The biochemical and hydrodynamic errors of the gradient chip

In several studies, the microfluidic channel is directly connected to the reservoir and shares the substrate at the bottom of the microfluidic chip [31, 32]. Thus, the cells become attached and cultured on the reservoir bottom. In the present experiment, the cell suspension was injected into the channel outlet and filled the entire channel, which evenly distributed the cells in the channel (Fig. 2b). Therefore, the cell suspension occupied the inlet port in the gradient chip (Fig. 4a, i). On the other hand, in the gradient chip without ports, the cells reached the reservoir and occupied the entire bottom of the reservoirs (Fig. 4a, ii). Since the condition of the reservoir without an inlet port is not a microfluidic environment, in this reservoir, cells were cultured in a 2D environment unintended for the utilization of this microfluidic system (The supplementary figure S5). The simulation was performed with flow rates of 0.342 μL/h and 3.42 μL/h. At a flow rate of 0.342 μL/h (Fig. 4a, iii and iv), the biomolecules were continuously diluted with the culture medium and introduced into the microfluidic channel during 10 h. The biomolecules had diffused through the outlet of the microfluidic channel, and at 48 h, the biomolecules were present over the entire area of the microfluidic channel in both gradient chips with and without ports. However, at a flow rate of 3.42 μL/h, since the cell secretion had been diluted with a larger amount of the culture media than at the lower flow rate, the biomolecules flowing into the channel were reduced (Fig. 4a, v and vi). Especially, in the gradient chip with the port, the biomolecules were diluted with the culture media, and the concentration was maintained for 48 h (Fig. 4a, v). In the gradient chip without ports, although the biomolecules had been diluted, their mass fraction gradually increased, and the reservoirs became considerably saturated at 48 h (Fig. 4a, vi). With a flow rate of 0.342 μL/h, the mass fraction of the biomolecules gradually increased and became saturated after 24 h (Fig. 4b, i). The saturated mass fractions in the gradient chip with and without ports were predicted to be 9. 81 × 10–2 and 9.97 × 10–2, respectively. These data show that the cells in the port affect the cells in the channel by 1.55% less than the cells in the reservoir without port. Although it was difficult to confirm this trend on the mass fraction contour (Fig. 4a, iii and iv), the quantitative data confirmed the positive effect of the inlet port. This reduction in unnecessary biomolecules became apparent as the flow rate was increased (Fig. 4b, ii). At a flow rate of 3.42 μL/h, the secretions from the cells cultured in the port became saturated with a mass fraction of 6.61 × 10–2 in the channel whereas in the gradient chip without ports, the saturated mass fraction increased by 29.5% (9.38 × 10–2). This CFD analysis indicates that more careful evaluation of the reality of the experimental procedure into the microfluidic chip design can improve the accuracy of the experiment. In conclusion, it was predicted that the benefit of the port would be more effective at higher flow rates. However, since the higher the flow rate inside the gradient chip, the sharper the concentration gradient is formed, researchers need to consider all of these issues when determining the flow rate of the channel [12].

Fig. 4

The functions of the inlet port. a The inflow of biomolecules cultured at the bottom of the gradient chip inlet. The unintended cell culture area of (i) the inlet port of the improved gradient chip and (ii) the reservoir without the port. Mass fraction contours of biomolecules under flow rate conditions of (iii and iv) 0.342 μL/h and (v and vi) 3.42 μL/h. b Mass fraction graphs measured at the center of the microfluidic channel (red colored point) under flow rates of (i) 0.342 μL/h and (ii) 3.42 μL/h. The red points in the inset images indicate where the mass fractions of the biomolecules were measured. c As a passive valve to prevent flow between the inlets during chip handling: schematics of (i) the capillary force on the inlet port fluid and (ii) the hydrostatic force generated by the height difference between the inlet ports of the inclined gradient chip. d A graph of the forces applied to the fluid in the gradient chip inlets

This narrow port design compensated for another error factor in the microfluidic chip. We demonstrated the applicability of the gradient chip with its reservoirs open to the atmosphere to reduce hydrodynamic instability and interference from redundant cells cultured in the reservoirs. The capillary force in the port connecting the reservoir and the microfluidic channel functions as a passive valve to prevent gravity-driven flow (Fig. 4c). Two forces act on the fluid in the ports. The first is capillary force (FC) that holds the fluid in the port (Fig. 4c, i), which is defined as

$$_=\gamma 2\pi rcos\theta$$

(5)

where γ is the surface tension (0.07 N/m) [33], θ is the contact angle (81°) [34], and r is the port radius. Using this equation, the FC value of the fluid in the ports and reservoirs without ports were calculated as 0.103 and 0.550 mN, respectively. On the other hand, the gravity force (FG) applied to the fluid in the inclined chip is obtained as

$$_=\rho g\Delta h\pi ^$$

(6)

where ρ is the fluid density, g is the gravitational acceleration (9.81 m/s2), and ∆h is the height difference between the top surface of the fluid in the ports (Fig. 4c, ii). The reservoirs contain a greater weight of fluid in the gradient chip without ports than in the one with ports (Fig. 4c, ii and iii). Depending on the ∆h value, the FG value of the fluid in the reservoirs of the chip without ports increases significantly compared to the force in the ports (Fig. 4d). In the ports, the fluid remains static when the capillary force is greater than the gravity, and derived from Eqs. (5) and (6), the height difference at which the FC value and gravity are at equilibrium is calculated as

In the gradient chip without ports (8 mm-diameter reservoirs), the ∆hE at which the FG and the FC values are balanced is 1.12 mm. However, in the inclined device with ports, the effect of capillary force is enhanced, and so ∆hE is increased to 5.99 mm. We observed the passive valve function by surface tension in a narrow port through experiments (supplementary figure S6). Consequently, the flow instability during chip handling is significantly inhibited by using the port as a passive valve.

We verified the effect of the port on cell distribution and flow stability, which suggests that, not only the cell manipulation technology, but also the delicate design of unnoticed geometric factors constituting the microfluidic chip should be considered for the precise realization of the experimental conditions of the microfluidic chip.

The diffusion simulation of BPA concentration

We performed the diffusion simulation to predict the BPA concentration in the gradient chip. To verify our CFD model, the experiment was performed using trypan blue, which is easy to analyze with the naked eye. In the CFD model, the diffusion coefficient of trypan blue was applied, and the mass fraction contour confirmed in the CFD model predicted the color intensity of the trypan blue gradient similarly (supplementary figure S7). The gradient profile of the CFD model was consistent overall with the normalized intensity measured in the image (supplementary figure S7). Therefore, we used this CFD model to calculate the BPA concentration in the gradient chip (Fig. 5). The culture medium and 200 μM BPA solution were introduced through separate inlets (Fig. 5a). The profile of the BPA gradient was measured upstream (line A), midstream (line B), and downstream (line C) of the channel. The concentration contour shows that the BPA concentration gradient was formed in the cell culture area, and diffusion-mixing occurred gradually as the BPA flowed downstream (Fig. 5b). The diffusion-mixing reduced the concentration gradient in the flow; the BPA concentration ranges were calculated as 120.8–79.3 μM upstream, 108.0–92.1 μM downstream, and 103.1–97.1 μM upstream and was predicted as 100 μM at the center of each area (Fig. 5c, i–iii).

Fig. 5

The diffusion simulation results. a Schematic of the gradient chip. The normal culture medium and the culture medium containing 200 uM BPA flow into each inlet of the gradient chip. The GC-1 cells were observed in the square Sects. (1.6 × 0.29 mm) with three columns (aligned on lines A, B, and C) and 10 rows. b The mass fraction contour of the BPA concentration in the gradient chip. c The graphs of BPA concentration (i) upstream, (ii) midstream, and (iii) downstream (lines A, B, and C in the insets, respectively)

Since only an infinitesimal amount of BPA was required in the experiment, it is difficult to prepare several BPA solutions with slight differences in concentration. However, the concentration gradient created by this chip easily provides the experimental conditions required to test slight differences in BPA concentration simultaneously. In addition, it is possible to control the concentration range in the microfluidic channel by adjusting the concentration of the BPA solution loaded on the inlet reservoirs.

Testing the effect of BPA concentration on GC-1 cells

In the main test section where the GC-1 cells attached uniformly for 12 h (Fig. 6a, i–iii), 200 μM BPA was introduced in the channel through one of the inlets, then the laminar flow in the channel generated the BPA gradient. After exposure to the BPA gradient for 2 days, the GC-1 cells were damaged by the BPA (Fig. 6a, iv–vi); the cells had shrunk, and their pseudopodia had almost disappeared. In addition, it can be observed that the number of cells decreased depending on the BPA concentration in each section. Compared to columns B and C, the morphological alterations and reduction in the number of cells were prominent in column A (Fig. 6a, iv). From the results of the CFD analysis, the concentration difference in BPA between Sections A1 and A10 was 41.5 μM, and so the BPA gradient had obviously formed in column A (Fig. 5c, i). On the other hand, the BPA molecules were mixed as the flow proceeded downstream, and so columns B and C showed reductions of the concentration difference between rows 1 and 10 of 15.9 and 6.0 μM, respectively (Fig. 5c, ii and iii). The number of cells in each section was counted and the cell proliferation rate was then calculated (n = 4). The GC-1 cells were generally reduced by the BPA. Since the BPA gradient was formed steeper in the upstream (Fig. 5c), the decrease in cells according to the BPA concentration was also evident in column A, corresponding to the upstream region (Fig. 6b). In the proliferation rate graph, the slopes of the trend lines was calculated as 2.2126, 0.3892, 0.0481 for columns A, B, and C, respectively. The experiment using the gradient chip facilitated cell observation in a designated local area. Therefore, the morphological adjustment of GC-1 cells could be directly traced, and the data provided by the CFD technique supported the analysis of the cell proliferation depending on the BPA concentration. Although many biological studies have been conducted on the microcellular environment of the testis from which GC-1 is derived, more research on the mechanical phenomena in this tissue needs to be performed. This study is expected to contribute to the diversification of testicular cell-related experimental techniques through the application of a microfluidic chip. Determination of optimal conditions to simulate the testis microenvironment is considered a challenge to be achieved with this experimental methodology in the future.

Fig. 6

The effect of BPA on GC-1 cells in the gradient chip. a Bright field images of GC-1 cell reduction in the upstream region (sections A1, A5, and A10), the midstream region (section B1, B5, and B10), and the downstream region (sections C1, C5, and C10). b Graphs of the GC-1 cell proliferation rates observed in (i) the upstream region (column A), (ii) the midstream region (column B), and the downstream region (column C). Each equation in the graph indicates the linear trend (the dashed lines)

For comparison, the effect of BPA on the GC-1 cells cultured in a static 2D environment was tested. CG-1 cells were cultured with 0, 100, and 200 μM BPA. In the brightfield images in Fig. 7a, it can be observed that the GC-1 cells were damaged by the BPA. These results show that both 100 and 200 μM BPA were toxic to the GC-1 cells, and especially, most of the cells had become detached from the bottom when dosed with 200 μM BPA. The proliferation rates were 55.7% in 100 μM BPA and 0.27% in 200 μM BPA, the latter being extremely low (Fig. 7b). In the gradient chip, the middle of the channel (between rows 5 and 6 in columns A, B, and C in Fig. 5a) is the area where a concentration of 100 μM BPA formed (Fig. 5c), and on average, the cell proliferation rate was estimated in the middle of the channel of the trend line at 55.1%, which is 1.08% different from the value obtained from the static culture test. It was confirmed that the toxicity of BPA to statically cultured GC-1 cells was maintained even in cells in a microfluidic environment.

Fig. 7

The effect of BPA on GC-1 cells in 2D culture conditions. a Images showing the effects of BPA according to concentration. b A graph of the proliferation rate of CG-1 cells exposed to BPA. The proliferation rate is the ratio of the harvested GC-1 cells after culturing to the number of intact GC-1 cells after treatment with BPA for 48 h (P < 0.05)

Potential applications of the EDs study and challenges

Continuing exposure to EDs deleteriously affects thyroid function, corticosteroid production, and nervous system function in the human body [35,36,37]. EDs are present in synthetic chemicals such as solvents, lubricants, plastics, and pharmaceuticals. Especially, BPA, polybrominated biphenyls, polychlorinated biphenyls, dichlorodiphenylltrichloroethane, dioxin, polychlorinated biphenyls, phthalate esters, Endosulfan, Zeranol, and Atrazine are known to cause conditions such as cancer and infertility [36, 38]. Our experimental methodology accurately predicted and measured the effect of concentration of BPA on GC-1 cell efficiently in a single microfluidic chip. This is the first trial using our microfluidic chip system to explore the harmfulness of BPA on GC-1 cells. Thus, determining the harmful concentration not only of BPA but also of other EDs can be determined using our microfluidic chip system. Therefore, we expect that our approach can be utilized in the study of reproductive diseases caused by EDs.

Although functional improvement of the microfluidic chip was provided by our design, other improvements could enhance its practicability. For example, although using a reservoir port stabilized the flow of the culture medium during cell loading, flow instability was still present during the entire experimental process. Furthermore, in the channel, the interstitial flow depends on osmosis, which makes it difficult to consistently maintain the flow conditions during the entire experimental period.