2.1. Design, Fabrication and Characterization of Functionalized 3D Structures
The starting point is represented by the optimized microstructures obtained in our previous studies [14
]. These structures were fabricated by LDW via TPP of IP-L780 photopolymer, using a specific design and fabrication procedure that promoted the cell attachment and interconnections, while retaining good structural integrity. Elements of the microstructures were either elliptical or hexagonal having a length of 80 µm, width of 40 µm, about 14 µm tall, separated by 20 µm cylindrical pillars, placed in the overlapping areas (Figure 1
The 3D structures were fabricated by two-photon polymerization, where the size of the volume pixel (voxel) is 2 µm width and 4 µm height [14
]. However, neighboring lines were not as tall as the voxel itself, as we imposed a 2 µm overlap to provide good structural integrity and polymerization degree.
Following dip coating of these 3D structures in Col/CT blends, the backbone architecture of the laser-imprinted structures changed drastically. In addition, the nature of these changes was found to depend on the blending ratios between Col and CT (Figure 1
b–d). Higher amounts of Col in the blend did not change the architecture of the 3D structure underneath; in this case, we found a continuous sheet that covered and “sealed” the whole structure and having several micrometric pores (as evidenced by the insets from Figure 1
b). On the other hand, the geometry of the structure was affected quite strongly when the amount of CT in the blend was increased. Ellipsoidal and hexagonal units were significantly deformed, while the vertical pillars remained almost unchanged (Figure 1
c). Moreover, remaining of the continuous sheet (similar with that observed in the case of pure Col) sealed some parts of the 3D structure underneath (as shown by the insets from Figure 1
c). With increasing CT content in the blend, the structure deformation was less dramatic and the free spaces between the unitary elements were no longer sealed by a coating sheet (as evidenced in Figure 1
The optimized structures design was determined in previous experiments [14
]. The layers are 14 µm tall and consecutive layers are 16 µm apart. The ellipses are 80 µm long and 40 µm wide, along the major and minor axes. Wall thickness is determined by the size of the voxel itself, which is 2 µm in this case. Col/CT functionalization determined strong deformation of the structures. For only Col functionalization, the structures are enclosed in a layer of collagen, and therefore rendering nonexistent structures porosity, from a practical point of view. Col/CT blends do not enclose the structures, but in turn determine strong deformations. As a result, porosity is difficult to measure and strongly varies for each sample. SEM micrographs do not indicate a noticeable increase in wall thickness.
The shrinkage is a general problem for the fabrication of complex micro/nanostructures and is mainly caused by the material densification as compared to the material before polymerization that results in volume reduction; the geometrical deformations of 3D printed structures mostly appear because of the surface tension effects that occur during the developing and drying processing steps [26
], which we described in Materials and Methods section. In our experimental conditions, the developer (PGMEA) was especially designed for IP-L780 polymeric structures fabricated using Nanoscribe technology and therefore its evaporation during sample developing process does not cause the shrinkage of the 3D structures. This idea is sustained be the fact that in Figure 1
a the uncoated 3D structure has a sharp architecture, with no signs of shrinkage. In contrast, structure deformation and especially shrinkage along the long axis of the ellipsoidal elements occurred for Col/CT functionalized structures (Figure 1
c,d). The shrinkage becomes stronger with increasing CT content in the blend (Figure 1
c,d). Since both Col and CT were dissolved in acetic acid, the solvent is certainly one reason causing the structures deformation. However, the fact that the structures functionalized with only Col were not deformed (Figure 1
b) means that the acetic acid is not the only factor responsible for structures’ deformation. Instead, we suspect that the combinations of Col/CT with a certain amount of acetic acid deformed the structures. As we stated before, it seems that CT was more influent than Col in this regard. An important aspect is that this deformation did not interfere with the cellular attachment. On the opposite, it seems that the design of the structures allowed for good volumetric cell migration, not only due to the free spaces between neighboring elements, but also due to the flexibility of the structures’ walls. More precisely, while the structures showed good mechanical resistance in the up and down direction, they are prone to side-to-side movements (Figure 1
c,d). This is advantageous for cell migration (Figure 1
d), but it also makes the scaffold prone to deformations determined by fluid movements and surface tension of any evaporating liquid, such the acetic acid used for preparing the Col/CT blends for structures functionalization. Moreover, structure deformation due to shrinkage may occur following the dipping procedure in the Col/CT solutions used for structures functionalization. Further investigations are required to determine and quantify the set of factors that result in structure deformation.
A major point is related to the chemical composition of the Col/CT coating that may account, at least to a certain extent, for samples’ wettability. In this regard, we argue that the dip coating technique used for the functionalization our structures is generally recognized as suitable for coating complex shapes and curved parts (like those in the 3D geometries presented in this study) that could not be coated by any technique [27
]. Moreover, the chemical composition of other complex structures functionalized by dip coating in Col and CT has been extensively investigated by Fourier transformed infrared spectroscopy, for both individual materials as well as in blends, and therefore the results could be easily extrapolated to our experimental conditions [28
]. In our case, the optical inspection provided by the detailed SEM analysis provide evidence on the degree of uniformity of the coating and of other particular aspects regarding the role of Col/CT blending ratio on the uniformity of the coating of the 3D structures and that were discussed in this section.
To avoid any influence of the structure geometry on the wettability of the samples, we measured the water contact angles on flat surface made by drop casting of the same materials as used for the functionalized structures. As such, water contact angles measured on Col/CT blends casted on flat photopolymer surfaces are depicted in Figure 2
a. Figure 2
b shows water drops on Col/CT blends with different blending ratios; the morphology of these surfaces and the seeded cells are also shown. The mean ± standard deviation of the contact angle measured on the material used for building the backbone (IP-L780 polymerized in the form of a flat surface) was of 105° ± 3°, indicating that the hydrophobic character of the backbone 3D structure. This is in fact a major problem of 3D printed structures for bone tissue engineering that generally have hydrophobic surfaces that impede the cellular attachment and differentiation. This was also our case, since the uncoated photopolymer which was in fact the backbone material of the 3D structures had the water contact angle above 90°.
The contact angles measured on Col/CT surfaces casted on flat photopolymer substrates were statistically lower than on the photopolymer surface, proving that the functionalization of the photopolymer with Col/CT increased the hydrophilicity of the samples. Moreover, the contact angles decreased with decreasing Col and increasing CT contents in the blend. Specifically, for pure Col surfaces the contact angle was of 87° ± 4°and continued to decrease with increasing CT content in the blend until reaching 43° ± 3° for pure CT surface. These results are in fair agreement with previously reported values. It has been shown that the water contact angle on Col substrates varies between 83° and 87° [23
]. It is also know that CT has a hydrophilic nature that supports the attachment and proliferation of bone-forming osteoblast cells as well as formation of mineralized bone matrix in vitro [20
In our experimental conditions although both CT and Col are hydrophilic but with quite high contact angles i.e., above 70°, the hydrophilic character of the Col/CT blends was more pronounced than each of the individual Col and CT components reported by the above cited literature. Most likely that also the morphology of the Col/CT surfaces played an important role in what concerns the wettability. As evidenced by Figure 2
, the contact angle for the Col/CT surfaces decreased with increasing CT content in the blends, while the surface becomes smoother.
Moreover, the cells attachment also increased with CT content in the blends; likely it was promoted by the increase of surface hydrophilicity, which in turn corresponded to an increase in surface smoothing. The influence of surface wettability has been intensively studied in the context of favouring the cell attachment [29
]. A hydrophilic surface is characterized by the fact that the forces of adhesion overcome the forces of cohesion at the surface of the material, thus supporting the attachment of cells. Previous studies reported that CT surfaces increase the protein adsorption, the cell attachment and act in the benefit of osseointegration [32
]. In our case, a similar effect of cell attachment promotion was noticed in case of high content CT in blend (Figure 3
The reason for which we presented the wettability measurements of flat Col/CT surfaces is because the height of the ellipses and hexagons within the 3D structures goes up to 25 μm and the thickness (related to the minimum volume of polymerized material named voxel [10
]) of few micrometers. These dimensions, much smaller than any water droplet, make impossible to actually measure the wettability directly on the 3D structures. Another reason is that any droplet of liquid used for contact angle measurements would be much higher in diameter that the lateral size of our structures which goes up to 200 μm (as observed from the scales in Figure 1
). An additional limiting factor for measuring the water contact angle directly on the 3D structures is that the complexity of the structures triggers the formation of large air pockets under the water droplet, therefore the contact angle measurement on such complex structures would not really reflect the real wettability of the structures’ walls on which the cells actually adhered. To sustain the validity of the contact angle measurements on flat surfaces made of Col/CT blends, we performed preliminary in vitro studies regarding the cells’ adhesion by fluorescence microscopy visual inspection. As it can be observed in the Supplementary Material
, the cells attachment on the flat Col/CT surfaces follows the same trend as the cells attached on the 3D structures illustrated in Figure 3
. Specifically, for both flat Col/CT surfaces and 3D structures functionalized with similar Col/CT blends, the cellular attachment increased with increasing CT content in the blends. This finding, along with the generally accepted hypothesis that the cellular attachment directly relates to substrate wettability, provides a reasonably strong base to correlate the wettability trends observed for the flat Col/CT surfaces to the wettability of the 3D structures functionalized with similar Col/CT blending ratios.
It was found that the biological performances of the photopolymerizable materials were limited by the characteristics of the photopolymers employed for the laser writing process, which were generally biocompatible, but did not possess specific properties (such as the water affinity, surface roughness, the ability to be modeled, biodegradability, elasticity or stiffness) that sustain cell growth and differentiation into functional tissues [9
]. As emphasized by the contact angle measurements (Figure 2
a), the IP-L780 photopolymer used in this study has a strong hydrophobic character, generally causing a rejection of water-based liquids such as the cell culture medium, thus affecting the interaction between the cells and the surface of the structure. The coating with Col/CT blends aims to improve the wettability of the IP-L780.
2.2. Biological Assessments of Functionalized 3D Structures
The morphological investigations of MG-63 osteoblast-like cells cultured on Col/CT functionalized structures are shown in Figure 3
. The cells were able to attach and grow on all types of structures. However, some differences in shape and density were noticed. In the case of uncoated structure with hexagonal units, numerous attachment points were available for the cells to bind to, resulting in a dense and interconnected network after three days in culture (Figure 3
a lower panel). On the other hand, in the case of the uncoated ellipsoidal 3D structure (Figure 3
a upper panel), the aspect of the cells network was slightly less dense. Probably due to less attachment points available, but favored by the dimension of the component units, the cells positioned themselves in the round units of the 3D structures, showing a stretched morphology.
Structures functionalization changed the cells behavior. An increase in cells density was clearly evidenced for Col/CT 50/50 (Figure 3
c) and Col/CT 0/100 (Figure 3
d) as compared to the non-functionalized (uncoated) structures.
On the Col/CT 100/0 samples, the cells formed a thick continuous monolayer at the surface of the 3D structures, which suggested that the cells were not able to bind to the morphological elements of the structures and rather cohered to each other, forming a dense network covering the whole structures.
This phenomenon can be explained by properties of the coating material such as its hydrophobic character (Figure 2
), viscosity of the coating solution, but also by the morphology of the 3D structures, namely the porosity (Figure 1
). Figure 1
b shows that the blend made of 100/0 Col/CT sealed the entire surface of the 3D structure, disabling many of its binding points that could have helped the cells attachment. While the viscosity of the Col/CT solution might have been too high and added to the hydrophobicity of the resulting coating, the introduction in vacuum of the collagen dip-coated structure we would only have unblocked the pores of the structures, whereas the inner walls of the structures would remain uncoated. Thus, although we would have facilitated the access of the cells inside the structure, the interaction with the inner walls would be the same as in case of the uncoated structures. The final result would then be partially false. Further studies on quantitative evaluation of structures’ porosity and viscosity of the dip coating solution would be helpful for better understanding the behavior of these samples and will be the subject of future experiments.
We chose this particular time interval of three days of cell culture for having an optimum cell number density attached on the 3D structures. The cellular behavior observed at shorter time intervals would not be reliable because there would be too few cells attached on the structures. On the other side, longer time intervals were not used for SEM analysis for avoiding too numerous cells that overlap on and within the 3D structures, since the cells divide and grow continuously; a large number of cells growing on and inside the 3D structures would impede detailed observations about the morphology and attachment of individual cells on the structures.
The cell culture density was controlled by using the same cell number/cell medium volume ratio for all samples. Moreover, the same cell number was diluted in the same volume. The attachment time of cells was one hour and afterwards more culture medium was added. This protocol was applied for all samples. The porosity of the structures is the same for all samples in each morphological category, as we used the same 3D architectures for all structures in the same group (ellipsoid respectively hexagonal units). The coating (Col/CT at different ratios) changed the surface (the walls) character of these structures and thus altering their response to cell culture medium. In our previous study [14
] we proved that cells were able to penetrate these structures, due to the dimension of its architectural elements. Depending of their morphology (ellipsoid or hexagonal), they enabled different attachment points for the cells, thus affecting the way the cells interacted with the structures. Here, the interaction between cells and the structures was altered by the wettability of the coated surface, rather than its porosity. In Figure 3
b is represented the Col/CT 100/0 sample, which was characterized by a strong hydrophobic character. As observed in Figure 3
b, the coating covered the exterior walls of the scaffold, sealing its pores. Thus, in this case, the cells laid on the coating and were not able to penetrate the interior of the structures.
The two architectures of the structures we used in this study, namely ellipsoidal and hexagonal, have also been described in a previous work [14
] where we proved to be a close relationship between the morphology of the substrates and the cell development. The structures enable different attachment points depending on their architecture. The hexagonal structure enables more attachment points for the cells whereas cells showed a fragmented morphology. On the other hand, the ellipsoidal structure enables a circular shape of the attached cells, in both cases the cells penetrating the interior of the structures.
The weak ability of the cells to bind to a hydrophobic material has been ascribed to an entrapment of air bubbles in these samples due to hydrocarbon contamination, which interferes with the protein adsorption and the cell-receptors [32
Our experimental findings integrate well into the broader scientific context of wettability-cell behavior. It has been shown that both wettability and topography of a surface can influence cell adhesion and spreading: cell adhesion increases with surface roughness, while cell spreading ability decreases [29
]. In terms of wettability, a medium character was shown to be the most favorable for cells growth, extreme hydrophilicity/hydrophobicity was shown to prevent the cells attachment.
Next, the soluble tetrazolium salt viability (MTS) assay was used to investigate the viability of the cells cultured on the Col/CT functionalized structures, as relative to uncoated (control) ones (Figure 4
a). Results showed an increase in relative viability with increasing CT content in the blend. Samples with preponderant CT in the composition revealed a biocompatible behavior (viability above 80%, according to [34
]). On the opposite, the cells cultured on samples with higher Col content in the blend revealed a decrease in viability, which can be due to the inability of cells to penetrate the structure because of the continuous sheet that sealed the whole structure that also provided less binding points for the cells (as already evidenced in Figure 1
The bone tissue forming potential of the Col/CT functionalized 3D structures was evaluated through the cells’ differentiation and mineralization ability after four weeks in culture. This was achieved by measuring the alkaline phosphatase (ALP) and osteocalcin production as biomarkers for these processes along with the alizarin red staining (ARS) assay as direct indicator of mineralized tissue [35
Our investigations showed an increasing ALP production with increasing CT content in coating blends (Figure 4
b). The maximum ALP amount was noticed for Col/CT blending ratio of 20/80. Moreover, a significantly higher amount of ALP was measured in case of 3D structures with ellipsoidal units compared to those with hexagonal units. By extrapolating these results to the morphological investigations discussed above, the increase of ALP production in the cells seeded on 3D structures with ellipsoidal unit is correlated with the elongated morphology of the cells (as evidenced in Figure 2
c,d upper panels). Thus, it appeared that the morphology of the ellipsoidal units favored the early differentiation of osteoblast-like cells.
Similar observations were made for osteocalcin and alizarin red staining (ARS) measurements (Figure 4
c,d). Osteocalcin is a late biomarker for osteoblast cells differentiation and its presence is directly proportional to the deposited mineral [36
]. On the other hand, ARS binds to the mineral depositions in the cell cultures and is a direct indicator of the quantity of minerals in the samples [36
]. Structures with ellipsoidal units showed better results in terms of ALP production than the structures with hexagonal units. Again, for both types of structures, the best results were obtained for 3D structures functionalized with Col/CT blends of 20/80 blending ratio, which resulted in highest osteocalcin secretion and ARS intensity. Furthermore, for 3D structures with ellipsoidal units the osteocalcin secretion and ARS intensities were higher than for the structures having hexagonal units. Similar to the experimental data concerning the ALP production, this result can be explained by means of the elongated cells morphology observed in the case of these samples.
As shown in our previous study [13
], the ellipsoidal structures enabled a better behavior of the cells’ attachment, as they followed the architecture of the structure’s walls. Although the hexagonal structure enabled more attachment points, their morphology was rather fragmented, affecting their biological behavior. These could explain the better osteogenic activity of the ellipsoidal structures.
Our results are in good agreement with previous studies on Col and CT effects on the osteogenesis. CT is a naturally occurring polymer with minimal foreign-body response, non-toxic degradation products, high biocompatibility, biodegradability and osteoinduction. CT is particularly attractive for bone regeneration because it promotes the attachment and proliferation of osteoblasts and the formation of mineralized bone matrix in vitro [20
]. For example, CT scaffolds exhibited high osteoconductivity in surgically created bone defects [39
]. CT accelerated the wound healing and showed antimicrobial properties [40
]. Nevertheless, CT is mechanically weak, with compressive modulus of pure CT structures significantly lower magnitudes than the cancellous bone [42
]. In addition, CT does not have the structural stability and is not able to preserve a specific shape in aqueous environments, which is a critical requirement in the implantation at a bone defect site. In order to improve its mechanical properties and structural integrity when hydrated, CT is sometimes blended with other materials, an interesting candidate for this purpose being Col [43
]. Col has been extensively used in the biomedical field mainly because it forms fibers with high strength and stability. In addition, Col is a good surface-active agent, with an important role in the functional expressions of cells leading to the formation of tissues and even organs [44
]. Col exhibits biodegradability, weak antigenicity, good biocompatibility and increases the osteogenic gene expression and the alkaline phosphatase activity in bone-forming cells [21
]. An important aspect for the engineering of bone tissue is that Col is the most abundant extracellular matrix protein of bone and plays a major role for its strength [46
]. Combinations of Col/CT fused with glycolic acid were used as bone graft substitute to promote bone tissue regeneration [43
]. The results of our work demonstrate that the Col/CT blends are suitable candidates for the functionalization of complex 3D structures for bone tissue engineering and that the structures functionalization can be achieved via a simple i.e., non-chemical dip coating process.
For morphological examinations by SEM, we selected the most representative samples, namely the extremes and middle of the Col/CT blending ratio range. Moreover, SEM is a qualitative investigation, while the other in vitro methods employed in this study have a statistic and quantitative character. A difference of 20% in the concentration of each polymer in the blend was generally used for these quantitative investigations. A difference of only 10% (between 60/40 and 50/50 or 50/50 and 40/60) fits into the 10% interval of error which is generally accepted in biological investigations. In our experimental conditions, an interval equal or below 10% between Col/CT blending ratios is not relevant for the observed trends of the quantitative measurements such as wettability and biological assessments nor for qualitative morphological investigations.