Simple-Flow: A 3D-Printed Multiwell Flow Plate to Coculture Primary Human Lung Cells at the Air–Liquid Interface

Abstract

Immortalized epithelial cell lines and animal models have been used in fundamental and preclinical research to study pulmonary diseases. However valuable, though, these models incompletely recapitulate the in vivo human lung, which leads to low predictive outcomes in potential respiratory treatments. Advanced technology and cell culture techniques stimulate the development of improved models that more closely mimic the physiology of the human lung. Nonetheless, most of these models are technically demanding and have a low throughput and reproducibility. Here, we describe a robust fluidic device consisting of a biocompatible and customizable 3D-printed cell culture plate, the Simple-Flow, which has medium throughput, is simple to manufacture, and is easy to set up. As a proof of principle, human primary bronchial epithelial cells (hPBECs) and human pulmonary microvascular endothelial cells (hMVECs) were cocultured on the apical and basolateral sides of the inset membranes, respectively. While hPBECs were cultured at the air–liquid interface to induce mucociliary differentiation, hMVECs were exposed to flow medium for up to 2 weeks. We show the versatility of 3D-printing technology in designing in vitro models for cell culturing applications, such as pediatric lung diseases or other pulmonary disorders.

Introduction

The human lung is an essential organ for life and is required for proper gas exchange between the environment and the internal milieu. It is composed of a branched tubular structure, called conducting airways, that warms and filters the inhaled air and transports it through even finer branched tubes, termed small airways, and into the smallest structures, the alveoli, where gas exchange takes place.^1,2 The airways and alveoli are lined by an epithelium consisting of various epithelial cell types, which varies in composition along the proximo-distal axis to fulfill the different pulmonary functions. These tubes are intermingled by a network of blood vessels that provide oxygen and nutrients, and these tubular structures are embedded in connective tissue made of a variety of mesenchymal cells, such as different fibroblasts and smooth muscle cells.^2,3

Respiratory-associated diseases, including chronic obstructive respiratory disease (COPD), lower airway infections, neonatal conditions, and lung cancer, are among the leading causes of death globally emphasizing the importance of lungs in sustaining life.⁴ In addition, congenital and neonatal pulmonary abnormalities, like congenital diaphragmatic hernia (CDH) and congenital pulmonary airway malformation (CPAM), cause significant structural abnormalities, which results in lifelong morbidities for survivors.^5,6

To better understand the origin and subsequent progression of respiratory diseases, it is imperative to have representative models. Several in vivo and in vitro models have been developed over the years, which all have made contributions to the field. Continuously, emphasis is placed on developing in vitro models that increasingly mimic the human in vivo situation, providing opportunities to study patient-specific situations and replace animal models. Despite their substantial impact on research, nearly all animal models lack reliable extrapolation to human disease and, thus, are insufficient to predict outcomes, identify preventive measures, or develop effective treatment modalities.⁷ Moreover, most therapeutic agents tested in animals failed to prove therapeutic efficacy in clinical randomized trials.⁸⁻¹⁰

Several alternatives to replacing animal models of pulmonary diseases are being explored. The most widely used and accepted standardized in vitro model is the air–liquid interface (ALI) culture method, which is a static, air-exposed cell culture system that promotes mucociliary differentiation of the human airway.^11,12 Originally, this model used epithelial cells only, which failed to recapitulate the various cell types found in the lung. Currently, more complex ALI models are being developed that integrate not only epithelial cells but also mesenchymal, endothelial, and immune cells to more closely mimic in vivo lung conditions.^11,13,14

Other complex human in vitro models, like organoids and organ-on-chips (OoC) technologies, aim to integrate three-dimensional structures, specialized culture substrates, cocultures of different cell types, and biomechanical dynamics, such as flow shear stress and stretch, to more closely mimic in vivo organ conditions.¹⁵⁻²⁰ These technologies are under development and are not yet mature enough to have a standard consensus on how to be assessed. Therefore, different OoC devices are being developed and explored, mainly using cell lines as a proof of principle. However, immortalized cell lines do not properly recapitulate the characteristics of the human lung, as cell lines often present genetic aberrations, such as chromosomal aberrations, while primary cells are known for their high biological relevance and preservation of their genetic integrity.²¹⁻²³ For instance, primary bronchial epithelial cells present higher mucociliary differentiation and stronger tight junctions than cell lines.²⁴⁻²⁶ Finally, many OoCs are very specialized devices requiring advanced technological support with low throughput and reproducibility.^17,18,27 Polydimethylsiloxane (PDMS) chips, for example, present challenges in providing medium- or high-throughput devices, as they typically support one condition per device due to size and 3D structure fabrication limitations. In addition, their fabrication is labor intensive, involving the modeling of 3D structures, which can lead to reproducibility issues.^28,29

Therefore, we aimed to design an easily accessible, reproducible, and with medium-throughput lung model based on standard ALI cultures with basolateral fluidics. Based on a standard 24-well plate, we designed and 3D-printed a biocompatible, customizable, and easy-to-set up fluidic cell culture plate, Simple-Flow, that uses standard cell culture inserts and has flow of cell culture medium. As a proof of principle, we seeded human primary bronchial epithelial cells (hPBECs) and human lung microvascular endothelial cells (hMVECs) on the apical and basolateral sides of insert membranes, respectively. The hPBECs were cultured for extended periods at the ALI to allow differentiation to a mucociliary epithelium, and the hMVECs were cultured under flow while maintaining their endothelial phenotype.

Materials and Methods

3D-Printing and Postprocessing of the Multiwell Flow Plate

The 3D-printed multiwell flow plate Simple-Flow was designed with Inventor Professional and Fusion 360 (AutoDesk, student license) and then converted to .stl format. The “Luer Fitting Generator” plug-in for Fusion 360 was downloaded from AutoDesk App Store as an open source and used for the design of the internal male Luer lock connectors with a hole diameter of 3.00 mm and a wall thickness of 0.50 mm. Dimensions of the Simple-Flow are found in Figure S1.

The Simple-Flow.stl file was uploaded to the PreForm (FormLabs) software and was printed at a resolution of 0.05 mm (printing time: ∼7.5 h; resin volume: ∼89 mL/print). Supports were used at a density of 0.7, and the touching point size was set at 0.35 mm to ensure that the male Luer lock connectors were printed properly. Only external supports were applied, and one support per Luer connector was enough for a successful printing. The Simple-Flow was 3D-printed using a Form 3B+ printer (FormLabs), biocompatible BioMed Clear resin^30,31 (FormLabs), and a Tank V2.1 (FormLabs).

Once the plate was printed, the excess of uncured resin was removed with 100% isopropanol (IPA) using the Form Wash (FormLabs) according to the manufacturer’s instructions for BioMed Clear resin.³² This type of resin is transparent, hard, wear-resistant, and suitable for cell culture applications (see manufacturer’s Web site for technical details). Therefore, the device was washed twice for 15 and 5 min, respectively. Then, the supports were cut off with a scalpel (Aesculap, Cat No. 5518059) and the plate and Luer holes were air-dried. Finally, the plate was further cured with UV light using the Form Cure (FormLabs) at 60 °C for 60 min following the manufacturer’s instructions to ensure biocompatibility.³³ The Simple-Flow platform was stored in the dark until further use to prevent light-induced alterations of the materials.

Setting up and Maintenance of the 3D-Printed Simple-Flow

Under sterile cell culture hood conditions, the Simple-Flow was sterilized by soaking it for 15 s in 70% IPA and then air-dried. Meanwhile, Pharmed extension tubing (1.02 mm ID, Ismatec, Cat. No. SC0343) was connected to Pharmed three-stop tubing (1.02 mm ID, Ismatec, Cat. No. SC0309) using male (Ibidi, Cat. No. 10826) and female (Ibidi, Cat. No. 10825) Luer lock connectors. Then, female Luer lock connectors were attached to the tubing and screwed into the Simple-Flow. Tubing was connected to an IP65-rated peristaltic pump (Longer G100-1J with six rollers and eight channels, Darwin Microfluidics, Cat No. LG-G100-1J-EU/DG-8-A). The system was sterilized by pumping 70% IPA through the tubing for 5 min, followed by DPBS to remove excess IPA, and finally 8 mL of cell medium was added to each channel. The flow rate was set at a flow rate of 4 μL/min (Figure S2). Ready-to-put at ALI coculture inserts were placed into the Simple-Flow and a normal, sterile 24-well cell culture lid was used to cover the plate. Autoclaved high-vacuum grease was added into the groove between the 3D-printed plate and the female Luer lock connectors to prevent possible leakages. The whole setup was placed on a tray and put into the incubator (Figure 2F). Cocultures were left to differentiate at ALI for 14 days. Coculture medium was refreshed twice a week, and the cell layer integrity was checked under a bright-field microscope using the original Corning insert plates due to the poor transparency of BioMed Clear resin.

Figure 2.

Final design and setup of Simple-Flow. (A) Top view of the 3D-printed plate. (B) Side view of the 3D-printed plate showing the internal male Luer lock connectors. (C) Transverse plane of the 3D-printed plate. (D) Sagittal plane of the 3D-printed plate showing how the wells are connected in rows through channels. (E) Autoclaved high vacuum grease was added between the groove of the plate and Luer connectors to further prevent leakages. Dotted line shows where exactly the grease was added. (F) Final setup of the Simple-Flow: the printed plate is connected to a peristaltic pump, and the whole setup is placed inside an incubator for 2 weeks.

The relatively large dimension of the flow channel combined with the low flow rate (4 μL/min) prevents the development of turbulence in the system. The dimensionless Reynolds number (Re) determines that a fluidic flow is laminar if the Re is below 2300, and turbulent if it is above 2900. We calculated that the Re is 1.1 × 10^–2, which is well below the 2300 threshold, and thus it is most likely that the Simple-Flow has laminar flow.

For the Simple-Flow to have a turbulent flow velocity, the required flow rate for the flow to become turbulent is approximately 1.45 × 10¹⁵ μL/min, which is far from our 4 μL/min flow rate (for details of this calculation, see the Detailed Calculation of the Flow Dynamics of the Simple-Flow section in the Supporting Information).

Patient Samples

Tumor-free human lung tissue and bronchus rings were obtained from nontumorigenic lung resection material of patients diagnosed with lung cancer requiring surgery at the Erasmus University Medical Center (Rotterdam, The Netherlands) and approved by the Medical Ethical Committee (METC no. MEC-2012-512, MEC-2023-0112). All lung samples were processed according to the “Human Tissue and Medical Research Code of conduct for responsible use (2011, www.federa.org)” guidelines and used anonymously, and informed consent was obtained from patients. Donor characteristics can be found in Table 1.

Table 1. Summary of Donor Characteristics that Underwent Lung Resection Surgery.

patient	hPBECs	hMVECs	hECFCs	sex	age
donor 1 (D1)	yes	yes	no	female	66
donor 2 (D2)	yes	yes	no	female	52
donor 3 (D3)	yes	no	no	female	63
donor 4 (D4)	yes	no	no	male	73
donor 5 (D5)	no	yes	no	male	72
donor 6 (D6)	yes	no	no	female	56
donor 7 (D7)	yes	no	no	N/A	N/A
donor 8 (D8)	no	no	yes	N/A	N/A

Primary Cell Culture

Human lung primary cells were isolated and cultured, as previously described.^34,35 Briefly, all lung cell types were isolated from tumor-free human lung tissue. HPBECs were expanded in precoated 10 cm dishes using complete KSFM medium (see Table 2 for all culture media and components). To differentiate, hPBECs were then plated on precoated inserts and put at ALI once confluent using complete BEGM with 50 nM EC-23 (Tocris, Cat No. 4011, 10 mg). HPBECs were trypsinized using 1× soft trypsin for 12–15 min and quenched with a double amount of soy bean trypsin inhibitor (SBTI). HPBECs were used at passages 2–3 at the ALI. HMVECs were expanded in precoated 10 cm dishes using complete EGM-2MV medium and were trypsinized with Trypsin-EDTA (TE solution) for 3–5 min. HMVECs were used in cocultures at passages 5 to 7. Human endothelial colony forming cells (hECFCs) were isolated from healthy human lung tissue following the protocol of Alphonse et al.³⁶ and cultured on collagen type-I based coated surfaces in EGM-2 BulletKit medium (Lonza) with 10% FBS. Table 2 provides details on the composition of cell culture media, surface coatings, and types of trypsine used for hPBECs and endothelial cells (ECs).

Table 2. Overview of Cell Culture Reagents Used for Each Primary Lung Cell Type.

name	component	final concentration	supplier (cat. no.)
Human primary bronchial epithelial cells
hPBECs and coculture coating	PureCol, 3 mg/mL	30 μg/mL	Advanced BioMatrix (5005)
	human fibronectin	10 μg/mL	EDM Millipore (FC010)
	BSA	10 μg/mL	Sigma-Aldrich (A7030—10 g)
	DPBS	1×	Sigma-Aldrich (D8537—500 mL)
complete KSFM medium (proliferation)	KSFM basal medium with l-glutamine	1×	Thermo Fisher Scientific (17005)
	BPE	25 μg/mL	Thermo Fisher Scientific (13028-014)
	EGF	0.2 ng/mL	Thermo Fisher Scientific (10450-013)
	IP	1 μM	Sigma-Aldrich (I-6504—100 mg)
	penicillin/streptomycin	P: 100 U/mL	Sigma-Aldrich (P0781)
		S: 100 μg/mL
complete BEGM medium (differentiation)	BEpiCM medium	0.5×	ScienCell (3211)
	DMEM medium	0.5×	StemCell (36250)
	2× BEpiCGS supplement	1×	StemCell (3262)
	HEPES buffer	12.5 mM	Gibco, Thermo Fisher Scientific (15630-056)
10× soft trypsin*	Difco Trypsin 1:250	0.3%	BD Biosciences (215240—100 g)
	EDTA	0.1%	Sigma-Aldrich (E1644–1KG)
	d-(+)-glucose	1%	Sigma-Aldrich (G6152–100G)
	DPBS	1×	Sigma-Aldrich (D8537–500 ML)
soy bean trypsin inhibitor (SBTI)	SBTI	1.1 mg/mL	Sigma-Aldrich (T9128—1 g)
	KSFM basal medium with l-glutamine	1×	Thermo Fisher Scientific (17005)
	penicillin/streptomycin	P: 100 U/mL	Sigma-Aldrich (P0781)
		S: 100 μg/mL
Human microvascular endothelial cells
hMVEC coating for 10 cm dishes	collagen R solution (0.4%)	50 μg/mL	Serva (47256.01)
	glacial acetic acid in sterile water	0.02 N	Fisher Scientific (17.4 N, 505216)
Complete EGM-2MV medium	EGM-2MV medium	1× all supplements	Lonza (CC-3202)
	penicillin/streptomycin	P: 100 U/mL	Sigma-Aldrich (P0781)
		S: 100 μg/mL
Trypsin-EDTA (TE) solution	TE solution	1×	Sigma-Aldrich (T3924)
Human endothelial colony forming cells
hECFC coating	collagen R solution (0.4%)	50 μg/mL	Serva (47256.01)
	glacial acetic acid in sterile water	0.02 N	Fisher Scientific (17.4 N, 505216)
Complete EGM-2 medium	EGM-2 medium	1× all supplements but EGF + 10% FBS	Lonza (CC-3162)
	penicillin/streptomycin	P: 100 U/mL	Sigma-Aldrich (P0781)
		S: 100 μg/mL
Trypsin-EDTA (TE) solution	TE solution	1×	Sigma-Aldrich (T3924)

Coculture Setup of the Simple-Flow Platform

Cell culture inserts (Costar, Corning, Cat. No. 3470), with 0.4 μm pore size and polyethylene terephthalate (PET) membranes, were coated on the basolateral side of the membrane at 37 °C overnight (ON). Next day, the coating was removed and allowed to dry in a sterile cell culture hood for 2–3 h. HMVECs were seeded at a cell density of 2 × 10⁵ cells/cm² on the basolateral side of the membrane and allowed to attach for 3.5–4 h. Inserts were then flipped back, and 500 μL of complete EGM-2MV medium was added on the basolateral compartment. After 4 days, the apical side of the insert membrane was coated at 37 °C for 2 h and washed once with DPBS. Subsequently, hPBECs were seeded at a cell density of 1.5 × 10⁵ cells/cm² on the apical compartment in 150 μL complete BEGM medium (Table 2) with 1 nM EC-23 (Tocris, Cat No. 4011, 10 mg) and 5 μM ROCK inhibitor (Y-27632, AbMole, Cat No. Y-27632). On the same day, coculture medium BEGM:EGM-2MV (2:1 ratio) with 1 nM EC-23 was added on the basolateral compartment. After the hPBEC cell layer was confluent, cocultures were put at ALI, and BEGM:EGM-2MV (2:1 ratio) with 50 nM EC-23 coculture medium was used. Flow inserts were transferred to the Simple-Flow and cultured at ALI for 14 days. A more detailed description of the hPBEC–hMVEC cocultures can be found in Table 3.

Table 3. Overview of hPBEC–hMVEC Cocultures Used.

number of coculture	cell type	donor	passage	static plate
1	hPBECs	1	3	cell culture
	hMVECs	1	6
2	hPBECs	2	2	3D-printed
	hMVECs	2	5
3	hPBECs	2	3	3D-printed
	hMVECs	2	6
4	hPBECs	2	3	3D-printed
	hMVECs	2	7
5	hPBECs	3	3	cell culture
	hMVECs	1	6
6	hPBECs	4	2	3D-printed
	hMVECs	5	5
7	hPBECs	4	3	3D-printed
	hMVECs	5	7

Lentiviral Production and Transduction of Primary Cells

HEK293T, cultured in DMEM + 10% FBS + 1% P/S + 1% sodium pyruvate (NaPyr, Gibco) + 1% nonessential amino acids (NEAA, Gibco), were transfected with VSV-G envelope plasmid, packaging plasmid Pax2, and transfer plasmid with an expression cassette for a fluorescent protein (EGFP, RFP, or BFP) in a 1:3:4 ratio in 2× HBS buffer (1.6 g of NaCl, 0.74 g of KCl, 0.27 g of Na₂HPO₄, 10 g of HEPES buffer in 1 L of H₂O) containing 100 μg of PEI and incubated with transfection medium (DMEM + 1% FBS + 1% P/S + 1% NaPyr +1% NEAA) for 4 h. Subsequently, the produced lentiviruses were purified from the medium by ultracentrifugation (20,000g for 2 h and 15 min at 4 °C), after which the viral particles were dissolved in 100 μL of DPBS. This was used to transduce semiconfluent primary cells in a 10 cm dish.

Micronit, MIVO, and Simple-Flow Prototype Systems

OoC Micronit PET membranes with a 3 μm pore size and a density of 1.6 × 10⁶ (02406, Micronit) or 8 × 10⁵ (01206, Micronit) pores per cm² were coated and used to coculture hPBECs and hECFCs on the apical and basolateral sides of the Micronit membranes, respectively, until confluency. Membranes were sandwiched in between OoC top and bottom glass layers of the OoC (00739, Micronit), and then the OoC was placed in a Fluidic Connect Pro Chip OoC Holder (00750, Micronit) and connected to the IP65-rated peristaltic pump (Longer G100-1J, Darwin Microfluidics, Cat. No. LG-G100-1J-EU/DG-8-A) for fluidic experiments.

Standard 6.5 mm Transwell inserts (Costar, Corning, Cat. No 3470) were coated, and hPBECs and hECFCs were plated on opposite sides of the membrane and grown until confluency. Subsequently, flow was created using the IP65-rated peristaltic pump connected with tubing and inserts placed at ALI.

Immunofluorescence (IF) of Cocultures

Insert membranes were processed as previously reported.³⁴ Briefly, the insert membranes were washed once with DPBS plus Mg²⁺ and Ca²⁺ (Gibco, Cat. No. 14287-072) and fixed with 4% (w/v) paraformaldehyde (PFA; Sigma-Aldrich, Cat. No. 441244) at room temperature (RT) for 10 min. Inserts were washed three times with DPBS (Sigma-Aldrich, Cat. No. D8537-500 ML) to remove excess PFA. Then, membranes were removed from the insert housing with a scalpel blade (Swann-Morton, Cat. No. 0103) and incubated three times for 5 min each with 0.1% (v/v) Triton X-100 (Sigma-Aldrich, Cat. No. T8787-25) in DPBS (0.1% T-PBS). Membranes were blocked with 1% (w/v) bovine serum albumin (BSA; Roche Diagnostics, Cat. No. 10735086001) and 5% (v/v) normal donkey serum (EMD Millipore, Cat. No. S30-100 ML) in 0.1% T-PBS at RT for 30 min and subsequently incubated with primary antibodies (Table 4) in blocking buffer at 4 °C ON. Then, membranes were rinsed three times and subsequently washed three times for 10 min with 0.03% (v/v) T-PBS, followed by an incubation with an appropriate secondary antibody (Table 4) and 1:2000 DAPI solution (BD Biosciences, Cat. No. 564907) in blocking buffer at RT in the dark for 2h. Again, membranes were rinsed three times and subsequently washed three times for 10 min with 0.03% T-PBS. Finally, membranes were mounted on slides (Menzel-Gläser SuperFrost Plus) with mounting media (2.4% (w/v) Mowiol (Sigma-Aldrich, Cat. No. 81381), 12% (w/v) 0.2 M Tris pH 8.5 (Sigma-Aldrich, Cat. No. T6066), and 4.75% (v/v) glycerol (Honeywell, Cat. No. 49770-1L) in ddH₂O) and covered with coverslips (Menzel-Gläser 24 mm × 60 mm). Samples were left to dry in the dark at RT ON and then stored at 4 °C (or −20 °C for long-term storage). Samples were imaged with an ECHO Revolve fluorescence microscope and a Leica SP-5 confocal microscope. Images were processed with ImageJ.

Table 4. Overview of Primary and Secondary Antibodies.

	dilution	supplier	cat. no.
Primary antibody
TUBIV, mouse monoclonal	1:200	BioGenex	MU178-UC
ZO-1, rabbit polyclonal	1:100	Zymed, Invitrogen	40-2300
MUC5AC, mouse monoclonal	1:500	Abcam	ab3649
KRT5, rabbit polyclonal	1:500	ITK diagnostics BV, BioLegend	905501
CD31 (PECAM-1), mouse monoclonal	1:100	BioLegend	303101
ERG-1, rabbit monoclonal	1:500	Abcam	ab92513
Secondary antibody
Alexa Fluor 488 Donkey anti Mouse IgG	1:500	Jackson ImmunoResearch	715-545-151
Alexa Fluor 594 Donkey anti Rabbit IgG	1:500	Jackson ImmunoResearch	711-585-152

Trans-Epithelial Electrical Resistance (TEER)

200 μL DPBS with Mg²⁺ and Ca²⁺ was added at the apical compartment of the culture inserts. After short acclimatization in the incubator, TEER values were measured with an STX2 electrode (World Precision Instruments, WPI) connected to an EVOM2 Epithelial Voltohmmeter (WPI). The resistance measurements were corrected to Ohm × cm² by multiplying the surface area of the insert membrane (0.33 cm²) with the raw TEER values measured. Finally, the background value was subtracted from the TEER values measured. TEER measurements plots were created with GraphPad Prism 9.

Measurement of hPBEC Thickness

XZ images were generated with ImageJ from SP-5 confocal images. A line was then drawn from top to bottom of the layer, and length (in pixels) was obtained by using the Measure function (Ctrl + M). Then, pixels were converted to micrometers by dividing by 4.1580, as the SP-5 image scale is 4.1580 pixels/micron.

Results

Design, Optimization, and Setup of the Simple-Flow for hPBEC–hMVEC Cocultures

In order to investigate the effect of fluidics on primary epithelial and ECs in cocultures, we first tested two commercially available flow systems (Figure 1 and Table 5). HPBECs and hECFCs were transduced with lentiviruses containing an expression cassette with a fluorescent protein (EGFP, RFP, or BFP) to monitor the cells in real time during the culture (2 weeks). HPBECs were plated on the apical side and hECFCs on the basolateral side of the semitransparent PET membrane of the organ-on-chip device from Micronit. Initially, hPBECs (green) and hECFCs (magenta) attached under static conditions, but after applying flow, the ECs eventually detached within a few days (Figure 1). In addition, it was challenging to put hPBECs at ALI in the Micronit device because it is a closed system and media kept leaking through the semipermeable membrane to the air-exposed chamber despite the confluent cell layers.

Figure 1.

Fluidic devices used for hPBEC–EC cocultures before the Simple-Flow. Representative images of hPPBECs–BFP or hPBECs–RFP cocultured with hECFCs–GFP (for details, see Table 1). Merged images (hPBECs green and hECFCs magenta) of the Micronit membranes after 11 days in static and flow conditions and of the insert membranes in the MIVO (React4Life) device and in the Simple-Flow prototype after 5 days in static and flow conditions. Scale bar = 1000 μm.

Table 5. Comparison of the Characteristics of the Different Fluidic Devices^a.

fluidic system	setup	handling	visibility	throughput	use of inserts
Micronit	>2 h	>1–2 h	crystal clear	low (1 condition/device)	no
MIVO	<1 h	<30 min	crystal clear	low–medium (3 conditions/plate, new model up to 8)	yes
Simple-Flow	<1 h	<30 min	cell shape can be appreciated but image blurred	medium (up to 16 conditions/plate)	yes

^aHandling refers to the ease of setting up the cultures and the time it takes to refresh cell culture medium. Visibility refers to the clarity of visualizing cells during the culture using a bright-field microscope.

Since we were already familiar with static ALI cultures with commercially available cell culture inserts, we used the MIVO technology (React4Life) device in which these can be placed under flow conditions. As with the Micronit device, cells remained attached to the coated membrane under static conditions, but the integrity of the hECFC layer was soon lost after flow was initiated and cells started to rapidly detach after a couple days (Figure 1). Aside from these challenges, both systems have quite low throughput, allowing only one culture condition per device. To overcome the limitations of usability and throughput, we adapted a commercial 24-well insert cell culture plate by connecting adjacent wells, and added standard inserts with cells (Figure S3A,B).

Subsequently, flow was applied using a peristaltic pump, and the primary hECFCs remained attached to the membrane even under flow conditions (Figure 1). Based on this successful pilot experiment with the prototypes, we designed a 16-well device based on the standard 24-well plate format and 3D-printed it with a standard stereolithography (SLA) printer.

The first design of the 3D-printed Simple-Flow platform (Figure S3C) had three adjacent wells connected per row, a 3D-printed lid to cover the plate (Figure S3D), and female Luer flip connectors (Figure S3C,H). Each row had two medium reservoirs flanking the wells to guarantee a sufficient supply of medium during the culture. This 3D-printed first design was reviewed, and based on observations, some improvements were included. The plate was adjusted to fit the lid of a standard 24-well plate (Figure S3E,F). In addition, four wells per row were connected instead of three to increase the number of parallel cultures (Figure S3G), and the female Luer slip was replaced with a male Luer lock connector to prevent leakage (Figure S3I). The new connections significantly decreased the level of leaking of the system, although occasionally some leakage still occurred. Therefore, autoclaved high vacuum grease was applied between the connectors and the plate to completely seal the connections (Figure 2E). The final prototype of the Simple-Flow is represented in Figure 2A–D, while a version without flow connectors was printed to serve as a control for static conditions (Figure S3J). We applied a flow of approximately 4 μL/min, and combined with the relatively large dimension of the flow channel, this results in a Reynolds number of 0.11 (calculation: see Materials and Methods). This indicates that there is a laminar flow in the chambers of the Simple-Flow device.

HPBECs Differentiate to Mucociliary Cells in the Simple-Flow Platform

Cocultures of hPBECs and hMVECs were set up (Figure 3A), and the integrity of the epithelial layer was evaluated by measuring TEER at three different time points during the culture. All static and flow cocultures showed similar trends of TEER values peaking on day 7 with the exception of flow coculture 4, which peaked on day 14 (Figure 3B). In addition, TEER values were the lowest on day 0 (at ALI) and generally ranged between 150 and 2500 Ω·cm² (Figure 3B). Finally, in two out of three cocultures, flow conditions showed a lower TEER values trend compared to static conditions.

Figure 3.

HPBECs differentiate in coculture with hMVECs in the Simple-Flow. (A) Scheme showing the setup of the coculture and fluidic system. (B) TEER values (Ω·cm²) of three independent cocultures in static and flow conditions on days 0, 7, and 14 of ALI. (C) Representative confocal microscope images of immunofluorescence-stained inserts under static or flow conditions showing differentiated ciliated (TUBIV) cells and hPBEC tight junctions (ZO-1). Coculture 2 was used. Scale bar = 50 μm. (D) HPBEC height for four independent hPBEC–hMVEC cocultures (1, 2, 3, and 5). (E) Representative fluorescent microscope images showing goblet (MUC5AC) cells in static and flow conditions. Coculture 2 was used. Scale bar = 200 μm. (F) Representative images showing basal (KRT5) cells in static and flow conditions. Coculture 3 was used. Scale bar = 200 μm. Scheme was created using BioRender.com.

After 14 days of coculture at ALI, the cultures still maintained ZO-1 tight junctions under static or flow conditions, and in two out of four cocultures, the hPBECs formed thicker epithelial layers in flow conditions compared to static (Figure 3C,D and Figure S4A). HPBECs differentiated into a mucociliary epithelium as exemplified by positive IF staining for basal (KRT5), ciliated (TUBIV), and secretory cells (MUC5AC) (Figure 3C,E,F and Figures S4 and S5). Interestingly, hPBECs showed less MUC5AC+ goblet cells under flow conditions for all cocultures, but no differences were observed in KRT5 expression between conditions.

Although donor-to-donor variations are clearly present, our results show that hPBECs in coculture with hMVECs maintain expression of basal cell markers, differentiate toward ciliated and goblet cells, and form tight junctions in the Simple-Flow at 14 days of ALI.

HMVECs Are Maintained in the Simple-Flow Platform after 14 Days of Coculture

Next, we analyzed the integrity of the EC layer in the different cultures using the nuclear marker ERG-1 and cell membrane marker CD31 (a.k.a. PECAM-1). HMVECs expressed both markers after 14 days of coculture with hPBECs in both static and flow conditions (Figure 4 and Figure S6).

Figure 4.

HMVECs are maintained in coculture with hPBECs in the Simple-Flow. (A) Representative images showing one-half of a 6.5 mm insert membrane stained with ERG-1 and CD31 in static and flow conditions after 14 days at ALI. Scale bar = 1000 μm. (B) Representative images showing that hMVECs in coculture with hPBECs express endothelial cell markers ERG-1 and CD31 in static and flow conditions after 14 days at ALI. Scale bar = 200 μm.

ECs remained attached to PET insert membranes for the total length of the cultures, as shown with ERG-1 staining. However, the hMVECs of two donors did not maintain a CD31-positive signal, suggesting that the cell–cell contacts between adjacent ECs were partially lost (Figure S6B). In addition, some patches without ECs (CD31- and ERG-1-) were also spotted (Figure S6B). These observations were donor dependent and occurred more frequently during flow conditions, probably due to different attachment properties of hMVECs to PET membranes compared to normal cell culture plastics and the shear stress induced by the flow medium.

In conclusion, we developed an easy-to-use and medium-throughput fluidic device that maintained primary cultures of human lung-derived cells. Our results show that hMVECs remain in coculture with hPBECs for long culturing periods (14 days) in static and flow conditions and maintain their endothelial phenotype.

Discussion

In this study, we describe the development of an easy-to-use medium-throughput culture device to culture pulmonary cells at ALI with medium flow in the basal compartment. As a proof of principle, cocultures of human primary PBECs and MVECs were maintained for 2 weeks in which the hPBECs formed a mucociliary epithelium and the hMVECs an endothelial monolayer.

Dynamic culture systems, such as the OoC, are a rapidly expanding field, as these types of cultures more closely represent the physiological properties of the organ of interest. One of the first pulmonary OoC devices was reported in 2010 by Huh and colleagues, and since then, others have followed in generating PDMS-based OoC with soft lithography technology.³⁷⁻⁴³ Although PDMS chips have proved to be valuable tools in this emerging field, they still present a series of disadvantages as fabrication and molding of complex 3D structures in soft lithography is a limited and tedious process and is hard to automate, and PDMS absorbs small hydrophobic molecules (e.g., drugs), which can bias results in drug discovery and development studies.^28,29,44,45

Commercially available OoC platforms like Micronit and React4Life’s MIVO technology have emerged as an alternative to PDMS chips. We were interested in generating an in vitro airway model of the human lung by coculturing hPBECs and hMVECs under ALI and flow conditions, respectively. We therefore decided to acquire Micronit and MIVO technology platforms to test our fluidic airway cocultures. Although both systems have advantages, it was challenging to set up the primary cocultures in the Micronit platform at ALI and to scale up both platforms to perform several flow cultures in parallel.

We wanted to have an OoC platform that was easy to set up and use, could include multiple conditions into one device, and was affordable. To overcome these limitations, we decided to design and print the Simple-Flow fluidic platform, which could serve as an intermediate between traditional OoC devices and standard ALI cultures. In addition, the Simple-Flow platform uses 3D-printing technology for its fabrication process, which holds promise in the future of OoC devices as it allows to design more complex 3D designs than soft lithography.^28,29 The Simple-Flow platform is based on a standard 24-well plate, is simple to print and set up, and allows up to 16 individual cultures using commercially available cell culture inserts. In addition, the design of the Simple-Flow could be easily modified according to our needs, e.g., increase the number of wells, making it a very versatile and easy-scalable device. The use of 3D-printed chips for cell culturing applications has been previously reported.⁴⁶⁻⁴⁸ However, few have used this technology for the fabrication of customized plates for biomedical applications.⁴⁹⁻⁵⁸

Human primary PBECs and MVECs were seeded on opposite sides of insert membranes and cultured in the Simple-Flow for 14 days. The hPBECs were air-exposed at ALI and differentiated toward a mucociliary epithelium, and the hMVECs tolerated fluidic conditions and formed a monolayer. Similar mucociliary differentiation of hPBECs was observed by others in OoC devices.⁵⁹⁻⁶¹ In addition, TEER values to quantify barrier integrity were measurable in our cocultures and detectable within Simple-Flow. This model was reported to develop a TEER range between 150 and 2500 Ω·cm², and monocultures of hPBECs have displayed various TEER values ranging from 100 to 950 Ω·cm².^25,62−64 Interestingly, the peak TEER value phenomenon at around 1 week of ALI was also reported by others.^64,65 However, TEER showed considerable variability depending on the donor, medium, passage, and angle and position of the electrode.⁶⁶ Nonetheless, the variation in TEER, it is a good method to evaluate whether the monolayer has formed tight junctions, which is required for the epithelium to differentiate once it is exposed to air at the apical side.

Complex in vitro models, such as the OoC technology, are not developed enough to have a standard consensus on how to be assessed, and there is still room for improvement in optimizing the biological and technological conditions. For instance, we experienced attachment problems with human primary lung ECs to PET membranes when exposed to medium flow conditions. Many factors may influence EC attachment to cell culture membranes, such as material and stiffness of the membrane, pretreatment of the membrane (plasma treatment, type of coating, coating concentration, and duration), pore size and density, and shear stress.⁶⁷⁻⁷² In addition, the nonoptimal coculture medium used and long culturing periods could have also influenced EC attachment.⁷³

Apart from cell attachment problems, the ease of use associated with immortalized cell lines still makes them frequently used for short culturing periods in OoC platforms, serving as a proof of principle.⁷⁴⁻⁷⁷ However, it needs to be emphasized that in order to better recapitulate the physiological conditions of the in vivo human lung in vitro, the use of primary cells should be preferred over cell lines and longer culturing periods are needed to allow bronchial cells to differentiate into the mucociliary epithelium. Despite the challenging circumstances, we and others developed airway-on-chip cultures that allowed for long culturing periods to induce mucociliary differentiation.^59−61,78

In addition of using primary cells, it is also important to highlight the need of using pulmonary cells for lung in vitro models, as data show that many cell types maintain tissue-specific characteristics, such as expression of adhesion molecules, EC barrier function, and trans-endothelial migration of immune cells.^79,80 Even though primary cells from other parts of the body, such as HUVECs,^74,81−86 have been extensively used to develop their lung-on-chip models, we and others selected lung cells hoping to more closely mimic lung conditions.^{42,59−61,78}

In the technological aspect, we also faced some leakage between the Simple-Flow platform and the tubing connections. That is why we redesigned the female Luer slips to a more leakproof torque connector like male Luer locks. Others have also used Luer lock connectors for 3D-printed devices,^28,87,88 and experienced leakage as well.^89,90 Even though the situation improved with the new connectors, sometimes the leakages did not completely vanish. Therefore, we used high vacuum grease as well to completely seal the connection, similarly to what Chan and colleagues did.⁹⁰ An alternative approach to prevent leakage could be using leak-tight connectors like ferrule lock rings⁹¹ or O-rings⁹² instead. In addition, printing material, resolution, orientation and design, postcuring deformations, and thermal expansion caused by incubator conditions could also contribute to the leakages.³¹

In summary, here we showed that the 3D-printed Simple-Flow can be used as a platform for the coculturing of primary lung epithelial and ECs that can be put at ALI and flow conditions, respectively, for long culturing periods. After 14 days of coculture, hPBECs differentiated toward ciliated and goblet cells and maintained their cell layer integrity and identity, and hMVECs still expressed EC markers. Finally, the ease of design and fabrication makes the Simple-Flow platform a very versatile and easy-to-use tool for culturing. Finally, our Simple-Flow device is in line with current progress in the field to develop low cost or alternatives to commercially available ALI platforms.⁹³⁻⁹⁸

Conclusions

In this work, we developed and optimized a customizable 3D-printed cell culture plate, Simple-Flow, that allows the use of commercial cell culture inserts and medium flow in the basolateral compartment. Compared to other OoC devices, the Simple-Flow has a higher throughput and is easy to manufacture and set up without requiring high-end technology. As a proof of principle, human primary lung-derived PBECs and MVECs were cocultured on either side of cell culture insert membranes for up to 14 days and were exposed at ALI and medium flow, respectively. HPBECs fully differentiated to a mucociliary epithelium, and hMVECs still preserved their endothelial phenotype.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsbiomaterials.4c01322.

• (Figure S1) Drawing showing the dimensions in millimeters of the Simple-Flow, (Figure S2) empirically calculated flow rate of Simple-Flow using a Longer G100-1J peristaltic pump with 1.02 mm ID tubing, (Figure S3) evolution of the Simple-Flow design, (Figure S4) additional primary cocultures showing hPBECs differentiating into ciliated cells in the Simple-Flow after 14 days at ALI, (Figure S5) additional primary cocultures showing hPBECs differentiating into goblet cells in the Simple-Flow after 14 days at ALI, and (Figure S6) additional pictures showing hMVECs in the Simple-Flow after 14 days of static and flow coculture (PDF)

All authors declare that they have no conflicts of interest

The authors declare no competing financial interest.