A direct-writing electrospinning system for designing complex architectures in tissue engineering

Recently, direct-writing electrospinning has been pursued to reach a higher accuracy and complexity in fiber scaffold fabrication compared to other extrusion techniques more frequently encountered in tissue engineering. However, to date, direct-writing electrospinning lacks a wide application to process materials such as nature-derived polymers, of huge importance in tissue engineering given their chemical properties similar to that of native tissues. In this work, a setup to perform direct-writing electrospinning was developed and demonstrated versatility and efficiency in obtaining submicrometric fibers and guiding their deposition along various types of paths and patterns, resulting in a user-friendly method to create structures closely resembling tissue architecture.


Introduction
By applying an electric field between a spinneret and a moving platform, direct-writing electrospinning (DWE) combines the typical control over material deposition of additive manufacturing (AM) with the capability of electrospinning (ES) to lay down submicrometric filaments.In this way, DWE enables to control fiber deposition and fabricate complex fibrous structures which are challenging to obtain with ES and replicate more realistically the fibrous architecture of biological tissues with respect to AM.
Moreover, if compared to cell-electrospinning, DWE does not imply cells embedding directly inside the ink which, under the application of a voltage difference and in direct contact with the solvents usually employed for electrospinning [1], can undergo necrosis, but it is able to reach a high structures resolution without impairing cell survival.
To control fiber deposition in DWE, maintaining the electrospun jet in its straight region is essential and this can be obtained by near-field electrospinning (NFES) or melt electrowriting (MEW).The absence of a whipping phase results in generally larger diameter compared to traditional electrospinning, but significantly smaller if compared to other widely used extrusion techniques (the minimum obtainable filament diameter being roughly less than 0.1 μm [2] in NFES or 5 μm [3] in MEW versus 100-200 μm [4][5][6][7]), reaching a higher resolution.
DWE potential to guide cell orientation and maturation has been widely confirmed by processing synthetic polymers, especially poly(ε-caprolactone) (PCL).Using MEW, diagonal PCL fibers led to a pronounced fibroblasts elongation and scaffold morphological features made cells interact as a 3D construct rather than a flat substrate [8] whereas seeded on hexagonal PCL scaffolds had a positive influence on human induced pluripotent stem cell-derived cardiomyocytes in terms of beating rate, cell alignment, sarcomere content and organization as compared to rudimentary fiber scaffolds [9].
Adopting NFES, poly-L-lactic acid (PLLA) quadrangular or rhomboidal (0°/45°) grids were also collected under optimized process conditions [10], which made human umbilical artery smooth muscle cells largely aligned and elongated and cell infiltration 4.5-fold greater than control electrospun scaffolds, whereas triangular or diamond-shaped porous scaffolds made by PCL and hydroxyapatite [11] showed excellent connectivity with no closed pores and mouse pre-osteoblasts could spread, proliferate and differentiate for an extended time, in vitro.
Despite its potential, to date DWE applicability is still very limited, especially in dealing with naturederived polymers, which are typically more challenging to be processed via electric-field based techniques [12], even though they possess extracellular matrixlike features, mechanical tunability, high biocompatibility, and high water holding capacity, all appealing for tissue regeneration [13].Some applications of DWE to process natural polymers have been reported in [14], where gelatin fibers were parallelly placed onto silicon or polydimethylsiloxane (PDMS) substrates and a significant endothelial cell line attachment was observed, not present on featureless 2D surfaces.Fibrous alginate patterned in parallel lines [15] demonstrated how interfiber spacing influenced cell alignment and viability, defining an optimal range to guide cellular orientation and maintain an adequate viability not only with clusters but even at single cell level.The remaining studies involving DWE and nature-derived polymers report blends of the latter with synthetic polymers: PCL and collagen grids obtained via NFES [16] not only guided human periodontal ligament stem cells disposition, but also promoted their differentiation in periodontal ligament cells with respect to randomly electrospun fibers.NFES was also used to fabricate scaffolds made by a blend of PCL and an anti-inflammatory drug (F101) [17] which after being implanted in a mouse supported an improved wound recovery.
Here, a simple but highly controllable and userfriendly DWE setup was developed.Components were assembled in a configuration that ensures easy replicability, transferability of the system and fast set-up of the DWE procedure, to provide an alternative easily adaptable to experimental conditions.The capabilities of control and actuation system to accurately reproduce a desired motion trajectory and exert a control over fiber placement were here assessed with a synthetic polymer (polyethylene oxide, PEO) and then the possibility of processing more complex materials such as natural polymers was validated, using a blend of Gelatin B and Bovine Serum Albumin, never tested before with DWE: the solution was chosen as a model for large availability and favourable properties of both polymers, such as the similarity of Gelatin B to collagen, the most naturally found protein in humans and animals [18] and the non-oxidant properties of Bovine serum Albumin [19] that tend to reduce cell damaging.
Structures obtained with the system were also biologically tested, proving DWE setup efficiency in depositing highly organized and controlled fiber patterns able to guide cellular orientation and disposition.
A 3D printed polylactic acid (PLA) plate was screwed on the XY translating element of the platform (figure 1) and an aluminum plate (12 × 12 × 0.5 cm 3 ) was assembled on the PLA one with nuts and screws to allow the levelling of the fiber collecting surface.
The control system was made by Arduino Uno (V in = 5-12 V) and CNC Shield v3.0 (V in = 12-36 V) boards with 4 A4988 drivers (V in = 8V-35V) assembled, connected to the PC through a USB 2.0 cable Type A/B and fed by a voltage supply (12 V, 20 A).The reduction of the control system was set as 1/16, so that each motor step is converted into a movement of the driven wheel equal to 1/16 of the pulley rotation, the smallest motion entity possible with NEMA 17 motors.
LaserGRBL software was used to send motion trajectories to the XY motion platform and to set other key motion parameters.The wide variety of file extensions (.SVG, .JPEG, .PNG or .STL file) accepted and authomatically converted into .GCODE results in an easy-to-use and versatile PC interface.Acceleration was set as 1500 mm/s 2 to ensure a fast reaching of the set velocity value, while step/mm value was equal to 53.33 for X and Y motion axes to make the XY moving platform accomplish motion translations coherent with the trajectory extension.
On the top of the XY moving plate, a syringe pump (NewEra Pump Systems) was positioned to dispense the polymeric solution for DWE and electrostatic generators (Enrico Pesatori Costruzioni Elettriche, V max = 20 kV) were used to initiate the DWE procedure between the spinneret and the XY moving plate.

XY motion system characterization
To control the correct command transfer from PC to XY platform, speed values were increasingly set on LaserGRBL and the consequent control system output signal frequency was measured by detecting the signal with an oscilloscope (XDS3104E, Owon), verifying proportionality between speed and frequency increments.The actual XY moving plate velocity was recorded by measuring the time necessary for the plate to complete the trajectory while performing specific squared trajectories (3 measurements per velocity value).amount of deposition cycles were optimized and set as 300 μm and 300 respectively towards fibers with suitable diameters and straightness to address anisotropic tissues engineering.GelB was substituted by GelMA to make structures not soluble in aqueous environment: in particular, the so-obtained structures were crosslinked by immersing them in Irgacure 2959 (Fluorochem, CAS number: 106797-53-9, PCode: F234409) 10% w/v in ethanol for 2 h and exposed at UV light at 10 cm for 10 min, still being immersed in photocuring agent solution.This procedure made structures resistant against PBS immersion for up to 23 d at 37 °C.

Biological evaluation
Multilayered samples were then UV sterilized, immersed in DPBS and DMEM (EuroClone, PCode; ECB7501L) for 5 min and 4 h respectively, to avoid the risk of a possible acetic acid residual release in culture.C2C12 murine cells were seeded on them (density = 20000 cells/cm 2 ), and at 1, 3 and 7 d of culture AlamarBlue resazurin assay (Resazurin sodium salt, BioReagent, suitable for cell culture, Sigma Aldrich, PCode: R7017) was performed to evaluate cell growth, by dividing the fluorescence (RFU) signal detected by a spectrophotometer at each time point to that recorded at culture day 1, at the excitation and emission wavelength of 554 and 593 nm respectively, according to the below equation (equation ( 1)):

Morphological characterization
Fiber diameter and alignment along the motion direction were evaluated by Stereo Microscopy (S9i, Leica) and Scanning Electron Microscopy (SEM, EVO 50, Zeiss) and image analysis was performed through ImageJ [20]: for each filament, the diameter was measured in three points along the entire filament length and directionality hystograms were generated through the Directionality tool to evaluate fiber straightness and alignment along the motion trajectory.Cells nuclei were quantified via the Analyze Particles tool, selecting only particles with area greater than 30 μm 2 to exclude particles different from cells and after a proper thresholding and watershed binary analysis.

Statistical analysis
Data were expressed as mean ± standard deviation.Statistical analyses were performed with GraphPad Prism 8.A value of p < 0.05 was considered statistically significant.

Results
The developed setup is reported in figure 3. The control system evaluation through oscilloscope highlighted a signal frequency increase directly proportional to velocity values set at the PC interface up to 9000 mm/min: at higher speed values, the proportionality with the set speed was not maintained.Thus, 9000 mm/min can be regarded as the maximum speed value applicable to the XY motion platform.
Considering the XY plate movement, the error between the effective and the set XY velocity increased with velocity values, remaining down to 7% for speed below 4000 mm/min; at 8000 mm/min, the highest error (16%) was found (figure 4).
In table 3, the effective XY velocities of the motion platform corresponding to the values set on LaserGRBL software with the relative error between the two velocities are reported.After DWE system development and XY motion platform characterization, setup capability of fiber guidance was assessed.In tests conducted with PEO solution varying needle-collector distances and XY motion velocities while maintaining a constant electric field (1 kV/mm) and performing a 40-mm squared motion trajectory, it was observed that by reducing the travel distance and increasing XY collector velocity, fiber straightness and alignment were improved.
At 1 mm, straight submicrometric fibers were reached at each velocity value tested, as demonstrated by the sharp peaks present in directionality hystograms (figure 5).
At trajectory corners, shape accuracy was maintained but fiber straightness is impaired, due to the XY motion platform deceleration when it approaches the direction inversion point (figures 6(a), (b)), with the collection of coiled and discontinuous fiber.However, from literature [21], it is recommended the use of patterns near the high-speed region only, discarding the fiber portions near acceleration and deceleration regions.
The system was able to lay down fibers according even to intricated motion trajectories (figures 7(c), (d)), maintaining a dimensional correspondance with the original trajectory image (figures 7(a), (b)) at a needle-collector distance of 10 mm applying 2 kV/mm as electric field, even if fibers lost their straightness appearing coiled and wavy.Nevertheless, an important advantage of the DWE setup consists in the versatility ensured in terms of motion trajectory file format, which does not need to be converted in GCode file with auxiliary programmes as in other techniques to be read properly by the XY platform (e.g.slicing software for 3D printing) but can be authomatically translated in form of GCode commands by the PC interface itself.This overcomes a great limitation of current DWE systems: the lack of easy programming strategies and of predefined command libraries for the XY plate motion, which makes it hard to reach complex scaffold geometries with DWE [22].Indeed, literature shows plenty of grid-like scaffolds built via DWE [23][24][25][26][27][28] and only few exceptions show a higher degree of pattern intricacy [11,[29][30][31][32][33].
Moreover, the vast majority of DWE systems are built 'in-house' with broad differences in how each machine performs functions, which complicates the direct comparison of data and results among different research group and hampers the high-throughput production of structures, preventing the technique from becoming standard in scaffolds fabrication.Having a DWE setup which simplifies the motion commands transfer for the user can bring to increase consistently DWE usability and control.
For GelB/BSA solution, 22% w/v as total concentration allowed process stability and deposition of continuous fibers along different motion trajectories, even curved which have not yet been widely explored with DWE of nature-derived polymeric solutions and are difficult, if not impossible, to be obtained with conventional electrospinning.As electric field, 1 kV/mm gave rise to a not continuous fiber writing and to fiber diameters not uniform and micrometric (1.37 ± 0.97 μm).Straight fibers were obtained at 1.5 kV/mm, 8000 mm/min, 22 μl/h, 22 G for both parallel fibers and circles (figures 8(a), (b)), apart from some confined coiled regions in correspondence with deceleration of the XY motion platform; in linear trajectories, fiber diameters were submicrometric, while for circular paths the mean fiber diameter resulted in falling into the micrometric range.Mean fiber diameters (mean fiber Ø) and standard deviation (SD) for both trajectories are reported in table 4.
After having validated the system ability to precisely lay down natural polymers according to a  In particular, a 300 μm interfiber spacing was selected as the best to ensure both fiber straightness and fiber diameters thinner.300 repetitions of motion trajectory were estimated as the maximum to keep fiber alignment over the whole process.
From tests with C2C12 cell line, fiber-patterned structures demonstrated a progressive cell growth on them highlighted by the resazurin assay (figure 9), reaching after 7 d a comparable value with respect to the positive control (multiwell bottom) and to the hydrogels without fibers on top.
By Hoechst/Phalloidin staining at 8 d (figure 10), fiber-patterned samples were observed in their ability to guide cell orientation along with the fiber pattern direction.In fact, on average, 3-fold cells (1239 cells/mm 2 versus 430 cells/mm 2 ) were adhered on fiber-patterned samples with respect to plain hydrogels and moreover cells were spread and well elongated along fiber pattern direction, which confirms that the DWE system can create fiber patterns suitable for cell guidance.

Conclusion
The DWE system developed was found to be simple and cost-effective, and allows efficient control over process parameters for material lay-down.Submicrometric and very straight fibers can be deposited according to various shapes, even complex architectures, with only minor coiling when motion direction abruptly changes sometimes occurring.The DWE setup proved its capability to prepare patterns capable of exerting cell guidance.
On the whole, the DWE system presented can contribute to extend the use of this technique in the field of regenerative medicine and tissue engineering.

2. 4 .
Fabrication of protein-based scaffoldAfter system optimization and validation, the DWE system potential of processing even nature-derived polymers in addition to synthetic polymers was also evaluated.In particular, two natural polymers not yet investigated via DWE, Gelatin from bovine skin (GelB, Sigma Aldrich, powder, CAS number: 9000-70-8, PCode: 1002978963) and Bovine Serum Albumin (BSA, Sigma Aldrich, lyophilized powder, CAS number: 9048-46-8, PCode: 1002494455) were blended 10:1 in 70% v/v aqueous acetic acid (total concentration 22, 27.5, 44% w/v) and both solution and DWE process (electric field, flow rate, needle gauge and XY speed) were optimized (range of parameters reported in table 2) to deposit straight submicrometric fibers in parallel arrays (20 × 20 mm 2 ) at different interfiber spacings (0, 500 and 1000 μm) and in circles (Ø = 26 mm), to demonstrate whether the DWE system was able to guide submicrometric fibers along straight and curved trajectories.To address biological characterization, multilayered oriented fibrous samples were obtained by depositing methacrylated gelatin and BSA 22% w/v in 70% v/v aqueous acetic acid in parallel arrays of fibers on a soft hydrogel (h wet = 500 μm, Ø = 6 mm) made by the same natural polymers dissolved in DPBS (EuroClone, PCode: ECB4053) to allow an easy handling of patterns and preventing their damage during in vitro cellular tests (figure2).Both fiber spacing and

Figure 1 .
Figure 1.XY translating plate of the DWE system: the aluminum collector plate, the white 3D printed PLA plate fixed to the XY translating element and the 4 screws with nuts for fine leveling of the collecting plane.

Figure 2 .
Figure 2. Scheme of DWE fiber-writing strategy and subsequent photo-crosslinking to obtain multilayered oriented samples.

Figure 3 .
Figure 3. Scheme of the DWE setup.

Figure 4 .
Figure 4. Comparison between set XY velocity (dashed line) and effective XY velocity (yellow-coloured line) of the XY motion platform, with the relative error between them (blue-coloured histogram).

Figure 5 .
Figure 5. SEM images of single PEO fibers collected, with the corresponding directionality hystograms (electric field: 1 kV/mm, flow rate: 30 μl/h, n°of motion trajectory executions: 1).The presence of a sharp peak indicates high fiber orientation.

Figure 8 .
Figure 8. Motion trajectory schemes and photos of fiber composing patterns (a): parallel array (b): circle (taken at stereoptical microscope and at SEM).

Figure 9 .
Figure 9. Cell growth on fiber-patterned samples, plain hydrogels and controls.

Table 1 .
Range of parameters varied for DWE optimization and system validation with PEO solution.

Table 2 .
Range of parameters varied for DWE optimization and system validation with GelB/BSA solution.

Table 3 .
XY velocity values set on LaserGRBL and the corresponding effective XY velocities of the XY motion platform.
fibrous pattern of interest, multilayered oriented structures mimicking anisotropic tissues architecture (e.g.skeletal muscle tissue) were fabricated through DWE with a high degree of fiber alignment (70%) and in an automated and reproducible way.