vessels forming Y-junction were located and slightly scratched with a cotton swab to rupture vessels and expose blood. The sterilized scaffold (Ø 5 mm, height 2 mm) was carefully placed on the top of exposed blood on CAM without disturbing the membrane. Implanted CAMs were incubated further in a humidified incubator. On day 7, implanted scaffolds were imaged, followed by harvesting for subsequent sectioning and staining. The scaffolds were fixed in neutral buffered formalin, embedded in paraffin, and sectioned for Masson’s trichrome staining.

2.8. In vivo bone formation: Rat calvarium defect model

Highly porous 3D printed PCLF scaffolds were implanted to 5 mm calvarium defects in rats. Animal surgical work in this study was performed under the protocol reviewed and approved by Institutional Animal Care and Use Committee (IACUC) at Mayo Clinic, USA. Twelve 8-week-old Sprague Dawley (SD) rats (Charles River Laboratories, Wilmington, MA) were used to create the bilateral critical-sized 5 mm defects on the calvarium for scaffold implantation. In each rat, one defect was left un-implanted, serving as the void control group, and the other was implanted with PCLF scaffolds. The skin incision was closed with 4-0 vicryl sutures, and the rats were allowed recover post-surgery. Recovered animals were housed in a twelve-hour light/dark cycle and fed a normal diet. At 4- and 8- weeks post-surgery, rats were sacrificed, and the calvarium bone was harvested. After the tissue fixation in 4% buffered formalin solution, the bone samples were subjected to further analysis.

To evaluate the bone formation at the defect site, fixed specimens were scanned using a Micro-CT system (Bruker Skyscan 1276, Germany). The slices were reconstructed to generate the CT images. The quantitative assessment of newly formed bone was conducted by measuring the CT parameters such as bone volume/total volume (BV/TV) and bone mineral density (BMD).

For tissue morphometry analysis of newly formed bone, bone specimens were decalcified in EDTA hydrochloric acid decalcifying solution (Thermo Fisher Scientific, Waltham, MA), followed by the paraffin embedding and sectioning. The de-paraffinized sections were stained with hematoxylin and eosin and scanned using a slide scanner (AxioScan, Carl Zeiss, Germany).

2.9. Statistical analysis

The statistical analysis of quantitative data was performed using IBM SPSS statistics software. Unless mentioned otherwise, one-way analysis of variance (ANOVA) followed by Tukey’s post hoc was done to test the significance of differences among the groups, and the groups with p<0.05 were marked as significantly different.

3. Results and discussion

3.1. Synthesis and characterization of PCLF

The method of PCLF synthesis from PCL diols has been well-established in our laboratory [18]. In this study, PCLF 2000 was synthesized from PCL diol with a molecular weight of 2000 g/mol. A systematic investigation of this synthesis has been reported earlier [17,18]. As shown by the FTIR results in Figure 2A, an absorption band at 1630 cm⁻¹ was observed for PCLF that corresponds to the stretching vibration of -CH=CH- bond. This band was absent on the FTIR spectrum of PCL. Furthermore, a broad -OH band at 3400 cm⁻¹ was missing on the PCLF spectrum which further confirmed the esterification reaction.

Figure 2:

Printable ink development and characterization. (A) FTIR spectra of PCL and PCLF highlighting the presence of C=C bond on PCLF (long arrows and onset figure) and absence of -OH band (small arrow) confirming fumarate modification of PCL. (B) DSC characterization of PCL and PCLF, and their blends at two different ratios. (C) Viscosity vs shear rate of PCL and PCLF, and their blends at two different ratios. (D) 3D printed scaffolds with two different nozzle sizes. (E) Post processing of scaffolds and (F) rolled up post-processed scaffold. (G) Comparison of flexibility of PCLF-PCL and PCLF scaffolds. (H) Degradation of scaffolds under neutral (PBS) and alkaline pH (NaOH) conditions.

3.2. Development and characterization of PCLF ink

3D printable PCLF ink constituted of PCLF, PCL, and BAPO as a photoinitiator. The thermal characterization of polymers using DSC is shown in Figure 2B. The T_m of PCLF and PCL was observed as 47 °C and 58 °C, respectively. The melting curve of composite PCLF-PCL blend showed slight changes in the T_m of PCLF and PCL. This indicates the miscibility of two components and non-significant micro-phase separation [19]. Taking the highest temperature of exothermal peaks, the T_m of 70:30 and 75:25 composition was 56.4 °C and 57.6 °C, respectively.

The rheological characterization of polymers at 75 °C, the temperature used during printing, is shown in Figure 2C and supplementary Figure S1. The viscosity of pure PCL was several folds higher than that of pure PCLF. This was expected, as the molecular weight of PCL is about 10 times higher than that of PCLF. No shear thinning behavior was observed, as the viscosity almost remained constant with the increase in the shear rate for both pure PCL and PCLF. This behavior inflicted the pure PCL to have irregular printing at 75 °C and 200 KPa. The extrusion printing of high molecular weight PCL has been carried at pressure ranging from 350-500 KPa and temperature ranging from 80-100 °C, depending on the polymer concentration and nozzle size [20].

The viscosity vs. shear rate curve of PCLF-PCL blend was observed to align well between that of pure PCLF and PCL. Because of the higher PCL amount, PCLF-PCL at a 70:30 weight ratio showed slightly higher viscosity than a 75:25 formulation. Also, these formulations showed shear thinning behavior, the viscosity decreased at higher shear rates. This rheological property of the blends correlated well with their excellent extrusion printability at 75 °C.

3.3. Extrusion printing

As indicated earlier, printing parameters, including polymer composition, printhead temperature, extrusion pressure, and printing speed, were optimized for 400 μm and 200 μm nozzle. To identify the accurate heating temperature of the thermoplastic printhead that would allow for uninterrupted extrusion of PCLF-PCL, polymer pellets were loaded into a 10 ml stainless steel cartridge for printer-controlled heating. The extrusion pressure was set close to 200 KPa. It was observed that at a temperature close to the melting temperature of PCLF, the PCLF-PCL melt could not be extruded from either 400 μm or 200 μm nozzles due to incomplete melting of the blend. Pure PCLF could be extruded but did not retain the original design due to excessive fiber spreading and eventual structural collapse after extrusion. From this optimization, the ideal printhead temperature for printing was identified to be 75 °C to 80 °C using both nozzle sizes at the pneumatic pressure ranging from 200 KPa to 220 KPa.

It should be noted that initial optimization work on polymer ratios, thermoplastic printhead temperature, extrusion pressure, and printing speed was contingent on the 400 μm nozzle size. This optimization elicited a 75:25 weight ratio of PCLF-PCL as a printable formulation with a temperature set to 75 °C, pressure set to 200 KPa, and printing speed set to 4 mm/s. A simple 10 mm lattice printed with this formulation and specified parameters is shown in Figure 2D. However, this formulation at specified parameters was not printable with a 200 μm sized nozzle. The changes in temperature and pressure within the appreciable range from the original parameters failed to print the 75:25 formulation with a smaller nozzle size.

Consequently, we modified the PCLF-PCL blend composition from a 75:25 to a 70:30 weight ratio. This composition made it possible to print the structures using the 200 μm nozzle. The printing parameters included printhead temperature at 80 °C, pressure set to 230 KPa, and printing speed set to 4 mm/s. A representative printed lattice for this composition is shown in Figure 2D. For the rest of this study, only PCLF-PCL scaffolds printed with a 400 μm nozzle will be used.

3.4. Post-processing of 3D printed scaffolds

The post-processing steps involved treatment with methylene chloride, acetone, and UV-curing. Since PCLF possesses the crosslinking fumarate groups, the uncrosslinked PCL component was washed out by the treatment of printed PCLF-PCL constructs with methylene chloride. During this step, the scaffolds underwent swelling, as shown in Figure 2E. The swelling ratio was about ten times their initial weight. An experimental weight loss of about 20% was observed during this step. The swollen scaffolds were further subjected to overnight acetone treatment to ensure the complete removal of PCL (25%). As observed in Figure 2E, the PCLF scaffolds obtained after post-processing of PCLF-PCL scaffolds completely retained their printed shape and design with minimal artifacts. In addition, printed porous mesh could be easily rolled into a tubular structure in a swollen state (Figure 2F) and maintained flexibility after drying compared to PCLF-PCL scaffolds (Figure 2G).

3.6. Physical characterization of the scaffolds

Figure 2H shows the degradation kinetics of printed scaffolds in PBS and NaOH at 37 °C. The homopolymer of PCL is primarily known for its slow degradation rate, with studies demonstrating several years for its significant degradation [21]. However, the degradation kinetics of PCL can be tailored by its copolymerization with faster degrading polymers [21,22]. Regardless, in this study, the blending of PCL and PCLF had non-significant effects on their hydrolytic degradation. Both PCLF and PCLF-PCL showed similar degradation kinetics up to 6 months with an average reduction of 6% weight. Between 6 months to 1-year, the degradation rate of PCLF scaffolds was off the pace with that of PCLF-PCL scaffolds, as a 5.5% reduction in remaining weight was observed for the former compared to a 10% reduction for the latter. With no data on the hydrolytic degradation of crosslinked PCLF, this study suggests that crosslinked PCLF degrades slower than the uncrosslinked PCL. Under the alkaline pH condition, both the scaffolds degraded rapidly, with a 40% reduction in weight observed after 10 days.

SEM images of the scaffolds are shown in Figure 3A. Both the surface and cross-section of PCLF-PCL were observed to be markedly smooth. In contrast, the surface of the PCLF scaffold was relatively porous, with largely micropores distributed along the struts. The cross-section was not as porous as the surface, but few pores were observed. As shown in Figure 3B, the average strut size of PCLF-PCL scaffolds was 417 μm. This was comparable to the extrusion nozzle size of 400 μm and demonstrates the printing capability with high precision and fidelity. After post-processing, the struts’ average size of PCLF was reduced to 354 μm, 11.5% less than the original extrusion size (Figure 3C). As shown in figure 3D, the PCLF scaffolds had an average pore size of 928 μm compared to 894 μm for PCLF-PCL scaffolds because of the decrease in strut size after post-processing. The surface topography of the scaffolds was further characterized using AFM. As shown in figure 3E and quantitative data in figure 3F, the nano surface roughness of PCLF scaffolds was significantly higher than that of PCLF-PCL scaffolds.

Figure 3.

Physical characterization of 3D printed scaffolds. (A) SEM images of scaffolds showing highly porous PCLF scaffolds. Comparison of (B) strut diameter, (C) change in fiber diameter, and (D) pore size between PCLF-PCL and PCLF scaffolds. (E) and (F) AFM characterization of scaffolds. (G) Comparison of protein adsorption to PCLF-PCL and PCLF scaffolds. * and # indicate significantly higher than PCLF-PCL group at p<0.001 and p<0.05, respectively

The surface property of the scaffolds was further characterized by assessing protein adsorption. As observed in Figure 3G, a higher quantity of protein per unit scaffold weight adhered to the surface of PCLF scaffolds compared to PCLF-PCL scaffolds. Higher surface roughness is directly related to the quantity of protein adsorbed to the surface, making those scaffolds well-suited for supporting cell attachment and proliferation.

The scaffolds’ compressive properties were measured according to ASTM International Standard D695-10. Cylindrical specimens with a diameter of 5 mm and a height of 10 mm were prepared either by injection molding or 3D printing (Figure 4A). Representative stress-strain curves for the specimens are presented in Figure 4B, and quantitative measurements of compressive strength and modulus are shown in Figure 4C. The compressive strength and modulus of PCLF scaffolds were lower than that of PCLF-PCL scaffolds. As the PCL component is eliminated during post-processing, the resulting porosity along the struts of PCLF scaffolds resulted in lower compressive parameters than PCLF-PCL scaffolds. For similar reasons of high porosity in 3D printed scaffolds, the compressive strength of molded PCLF scaffolds was observed to be higher. For PCLF-PCL scaffolds, both the compressive strength and modulus were similar for molded, and 3D printed scaffolds. The mechanical behavior of 3D printed PCLF-PCL, and PCLF scaffolds were investigated under cyclic compressive loading. For all the study groups, a large hysteresis loop was observed in the first loading-unloading cycle, however, the remaining cycles showed a similar hysteresis loop with insignificant change in the recovery profile (Figure 4D). Molded and 3D printed PCLF scaffolds (Figure 4D i and ii) showed slightly smaller residual strain compared to molded and 3D printed PCLF-PCL scaffolds (Figure 4D iii and iv).

Figure 4.

Mechanical characterization of the scaffolds. (A) 3D printed specimen for compression testing. (B) Representative stress-strain curves for molded and 3D printed PCLF-PCL and PCLF scaffolds. (C) Quantitative assessment of compressive strength and modulus for the scaffolds. (D) Evaluation of mechanical response to cyclic compressive loading for molded PCLF (i) and PCLF-PCL (iii), and 3D printed PCLF (ii) and PCLF-PCL (iv). ‘a’ and ‘b’ indicate significantly higher (p<0.05) compressive strength compared to molded and 3D printed PCLF, respectively. ‘c’ and ‘d’ indicate significantly higher (p<0.05) compressive modulus compared to molded and 3D printed PCLF, respectively.

3.7. In vitro assessment of scaffold properties

Rat mesenchymal stem cells (rMSC) were used throughout the in vitro studies. rMSC co-culture with scaffolds placed in a 24-well plate showed normal cell growth, indicating the cytocompatibility of scaffolds. As shown from the MTS assay result in Figure 5A, no adverse effects on cell growth were observed over 5 days as demonstrated by a comparable or higher cell viability percentage for both scaffolds groups compared to the TCPS group. Furthermore, when seeded with rMSC, the surface of the scaffolds was highly conducive to cell growth, evident through the cytoskeletal stain (red) and nucleus stain (blue) in Figure 5B. Excellent spreading of cellular filaments was observed on the surface of scaffolds. In particular, cell attachment and proliferation were observed to be higher on PCLF scaffolds than PCLF-PCL scaffolds. This was further evident in the quantitative assessment of cell proliferation. A significantly higher fluorescence intensity (Figure 5C) and corresponding cell number (Figure 5D) were observed on the PCLF scaffolds compared to PCLF-PCL scaffolds, specifically on day 7.

Figure 5.

In vitro characterization of 3D printed scaffolds. (A) MTS assay results showing comparable OD values for rMSC cultured on TCPS surfaces with or without co-culture with scaffold groups. (B) Actin staining showing the spreading of rMSC directly cultured on the surfaces of 3D printed scaffolds. (C) and (D) Quantitative assessment of cell number on the surfaces of the scaffolds. The curve on the onset shows the calibration curve generated for the quantitation of cell number from fluorescence intensity. * indicates significantly higher (p<0.05) values between the groups.

3.8. Vascularized membrane formation on the scaffolds

CAM model was used to assess the initial biomaterial-tissue interaction and the conduciveness of PCLF scaffolds to vascularized membrane (Figure 6A). Approximately 95% of CAM survived the implantation of the scaffolds (5% non-viability was not associated with the implantation of the scaffolds). The microscopic observation of implanted CAM on day 5 demonstrated an excellent membrane formation on the membrane contact side of both PCLF-PCL and PCLF scaffolds. Interestingly, better membrane ongrowth and ingrowth were observed on both the top and bottom sides of PCLF scaffolds compared to PCLF-PCL scaffolds. As shown in Figure 6B and 6C, the coverage of vascularized membranes on the top side of PCLF scaffolds was higher than the PCLF-PCL scaffolds. The scaffolds covered with the vascularized membrane recovered on day 5 of implantation are shown in Figure 6D. The histological observation of the membrane using Masson’s trichrome staining further demonstrated that the membrane was highly vascularized, evident through several blood vessels on the membrane (Figure S2). Through the qualitative observation of the images, no rigid differences in histomorphometry were observed between the two scaffold groups. However, it was evident that ongrowth of the membrane along the struts of the scaffolds was higher on PCLF scaffolds compared to PCLF-PCL scaffolds.

Figure 6.

Ex ovo and in vivo functional assessment of 3D printed scaffolds. (A) Scaffold implantation to CAM specimen. Microscopic observation of (B) PCLF-PCL and (C) PCLF scaffolds at day 5. (D) Photograph of harvested PCLF scaffold showing the complete coverage of scaffold bottom and side with highly vascularized membrane. (E) 3D-reconstructed micro-CT images showing new bone formation on the calvarium defects implanted with 3D printed PCLF scaffolds on the left and void on the right at 4 (top) and 8 (bottom) weeks. (F) Quantitative assessment of bone parameters for the regenerated bone. (G) Micro-CT images showing mapping of bone mineral density. (H) H&E staining of the bone sections from the defect site. * indicate significantly higher (p<0.05) observation for PCLF group compared to void group.

3.9. Calvarium bone formation

As demonstrated on the reconstructed micro-CT images in Figure 6E, the onset of new bone formation was observed along the struts of scaffolds at 4 weeks post-implantation. New bone formation continued to persist along the struts and became more prominent at 8 weeks post-implantation. In comparison, such accelerated bone formation was not observed on the void groups at either time point, though new bone formation was observed to occur along the edges of the created defect (Figure 6E). The quantitative assessment of new bone formation further demonstrated a significantly higher BV/TV at both time points and significantly higher BMD at 8 weeks for the PCLF scaffold group compared to the void group (Figure 6F).

Furthermore, the 2D mapping of mineral density showed some interesting variability in the mineral density of newly formed bone among two groups at 8 weeks. As observed in Figure 6G, the mineral density of newly formed bone along the struts of PCLF density appeared similar to native calvarium bone (mostly red). The mineral density of newly formed bone for the void group was different and at the lower end compared to the native bone (which appeared mostly green). The excellent bone formation along the struts of PCLF scaffolds was further demonstrated by H&E staining (Figure 6H).

Discussion

In general, PCL diols with lower molecular weight are used for PCLF synthesis to overcome the steric hindrance effects with high molecular weight PCL diols that causes a lower degree of fumarate functionalization [17,23]. Three-different molecular weights of PCL diol, 530 g/mol, 1200 g/mol, and 2000 g/mol, have been successfully used to synthesize PCLF [24]. Due to the lower molecular weight, the viscosity of PCLF solution or PCLF melt was also lower, preventing the fine-fiber extrusion needed for 3D printing of PCLF (results not shown). For instance, PCLF synthesized from PCL diol of molecular weight 2000 g/mol has a molecular weight of 7300 g/mol which has been shown to be not suitable for melt extrusion printing [17,18]. As a comparative reference, PCL, with a molecular weight of 40,000 g/mol and higher, can be conveniently 3D printed using extrusion printing [25]. Consequently, limited success has been achieved in the 3D printing of PCLF.

Unlike PCL, PCLF is a crosslinkable polymer with controllable mechanical and thermal properties suitable for tissue engineering applications. Depending on the molecular weight of precursor PCL diols, the tensile modulus of PCLF can vary from 0.87 MPa to 138 MPa. This range of mechanical properties enables its application in nerve and bone regeneration applications [8,17,18]. PCLF nerve conduits developed using injection molding are currently under clinical trial for the repair of a 6-cm sural nerve defect [24]. To specifically expand the utility of PCLF and to better design the PCLF grafts for state-of-the-art tissue regeneration applications, we strived to develop a 3D printable PCLF ink.

PCL of high molecular weight was used in this study to support the pneumatic extrusion of PCLF. PCL and PCLF showed an excellent miscibility without any noticeable phase separation. Two compositions of PCLF-PCL blend were identified to be optimal for printing at two different resolutions. PCLF-PCL blend ratio of 75:25 was printable with a 400 μm but not 200 μm nozzle. The blend composition of 70:30 was ideal for printing with a 200 μm nozzle. This composition was also observed to be extrudable and printable with a 400 μm nozzle.

The post-processing steps involved treatment with methylene chloride and acetone, and UV-curing. The removal of the PCL component during post-processing provided PCLF scaffolds with extra physical flexibility. The high porosity of PCLF scaffolds, especially along the struts, contributed to their high flexibility [26]. This further demonstrates that the post-processing of PCLF-PCL scaffolds intended to eliminate the PCL component enhanced the surface properties for tissue engineering applications. Furthermore, when seeded with rMSC, the surface of the scaffolds was highly conducive to cell growth. It has been well acknowledged across the orthopedic research community that porous/rough substrate can facilitate better cell attachment by providing favorable anchorage points for cells, eventually resulting in better downstream cellular responses [27,28]. Accordingly, several research endeavors have strived to modify the PCLF surface porosity using various techniques, including surface decoration with biological factors, and nano-surface modification with inorganic materials [29].

CAM model was used to assess the initial biomaterial-tissue interaction. Integration of the scaffold to the embryonic tissue and the growth of vascularized membrane on the scaffold surface was observed on both PCLF and PCLF-PCL scaffolds. Early angiogenic activities around the implantation site have been shown to elicit better bone remodeling and healing [30–32]. The on-growth and in-growth of new blood vessels facilitate the delivery of stem cells, growth factors, and cytokines needed for the reparative process to embark [32,33]. Additionally, the Masquelet technique, one of the proven techniques of repairing complicated bone defects, uses a synthetic spacer to induce vascularized membrane formation and eventually implant such membranes to the bone defects [34,35]. While traditionally, polymethyl methacrylate (PMMA) spacers have been dominantly used for this application, several recent studies have demonstrated thicker membranes with improved osteoinductivity can be achieved with alternative materials and different surface properties [36]. Calvarium bone defect model was used to further assess the long-term response of scaffold implantation to the bone tissue. The comparative study of bone formation due to implantation of PCLF scaffolds with the un-implanted void group showed the accelerated bone formation with PCLF scaffolds resulting in higher bone volume at the terminal time points.

Conclusion

We have successfully developed a robust new method of 3D printing of PCLF, a fumarate-derivative of PCL, which is a widely used polymeric biomaterial in tissue engineering applications. This is the first report describing an extrusion printing of 3D PCLF scaffolds. Furthermore, the unique post-processing method reported here facilitated the development of macro and micro-porous scaffolds with high flexibility and suitable mechanical properties for bone tissue engineering applications. In-vitro evaluation of these porous scaffolds demonstrated improved MSC responses including cell attachment and proliferation. We also demonstrated that highly porous PCLF scaffolds can facilitate excellent vascularized membrane formation upon implantation to the CAM model compared to microporous PCLF-PCL scaffolds. Additionally, when these highly porous PCLF scaffolds were implanted into rat calvarium defects, significantly improved and accelerated bone regeneration was observed compared to non-implanted defects. Overall, the unique method reported here for 3D printing of PCLF scaffolds will help design PCLF-based polymeric biomaterials with diverse biomedical applications.