The VersaLive platform enables microfluidic mammalian cell culture for versatile applications

The VersaLive platform enables microfluidic mammalian cell culture for versatile applications


Design and computer simulations of the device

The VersaLive microfluidic platform was designed using Autodesk Fusion 360. The velocity profile of the chip was modeled using the finite element method (COMSOL Multiphysics 5.4). In the simulated utilization modes (Fig. 2), 0.5 mbar of pressure was applied at every inlet while the outlets were set at zero pressure. This value of pressure was chosen because equivalent to the 5 mm of hydrostatic pressure when the reservoirs are filled to the maximum of their volume capacity. Flow direction in the simulation plots was indicated using arrows whose dimensions reflect the magnitude of the velocity in a logarithmic scale. A 2D CAD model of the chip was exported into Autodesk AutoCAD 2020 to design the printable photolithography mask.

Fabrication of the VersaLive microfluidic platform

Microfluidic chips were fabricated by using a combination of standard photolithography and soft lithography procedures24. The master mold was microfabricated via mask photolithography of SU-8 negative photoresist (SU-8 3035, Kayaku Advanced Materials Inc.) on silicon wafer substrate. The photoresist was spin coated to reach a final thickness of 25 µm and processed following the guidelines of the manufacturer. Feature height was confirmed by measurement via optical microscopy of the cross-section of a silicone replica of the channels. Before the first use of the master, the passivation of its surface is required to facilitate the release step during soft lithography. The master was passivated by vapor deposition of perfluorosilane (1H,1H,2H,2H-perfluorooctyl-trichlorosilane, Merck kGaA). Specifically, the master was placed in a desiccator with a small vial containing 20 µL of perfluorosilane. Vacuum was then applied overnight to allow the perfluorosilane to evaporate and to react with the surface of the silicon wafer, forming a covalently bound super-hydrophobic coating. The passivated master was then used as a mold for the soft lithography part of the fabrication process of the microfluidic device. The VersaLive microfluidic platform is formed by a silicone elastomer (PDMS) bonded to a glass slide. The elastomer base of the PDMS was thoroughly mixed with the curing agent in a 10:1 ratio as reported by the datasheet of the manufacturer (Sylgard 184, Dow Corning). Air bubbles were removed from the uncured polymer applying vacuum to the mix for 2 h or until no bubbles were visible. The mix was then poured onto the master mold for a final thickness of about 5 mm. If required, vacuum was applied a second time to remove the bubbles formed during the pouring step. The mold with the uncured PDMS were then placed in oven at 80 °C for a minimum of 2 h to accelerate the crosslink of the polymer mix. Once cured, the PDMS was peeled off the master mold. The PDMS slab was then placed with the pattern features facing up to cut out the single chips. Similarly, access ports were opened using a 3-mm biopsy punch (ref. 504649, World Precision Instruments). The channels of the PDMS chips were sealed by plasma bonding the chips to round glass cover slide (30 mm in diameter, thickness no. 1, Marienfeld). Glass slides and PDMS chips were first cleaned from dust particles using adhesive tape. For the surface activation of glass slides and PDMS chips, 85 W air plasma at 0.4 mbar or lower (ZEPTO version B, Diener electronic GmbH & Co. KG) were used. For the permanent or reversible bonds, 30 or 10 seconds of air plasma were used, respectively. The chip was then placed in an aluminum lens tube (SM30L05, ThorLabs) used as chip holder. The lens tube allowed the safe transfer of the chip during the experimental workflow (e.g., workbench, incubator, microscope). To allow an easy and reliable image acquisition procedure, the Nikon microscope was equipped with a custom holder for the lens tubes made of 2-mm laser-cut acrylic.

Cell culture

HeLa WT, together with AU565 and HCC38 breast cancer cell lines were purchased from ATCC and grown in RPMI 1640 (without L-glutamine, EuroClone) cell medium. CHO-K1 reporter cells (kind gift from Prof. David Ron)18 were grown in F-12 (Gibco) cell medium. Both cell media were supplemented with 10% fetal bovine serum (FBS, EuroClone), 1% L-glutamine (EuroClone) and 1% penicillin-streptomycin (EuroClone).

Human Proximal tubule epithelial (HK2) cell line was bought from ATCC (#CRL-2190). HK2 cells were cultured in Dulbecco’s Modified Eagle Medium/F12 (DMEM/F12, Gibco) supplemented with 5% FBS, 2 mM L-glutamine, 1 U/ml antibiotics (penicillin/streptomycin), and 1% insulin-transferrin-selenium (ITS-Sigma Aldrich).

Cells were maintained in a cell incubator at 37 °C, 100% humidity and 5% CO2 atmosphere. Prior to the loading of a microfluidic device, the content of a T25 flask at 90% confluence was collected according to the following procedure. First, the cell medium in use was removed and the cells were rinsed with 2 mL of phosphate-buffered saline without calcium and magnesium (PBS, EuroClone). To detach the cells from the flask, 0.5 mL of 0.05% trypsin-EDTA (Gibco) were added and the flask was placed back in the incubator for two minutes. To deactivate the trypsin, 2 mL of cell medium was added to the flask and the cells were thoroughly mixed until all visible aggregates were debulked. The resulting cell suspension was directly used for the loading onto the chip. Primary mouse retina pigment epithelium (RPE) were a gift from Sabrina Carrella, PhD (TIGEM) and were obtained as previously described25.

Channel wetting and long-term storage of the chips

The plasma treatment used for the PDMS-to-glass bonding also gives a temporary hydrophilicity to the surface of the channels24,26. A hydrophilic channel surface is key for allowing any water-based solution to flow into the microchannels (i.e., wetting). To optimize the fabrication workflow of a batch of chips, the wetting process was carried out within minutes after the bonding procedure. For the wetting of the microfluidic channels, 10 µL of PBS were added in port B of the chip until all channels were filled. If required, all ports were filled with 10 µL of PBS and the chip was placed in a desiccator and vacuum was applied for 15 min or until all air bubbles disappeared. For long term storage (4–6 weeks), all reservoirs of the chip were filled with 20 µL of PBS, the PDMS side of the chips were sealed with office tape (Scotch Magic Tape, 3 M) and the chips were kept at +4 °C until use.

Loading of the cells onto the microfluidic chip and static cell culture

For the loading of the cells, the PBS used for the wetting of the channels is removed from all reservoirs and 10 µL of cell suspension (1–5 × 106 cells/mL) is added to port B of the device. Upon filling of the reservoir, cells immediately start to flow through the main channel and to enter into the chambers (Fig. 2b). When a given chamber is filled with the suited number of cells, 10 µL of cell medium are added to the respective port to decrease the flow rate across that chamber. When all chambers are filled with cells, 20 µL of cell media are added to port A and port B is emptied to wash the main channel from undesired cells. Port B is then rinsed from the residual cells and filled with 20 µL of cell media. Next, all input ports from #1 to #5 are filled up to a final volume of 20 µL of fresh cell media. An equal volume of cell media in all ports prevents the formation of pressure drops across the chip and enables the static cell culture. Ultimately, 2.5 µL of mineral oil for cell culture (M8410, Merck kGaA) are added to each reservoir to prevent the evaporation of their content. The chips stayed in the incubator overnight at 37 °C, 100% humidity and 5% of CO2 atmosphere prior to the beginning of the experiments. The outcomes of the wetting and the cell loading operations were checked using an inverted stereomicroscope (Leica Microsystems).

Live cell imaging acquisitions

Time lapse acquisitions of CHO-K1 cells were acquired via a Nikon Ti Eclipse microscope equipped with a mercury lamp (Intensilight, Nikon), a EMCCD digital camera (iXon Ultra 897, Andor Technology Ltd) and an incubation chamber (H201-OP R2, Okolab). Prior to the mounting of the chip, the incubator was equilibrated to a temperature of 37 °C and 5% CO2 humidified air. Time-lapse acquisitions up to 20 h were acquired using a ×40 air objective (CFI Plan Fluor DLL ×40, 0.75 NA, Nikon Instruments) to collect the highest possible signal. However, because of the off-center position of chamber #5 in this specific device, the acquisition of this chamber with this objective was not achievable in our microscope. For the single-cell analysis, the fluorescence of the cells from all chambers was acquired at the end of the 20 h treatment using a ×20 air objective (CFI Plan Fluor DLL ×20, 0.5 NA, Nikon Instruments). Depending on the experiment, images were collected in phase contrast (PC) and epifluorescence for green, red and yellow wavelengths using the exposure times and filter sets respectively reported in Table 2.

Table 2 Summary of the microscopy settings used in this work.

Stress response assay of CHO-K1 cells under tunicamycin stimulus

After overnight seeding, VersaLive chips with CHO-K1 cells in static culture were removed from the incubator. To each chip, reservoirs A and B were emptied. Next, the content of reservoirs #2 and #4 was renewed with 15 µL of fresh F12 cell medium. Then, the content of reservoirs #1, #3, and #5 was replaced with 15 µL of tunicamycin 0.5 µg/mL in F12 cell medium. Then, 1 µL of sulforhodamine B 0.1 mg/mL (Merck kGaA) was added to reservoir #3 as flow tracer. To all filled reservoirs (#1 to #5), 2.5 µL of mineral oil for cell culture were added to prevent their evaporation. Reservoirs A and B were left empty. The chip was then mounted on the microscope for live cell imaging for a time lapse acquisition of 20 h at intervals of 15 min between each acquisition.

To quantify the stress response of CHO-K1 cells treated and untreated with tunicamycin, live cell microscopy images were analyzed using the Trainable WEKA Segmentation27 plugin in the ImageJ software. In brief, the software was trained in distinguishing CHO cells within the culture chambers of VersaLive. The probability map returned by the software was thresholded to be converted in a segmented mask. The mask was then superimposed onto the original image. The average intensities of the resulting single cells were used to quantify their stress response.

On-chip immunostaining of AU565 and HCC38 breast cancer cells

AU565 and HCC38 breast cancer cell lines were loaded on VersaLive chips and grown overnight in static cell culture as described in a previous section of this work. For the on-chip chemical fixation, after removal of the cell medium, all ports were washed with 20 μL of PBS. Cells were perfused in multi-input mode with paraformaldehyde (PFA, 4%w/v in PBS) for 10 min. Next, all ports were washed twice with 20 μL of PBS followed by 5 min of perfusion of PBS in multi-input mode. To quench the fluorescence of PFA, a blocking solution (50 mM NH4Cl; 0.5% BSA in PBS) was perfused in multi-input mode for 30 min. Ultimately, all ports were washed with 20 μL of PBS. For the on-chip immunostaining, anti-HER2 antibody (BB700 Mouse Anti-Human Her2/Neu, BD OptiBuild) diluted 1:100 in blocking buffer was perfused in multi-input for a time defined by the experiment (10 or 30 min). Then, all ports were washed twice with 20 μL of PBS. To preserve the sample, all ports were filled with 20 μL of PBS and 2.5 μL of mineral oil for cell culture to prevent evaporation. Samples were stored at 4 °C.

On-slide immunostaining of AU565

AU565 (ERBB2+, positive control) and HCC38 (ERBB2−, negative control) breast cancer cell lines were seeded at 60–70% confluency on glass coverslips (EXACTA-OPTECH, area 10 × 10 mm2, thickness 0.13–0.16 mm) in 24-well plate.

Following PBS 1× washing, cells were fixed with PFA 4% for 10 min, and then washed with PBS 1×. After fixation, cells were then blocked with blocking buffer, prepared with BSA 0.5% (Sigma-Aldrich) and NH4Cl 50 mM (Sigma-Aldrich) in PBS 1×, for 1 h at room temperature (RT). Cells were incubated in humidified environment at RT in the dark with 1:100 primary antibody (BB700-conjugated mouse antihuman-ERBB2 Ab, BD Bioscience). Incubation time was set according to the experimental plan. Primary antibody was prepared in blocking buffer.

Following antibody incubation, cells were washed three times with PBS 1× and once more with ddH2O. Then, coverslips were mounted in Fluoroshield with DAPI (Sigma-Aldrich) and fluorescence images were acquired using a Nikon Ti Eclipse microscope equipped with a mercury lamp (Intensilight, Nikon), a EMCCD digital camera (iXon Ultra 897, Andor Technology Ltd) and a ×20 air objective (CFI Plan Fluor DLL ×20, 0.5 NA, Nikon Instruments).

On-chip immunostaining of primary mouse RPE cells

Primary mouse RPE cells were loaded on VersaLive chips and let adhere to the microfluidic chip surface overnight in static cell culture, as described in a previous section of this work. On-chip chemical fixation was carried out as previously described for the cancer cell lines. Immunostaining on chip was initiated by perfusing for 10 min the permeabilization buffer (0.3% Triton X-100, 5% FBS in PBS) in multi-input mode. Next, the anti-citrate synthase primary antibody (ab96600, Abcam) in blocking buffer (0.5% BSA, 0.05% saponin, 50 mM NH4Cl, 0.02% NaN3 in PBS) was perfused in multi-input mode for 10 min. All ports were then washed with 20 μL of PBS. Successively, fluorescently labeled donkey anti-rabbit Alexa Fluor 488 (A-21206, Thermo-Fisher Scientific), Alexa Fluor 568 phalloidin (A-12380, Thermo Fisher Scientific) and DAPI (D1306, Thermo Fisher Scientific) were perfused for 10 min in multi-input mode to stain mitochondria, f-actin and nuclei, respectively. All ports were then washed twice with 20 μL of PBS. To preserve the sample, all ports were filled with 20 μL of PBS and 2.5 μL of mineral oil for cell culture to prevent evaporation. Samples were stored at 4 °C.

Primary mouse RPE cells were imaged via confocal microscopy (LSM 700, ZEISS Microscopy) equipped with 405, 488, 555 nm lasers and a ×40 oil objective (EC Plan-Neofluar ×40/1.30 Oil DIC). For the acquisition, DAPI (blue), FITC (green) and Texas Red (red) filters were used.

Transferrin recycling assay

VersaLive chips were coated with truncated human recombinant vitronectin (5 μg/mL in PBS 1×) using the chip in multi-input mode for 30 min at room temperature prior to cell loading. After cell loading on chip, cells were let adhere overnight in incubator (37 °C, 5% CO2). At the moment of the experiment, HK2 cells were perfused in multi-input mode at 37 °C for 30 min with serum-free medium with 50 μg/mL Alexa Fluor 488-conjugated transferrin (T13342, Thermo Fisher Scientific) to allow endocytosis and loading of recycling endosomes. Then the medium was changed and serum-free medium with 50 μg/mL unlabeled transferrin was added to allow the recycling of Alexa Fluor 488-conjugated transferrin that was evaluated by live imaging by using a Nikon Ti microscope equipped with a spinning disk module and a ×100 1.5 NA oil objective.

On-chip plasmid transfection

For the on-chip transfection, HeLa cells were loaded one day prior to transfection following the already described procedure. The transfection mix (Lipofectamine™ 3000 Transfection Reagent, ThermoFisher Scientific) was prepared according to the manufacturer guidelines and optimized to the VersaLive volumes. The mix contained 300 ng of DNA, 0.5 μL of Lipofectamine reagent, 0.6 μL of P3000 reagent and 10 μL of Opti-MEM medium. This mix was then diluted in complete growth medium to a final volume of 50 μL. The cells were perfused via multi-input mode in a cell incubator for 2 h adding 5 μL of diluted transfection mix in each chamber port and mineral oil to prevent their evaporation. After transfection, the mix was replaced by complete growth medium and the chip was left in static single-input mode in a cell incubator for 24 h. Then, images were acquired using the Nikon Ti Eclipse microscope.

Multi-well plasmid reverse transfection

Hela cells were reversely transfected with pCMV-NLS-mcherry-ccp3 obtained from Golden gate cloning (EMMA toolkit) and an empty vector using Lipofectamine 3000 (ThermoFisher Scientific) according to the manufacturer’s instructions. Twenty-four hours post-transfection, cells were harvested to measure fluorescence via fluorescence activated cell sorting (FACS) analysis.

Evaporation of the inlet reservoirs

The effect of the evaporation at the inlet reservoirs was assessed by running two different microfluidic devices in multi-input mode. Input reservoirs #1, #2, #4 and #5 were filled with 25 µL of DI water; input reservoir #3 was filled with 25 µL of rhodamine B (0.1 mg/mL). For one of the two devices, 2.5 µL of mineral oil for cell culture were added to the inlet reservoirs to prevent their evaporation. The intensity of rhodamine B was measured within the culture chamber over 15 h at 15 min intervals. Measurements were performed at the same experimental conditions used for live cell microscopy experiments. Data was processed using ImageJ software.

PDMS removal upon reversible bonding

To facilitate the removal of the PDMS, a single-edge razor blade was gently inserted at the interface between the PDMS and the glass slide all along the perimeter of the chip (Supplementary Fig. 1a). Successively, the PDMS chip was pinched to peel it off the glass slide that sealed the channels (Supplementary Fig. 1b). For the out-chip chemical fixation, the PDMS chip was first removed from the glass slide. Then, the glass slide with the exposed cells was rinsed with PBS and later immerged in the 4%wt PFA solution for 30 min at room temperature. The slide was then rinsed with PBS and gently dried.

Chip content recovery for RNA extraction

Poly(A)-RNA was captured using poly(T)-resin microparticles, as previously described by the authors in refs. 11,28. The lysis buffer, with the composition described in literature28, was diluted 1:1 in PBS 1× prior to start the experiments. Before each extraction run, a stock suspension of beads in lysis buffer (3000 beads/μL) was prepared.

The sequence for recovering the chip content starts with the chip in multi-input mode (10 μL per reservoir, Supplementary Fig. 2a). Next, port A is filled with lysis buffer (7.5 μL). In this configuration, the lysis buffer is able to flow through the main channel but not to enter the chambers. To start the lysis of a chamber, however, it is sufficient to empty the corresponding chamber reservoir (Supplementary Fig. 2b–f). To collect the poly(A)-RNA and preserve it from degradation, an aliquot of beads (3 μL) was placed at each port (#1 to #5). This amount was sufficient to cover the bottom surface of the port without creating backpressure. After 10 min, no cells were still visible in the chambers and the lysis process was considered completed.

Resin microparticles were recovered from each chamber reservoir (#1 to #5) and put into 200 μL tubes. The microparticles were then washed twice with 100 μL of SSC 6X and once with 100 μL of RT buffer. All centrifugation steps were performed at 1000 × g for 1 min. The reverse transcription, the exonuclease, PCR and cDNA library purification steps were performed as reported in literature28. The amplified cDNA was purified with the Ampure XP bead protocol. A 0.6× (sample volume/Ampure XP beads volume) volume of Ampure XP beads was added to each sample. Finally, the cDNA was eluted in 10 μL of RNAse-free water. Quantitative and qualitative analysis of each sample was performed using a TapeStation D5000 high sensitivity chip.

Statistics and reproducibility

Image analysis and data processing was performed using ImageJ and Microsoft Excel and error bars represent median ± SD unless otherwise noted. Replicates are biological, representing experiments on the same cell line but performed on different days. Experiments were repeated at least three times. Sample size varied depending on methodology and is defined in figure legends.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.


Source link

Jorge Oliveira