CNT incorporation improves the resolution and stability of porous 3D printed PLGA/HA/CNT scaffolds for bone regeneration

In this study, 3D printed porous poly(lactide-co-glycolide) (PLGA) and its nanocomposites with 5 wt. % hydroxyapatite (HA) and 0.5, 1 and 2 wt. % carboxyl-functionalized multi-walled carbon nanotube (CNT) scaffolds were fabricated by using extrusion-based printing. The printing parameters were optimized by rheological studies. The rheological studies demonstrated shear thinning properties for all compositions and an increase in storage modulus was observed after the addition of CNT. Porous PLGA/HA/CNT scaffolds were printed by applying a pressure of 4.76 bar at 125 °C. The addition of 0.5 wt. % of CNT reduced the strut size and increased the porosity from 42% to 60%. The increase in storage modulus and decrease in strut size were related to hydrogen bonding between CNT, HA and PLGA which ultimately improved shape fidelity. The scaffolds were characterized by analysis of their chemical structure, water contact angle measurement, in vitro bioactivity test, biodegradation test, mechanical analysis, and in vitro cell studies. The scaffolds were found to be more hydrophilic by the incorporation of CNTs. Also, degradation studies showed that the microstructure of the scaffold became more stable with the addition of HA and CNT. The compressive modulus of PLGA/HA/CNT2 scaffold was found to be 548.5 MPa, which is found suitable to replace cancellous bone. The scaffolds were found to be highly biocompatible which is possibly due to alignment of CNT and PLGA during 3D printing process. Alizarin red staining indicated improvement of mineralization of MC3T3-E1 cells on the CNT incorporated porous 3D scaffolds. The results suggest that the produced porous 3D printed PLGA/HA/CNT scaffolds are promising for bone regeneration applications.


Introduction
When bone defect due to trauma, resection or complex fracture is larger than critical size, it is incapable to regenerate itself [1][2][3]. To treat bone defects usually autografts and allografts are utilized, which are considered as gold standard [4]. However, they have some limitations in practice such as insufficient donor site, morbidity and induction of immunogenic reaction [5]. Bone tissue engineering may offer the best solution due to tunable mechanical and biological properties of designed porous 3D scaffolds [5,6]. Those provide a temporary structure in which the cell can adhere, differentiate and grow leading to new tissue formation while the biomaterial gradually degrades [7,8]. It is critical to design a biocompatible, biodegradable scaffold with interconnected microporosity with a pore size of 450-700 µm to deliver the nutrient and waste exchange effectively, promote vascularization [9,10] and osteogenesis [7,11]. In order to obtain such a unique design with desired pore size and shape, the computer assisted method is required [12].
Among many manufacturing techniques used, 3D printing comes forth as an efficient method to design the required porous 3D scaffolds since complex structures can be fabricated successfully. For 3D printing, different techniques are applied such as stereolithography, extrusion-based and ink-jet-based printing [13]. Polyesters such as PLGA, poly(lactic acid) (PLA) and poly(caprolactone) (PCL) which are usually 3D printed by extrusion-based techniques [14][15][16][17]. However, the pristine polymeric scaffolds are generally insufficient in terms of their biological capacity and mechanical strength. Polyesters are usually combined with ceramics such as bioactive glass, and calcium phosphates in bone substitution design to improve the bioactivity and mechanical strength of the ultimate composite [18]. Furthermore, carbonbased materials, such as carbon nanotube (CNT), graphene oxide are also employed to enhance the physical and biological performance of these scaffolds [19,20].
Calcium phosphates have commonly incorporated into polyester scaffolds, using different 3D printed techniques such as fused deposition modeling and selective laser sintering [21][22][23][24]. Park et al [22] bioplotted 3D scaffolds of 40 wt. % HA/PCL with uniform porous structure and interconnectivity. In vitro studies with MG-63 cells indicated higher % cell viability for porous HA/PCL scaffold compared to pristine porous scaffold. In the study carried out by Huang et al [25], 3D printed PCL/multi-walled carbon nanotubes (MWCNT) scaffolds were produced by blending 0.25, 0.75 and 3 wt. % MWCNT. This study indicated that 3D printing process led to alignment of MWCNTs which improved the in vitro response in cell culture. Authors reported that aligned MWCNTs were embedded in the polymeric matrix which reduced cellular penetration of MWCNTs. In fact, Alamar blue assay indicated that % cell viability increased up to 3 wt. % CNT gradually on day 7 and 14. This led to a reduction of toxic response compared to scaffolds produced with earlier methods such as solvent casting, gas foaming, freeze drying [25]. In another study, PCL/HA/CNT nanocomposites with up to 3 wt. % CNT were 3D printed and the authors indicated that alignment of CNT in the nanocomposites improved cytocompatibility due to alignment of CNT in the polymeric matrix. Alamar blue assay showed that 0.75 wt% CNT incorporated 3D scaffolds had higher % cell viability on day 3 and 7. Moreover, mineralization was higher for PCL/HA CNT scaffolds compared to solely PCL or HA/PCL scaffolds [26].
Zhang and colleagues [27] produced 3D bioprinted alginate/chitosan scaffold with 0.5, 1 and 2 mg ml −1 concentration of GO. The results revealed that GO containing groups had higher shape fidelity. Liu et al produced polyesters with 'peptide-like' functional groups. The authors indicated that presence of hydrogen bonding improved the shape fidelity of these 3D printed scaffolds [28].
Since up to 2 wt. % of CNT incorporated 3D printed scaffolds were found to be cytocompatible, in this study for the first time, PLGA/HA/CNT nanocomposites were 3D printed by extrusionbased technique to produce biocompatible scaffold. Moreover, due to CNT incorporation, higher shape fidelity may be achieved for our porous scaffolds. Functional groups on CNT and HA tend to form hydrogen bondings which expected to increase the stability and shape fidelity after 3D printing This would enable achieving desired architecture more efficiently than CNT free porous scaffolds. Then, rheological properties of inks were analyzed to be able to determine the printing parameter of the inks and to evaluate the effect of additional CNT, and the printed scaffolds were examined in terms of their physical and in vitro properties with multiple experimental methods.

Preparation of nanocomposite inks
Nanocomposites were blended via melt mixing technique [26,29]. PLGA granules were placed into a Teflon beaker and heated up to 135 • C on a hot plate for 30 min. Afterwards, 5 wt. % of HA and three different concentrations of 0.5, 1 and 2 wt. % CNT powders were added into molten PLGA. The mixture was blended manually by means of Teflon spatula at least 20 min at three different time intervals to assure homogenous dispersion of CNT and HA powders [25]. The obtained compounds were cooled at room temperature and chopped into small pieces to feed to the 3D printer cartridge. The scaffolds including PLGA, PLGA with 5 wt. % HA, and 0.5, 1 and 2 wt. % CNT addition with 5 wt. % HA into PLGA were denominated as PLGA, PLGA/HA, PLGA/HA/CNT0.5, PLGA/HA/CNT1 and PLGA/HA/CNT2, respectively.

Fabrication of 3D printed nanocomposites
3D porous scaffolds were fabricated by using extrusion-based fused deposition modeling with Axolotl Biosystems A1 Bioprinter. A square-shaped porous structure was obtained with lay-down pattern of 0/90 • in which there is 90 • angle between successive layers. Layer thickness was chosen as 100 µm for adequate print quality [23,30]. The nozzle temperature with a speed rate of 5 mm s −1 was set as 110 • C for PLGA and PLGA/HA samples and set as 125 • C for CNT incorporated study groups. The pressure to drive the ink through nozzle of 400 µm was set as 3.4 ± 07 bar, 4.08 ± 0.7 bar and 4.76 ± 0.7 bar for PLGA, PLGA/HA and PLGA/HA/CNT samples, respectively. Table 1 summarizes the printing parameters and condition.

Strut size, pore size and porosity
The morphological structure of porous 3D printed samples was examined by optic microscope and Scanning Electron Microscopy (SEM) (FEI-Philips XL30). Strut and pore sizes were measured from ten different regions by using microscope image. The porosity of the porous 3D nanocomposites was measured by performed liquid displacement method with ethanol. Porous 3D samples were weighed (M i ) and placed in a graded cylinder containing a given volume of ethanol (V 1 ). After submersing for 24 h, samples were taken out and weight (M f ). Porosity was calculated by using the following equation: where ρ is a density of ethanol (0.789 g cm −3 ) V 1 is the given ethanol volume and V 2 is the volume of given ethanol and scaffold [10,31].

Fourier-transform infrared spectroscopy (FTIR)
Chemical structure of the 3D porous neat PLGA, HA added PLGA and both HA and CNT added PLGA scaffolds was identified via Fourier infrared (FT-IR) spectrophotometer (Nicolet FTIR Instruments, Thermofisher) in the attenuated total reflection mode (FTIR-ATR). Transmittance spectra of samples were collected with 32 scans with a wavenumber range varying between 4000 and 400 cm −1 with 4 cm −1 resolution, operating in transmittance mode using OMNIC software.

X-ray diffraction spectrometry (XRD)
Crystallographic structural properties of the fabricated 3D PLGA, PLGA/HA and PLGA/HA/CNT scaffolds were examined by using an XRD spectrometer (Rigaku D/MAX-Ultima+/PC). XRD spectra were obtained with Cu-Kα (λ = 0.154 nm) equipped xray radiation source operated at 40 kV and 30 mA current. The diffraction data was collected by scanning of 2θ angles between 3 • and 90 • with a step size of 0.02 • . Phase composition of scaffolds was detected [11,32,33].

Water contact angle measurement
Surface wettability of the 3D printed porous scaffolds was examined by using the static sessile drop method with a contact angle goniometer (CAM 101 KSV Instruments). Thereby, surface hydrophilicity of the scaffolds was characterized by measuring the degree of droplet spreading on the scaffolds' surface. 5 µl of deionized water was pumped from the needle mounted to the instrument on the circular shaped scaffolds with a 10 mm diameter and 2 mm height at ambient temperature. The image of droplets was captured with a digital camera 40 s after they drop. Surface contact angles of the droplet were measured from the taken images. The experiment was performed for ten different regions on the sample for each experimental groups [34][35][36].

Bioactivity studies
In vitro bioactivity of the samples in terms of the bone-bonding ability was evaluated by using simulated body fluid (SBF). SBF including various ionic concentration has similarity to that in human plasma were prepared based on the method developed by Kokubo and Takadama [37]. To predict bone-bonding tendency, the formation of apatite-like crystals on the sample surface was tested by immersion of samples in SBF for 28 d at 37 • C. Samples with dimensions of 8 × 2 mm 2 (D × H) were immersed in 15 ml of 1xSBF. SBF was changed every other day. At the end of the test, samples were taken out and washed with distilled water. After drying at 37 • C in an oven, images were taken with SEM (FEI-Philips XL30) to analyze apatite deposition on the scaffold surface [38].

In vitro biodegradation studies
Biodegradation rate of samples were tested by immersing cylinder-shaped samples with dimension of 8 × 2 mm 2 in phosphate buffer solution (PBS) (pH: 7.4), at 37 • C for four weeks. The 3D printed porous PLGA, PLGA/HA and PLGA/HA/CNT scaffolds were immersed in PBS with a weight to volume ratio of 1 g/20 ml and the PBS was changed every three or four days [18]. The remaining weight was measured by a sensitive scale (Mettler Toledo) [39][40][41]. The samples were weight and then immersed in PBS.
The specimens were extracted out and rinsed with distilled water each week. Then, samples were gently wiped and weighted. Following this, the samples were lyophilized for two days and maintained in a desiccator. The pH was also measured using a pH meter (Mettler Toledo) [42]. There were three samples for each group. The PBS uptake and weight loss percent was calculated according to the following equations: where W i is the weight measured before immersion, W w is the weight of wet state as soon as taken out from PBS at testing time and W d is the weight of the freezedried samples at a testing time.

Mechanical analysis
Electromechanical universal test machine (MTS C43.104) was used to evaluate the influence of inclusion of HA and both HA and CNT on mechanical properties of PLGA scaffold. Accordingly, 39 layered samples with the dimension of 12 mm in diameter and 4 mm in thickness were printed for mechanical analysis. Then, samples between crosshead plate were compressed with 10 kN load cell of which speed was set as 0.5 mm min −1 . Load/displacement data were recorded during test. Compressive modulus, compressive strength and strain at ultimate strength were calculated. Four samples were used for each study group scaffolds.

Cell culture and seeding
MC3T3-E1 cell line (Calvaria newborn mouse derived, ATCC CRL2593) was cultured in α-minimal essential medium containing 10% fetal bovine serum and 1% penicillin/streptomycin to carry out the cellular studies under 37 • C, 5% CO 2 and 95% humidified atmosphere condition. Cells with passage numbers in a range of 3-5 were used to assess biological behavior of the PLGA/HA/CNT nanocomposites. Scaffolds with dimensions of 6 mm in diameter and 2 mm in height were sterilized by UV radiation exposure for 45 min on each side. Then, samples were pre-soaked for 2 h in complete medium [43]. Subsequently, cells with 35 × 10 4 cell well −1 were seeded by pipetting cell suspension of 50 µl aliquots top of each scaffold in 48-well plate and allowed to attach on the scaffolds in incubator for 2 h [44]. Then, the wells were filled by adding 350 µl of complete medium on scaffolds [25].

Cell proliferation and morphology
The proliferation capacity of MC3T3-E1 cell line seeded on the 3D porous scaffold was determined by Alamar Blue assay on day 1, 3 and 5 of the cell culture.
Here, the resazurin molecule is used as an indicator of cell viability and proliferation through the reduction of resazurin by mitochondrial metabolic activity in living cells [45,46]. Briefly, the old medium on the scaffold-cell construct was refreshed with new culture medium including 10% Alamar Blue and then cellseeded scaffolds were incubated for 4 h. At the end of incubation time, 100 µl of solution from each well was transferred into 96 well-plates. The optical density of solution was measured at 570 nm excitation and 595 nm emission wavelengths with microplate reader (Bio-Rad, iMark™) [26,31].
In addition, to investigate cell adhesion and morphology of the cytoskeleton, fluorescence imaging of cells cultured for five days was performed using an inverted fluorescence microscopy at 5× and 10× magnification. After five days in cell culture scaffold were fixed in 4% PFA for 10 min followed by rinsing with PBS. Then, the samples were immersed in Triton-X (0.1% in PBS) for permeabilization for 5 min. The fixed samples were dipped in 1% BSA to block unspecific bonding. Finally, samples were stained with Phalloidin-iFluor 488 reagent for actin filament staining (F-actin) for 1 h and nucleus was counterstained with DAPI in the dark for 20 min [47].

Cell mineralization assay
Alizarin Red S (ARS) staining was employed to evaluate calcium formation on 3D printed scaffolds. After 24 h of cell seeding on different scaffolds, the culture medium was replaced by osteogenic culture medium which consisted of complete medium supplemented with 50 µg ml −1 L-ascorbic acid (Sigma Aldrich), 10 mM β-glycerophosphate and 100 nM dexamethasone [48]. The osteogenic inductive medium was changed every third day for up to 14 d. On day 14 after osteogenic induction, cellscaffold substrates were rinsed with PBS, then fixed with 4% PFA for 15 min. at ambient temperature. For the qualitative and quantitative analysis substrates were stained with ARS for 30 min. with a gently shaking/on an orbital shaker. Then the samples were washed with plenty of DI water until the excess dye was removed. Finally, dried samples were observed with an optic microscope. Calcium concentration in deposited minerals was quantified by adding 10% acetic acid (Sigma Aldrich, A6383). The absorbance of the dissolved calcium nodules was measured via microplate reader at 415 nm wavelength.

Statistical analysis
All data were represented as mean ± standard deviation. Firstly, the normal distribution of the variance was checked by Shapiro-Wilk test, the differences between experimental groups were then analyzed by applying ANOVA with post-hoc Tukey's multiple comparison test using GraphPad Prism software. The significant level of p value * < 0.05, * * < 0.01 and * * * 0.001 was referred to as statistically significant difference. All experiments were repeated at least three times.

Rheological measurements
A successful 3D printed construction of scaffolds depends on the rheological properties of the prepared ink. The ink must flow and should not clog the nozzle during printing while being steady in desired shape after deposition on the building platform [49]. To evaluate the flow behavior of the molten samples, rotational analysis was conducted. This enabled determination of the optimized printing parameters.
To determine linear viscoelastic region, it is necessary to carry out an oscillatory amplitude sweep test [25]. The amplitude-sweep test graph in figure 1(A) demonstrates the dependence of storage modulus of PLGA, PLGA/HA and PLGA/HA/CNT nanocomposite inks on the % strain. The storage modulus increased with the rise of CNT loading. Additionally, the higher storage modulus lead to improvement of the shape fidelity of the 3D printed scaffolds which was also observed in previous studies [50]. Moreover, the nanocomposite inks had broad linear viscoelastic region. From this region, a strain value of 1% has been chosen to carry out frequency, temperature and strain sweep rheological tests.
The angular frequencies, ω, were swept from 100 to 0.1 rad s −1 at temperature of 110 • C. Figure 1(B) displays the storage modulus of samples as a function of frequency under a constant strain of 1%. This figure also shows that as frequency increased, storage modulus also elevated which indicates viscoelastic behavior of the biomaterial. Moreover, with the addition of HA and CNT, storage modulus was increased compared to pristine PLGA ink [25]. Figure 1(C) displays change in viscosity depending on temperature. The viscosity of the samples was decreased with the increase of the temperature for each experimental group [51]. This figure also indicated that the addition of both HA and CNT into PLGA increased the viscosity for all the temperature range of measurement.
The shear thinning properties of the biomaterial was assessed under rotational shear rate by recording of the change in viscosity [25]. According to figure 1(D), each group exhibits Newtonian flow regime at low shear rate which means inks resist to flow through the nozzle of printer at low shear rates. However, above a certain shear rate, the viscosity of the samples began to decrease linearly as the shear rate was increased. Non-Newtonian flow regime can be observed for all groups at higher shear rate, which means the materials exhibit shear thinning properties. Shear thinning properties are retained after addition of CNT. This is because polymeric chains and CNT in the polymeric matrix are aligned in the direction of flow with increasing shear rate which leads to reduction of resistance to flow [52]. This phenomenon is necessary for achieving a printable ink since the compound should have a suitable viscosity to be easily dispensed from the nozzle [53]. The inks with the highest wt. % of CNT had the highest viscosity for each shear rate, however the difference was marginal between study groups. It can be seen from the figure 1(D) that a relatively higher pressure is needed to extrude CNT reinforced samples without affecting printability of the scaffolds [25].

Strut size, pore size and % porosity of the scaffolds
The photo image of the 3D printed scaffolds was shown in figure 2(A). The strut and pore sizes of 3D printed specimens were calculated by images taken from optical microscope and SEM. As demonstrated in figure 2(B), the strut size was dramatically reduced by 8.5% after loading with 0.5 wt. % CNT [20]. The decrease of strut size with addition of CNT may be due to hydrogen bonding between CNT, HA and PLGA which leads to increase in viscosity. This ultimately enables biomaterials to be printed with enhanced shape fidelity. In the literature, hydrogen bonding between carbon-based materials such as graphene oxide is explained to improve shape fidelity [27,28,31]. Moreover, it is reported that a welldefined structure with high precision was obtained by decreasing strut size as a result of holding PCL polymer chains by CNT [54]. Additionally, pore sizes were calculated as in figure 2(C). The pore size of the PLGA with a mean value of 591 ± 37 µm was statistically significantly smaller than that of the nanocomposite samples. A drastic increase in pore size was observed after blending PLGA with HA and CNT, which increased to 705.8 ± 32.6 µm for PLGA/HA/CNT0.5. The pore sizes were between 450 and 700 µm, indicating a bone-like, interconnected porous structure [11]. The pore size within the specified range was also reported to promote cellular response including adhesion, proliferation of the cells and synthesis of extra cellular matrix [7]. Figure 2(D) shows that PLGA, PLGA/HA and PLGA/HA/CNT0.5 scaffolds had similar % porosity which were 42.01 ± 3.3, 46 ± 4.7 and 49 ± 2.2, respectively. On the other hand, there was a dramatic increase in % porosity with a 30% increase when 2 wt.% of CNTs were added to the PLGA/HA nanocomposite. The increase of porosity is also found to be relevant to decrease in strut size which directly affected the ultimate % porosity.  [55]. The sharp peak at 1746 cm −1 was for -C=O stretching vibrations of ester carbonyl group [48]. The absorption band around at 869 cm −1 was attributed to -C-H bending vibrations [56]. The absorption peaks at 1080 and 1165 cm −1 were assigned to stretching vibration of -C-O and asymmetrical stretching of C-O-C ether group [42,[57][58][59]. The peak at 1045 cm −1 was corresponding to -OH bending in -CH(CH3)-OH end groups [60]. The peaks at 1376, 1427 and 1450 cm −1 correspond to bending vibration of -C-H from CH 2 , CH 3 and glycolic acid group for PLGA [48,61].

FTIR analysis
After addition of HA into PLGA matrix, four new signatory bands of phosphate group (PO −3 4 ) were observed in the composite structure. The characteristic bands of HA in PLGA matrix at 568 cm −1 (ʋ 4 ), 602 cm −1 (ʋ 4 ) were ascribed to P=O bending vibration of phosphate group [62]. The absorption band of 960 cm −1 (ʋ 1 ) and 1045 (ʋ 3 ) indicate the symmetric and asymmetric stretching vibrations of P-O bonds in phosphate group (PO −3 4 ), respectively [63,64]. As it has been seen, in PLGA/HA nanocomposite' spectra, -C=O stretching peak split up into two peaks. While the peak at 1746 represent free C=O vibration, the second peak appeared at 1706 cm −1 indicated the hydrogen bonding of carbonyl group of PLGA and -OH group of HA [65]. However, the characteristic peaks of CNT were not observed due to its low concentration and overlap with PLGA and HA peaks [66].   peaks suggests that the incorporation of crystalline HA into 3D PLGA and PLGA/HA/CNT matrix [32,63].

Water contact angle measurement
An appropriate surface hydrophilicity promotes cell adhesion and proliferation on the scaffolds [57,58]. The water contact angle of experimental groups was calculated to assess whether the addition of CNT increased the hydrophilicity of the PLGA and PLGA/HA. As plotted in the figure 5, pristine PLGA had a contact angle of 101.4 ± 4.2 • , indicating a hydrophobic nature. The presence of HA in the nanocomposites drastically improved the wettability, with the contact angle decreasing to 78.6 ± 3.3 • . In addition, a more hydrophilic surface has been achieved by the incorporation of CNT. The contact angle of the droplets was significantly reduced to 66.8 ± 4.1 • , 68.9 ± 2.8 • and 69.9 ± 3.4 • for PLGA/HA/CNT0.5, PLGA/HA/CNT1 and PLGA/HA/CNT2, respectively. This is attributed to the presence of hydrophilic -COOH groups in the CNT structure. However, no significant difference was observed in the contact angle of the CNT loaded groups [41,68].

Mechanical analysis
An appropriate mechanical strength is an important requirement for scaffold to be able to withstand physiological stress during regeneration process. The compressive modulus of the samples calculated from the linear region of stress-strain curve presented in figure 6(A). The compressive modulus of the scaffolds significantly increased with the incorporation of HA into the scaffolds as shown in figure 6(B). A compressive modulus 548.5 ± 15.4 MPa was achieved for PLGA/HA/CNT2. The compressive modulus values were in the range of compressive modulus of cancellous bone (100-5000 MPa) which implied suitability of these scaffolds in terms of their mechanical properties [25].

In vitro biodegradation test
The degradation properties of the samples including remaining weight %, swelling % and pH change are represented in figure 7. The images of degraded samples in figure 7(A) shows that PLGA scaffolds were totally degraded at week 4 whereas HA and CNT added scaffold maintained their structure. At week 4, PLGA/HA/CNT scaffolds mostly remained their structure; however, PLGA/HA scaffolds started to lose their shape. According to figure 7(B), PLGA scaffolds began to degrade as of week 2. PLGA/HA/CNT1 scaffolds almost maintained their weight after week 4. This group also had the highest degree of swelling. Degradation studies indicate that 1% CNT loaded scaffolds can provide structural integrity which means sufficient mechanical support over four weeks. The literature indicates that structural integrity increased due hydrogen bonding between CNT, HA and PLGA [27,28,31,52]. The variance in pH values was examined every three or four days throughout the degradation process. The results in figure 7(D) show that the addition of CNT especially for 1% loading leads to higher pH values throughout the study. This effect may prevent the inflammation caused by acidic surroundings near the degradable bone substitutes. Moreover, neutral pH around 7 up to day 21 prevents the inflammation [69].

In vitro bioactivity test by SBF
SEM was used to visualize the morphology of the surface of the samples after immersion in SBF for 28 d. SEM images in figure 8 display the bone-like     apatite growth on 3D printed PLGA, PLGA/HA and PLGA/HA/CNT2 scaffolds. According to SEM analysis, there was almost no apatite formation on PLGA scaffold surface, which in turn suggests a highly biologically inactive scaffold. After addition of nano-HA into PLGA matrix led to increase in the bioactivity which is evident from needle-like apatite formation on the surface at the end of 28 d of incubation in SBF. For PLGA/HA/CNT2 samples, the nucleation of crystals occurs due to the presence of negatively charged carboxyl groups of functionalized CNT [8]. The formation of both hexagonal columnar (packed rod-like) and needle-like crystals of apatite may occur due to a relatively more acidic environment of PLGA/HA/CNT2 compared to that of PLGA/HA [70,71]. Figure 9(A) shows Alamar blue reduction of cells after incubated for 1, 3 and 5 d. The result indicates that adding 0.5, 1 and 2 wt. % CNT and 5 wt.% HA promoted the cellular proliferation from day 1 to day 5 [23]. The presence of CNT on the surface of the porous 3D nanocomposite enhanced cell affinity which led to cell proliferation. It is reported that Alamar blue reduction of cells on PLGA scaffolds were lower than control group during cell culture studies. Moreover, it is observed that cell proliferation on porous 3D printed PLGA/HA/CNT2 nanocomposite revealed statistically significant difference compared to porous 3D PLGA and PLGA/HA scaffolds. On day 7, 1% CNT reinforced nanocomposite scaffolds showed significant increase by serving as a favorable surface for cell attachment. Even for relatively high CNT concentrations, the scaffolds were found to be cytocompatible. This is probably due to alignment of polymeric chains and CNTs during the printing process. CNT tips were embedded in polymeric matrix which reduced penetration of CNT tips into the cellular membrane which prevented cell death [25]. The cells were interacting with hydrophilic functional groups of CNT which was favorable for cell adhesion and proliferation. This is illustrated in figure 9(B).

Cell proliferation and morphology
After 5 d of cell culture studies, the fluorescence images of the cells were evaluated for cell adhesion as shown in figures 10 and S2. MC3T3-E1 cells living on the scaffolds loaded with 1 and 2 wt. % CNTs appeared to be significantly more abundant  than on the plain PLGA and PLGA/HA scaffolds. The intercellular bridges on the PLGA and PLGA/HA scaffolds were much fewer and poorer, with limited spreading. After mixing with CNTs, the expansion of the cell cytoskeleton on 3D porous scaffolds with CNTs was much greater. This finding explains that the favorable surface of the PLGA/HA/CNT scaffolds led to enhanced cell adhesion and survival, which is consistent with the results of the Alamar blue study. Figure 11(A) demonstrates the calcium concentration on the surface of scaffolds on day 14. With the blending of HA into PLGA, there was a slight increase in the calcium concentration. On the other hand, a significant increase in calcium concentration was observed with the addition of CNTs at three different loading levels compared to pure PLGA for 0.5,1 and 2 wt. % of CNT. These results demonstrate the effective role of CNTs in promoting cell differentiation. As a result of the synergistic effect of the combination of CNT and HA, there was a remarkable increase in the accumulation of calcium content. Figure 11(B) presents the ARS staining samples images taken by optic microscopy at 20× magnification. The red color is the evidence of deposited calcium layer on the samples. While very little calcium deposition was observed on PLGA surface, calcium nodules can be clearly seen on PLGA/HA/CNT1 scaffolds. This result indicated that PLGA/HA/CNT1 could improve mineralization in comparison to PLGA/HA scaffolds [68]. Since HA also contributes to alizarin red staining, in the future Alizarin staining should be also performed without cell seeding. This way, solely effect of cell mineralization can be determined.

Cell mineralization
The nanocomposite inks with enhanced physical interactions have been used to achieve high porosity with improved shape fidelity by 3D printing technique. Besides this, enhancement of surface hydrophilicity promoted cell proliferation and differentiation on the developed 3D porous nanocomposite scaffold. Overall, 3D printed porous PLGA/HA/CNT nanocomposites had suitable physical, chemical and biological properties for bone regeneration.

Conclusions
In this study, porous 3D printed PLGA, PLGA/HA and PLGA/HA/CNT with 0.5, 1 and 2 wt. % CNT were produced successfully by using extrusion-based 3D printing technology. Strut size was reduced, and pore size was increased by the incorporation of CNT in the scaffolds. This was explained to be due to introduction of hydrogen bonding after incorporation of CNT which also increased shape fidelity of the scaffolds. Additionally, hydrogen bonding improved stability of the scaffolds in the PBS medium. By 3D printing technique, even at relatively higher CNT concentrations, highly biocompatible scaffolds were obtained with high degree of mineralization. This is due to alignment of PLGA chains and CNT in the direction of printing which also ultimately reduced cellular penetration of CNT tips and improved biocompatibility. In conclusion, 3D printed porous PLGA/HA/CNT nanocomposites were found to be promising candidates for bone tissue engineering applications.

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.