Rapid fabrication by digital light processing 3D printing of a SlipChip with movable ports for local delivery to ex vivo organ cultures
Megan A. Catterton, Alexander G. Ball, and Rebecca R. Pompano
Abstract
SlipChips are two-part microfluidic devices that can be reconfigured to change fluidic pathways for a wide range of functions, including tissue stimulation. Currently, fabrication of these devices at the prototype stage requires a skilled microfluidic technician, e.g. for wet etching or alignment steps. In most cases, SlipChip functionality requires an optically clear, smooth, and flat patterned surface that is fluorophilic and hydrophobic. Here, we tested digital light processing (DLP) 3D printing, which is rapid, reproducible and easily shared, as a solution for fabrication of SlipChips at the prototype stage. As a case study, we sought to fabricate a SlipChip intended for local de-livery to live tissue slices through a movable microfluidic port. The device was comprised of two multi-layer components: an enclosed channel with a delivery port and a culture chamber for tissue slices with a permeable support. Once the design was optimized, we demonstrated its function by locally delivering a chemical probe to slices of hydrogel and to living tissue with up to 120-µm spatial resolution. By establishing the design principles for 3D printing of SlipChip devices, this work will enhance the ability to rapidly prototype such devices at mid-scale levels of production.
Keywords
SLA printing, resin printing, tissue culture, local stimulation, Two-phase microfluidics
Introduction
The ability to produce microchips easily and with minimal manual assembly, while retaining rapid prototyping capabilities, is highly desirable for pushing microfluidic devices past the first hand-built prototype stage [1–3]. Scaled up fabrication is critical to conducting experiments at moderate scale (dozens of devices) and for propagating such technology to collaborators. In particular, this scale of fabrication would be useful for SlipChips, which are two-phase, reconfigurable microfluidic devices [4–9]. SlipChips usually comprise two planar components that can be “slipped” relative to one another, contain recessed features to hold droplets or streams of aqueous solution, and are separated by a thin layer of oil [4]. SlipChip devices were first developed in the Ismagilov lab [4] as a new technology to perform in low resource settings [5– 7]. The first SlipChips were fabricated from glass plates, which offer ideal surface properties and optical clarity but require wet etching with HF, a hazardous procedure that requires a skilled technician [4,10]. Since then, many different Slip-based designs have evolved, including rotational based Slipdisc and paper-based SlipPADs, to perform a wide range of laboratory processes such as PCR, cell culture and local delivery to tissue slices [8,9,11–18]. Fabrication is especially challenging for novel slip-based devices that have multiple layers per component [9,17]. While injection molding can simplify fabrication at large scale [19], an alternative method is needed to fabricate SlipChips at a moderate scale, while retaining the ability to rapidly prototype.
Any fabrication system for SlipChips must be able to meet four platform requirements, in addition to producing the specific features needed for the intended application. To prevent the aqueous phase from spreading into the oil-filled gap between components, high capillary pressure at the oil-water interface must be maintained. Therefore, the surfaces in contact with 3 the oil layer must be flat and smooth enough to create a gap height between ~1-10 µm across the entire face of the chip [5]. Furthermore, these surfaces must be hydrophobic, and if a fluorinated oil is used [4], a fluorophilic surface is preferred. Finally, for SlipChips that rely on visual alignment or optical detection, the layers must be optically transparent.
Considering these requirements, we reasoned that digital light projection (DLP) 3D printing, which uses UV or blue light to cure photocrosslinkable resins layer-by-layer [20,21], may facilitate SlipChip fabrication and allow for rapid prototyping. This additive method is quickly gaining popularity for fabricating small parts and microfluidic devices, because of the high feature resolution and reproducibility in parts and rapid fabrication speed compared to traditional soft-lithography [3,22–24]. While 3D printing has not been reported previously for SlipChips, two of the four fabrication requirements are already met: We recently described a method for fluorination of a DLP-printed surface based on solvent-based deposition of a fluoroalkyl silane [25], and others have demonstrated transparent prints by using clear resins on a glass surface [26]. The latter process generated a smoother surface with less light scattering.
As a case study for fabrication of a SlipChip by 3D resin printing, we considered a microfluidic movable port device (MP device) previously developed by our lab for local stimulation of ex vivo organ slices [9]. Local delivery devices like this one are used to study intrinsic tissue properties and to screen for potential drugs, by delivering aqueous solutions to specific regions of a tissue sample [9,18,27–30]. Compared to a device with stationary ports, the MP device lessens the amount of user handling of a tissue slice, by repositioning the port below a tissue slice for local delivery with the device. The MP device is a SlipChip that consists of two multilayer components. The bottom component contains an enclosed channel terminating in a 4 single microfluidic delivery port (delivery component), and the top component features a semipermeable tissue culture well (chamber component). In the original hand-built prototype, an extensive fabrication process limited the accessibility and distribution of the MP device to other labs and collaborators [9].
Here, we established an approach to fabricate a 3D printed SlipChip for the first time, using the MP device as a case study. First, we validated the selection of a DLP resin designed for microfluidic devices to meet the optical transparency, surface roughness, surface chemistry, and biocompatibility requirements of the tissue-specific movable port device. Next, the device design was optimized to maximize the functionality of the required ports and channels, while minimizing the fabrication time complexity with DLP printing. The ability of the assembled device to deliver aqueous solutions without leaks into the gap was tested, and finally we tested the ability to stimulate live organ cultures locally and with the position selected on demand.
Material and Methods
Device design, 3D printing, and laser etching
All 3D printed parts were designed using Autodesk Inventor 2018. The CAD files were sliced at 50 µm intervals using MII Utility Shortcut V 3.27 and printed using a CADworks3D M50-405 printer (30 µm xy-resolution, CADworks3D, Toronto, Canada) in BV-007A resin (MiiCraft, via CADworks 3D). The printer settings for the BV-007A resin at a 50-µm slice height was a slow peeling speed, 0.1 mm gap adjustment (unless printing on glass which was 0.27 mm), 1.15 s curing time, 1 base layer, 9.0 s base curing time, 1 buffer layer, and 75% light power. To print parts on glass, a cover glass slide 36 x 60 mm with a thickness of 0.13-0.17 mm (Ted Pella, Redding, CA, USA) was attached to the baseplate by curing a thin layer of BV-007A 5 with a 405 nm UV-light (Amazon, Seattle, WA, USA) [26]. The parts were rinsed with methanol (Fisher Chemical) and post-cured in a UV-light box for 20 s. No additional leaching steps were applied to the printed pieces used in this work. In preliminary experiments, we found that solvent washes at varied temperatures or extended UV light exposure did not substantially improve the biocompatibility of the BV-007A resin. To complete the chamber component, an array of ports with an 80-μm diameter were laser etched (Versa Laser 3.5, Universal Laser Systems, Scottsdale AZ, USA) into the printed BV-007A part, using a power setting of 7% and a speed of 10%.
Fluorination of resin surface and contact angle measurements
Parts printed in BV-007A were silanized using our recently described method [25]. The parts were submerged into a 10% v/v solution of tridecafluoro-1,1,2,2-tetrahydrooctyl trichlorosilane (Gelest Inc., Morrisville PA, USA) in Fluorinert FC-40 (Sigma Aldrich, St. Louis, MO, USA) for 30 min at room temperature. The surfaces were rinsed with 95% ethanol (Koptec) and DI water and finally dried with a nitrogen gun.
Surface air/water contact angles and three-phase contact angles were measured on cubic printed pieces (8 x 8 x 15 mm3 ) using a ramé-hart goniometer (model 200-00, ramé-hart instrument co., Succasunna NJ, USA) and DROPimage Advanced software. For consistency, the smooth, flat face of the cube produced against the polytetrafluoroethylene (PTFE) sheet was tested in all cases; this was also the side of the print that faced the oil layer in the SlipChip. The contact angle was measured in triplicate (3 separate printed pieces per condition), by pipetting one 5-µL droplet of 1x phosphate buffered saline (PBS) (Lonza, Walkersville, MD, U.S.A.; DPBS without calcium or magnesium) onto the printed surface. For three-phase contact angle, 6 the printed cube with a droplet was inverted into a cuvette filled with FC-40 oil containing 0.5 mg/mL triethyleneglycol mono[1H,1H-perfluorooctyl]ether (RfOEG). RfOEG was synthesized in-house as reported previously [9].
Surface profilometry
To assess surface roughness, the root mean square deviation of the surface height of the printed parts were measured with a Zygo optical surface profilometer (Zygo, Berwyn, PA, USA) at the Nanoscale Materials Characterization Facility at the University of Virginia. Cubes of 8 x 8 x 8 mm3 were printed, and surface roughness was measured on all sides, specifically the surfaces printed against the aluminum baseplate or printed against glass, closest to the PTFE sheet at the bottom of the vat, and the sides of the print. As a positive control, a glass microscope slide was also analyzed after plating with 30 nm of Au/Pd by a Technics sputter coater (Technics).
Measurement of curvature of printed pieces
Images of the side profiles of 3D printed 30 x 30 mm prism with various heights (2 – 5 mm) were collected using a Zeiss AxioZoom microscope. The displacement from horizontal due to curvature was manually measured in Zen 2 software.
Animal work and tissue slice collection
All animal work was approved by the Institutional Animal Care and Use Committee at the University of Virginia under protocol #4042, and was conducted in compliance with guidelines of the Office of Laboratory Animal Welfare at the National Institutes of Health (United States). Both male and female C57BL/6 mice aged 19 - 21 weeks (Jackson Laboratory, USA) were housed in a vivarium and given water and food ad libitum. Lymph nodes were harvested from the mice following humane isoflurane anesthesia and cervical dislocation. The tissues were 7 sliced according to a previously published protocol [31]. Briefly, peripheral lymph nodes were collected and embedded in 6% w/v low melting point agarose (Lonza, Walkersville MD, USA) in 1× PBS. After the agarose had hardened, agarose blocks containing lymph nodes were extracted with a 10 mm tissue punch (World Precision Instruments, Sarasota, FL, USA). The blocks were mounted with super glue on a stage and sliced into 300-μm thick sections using a Leica VT1000S vibratome (Bannockburn, IL, USA) in ice-cold 1× PBS. The lymph nodes were sliced at a speed setting of 90 (0.17 mm/s) and frequency of 3 (30 Hz). Slices were cultured in “complete RPMI”: RPMI (Lonza, 16-167F) supplemented with 10% FBS (VWR, Seradigm USDA approved, 6 89510-186), 1× L-glutamine (Gibco Life Technologies, 25030-081), 50 U/mL Pen/Strep (Gibco), 50 μM beta-mercaptoethanol (Gibco, 21985-023), 1 mM sodium pyruvate (Hyclone, GE USA), 1× non-essential amino acids (Hyclone, SH30598.01), and 20 mM HEPES (VWR, 97064–362). Slices of 6% agarose were collected in a similar manner but were stored in 1 x PBS instead of complete media.
Analysis of tissue viability
Prior to assembling the SlipChip, the channel in the delivery component was filled using pressure-driven flow via a Chemyx syringe pump (Fusion 200, Houston TX, USA). A 0.5 mg/mL solution of FITC-conjugated dextran (150-kDa and 70-kDa for agarose and tissue deliveries experiments respectively) was flowed into the channel using a 50 μL Hamilton syringe (model 1705 RN; 26 s gauge, large hub needle) and non-shrinkable PTFE TT-30 tubing (0.012” I.D., 0.009” wall thickness, Weico Wire, Edgewood NY, USA). Next, 500 µL of FC-40 oil containing 0.5 mg/mL RfOEG was pipetted onto the top of the filled delivery component. The chamber component was lowered onto the delivery component, and the two components were clamped together with two binder clips, sandwiching a thin layer of oil between them. The culture chamber on the top of the chip was then filled with 1× PBS. A sample of agarose gel or tissue was placed into the chamber and weighed down using a small stainless-steel washer (10 mm O.D. and 5.3 mm I.D., Grainger USA). The chamber component was manually slipped 9 relative to the delivery component and visually aligned under a microscope to align to a desired port. To initiate a delivery, the syringe pump was turned on at the desired flow rate. After 5 seconds, the pump was turned off and the device was slipped away, to reposition for another delivery or to a reach a closed position. After all deliveries were complete, the sample was removed, and the chamber was flushed with 1× PBS and refilled for the next sample. All delivery experiments were performed at room temperature.
All deliveries were monitored in real time using a Zeiss AxioZoom upright microscope with a PlanNeoFluor Z 1×/0.25 FWD 56 mm objective, Axiocam 506 mono camera and HXP 200 C metal halide lamp (Zeiss Microscopy, Germany), using filter cubes for GFP (Zeiss filter set #38), and Violet Chroma Filter (49021, ET-EBFP2). Images (16 bit) were collected before, during, and after delivery. During deliveries, time lapse images were collected at 1 s intervals. All images were analyzed in Zen 2 software.
Analysis of tissue viability
After alignment of the delivery port to an array port below a 6% agarose slice, with a 5 sec pulse of fluorescein (FITC)-labeled 150-kDa dextran was delivered to the slice at flow rates ranging from 0.2 to 1 μL min−1 (n = 3). After delivery, the device was slipped prior to imaging, to avoid the fluorescent signal from the underlying channel. Delivery width was determined from image analysis as previously described [30]. Briefly, line scans were drawn radially across the delivery region, and the background autofluorescence of the resin was subtracted. The data was fit to a Gaussian curve in GraphPad Prism version 8. The width was defined as 2 standard deviations of the Gaussian curve.
To fit the curve of the spread of analyte with respect to time, we used the previously published analytical model [9]. First, we considered the volume delivered per unit time as described by a cylinder
Parts printed in BV-007A were silanized using our recently described method [25]. The parts were submerged into a 10% v/v solution of tridecafluoro-1,1,2,2-tetrahydrooctyl trichlorosilane (Gelest Inc., Morrisville PA, USA) in Fluorinert FC-40 (Sigma Aldrich, St. Louis, MO, USA) for 30 min at room temperature. The surfaces were rinsed with 95% ethanol (Koptec) and DI water and finally dried with a nitrogen gun.
Surface air/water contact angles and three-phase contact angles were measured on cubic printed pieces (8 x 8 x 15 mm3 ) using a ramé-hart goniometer (model 200-00, ramé-hart instrument co., Succasunna NJ, USA) and DROPimage Advanced software. For consistency, the smooth, flat face of the cube produced against the polytetrafluoroethylene (PTFE) sheet was tested in all cases; this was also the side of the print that faced the oil layer in the SlipChip. The contact angle was measured in triplicate (3 separate printed pieces per condition), by pipetting one 5-µL droplet of 1x phosphate buffered saline (PBS) (Lonza, Walkersville, MD, U.S.A.; DPBS without calcium or magnesium) onto the printed surface. For three-phase contact angle, 6 the printed cube with a droplet was inverted into a cuvette filled with FC-40 oil containing 0.5 mg/mL triethyleneglycol mono[1H,1H-perfluorooctyl]ether (RfOEG). RfOEG was synthesized in-house as reported previously [9].
To assess surface roughness, the root mean square deviation of the surface height of the printed parts were measured with a Zygo optical surface profilometer (Zygo, Berwyn, PA, USA) at the Nanoscale Materials Characterization Facility at the University of Virginia. Cubes of 8 x 8 x 8 mm3 were printed, and surface roughness was measured on all sides, specifically the surfaces printed against the aluminum baseplate or printed against glass, closest to the PTFE sheet at the bottom of the vat, and the sides of the print. As a positive control, a glass microscope slide was also analyzed after plating with 30 nm of Au/Pd by a Technics sputter coater (Technics).
Images of the side profiles of 3D printed 30 x 30 mm prism with various heights (2 – 5 mm) were collected using a Zeiss AxioZoom microscope. The displacement from horizontal due to curvature was manually measured in Zen 2 software.
To fit the curve of the spread of analyte with respect to time, we used the previously published analytical model [9]. First, we considered the volume delivered per unit time as described by a cylinder