Design of a microfluidic lung chip and its application in assessing the toxicity of formaldehyde

Abstract In this work, a microfluidic lung chip with membrane supporting cell growth that can produce multiple concentration gradients of gas and liquid is introduced. The chip is composed of a gas gradient layer in the upper part, a porous membrane supporting cell growth in the middle and a liquid gradient layer in the lower part. The gas-liquid interface environment of the cells on the membrane can expose the cells to the gas in the upper layer and the liquid in the lower layer at the same time. Then, the chip is applied to the toxicity testing of formaldehyde in A549 cells. The results showed that at 6 × 10−5 mol/L formaldehyde, the survival rate of the cells in four channels were 90, 87, 81, and 71%, which shows a dose-response trend under the influence of different concentrations of formaldehyde. ROS staining results also showed that formaldehyde exposure at 6 × 10−5 mol/L lead to the increase of ROS level in the cells. These results suggest that the chip based on cell growth on membrane could be used for toxicological evaluation of environmental polluting gases.


Introduction
A substantial proportion of the global burden of disease is directly or indirectly attributable to exposure to air pollution (Goldizen et al. 2016), epidemiological links between exposure to air pollution (both indoor and outdoor) and adverse respiratory outcomes are strong. At present, the two-dimensional monolayer cell culture model, which is widely used in the toxicity testing of environmental pollutants in vitro, cannot really simulate the microenvironment of the cells in vivo. Although the experimental cost is low and the cycle is short, in the evaluation of the impact of air pollution on the body may cause a certain deviation (Fricker et al. 2014;Benam et al. 2015Benam et al. , 2016Kapałczy nska et al. 2018).
Biomimetic systems built on microfluidic chips that could mimic the main functions of human organs, also known as organ chips, which can be micromachined to precisely manipulate cells and microenvironments. In order to simulate the complex structure, microenvironment and physiological function of human organs, chip design always includes chemical concentration gradient, cell attachment, mechanical stimulation. (Ingber 2020(Ingber , 2022. In previous studies, there have been a number of chip studies for assessing environmental pollutants, such as in 2012, when Shintu et al. (Shintu et al. 2012) combined microfluidic chips with 1 H Nuclear Magnetic Resonance Spectra ( 1 HNMR), metabolic markers of exposure and toxicity of several toxic molecules (ammonia, dimethyl sulfoxide, paracetamol) were studied in isolated and co-cultured liver and kidney cells. In 2018, Theobald et al. (Theobald et al. 2018) developed a lung and liver microarray model to study the biotransformation and toxicity of aflatoxins B1 and benzopyrene. In 2018, Liu et al. (Liu et al. 2018) designed and fabricated a microfluidic chip that could realize gas-liquid dual gradient exposure and employ the agarose for cell culture, and used the constructed chip platform to assess the oxidative damage of A549 cells exposed to cigarette smoke. Organ chips can be used for toxicological prediction and mechanism research of environmental pollutants. At present, there are some applications of organ chips in organic compound pollutants, particulate matter and metal ions. In contrast to the conventional cytotoxicity experiments in static culture, onchip systems can increase the sensitivity of cells to exposure by increasing shear stress, providing a more sensitive platform for the quantitative analysis of cytotoxicity, especially in the evaluation of the toxicity of gaseous pollutants. However, the development of gas-liquid chips for researching direct exposure to gaseous pollutants is still in progress (Iskandar et al. 2017;Hou et al. 2019;Zhang et al. 2020).
Formaldehyde is a common air pollutant that can cause severe damage to the respiratory system (Tang et al. 2009). It has been identified as a Class I carcinogen in the International Agency for Research on Cancer (IARC) classification (Miller et al. 2017;Nielsen et al. 2017). An IARC Class I carcinogen refers to the substance that has sufficient evidence to prove carcinogenic to humans. In addition, formaldehyde is an irritant and one of the 93 harmful and potential harmful constituents (HPHCs) found in tobacco products and tobacco smoke (FDA and Center for Tobacco Products 2012). Due to its hydrophilicity, formaldehyde can not only be absorbed by the upper respiratory tract, but may also be adsorbed on inhalable particles and deposited in the lungs with the particles. Therefore, the toxicological evaluation of formaldehyde is of great significance in its health risk assessment. In the present study, a lung chip based on membrane-supported cell growth is introduced, which can directly expose cells to gas at the gas-liquid interface. And formaldehyde is selected as an appropriate gaseous pollutant for our lung chip assessment. The chip is used for the preliminary toxicological evaluation of formaldehyde.

Reagent and materials
About 5 lm polycarbonate (PC) membranes were from Millipore, America; silicone tube was from Jiangsu Runze Co., Ltd.; polymethylmethacrylate (PMMA) substrate was brought from Dongguan Nanya Insulation Plastic Material Co., Ltd.; polydimethylsiloxane (PDMS) was from DOWSIL; double valve gas collection bag (3 L) was from Ningbo Hongpu Instrument Technology Co., Ltd.
A549 cells were brought from the cell resource center of Shanghai Academy of Biological Sciences, Chinese Academy of Sciences. Calcein acetoxymethyl ester (Calcein-AM), propidium iodide (PI) were purchased from Beyotime Biotechnology, Nanjing, China. RPMI-1640 cell culture medium and 0.05% Trypsin-EDTA were from Gibco, USA. 0.01 M PBS and rat tail tendon collagen type I were from Solarbio, Beijing, China. Fetal bovine serum, penicillin, and streptomycin were purchased from Gibco, USA. Bromothymol blue was from Shanghai Reagent Co., Ltd. Formaldehyde was provided by Sigma-Aldrich.

Equipments and settings
TH4-200 microscope was from Japan Olympus corporation; laser engraving machine (Universal Laser Systems VLS 2.30, America) was used for engraving the pattern of the underlying chip; micropipette was from Eppendorf, Germany; multichannel microfluidic pressure pump (OB1 MK3þ) used for transporting gas was from Elveflow; liquid injection pump (70-4501) used for transporting liquid was from Harvard; gas flowmeter for detecting gas velocity was from Beijing Huibolong Instrument Co., Ltd., China; cryogenic plasma processor DT-02 for bonding between layers of a chip was from Suzhou OMG Mechanical & Electrical Technology Co., Ltd., China. All the images were analyzed further by Image-J software.

Design for the chip
The gas-liquid dual gradient chip is illustrated in Figure 1, a three-layer sandwich lung microfluidic chip with gas-liquid dual dimensional-concentration gradient is constructed for the study of cell culture and exposure experiment. The chip design consists of three layers and a substrate at the bottom. The upper layer is a gas pipeline layer, the middle layer is a porous membrane, and the lower layer is a liquid channel layer. The gas-layer channel is perpendicular to the liquidlayer channel, but not identical, separated by a porous membrane in the middle. The chip simulates the structure of the alveoli, the upper gas layer simulates gas exposure at the gas-liquid interface, the middle membrane supports cell growth, and the lower liquid channel simulates the capillary and provides nutrients for cell growth. Four concentration gradients could be formed in the upper gas pipe and the lower liquid pipe respectively.
The upper chip is a gas pipeline layer, which is designed to simulate the alveolar cavity as shown in Figure 2(a). Two gas inlets and one gas outlet are arranged on the chip. The width of the gas layer is 800 lm. The two kinds of gas in channels can mix and dilute each other in the middle pipes, forming four gas concentrations with different proportions at the same time. The lower layer chip is a liquid channel layer shown in Figure 2(b), used to simulate the structure of the capillary. The liquid channel is 600 lm wide, 150 lm deep. In the lower chip, two liquids can be fully mixed and diluted with each other, forming different liquid gradients. In addition, the back end of the oval cell culture area, also help to reduce fluid flow on the cell fluid shear force. The middle layer is a porous membrane, which simulates the gas-liquid interface. In this work, PC porous membrane with 5 lm aperture was selected as the supporting material for A549 cells. The mechanical stability of PC membrane enables it to be used as a cell carrier, to provide a stable environment for cell growth.

Chip fabrication
According to the chip design drawing, the upper and lower layers are gas layer and liquid layer cavities respectively. In this experiment, PMMA with better hardness and transparency was selected as the bottom substrate, and both upper and lower layers were made of PDMS material which is widely used in literature due to its high transparency and good biocompatibility. The lower mold was made by photolithography (substrate cleaninggluingpredryingexposuredevelopment), the upper mold was made by laser engraving by Laser engraving machine (Universal Laser Systems VLS 2.30, America), and then the upper and lower layers were made by PDMS pouring-curing (70 C, 1 h)demouldingdrillingcleaning, and then the upper and lower layer and the middle layer film were bonded in cryogenic plasma processor DT-02.

Characterization analysis of gas gradient and liquid gradient
According to the discoloration of bromothymol blue solution from blue to yellow in the presence of carbon dioxide gas (CO 2 ), the gas of CO 2 and the alkaline solution of bromothymol blue were employed to indicate the formation of gas gradient. First, gel containing bromothymol blue solution was dropped into the 16 positions of the lower layer of the chip. Then, Gas A inlet (Figure 2(a)) was connected with an air bag to supply air, Gas B inlet (Figure 2(a)) was connected with a CO 2 bag to supply CO 2 , and a gas pump with constant flow rate was connected at the gas outlet to pump gas at a flow rate of 2 mL/ min. Because the upper four gas channels are independent of each other, CO 2 and air form different CO 2 gas gradients in the upper layers. The bromothymol blue solution in the gel changes color depending on the CO 2 concentration in the channel. The greater the CO 2 concentration, the more significant the discoloration reaction and the more obvious the yellow color. The formation of gas concentration gradient was judged by the chromatic value analyzed with ImageJ software.
Characterization of the formation of the liquid concentration gradient was simulated by liquid diffusion with bromothymol blue solution and water. The bromothymol blue solution was introduced into one inlet, and the water was introduced into the other inlet. Due to the different distribution, the color of the liquid in the four channels is different, and the formation of concentration gradient was judged by chromaticity. ImageJ software was used for quantitative analysis. In order to more accurately determine the formation of the liquid concentration gradient, the above experiments were carried out using 20 lg/mL sodium fluorescein solution instead of bromothymol blue solution, and the fluorescence intensity in each channel was observed by a fluorescence microscope and photographed for analysis.

Chip preprocessing
Before cell inoculation, the chip should be sterilized. The prepared chip was irradiated with ultraviolet light for 30 min first, and then, added 75% ethanol solution from the gas  outlet of the upper gas chamber and the liquid inlet of the lower liquid chamber respectively to fill the whole upper and lower channels. Ethanol was retained in the channels for 5 min, then sucked out, replaced with 75% ethanol solution, repeated three times. Then the chip was cleaned with sterilized water and PBS successively, and the operation was the same as above, and the duration was 30 s for each. After cleaning, the coating material was added into the channel to coat the middle layer porous membrane. In this experiment, we selected rat tail collagen as the coating material. 200 lL 0.2 mg/mL rat tail collagen/acetic acid solution was added into the upper channel (acetic acid concentration: 0.006 mol/ L) and incubated at 37 C and 5% CO 2 for 2 h, and then the residual solution in the upper chamber was washed with sterile water and PBS successively, repeated three times. Finally, PBS in the upper and lower chambers was replaced with complete medium to rinse the porous membrane.

Cell culture
Before loading to the chip, A549 cells were cultured in RPMI-1640 medium containing 10% fetal calf serum, penicillin (100 mg/mL) and streptomycin (100 mg/mL). Before loading, the cells were resuspended in the culture medium with cell density adjusted to 0.5 Â 10 6 cells/mL, and inject 200 lL cell suspension through the gas outlet of the upper gas channel, then it was cultured in the incubator at 37 C and 5% CO 2 .
The cell state on the chip was observed at 2, 24 and 48 h after cell seeding. After 48 h of culture, the cells on the chip were stained with Calcein-AM/PI to characterize the growth of cells on the chip.
2.8. Application in toxicological testing of formaldehyde 2.8.1. Formaldehyde exposure device The formaldehyde exposure device is shown in Figure 3, which is composed of a chip, a gas conveying unit, and a liquid delivery unit.

Construction of formaldehyde exposure model
Cells were loaded on the chip that has been sterilized and coated. After 48 h of cell growth, the exposure device was connected. Two syringe pumps with constant and adjustable flow rate were placed on the left side of the chip, and connected respectively to each of two liquid inlets of the liquid layer through the liquid pipe extended on the syringe. Then, each of two gas inlets of the chip gas layer was connected with a gas collection bag containing a certain volume of experimental gas. In this experiment, clean air-contained collection bag and formaldehyde gas-contained collection bag were respectively connected to each of two inlets on the upper layer of the chip. Gas pump was connected with the gas outlet of the chip through the gas pipe, and the gas was pumped at a constant flow rate. In the experiment, the gas flow rate was 2.5 mL/min and the liquid flow rate was 80 lL/h.

Detection of cell viability and ROS level
After 30 min exposure, the cells on the chip were washed with PBS once, the medium containing phenol red in the chip was removed, the mixed solution of Calcein-AM, PI and detection buffer solution was prepared (working concentration: Calcein-AM was 2 lM, PI was 8 lM), and then was slowly added to the chip and incubated in dark for 30 min. After incubation, the chip was washed with PBS for 3 times, and then the green fluorescence intensity from the cells was observed under the blue light excitation of the microscope, as well as the red fluorescence intensity under the green light excitation(Calcein AM: Ex/Em ¼ 494/517 nm, PI: Ex/Em ¼ 535/617 nm). The cell survival was analyzed according to the intensity of green fluorescence and red fluorescence. The living cells showed green fluorescence and the dead cells showed red fluorescence.
For ROS level detection, the cells on the chip washed twice with PBS, and 10 lM highly sensitive 2 0 ,7 0 -Dichlorodihydrofluorescein diacetate (DCFH-DA) working solution was added, and than the chip was incubated for 30 min. Removed the working solution, washed the chip twice with PBS, added PBS again, and observed it with fluorescence microscope (Ex/Em ¼ 505/525 nm).

Statistical analysis
The fluorescence intensity of fluorescent images was analyzed with ImageJ software.
The data were presented in the form of mean value ±standard deviation. The data were analyzed with one-way analysis of variance to compare the difference between different channels by using SPSS 25.0. Statistically significance was set at p < 0.05.

Theoretical simulation
The COMSOL Multiphysics software was used to simulate the gas-liquid concentration distribution of the upper and lower layers of the designed chip. Figure 4 represents the formation of the theoretical concentration gradient of the gas layer and the liquid layer respectively. Blue represents the concentration of 0 and red represents the maximum concentration. It can be seen from Figure 4(a) that the concentration ratio of the four channels is about 0:2.9:7.2:10, which can form a significant gas concentration difference. Figure 4(b) shows that the concentration ratio of the four channels of the liquid layer is about 0:3.2:6.7:10, which can form a significant concentration gradient difference. Theoretical simulations proved that the design of the lung chip was feasible in theory.

Characterization of the gas gradient and liquid gradient generation from the chip
As the alkaline bromothymol blue solution turns yellow green when encountering CO 2 , the color of bromothymol blue solution in the channel will change continuously with the accumulation of CO 2 . It can be seen from Figure 5(a) that the gradient is formed stably, the CO 2 concentration from A to D is from minimum to maximum, and the color is from blue to yellow green. The chromaticity analysis results are shown in Figure 5(b), and the fitting linear equation at 4 min is y ¼ 34.014x-36.634 (R 2 ¼0.9947). The gas concentration gradient in the four channels show a good linear relationship.
Liquid gradient characterization is shown in Figure 5(c,d). PBS and bromothymol blue solution are respectively introduced into the two liquid inlets of the lower channel at the flow rate of 200 lL/h, monitor the color change of the liquid in the channel, analyze the chromaticity of the liquid in each channel and indicate its gradient change. The fitting linear

Cell culture
Most researchers cultured respiratory system cells in vitro to study the biological effects of environmental pollutants. Human lung adenocarcinoma cell line A549 was widely used to study the respiratory system damage caused by inhalable pollutants such as automobile exhaust and PM 2.5 . In this study, A549 cells were selected as the target cells to construct the model. Figure 6(a) shows the cell state diagram of cells growing on the chip for 2, 24, and 48 h. It shows that the cells can achieve good adhesion. From the 24 and 48 h state diagrams, it can be seen that the cells proliferate obviously, and after 48 h, the cells are closely connected and form a dense barrier like layer on the membrane. The cell survival rate in the four channels is as Figure 6(b). It can be observed that the cells in the four channels survive well (detailed in SI S1), and the cell survival rate is more than 95%. Also, the cells in the four channels grow evenly. It is proved that the cells can achieve good and stable growth on the chip, and the cells in the cell area of each channel are uniform. After 30 min of air exposure under the above conditions (gas velocity: 2.5 mL/min, liquid velocity: 80 lL/h), the cell viability in the four channels on the chip is shown in Figure  6(c), as can be seen, most of the cells remain alive and the cell viability in the four channels is over 90%, it can meet the requirements of the next exposure experiments.

Effects of formaldehyde exposure on cell survival and ROS level generated
Under the above exposure conditions, the cells in the four channels of the membrane were exposed to formaldehyde gas with different concentrations formed by air diffusion. As shown in Figure 7(a,b), in the 2Â10 À3 mol/L formaldehyde exposure test, a large number of cells became round and stained with red fluorescence. Because of the high concentration of formaldehyde and the strong diffusion ability of formaldehyde, the cell injury degree of air group was also very high. When the concentration of formaldehyde was reduced to 6Â10 À5 mol/L, as shown in Figure 7(c), the cell survival rate showed obvious gradient. When the concentration was 6Â10 À5 mol/L, the cell survival rate was 90, 87, 81, and 71%, respectively, from A to D channel. The concentration of formaldehyde increased, and the cell viability decreased. It can be seen that the air channel A has the least cell damage and the formaldehyde channel D has the most serious cell damage, which is confirmed that formaldehyde diffuses with air to form a stable concentration gradient (detailed in SI S2).
After exposed to 6Â10 À5 mol/L formaldehyde for 30 min, ROS staining was performed. Figure 7(d) is a column graph showing the relative fluorescence intensity of ROS in each region. After exposed to formaldehyde for 30 min, ROS level in cells was significantly higher than that in the air channel group (Figure 7(d,e)). And under the condition of this concentration, the ROS level increased with the increase of formaldehyde concentration, which indicated that the acute stimulation of formaldehyde could induce the increase of ROS in cells. The changing trend of ROS content was consistent with the increasing trend of cytotoxicity induced by formaldehyde.

Discussion
The microfluidic chip system combines the bionic microenvironment structure with human cell culture to improve physiological emulation and avoid the use of animals, which has a future in new approach methodology (NAM) testing. The NAM data can be applied for prioritization of toxic chemicals and regulatory decisions (Graepel et al. 2019;Westmoreland et al. 2022), as well as Next-Generation Risk Assessment (Nitsche et al. 2022). This study aimed to utilize a NAM such as in vitro techniques to generate data on exposure to hazardous substances. The chip we designed provides a new method to evaluate the toxicological effect of gaseous pollutants in vitro.
Lung-on-a-chip models can simulate the lung's microenvironment and functions in vivo, and have great application value for respiratory disease research, drug screening, toxicity assessment and other aspects (Nawroth et al. 2020;Francis et al. 2022;Li et al. 2022;Xia et al. 2023). The physiological microenvironment of the lung is very complex (Martinez et al. 2011). In order to simulate the physiological microenvironment realistically, Sakolish et al. (Sakolish et al. 2022) designed a microfluidic device. It could realize the coculture of primary human small airway epithelial cells and lung microvascular endothelial cells, which recreates the parenchymal-vascular interface in the end of lung tissue. Varone et al. from Emulate Inc. (Varone et al. 2021) developed a novel organ-chip system that emulates three-dimensional architecture of the human epithelia, and the chip also has mechanical forces function including mechanical stretch and fluidic shear stress. Compared to the chip designed by Varone et al. our chip has no physical forces function. Indeed, the tissue-relevant mechanical forces acting on the chip is a critical element for biomimetic reconstruction of native tissue. Nevertheless, the advantage of the chip we designed is that multiple concentration gradients of gas and liquid can be achieved, as well as air-liquid interface exposure.
In the application of our designed chip, the toxicity of formaldehyde was accessed. There are many studies on the acute toxicity of formaldehyde. It has been reported that formaldehyde at 60-240 lM for 24 h could induce ROS accumulation and cytotoxicity in PC12 cells (Tang et al. 2011). Zhang et al.'s results (Zhang et al. 2013) showed that 100 lM formaldehyde induced genotoxicity through its ROS and lipid peroxidation enzyme activity in A549 cells. Similar results were found in animal experiments, for example, Gulec et al. (Gulec et al. 2006) found that 10 mg/kg formaldehyde treatment for 10 days significantly increased the activities of malondialdehyde (MDA), nitric oxide (NO), and decreased the activities of superoxide dismutase (SOD) and catalase (CAT) in the liver of rats. Our results from the chip are consistent with previous studies, formaldehyde exposure leads to cell damage and ROS increase.
However, there are some limitations in the present work, such as the following. The determination of the actual gas concentration gradient on the chip has not yet been achieved, which still needs to be solved. Only one type of cell is involved in the current chip, and co-culture of more than two types of cells will better reflect the actual physiological environment. Ideally, primary human alveolar epithelial cells should be employed. Calibration of the chip should also be considered to minimize the error of experimental results due to different batches of the chips. In terms of the chip application, we only performed multi-concentration gas exposure experiment, but haven't carried out multiple concentrations of liquid exposure experiment as well as multiconcentration gas-liquid exposure experiment. This is an important aspect of the chip application. For special application scenarios of the chip, some key parameters and validation experiments should be considered, such as gas flow rate, liquid flow rate, exposure duration, dose setting, and comparison with the results of traditional in vitro and in vivo studies. In the future, the aspects mentioned above will be the focus of our research work. All the same, the application of the chip in formaldehyde exposure experiment still has important significance.

Conclusion
In summary, a gas-liquid interface three-layer lung chip was designed and used for assessing the exposure to formaldehyde in the present work. The gas layer of the chip can form four concentration gradients through pattern design, and the liquid layer can also form four concentration gradients. The middle layer can support cell growth with porous PC membrane. We applied the chip to the evaluation of the toxicity of formaldehyde, which has the advantages of direct exposure and high flux compared with traditional formaldehyde toxicity assessment method. The results showed that the survival rate and ROS level of A549 cells in different channels were different, and the damage increased with the increase of formaldehyde concentration. The results indicate that the chip we designed enable the concentration gradient formation of formaldehyde in the gas layer, and this also suggests that the chip can be used as a new platform for toxicological evaluation of gas pollutants.