A Microfluidic Model of Biomimetically Breathing Pulmonary Acinar Airways

Summary

Soft-lithography was utilized to produce a representative true-scale model of pulmonary alveolated airways that expand and contract periodically, mimicking physiological breathing motion. This platform recreates respiratory acinar flows on a chip, and is anticipated to facilitate experimental investigation of inhaled aerosol dynamics and deposition in the pulmonary acinus.

Abstract

Quantifying respiratory flow characteristics in the pulmonary acinar depths and how they influence inhaled aerosol transport is critical towards optimizing drug inhalation techniques as well as predicting deposition patterns of potentially toxic airborne particles in the pulmonary alveoli. Here, soft-lithography techniques are used to fabricate complex acinar-like airway structures at the truthful anatomical length-scales that reproduce physiological acinar flow phenomena in an optically accessible system. The microfluidic device features 5 generations of bifurcating alveolated ducts with periodically expanding and contracting walls. Wall actuation is achieved by altering the pressure inside water-filled chambers surrounding the thin PDMS acinar channel walls both from the sides and the top of the device. In contrast to common multilayer microfluidic devices, where the stacking of several PDMS molds is required, a simple method is presented to fabricate the top chamber by embedding the barrel section of a syringe into the PDMS mold. This novel microfluidic setup delivers physiological breathing motions which in turn give rise to characteristic acinar air-flows. In the current study, micro particle image velocimetry (µPIV) with liquid suspended particles was used to quantify such air flows based on hydrodynamic similarity matching. The good agreement between µPIV results and expected acinar flow phenomena suggest that the microfluidic platform may serve in the near future as an attractive in vitro tool to investigate directly airborne representative particle transport and deposition in the acinar regions of the lungs.

Introduction

A detailed quantification of respiratory flow dynamics in the distal, alveolated regions of the lungs is paramount towards understanding airflow mixing in the pulmonary acinus and predicting the fate of inhaled aerosols in the deepest airways^1–3. This latter aspect is of particular concern when addressing on the one hand the hazards of inhaled pollutant particles or conversely in seeking novel strategies for improved and targeted drug delivery of inhaled therapeutics to localized lung sites^{4, 5} as well as for systemic delivery.

To date, respiratory flows in the deep pulmonary acinar regions have been typically investigated in silico using computational fluid dynamics (CFD) or alternatively in vitro with scaled-up experimental models following hydrodynamic similarity matching. In the past few decades, CFD methods have been increasingly applied to study acinar flow phenomena, from single alveolar models^{6, 7} and alveolated ducts^8–12 to more elaborate in silico models that capture anatomically-realistic acinar tree structures with multiple generations of alveolated ducts and up to several hundreds of individual alveoli^13–15.

Together, numerical efforts have been pivotal in shedding light on the role and influence of wall motion during breathing movements on ensuing acinar airflow patterns. In the absence of breathing motion, static alveoli feature recirculating flows within their cavities that exhibit no convective exchange of air between the acinar duct and the alveolus^{6, 7}; in other words, alveolar flows would be entirely isolated from flows within the acinar trees and exchange of air would result uniquely from diffusive mechanisms. With the existence of cyclic expansions of the alveolar domain, however, alveolar flow topologies are drastically modified and the resulting flow patterns inside alveoli are intimately tied to the location of an alveolus along the acinar tree (e.g., proximal vs. distal generations).

In particular, it has been hypothesized in simulations that alveolar flow patterns are strongly influenced by the ratio of alveolar to ductal flow rates such that proximal generations of the pulmonary acinar tree, where this ratio is relatively large following mass conservation across a tree structure, feature complex recirculating flows inside the alveolar cavities with irreversible fluid pathlines. With each deeper acinar generation, the ratio of alveolar to ductal flow rates gradually decreases such that distal acinar generations exhibit more radial-like streamlines that are reminiscent of simple inflations and deflations of a balloon. With advances in modern imaging modalities, lung imaging data^{16, 17} of rodents, including rat and mouse, have given rise to some of the first CFD simulations of anatomically-reconstructed acinar flows in reconstructed alveoli. Despite such promising progress, these recent studies are still limited to addressing airflow phenomena in terminal alveolar sacs only^{18, 19} or a few alveoli surrounding a single duct²⁰. As a result, state-of-the-art investigations of respiratory flow phenomena in the acinus remain dominated by studies focusing on generic anatomically-inspired geometries of the acinar environment².

On the experimental side, various setups featuring an airway with one or several alveoli have been developed over the years^21–24. Yet, there exists no experimental models of bifurcating alveolated airways that are capable of mimicking physiological respiration by expanding and contracting in a breathing-like fashion. Given a lack of attractive experimental platforms at hand, the study of acinar transport phenomena remains limited with regards to validating computational studies and critically, there remains a dearth of experimental data available. In recent years, Ma et al. (2009) have constructed a scaled-up rigid-wall model of an acinus consisting of three acinar generations; however, the lack of wall motion in this model limited its capability to capture realistic alveolar flow patterns under breathing conditions.

Other scaled-up experiments including a moving wall model based on anatomical data from cast replica were recently introduced²⁵; however, since the model only captured the last two acinar generations (i.e., terminal sacs), it failed to capture the complex recirculating flows that characterize more proximal acinar generations. These latter examples of scaled-up experiments further underline the ongoing limitations with such approaches. Specifically, no existing experiment has thus far demonstrated the hypothesized transition from recirculating to radial flows along the acinus and thereby confirm numerical predictions of flow topologies hypothesized to exist in real pulmonary acinar trees^{7, 15}. Perhaps most critically, scaled-up experiments are extremely limited in investigating inhaled particle transport and deposition dynamics²⁶ due to difficulties in matching all relevant non-dimensional parameters (e.g., particle diffusion, a critical transport mechanism for sub-micron particles, is completely neglected).

With ongoing experimental challenges, new experimental platforms that permit investigations of respiratory air flows and particle dynamics in complex moving walls acinar networks are sought. Here, an anatomically-inspired in vitro acinar model is introduced. This microfluidic platform mimics pulmonary acinar flows directly at the representative acinar scale, and broadens the growing range of pulmonary microfluidic models²⁷, including bronchial liquid plug-flows^28–30 and the alveolar-capillary barrier³¹.

Namely, the present design features a simplified five generation alveolated airway tree with cyclically expanding and contracting walls, where cyclic motions are achieved by controlling pressure inside a water chamber which surrounds the thin PDMS lateral walls and where the top wall is deformed by an additional water chamber sitting directly above the acinar structure. Unlike common multilayer microfluidic devices, this chamber is simply formed by embedding the barrel section of a syringe inside the PDMS Device, and does not require preparation of an additional PDMS mold.

The miniaturized approach presented here offers a simple and versatile means for reproducing complicated acinar structures with moving walls as compared to scaled-up models while capturing the underlying characteristics of the acinar flow environment. This platform can be used for flow visualization using fluid-suspended particles inside the airways (see Representative Results below). In the near future, the model will be used with airborne particles for studying inhaled acinar particle dynamics.

Protocol

1. Master Fabrication

1. Use deep reactive ion etching (DRIE) of a silicon on insulator (SOI) wafer to fabricate a master silicon wafer as described in former works^{32, 33}.
   NOTE: DRIE is preferred to standard SU-8 micromachining due to the high aspect ratio features (40 µm wide and 90 µm deep trenches).

2. Casting and Sealing of the Microfluidic Device

1. Mix PDMS and curing agent at a 10:1 weight ratio inside a clean small container such as a plastic weighing dish.
2. Degas the mixture in a desiccator under vacuum until all air bubbles are removed.
   NOTE: Prepare enough PDMS for all subsequent steps. Here below, the acronym "PDMS" refers always to the degassed 10:1 PDMS:curing-agent mixture that was prepared in steps 2.1 and 2.2.
3. Pour the degassed-mixture to a height of approximately 1 mm above the master wafer. Degas once again for at least 40 min to remove all the air bubbles above the wafer and minimize the bubbles below the wafer.
   NOTE: Make sure that the wafer is as close as possible to the bottom of the plate. If necessary press the wafer gently to the bottom using 2 stirring sticks and degas once again.
4. Bake at 65 °C for 20 min in a natural convection oven.
   NOTE: After 20 min the PDMS is hardened and almost fully cured. While a longer baking time is possible baking for 20 min saves time and improves the adherence of the second PDMS layer (see below) to the first one.
5. File the barrel section of a plastic 2 ml syringe using a fine grit sand paper to improve adherence to PDMS. In addition, use the sand paper to flatten the base of the syringe barrel by placing the sand paper on a flat surface and sliding the base of the syringe barrel on top of it. Clean the syringe using pressurized air.
6. Place the barrel section of the syringe on top of the first PDMS layer with the large opening facing the surface of the PDMS, and pour a second layer of PDMS on top of the first one to a height of ~5 mm, and degas the PDMS once again in a desiccator.
   NOTE: The second PDMS layer should be poured from the small container around the barrel, and should not enter within it. 
7. Bake the entire setup at 65° C for at least 2 hr in a natural convection oven.
   NOTE: There is no need to hold the barrel in place during the curing processes since the weight of the PDMS pressing against the wide base of the barrel holds the barrel firmly in place.
8. Cut through the PDMS mold around the patterned region of the master wafer using a scalpel. While cutting, the scalpel should weakly touch the surface of the wafer. Then, gently insert a thin tool such as wafer forceps in the notch created by the scalpel, and peel off the PDMS cast from the master wafer.
9. Place the cast on a soft surface covered with aluminum foil with the patterned side facing up (i.e., the barrel should hang from the edge of the table), and punch a hole in the PDMS at the chamber inlet and channel inlet using a 1 mm biopsy punch. 
10. Coat a clean glass slide with a (degassed) 10:1 PDMS:curing-agent mixture using a spin coater programmed at 3,000 rpm for 30 sec, and bake for >1 hr at 65 °C. Then, clean the slide and PDMS cast using clear tape.
11. Treat the surface of the PDMS mold and PDMS coated glass slide with O₂ plasma (e.g., using a hand-held corona treater) for 1 min, and then gently press the surfaces together and bake at 65 °C overnight (O/N).

3. Device Filling and Actuation

1. Mix water suspended fluorescent polystyrene particles with water and glycerol in a glass vial to obtain a 64/36 (v/v) glycerol/water mixture with 0.25% (w/w) particles..
2. Place a drop of the glycerol solution on top of the channel inlet and a drop of DI water on the chamber inlet, then place the apparatus inside a desiccator and vacuum for ~5 min.
   NOTE: Before releasing the vacuum wait for the bubbles that form in the drops of glycerol solution and DI water to pop. Upon vacuum release the liquids are sucked into the voids inside the device. If residual air remains inside the channels, eliminate it by applying external pressure on the fluids (e.g., using a syringe) and allowing the air to diffuse into the PDMS.
3. Inject ~2 ml of DI water into the top chamber (i.e., the syringe barrel, Fig. 2b) until it is fully filled with water. Then cover the top chamber with a 19 gauge blunt syringe tip, cut the tip of another blunt 19 gauge syringe tip and insert this tip to the side chamber inlet. Connect both syringe tips to a 1 ml syringe via thin Teflon tubing and a T-shaped connector.
   NOTE: Make sure that the 1 ml syringe, Teflon tubing, T-shaped connector and top chamber (2 ml syringe barrel) are all filled with water without bubbles. This may be achieved by opening connection points, pushing water through empty sections of tubing and reconnecting the connection points.
4. Connect the 1 ml syringe to a syringe pump pre-programmed to mimic for example a quiet tidal breathing cycle (with a period of T = 4 sec) constructed of linear ramps, i.e., from zero to 1.8 ml/min in 1 sec, from 1.8 ml/min to -1.8 ml/min in 2 sec and from -1.8 ml/min back to zero in 1 sec.

4. Flow Visualization Experiments: Micro-particle Image Velocimetry (µPIV)

1. While the device is being actuated, obtain a series of 9 - 12 phase-locked, double-frame images of the particle-seeded flow using a micro-particle image velocimetry (µPIV) system consisting for example of a dual frame-multiple exposure CCD camera (e.g., 1,600 × 1,200 pixels to achieve sufficient resolution), a double pulsed Nd-YAG laser (wavelength: 532 nm, output energy: 400 mJ, pulse duration: 4 nsec), and an inverted microscope.
   NOTE: Such a system is capable of obtaining frame pairs with a time lag of down to a few microseconds between the first and second frames. To achieve phase-locked double frame images, it is useful to acquire a double frame series at e.g., 10 Hz (frame pairs are separated by 0.1 sec from one another). Then, the data may be reorganized so that all frame pairs that are separated by a full cycle time (here T = 4 sec) form a new time series. Image acquisition should be repeated several times while modifying the lag time between the first and second frames in each frame pair (i.e., 100 µsec to 0.1 sec) for resolving different flow regions inside the alveolar cavity.
   Note: alternative setups with regards to best combinations of image acquisition systems (i.e., camera) and illumination sources (i.e., lasers) to image such microflows are also available^{34, 35}.
2. Use a sum-of-correlation algorithm to compute phase-locked velocity vector maps of the resulting flow field from the image series for each time lag used. Repeat this process several times with varying lag times between the first and second frames in each frame pair for resolving different flow regions inside the alveolar cavity. Next, use a data analysis program to stitch together the individual flow maps into a complete and high-detailed map of flow patterns by averaging overlapping data points³³.

Results

Computer-aided design (CAD) and microscope images of the in vitro acinar platform are presented in Fig. 1. The biomimetic acinar model features five generations of branching rectangular channels lined with alveolar-like cylindrical cavities (Fig. 1). Here, the model generations are numbered from generation 1 (for the most proximal generation) to generation 5 (for the most distal generation). Note that only the channel inlet leading to generation 1 is open to the outer environmen...

Discussion

A critical feature of the microfluidic acinar platform presented here is its ability to reproduce physiologically-realistic breathing motions that give rise to physiological flow profiles and velocities within acinar ducts and within alveoli. Since the microfluidic channels are produced with a relatively low aspect ratio (i.e., w_d/h ≈ 3.9, where w_d is the duct width and h is the duct height), the measured flows show more plug-like ...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Polydimethylsiloxane (PDMS) and curing agent	Dow Corning	(240)4019862	Sylgard® 184 Silicone Elastomer Kit
Plastipak 2 ml syringe	BD	300185
Norm-Ject Luer slip 1 ml syringe	Henke Sass Wolf	4010-200V0
1 mm Biopsy punch	Kai Medical	BP-10F
Laboratory Corona Treater	Electro-Technic Products	BD-20AC
PHD Ultra Syringe pump	Harvard apparatus	703006
Dyed red rqueous fluorescent particles	Thermo-Scientific		Uncatalloged 0.86 µm beads were used
Glycerin AR	Gadot	830131320
FlowMaster MITAS micro-particle image velocimetry (µPIV) system	LaVision	1108630