Microfluidic Lung Airway Coculture System with Arrayable Suspended Gels for Studying Epithelial and

Smooth Muscle Cell Interactions

Mouhita Humayun1†, Chung-Wai Chow2.3, Edmond W.K. Young1,4*

1 Department of Mechanical & Industrial Engineering, University of Toronto, Toronto, ON, Canada

† Current location: Department of Biomedical Engineering, University of Wisconsin-Madison, WI, USA

2 Department of Medicine, Faculty of Medicine, Toronto General Hospital Research Institute, University

of Toronto, Toronto, ON, Canada

3 Department of Chemical Engineering, University of Toronto, Toronto, ON, Canada

4 Institute of Biomaterials & Biomedical Engineering, University of Toronto, Toronto, ON, Canada

Keywords: Organ-on-a-chip, respiratory system, suspended microfluidics, air-liquid interface, airway

epithelium, bronchial smooth muscle, goblet cells

* Corresponding author:

Prof. Edmond W.K. Young

Department of Mechanical & Industrial Engineering

Institute of Biomaterials & Biomedical Engineering

University of Toronto, Toronto, ON, Canada

E-mail: eyoung@mie.utoronto.ca

Tel.: +1 (416) 978-1521

Fax: +1 (416) 978-7753

2 Abstract

Chronic lung diseases (CLDs) are regulated by complex interactions between airway epithelial cells

(ECs) and airway smooth muscle cells (SMCs), but underlying molecular mechanisms are not well

understood. To advance our understanding of lung pathophysiology and accelerate drug development

processes, new innovative in vitro tissue models are needed that can reconstitute the complex in vivo

microenvironment of human lung tissues. Organ-on-a-chip technologies have recently made significant

strides in recapitulating physiological properties of in vivo lung tissue microenvironments. However,

novel advancements are still needed to enable the study of airway SMC-EC communication with matrix

interactions, and to provide higher throughput capabilities and manufacturability. We have developed a

thermoplastic-based microfluidic lung airway-on-a-chip model that mimics the lung airway tissue

microenvironment, and in particular, the interactions between SMCs, ECs, and supporting extracellular

matrix (ECM). The microdevice is fabricated from acrylic using micromilling and solvent bonding

techniques, and consists of three vertically stacked microfluidic compartments with a bottom media

reservoir for SMC culture, a middle thin hydrogel layer, and an upper microchamber for achieving air-

liquid interface (ALI) culture of the epithelium. A unique aspect of the design lies in the suspended

hydrogel with upper and lower interfaces for EC and SMC culture, respectively. A mixture of Type I

collagen and Matrigel was found to promote EC adhesion and monolayer formation, and SMC adhesion

and alignment. Optimal culturing protocols were established that enabled EC-SMC coculture for more

than 31 days. Epithelial monolayers displayed common morphological markers including ZO-1 tight

junctions and F-actin cell cortices, while SMCs exhibited enhanced cell alignment and expression of α-

SMA. The thermoplastic device construction facilitates mass manufacturing, allows EC-SMC coculture

systems to be arrayed for increased throughput, and can be disassembled to allow extraction of the

suspended gel for downstream analyses. This airway-on-a-chip device has potential to significantly

advance our understanding of SMC-EC-matrix interactions, and their roles in the development of CLDs.

3 Introduction

The human respiratory system is one of the major organ systems of the body, and consists of specific

organs and structures that enable gas exchange between inhaled air from the atmospheric environment

and blood within the circulatory system. The structures of the respiratory system are most often divided

into two regions: (i) the conducting zone, comprised of larger airway tubes that serve the function of

carrying inhaled (or exhaled) air to (or from) the gas exchange region of the lung; and (ii) the

respiratory zone or acinus region, which consists of respiratory bronchioles, alveolar ducts, and alveoli,

and is responsible for actual gas exchange and transport through thin membranous tissue walls that

separate inhaled air from the vasculature.1 Both regions of the lung are susceptible to disease

development, with asthma and chronic obstructive pulmonary disease (COPD) affecting the airways,

leading to airway hyperresponsiveness (AHR), inflammation, and airflow obstruction, and airway tissue

remodeling.2,3 The global burden of asthma and COPD is expected to rise in coming years due to an

aging population and growing exposure to indoor and outdoor air pollution and various allergens.4 Thus,

both fundamental advances in lung biology research and novel therapies for managing lung diseases are

urgently needed.

Development of chronic lung diseases (CLDs) such as asthma and COPD is ultimately regulated

by cell and tissue processes within the airways. Different cell types comprise the airway tissue

microenvironment, including airway epithelial cells (ECs), smooth muscle cells (SMCs), endothelial

cells, and resident immune cells. Of these types, ECs and SMCs have been studied extensively, and have

been shown to play major roles in the pathogenesis of both asthma and COPD.2,3 Importantly, the

communication between ECs and SMCs is a crucial aspect in CLDs triggered by exacerbations such as

air pollutants and pathogens. These stimulants interact directly with the airway epithelium, and lead to

clinical manifestations such as airway narrowing and obstruction that involve SMC proliferation and

4 contraction.2,3 SMC-EC interactions are thus central to elucidating mechanisms in CLDs, and important

for revealing new therapeutic opportunities.

Despite this progress in our understanding of SMC-EC interactions in lung pathophysiology,

many questions regarding the molecular mechanisms regulating lung airway disease processes remain

challenging to address, especially in the context of disease development and exacerbations caused by

pollutant and pathogen exposure. Tackling these challenging questions requires appropriate tissue and

organ models that are convenient, affordable, and more importantly, can accurately represent the

structure and function of the human lung airway microenvironment found in vivo. Most experimental

lung airway models, however, have limitations that do not accurately recapitulate critical aspects of the

in vivo tissue microenvironment. For example, Transwell membrane inserts are used widely for

establishing cocultures between SMCs and epithelia grown on the membrane at the air-liquid interface

(ALI).5,6 However, the membrane is typically made of rigid polymeric materials (e.g., polyester or

polyethylene terephthalate (PET)), which are bio-inert and do not properly mimic the extracellular

matrix (ECM)-laden lamina propria separating the SMCs and ECs within airway tissue. Physiological

airflow may contribute to mechanoregulation of the epithelial layer, but such airflow cannot be readily

applied in typical well-plate or Transwell insert culture formats. Aside from in vitro models, in vivo

animal models are also commonly used in lung airway research studies. Mouse models are especially

valuable due to the wide availability of transgenic models.7 In addition, ex vivo lung slices may be

obtained from mice to allow controlled experimental exposure studies directly on whole tissue

constructs.6 However, while the structure and function of ex vivo tissue slices are more physiologically

relevant than existing culture models, animal tissues inherently differ from human tissues.8 Anatomical,

physiological and immunological differences need to be addressed when translating results from these

tissue sources to human diseases.

5 Recent advances in microengineering and microfluidic systems have enabled researchers to

engineer increasingly more physiological cell culture microenvironments that can provide cells with

various nutrients and physiological stimuli.9,10 In addition, these technologies can be leveraged to build

organ-level tissue structures and geometries with the potential for parallelization and increased

throughput.11,12 Previous microfluidic lung models have been used to investigate the effects of vascular

perfusion, liquid plug flows and cyclic mechanical strain on recapitulating healthy and damaged lung

epithelium.13,14 Recently, much more attention has focused on studying the synergistic effects of

multiple lung airway tissues in the progression of lung inflammatory disorders, particularly the lung

epithelium and the underlying vasculature, on inflammatory responses to pathogens, neutrophil

recruitment and molecular secretion profiles, providing insights into the mechanisms of inflammatory

lung disorders.14,15 In these models, the two tissue types are compartmentalized in vertically stacked

chambers separated by a thin porous polyester membrane coated with exogenous ECM to allow

crosstalk between the tissue types. These models demonstrate the potential for organ-on-chips, and have

opened new opportunities to extend microfluidic technologies to achieve even more complex and

physiologically relevant models.11,15,16 Despite their transformative potential, these designs still have

room for significant advancement in biological complexity, ease of fabrication, increase in throughput,

and additional sample manipulation. The biologically inert polymeric membranes used in current

microchip designs have demonstrated support of cell attachment and growth, but are not representative

of the structure and stiffness of native ECM. Furthermore, coculture between epithelial and

microvascular endothelial cells have been shown, but coculture between ECs and SMCs have yet to be

demonstrated. This is crucial for studies of airway injury and repair after pollutant exposure because

SMCs rather than endothelial cells are present in large and medium airways; endothelial cells are only

present in the smallest airways of the respiratory tree.1 Additionally, current designs are not often as

scalable as desired, and can be time consuming and laborious to fabricate due to challenges of

6 incorporating the polymeric membrane. Therefore, to push the boundaries of organ-on-a-chip

technology, new innovative designs are necessary that can demonstrate coculture of other cell types,

enable transition to biocompatible growth surfaces, and offer increased throughput and reduced

fabrication time to maximize experimental efficiency.

Here, we describe a novel microfluidic cell-based model that offers three unique advantages

from other recently developed microfluidic systems. First, this model uniquely recapitulates airway EC-

SMC interactions in microfluidic cell culture, which can be distinguished from the other models that

have focused on epithelial-endothelial interactions in the small airway. Achieving appropriate SMC

morphology and function on the chip in co-culture with airway epithelia is important for studying

airway constriction and matrix remodeling. Second, rather than separating top and bottom microfluidic

compartments with a biologically inert polyester membrane, our design employs a unique suspended

hydrogel to separate top and bottom compartments. The advantage of a suspended hydrogel is its

biocompatibility, and the fact it can be comprised solely of naturally derived matrix components. Third,

we demonstrate the use of a thermoplastic construction with an arrayable design, features which enable

potential increases in throughput and allow a path towards manufacturability.

7 Materials & Methods

PMMA Micromilling

Our lung airway microdevice (or “airway-chip”) was fabricated from two 1/16″ thick and one 3/32″

thick flat pieces of poly(methylmethacrylate) (PMMA) (#8560K173 and #8560K183, McMaster-Carr,

Elmhurst, IL, USA), which were micromilled with features using in-house protocols.17,18 The device

features were modeled in SolidWorks (Dassault Systèmes, Velizy-Villacoublay, France) and converted

to G-code in SprutCAM (SprutCAM, Naberezhnye Chelny, Russia) for computer numerical control

(CNC) milling. Microfeatures were generated using a Tormach PCNC 770 vertical milling machine

(Tormach, Waunakee, WI) with 4-flute carbide endmills of the following sizes: 1/16″ (1.5875 mm,

#01982156), 3/64″ (1.1906 mm, #07765431) and 0.015″ (381 µm, #37289501) (MSC Industrial Supply

Co., Melville, NY, USA). Key microchannel feature dimensions were as follows: (1) for the top layer of

the device, the channel designed for air exposure (for ALI) was 1.2-mm deep, 5-mm wide by 10-mm

long; (2) for the middle layer of the device, the main channel was 0.7-mm deep, 5-mm wide by 10-mm

long, and the inner microchannel for the suspended hydrogel was 0.65-mm deep, 2.0-mm wide by 6-mm

long. The exposed region to the media reservoir was 0.85-mm wide and 4-mm long; (3) for the bottom

layer, the lower media reservoir channel was 1.2-mm deep, 1.6-mm wide x 12.5 mm long. The inlet and

outlet ports in each layer were 2.0-mm and 2.75-mm in diameter, respectively.

Solvent-Assisted Thermal Bonding

To bond the layers of PMMA, we employed a solvent-assisted thermal bonding method developed in

our lab.19 Briefly, 99% ethanol (Commercial Alcohols Inc., Toronto, ON, Canada) was pipetted between

two layers of PMMA, aligned and pressed with 1000 lbf at 70 °C for 1 min with a Carver Automatic

Hydraulic Laboratory Press (#3889, Carver Inc., Wabash, Indiana, USA). The middle and bottom

8 PMMA layers were bonded first, followed by the bonding of the top layer to the middle layer. See Wan

et al. for step-by-step procedures.19

Standard Cell Culture

Human airway epithelial cells (Calu-3) were obtained from ATCC (HTB-55, Manassas, VA, USA) and

maintained in Minimum Essential Medium α (MEM-α) (#12561-056, Thermo Fisher Scientific,

Burlington, ON, Canada) supplemented with 10% fetal bovine serum (FBS) and 1%

penicillin/streptomycin (P/S) (Thermo Fisher Scientific). Human bronchial smooth muscle cells

(hBSMCs) were obtained from Lonza (#CC-2576, Cedarlane Labs, Burlington, ON, Canada), and

maintained in smooth muscle basal medium (SmBM) supplemented with BulletKit (SmGM-2, Lonza)

and 1% P/S. All cells were maintained in an incubator at 37 °C and 5% CO2. Calu-3 cells were used

between passages 11 to 14, and hBSMCs were used between passages 9 to 12.

Hydrogel Preparation

Rat-tail Type I collagen (CACB354249, Corning) and growth-factor-reduced Matrigel (CACB354320,

Corning) were purchased from VWR (Mississauga, ON, Canada). Collagen was mixed with 20% 0.5-N

NaOH (Sigma-Aldrich, Oakville, ON, Canada) to neutralize the solution, and then incubated at 4 °C for

1 h to allow the formation of thicker collagen fiber structures.20 This collagen solution was then mixed

with Matrigel for preparations of the mixed hydrogels that were tested.

Loading and Culture in Microdevice

To disinfect the lung microdevice, all the internal chambers of the device were flushed at least three

times each with 70% ethanol and 1X PBS. During the last wash, a thin film of PBS was left to render the

walls of the hydrogel suspension channel hydrophilic. To improve adhesion of the hydrogel to the

9 channel walls, 100 µg/mL bovine fibronectin (FN) (F1141, Sigma Aldrich) was pipetted into the

channel immediately after the last PBS flush, and the device was incubated at room temperature for 30

min. After incubation, the FN was aspirated out of the hydrogel suspension channel and a thin coat

along the channel walls was left to improve capillary flow of the hydrogel. The hydrogel was dispensed

into the hydrogel suspension channel, and polymerized in a humidified incubator at 37 °C, 5% CO2 for 1

h, resulting in a suspended polymerized exogenous hydrogel (Fig. 1E). After polymerization, the

chambers of the devices were filled with Calu-3 cell culture media and incubated overnight to keep the

hydrogel hydrated.

Airway Epithelial Cell Culture

Calu-3 cells were seeded as an 80-µL suspension at 3000 cells/µL on top of the suspended hydrogel of

the lung microdevice. The cells were incubated at 37 °C and 5% CO2 overnight in the device, and the

media was replaced the next day to remove unattached cells from the top compartment. The top and

bottom chambers were filled with Calu-3 cell culture media during normal epithelial cell culture and

replenished every 24 h. Cells were cultured for 14 days prior to immunostaining. For air-liquid interface

(ALI) cultures, Calu-3 cells were cultured on the suspended hydrogel for 5 to 7 days before removing

the media from the top chamber and exposing the cells to air. During ALI cultures, the apical surface

was rinsed with PBS daily and maintained at ALI for 3 weeks prior to immunostaining. Calu-3 media

was used in the basolateral side (bottom) compartment and was replenished every 24 h.

Coculture of Lung Epithelial and Smooth Muscle Cells

Calu-3 cells were cultured on the suspended hydrogel, as detailed above, for 7 days prior to seeding of

hBSMCs. At Day 7, hBSMCs were seeded to the bottom chamber as a 50-µL suspension at 750 cells/µL

and the device was turned over to facilitate adhesion of hBSMCs to the underside of the suspended

10 hydrogel. The device rested on two PDMS posts during upside-down culture, and was incubated for 4 h.

Subsequently, the media was replaced in both compartments to remove any unattached cells. The co-

culture was maintained for 1 week prior to immunostaining, and media in both the top and bottom

compartments was replaced every 24 h.

Device Disassembly and Sample Removal

Cells inside the lung microdevice were fixed with 4% paraformaldehyde (Commercial Alcohols Inc.,

Toronto, ON, Canada) for 20 min and permeabilized with 0.1% Triton X-100 (Sigma Aldrich) for 10

min. Cells were washed with PBS three times between each step by flushing the solution through the top

and bottom chambers of the device. To extract the suspended hydrogel sample with cultured airway cells

intact, the top and middle layers of the PMMA device were separated using a single-edge steel razor

blade (Fig. 2A). The blade edge was inserted between each layer at the corner of the device, and gently

moved along the four device edges to slowly open the PMMA-PMMA bonds between layers. Once the

top layer was removed, a scalpel was used to separate the hydrogel from the inner walls of the

microchannel. The hydrogels were then gently rinsed with PBS using a pipette until the hydrogels

completely detached from the device. The released hydrogel was subsequently suspended in PBS in a

0.6-mL centrifuge microtube.

Immunostaining

Standard immunofluorescence staining procedures were performed on hydrogel samples while they

were suspended in 0.6-mL centrifuge microtubes. To visualize tight junctions, samples were blocked

with 3% bovine serum albumin (BSA) & 0.1% Tween-20 in PBS (Sigma-Aldrich), and stored at 4 °C

overnight. ZO-1 primary antibody (Rabbit anti-ZO-1 pAb; Life Technologies, Carlsbad, CA, USA) was

diluted to 10 µg/mL in the above-mentioned blocking buffer and added to the sample. The samples were

11 stored at 4 °C for 48 h. Samples were then labeled with AlexaFluor 568 goat anti-rabbit IgG (10 µg/mL

dilution in blocking buffer; Life Technologies, Carlsbad, CA, USA) and Hoechst 33342 (1 µg/mL

dilution in blocking buffer) at 37 °C, 5% CO2 for 2 h. Samples were flushed with PBS prior to and after

labeling with secondary antibody before imaging. The same procedure was used for α-smooth muscle

actin (α-SMA) staining in hBSMCs, where mouse anti-α-SMA mAb (Sigma, St. Louis, Missouri, USA)

was diluted at 10 µg/mL in blocking buffer and added to the sample. AlexaFluor 488 goat anti-mouse

IgG (Life Technologies, Carlsbad, CA, USA) at 10-µg/mL dilution in blocking buffer was used for

labeling. MUC5AC primary antibody (Mouse anti- MUC5AC mAb; Life Technologies, Carlsbad, CA,

USA) at 20-µg/mL dilution was used to label mucin-producing airway epithelial cells, and conjugated

with AlexaFluor 488 goat anti-mouse IgG using the above-mentioned procedure. To identify F-actin in

both airway cells, AlexaFluor 488 phalloidin (Life Technologies, Carlsbad, CA, USA) at 2.5% (v/v) in

blocking buffer was used.

Imaging and Data Analysis

For imaging, excess fluorescent labeling solution was removed with PBS and samples were mounted on

a 1.2-mm thick 75 x 25 mm glass microscope slide (Fig. 2C). Images were obtained using a fluorescence

microscope (EVOS FL Auto Cell Imaging System). For hBSMC alignment analysis, Hoechst (nuclear)

stained images were analyzed using the ImageJ Orientation J plugin. To measure the angles of each

nuclei, an ROI was selected covering >90% of each stained image.

Statistical Analysis All values are presented as mean ± standard deviation. In total, four independent experiments (n = 4)

were conducted, which means that results from four separate culture systems were used for each

condition on each day for each of the two different cell types. Single-variable analysis of variance

12 (ANOVA) was used for multiple comparisons within a study (e.g., effect of hydrogel on cell adhesion of

each cell type), and multiple comparison tests were performed using Tukey’s method. To analyze the

effects of hydrogel compositions and culture time on the area covered by adhered cells, two-way

ANOVA was performed followed by Sidak’s multiple-comparison test. Statistical differences were

considered significant for p < 0.05.

13 Results and Discussion

Lung Microdevice Design and Fabrication

To recapitulate the in vivo microenvironment of the lung airway within the conducting zone, we

considered the functional roles of major cellular constituents of the airway tissue. The tissue structure of

the airway includes a pseudostratified differentiated epithelium lined with a band of smooth muscle cells

(SMCs) separated by a thin layer of connective tissue called the lamina propria (Fig. 1A).21 These tissue

layers together perform critical roles in the adaptive and innate immune system responses of the airway

that drive development of CLDs.22

Given the roles of these tissue layers in airway cellular responses, we incorporated in our model

a bronchial epithelial cell type (Calu-3 cell line), and a primary human bronchial smooth muscle cell

(hBSMC) type, cultured on opposing sides of a suspended hydrogel. Our device design consisted of

three vertically stacked compartments, fabricated from three milled PMMA layers, where the top and

bottom chambers compartmentalized two different airway cell types and were separated by a suspended

hydrogel in the middle compartment (Fig. 1B-D). Separate inlet and outlet ports were incorporated to

accommodate daily media replenishment for the cells in both top and bottom compartments (Fig. 1D).

The inlet port for the epithelial cells cultured in the top chamber were located above the hydrogel port to

allow easy access to the hydrogel chamber during loading. The outlet port of the top chamber was

located away from the hydrogel outlet port to avoid disturbing the hydrogel during media changes.

14

Figure 1. Lung airway-on-a-chip device design and fabrication. (A) Schematic of airway lumen tissue structure including a thin ECM layer (lamina propria) sandwiched by a ciliated airway epithelium on one side and aligned bronchial smooth muscle cells on the other. (B) Photograph of an assembled lung airway microdevice with the top, middle and bottom compartments highlighted in yellow, green, and red, respectively (scale bar = 10 mm). (C) Lung airway microdevice schematic design showing three vertically stacked layers of PMMA with a top airflow chamber, middle suspended hydrogel chamber and bottom media reservoir. (D) Exploded view of the three PMMA layers that comprise the microdevice, and assembled view of the solvent bonded PMMA layers. (E) The suspended hydrogel is introduced into the middle layer by simple pipette loading. Magnified image of device cross-section showing hydrogel loaded with red-fluorescent microparticles, supported by surface tension between hydrogel and the surfaces of the two protruding ledges (scale bar = 500 µm).

15 PMMA was chosen as our device material because of several desirable properties, including

bioinertness, compatibility with existing mass production infrastructure, low material cost, and optical

transparency. While PDMS has been a popular choice for biomicrofluidic systems, we (and others) have

observed how PDMS can absorb small hydrophobic molecules from solution, which can influence cell

biology readouts, and lead to unintended biases.23,24 Thus, for certain biomicrofluidic applications,

thermoplastics may be more preferable than PDMS.25 Our combined approach of micromilling and

solvent bonding allowed us to align various plastic layers during the bonding process repeatedly. With

micromilling, microscale features can be machined into plastic substrates within the timeframe of

minutes to hours, depending on the complexity of the design. Moreover, we have previously

demonstrated that these milled plastic layers can be solvent-bonded with high bond strength by

employing retention grooves, thereby achieving uniform bond coverage between layers and preserving

the microscale features that have been milled into the layers.19,23,26 This process helped to ensure a leak-

free bond in our devices, which enabled us to incorporate an array of culture systems on a single chip

without cross-contamination.

To incorporate a more biologically relevant component in our system, we leveraged the concepts

of suspended microfluidics27 to integrate a suspended hydrogel rather than a polymeric membrane to

separate the two cell types, and thus attempt to model the lamina propria using matrix proteins in a

microfluidic system. The thickness of the suspended hydrogel can be modified by changing the

thickness of the protruding ledges of the middle plastic layer. To our knowledge, this is the first

demonstration of airway epithelial and smooth muscle cell co-culture on opposing sides of a suspended

hydrogel in a microdevice. By simply pipetting a thin layer of hydrogel between the two protruding

PMMA ledges, the pre-polymerized hydrogel flowed by capillary action along the ledges, thus creating

a thin film of gel with two open interfaces for cell culture (Fig. 1E).

16

Arrayability and Sample Extraction

Using the thermoplastic fabrication techniques described above, we fabricated an array of 12

independent culture systems on a single PMMA device, which matched the format of a 75 x 25 mm

microscope slide (Fig. 2; same channel design with increased array throughput compared to Fig. 1B).

The array increased throughput without risking cross-contamination between adjacent systems due to

leak-free solvent bonding. Cells in culture can be monitored for viability inside each system at various

time points via immunofluorescence staining within the optically clear device. To demonstrate the

versatility of our device in enabling downstream sample processing, we disassembled our device and

extracted cell culture samples with minimal disruption to the sample itself by lodging a single edge razor

blade and severing the interfacial bonds between the top and middle PMMA layers (Fig. 2A-B). The

samples consisted of both cell types adhered to the hydrogel material in between, and effectively

represented a small engineered “tissue” sample that can be readily handled and manipulated (Fig. 2C).

For endpoint analysis, we fixed and permeabilized the cell cultures on the suspended hydrogel inside the

device prior to disassembling the chip, ensuring that the cells were not affected during device

disassembly. After detaching the top and middle layers of the device, the suspended gels in the middle

layer were easily detached from the device walls with tweezers, and then submerged in PBS for

immunofluorescence staining (IFS) (Fig. 2C-D). With this approach, extracted culture samples from

each system can be combined or individually stained in an Eppendorf tube by flushing the samples with

IFS solutions. Therefore, our device fabrication approach enabled simple extraction of culture samples

with minimal damage to the samples, and can allow further downstream sample processing for DNA,

mRNA, protein and molecular analyses if desired to assess more complex cellular functions.

17

Figure 2. Sample gel extraction for immunofluorescence staining. (A) Photograph of a lung airway microdevice with 12 arrayed systems after sample fixation. The top and middle layers of the device were detached by lodging a single edge blade between the layers. (B) Photograph of a disassembled device after sample fixation. (C) A fixed sample (airway cells and ECM hydrogels) that was detached from the side walls of the channels from the middle PMMA layer using a sharp tip of a scalpel. (D) Stained samples on a microscope slide, ready for microscopy.

18

Cell Adhesion to Hydrogel

We studied different hydrogel mixtures to determine an optimal gel composition that could

support co-culture of both Calu-3 cells and hBSMCs over 7 days. Specifically, we chose to test mixtures

of Type I collagen (Col-I) and Matrigel because of the abundance of Col-I in ECM, and because

Matrigel is known to contain basement membrane proteins and is widely used in in vitro assays due to

its ability to promote adhesion.28,29 We tested the adhesion of both airway cell types on four different

hydrogel compositions: (i) Col-I alone; (ii) Matrigel alone; (iii) a mixture of 6 µg/µL Col-I and 3 µg/µL

Matrigel (i.e., referred to below as “high Col-I mix”); and (iv) a mixture of 6 µg/µL Matrigel and 3

µg/µL Col-I (i.e., referred to below as “high Matrigel mix”). Calu-3 cells and hBSMCs were cultured

and stained for nuclei after 2 days and 7 days, and either counted for hBMSCs, or measured for area

coverage in the case of Calu-3s.

After 2 days, Calu-3 cells formed networks with neighboring cells and established partial

monolayers on Col-I, the high Col-I mix and the high Matrigel mix, covering large regions of the

hydrogel surface area. Cells displayed the least area coverage on Matrigel alone (Fig. 3A-B).

Morphological analysis of Calu-3 cells after 2 days confirmed significant statistical differences (**** p

< 0.0001) in area coverage by cells on Matrigel versus all other hydrogel solutions. Interestingly, after 7

days, cell detachment was observed on the Col-I and the high Col-I mix hydrogels, but became more

interconnected as a monolayer on the high Matrigel mix. The low adhesion behavior on Matrigel alone

remained unchanged over the 7-day culture period. After 7 days, significant differences were observed

in area coverage between the high Matrigel mix and all other compositions tested in this study (**** p <

0.0001). No significant difference (p > 0.05) was found in area coverage between Col-I and the high

Col-I mix after 7 days. Area coverage on the high Matrigel mix remained above 80% from Day 2

through to Day 7 (Figure 3B).

19

Figure 3 Calu-3 epithelial cell and hBSMC adhesion on hydrogel in a PMMA lung airway microdevice. Adhered cells were stained with Hoechst 33342 to label the nuclei and observed under fluorescence microscopy on Days 2 and 7. (A) Fluorescent images represent Calu-3 cell adhesion on Type I Collagen alone (Col-I); mixture of 6µg/µL Col-I and 3 µg/µL Matrigel (C6:M3); mixture of 6 µg/µL Matrigel and 3µg/µL Col-I (M6:C3); and Matrigel alone (Matrigel). Dashed lines represent edge of hydrogel. Scale bar = 400 µm. (B) Comparison of area covered by adhered Calu-3 cells (%) on different hydrogel compositions after 2 d and 7 d. Data presented as mean ± SD (n = 4, **** p < 0.0001). (C) Cell adhesion and growth of hBSMCs cultured for 7 d on different hydrogels. Graphs represent a comparison of the number of attached hBSMCs per mm2 of hydrogel surface inside a PMMA microfluidic device after 2 d and 7 d of culture. Data presented as mean ± SD (n = 4).

20 For primary hBSMCs, cells appeared to be uniformly dispersed (without networks) on Col-I, the

high Col-I mix and the high Matrigel mix. Similar to Calu-3s, hBSMCs appeared to form aggregates in

small regions on Matrigel alone instead of evenly across the surface. Combined with the results obtained

for Calu-3 adhesion, the low adhesion of both cell types on Matrigel alone warranted excluding Matrigel

alone from further study in our device. After 2 days on the other three gel compositions, hBSMCs

appeared to be well dispersed on all three hydrogels. Quantitative analysis of attached cells per mm2

revealed no significant differences (p > 0.05) between the three hydrogel compositions after both 2 and

7 days (Fig. 3C). However, cells appeared to be more confluent after 7 days, suggesting some

proliferation may have occurred. Nuclei count confirmed these observations of proliferation from Day 2

to Day 7, as well as no significant differences in cells attached per mm2 between the three hydrogels

tested on both Day 2 and Day 7.

While previous studies have shown increased Calu-3 cell adhesion, proliferation and confluent

monolayer formation on Col-I-coated substrates compared to non-coated substrates,30,31 our results

suggested that cell adhesion may be quite different when cultured on a Col-I hydrogel of ~500-µm

thickness compared to a more conventional thin coating of Col-I. Furthermore, this indicated that aside

from integrin signaling between cells and the ECM that are known to influence cell adhesion, other

factors such as substrate stiffness, surface topography and availability of growth factors may also play

important roles. The potential impact of Col-I gel stiffness on Calu-3 epithelial adhesion may thus

require further investigation.

In addition, airway smooth muscle cell adhesion studies on 2D substrates in the past have

suggested that interstitial ECM proteins like Col-I promote proliferation and expression of less

contractile phenotypes while basement membrane proteins stimulate expression of a differentiated

contractile phenotype and inhibits proliferation.32,33 This is due to the effects of laminin, a constituent of

Matrigel, and its primary integrins that inhibit airway SMC growth.34 Consistent with these findings, we

21 found that hBSMCs seeded on Matrigel did not adhere and spread over a 7-day culture period compared

to other hydrogels tested. Interestingly, hBSMC adhesion was not affected by the hydrogel mix with

high Matrigel concentration. A possible explanation for this observation is that the primary integrin

receptors, more specifically α1β1 and α2β2, that mediate growth inhibition of laminin also interact with

Col-I 35, which is present in our hydrogel mixture. These interactions may have prevented the growth

inhibitory effects of laminin on hBSMCs, leading to improved cell adhesion compared to Matrigel

alone. This may also explain previous results that showed successful endothelial cell adhesion on a

mixture of Col-I and Matrigel within a microfluidics context.36

Together these results showed that a mix with Col-I and high Matrigel concentration facilitated

increased initial and sustained Calu-3 cell and hBSMC adhesion over a 7-day culture period.

Specifically, our results showed that Col-I was important for initial cell adhesion to matrix, whereas the

ECM proteins in Matrigel facilitated sustained cell adhesion over a longer culture period. Based on these

findings, the high Matrigel mix was used for all subsequent work in this study.

ALI Culture Protocol and Characterization

Calu-3 cells were cultured on the top surface of the suspended hydrogel for 14 days, with daily media

replacements on both apical and basolateral sides, before exposing to ALI for more than 2 weeks (17

days) (Fig. 4A). During ALI culture, only the basolateral media was replaced, and the apical side was

flushed with 1X PBS every 3 days to remove accumulating mucus. To examine both intracellular and

barrier structures of the epithelium, we immunostained for F-actin and ZO-1 tight junctions after 7 and

14 days, respectively (Fig. 4B). F-actin staining revealed strong localization around the periphery of

each epithelial cell, consistent with a typical cell cortex found with cells within a monolayer (Fig. 4B).

ZO-1 staining showed cobblestone morphology and well-defined polygonal rings localized to the outer

periphery of the cells. This morphology was less pronounced after 7 days when ZO-1 was more diffuse,

22 and more pronounced after 10 days of submerged culture when both F-actin and ZO-1 rings were clearly

visible (Fig. 4B, 14 days shown). This type of morphology is common in Calu-3 monolayers, and is

indicative of stable barrier properties formed in these monolayers.37,38 We chose to expose epithelial

cells to ALI after 14 days of submerged culture to ensure stable tight junction barrier formation prior to

ALI exposure.

At 31 days (i.e., 14 days of submerged culture followed by 17 days of additional ALI culture),

Calu-3 cultures were fixed, extracted and immunostained for MUC5AC, a goblet cell marker and a

protein abundant in the mucosal layer of the airway. We observed that dispersed groups of Calu-3 cells

on the monolayer stained positive for MUC5AC expression (Fig. 4C), suggesting that our system was

capable of accommodating ALI culture conditions that can facilitate airway epithelial differentiation into

goblet cells. This is consistent with cellular compositions typically found in native small airways.

Furthermore, to ensure that the tight junctions that appeared by 14 days were not compromised during

ALI culture, we co-stained the culture for ZO-1, and found similar tight junction formation after 31 days

(Fig. 4C). These results showed that the airway epithelium in our lung device can establish tight

junctions between cells, and differentiate into mucus-producing goblet cells. We also demonstrated the

ability of our device to accommodate long-term airway epithelial cell cultures, including ALI culture

that can last several weeks (> 4 weeks) without losing barrier structure or differentiation capacity.

23

Figure 4. ALI culture protocol and characterization of Calu-3 cell cultures on suspended hydrogel. (A) Protocol for air-liquid interface (ALI) culture of epithelial cells (dotted line represents longitudinal cross-sectional plane in illustrations). Cells are seeded and cultured under submerged culture conditions for 14 days prior to ALI exposure for an additional 17 days with daily media replacement. (B) Immunofluorescence image of monolayers formed by cultured Calu-3 cells after 7 and 14 days. Green: F-actin; Red: ZO-1 tight junctions; Blue: Hoechst nuclei (scalebar = 20 µm). (C) Immunofluorescence image of Calu-3 cells in monolayer expressing goblet cell marker MUC5AC (green) and ZO-1 tight junctions (red) (scalebar = 50 µm).

24 Coculture Protocol and Characterization

For co-culture of both airway ECs and SMCs, Calu-3 epithelial cells were first seeded and cultured on

the suspended hydrogel for 7 days under submerged culture conditions. Subsequently, hBSMCs were

seeded in the bottom compartment, and the device was flipped upside-down for 4 h to allow the

hBMSCs to settle due to gravity and adhere to the underside of the gel. After adhesion, the device was

flipped back to the upright position for the remainder of the culture period, i.e., an additional 14 days

prior to fixation, extraction and staining (Fig. 5A). We performed co-culture characterization to

highlight phenotypic differences between the two cell types and to determine if the characteristic

properties of Calu-3 monolayer and hBSMCs can be reproduced and maintained under coculture

conditions. At 14 days, the coculture samples were co-stained for ZO-1, nuclei and alpha smooth muscle

actin (α-SMA), a phenotypic marker for SMCs. Similar to the monocultures, Calu-3 cells displayed

well-defined polygonal rings of ZO-1 localized to the periphery of the cells under co-culture conditions,

indicating that hBSMCs did not interfere with epithelial tight junction formation (Fig. 5B). This

morphology was observed uniformly across the Calu-3 monolayers in coculture. Additionally, hBSMCs

were positively stained for α-SMA (Fig. 5C). Confocal 3D images of our cocultures stained with both α

-SMA and ZO-1 displayed a clear distinction between the two cell types separated by the suspended

hydrogel (Fig. 5D-E). Importantly, the integrity of the suspended hydrogel appeared undisturbed

throughout the culture, extraction and staining processes.

25

Figure 5. Protocol for coculturing Calu-3 epithelial cells with hBSMCs (dotted line represents cross-sectional plane in illustrations). (A) Protocol for coculture Calu-3 cells and hBSMCs (dotted line represents cross-sectional plane in illustrations). Calu-3 cells were seeded and cultured under submerged culture conditions for 7 days prior to hBSMC seeding. The device was flipped over for hBSMC adhesion (4 hours) and cocultured with Calu-3 cells in the upright positions for 7 more days with daily media replenishment. (B) Immunofluorescence image of monolayers formed by cultured Calu-3 cells after 14 days of total culture and 7 days coculture (Green: F-actin; Red: ZO-1 tight junctions), scalebar = 20 mm). (C) Immunofluorescence image of alpha smooth muscle actin (α-SMA) in hBSMCs (Red: α-SMA; Blue: Hoechst nuclei) cocultured with Calu-3 cells for 7 days), scalebar = 50 mm. (D) 3-D rendered confocal laser micrograph of Calu-3 and hBSMC coculture on two sides of the suspended ECM hydrogel. (E) Confocal laser micrograph of coculture sample viewed from the x-z plane. (F) Polar graphs showing frequency distribution of the angle of hBSMC alignment with respect to the inlet-outlet axis of the microfluidic channel. Radial (horizontal) axis represents the frequency of angle of alignment. Each bar represents a 10°-range. Arrows point to the outlet of the microfluidic channel. Polar graphs represent frequency distribution for one representative sample on Day 4 (top panel) and Day 7 (bottom panel) of co-culture. A narrower distribution is observed for 7-day co-culture compared to a 4-day co-culture (n = 134). Mean alignment angles and standard deviations of each system are indicated on the top right corner of each graph for each system.

26 Interestingly, in our analysis of α-SMA in hBSMCs, the actin stress fibers appeared to be

predominately oriented in the direction parallel the length of the channel (Fig. 5F). Quantitative

assessment of angle of orientation revealed that this preferential alignment was more significant with

longer culture times and higher cell densities. Actin fiber orientation angles were measured and the

frequency distributions were plotted in polar graphs (Fig. 5F). A total of 134 cells were analyzed for

their fiber orientations from each image obtained and grouped into 10º bins. After 4 days of co-culture,

cells appeared to be more randomly oriented, after 7 days of co-culture, however, a narrower distribution

of alignment angles were observed (Fig. 5F). The angle of orientation spanned from the northwest

direction to the northeast direction with a mean alignment angle of 78º and standard deviation of 43º for

our representative sample of 4-day co-culture. For a 7-day culture sample, a mean alignment angle of

73º and standard deviation of 21º was observed. These observations were consistent throughout replicate

experiments for each day.

An important distinction between the expansion of Calu-3 epithelial cultures in the monoculture

setup (Fig. 4) and coculture setup (Fig. 5) is that there is crosstalk between two different cell types that

may influence the growth and development of these cultures. Here we show that hBSMC coculture with

Calu-3 cells in our lung device did not affect the ability of the epithelial cells to from tight junctions.

Furthermore, hBSMCs appeared to align along the longitudinal direction of the device with extended

culture time, taking on a morphology similar to that seen in vivo.

With this device design, coculturing of SMCs and ECs, hydrogel extraction, and costaining for

various phenotypic markers can easily be done with our chip without cross-contamination between

culture systems. Importantly, the ability to remove the cell-laden hydrogel from the device – with both

cell layers intact and without disruption of the gel – enables possible downstream biological analyses to

study gene and protein expression of either cell type, as well as matrix deposition and remodeling events

that are critical to understanding airway disease development. Of particular interest is the future

27 potential to incorporate airflow over the airway epithelium at the ALI of the upper chamber, which is

possible due to the geometry of the microchannel features with the current design.

Conclusion

We have developed a plastic airway-on-a-chip device for culturing airway ECs and SMCs on opposing

sides of a suspended hydrogel, and performed preliminary experiments to demonstrate long-term

viability and assess its fitness as an in vitro lung airway model. We used a combined micromilling and

solvent bonding approach to fabricate a thermoplastic-based platform that is suitable for making arrays

of culture systems in a single microfluidic device within a matter of hours. The key feature of the design

is the use of “open microfluidics” to suspend a thin layer of hydrogel that can support long-term culture

of both airway ECs and SMCs. The device was also designed to accommodate ALI culture, which was

shown to lead to the differentiation of airway epithelial cells into mucus-producing goblet cells inside

the microfluidic device. As part of preliminary assessment, the device was used to determine an optimal

hydrogel composition that promoted airway cell adhesion and proliferation. In monoculture, ECs

displayed cobblestone morphology with tight junction proteins, and stained positive for the goblet cell

marker MUC5AC under ALI culture conditions without the loss of tight junction markers. In coculture,

ECs retained tight junctions in monolayer, while SMCs appeared highly aligned and expressed α-SMA

proteins, which are critical for motility and contractility functions. These findings demonstrated the

usefulness of the designed platform for its ability to accommodate long-term SMC-EC coculture, to

allow immunofluorescence staining and sample handling without disrupting the matrix or cells, and to

offer increased throughput due to the arrayability of the design. The platform has significant potential

for use as a lung airway tissue model that can incorporate other microenvironmental cues, including

airflow and pollutant exposures for studying mechanisms associated with CLDs.

28 Conflict of Interest

There are no conflicts of interest to declare.

Acknowledgements

We acknowledge financial support from the Natural Sciences and Engineering Research Council of

Canada (NSERC) Discovery Grant, and from the Canadian Lung Association / Ontario Thoracic Society

Grant-in-Aid Award to EY.

29 References

1 P. Farrell, Lung Development Biological and Clinical Perspectives : Biochemistry and Physiology., Elsevier Science, 2012.

2 K. F. Chung and I. M. Adcock, Eur. Respir. J., 2008, 31, 1334–1356. 3 D. J. Erle and D. Sheppard, J. Cell Biol., 2014, 205, 621–631. 4 D. M. Mannino and A. S. Buist, Lancet, 2007, 370, 765–773. 5 N. K. Malavia, C. B. Raub, S. B. Mahon, M. Brenner, R. A. Panettieri and S. C. George, Am. J.

Respir. Cell Mol. Biol., 2009, 41, 297–304. 6 C. Blume and D. E. Davies, Eur. J. Pharm. Biopharm., 2013, 84, 394–400. 7 C. S. Dela Cruz, M.-J. Kang, W.-K. Cho and C. G. Lee, Immunology, 2011, 132, 9–17. 8 D. B. Corry and C. G. Irvin, Immunol. Res., 2006, 35, 279–294. 9 E. W. K. Young and C. A. Simmons, Lab Chip, 2010, 10, 143-160. 10 E. W. K. Young and D. J. Beebe, Chem. Soc. Rev., 2010, 39, 1036-1048. 11 D. Huh, G. A. Hamilton and D. E. Ingber, Trends Cell Biol, 2011, 21, 745–754. 12 C. Moraes, G. Mehta, S. C. Lesher-Perez and S. Takayama, Ann Biomed Eng, 2012, 40, 1211–

1227. 13 D. Huh, H. Fujioka, Y. Tung, N. Futai, R. Paine, J. Grotberg and S. Takayama, Proc. Natl. Acad.

Sci. U. S. A., 2007, 104, 18886–18891. 14 D. Huh, B. D. Matthews, A. Mammoto, M. Montoya-Zavala, H. Y. Hsin and D. E. Ingber,

Science, 2010, 328, 1662–1668. 15 K. H. Benam, R. Villenave, C. Lucchesi, A. Varone, C. Hubeau, H.-H. Lee, S. E. Alves, M.

Salmon, T. C. Ferrante, J. C. Weaver, A. Bahinski, G. A. Hamilton and D. E. Ingber, Nat. Methods, 2016, 13, 151–157.

16 D. Trieu, T. K. Waddell and A. P. McGuigan, Biomicrofluidics, 2014, 8, 64104. 17 D. J. Guckenberger, T. de Groot, A. M.-D. Wan, D. Beebe and E. Young, Lab Chip, 2015, 15,

2364–2378. 18 D. Konstantinou, A. Shirazi, A. Sadri and E. W. K. Young, Sensors Actuators, B Chem., 2016,

234, 209-221. 19 A. M. D. Wan, T. A. Moore and E. W. K. Young, J. Vis. Exp., 2016, 119, e55175. 20 K. E. Sung, G. Su, C. Pehlke, S. M. Trier, K. W. Eliceiri, P. J. Keely, A. Friedl and D. J. Beebe,

Biomaterials, 2009, 30, 4833–4841. 21 Q. Hamid, J. Shannon and J. Martin, Physiologic Basis of Respiratory Disease, McGraw-Hill

Europe, 2005. 22 B. N. Lambrecht and H. Hammad, Nat. Med., 2012, 18, 684–692. 23 T. A. Moore, P. Brodersen and E. W. K. Young, Anal. Chem., 2017, 89, 11391–11398. 24 K. J. Regehr, M. Domenech, J. T. Koepsel, K. C. Carver, S. J. Ellison-Zelski, W. L. Murphy, L.

A. Schuler, E. T. Alarid and D. J. Beebe, Lab Chip, 2009, 9, 2132–9. 25 E. Berthier, E. W. K. Young and D. Beebe, Lab Chip, 2012, 12, 1224–1237. 26 A. M. D. Wan, A. Sadri and E. W. K. Young, Lab Chip, 2015, 15, 3785–3792. 27 B. P. Casavant, E. Berthier, A. B. Theberge, J. Berthier, S. I. Montanez-Sauri, L. L. Bischel, K.

30 Brakke, C. J. Hedman, W. Bushman, N. P. Keller and D. J. Beebe, Proc Natl Acad Sci U S A, 2013, 110, 10111–10116.

28 C. Bonnans, J. Chou and Z. Werb, Nat. Rev. Mol. Cell Biol., 2014, 15, 786–801. 29 G. Benton, I. Arnaoutova, J. George, H. K. Kleinman and J. Koblinski, Adv. Drug Deliv. Rev.,

2014, 79–80, 3–18. 30 B. I. Florea, M. L. Cassara, H. E. Junginger and G. Borchard, J. Control. Release, 2003, 87, 131–

138. 31 L. Zhang, S. Du, Y. Lu, C. Liu, Z. Tian, C. Yang, H. Wu and Z. Wang, Drug Des. Devel. Ther.,

2016, 10, 2227–2237. 32 M. Yamamoto, K. Yamamoto and T. Noumura, Exp. Cell Res., 1993, 204, 121–9. 33 X. Li, P. Tsai, E. D. Wieder, A. Kribben, V. Van Putten, R. W. Schrier and R. A. Nemenoff, J.

Biol. Chem., 1994, 269, 19653–8. 34 S. J. Hirst, C. H. C. Twort and T. H. Lee, Am. J. Respir. Cell Mol. Biol., 2000, 23, 335–344. 35 J. Thyberg, Int. Rev. Cytol., 1996, 169, 183–265. 36 L. L. Bischel, E. W. K. Young, B. B. R. Mader and D. J. Beebe, Biomaterials, 2013, 34, 1471–

1477. 37 C. I. Grainger, L. L. Greenwell, D. J. Lockley, G. P. Martin and B. Forbes, Pharm. Res., 2006, 23,

1482–1490. 38 C. E. Stewart, E. E. Torr, N. H. Mohd-Jamili, C. Bosquillon and I. Sayers, J. Allergy, 2012, 2012,

943982.