Customized 3D-printed stackable cell culture inserts tailored with bioactive membranes

Fabrication of 3D-printed inserts

3D-printed inserts with different formats and dimensions (Fig. 1c) were designed for tissue-culture-treated (TCT) well plates (Nunc, Thermo Fisher, Slangerup, DK) using Fusion 360 (Autodesk, California, USA) and exported as .STL format files. The print files were processed by adding external supports (with 0.5 density and 0.4 mm touchpoint size and mini rafts with 0.5 mm thickness) and converted to .FORM files in PreForm (v.3.11.0, Formlabs, Somerville, Massachusetts, USA) before printing by a Form 3B 3D printer (Formlabs). Inserts were printed in Dental LT v1 resin (Formlabs) with a layer thickness of 0.1 mm. After printing, inserts were washed in the isopropyl alcohol (IPA) tank (Formlabs) and the support structures were cut off. All prints were subsequently cleaned twice in fresh IPA for 1 h, respectively. After washing, inserts were air-dried and post-cured in the Form Cure UV-oven (Formlabs) for 2 h at 60 °C. After the post-curing step, inserts were submerged in sterile water with a 30 mL resin to 1 L water ratio for a day. Once the prints were air dry, a further 15 min UV exposed for each side for sterilization was performed using Asiga Flash UV Curing Chamber (New South Wales, AUS) under the laminar flow biosafety level 2 (BSL-2) cabinet.

Assembling synthetic and hydrogel membrane

Hydrophilic Polytetrafluoroethylene (PTFE) membrane (BioPore Membrane Filter, with a pore size of 0.4 µm, Millipore®, Sigma-Aldrich, Missouri, USA) was assembled on the bottom of the 3D-printed insert and referred to as Filter inserts). Dental LT was applied to the insert walls and punched out PTFE membrane discs was affixed. Afterwards, the membrane was fixed towards the wall of inserts using a printed cylindrical ring that also stretched the filter. It was post-cured in a Form Cure UV-oven (Formlabs) for 20 min at 60 °C. Sequential IPA washing steps and UV post-curing were conducted as described above.

The gelatin (48723, Sigma-Aldrich) was mixed with 5 mL phosphate-buffered saline (PBS) (D8537, Sigma, St Louis, USA) and heated to 37 °C. The solution was adjusted to pH 7 by titrating with a 10 M NaOH. For a 5 mL 10% solution, approximately 3.7 µL of 10 NaOH was added. For 5 mL 15% and 20% gelatin solutions, the corresponding values were 5.7 and 7.7 µL. Chloroform was added (0.5% v/v) to sterilize the solution. These standardized volumes made it possible to mix all components and place the tubes a in a 37 °C water bath until the gelatin was fully dissolved to obtain a sterile working solution with the correct pH. Once dissolved, gelatin solutions were poured to a pre-heated sterile petri dish (Nunc, Thermo Fisher Scientific, USA) on the homemade hot plate (Supplementary Fig. S3) set at 50 °C under the laminar flow biosafety cabinet. Note, the LAF bench requires an exhaust into free air as the chloroform will evaporate quickly. Sterile 3D-printed inserts without PTFE membrane were immersed in the gelatin solution for 5 s until the bottom was covered with hydrogel liquid. The inserts were fully withdrawn and pressed on a sterile pre-warmed petri dish lid on the homemade Peltier unit at 50 °C to form a thin hydrogel membrane by removing the extra hydrogel liquid. After dip-casting the membranes, inserts were placed to the cell culture plate and placed at 4 °C for at least 20 min to thermally (reversible) crosslink the gelatin. In the meantime, 15 U/mL microbial Transglutaminase (mTG) solution was prepared by mixing mTG (1002, 100 U/g, ACTIVA® TI, Ajinomoto Food Ingredients LLC, Illinois, USA) with PBS. The mTG solution was vortexed until the mTG powder was fully dissolved and filtered through a 0.45 µm sterile syringe filter (Avantor, VWR, USA). After the reversible crosslinking of the gelatin membranes, the mTG solution was added to wells with the amount to cover insert well surfaces (for 96-well: 50 µL; for 48-well: 100 µL; for 24-well: 200 µL). The plates were placed in an incubator at 37 °C for 1 h to crosslink membranes irreversibly. After crosslinking, the sterile hydrogel membranes were washed in PBS for 10 min three times. All inserts with gelatin membrane (Gelatin-inserts) were either used for cell culture experiments directly or stored in the fridge for max. 2 weeks or dried overnight in the incubator and stored at room temperature for maximum 1 month prior to use.

Mass transport through membranes

The permeability of 96-well strip inserts was tested by mass transport studies of Fluorescein sodium salt (F6377, Sigma-Aldrich). Inserts were placed in a 96-well plate and, 225 µL of pre-heated (to 37 °C) PBS was added to the basolateral side, and 75 µL of 50 µM Fluorescein sodium salt was added to the apical side in each insert. The transport study was performed at 37 °C with 100 rounds per minute (rpm) shaking under dark conditions for 2 h. After the incubation, samples of 50 µL were transferred to a 96-well plate to analyze their Relative Fluorescence Units (RFU) emission at an excitation wavelength of 460 nm and emission wavelength of 515 nm (Spark® multimode microplate reader, TECAN, Männedorf, CH). A standard curve was obtained by evaluating RFU of fluorescein sodium salt in different concentrations from 50 µM to 0 µM, and RFU data was collected and analyzed using GraphPad Prism (version 9.0.0 for macOS, GraphPad Software, La Jolla California USA, www.graphpad.com).

Mechanical testing of hydrogel membranes

Gelatin hydrogel discs were prepared using the sandwich casting method. Two polymethyl methacrylate (PMMA) slides with a thickness of 3 mm were coated with 5% (w/v) PVA (Polyvinyl alcohol)–EtOH (ethanol) solution used as a mold released agent and air-dried. Afterwards, two 1 mm thick-PMMA slides were glued on two pairs of 3 mm-thick PMMA slides (base) with PVA solution. 500 µL gelatin solution was cast on one of the base slides, and the other base slide was closed on top of the other pair. Then, gelatin discs with 1 mm thickness were crosslinked in the fridge at 4 °C. After crosslinking, the gelatin discs were carefully removed by flushing mTG solution between the base slides and incubated in mTG loaded petri dish for 1 h in the incubator at 37 °C. Irreversibly crosslinked gelatin discs were washed with PBS twice and stored in a closed petri dish with PBS to avoid dehydration. Rheological properties were characterized using a rheometer (TA Instruments, DHR 20, New Castle, Delaware, USA) equipped with a 20 mm diameter cylinder with parallel plate geometry and a gap of 1 mm. Tests in gelatin discs were performed at 37 °C. Manufactured samples were loaded on the lower Peltier plate of the rheometer and trimmed to adjust the shape to the upper parallel plate geometry. First, a frequency sweep test was performed from 200 rad/s to 1 rad/s at the strain of 0.01% to calibrate the instrument. Then, the amplitude/strain sweep test was performed from 0.2 to 200% at 1 rad/s, and the results were used to evaluate material properties. Finally, time sweep up to 1000 s (about 16 and a half minutes), at 5% strain and 1 rad/s to ensure that the material was stable in the estimated time. Stiffness and elasticity of the developed hydrogels were assessed from shear modulus, and Young’s modulus of each disc was obtained as previously described¹⁶. The shear modulus or complex modulus (G^∗) was calculated by:

where, G’ is the storage modulus, and G” is the loss modulus.

Then, Young’s modulus (E) was obtained as:

where, μ is the Poisson’s ratio, assumed to be 0.5 as the hydrogels are considered incompressible materials.

Thickness measurement of hydrogel membranes

For measurement of gelatin membrane thickness, fluorescence intensity analysis was performed. Fluorescent gelatin was prepared by mixing 15% (w/v) gelatin powder in 1 µg/mL Fluorescein sodium salt (F6377, Sigma-Aldrich)—PBS solution. First, 6 liquid coupled slides were created by using the similar sandwich method described before to form the standard curve of fluorescence intensity of different heights provided by stacking coverslips (150 µm thick) onto a 1 mm thick glass microscope slide (base) and 500 µL fluorescence gelatin was pipetted onto the base slide and covered with the other pair of the base slide. Inserts (96- and 48-well format strips) were fabricated as described above and placed onto a base slide (Avantor, VWR, USA). Gelatin membranes were imaged with Zeiss™ AxioObserver Z1 epifluorescence microscope (Carl Zeiss MicroImaging GmbH, Gottingen, Germany). All images (n = 5) were obtained at an emission wavelength range of 495 nm to 517 nm by a LED (Light-emitting Diode) laser with filter set 38 HE (Carl Zeiss MicroImaging GmbH) and within 150–900 µm thickness range. All the .TIF files were processed with Zeiss Zen Blue 3.4 Lite Digital Imaging Software (version 3.4.91.00000, Carl Zeiss Microscopy GmbH) to correlate membrane height according to fluorescence intensity.

Swelling ratio of hydrogel membranes

For determination of thickness change of hydrogel membranes, 15% (w/v) Gelatin membranes were assembled to 48-well format Gelatin-inserts as mentioned above. Membranes were punched with 3 mm diameter biopsy skin punch (Acu-Punch®, Acuderm Inc., USA)) as freshly crosslinked (wet, n = 5), and swollen (n = 5) in PBS for 1 h at 37 °C to mimic equilibrate status of membranes and dehydrated overnight at room temperature (dry, n = 5). The thickness of punched membrane discs was measured by using a micrometer screw (Mitutoyo, JP) and the swelling ratio (SF_v) was calculated by deriving the equation below⁴³ (diameter was taken constant due to the dry, wet, and swollen membranes were taken out of the inserts using the puncher:

where, r_f is the final radius (µm), h_f is the final height (µm), r₀ is the initial radius (µm), h₀ is the initial height (µm), d_f is the final diameter (µm), d₀ is the initial diameter (µm).

Cell types and culture conditions

Human epithelial colon carcinoma cells (Caco-2, passage 55–65, 09042001, European Collection of Authenticated Cell Cultures (ECACC), Salisbury, UK) were cultured in T-75 cell culture flasks (Sarstedt, Nümbrecht, Germany) in High-glucose Dulbecco’s DMEM medium (Sigma-Aldrich) with 10% (v/v) fetal bovine serum (FBS, Hyclone™, CA), 1% (v/v) non-essential amino acids (NEAA, Gibco, Fisher Scientific, Slangerup, Denmark), and penicillin (100 U/mL)-streptomycin (100 µg/mL) (P/S, Sigma-Aldrich). The Caco-2 cells were seeded on hydrogel membranes at a density of 80,000 cells/cm² and incubated for 3 weeks before analyses. Human umbilical vein endothelial cells (HUVEC, passage 5–9, Cell Applications Inc., California, USA) were cultured in endothelial culture growth medium (ECGM, Cell Applications) with 10% (v/v) FBS and 1% (v/v) P/S. The HUVEC cell cultures were seeded on gelatin micro-membranes and Matrigel-coated PTFE membranes at a density of 100,000 cells/cm² and incubated for up to 1 week before analyzes. Cells were split with trypsin-ethylenediaminetetraacetic acid (EDTA) for 3–5 min upon ~ 90% confluency. After use, the gelatin membranes could be dissolved using trypsin–EDTA to reuse the 3D-printed parts. Mouse intestinal organoids were cultured according to the protocol from StemCell Technologies (Cambridge, UK). Intestinal crypts were thawed, centrifugated and re-suspended in Matrigel (Corning, New York, USA) and transferred into 24-well PTFE membrane assembled insert plates. After polymerization, IntestiCult mouse organoid growth medium (StemCell Technologies) supplemented with 1% (v/v) P/S was overlaid on the gel in each well. All cell lines and organoids were maintained in an incubator (37 °C, 100% Humidity, 5% CO₂) with the culture medium replaced every 2 days.

F-actin, nuclei, live/dead stains

For F-actin (Alexa Fluor™ 594 Phalloidin, Invitrogen) and Hoechst 33342 (nuclei, Invitrogen™, Thermo Fisher) staining, samples were washed for 15 min in PBS followed by fixation in 2% (v/v) PFA in PBS for 2 min and fixation in 4% (v/v) PFA in PBS for 13 min at room temperature. PFA was aspirated from the samples and washed three times in PBS. Then, the samples were incubated with the stain (5 µL phalloidin and 1 µg/mL Hoechst 33342 in PBS diluted in 200 µL PBS with 1% (v/v) bovine serum albumin (BSA, Sigma)) for 20 min at room temperature. The samples were washed three times in PBS and left in PBS.

For live/dead staining (LIVE/DEAD™ Viability/Cytotoxicity Kit, L3224, Invitrogen™, Thermo Fisher) was used. Samples were washed three times in PBS and stained in 500 µL of a solution with Ethidium homodimer-1 (8 nM, EthD-1) and Calcein AM (4 mM) for 1 h at room temperature. The samples were washed in PBS two times and then kept in PBS to keep the hydrogel growth-matrices moisturized.

Microscopy imaging

Phase-contrast bright-field micrographs were obtained using a Zeiss Primovert microscope (Carl Zeiss MicroImaging GmbH, Gottingen, Germany) with the following objective: Plan-Achromat 4x/0.10 while staining images were acquired with Zeiss™ AxioObserver Z1 epifluorescence microscope (Carl Zeiss MicroImaging GmbH) with the following objectives: EC Epiplan-NEOFLUAR 5x/0.16 Ph1 M27: EC Epiplan-NEOFLUAR 10x/0.3 Ph1, LD Epiplan-NEOFLUAR 20x/0.4 Korr M27. The obtained images were fitted with scale bars and were processed in Zeiss Zen Blue 3.4 Lite Digital Imaging Software (v. 3.4.91.00000, Carl Zeiss Microscopy GmbH) and ImageJ/Fiji⁴⁴. Surface area analysis of organoids were performed using the macro; OrgM⁴⁵ on the ImageJ/Fiji.

Transport of lucifer yellow

Single inserts were placed in a 96-well plate and 225 µL (well compartment) of pre-heated (to 37 °C)) HBSS transport buffer (HBSS (1x), Sodium bicarbonate (0.0375% w/v), HEPES (10 mM), BSA (0.05% w/v, pH 7.4) was added to the basolateral side and 75 µL of 60 µM lucifer yellow was added to the apical side in each insert. The transport study was performed at 37 °C with 100 rounds per minute (rpm) shaking under dark conditions for 2 h for Caco-2 cells. After the transport experiment, the 100 µL samples (from 1st insert and the well compartments), 50 µL samples (from 2nd insert compartment) and 50 µL HBSS transport buffer were transferred to a 96-well plate to analyze their Relative Fluorescence Units (RFU) emission at an excitation wavelength of 428 nm and measuring emission at 536 nm. The permeability coefficients (Pc, nm/s) were calculated by the equation:

where, Pc is the permeability coefficient (nm/s), V_r is the receiver volume in mL, A is the membrane growth area in cm², C_i is the initial apical concentration (µM), C_f is the final receiver concentration (µM), t is the assay time in seconds.

Statistical analysis

The data are presented as the sample sizes (n), means, standard deviations (SDs). Calculations were done using Microsoft Excel (Version 2016, Microsoft Office, Seattle, Washington) and data were analyzed using GraphPad Prism (version 9.0.0 for macOS, GraphPad Software, La Jolla California USA, www.graphpad.com). Saphiro–Wilk for normality testing of multiple comparisons. Data was analyzed by ordinary one-way ANOVA or 2way ANOVA (α = 0.05) with post-hoc Tukey’s multiple comparison tests for equality of means among groups in membrane characterization, lucifer yellow transport of stacked co-culture and biocompatibility of organoids experiments. P-values were obtained using Bartlett’s corrections and determined significant differences when p-value < 0.05. F-test for Welch’s t-test was performed to compare variances of membrane thicknesses. The significant differences are highlighted with symbols on figures.