The microfluidic chip made of PDMS was fabricated by photolithography and replica molding using photoresist patterns as master templates. SU-8, a negative photoresist on a silicon wafer, was patterned by photolithography, and 50 µm deep main channels were generated. After the photolithography process, the photoresist pattern was developed and washed to clear debris. Well-mixed PDMS prepolymer and curing agent (Sylgard184; Dow Corning, Midland, MI, USA) were poured onto the master template and cured in an oven at 120℃ for 1 hour. The PDMS plate was then peeled from the master template, and an inlet and an outlet were punched with a sharp needle. The surfaces of the PDMS plate and slide glass were oxidized by O2 plasma using a plasma reactor. After binding the PDMS piece to the slide glass, the device was kept at 90℃ for 1 h to allow complete annealing. Syringe pump and the inlet of the chip were connected by flexible polytetrafluoroethylene (PTFE) tubing (inner diameter, 1.0 mm; outer diameter, 1.5 mm). The outlet of the chip was connected to a reservoir by PTFE tubing.

We computationally simulated the wall shear stress distribution in the channel to identify the region of the flow channel in which a uniform shear stress level could be expected. Three-dimensional flow in the microfluidic channel was modeled using a finite volume method with microfluidics module of COMSOL Multiphysics 4.3 (Stockholm, Sweden) (Fig. 1c). A structured grid system was created to model the main region of the channel for high computing performance. The steady-state Navier–Stokes equations were solved to analyze the flow field. The working fluid was assumed to be a homogeneous, incompressible Newtonian fluid. The physical properties of the culture medium was as follows; density is 993.2 kg/m3 (same as water), dynamic viscosity is 0.000785 kg/m·s (1.04 times that of water) at 37℃ [35]. The boundary conditions consisted of a constant flow rate at the inlet and atmospheric pressure at the outlet. The no-slip boundary condition was assumed for other surfaces. The shear stress generated in the main channel was directly dependent on the velocity profile, because the flow could be modeled as Poiseuille flow in our microchip. For accurate prediction of WSS in the calculation, an appropriate grid density near the wall (channel bottom) was very important. Because wall shear stress was calculated as the first derivative of the flow velocity near the wall, an appropriate grid density near the wall was essential to accurately follow the velocity profile. Therefore, special care was given to the wall shear stress evaluation due to its strong dependency on the vertical grid density (Fig. 1c, inset). To quantify the detailed local distribution of shear stress, stress was computed along two orthogonal lines (A–B and X–Y in Fig. 1b). The results showed a flat shear stress profile everywhere within uniform shear region (Fig. 1b). Therefore, the cells were exposed to a uniform shear stress in a given observation region. Shear stress was calculated at 10 µm above the bottom of the channel, allowing for the height of the cell.

Primary cortical NSCs were isolated from cerebral cortical regions of embryonic day 16 rat embryos (Sprague-Dawley, Koatech, South Korea) using a surgical procedure as previously described [36] and authorized by the Institutional Animal Care and Use Committee (IACUC) of Korea University. NSCs were maintained in DMEM/F12 (Gibco) supplemented with B-27 Supplement w/o vitamin A (Gibco), Penicillin-Streptomycin-Glutamine (Gibco), BSA 7.5% (Gibco), N2 supplement (Gibco), epidermal growth factor (EGF) (Peprotech) and basic fibroblast growth factor (bFGF) (Peprotech). The chip was coated overnight by filling the channels with a solution of laminin (5 µg/ml, Sigma), followed by being coated with a solution of poly-D-lysine (10 µg/ml, Sigma) and washed with PBS before seeding cells. This surface treatment induces neural cell adhesion. A NSC suspension (200 µl of 5.7×10⁴ cells/ml) was loaded into the inlet of the chip using a micropipette tip. The cells in the suspension medium sank, attached to the bottom of the channels, and stabilized by being left undisturbed for about 2 hours, after which a constant flow of the medium from the inlet to the outlet was applied to wash out the untrapped cells. Finally, NSCs were cultured for 5 days under different shear stresses. For the Ca²⁺ imaging experiments we identified morphologically for the cells that are elongated and bipolar-shaped as RGCs.

Cultured RGCs were fixed with 4% paraformaldehyde (PFA) for 20 min at 4℃ and then washed with 0.1% bovine serum albumin diluted in phosphate-buffered saline (PBSA). The cells were treated with 0.1% Triton X-100 diluted in phosphate-buffered saline for 20 min at room temperature and then washed with 0.1% PBSA. The cells were blocked with 3% PBSA for 30 min at room temperature to reduce non-specific protein adsorption and washed with 0.1% PBSA. Afterwards, the cells were incubated overnight at 4℃ with primary antibodies: mouse anti-Nestin (1:500 diluted in 1% PBSA, Abcam). Cells were rinsed with 0.1% PBSA and incubated with Alexa Fluor 488-conjugated secondary antibody (1:1000 diluted in 1% PBSA, Invitrogen) for 90 min at room temperature. Fluorescence images were taken using a confocal laser-scanning microscope (Olympus, Japan) after counterstaining the cell nuclei with 4′, 6-diamidino-2-phenylindole dihydrochloride (DAPI) (1:1000 diluted in 1% PBSA, Invitrogen).

For imaging, cultured RGCs were plated as a monolayer on 12-mm glass coverslips coated with poly-D-lysine (Sigma). Astrocytes were incubated with 5 µM Fura-2 AM plus 1 µM pluronic acid (Molecular Probes) for 30 min at room temperature. External solution contained 150 mM NaCl, 10 mM HEPES, 3 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 22 mM sucrose, and 10 mM glucose (pH adjusted to 7.4 and osmolarity to 325 mOsm). Intensity images of 510 nm wavelength were taken at 340 nm and 380 nm excitation wavelengths using iXon EMCCD (DV887 DCS-BV, ANDOR technology, UK). Imaging Workbench version 6.0 (Indec System, CA, USA) was used for acquisition of intensity images and conversion to ratios as previously described [37].

To quantify the cell number inside the main channel (Fig. 2c), the DAPI staining images were further analyzed by the image analysis software ImageJ.

For measurements of peak amplitude of Ca²⁺ signal, each 10 seconds of data right before the mechanical stimulation were averaged and subsequently used as a reference baseline from which changes were measured. Search region was set to find the maximum value of Ca²⁺ signal and peak amplitude was calculated using Clampfit 10.4 (Axon Instruments, USA).

All quantitative data were expressed as means±standard error of the mean. D'Agostino & Pearson normality test was performed to judge if data are normally distributed or not. Statistical differences were determined by two-tailed, unpaired t-test with Welch's correction since our data are normally distributed but the variances are unequal.

To mimic the physiologically relevant dynamic shear stress experienced by RGCs in ventricular zone, we fabricated a novel microfluidic chip. Fig. 1a illustrates the chip design concept. The chip contains 8 identical main channels for high throughput experiments. And each main channel is 2.5 mm in width, 8 mm in length, and 0.05 mm in height. The aspect ratio, the proportional relationship between the width and height of the channel, is large enough to give uniform shear stress distribution throughout a large area of the middle region of the channel termed uniform shear region. The wall shear stress (τ) on the cell surface was estimated from τ≈µ(∂u/∂y) at y=10 µm, where µ is the dynamic viscosity of the fluid, u is the axial velocity, and y is the height (Fig. 1a). As predicted in the simulations, only the cells in uniform shear region were assumed to be exposed to the same level of shear stress (Fig. 1b). And desired levels of shear stress over RGCs can be induced by applying corresponding flow rates (Fig. 1c). The microfluidic chip made of PDMS was fabricated by photolithography and replica molding using photoresist patterns as master templates (Fig. 1d). RGCs were seeded into the microfluidic chip and continuous, in vivo-mimicking shear stress powered by a syringe pump was applied on RGCs (Fig. 1e).

To study the effect of shear stress on the proliferative potential of RGCs, we applied an in vivo-mimicking shear stress (0.01 dyne/cm²) continuously over RGCs in culture for 5 days. The estimated physiological ventricular wall shear stress of CSF in vivo (>Postnatal day 20 in mouse) is in the range of 0.01~0.018 dyne/cm² [28]. We also prepared cultures of RGCs under extremely low shear stress (0.001 dyne/cm²) as a control condition, instead of those under static condition to avoid depletion of nutrient, oxygen, and soluble factors. Fig. 2a shows the daily bright-field images of RGCs exposed to 0.001 dyne/cm² shear stress for 5 days. Under 0.001 dyne/cm² shear stress condition, the number of cells did not change dramatically over 5 days. However, the number of RGCs cultured under 0.01 dyne/cm² shear stress was dramatically increased (Fig. 2b).

To accurately quantify the number of proliferated cells and to identify the type of cells, we immunostained the cultures of RGCs against neural stem cell marker, nestin, and nuclear marker, DAPI. The intensity of nestin and DAPI immunostaining was significantly higher in the cultures under 0.01 dyne/cm² compared to 0.001 dyne/cm² (Fig. 2c). The number of DAPI-positive cells on DIV5 was counted and there was a significant difference between 0.001 dyne/cm² and 0.01 dyne/cm² (Number of cells (DIV5), 0.001 dyne/cm²: 1691±58.52, n=8 chips; 0.01 dyne/cm²: 5244±140.8, n=8 chips) (Fig. 2c). These results indicated that the increase in the number of cells was a direct consequence of enhanced proliferation, suggesting that flow shear stress is one of the critical factors for enhancing RGC proliferation.

The most well-known biosensor for flow shear stress is a mechanosensitive channel. To study if RGCs have mechanosensitive Ca²⁺ channels, we gave mechanical stimulation on RGCs using glass pipette while performing Fura-2 Ca²⁺ imaging. We identified the cells with elongated and bipolar processes as RGCs as shown in insets of Fig. 3a. A glass pipette was placed above a cultured single RGC and lowered slowly while the pipette resistance (R_t) was continuously monitored to determine when the pipette touched the cell membrane as previously described [38]. Once the Fura-2 signal started to change, the pipette was immediately withdrawn from the cell membrane [38]. Poking was given until the change in pipette resistance (ΔR_t) reached about 0.5~1.5 MΩ (Fig. 3a). Fig. 3b shows the representative Fura-2 intensity images before and after mechanical stimulation on a cultured RGC reflecting the increase of Ca²⁺ signals in cytoplasm of RGC evoked by poking (Fig. 3b). Some RGCs were responsive to mechanical stimulation (n=27 out of 48 single cells), while other cells showed no Ca²⁺ response to mechanical stimulation (n=21 out of 48 single cells) (Fig. 3c). To determine the source of mechanically-induced Ca²⁺ increases, we tested whether Ca²⁺ increases by mechanical stimulation can be measured under extracellular Ca²⁺-free condition. Under normal condition, Ca²⁺ increases were evoked by two sequential mechanical stimulations separated by at least a 20 minute break (Peak ratio (2^nd/1^st), 1.12±0.2082, n=8 cells). However, under extracellular Ca²⁺-free condition, Ca²⁺ increases evoked by 2^nd poking almost completely disappeared (Peak ratio (2^nd/1^st), 0.09025±0.02227, n=8 cells) (Fig. 3d). These results suggest that RGCs have mechanosensitive Ca²⁺ channels which give rise to a Ca²⁺ entry evoked by a mechanical stimulation.