Voxels/Unit Cell Porosity (Phase I approach)
General Unit Cells
Complexity with Unit Cell Design
During Phase I of our project we identified scaffold requirements based on our human spinal cord research, such as biodegradability, biocompatibility, mechanical properties, as well as suitable macro- and micro-architectures. Consideration of these factors led to polycaprolactone (PCL) being the material of choice. PCL is an FDA-approved material, commonly used in medical implants. It is non-toxic, bioresorbable, and has suitable mechanical properties. Within Phase I, we also defined the macro- and micro-architecture of our scaffold, which consisted of gyroid-shaped unit cells iterated within an open path with a core design, as indicated in Figure 1. Due to COVID-19 restrictions, we were unable to perform wet-lab testing of our design, however, we conducted Finite Element Analysis (FEA) of our scaffold, as well as a degradation model. From the degradation model, we calculated that our scaffold should have a minimum molecular weight, Mr = 9.94kDa to last for a year, the period required in which axon regeneration occurs("Team:KCL UK - 2020.igem.org", 2021).
Figure 1: Scaffold Design from Phase I.
Within Phase II of our project we aimed to print our scaffold and conduct wet-lab experimentation to provide further validation for our design. However, we discovered that our initial scaffold design was too complex to feasibly print. As such, we proposed a new micro- and macro-architecture, i.e., a cross-hatch lattice iterated within an open path without a core design, as is indicated in Figure 4. Following these changes, we printed the scaffold and tested our design via a compression test, the results of which are shown here.
The following document outlines the scaffold engineering process underlying the aforementioned changes to the Phase I design, as well as the validations performed of our new design in Phase II. Initially, we will discuss the changes to the scaffold micro-architecture, followed by the macro-architecture, with consideration of pore size, porosity, and unit cell designs. Subsequently, our printing process, including the parameters we used, and the testing process is discussed.
Table 1: Table comparing the scaffold from Phase I to scaffold from Phase II.
The micro-architecture, such as the porosity, of our scaffold design, plays a key part in promoting axonal growth and regeneration. Pores allow the manipulation of stiffness and cell adhesion properties in our polycaprolactone scaffold (Shahriari et al., 2017) to ensure the scaffold mechanical properties match those of the spinal cord. The porosity percentage, therefore, must be adjusted according to the elastic properties required by the spinal cord. A porosity percentage of 60% has been shown to give polycaprolactone an elastic modulus of approximately 1.48 MPa (Guarino et al., 2007) which matches the experimentally obtained value of 1.40 MPa for the average elastic modulus of a spinal cord (Mazuchowski and Thibault, 2002).
The pore size is also another important consideration for our scaffold design. The ideal pore size would be within the range of 100µm and 400µm (Bayram et al., 2019). The 3D printing of our scaffold design can result in variations in the printed scaffold, especially during the cool down of layers, and therefore cause the resultant pore size to be smaller (Cubo-Mateo and Rodríguez-Lorenzo, 2020). Additionally, because our scaffold will be coated with a PVFP-5 bioadhesive, the pore size would become smaller than what was printed. To anticipate these factors, the pore size of our scaffold design is 500µm. The decreased pore size following the coating of our adhesive would then ideally fall into the optimal range.
The pore structure must be interconnected and continuous to match the properties of native spinal cord tissue and create the best environment for axonal regrowth and regeneration (Madigan et al., 2009). This type of pore structure is important for the nutrition and proliferation of regenerating cells. It promotes the formation of new blood capillaries for the supply of oxygen and aforementioned nutrients to the spinal cord tissue which better facilitates the bridging of the spinal cord (Loh and Choong, 2013).
Additive manufacturing (3D printing) has many benefits over traditional scaffold fabrication methods, including freeze-drying, solvent-casting, and phase separation (Jammalamadaka and Tappa, 2018), because of its ability to produce scaffolds with regular, complex, microscale geometry (Kundu et al., 2013). There are a number of techniques within computational modelling to implement micro-architecture, including computer-aided design (CAD), image-based methods, and topology optimisation algorithms (Ambu and Morabito, 2019). Of these, CAD was selected as an appropriate choice, being relatively simple.
One method of creating porosity within a CAD environment is through the application of unit cells. These unit cells are analogous to ‘building blocks’ (May, 2006), whereby they can be placed in a repeatable manner to create a predefined macro-architecture geometry. The process of applying these cells to a shape is as follows:
Figure 2: Method of creating unit cells within the CAD software.
Figure 3: Schawarz primitive minimal surface.
Solid scaffolds consisting of unit cells with sharp edges such as cubic lattice have been shown to possess machining difficulties (Shin et al., 2012); unit cells with curved edges and pores may be able to circumvent these issues. One such unit cell shape is the Schwarz Primitive minimal surface (Schwarz P). The Schwarz P surface is an example of a triply periodic minimal surface (TPMS), a term for surfaces defined by the minimal area between a set of boundaries, where the sum of the principal curvatures at each point is equal to 0. These surfaces also possess crystallographic group symmetries and are found within many naturally forming constructs such as biological membranes (Deng et al., 1998) and block copolymers (Jiang et al., 2003). In particular, Schwarz P cells are thought to possess the largest fluid permeability and the highest load bearing capacity across all other TPMS (Jung et al., 2005; Felfel et al., 2016). Additionally, researchers have demonstrated the Schwarz P surface to have excellent mechanical stability (Shin et al., 2012). Although there is very little research investigating the viability of Schwarz P surface scaffolds in vitro or in vivo, some preliminary research has indicated high cell viability (>95%) for cells adhered to the scaffold surface (Rajagopalan and Robb, 2006). Although theoretically a promising option, such scaffolds are limited by the need for fairly advanced printers in order to print this complex shape on a small enough scale.
Figure 4: Kagome structure.
A kagome structure consists of a combination of interlaced triangles where each lattice point has four neighboring points, giving a hexagonal appearance. Naturally, such a design has an abundance of interconnected and open pores, as well as a high surface-to-volume ratio, making it ideal for application within a scaffold. Additionally, such structures are thought to possess excellent mechanical properties with respect to their lightweight design (Wang et al., 2003; Hyun et al., 2003), as well as superior load-carrying capacities and lower softening rates than alternative pyramidal and tetrahedral structures (Moon et al., 2014). Researchers have also investigated kagome structures in regard to two key features of scaffolds, the fracture toughness, and compressive stress strength, finding them superior to both triangulated and octet structures (Fleck et al., 2010; Hyun et al., 2008). Similar to the Schwarz P surface, little in vivo or vitro investigation of kagome scaffolds has been performed. Nevertheless, researchers such as Cho et al. (2020) reiterated the aforementioned mechanical findings and also found evidence of excellent initial cell attachment to this structure, as well as superior cell proliferation in comparison to a standard log-based lattice.
Figure 5: Gyroid structure.
Of the number of unit cell geometries that may be used (some of which were described above), the shape chosen within Phase I was the gyroid. Similar to the Schwarz Primitive minimal surface, the gyroid is a triply periodic minimal surface—i.e. it has a mean curvature equal to zero and repeating structures in three coordinate directions (Rajagopalan and Robb, 2006). It is particularly advantageous for use in tissue engineering, owing to its high interconnectivity and mechanical integrity regardless of degradation (Germain et al., 2018). Further, it has been shown that micro-pores are introduced abundantly throughout the degradation of gyroid structures (Jin et al., 2019)—this is an important factor for cell attachment properties, and hence axonal regrowth (Shahriari et al., 2017).
In our initial scaffold design, the macro-architecture was filled with gyroid-shaped unit cells in order to generate a scaffold with interconnecting pores throughout. However, our current project is limited to the use of Fusion Deposition Modelling due to printer availability, an approach possessing a comparatively low spatial resolution. In this regard we found that the chosen software was unable to slice the mesh to print the scaffold, or even if this did succeed, print thin enough layers of our scaffold to generate an output sufficiently similar to the desired design. As such, this unit cell was no longer an appropriate design choice and simpler alternatives were investigated.
During our meeting with Dr Jia An from Nanyang Technological University, Singapore, he suggested looking into Selective Laser Sintering (SLS) printers since they have a better resolution and do not require supports that would allow us to print smaller pores. Following that we have done extensive literature research to establish which printing technique SLS or FDM would be more suitable for our scaffold.
Table 2: Comparison of Selective Laser Sintering (SLS) and Fusion Deposition Modelling (FDM) printing methods ("Introduction to FDM/SLS 3D printing | Hubs", 2021)
Overall SLS printing has an advantage in the printing parameters allowing it to print more intricate details. However, in the bigger picture FDM suits our project more. FDM printers are more widely accessible meaning that our therapy could be more widely distributed. FDM printing temperatures are lower, meaning that the scaffold does not require long post-processing making the scaffold ready for use within a few hours from the CAD design to the final product. Using FDM printing would also allow us to recycle excess PCL filament more efficiently than the PCL powder needed for SLS, making our therapy more sustainable and cost-effective.
Known commonly as the ‘cross-hatch’ method, log-pile approaches are the most common within tissue engineering for producing porosity in scaffolds (Meng et al., 2020). This scaffold design is often used alongside FDM printing (Kelly et al., 2018), and is much simpler than its unit-cell counterparts. Despite appearing much more basic, this design is accessible because it does not require challenging (and often expensive) printing specifications, yet still matches pore requirements, such as interconnectivity (Kelly et al., 2018). A further benefit is that the mechanical properties of cross-hatch structures can easily be tailored via altering a range of parameters - unlike unit cells, which are limited to features such as voxel size. Some of these features include log diameter, log spacing and the orientation (angle) between adjacent layers (Kelly et al., 2018). Specifically, a number of studies have looked into the effect of layer orientation - the angle at which adjacent layers are printed. Experimental data has determined that scaffolds produced in a 0-90 manner were stiffer and less plastic (had a smaller range of deformation beyond the elastic range) than scaffolds with rotating orientations of 0–45–90–135 (Kelly et al., 2018; Moroni et al., 2006).
To ensure our design was suitable for FDM, the log-pile/cross-hatch design was adopted in order to create interconnected pores throughout the scaffold. The design consists of 3D printing strips of PCL adjacent to each other with spaces between them. The 3D printer then rotates the angle by 90° and adds another layer of PCL strips. By doing this continuously up the scaffold it creates intersecting sets of parallel lines that make up the microarchitecture. Printing parameters include printing temperature and bed temperature and they are shown to be important in pore size, pore shape and efficacy of the scaffold (Soman et al., 2012). Through a process of trial and error, we decided on a printing temperature of 85°C and a bed temperature of 45°C. This ensured that the pore dimensions were consistent throughout the scaffold which is important in ensuring axonal regrowth.
The macro-architecture of a scaffold is its overall macro-scale geometry. Scaffold macro-architectures can vary in shape with the following five macro-architectures being studied by Wong et al. (2008): channel, cylinder, tube, open-path with core, and open-path without the core. The most commonly seen designs are scaffolds that have channels embedded within them. Moore et al. (2006) designed a multi-channel scaffold made of poly lactic-co-glycolic acid (PLGA). The results showed that the scaffolds lasted for 30 weeks in vitro before degrading fully. However, from Wong et al., the preferred designs were the open-path with core and the open-path without the core. These both displayed benefits to spinal cord injury progression when the defect length was measured. Axonal regeneration was seen in both of these open-path designs. Noticeably, fibres were seen crossing the defects. The open-path designs performed better than the channel, tube, and cylinder designs because there is space within these two, allowing for nerve roots to merge in the defect area. From the work completed last year by Renervate, it was shown that all five scaffolds would be suitable as a scaffold—in terms of mechanical yield, as none of them exceeded the yield strength of 17.82 MPa. In order to determine the best scaffold, we simulated each type and evaluated the following properties: solid von Mises stress, solid von Mises strain, displacement, and applied force. Comparing and evaluating the designs, the two open-path designs were the most optimal, thus agreeing with Wong et al. (2008). The open-path without core design was chosen to have the cross-hatch method of implementing pores, to be completed in Autodesk Fusion 360.
From Phase I, we designed our scaffold that would be ready for validation in Phase II. Having approached Professor Trevor Coward and Giovanni Gonnella regarding the printing of our PCL scaffold using their 3D printers, we encountered a problem with the design of our original scaffold. They informed us that the device was too complex to process and print. This STL file was ~1 GB in storage, which was not able to be sliced. To resolve this issue we had to redesign the scaffold. We decided to adopt Dr Lorenzo Veschini’s method of creating a log-like structure within the scaffold, known as cross-hatching. This is shown in the figures below.
Figure 6: The redesigned PCL Scaffold.
This design was then sent to Professor Coward and Mr Gonnella for printing. The printer available to us was the Cubicon. Initial attempts to print with the Cubicon were met with difficulty. As a control to see whether the scaffold prints at certain parameters, we printed a scaffold using PLA, which is much easier to print. The PCL was melting at a much lower temperature than expected which led to the material spreading out on the printing bed. Subsequently, this led to the covering of the pores. This issue persisted and it required a trial and error approach with changing the printing parameters to optimise the printing. After tampering with the parameters, the scaffold was printed successfully, using the following final values: the printing speed was decreased to 6 mm/s, the moving speed was 100 mm/s, both fans inside of the Cubicon were turned on to 100% capacity. Other variables include the printing bed at 45°C, and the extrusion temperature was at 80-85°C. This newly printed scaffold has an approximate printing time of 3 hours compared to 30 minutes for the equivalent PLA scaffold. PCL was expected to have the same printing time as PLA however, due to the changed printing parameters aggregated leading to an increased printing time.
Figure 7: The load extension graph resulting from our compressive tests.
Thanks to Dr Carlo Seneci and Mr John Bason, we were able to use the King’s College London Biomedical Engineering department’s Instron machine at St Thomas’ Hospital. Here, we carried out compressive tests on two of our printed polycaprolactone scaffolds with loads of up to 100N. We were able to confirm that our scaffold can withstand a load of 100N without being deformed. As the graph above does not display any signs of plateauing or peaking we are unable to extract values for any specific mechanical properties. Though we expect the load on the scaffold once implanted in the spinal cord to be greater than 100N, due to the limited lab access we had during this iGEM season we wanted to take full advantage of any opportunity available to us to test our scaffold mechanically.
Based on our literature review, in the future, we would complete further mechanical testing on our scaffold to obtain a stress-strain graph where we would be able to extract values for Young’s Modulus, yield stress, ultimate strength, and strain (Wieding et al., 2013). These values could then be used to compare with respective values for the spinal cord to ensure they match. We would also complete creep and recovery tests, whereby a load would be applied on the scaffold for a specified length of time and then taken away. This would give us the hysteresis curve for our scaffold which we could use to see how long the scaffold would be able to retain its mechanical properties and shape under a specific load (Niaza et al., 2017).