The open source bioprinter and syringe pump extrusion tool
The bioprinter (Fig. 1) was built around the E3D tool changer and motion system (E3D-online, London, United Kingdom)20,21, and enclosed in a custom-built polycarbonate cabinet with an integrated a HEPA filter with an air intake fan. The syringe pump extrusion tool (Fig. 2) was built using a stepper motor, a lead screw, and 3D-printed plunger housing and syringe holder. All computer-aided design (CAD) was done using Fusion 360 (Autodesk Inc, San Rafael, USA). All 3D-printed parts for the extrusion tools were printed on a Form 3 SLA 3D printer (Formlabs) in standard clear resin or tough 2000 resin (Formlabs) with a Z-resolution of ~ 100 µm. All prints with resin were washed and cured according to the manufacturer’s instructions. The baskets used for handling FRESH-printed constructs were 3D-printed in PEKK-A (Kimya, Nantes, France) on a miniFactory Ultra FFF 3D-printer (miniFactory Oy LTD, Seinäjoki, Finland). Brackets for the enclosure, camera housing, touch screen mount and HEPA-filter holder were printed in PLA (Add North 3D AB, Ölsremma, Sweden) on a Prusa I3 MK3S (Prusa, Prague, Czech Republic). STL-files for all 3D printed components as well as a full list of all components required to build the bioprinter and syringe pump extrusion tool are included in the Supplementary Materials section.
Bioprinter operation
Before powering up the printer, the stepper motors driving the syringe pump extrusion tools were connected to the control board and the tools placed in the tool dock. The E3D motion system is controlled by a Duet Wifi controller board together with the expansion board Duex 5. This enables control of up to 10 stepper motors. 4 stepper motors in the E3D motion system are used to control the x-, y- and z-axes as well as the tool changer mechanism. In addition, syringe pump extrusion tools use one stepper motor output each. Configuration of any additional tools are made via modification of the config.g-file in the DWC interface. For multimaterial prints, requiring the use of two or more tools, calibrations for tool offsets were required as described below.
Tool offset calibration
Tool offset identification was performed using the “Tool Alignment with Machine Vision” (TAMV) program originally developed by Danal Estes37, using a Logitech web camera mounted on the print platform and a Raspberry Pi 4B running OpenCV. A prerequisite for running the calibration commands is that all axes first are homed. The procedure for x- and y-axis tool offset identification is performed per the instructions on the creator’s website37. Z-axis offset calibration is facilitated by a microswitch mounted on the build plate, this enables probing of individual tools to identify differences in syringe and needle length. For the calibration to work correctly the leadscrew in the stepper motor must be engaged with the plunger via the plunger press (Fig. 2). This is done by running the “Tool 0—Select” macro from the “macros” section of DWC, followed by stepwise movement of the leadscrew until it applies pressure to the plunger, which is repeated for each syringe pump tool. The procedure for z-axis offset calibration is performed automatically via a custom-made G-code script, using the spatial positioning of the first tool as the reference for offset calibration for the other tools. The first tool is picked up and positioned with the needle tip above the microswitch, followed by movement of the build plate until the microswitch is engaged. After the reference position of the first tool has been logged, other tools are then picked up and used to trigger the microswitch. At the end of the procedure, a z-offset value is presented in the DWC web interface and can be inserted manually into config.g which stores the offset values for each tool. This procedure is repeated between each new print since needle length and position of needle tips in the tool will differ between prints. Once calibration is complete bioprinting can be initiated by using the macro to pick up the loaded tool, manually positioning it at the starting point, zero the z-axis and start the G-code file for the selected construct of interest.
Toolpath generation
The toolpath generation software (Simplify3D, Cincinnati, USA) was used to generate G-code files for bioprinting of the different constructs. Bioprinting was typically performed at a speed of 10 mm/s, as printing at this speed produced reproducible constructs of both the collagen and laminin bioinks. The printing parameters are available in a .factory file format, which is native to Simplify3D. Further information about software configuration and slicing settings can be found in the Supplementary Materials section.
Assessment of stepper motor step accuracy, minimum print volume, and practical print resolution
To determine the accuracy with which the stepper motor vertically displaced a syringe plunger, a dial indicator (Mitutoyo 543-125B, Kanagawa, Japan) was mounted beneath the lead screw and adjusted such that its contact point was in contact with the plunger press. The stepper motor was programmed to take 10 mm, 1 mm, or 0.1 mm steps sizes and for each step the measured displacement distance was recorded from the dial indicator. These measurements were repeated 10 times for each step size, and a linear plot was prepared (Fig. 2c). We calculated the minimum volume of bioink that the bioprinter could extrude through a Hamilton 0.25 ml syringe using the measured displacement values recorded when the step size was set at 0.1 mm, based on the formula for the volume of a cylinder (πr2h), where r is the inner radius of the Hamilton 0.25 ml syringe (1.15 mm). This yielded a value of 0.39 ± 0.02 μl (mean ± S.D.). To assess the practical lower limits of the resolution of the bioprinter we used a nozzle with a 50 μm cross-sectional diameter to extrude strings of collagen using FRESH bioprinting. The strings were imaged by differential interference contrast using a Zeiss LSM710 confocal microscope (Zeiss), and the average cross-sectional width of the finest collagen strings that we could produce and recover was calculated from multiple measurements along the length of the printed string (Fig. 2d).
Cell culture
The MDA-MB-231 breast cancer cell line was obtained from the American Type Culture Collection (ATCC) and cultured in Dulbecco’s Modified Eagle Medium (DMEM) Glutamax (Gibco, Thermo Fisher Scientific, Uppsala, Sweden), supplemented with 10% Fetal Bovine Serum (FBS; Thermo Fisher Scientific), referred to hereafter as culture medium, in a humidified incubator (5% CO2, 37 °C) according to the instructions provided by the supplier ATCC.
Bioink loading
The Collagen Lifeink 240 (Advanced Biomatrix) bioink and LaminInk + (Cellink) laminin bioink were first transferred to a 5 ml plastic syringe with male-male LuerLock. To reduce the presence of bubbles in the bioink, the loaded syringe was centrifuged for 4 min at 1200×g, with the aid of a custom-made centrifuge insert. The hydrogel was next transferred to a 1 ml Hamilton gastight syringe via a female-female LuerLock adapter. The syringe was then inserted into the 3D-printed syringe holder, which in turn was bolted into the syringe pump body with four M3 × 10 screws.
FRESH collagen bioprinting, cell seeding, and construct imaging
FRESH bioprinting was performed according to the protocol by Lee et al.16, using LifeInk 240 (Advanced Biomatrix). The FRESH LifeSupport powder (Cellink), from which the gelatin microparticle support bath was prepared, was used as per the manufacturer’s instructions. Following print completion, the tool was left at the tool dock and the build plate lowered. All constructs generated using FRESH bioprinting was made in a custom-designed and easily moveable basket. After a completed print, the basket holding the collagen construct was placed at 37 °C for 1 h to melt the gelatin support material and to thereby release the printed construct. Constructs were washed three times in 1× phosphate buffered saline (PBS, Gibco) and stored in PBS at 4 °C until further use. For fluorescent imaging, collagen constructs were stained for 1 h at 37 °C with 10 μM Col-F (ImmunoChemistry Technologies, Bloomington, USA) prepared in 1× PBS, and then washed twice (5 min each) with 1× PBS at room temperature. Constructs were imaged by confocal microscopy performed on a Zeiss LSM700 (Zeiss, Jena, Germany) instrument and images were captured using Zen imaging software (Zen 2011 SP7 FP3, version 14.00.22.201, Zeiss). For cell seeding, FRESH constructs were first typically stored overnight at 4 °C in 1× PBS supplemented with the antibiotics penicillin and streptomycin (PenStrep; Thermofisher, Uppsala, Sweden). Individual constructs were next moved to separate wells in a µ-slide 8-well coverglass-bottomed chamber (Ibidi) and a cell suspension of 30 × 103 MDA-MB-231 cells diluted in 250 µl culture medium supplemented with 1× PenStrep per well was deposited directly onto the collagen construct or into an adjacent, construct-free wells, and cultured for 20 h under standard cell culture conditions. Cell staining and analysis of viability and cell morphology by microscopy are described below.
Analysis of cell viability in seeded FRESH collagen constructs
To assess cell viability in FRESH bioprinted collagen constructs, cell-seeded constructs were incubated with the NucBlue nuclear stain to detect individual cells (Invitrogen, Thermo Fisher), propidium iodide (PI) to identify dead cells (1:1000; Invitrogen, Thermo Fisher), and the Cell event caspase-3/7 stain that reports on the activity of caspase 3 and 7 in apoptotic cells (10 μM; Cell event, Invitrogen, Thermo Fisher). Staining solutions were diluted in OptiMEM (Gibco, Thermo Fisher), supplemented with 1× PenStrep. To control for the effectivity of the viability stains, cells were treated for 5 h with staurosporine (10 μM; Abcam, Cambridge, United Kingdom), a potent protein kinase inhibitor known to induce apoptosis. Fluorescent signals for the respective stains were captured by confocal microscopy with an LSM700 instrument and Zen software (Zeiss). Multiple z-planes were captured for cell-laden constructs, while a single in-focus z-plane was captured for cells attached directly to the glass in construct-free wells. These experiments were conducted in duplicate on three independent occasions. Image analysis was conducted in the Fiji version of ImageJ38. For cell-laden collagen constructs, the z-plane containing the highest number of cells was selected for analysis. Individual cells were identified as regions of interest (ROI) by applying a threshold to the NucBlue signal. The ImageJ particle analysis function was applied to this thresholded image to include particles within a size range of 15–600 μm2. This range excluded non-specific signals from small debris and very large clusters of cells. However, it included smaller cell clusters that could be delineated as individual cells by applying the automatic watershed function to the thresholded image. This ROI mask of individual cells was then overlaid on the PI and Cell event caspase-3/7 channels and the mean fluorescence was recorded in each ROI for each channel. This data was exported to Excel, and mean intensity per ROI and channel was background adjusted using an average from three ROIs measured in cell-free areas of the collagen construct. Data was statistically analyzed using ordinary one-way ANOVA with Tukey’s multiple comparisons in GraphPad Prism (Prism).
Analysis of cell attachment and cell morphology
To visualize cells adhering to FRESH bioprinted collagen constructs the cells were labelled overnight with SiR-actin (500 nM; Spirochrome, Switzerland), which permits fluorescence imaging of the actin cytoskeleton. Samples were washed twice in OptiMEM cell culture medium, which was then exchanged for OptiMEM containing the NucBlue nuclear stain. The cells were imaged as described above by confocal microscopy.
Analysis of cell viability in long term culture of cells seeded on collagen
FRESH bioprinting of constructs for one week culture was performed as described above, however to reduce unnecessary handling, constructs were printed directly on a Labtek 8 well chamber coverslip. Onto each construct 24,000 cells/well were seeded in DMEM (10% FBS, 1% PenStrep). Half the media was exchanged every 1–2 days. Live/dead staining was performed on day 1 after cell seeding, and in separate constructs 1 week after cell seeding. Briefly, cells were stained with fluorescein diacetate (Sigma) 8 µg/ml and PI, diluted 1:1000 in OptiMEM, for 10 min. Then constructs were washed once with OptiMEM. Images were taken using an LSM 700 confocal microscope (Zeiss) and a 10× objective. For each construct two images were taken of a z-plane with high cell density and used for analysis in ImageJ. In ImageJ both channels were thresholded manually and the automatic watershed function applied. Subsequently, the particle analysis tool was used to count the number of live or dead cells. Using Excel, the percentage of viable cells was calculated for each image. To determine statistically significant differences between groups, data were imported to GraphPad Prism and analyzed by Student’s t-test. Experiments were performed in three independent repetitions with two timepoints and four to five constructs per timepoint and repetition.
Analysis of cell viability following extrusion of cell-laden laminin bioink
Cell-laden laminin bioink was prepared as follows; 100 µl of cell suspension containing 11 × 106 cells/ml was mixed with 1 ml Laminink + Bioink (Cellink) by passing back and forth between two 1 ml syringes. For control, the cell-laden bioink was directly deposited into wells from the plastic syringe without extrusion through a thin needle or nozzle. The remaining ink was loaded into a 250 µl Hamilton glass syringe and bioprinting performed with an 18 G needle. For crosslinking of all constructs, including non-printed controls, the recommended Crosslinking Agent (Cellink) was applied for 1 min. Subsequently, the constructs were washed once in warm (37 °C) DMEM Glutamax, supplemented with 10% FBS and 1× PenStrep, and then either cultured for one week in 100 µl of media, with half the media being exchanged daily, in a cell incubator (5% CO2, 37 °C) or directly used for live/dead staining, described above. Experiments were conducted in three independent repetitions with two timepoints and a total number of three to five printed constructs and non-printed control hydrogels for each group and repetition. For each construct two images were taken of a z-plane with high cell density and used for particle analysis in ImageJ, and statistical analysis in GraphPad Prism (Prism), as described above.
Multimaterial bioprinting with collagen and laminin bioinks
Multimaterial constructs were bioprinted using acidified collagen (Advanced Biomatrix) and Laminink + (Cellink) containing cells as described above. Printing was done directly onto a glass slide, followed by incubation in 50 mM HEPES (pH 7.4) containing 50 mM CaCl2, to induce gelation of collagen and laminin, respectively.
