This mechanical force–induced increase in nanoparticle translocation was reduced significantly when the cells were incubated with the antioxidant NAC (Fig. 4J). This finding implies that elevated intracellular production of ROS due to a combination of mechanical strain and nanoparticle exposure might be responsible for the observed increase in barrier permeability to these nanoparticles. We also found that preconditioning of endothelial cells with physiological levels of shear stress (15 dyne/cm2) slightly increased the rate of nanoparticle translocation in the presence of mechanical stretch (fig. S11), presumably due to a shear-induced increase in endothelial permeability, as previously demonstrated by others (39).

To determine the physiological relevance of these observations, we conducted similar studies in a whole mouse lung ventilation-perfusion model, which enables intratracheal injection of nebulized nanoparticles and monitoring of nanoparticle uptake into the pulmonary vasculature ex vivo (fig. S10). When 20-nm nanoparticles were injected into whole breathing mouse lung, they were delivered into the deep lung, and histological analysis revealed that they reached the surface of the alveolar epithelium, as well as the underlying interstitial space and microvasculature (Fig. 4H).

When the ultrafine (12 nm) silica nanoparticles were added to the alveolar epithelium in the absence of mechanical distortion, there was little or no ROS production. However, when the cells were subjected to physiological levels of cyclic strain (10% at 0.2 Hz), the same nanoparticles induced a steady increase in ROS production that increased by a factor of more than 4 within 2 hours (Fig. 4B), and this response could be inhibited with the free radical scavenger, N-acetylcysteine (NAC) (fig. S5). ROS levels in the underlying endothelium also increased by a factor of almost 3 over 2 hours, but initiation of this response was delayed by about 1 hour compared with the epithelium (fig. S5). Experiments with carboxylated Cd/Se quantum dots (16 nm) produced similar results (fig. S6), whereas cyclic strain alone had no effect on ROS even when applied for 24 hours (fig. S7). Nanomaterial-induced ROS production also increased in direct proportion to the level of applied strain (Fig. 4C). In contrast, exposure of alveolar epithelial cells to 50-nm superparamagnetic iron nanoparticles under the same conditions only exhibited a small transient increase in ROS production (Fig. 4D).

reactive oxygen species (ROS)

When we introduced fluorescent nanoparticles (20 nm in diameter) into the alveolar microchannel and monitored nanoparticle translocation across the alveolar-capillary barrier in air-liquid interface culture by measuring the fraction of particles retrieved from the underlying microvascular channel by continuous fluid flow (Fig. 4F), we observed only a low level of nanoparticle absorption under static conditions.

Because cytokines are produced by cells of the lung parenchyma, we simulated this process by introducing medium containing the potent proinflammatory mediator, tumor necrosis factor–α (TNF-α), into the alveolar microchannel in the presence of physiological mechanical strain and examined activation of the underlying microvascular endothelium by measuring ICAM-1 expression. TNF-α stimulation of the epithelium substantially increased endothelial expression of ICAM-1 within 5 hours after addition (Fig. 3A and fig. S3), whereas physiological cyclic strain had no effect on ICAM-1 expression. The activated endothelium also promoted firm adhesion of fluorescently labeled human neutrophils flowing in the vascular microchannel, which do not adhere to the endothelium without TNF-α stimulation (Fig. 3B and movies S5 and S6).

The level of applied strain ranged from 5% to 15% to match normal levels of strain observed in alveoli within whole lung in vivo, as previously described (25). Vacuum application generated uniform, unidirectional mechanical strain across the channel length, as demonstrated by measured displacements of fluorescent quantum dots immobilized on the PDMS membrane (Fig. 2E and movie S2). Membrane stretching also resulted in cell shape distortion, as visualized by concomitant increases in the projected area and length of the adherent cells in the direction of applied tension (Fig. 2F and movie S3).

Application of physiological cyclic strain (10% at 0.2 Hz) also induced cell alignment in the endothelial cells in the lower compartment (fig. S2 and movie S4) and, hence, mimicked physiological responses previously observed in cultured endothelium and in living blood vessels in vivo (26, 27). Cyclic stretching caused some pulsatility in the fluid flow, but this unsteady effect was negligible due to the small channel size and low stretching frequency (see SOM text).

We mimicked this subatmospheric, pressure-driven stretching by incorporating two larger, lateral microchambers into the device design. When vacuum is applied to these chambers, it produces elastic deformation of the thin wall that separates the cell-containing microchannels from the side chambers; this causes stretching of the attached PDMS membrane and the adherent tissue layers (Fig. 1A, right versus left). When the vacuum is released, elastic recoil of PDMS causes the membrane and adherent cells to relax to their original size. This design replicates dynamic mechanical distortion of the alveolar-capillary interface caused by breathing movements.

We also demonstrated that this system could mimic the innate cellular response to pulmonary infection of bacterial origin. Living Escherichia coli bacteria constitutively expressing green fluorescent protein (GFP) were added to the alveolar microchannel. The presence of these pathogens on the apical surface of the alveolar epithelium for 5 hours was sufficient to activate the underlying endothelium, as indicated by capture of circulating neutrophils and their transmigration into the alveolar microchannel. Upon reaching the alveolar surface, the neutrophils displayed directional movement toward the bacteria, which they then engulfed over a period of a few minutes (Fig. 3E and movies S8 and S9), and the phagocytic activity of the neutrophils continued until most bacteria were cleared from the observation area.

Real-time, high-resolution, fluorescent microscopic visualization revealed that soon after adhering, the neutrophils flattened (fig. S3) and migrated over the apical surface of the endothelium until they found cell-cell junctions, where they underwent diapedesis and transmigrated across the capillary-alveolar barrier through the membrane pores over the period of several minutes (Fig. 3C and movie S7). Phase-contrast microscopic visualization on the opposite side of the membrane revealed neutrophils crawling up through the spaces between neighboring cells and emerging on the surface of the overlying alveolar epithelium (Fig. 3D), where they remained adherent despite active fluid flow and cyclic stretching.

Addition of air to the upper channel resulted in increased surfactant production by the epithelium (Fig. 2B and fig. S1), which stabilizes the thin liquid layer in vitro as it does in whole lung in vivo, such that no drying was observed. This was also accompanied by an increase in electrical resistance across the tissue layers (Fig. 2C) and enhanced molecular barrier function relative to cells cultured under liquid medium (Fig. 2D).

When human alveolar epithelial cells and microvascular endothelial cells were introduced into their respective channels, they attached to opposite surfaces of the ECM-coated membrane and formed intact monolayers composed of cells linked by continuous junctional complexes containing the epithelial and endothelial junctional proteins, occludin and vascular endothelial cadherin (VE-cadherin), respectively (Fig. 2A). These cells remained viable for prolonged periods (>2 weeks) after air was introduced into the epithelial microchannel and the alveolar cells were maintained at an air-liquid interface (fig. S1).

To mimic delivery of airborne nanoparticles into the lung using our microdevice, we injected nanoparticle solution into the alveolar microchannel and then gently aspirated the solution to leave a thin liquid layer containing nanoparticles covering the epithelial surface. When alveolar epithelial cells were exposed for 5 hours to 12-nm silica nanoparticles that are commonly used to model the toxic effects of ultrafine airborne particles (34–36), the underlying endothelium in the microvascular channel became activated and exhibited high levels of ICAM-1 expression (Fig. 4A).

Fabrication begins with alignment and permanent bonding of a 10-μm-thick porous PDMS membrane (containing 10-μm-wide pentagonal pores) and two PDMS layers containing recessed microchannels (Fig. 1C). A PDMS etching solution composed of tetrabutylammonium fluoride and N-methylpyrrolidinone is then pumped through the side channels (Fig. 1D). Within a few minutes, PDMS etchant completely dissolves away portions of the membrane in the side channels, creating two large chambers directly adjacent to the culture microchannels.

microvascular endothelium

parenchymal

spatiotemporal

cytokines