# A hierarchically ordered compacted coil scaffold for tissue regeneration

#### ByAUTHOR

Aug 28, 2020

Our MEW device, schematically represented in Fig. 1a, consisted of an air pressure pump, a heated extruder, a high voltage supply, and a moving stage collector. The morphologies of the PCL fibers extruded at 90 °C, under different stage speeds varying from 200 to 4500 mm/min were investigated (Fig. 1b). Inconsistent with the previously reported coiling instability from electrohydrodynamics32,35,41, the fibers showed periodic coils at a low stage speed, lower coil density at an increased speed, and eventually turned into straight lines once the stage speed was increased above a threshold value, namely, the CTS23,42,43. Hrynevich et al. provided details about the determination of the CTS value32. The CTS of the PCL fiber was approximately 3500 mm/min at 90 °C.

The periodic coil density was quantified and marked on the right side of the corresponding SEM images (Fig. 1b), assigned with a color bar. The coil density as a periodic property could be fitted as a function of the collector speed (Fig. 1c). For the coils printed at 90 °C, the fitting equations were proposed as Equation 3 and Equation 4 (the fitting computation process is shown in Supplementary note 1 and Fig. S1). From the equations, the theoretical numbers of coils at speeds of 200 mm/min and 400 mm/min (which cannot be manually counted due to experimental conditions) were 66 coils/mm and 31 coils/mm, respectively.

$$n = frac{{2sQ}}{{cv{uppi}r_1r_2}} – frac{s}{c}$$

(3)

$${mathrm{Coil}};{mathrm{density}}frac{n}{s} = frac{{2Q}}{{cvpi r_1r_2}} – frac{1}{c}$$

(4)

where s is the distance traveled by the collector; Q is the volume flow rate of the melted PCL; c is the course length of one coil; v is the collector speed; r1 and r2 are the minor and major axes of cross-section for fibers, respectively; n is the number of coils in distance “s”; and coil density (n/s) is the coil number per unit length.

Furthermore, the effects of temperature, distance, and applied voltage on the morphologies, diameters, and coil densities of the PCL fibers were also studied (Supplementary notes23 and Figs. S23). With increased temperature, the PCL melt possessed lower viscosity and accordingly showed a higher flow rate, which led to an increase in the diameter. The effect of temperature, and collector speed on the diameter of straight PCL fibers was found to be consistent with previous studies44,45. At the same stage speed (800, 1000 or 1500 mm/min), the PCL coil density increased linearly with temperatures from 75 °C to 100 °C (Fig. S2d). When the temperature was fixed at 90 °C and the distance/applied voltage increased from 4 mm, 3.5 kV to 16 mm, 6.5 kV, the coil density at 1000 mm/min increased from 10 coils/mm (Fig. 1c) to 37 coils/mm (Fig. S3f). The coil density along the stage orientation could also be tailored by both temperature and distance/applied voltage. It is noteworthy that Hochleitner et al. used a constant speed below the CTS to print sinusoidal patterns, where the wavelength and peak-to-peak distance of sinusoidal shapes were affected by the collector speed23,46. The tubular structures of “figure of eight” looping and a sinusoidal fiber pattern were also observed47. Various designed soft networks with waving fibers were demonstrated for engineering biomechanically and biologically functional soft tissues48. However, those studies applied constant speeds below the CTS, while various speeds (below the CTS) during printing have not been previously systematically explored.

Here, based on the established relationship between stage speed and coil density (Fig. 1b, c), a programmed stepped speed between 200 and 4000 mm/min was applied to print unidirectional (Fig. 1d, e) and bidirectional (Fig. 1f, g) parallel coil gradients. In Fig. 1d, the stage speed was increased from 200 to 4000 mm/min, and the distance between adjacent paths was 300 μm. The number of coils decreased with increasing stage speed and eventually disappeared. For the bidirectional path (Fig. 1f), the stage speed increased in one path from left to right and then inversely increased in the adjacent path. As shown in Fig. 1e, g, after repeating the unidirectional and bidirectional printing 10 times, macroscopic isosceles triangle-like patterns were created along each path. The enlarged images (Fig. 1h, i) showed the fiber morphologies at variable stepped speeds (200, 600, 1000, 2500, and 4000 mm/min) after printing 1 and 10 layers, respectively, where the single layer showed a similar result to that in Fig. 1b. Therefore, using programmed stage speed, triangle-shaped patterns with coil density gradients were successfully fabricated.

Next, by controlling the programmed stepped stage speed along the horizontal printing path, rhombic patterns with tunable geometry and coil density were successfully created. Different structures of the rhombic patterns were characterized by each block having an interstructural gap (g) and a long periodic diagonal length (d) (Fig. 2a). To print the pattern with g = 0.2 mm and d = 1.2 mm, as an example, the stepped stage speed was programmed as shown in Fig. 2b. For each step distance of 0.1 mm, the speed gradually decreased from 4000 to 200 mm/min and then increased symmetrically back to 4000 mm/min to complete one rhombic structure. Various rhombic patterns (p 1.11.3) were produced through these blocks, as shown in Fig. 2e–k, with the corresponding stage speed program listed in Fig. S4. The length of the short diagonal for every rhombic structure was ~316 μm, which was determined by the intrinsic coiling instability, while g and d can be tailored by programming the stage speed. The faster the stage speed decreases/increases, the shorter d becomes. A total of 10 layers were used for all patterns; accumulation of layers the corresponding to the pattern in Fig. 2e is shown from the SEM images in Fig. S5, which shows that not only the rhombic shape but also the 3D coil density can be controlled by the tunable layer number. Further combination of both horizontal and vertical printing with rhombus crossing at either their centers (p 2.1–2.3) or gaps (p 3.1–3.3) led to more sophisticated patterns with enhanced durability for practical handling. As shown in Fig. 2l–q, we created a structure with a constant periodic diagonal of 1.2 mm and a tunable inter structure gap from 0.2 to 1.8 mm. These scaffolds were complex systems composed of different patterns and coil compactness, which were arranged in the x and y directions with a highly ordered structure.

Furthermore, three rhombic structures (g = 0.2 mm, d = 1.2 mm) crossing with straight vertical lines, p 1.1, vertical stepped speed printing crossing at the center, p 2.1, or at the gap, p 3.1, were used to assess cell patterning. SEM images of the three cell-free samples, epifluorescence imaging, and confocal fluorescence z-stack imaging for cell-laden samples are shown in Fig. 3a–c. sMEW technology provides a programmable speed-control strategy for the fabrication of fibrous patterned substrates with a proper 3D culture environment for cellular microarrays. The cells grown along the rhombic patterned scaffolds formed rhombic patterned cell constructs with a depth of ~50 μm.

Cytotoxicity tests (Fig. 3d) and cell proliferation experiments (Fig. 3e) were also conducted for these rhombic patterned scaffolds. All scaffolds showed toxicity lower than 6% with no significant differences compared to the tissue culture plastic (TCP) low control. It was observed that the initial cell proliferation from days 1 to 5 for cells cultured on the p 1.1, 2.1, and 3.1 samples and TCP were similar, with no significant differences between the groups. After 7 days of culture, cell proliferation in the p 1.1, 2.1, and 3.1 groups was significantly accelerated compared to that of the TCP group, and showed excellent 3D cell culture capacity. No significance was found among the p 1.1, 2.1, and 3.1 groups. The patterned scaffolds yielded high densities of viable cells, and the cell densities were sufficiently high to achieve multicellular architectures. The cell culturing process on grid scaffolds with speeds of 200 mm/min, 600 mm/min, 1000 mm/min, 2500 mm/min, and 4000 mm/min is also presented in Supplementary note 4 and Figs. S68. The day 1 adsorption values for all samples at 450 nm from the CCK-8 assay are given to show the initial cell attachment (Fig. S9). Last but not least, due to its flexibility, the scaffolds can also be used as injectable scaffolds (Video S1).

The p 1.1, 2.1, and 3.1 scaffolds were evaluated in vivo by testing the IL-6 response in BALB/c mice for their special patterns. IL-6 is a pro-inflammatory cytokine with multiple functions and is believed to be a pivotal mediator or host responder for injury and infection49,50. TNF-α is another cytokine known to play an important role in cellular and inflammatory responses51. Overproduction of IL-6 causes neurological disease in the central nervous system52. Additionally, expression of TNF-α is related to nuclear factor-κB that involved in tumour initiation and progression53. Therefore, the inflammatory response from implantation of the patterned scaffolds was further evaluated by analyzing the relative expression (compared with each mouse group before implantation, considering the individual differences) of TNF-α and IL-6 in mouse serum from blood before implantation and on day 7 after implantation (Fig. 3f, g). For IL-6 expression, the p 2.1 sample at day 7 showed no significant difference compared with the control group. IL-6 expression was significantly downregulated in groups p 1.1 (from 17.03 to 8.88 pg/mL, a 0.52-fold change on day 7) and p 3.1 (from 32.22 to 14.76 pg/mL, a 0.46-fold change on day 7). The expression levels of TNF-α from mouse serum in groups p 1.1 and p 3.1 before implantation and on day 7 showed no significant difference. For the p 2.1 sample, the level of TNF-α secretion showed a significant reduction from 4.37 to 2.24 pg/mL (a 0.51-fold change). The expression levels, shown in Fig. S10, of TNF-α and IL-6 in mouse serum before implantation and on day 7 after implantation were all in the normal range, confirming the biocompatibility of the PCL scaffolds54,55.

The p 2.1 samples exhibited significantly higher IL-6 expression than the p 1.1 and p 3.1 samples, possibly because the p 2.1 scaffold is thicker (~120.8 μm at the thickest point) than the p 1.1 scaffold (~79.4 μm at the thickest point) and p 3.1 scaffold (~91.0 μm at the thickest point). Previous reports have proposed that thicker materials could produce a proportionally higher magnitude of foreign body responses56,57. The increasing expression of IL-6 from p 2.1 on day 3 seemed to inhibit TNF-α protein expression, resulting in the downregulation of TNF-α levels on days 3 and 7. This is in agreement with the hypothesis proposed by Starkie et al.58 that IL-6 can inhibit TNF-α protein production. In turn, the decrease in the IL-6 levels from the p 1.1 and p 3.1 samples also correlated with the upregulation of TNF-α expression with a lower inhibitory effect on day 7, where the level of TNF-α after implantation on the 7th day exhibited a slight increase compared with the control. On day 3, the TNF-α level exhibited the lowest expression as a transitional period.

In order to investigate the microscale mechanical properties of the hierarchical coil compacted scaffolds, a local compression test using atomic force microscopy (AFM) with a modified probe (Fig. 4a) was conducted. The mechanical properties were determined by microindentation using AFM. Young’s modulus based on the Derjaguin–Muller–Toporov (DMT) model was calculated from the withdrawal curves using the equation in Fig. 4b, inset. Fig. 4d shows the tip locations on one rhombic pattern, where 10 points were measured. Representative withdrawal curves measured at marked positions 6, 8, and 10 are shown in Fig. 4c. With increasing coil density (Fig. 4e) from tip position 1 to tip position 5, the modulus of the scaffold (Fig. 4f) increased from 0.22 ± 0.03 MPa to 1.10 ± 0.06 MPa and then decreased back to 0.24 ± 0.06 MPa at tip position 10 with decreasing coil density (Fig. 4e). The Young’s modulus data at each point were therefore presented (Fig. 4f) in correlation with the theoretical fiber density (Fig. 4e). The mechanical properties of the scaffolds were tunable based on the coil density. The tensile stress-strain curve measurements for the grid scaffolds with different speeds are also shown in Supplementary note 5 and Figs. S1112. Such coil compacted scaffolds can be designed on top of customized primary shapes with tunable secondary interior structures and mechanical properties46,48, which makes sMEW a great contribution to MEW technology.

To demonstrate the tailoring ability, a lumbar vertebra was printed using sMEW, according to the anatomical structure (Fig. 5a), with the design shown in Fig. 5b. The G code is shown in Supplementary note 6. The printed structure with a stage speed of 200 mm/min can be observed by optical microscopy (Fig. 5c) and SEM (Fig. 5e), where short-range compacted coils were also obtained (Fig. 5d, f). Because the running paths are too dense to observe the details in the 10-μm design (Fig. 5b), the overall schematic diagram of G-code with 100 μm spacing is shown in Fig. S13a, which can be helpful to observe the line and piece details clearly. The units of arcs and lines were used to fill in the simplified rectangular, parallelogram, round geometries with different sizes as small pieces. Then, these small pieces were assembled and connected with continuous printing to construct the whole vertebrae structure. The running path and corresponding zoomed-in SEM images of the lumbar spine-like scaffold (Fig. 5e) in different regions are shown in Fig. S13b–f, and all regions exhibited a similar coiling density. At the speed of 200 mm/min, the pore size was 9.13 ± 4.43 μm, as shown in Fig. S7b, which is similar to the lacunar-canalicular pores surrounding the osteocytes that are in the range of 0.1–10 μm59. After culturing with EGFP-hMSC-TERT cells for 7 days, the SEM images (Fig. 5g, h) revealed cell spreading on the scaffold (Fig. 5h). The lumbar spine-like scaffold-patterned EGFP-hMSC-TERT cells monitored by fluorescence microscopy on day 1 (Fig. 5i, j), day 3 (Fig. 5k, l), day 5 (Fig. 5m, n) and day 7 (Fig. 5o, p) showed cell proliferation over the culture time.