© 2013 Nature America, Inc. All rights reserved.

Microfabrication of human organs-on-chips

By Dongeun Huh, Hyun Jung Kim, Jacob P Fraser, Daniel E Shea, Mohammed Khan, Anthony Bahinski,

Geraldine A Hamilton, & Donald E Ingber

Nature ProtocolsVol.8 No.11, 2013

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2135

INTRODUCTIONSurrogate models that can reproduce the complex structure and functionality of living human organs are indispensable for understanding diverse biological responses of the human body for a variety of biomedical, pharmaceutical, chemical and envi-ronmental applications. Conventional models used in these types of studies often use living cells cultured in 2D monolayers, 3D extracellular matrix (ECM) gels or multicellular spheroids1. Although these in vitro models provide controllable environments to probe biological processes at the cellular or tissue level, they lack the ability to replicate organ-specific structural organization of their constituent tissue types in three dimensions or to fully recapitulate integrated physiological functions at the organ level. Consequently, animal testing has been the method of choice for simulating and predicting human responses to drugs, chemicals, pathogens and environmental toxins; however, animal studies are costly, lengthy and controversial, and their results often fail to pre-dict human responses. These problems raise serious economical, ethical and scientific issues in areas ranging from environmental monitoring and biomedical devices to the development of new therapeutics and cosmetics.

The emergence of organs-on-chipsThe disadvantages of existing in vitro models led to new alliances between cell biologists and bioengineers, who adapted micro-fabrication methods from the computer microchip industry to create microengineered systems that can be used to culture living cells. Early studies using this microsystems approach demonstrated that cell viability can be maintained in microchannels, and that with appropriate ECM coatings, culture medium and flow condi-tions, these cultured human cells can be induced to express and maintain tissue-specific differentiated functions in vitro. Examples include microfluidic culture of human cells derived from bone2,3, skin4, cartilage5, kidney6, liver7–9 and blood vessel10.

Recent progress in microengineering technologies, however, has made it possible to considerably improve these rudimentary models. This has led to the development of much more sophis-ticated microfabricated systems that allow one to recapitulate tissue-tissue interfaces, along with complex organ-specific chemi-cal and mechanical microenvironments, to mimic key 3D func-tional units of living human organs. These microengineered organ models, which are now collectively known as ‘organs-on-chips’1,11–16, contain hollow microchannels created in polymeric or glass microdevices that are lined by living cultured cells. Because microengineering methods enable precise control of feature size on the same scale as that in which living cells and tissues normally reside (nanometer to micrometer scale), it is possible to use this strategy to generate 3D microfabricated patterns and structures that induce cultured cells to mimic organ-specific microarchitec-ture in vitro. For example, by using this approach, it has been pos-sible to regenerate in vivo–like epithelial or endothelial tissues that stably express differentiated functions in microengineered mod-els of kidney17–19, liver20–24, brain25, heart26–31, skeletal muscle32 and intestine33–38. Similar methods have been used to integrate polarized epithelium with living vascular endothelium or stro-mal cell–containing connective tissue in 3D microfluidic devices that reproduce key tissue-tissue interfaces in healthy and diseased organs including the eye39, breast40–44 and brain40,41,44–48.

The greater promise of this biomimetic microsystems approach, however, lies in the possibility of accurately recreating the physical and biochemical microenvironments of the key compartments of living organs that are crucial for reconstituting organ-level func-tions. Microfluidic systems designed to generate complex concen-tration gradients have been integrated with cultured living cells to create microengineered in vitro models that mimic physiological and pathological gradients of nutrients, oxygen, growth factors and chemokines in the functional units of the liver49, lung50 and

Microfabrication of human organs-on-chipsDongeun Huh1,2, Hyun Jung Kim1, Jacob P Fraser1, Daniel E Shea1, Mohammed Khan1, Anthony Bahinski1, Geraldine A Hamilton1 & Donald E Ingber1,3,4

1Wyss Institute for Biologically Inspired Engineering at Harvard University, Boston, Massachusetts, USA. 2Department of Bioengineering, University of Pennsylvania, Philadelphia, Pennsylvania, USA. 3School of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts, USA. 4Vascular Biology Program, Departments of Pathology & Surgery, Children’s Hospital Boston and Harvard Medical School, Boston, Massachusetts, USA. Correspondence should be addressed to D.H. (huhd@seas.upenn.edu) or D.E.I. (don.ingber@wyss.harvard.edu).

Published online 10 October 2013; doi:10.1038/nprot.2013.137

‘Organs-on-chips’ are microengineered biomimetic systems containing microfluidic channels lined by living human cells, which replicate key functional units of living organs to reconstitute integrated human organ-level pathophysiology in vitro. These microdevices can be used to test efficacy and toxicity of drugs and chemicals, and to create in vitro models of human disease. Thus, they potentially represent low-cost alternatives to conventional animal models for pharmaceutical, chemical and environmental applications. Here we describe a protocol for the fabrication, microengineering and operation of these microfluidic organ-on-chip systems. First, microengineering is used to fabricate a multilayered microfluidic device that contains two parallel elastomeric microchannels separated by a thin porous flexible membrane, along with two full-height, hollow vacuum chambers on either side; this requires ~3.5 d to complete. To create a ‘breathing’ lung-on-a-chip that mimics the mechanically active alveolar-capillary interface of the living human lung, human alveolar epithelial cells and microvascular endothelial cells are cultured in the microdevice with physiological flow and cyclic suction applied to the side chambers to reproduce rhythmic breathing movements. We describe how this protocol can be easily adapted to develop other human organ chips, such as a gut-on-a-chip lined by human intestinal epithelial cells that experiences peristalsis-like motions and trickling fluid flow. Also, we discuss experimental techniques that can be used to analyze the cells in these organ-on-chip devices.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2136 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

breast51. Recent advances in microengineering have also enabled the development of mechanically active microsystems that can reproduce organ-specific fluid flow and mechanical deformation to model pathophysiology of living organs such as the lung52,53 and kidney17. These studies leverage the power of microfabrica-tion to develop micromechanical actuators and precise hydrody-namic control as well as microfluidic channels that place cells of different tissues in their relevant 3D physical context. However, the challenge for the field has been to develop microsystems that reconstitute complex functionality of living organs by fully inte-grating multiple tissues, recreating crucial tissue-tissue interfaces characteristic of each organ’s specific 3D microarchitecture and reconstituting their physiological mechanical and biochemical microenvironments.

To provide the proof of principle for this type of integrated microdevice, we developed a microengineered ‘organ-on-a-chip’ model that replicates one of the major functional units of the living human lung: the alveolar-capillary interface and its dynamic mechanical activities including cyclic breathing and blood flow54 (see below for design details). The lung-on-a-chip provides unprecedented capabilities to reconstitute, directly observe and quantitatively analyze organ-level physiological functions such as gas exchange and immune responses to cytokines, bacteria and nanoparticulates. This biomimetic microsystem also has been used to develop a clinically relevant human disease model of chemotherapy-induced pulmonary edema that can be used to identify new therapeutics and reliably predict their in vivo efficacy and toxicity55.

This organ-on-a-chip strategy can be translated to mimic key functional units of other living organs. For example, a similar design and fabrication approach has been used to create a micro-engineered model of the human intestine. In this microsystem, called a human ‘peristalsing’ gut-on-a-chip, human intestinal epi-thelial cells can be cocultured with intestinal microbes under the influence of physiological peristaltic motions and fluid flow (see below for design details) to recapitulate intestinal microenviron-ment, which leads to cytodifferentiation, enhanced barrier func-tion and spontaneous intestinal 3D villus morphogenesis56,57. This biomimetic microengineering approach also opens up the possibility of integrating these individual organ-on-chip models in a single instrument to recapitulate multiorgan interactions and whole-body physiology. This concept has been demonstrated by the development of microsystems termed ‘micro–cell culture ana-logs’, or ‘body-on-a-chip’, that fluidically integrate multiple micro-fabricated cell culture chambers representing different organs in a physiological manner to reproduce and study the whole-body response to drugs and toxins13,16. These studies clearly demon-strate that organ-on-chip microdevices offer remarkable advan-tages over conventional cell culture models and might serve as promising alternatives to animal models for fundamental bio-medical research as well as practical applications such as testing of drug efficacy and safety.

In this protocol, we describe how to design and fabricate organ-on-chip microdevices as developed by our group, as well as gen-eral guidelines for operation and analysis of these systems. In the PROCEDURE we describe how to make a biomimetic 3D microfluidic system for the coculture of human alveolar epithe-lial cells and pulmonary microvascular endothelial cells to create the human ‘breathing’ lung-on-a-chip. In Box 1 we detail how

this protocol can be modified to produce other organ chips, such as a gut-on-a-chip microdevice that experiences peristalsis-like mechanical motions and physiological fluid flow. Our detailed protocols also include procedures for seeding, growing and dif-ferentiating human cells in these microdevices as well as proce-dures to analyze cell viability and differentiated functions during prolonged culture.

Experimental design Organs in the human body are complex living systems com-posed of different types of tissues that form complex tissue-tissue interfaces, such as between endothelium-lined blood vessels and parenchymal epithelial cells that exhibit organ-specific functions. Most organs are multimodular structures in that they consist of repeating smaller functional units that individually perform the major characteristic functions of the whole organ (e.g., gas exchange in the alveoli of the lung, absorption in the villi of the gut, metabolism in the hepatic triad of the liver, etc.). Typically, these functional units comprise different types of specialized tissues (e.g., epithelium, vascular endothelium, connective tis-sue, immune cells, nerves, etc.) that interface in organ-specific patterns and are subjected to dynamic changes in chemical and mechanical signals that vary depending on their particular spatial microenvironment.

To develop a useful organ surrogate device for in vitro analysis of complex human physiology, it is necessary to both reproduce normal tissue-tissue interfaces and mimic this complex physi-cal microenvironment in which cells are normally situated. We recreated the critical alveolar-capillary interface of the lung air sac in our human breathing lung-on-a-chip54 by developing a 3D microfluidic device that contains two parallel microchannels with the same dimensions (400 m wide × 100 m high) separated by a thin porous flexible membrane made of poly(dimethylsiloxane) (PDMS) ECM-coated (Fig. 1a). We chose the width of micro-channels to approximate the average diameter of an alveolus in human lungs58. These channels also are surrounded on either side by two hollow full-height chambers that permit application of cyclic suction to mechanically stretch and relax the flexible PDMS membrane in the central channel. This design was inspired by the mechanism of physiological breathing in living human lungs in which sub-atmospheric pressure in the intrapleural space gener-ated by inspiration causes the inflation of the alveolar air sacs, which induces filling of the lungs with air and stretching of the alveolar epithelium and the juxtaposed vascular endothelium in the surrounding capillaries54. Human alveolar epithelial cells and pulmonary microvascular endothelial cells are then introduced into the upper and lower microchannels, respectively, and grown on the ECM-coated membrane for at least 5 d to form two closely apposed tissue monolayers.

During cell culture, medium is perfused continuously through both channels to provide cells with nutrients and to remove their metabolic wastes. Once confluence is reached, the culture medium is removed from the upper channel, and this channel is filled with air to form an air-liquid interface at the apical surface of the alveolar epithelium, which induces the cells to differenti-ate and express tissue-specific functions, such as surfactant pro-duction. During this period, medium flow (1–15 dynes cm−2) is maintained in the lower channel to recapitulate the dynamic flow that normally perfuses the pulmonary capillary blood vessels

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2137

Box 1 | Fabrication of the gut-on-a-chip TIMING 8.5–18.5 dFirst, fabricate the SU-8 silicon master and prepare PDMS as described in Steps 1–23 of the PROCEDURE before proceeding to the following steps specific to gut-on-a-chip.

Fabrication of the upper and lower microchannels TIMING 5.5 h1. Perform Step 24 of the PROCEDURE to clean the silicon master.2. Pour 3 g of 15:1 (wt/wt) PDMS mixture onto the bottom surface of a Petri dish, and spread it uniformly.3. Place the cleaned master on the spread PDMS with SU-8 features facing up and wait for 10 min on a level surface.4. Pour 15:1 (wt/wt) PDMS mixture onto the silicon master and degas it in a vacuum desiccator for 30 min.

CRITICAL STEP Use 15 g and 3 g of PDMS for the fabrication of the upper and lower microchannel slabs, respectively.5. Fully cure PDMS in an oven maintained at 60 °C for at least 4 h.6. Peel the cured PDMS off of the master, and cut it into a 2-cm-wide × 3-cm-long rectangular block using a scalpel.

CRITICAL STEP Be careful when you are using a scalpel, in order to avoid injuries and fracture of fragile silicon masters.7. Punch holes through the upper channel layer by using a biopsy punch with a diameter of 2 mm.

Fabrication of porous PDMS membranes TIMING 1 d8. Pour degassed 15:1 (wt/wt) PDMS mixture into an empty Petri dish to generate a 1-cm-thick flat PDMS slab.9. Cure PDMS in a leveled dry oven at 60 °C for 4 h.10. Use a scalpel to cut the fully cured PDMS along the edge of the Petri dish, and remove it from the dish by using tweezers.11. Rinse the removed PDMS slab with 100% ethanol, and dry it completely using compressed nitrogen or air.12. Treat the PDMS block with oxygen plasma for 1.5 min, and silanize it in a vacuum desiccator overnight according to PROCEDURE Steps 52–54.13. Place a silicon master containing an array of microfabricated circular pillars (10 m in diameter and 30 m in height) with a center-to-center spacing of 25 m at the center of the bottom surface of a clean Petri dish.14. Pour 3 g of degassed 15:1 (wt/wt) PDMS mixture onto the wafer, and spread it evenly.

CRITICAL STEP Avoid the formation of air bubbles.15. Gently put a silanized PDMS slab from step 12 (in this box) on the surface of the silicon master covered with uncured PDMS.

CRITICAL STEP Release the slab very slowly to prevent formation of air bubbles.16. Place a frosted glass slide on the silanized PDMS block, and place 3 kg weight on the glass slide.17. Wait for 30 min to allow for intimate contact between the PDMS slab and microfabricated master surface.

CRITICAL STEP Perform this step on a level surface.18. Move the entire assembly to a dry oven at 60 °C, and incubate it overnight.19. Remove the sample from the oven, take off the weight, and cool the assembly to room temperature over 30 min.20. Use a scalpel to lift up a corner of the slab, and slowly peel it from the wafer.

CRITICAL STEP Apply 100% ethanol to the gap between the PDMS surface and the silicon wafer during this step to facilitate the detachment of the PDMS layer.

Alignment and assembly of the microdevice TIMING 1.5 d21. Clean the upper PDMS layer (Fig. 5a) and the porous PDMS membrane with packaging tape.

CRITICAL STEP Do not apply excessive pressure when you are cleaning the membrane surface, in order to avoid unwanted damage of the device and delamination.22. Treat the membrane surface and the channel side of the upper PDMS layer with plasma by using a corona generator for 3 s and 1 min, respectively (Fig. 5b).

CRITICAL STEP Use sweeping motions to achieve uniform treatment, and keep the tip of the electrode of the corona generator ~5 mm away from the sample surface for best results.23. Overlay the upper microchannel layer on the PDMS membrane, and bring them in contact.

CRITICAL STEP Press the PDMS slabs to permit intimate contact between layers and to remove trapped air.24. Incubate the assembled layers in a dry oven at 80 °C for at least 12 h.25. Remove the sample from the oven, and cool it down to room temperature for 1 h.26. Cut along the edges of the upper PDMS channel layer bonded to the membrane by using a scalpel, and gently peel the silanized flat PDMS slab from the assembly.

CRITICAL STEP Put a few drops of 100% ethanol between the layers for easier detachment.27. Tear off the portions of a porous membrane located over the lateral vacuum chambers using fine-tip tweezers under a stereoscope (Fig. 5c).28. Expose the membrane surface and the channel side of the lower PDMS layer to the corona for 1 min (Fig. 5d).29. Perform Steps 84 and 86 from the PROCEDURE to align and bond the device (Fig. 5e).

CRITICAL STEP After completion of this step, attempt to pull apart the upper and lower PDMS slabs to qualitatively determine the success of device bonding. Incomplete or unsuccessful bonding results in peeling off and separation of the PDMS slabs.30. Bend the tips of six 18-gauge blunt needles by 90° by using pliers, and break off the needle at a point near the syringe entry port.31. Insert the needles into the access ports of the central cell culture channels and side vacuum chambers.32. Cut six pieces of 4-cm-long silicone tubing, and connect them to the free ends of the needles.

(continued)

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2138 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

and supplies oxygen and nutrients to the epithelium through the intervening endothelium.

To mimic physiological breathing motions of the lung, vacuum is applied to the side hollow chambers to induce outward bending of the elastic vertical walls of the central microchannels (Fig. 1b). This deformation laterally stretches the intervening PDMS membrane along with the attached epithelial and endothelial

monolayers. When the vacuum is removed, the elastic recoil of PDMS causes the membrane to return to its original length, thus relaxing the tissue layers. By repeating this actuation cycle at a frequency of 0.25 Hz and controlling the level of stretching to produce ~10% cyclic strain, it becomes possible to generate dynamic mechanical distortions similar to those observed in the air sac of the living human lung.

Box 1 | (continued)33. Perform Steps 100–103 of the PROCEDURE to complete the fabrication of the device (Fig. 5f).

CRITICAL STEP To examine the quality and performance of the device, connect the device to the vacuum pump and test vacuum-assisted membrane stretching after the completion of this step. Leakage of vacuum from the side chamber owing to unsuccessful PDMS bonding results in little or no stretching of the membrane. Alternatively, fill the central cell culture channels with deionized water, and check for the leakage of water into the side chambers when vacuum is applied. Incomplete removal of the membrane in the side vacuum channels leads to varying degrees of membrane stretching along the channel length. These undesirable events can be visually detected by using a microscope.34. For sterilization, flow 70% ethanol through the channels, dry the device in an oven at 60 °C for 2 h and perform Steps 104 and 105 of the PROCEDURE.

Microfluidic cell culture: 5–15 d35. Introduce 500 l of the ECM coating solution (see Reagent Setup) into both the upper and lower central microfluidic channels, and pinch off tubing to prevent coating solution from leaking out of the device.

CRITICAL STEP Ensure that the entire microchannels are filled with the ECM solution without any trapped air bubbles. ? TROUBLESHOOTING36. Place the microdevice in a humidified 37 °C incubator for over 2 h.37. Flow culture medium through the channels to remove residual coating solution flow culture medium at 30 l h−1 overnight without mechanical stretching.38. Aspirate the culture medium from a T75 flask containing ~90% confluent Caco-2 cells, and wash the cells with Ca2+- and Mg2+-free PBS twice.39. Incubate the cells with 1 ml of prewarmed trypsin/EDTA solution (0.05% (wt/vol)) in a humidified incubator (37 °C, 5% CO2) for 10 min.40. Resuspend the trypsinized cells with 10 ml of culture medium containing 20% (vol/vol) FBS and antibiotics, transfer the sample to a 15-ml sterile conical tube, and centrifuge the cells at 500g for 5 min.41. Remove the supernatant and resuspend the cells with culture medium to yield a seeding density of ~5 × 105 cells cm−2.42. Load a 1-ml syringe with the cell suspension solution and attach a 25G 5/8 needle to the syringe.43. Inject the cells into the microchannel through the tubing connected to the outlet of the upper central microchannel. During this step, the inlet and outlet of the lower microchannel remain closed. After seeding, clamp both ends of the upper channel.

CRITICAL STEP Injection of the cell suspension solution should be performed very slowly and carefully in order to prevent introduction of air bubbles. Make sure that the needle attached to the syringe containing cell suspension is tightly fit into the outlet tubing.44. Incubate the device in a humidified incubator for ~1.5 h to allow the seeded cells to adhere on the ECM-coated membrane surface.45. After cell attachment, gently aspirate the culture medium from the upper and lower microchannels to remove unbound cells and cell debris.46. Connect the inlet tubing to a syringe pump, and perfuse the upper microchannel at 30 l h−1 in a humidified incubator for 24–36 h to allow the cells to grow to confluence. Close the tubing connected to the lower microchannel by using clamps during this step.47. Once the cells form a confluent monolayer, flow the culture medium in both the upper and lower microchannels at 30 l h−1.48. For coculture with microbes, switch the cell culture medium to antibiotic-free DMEM, and perfuse the device for at least 12 h before seeding the microbial cells. During this perfusion, proceed with the next step.49. Grow the microbial cells (e.g., Lactobacillus rhamnosus GG) in autoclaved MRS broth medium without shaking in a humidified incu-bator (37 °C, 5% CO2) for 12 h.! CAUTION Use a separate incubator for microbial cultures to prevent cross-contamination.50. Take 1 ml of microbial culture broth and spin it down at 12,000g at room temperature for 5 min.51. Aspirate the supernatant and add antibiotic-free culture medium to obtain a final cell density of ~1.0 × 107 c.f.u. ml−1. ? TROUBLESHOOTING52. In a microbial culture hood, seed the microbial cells into the upper microchannel by using the method described above, and incubate the mixture in a humidified incubator for ~1.5 h to allow the microbes to attach to the apical surface of the intestinal epithelial cells.! CAUTION Use a separate incubator for the coculture devices in order to prevent cross-contamination.53. Open the outlet tubing connected to microchannels, remove unbound microbial cells by aspirating culture medium, and perfuse both the upper and lower channels with antibiotic-free culture medium at 40 l h−1.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2139

Applications of the protocolThe ability of the lung-on-a-chip microdevice to recapitulate the key structural and microenvironmental features of the living human lung induces the cultured endothelium and epithelium to express complex integrated organ-level physiological func-tions not normally observed in conventional in vitro cultures. For example, this microdevice reproduced immune responses to bacteria and inflammatory cytokines introduced into the alveolar space by stimulating the expression of intercellular adhesion mol-ecule-1 on the microvascular endothelium surface, thus increasing the adhesion of circulating neutrophils and their transmigration across the capillary-alveolar interface and inducing phagocyto-sis of the infectious pathogens, all of which can be visualized in real time by using high-resolution phase-contrast and fluores-cence-microscopy imaging54. By using this approach, we devel-oped nanotoxicology models that revealed previously unknown detrimental effects of physiological breathing on nanoparticle toxicities and extrapulmonary absorption54, which could never have been identified using static culture systems.

In a separate study, the same lung-on-a-chip device was used to create a human disease model of pulmonary edema induced by drug toxicities, and this also provided important new mechanistic insights into progression of disease, in addition to enabling the identification and evaluation of new therapeutics55. Our detailed protocols to produce this device and to use it to analyze human lung pathophysiology in vitro are described below.

The same microdevice design can be modified easily to develop and analyze other microengineered human organ models. For example, this lung-on-a-chip design was adapted to build a human gut-on-a-chip56 by increasing the height (from 100 m to 150 m) and width (400 m to 1 mm) of the central channels to produce a trickling-like flow (0.02 dyne cm−2) and to exert differ-ent force regimens that more closely mimic peristalsis-like motions (10% strain at 0.15 Hz). Human intestinal epithelial Caco-2 cells cultured in this microdevice exhibited a highly polarized columnar epithelial morphology within 48 h, and then they spontaneously grew into folds that recapitulate the structure and function of intestinal villi, including basal proliferative cells in the crypt with all four cell types of the small intestinal epithelium as well as increased mucus production and enhanced intestinal barrier functions. In addition, normal intestinal microbes (Lactobacillus rhamnosus GG) can be cocultured in this device for extended periods (over 1 week) on the luminal surface of the cultured epi-thelium without compromising epithelial cell viability; in fact, the gut epithelium shows increased with this probiotic bacterium cultured on its surface. We describe our protocols to create and analyze this microengineered gut model below. These studies dem-onstrate our ability to modify experimental design as needed and

to quantitatively analyze the functional behavior of the cells in a physiologically healthy state as well as in response to insults or to the introduction of drugs into organ-on-chip microdevices.

Because the use of this multilayered microchannel approach allows one to introduce and control microenvironmental cues (e.g., surface coating and fluid flow) in the upper and lower culture chambers separately, this method can be used to model key functional units of any other organ that similarly contains a polarized tissue-tissue interface. For example, this design can be used with human proximal tubular or glomerular epithelial cells and kidney endothelium to create a kidney tubule-on-a-chip18,19 and a kidney glomerulus-on-a-chip, respectively. Similarly, hepa-tocytes and endothelium can be combined to produce a liver-on-a-chip, astrocytes and brain microvascular endothelium can be combined to generate a blood brain barrier-on-a-chip48, and so on. The application of this approach could be extended to virtually all organs that contain a vasculature interfaced with a parenchymal cell population (e.g., cartilage, bone, etc.), as well as other barrier tissues that contain two different types of cells in close apposition.

The ability to establish an air-liquid interface in the upper channel also is advantageous for inducing the differentiation of pulmonary bronchiolar epithelium in a small airway-on-a-chip53 or to produce a skin-on-a-chip with keratinocytes. In addition, the microchannel design permits sampling of fluids and cellular products in a polarized manner, which can be leveraged to sys-tematically examine polarized epithelial barrier functions, direc-tional molecular transport or absorption and dynamic changes in molecular behavior, which is critical for the analysis of drug phar-macodynamics. Although this type of analysis has previously been demonstrated in more simplified tissue models17–19,34,35, these systems did not incorporate organ-specific endothelium that is normally juxtaposed to the epithelium and that has an important role in regulating molecular transport across this polarized tis-sue-tissue interface in vivo.

The capability to independently regulate fluid flow in the upper and lower channels also provides a means to deliver physiologi-cally relevant cells, pathogens, toxins, cytokines and drugs to one or both tissues. This is particularly advantageous for developing models of disease or for assessing efficacy and toxicity of drugs. The newest feature of the organ-on-chip devices described here is that they recapitulate the dynamic mechanical microenviron-ment that cells are normally exposed to in vivo. This is important because these physiological forces appear to have a critical role in

Vacuum

Stretch

Upperchannel

Lowerchannel

Porousmembrane

Sidechamber

Sidechamber

VacuumEndothelium

Epithelium

a

b

Figure 1 | Mechanically active organ-on-chip microdevice with compartmentalized 3D microarchitecture. (a) The human lung-on-a-chip microsystem is constructed in a multilayered microfluidic device comprising the upper (blue) and lower (red) cell culture microchannels with a microfabricated porous elastic membrane sandwiched in-between. The microdevice is also equipped with two full-height, hollow microchambers alongside of the cell culture channels. (b) Physiological breathing motions in the living human lung are reproduced by the application of vacuum to the side chambers. This actuation causes the lateral elongation of the intervening elastic membrane, which induces mechanical stretching of the adherent tissue layers in the central channels.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2140 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

reconstituting complex organ-specific structures, differentiated functions and disease responses in vitro. A simple example is the finding that human Caco-2 intestinal epithelial cells only show the formation of villus undulations and folds reminiscent of intes-tinal villi56, as well as the expression of differentiated functions (e.g., mucus production and elevated CYP450 enzyme activities)57, when exposed to peristalsis-like mechanical motions and physio-logical fluid flows in the gut-on-a-chip microdevice. Similarly, the toxic effects of airborne nanoparticulates, their absorption across the alveolar-capillary interface and the ability of interleukin-2 to induce pulmonary edema were all found to be highly dependent on the presence of physiological breathing motions in the lung-on-a-chip. These responses could never be identified or studied in conventional static cultures such as Transwell chambers, and this is a major advantage of our organ-on-chip devices. In addi-tion, the vacuum-assisted microactuation scheme can potentially be used to mimic pathological mechanical forces and to produce injury responses to develop microengineered models of other dis-eases, such as ventilator-induced lung injury, that are currently challenging to study using existing in vitro models.

Limitations of the protocolAlthough our organ-on-chip microdevices are a robust platform for mimicking the key structures and functions of various living human organs, they also have limitations. First, the PDMS mem-branes with microfabricated pores used in our models differ in thickness and composition from their in vivo counterparts. For example, the 10- m-thick membrane used in our lung-on-a-chip is considerably thicker than the interstitium between the alveolar epithelium and capillary endothelium in vivo, which is usually thinner than 1 m. This substantial difference is due to technical challenges associated with fabricating ultrathin (<10 m) flex-ible porous PDMS membranes while maintaining their struc-tural integrity and ease of handling. Although our lung model effectively recapitulates transport of nanoparticles, proteins and fluids across the alveolar-capillary barrier in vivo54, there are still some quantitative discrepancies in the actual rates of transport. These problems could potentially be addressed by replacing the PDMS membranes with thinner membranes formed from natu-ral ECM molecules39 or ultrathin porous membranes generated by standard nanofabrication techniques48. However, as the large membrane pores used in our system allow large portions of cells of each cell layer to form direct physical contact across them, the thickness of the barrier in these regions is essentially only limited to that of the basement membrane deposited by the cells them-selves, which might be more physiologically relevant than using rigid nanoporous membranes.

Another potential limitation is that the microengineered organ models described here only have a subset of the cell types of the whole living organ. For example, the lung-on-a-chip lacks con-nective tissue containing fibroblasts between the epithelium and endothelium, which can have an important role in organ home-ostasis and pathogenesis. However, one of the greatest strengths of our organ-on-chip microengineering approach is that we start with as simple a model as possible and use reconstitution of organ-level functionality as our measure of success. For example, our finding that we can induce pulmonary edema using inter-leukin-2 in the lung-on-a-chip without immune cells revealed a new insight into the mechanism of action underlying this

toxicity55. At the same time, if we fail to obtain an in vivo–like response using a simplified system, we can then add additional cell types until we identify the correct combination of cells necessary to achieve the functionality of interest. For example, connective tissue can be readily integrated into the lung-on-a-chip by plating fibroblasts on the membrane (e.g., in a thin collagen gel) before seeding the epithelial cells. Because we already have introduced neutrophils into the lung model by perfusing them through the vascular channel, we could incorporate other immune cells (e.g., monocytes, lymphocytes and dendritic cells) in a similar manner or by plating them before forming the epithelial layer. Pericytes or vascular smooth muscle cells could be similarly integrated under the endothelium.

Irreversible permanent bonding between PDMS layers, which is required to construct organ chips described here, makes it tech-nically challenging to access, isolate and process intact human tissues produced in the microdevice for certain types of analysis, such as histology and electron microscopy. For this reason, we rely heavily on fluorescence microscopy, microfluorimetry and real-time imaging for analysis of our organ models fabricated in PDMS, but it should be possible to slice the whole device to obtain tissue sections using heavy-duty microtomes (e.g., those designed for whole-animal sectioning or materials research). Despite the closed-channel configuration, our organ-on-chip microdevices still allow one to analyze various tissue-specific functions. For example, barrier function of tissue-tissue interfaces created in our microsystems can be quantitatively analyzed by measuring electrical resistance or transport of fluorescent proteins and nanoparticles across the tissue layers54–56. In addition to analyz-ing changes in biochemical activities (e.g., production of reac-tive oxygen species) in situ in the chip using microfluorimetry54, it is also possible to collect secretory products from living cells cultured in sealed microdevices for the analysis of differentiated tissue function7,20,53, as well as to isolate tissues and cells the from microchannels to examine their morphology, and gene and pro-tein expression49.

The small size of cell culture chambers used in our models minimizes consumption of reagents and culture medium, and also provides the opportunity to achieve high-throughput and high-resolution analysis. On the other hand, this can be prob-lematic in the sampling and detection of secreted cellular prod-ucts owing to the small number of cells used. For example, our lung-on-a-chip device, whose channel width matches the average diameter of an alveolus in vivo (400 m), only contains ~10,000 cells. Continuous perfusion with culture medium required to maintain cell viability can exacerbate this problem by diluting samples collected from these microfabricated devices. Therefore, it is important to carefully consider channel dimensions and per-fusion rates in designing and operating organ-on-chip microsys-tems to meet the minimum requirements for detection and analysis of cellular products. Potential solutions can be found by lengthening cell culture chambers and/or decreasing the flow rate of culture medium while increasing the number of microchan-nels operating in parallel to obtain samples with sufficiently high concentrations and volumes. Previous studies have demonstrated the feasibility of using these approaches to detect and quantita-tively analyze various soluble factors produced by cells cultured in microfabricated tissue and organ models using conventional assay systems20,49,53.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2141

Another challenge is obtaining human organ–specific cells with both proliferative capacity and full differentiation capa-bility. Although the ultimate goal is to obtain primary human cells or induced pluripotent stem cells that retain the ability to differentiate into mature cells, we have been very successful at using established cell lines in our studies. For example, although we used Caco-2 intestinal epithelial cells that had been originally isolated from a human colon cancer, these cells were capable of differentiating and forming well-organized intestinal villi with basal proliferative crypts and multiple differentiated cell lineages in correct locations when they were provided with physiological microenvironment, in this case, with trickling flow and peri-stalsis-like motion57. Similarly, although we used human lung epithelial cell lines that had been derived from tumor specimens, these cells also increased the expression of surfactant when exposed to an air-liquid interface in the device54. These results suggest that established human cell lines, even isolated from human tumor tissues, can be used in organs-on-chips to replicate key physiological functions and disease processes by subject-ing those cells to physiologically relevant microenvironmental cues that induce them to re-express highly differentiated organ- specific structures and functions. Established cell lines also have the advantage of increased robustness and easy availability. However, induced pluripotent stem cells are potentially excit-ing because they can easily be genetically modified and provide

a way to obtain cells from specific genetic subpopulations (e.g., patients who exhibit specific disease processes or propen-sities). In the end, these limitations are a function of the cells and not of the devices, as any cell can be incorporated into the organ-on-chip model system.

Finally, PDMS used in the construction of our models may pose technical challenges when testing chemicals and drugs because of its poor chemical resistance to certain solvents. Absorption of small hydrophobic molecules by PDMS could compromise the accuracy of measuring drug efficacy and toxicity in these systems, depending on the physicochemical properties of the molecule being tested59–61. This potential variability may limit the utility of this material for fabricating organs-on-chips to be used for drug discovery and development applications. Although PDMS is an excellent material for rapid prototyping of microfluidic devices, it also might not be ideal for large-scale manufacturing and com-mercialization of these organomimetic microdevices. Therefore, it is crucial for the successful application of this technology beyond proof of concept to find alternative materials that retain the desir-able properties of PDMS yet do not exhibit chemical absorption issues and are amenable for the scale-up and commercial manu-facturing of organs-on-chips for pharmaceutical applications. We and others have recently identified promising alternatives that do not absorb small molecules and hence can potentially be used to create organ-on-chip microdevices62,63.

MATERIALSREAGENTS

Photoresist SU-8 2100 (Microchem) ! CAUTION Wear safety glasses, gloves and protective clothing when you are handling this material. Adequate ventilation is highly recommended in order to avoid breathing the vapors or mist.Photoresist developer (PGMEA; Microchem) ! CAUTION Wear safety glasses, gloves and protective clothing when you are handling this material. Adequate ventilation is highly recommended in order to avoid breathing the vapors or mist.Poly(dimethylsiloxane) (PDMS; Sylgard 184 Silicone Elastomer Kit; Dow Corning, cat. no. 3097358-1004)Tetrabutylammonium fluoride (TBAF; for PDMS etching, 75% (wt/wt) solution in water; Sigma-Aldrich, cat. no. 361399) ! CAUTION Wear safety glasses, impervious clothing and face protection. Handle the compound with gloves. This chemical causes severe skin burns and eye damage.1-Methyl-2-pyrrolidinone (MP; for PDMS etching, 99+% A.C.S. reagent; Sigma-Aldrich, cat. no. 443778) ! CAUTION Wear safety glasses and impervious clothing. Handle the compound with gloves. This chemical causes skin and eye irritation.Trichloro(1H,1H,2H,2H-perfluorooctyl)silane (Sigma-Aldrich, cat. no. 448931) ! CAUTION Wear safety glasses, gloves, face protection and a respirator if necessary when you are working with this material. Adequate ventilation is highly recommended in order to avoid breathing the vapors or mist.NCI-H441 human alveolar epithelial cell line (American Type Culture Collection, cat. no. HTB-174) CRITICAL This cell line is needed for lung-on-chip only.EGM-2MV human lung microvascular endothelial cells (Lonza, cat. no. CC-2527) CRITICAL These cells are needed for lung-on-chip only.Caco-2BBE human colorectal carcinoma line (Harvard Digestive Disease Center) CRITICAL This carcinoma line is needed for gut-on-chip only.Lactobacillus rhamnosus GG (American Type Culture Collection, cat. no. ATCC-53103) CRITICAL L. rhamnosus GG is needed for gut-on-chip only.RPMI-1640 medium (for culture of NCI-H441; American Type Culture Collection, cat. no. 30-2001) CRITICAL This medium is needed for lung-on-a-chip only.

•

•

•

•

•

•

•

•

•

•

•

EGM-2-MV medium BulletKit (for culture of human lung microvascular endothelial cells; Lonza, cat. no. CC-3202) CRITICAL This is needed for lung-on-a-chip only.DMEM containing 25 mM glucose and 25 mM HEPES (Gibco, cat. no. 10564-011) CRITICAL This medium is needed for gut-on- a-chip only.Difco Lactobacilli MRS broth (BD, cat. no. 288120) CRITICAL This broth is needed for gut-on-a-chip only.Ethanol (100%, 200 proof; KOPTEC, cat no. V1016)Ethanol (70% (vol/vol); VWR, cat. no. BDH1164-4LP)Dulbecco’s PBS (D-PBS; pH 7.4, Ca2+- and Ma2+-free; Gibco, cat. no. 14190-144)HBSS (Gibco, cat. no. 14025-092)Heat-inactivated FBS (Gibco, cat. no. 10082-147)Trypsin/EDTA solution (0.05% (wt/vol); Gibco, cat. no. 25300-054)Penicillin-streptomycin-glutamine (100×; Gibco, cat. no. 10378-016)Human fibronectin (BD Biosciences, cat. no. 356008)Matrigel (BD Biosciences, cat. no. 354234)Dexamethasone (water soluble and cell culture tested; Sigma, cat. no. D2915)CellTracker Green CMFDA (5-chloromethylfluorescein diacetate) (Invitrogen, cat. no. C2925)CellTracker Red CMTPX (Invitrogen, cat. no. C34552)Paraformaldehyde (4% (wt/vol); Electron Microscopy Science, cat. no. 15710)4 ,6-Diamidino-2-phenylindole dihydrochloride (DAPI; Molecular Probes, cat. no. D1306)BSA (2% (wt/vol); Sigma, cat. no. A7030)Triton X-100 (0.3% (vol/vol); Sigma, cat. no. T8787)Anti-occludin antibody conjugated with Alexa Fluor 594 (mouse monoclonal, isotype: mouse IgG1 ; Molecular Probe, cat. no. 331594)Anti-VE-cadherin antibody (rabbit polyclonal, isotype: rabbit IgG; Cell Signaling Technology, cat. no. D87F2)Alexa 488–conjugated goat anti-rabbit antibody (Invitrogen, cat. no. A11070)Phalloidin-CF647 conjugate (25 units ml−1; Biotium, cat. no. 00041)

•

•

•

•••

•••••••

•

••

•

•••

•

•

•

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2142 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

FITC-inulin (for permeability assay; Sigma, cat. no. F3272)Carbonate bufferType I collagen (Gibco, cat. no. A10483-01)Acetone (Sigma, cat. no. 179124)Methanol (Sigma, cat. no. 179337)Isopropyl alcohol (Sigma, cat. no. W292907)

EQUIPMENTSpin coater (Specialty Coating systems, Spincoat G3P-12)Desiccator (Space Saver Vacuum desiccator; Bel-Art, cat. no. 4201000000)Planetary centrifugal mixer (Thinky, Thinky Mixer ARE-310): handheld cooking mixers can be used as alternativesLaboratory corona treater (Electro-Technic Products, BD-20AC)Oxygen plasma etcher (SPI Supplies and Plasma Etch PE-100, Plasma Prep II)UV-ozone cleaner (Jelight Company, UVO Cleaner Unit 342): UV lamps installed in biological safety cabinets can be used as alternativesFlexcell FX-5000 tension system with a computer-controlled vacuum pump (Flexcell International Corporation, cat. no. FX5K)Dual chamber dry oven (FinePCR, cat. no. combi-D24)Syringe pumps (Braintree Scientific, cat. no. BS-8000)Inverted epifluorescence microscope (Zeiss Axio Observer Z1)Inverted microscope (Zeiss Axiovert 40CFL)Stereo microscope (Zeiss Discovery V20 stereo microscope)Millicell ERS meter (Millipore, cat. no. MERS00002)Microplate reader (SpectraMax M5; Molecular Devices)Volt-ohm meter (87 V, industrial multimeter; Fluke Corporation)Silicon wafers (3 inches in diameter; University Wafer, cat. no. 1080)Disposable scalpels (stainless steel blade with plastic handle; Feather, no. 11)Tweezers (SPI Supplies, cat. no. 2WFG.SA)Fine-tip precision tweezers (Aven, OOSA technik)Hole punchers (tip inner diameter = 2 mm; Harris Uni-core; Ted Pella, cat. no. 15076)Hole punchers (tip inner diameter = 1.5 mm; Harris Uni-core; Ted Pella, cat. no. 15075)Blunt needles (18 gauge; VWR, cat. no. KT868280)Silicone tubing (Saint-Gobain Tygon sanitary tubing, inner diameter = 1/32 inches; Fisher Scientific, cat. no. 02-587-1a)Epoxy glue (5-min epoxy; Devcon, cat. no. 14250): silicone glues can be used as alternativesSilicon wafer containing an array of microfabricated silicon pillars that are 10 m in diameter and 50 m in height with a center-to-center spacing of 40 m; the silicon pillars have cylindrical shape and are fabricated by first patterning a silicon wafer using photolithography and removing silicon in the exposed areas by deep reactive-ion etching CRITICAL This silicon master was obtained from the microfabrication foundry service provided by MEMS and Nanotechnology Exchange (http://www.mems-exchange.org/).Dakin fully frosted microscope slide glass (Thermo Scientific, cat. no. 2958-001)

••••••

•••

•••

•

•••••••••••••

•

••

•

•

•

Y-connectors (Cole-Parmer, cat. no. 30703-90)Glass coverslips (48 × 65 mm, no. 1; Gold Seal, cat. no. 3335)Glass slides (75 × 50 mm; Fisher Scientific, cat. no. 12-553-5B)Needles (for cell seeding, 25G, 5/8 inch; BD, cat. no. 305122)Sterile syringes (1-ml Tuberculin slip tip; BD, cat. no. 309659)Chopstick-like Ag/AgCl electrode (Millipore, cat. no. MERSSTX01)Ag/AgCl electrode wires (0.008 inches in diameter; A-M systems, cat. no. 530800)Inverted laser-scanning confocal microscope (Leica SP5 X MP DMI-6000)T75 flasks (BD Falcon, cat. no. 137787)SingleQuots vials (Lonza, cat. no. CC-3202)

REAGENT SETUPECM solutions Dilute human fibronectin in carbonate buffer (pH 9.4) to prepare a fibronectin solution at a final concentration of 5 g ml−1 for use in the lung-on-a-chip. For the gut-on-a-chip, mix type I collagen (50 g ml−1 final concentration) and Matrigel (300 g ml−1 final concentration) in prechilled serum-free DMEM. CRITICAL Prepare these ECM solutions imme-diately before use, and keep collagen and Matrigel-containing solution at 4 °C or on ice to prevent unwanted premature polymerization of matrix proteins.Alveolar epithelial cell culture medium Mix RPMI-1640 with FBS (10% (vol/vol)), 100 units ml−1 penicillin, 100 g ml−1 streptomycin and 292 g ml−1 glutamine. CRITICAL Store the medium at 4 °C and use it before the expiration date.Pulmonary microvascular endothelial cell culture medium Add growth supplements contained in each SingleQuots vial to the basal medium included in the EGM-2-MV medium BulletKit, as instructed by the vendor. CRITICAL Store the medium at 4 °C, and use it before the expiration date.Intestinal epithelial cell culture medium Mix DMEM containing 25 mM glucose and 25 mM HEPES with FBS (20% (vol/vol)), 100 units ml−1 penicil-lin, 100 g ml−1 streptomycin and 292 g ml−1 glutamine. For coculture with microbes, prepare antibiotic-free DMEM containing FBS (20% (vol/vol)).

CRITICAL Store the medium at 4 °C, and use it before the expiration date.Bacterial culture medium Prepare culture medium by dissolving 5.5 g of Lactobacilli MRS broth (Difco) in 100 ml of dH2O under agitation for mixing, and then autoclave it at 121 °C for 15 min. After cooling down the autoclaved MRS broth, store it at 4 °C until use (up to 6 months).PDMS etchant Mix TBAF and MP at a volumetric ratio of 1:3 (TBAF:MP). ! CAUTION Wear safety glasses, impervious clothing, face protection and gloves to prevent skin and eye damage. CRITICAL This reagent must be made fresh for each use.FITC-inulin solution To carry out permeability assays in the lung-on-a-chip, dissolve FITC-inulin in pulmonary microvascular endothelial cell culture medium at a concentration of 1 mg ml−1. CRITICAL This reagent must be made fresh for each use.

•••••••

•••

PROCEDUREFabrication of SU-8 silicon master and its silanization TIMING 17 h1| Place a 3-inch (76.2 mm) silicon wafer on the spinner chuck of a spin coater and apply vacuum.

2| Spin the wafer at 3,000 r.p.m. for 1 min while spraying acetone, methanol and isopropyl alcohol in a sequential manner to clean the surface of the wafer.

3| Dehydrate the wafer completely by placing it on a hot plate maintained at 200 °C for 10 min.

4| Remove the wafer from the hot plate and allow it to cool for 2 min.

5| Use wafer tweezers to pick up the cleaned wafer and put it back on the spin coater. Ensure that vacuum is on and the wafer is centered.

6| Pour ~10 g of SU-8 2100 photoresist on the center of the wafer. CRITICAL STEP Perform this step slowly in order to prevent the formation of air bubbles in the poured photoresist. When the

desired amount is dispensed, gradually change the angle of the SU-8 container while rotating it to cut off the stream of SU-8.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2143

7| Wait for 5 min to allow the dispended SU-8 to spread over the wafer.

8| Start the spin coater and spin the wafer at 500 r.p.m. for 10 s.

9| Ramp up the spinning speed to desired speed and hold. For the lung-on-a-chip, increase the speed to 3,000 r.p.m. at an acceleration of 300 r.p.m. s−1 and at 3,000 r.p.m. hold for 30 s to achieve the desired thickness of SU-8 photoresist (100 m). For the gut on-a-chip, spin at 2,000 r.p.m. for 30 s to achieve the desired thickness of SU-8 photoresist (150 m).

After spin-coating, remove the photoresist accumulated at the wafer edges by dispensing SU-8 developer solution to the edges of the wafer spinning at 1,000 r.p.m. for 1 min.

10| Place the wafer on a hot plate at 65 °C for 5 min. CRITICAL STEP Keep the hot plate leveled to prevent unwanted changes in the thickness of the spin-coated SU-8 film.

11| Transfer the wafer to a hot plate at 95 °C, and bake the wafer for 30 min.

12| Remove the wafer from the hot plate, and cool it at room temperature (23 °C).

13| Load a photomask and the SU-8–coated wafer onto a mask aligner and bring them in conformal contact.

14| Expose the wafer to UV light for 30 s at 260 mJ cm−2.

15| Place the exposed wafer on the hot plate at 65 °C for 5 min, and then bake the wafer at 95 °C for 12 min.

16| Place the wafer in SU-8 developer for 15 min.

17| Remove the wafer from the solution and spray it with SU-8 developer to rinse off undeveloped photoresist.

18| Spray and wash the wafer with isopropyl alcohol. CRITICAL STEP Underdevelopment results in the formation of white residue on the surface of the wafer during this step.

In this case, immerse the wafer back in the developer or spray it with additional developer and repeat the rinsing step with isopropyl alcohol.

19| Use pressurized filter nitrogen or air to dry the wafer.

20| Place the wafer in a desiccator connected to vacuum and place a glass coverslip adjacent to the wafer.

21| Put a small drop containing ~35 l of silane on the coverslip, and evacuate the chamber to induce the evaporation of silanizing agent. ! CAUTION Perform this procedure inside a ventilated chemical fume hood, and wear safety glasses, gloves and face protection.

22| Remove the wafer from the desiccator after overnight silanization.PAUSE POINT The silanized silicon master can be stored indefinitely at room temperature.

Mixing and degassing of PDMS TIMING 10–40 min 23| Prepare a degassed mixture of PDMS silicone elastomer base and curing agent for replica molding. This procedure can be performed manually (option A) or with the assistance of a Thinky centrifugal mixer (option B).(A) Manual mixing and degassing TIMING 40 min (i) Place an empty disposable plastic cup on a scale with appropriate precision and pour silicone elastomer base and

curing agent in a 15:1 (base:curing agent) weight ratio. CRITICAL STEP Use large enough amounts of elastomer base to produce PDMS channel slabs with desired thickness

(upper channel slab of ~5 mm and lower channel slab of ~1.5 mm). (ii) Mix silicone elastomer base and curing agent vigorously by hand using a disposable plastic fork for 5 min until PDMS

appears white. (iii) Place the cup containing the PDMS mixture in a vacuum desiccator for ~30 min. (iv) Remove the cup from the chamber to complete degassing.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2144 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

(B) Thinky mixer–assisted mixing and degassing TIMING 10 min (i) Dispense PDMS elastomer base and curing agent into a disposable cup provided by the manufacturer (Thinky cup) at a

weight ratio of 15:1 (PDMS prepolymer:curing agent). CRITICAL STEP Use large enough amounts of elastomer base to produce PDMS channel slabs with desired thickness

(upper channel slab of ~5 mm and lower channel slab of ~1.5 mm). (ii) Switch on the mixer and place the cup with PDMS mixture in a holder. (iii) Adjust balance in the mixer using a spin dial to reflect the total weight of the sample consisting of PDMS mixture,

Thinky cup and a holder. (iv) Set operation mode to ‘Mix’. Adjust ‘Mix’ speed and time to 2,000 r.p.m. and 2 min, respectively. Set ‘Defoam’ speed

and time to 2,200 r.p.m. and 2 min, respectively. (v) Close the lid and press ‘Start’. (vi) Retrieve PDMS mixture when mixing and degassing are complete (4 min total).

Creation of the upper microchannels of the lung-on-a-chip TIMING 5.5 h24| Attach a master silicon wafer with desired microfabricated channel features to the bottom surface of a Petri dish (100 mm) using double-sided tape.

25| Use compressed filtered air to gently blow off dust particles from the silicon master. CRITICAL STEP Avoid using large air pressure because this may cause unwanted detachment of SU-8 microchannel features

from a silicon wafer.

26| Pour 15:1 (wt/wt) PDMS mixture onto the master in a Petri dish (Fig. 2a). CRITICAL STEP Dispense ~16 g of PDMS mixture to achieve the desired thickness of the upper microchannel slab, which is

5 mm (Fig. 2a).

27| Cover the Petri dish with a lid and put it in an oven at 60 °C for 5 h to fully cure PDMS.

28| Remove the dish and allow it to cool to room temperature for ~10 min.

29| Cut the fully cured PDMS along the edge of the silicon wafer using a scalpel, and carefully peel it off. CRITICAL STEP Be careful when using a scalpel, in order to avoid injuries and fracture of fragile silicon masters.

30| Use a scalpel or a sharpened tile scraper to cut the microchannel slab to desired size and shape. Typically, cut the fully cured microchannel slab to a rectangular block (2 cm wide × 3.6 cm long) (Fig. 2b).

CRITICAL STEP Make sure that microchannels are located at the center of the rectangle.

31| Cut away the corners of the PDMS block at a 45° angle using a razor blade (Fig. 2c). This facilitates bonding between PDMS layers by removing surface irregularities created typically at the corners of PDMS slabs.

32| Bore holes through the PDMS slab using hole punchers. Use a 1.5-mm hole puncher to generate holes labeled with C and C* that provide access to the central upper and lower cell culture microchannels, respectively; use a 2-mm hole puncher to

200 µm

200 µm

100 µm

400 µm

SU-8

Siliconwafer

PDMSslab

PDMS

Side vacuummicrochannel

C

CV

V

V*

V

V

V*C*

C*

Cell culturemicrochannel

a b

c d e

Figure 2 | Fabrication of the upper microchannels of the lung-on-a-chip. (a) Prepolymer of PDMS mixed with curing agent is poured onto the photolithographically prepared SU-8 microchannel features and subsequently cured at an elevated temperature. (b) The fully cured PDMS is peeled off of the master and cut into a rectangular block embossed with three parallel upper microchannels that are 100 m in thickness. The widths of the central cell culture channel and two side vacuum channels are 400 m and 200 m, respectively. (c,d) The corners of the PDMS slab are removed (c), and access ports are made to the microchannels using hole punchers (d). Cell culture and vacuum channels are denoted by C and V, respectively, and asterisks mark access holes for the lower microchannels. (e) The upper PDMS slab is cleaned and wrapped in packaging tape for later use.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2145

create access ports designated with V and V*, which are upper and lower vacuum chambers, respectively. Bore these holes starting from the micropatterned side of the PDMS slab to ensure precise positioning of the holes relative to the microchannels.

CRITICAL STEP These holes serve as access ports to the different microchannels of the upper and lower PDMS layers for vacuum application as well as for delivery of cells, culture medium and other materials (Fig. 2d).

33| Put packaging tape on both sides of the PDMS block and then peel the tape off to clean the surfaces. Wrap the final PDMS slab in packaging tape to keep it clean until use (Fig. 2e).

PAUSE POINT The upper microchannel slabs wrapped in tape can be stored indefinitely.

Fabrication of the lower microchannel of the lung-on-a-chip TIMING 6.5 h34| Attach a master silicon wafer with desired microfabricated channel features to the bottom surface of a Petri dish (100 mm) using double-sided tape.

35| Use compressed filtered air to gently blow off dust particles from the silicon master. CRITICAL STEP Avoid using large air pressure because this may cause unwanted detachment of SU-8 microchannel features

from a silicon wafer.

36| Pour ~9 g of degassed 15:1 PDMS mixture onto the cleaned silicon master. This produces a 1.5-mm-thick PDMS slab.

37| Place the Petri dish on a level surface for 1 h to ensure uniform coverage of the master surface with PDMS.

38| Cure PDMS in an oven at 60 °C for 5 h. CRITICAL STEP It is crucial to keep the Petri dish level inside the oven during curing. Failure to do so results in a PDMS

slab with nonuniform thickness, which adversely affects imaging of living cells and tissues in the microdevice.

39| Remove the dish and allow it to cool to room temperature for ~10 min.

40| Cut the fully cured PDMS along the edge of the silicon wafer by using a scalpel, and carefully peel it off. CRITICAL STEP Be careful when using a scalpel, in order to avoid injuries and fracture of fragile silicon masters.

41| Cut the microchannel slab to a rectangular block of size 2 cm (width) × 4 cm (length) by using a scalpel or a sharpened tile scraper, while ensuring that microchannels are located at the center of the rectangle.

42| Cut away the corners of the PDMS block at a 45° angle using a razor blade. This facilitates bonding between PDMS layers by removing surface irregularities created typically at the corners of PDMS slabs.

43| Bore holes through the PDMS slab using hole punchers. Use a 1.5-mm hole puncher to generate holes labeled with C and C* that provide access to the central upper and lower cell culture microchannels, respectively. Bore these holes starting from the micropatterned side of the PDMS slab to ensure precise positioning of the holes relative to the microchannels.

44| Place the cleaned PDMS block on a glass slide with the microchannel side facing up, and put a small piece of packaging tape to cover the surface and prevent its contamination with dust particles.

PAUSE POINT The lower microchannel slabs covered with tape can be stored indefinitely until use.

Fabrication of porous PDMS membranes of the lung-on-a-chip TIMING 1.4 d45| Attach a cleaned flat silicon wafer without micropatterned features to the bottom of a 100-mm Petri dish with double-sided tape and pour 15:1 PDMS onto the wafer surface to generate a PDMS layer with a thickness of ~5 mm.

46| Repeat Steps 27–29.

47| Cut the PDMS slab into rectangular pieces with of size 2.5 cm (width) × 4 cm (length) by using a scalpel or a sharpened tile scraper.

48| Cut away the corners of the rectangular PDMS block at a 45° angle by using a razor blade.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2146 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

49| Clean the PDMS block with packaging tape as described in Step 33, and adhere it onto a clean glass slide.

50| Put a piece of packaging tape on the top surface of the PDMS slab to prevent the deposition of dust particles.

51| Remove the tape and place the PDMS block bonded to a glass slide in a desiccator.

52| Put a clean glass slide adjacent to the PDMS slab, and use a pipette to dispense ~35 l of undiluted silane on the glass surface. ! CAUTION Perform this procedure inside a ventilated chemical fume hood and be sure to wear safety glasses, gloves and face protection.

CRITICAL STEP It is important to use small amounts of silane during this process to prevent the formation of residues on the surface.

53| Apply vacuum to the chamber, and keep the PDMS slab in the evacuated chamber overnight.

54| Remove the silanized PDMS slab from the desiccator and keep it in a covered Petri dish until use.

55| Mix PDMS base with a curing agent at a weight ratio of 10:1 (prepolymer:curing agent) and degas it as described in Step 23.

56| Clean the silanized PDMS substrate attached to a glass slide from Step 54 by using compressed filtered air, and place it on a vacuum chuck of a spin coater.

57| Gently pour a 10:1 PDMS mixture over the silanized surface and wait for 10 min.

58| Ramp the spinning speed up to 500 r.p.m. at 100 r.p.m. s−1 and dwell for 20 s to allow PDMS to initially spread the PDMS mixture over the entire surface.

59| Increase the speed to 2,400 r.p.m. over 15 s and hold it for 10 min. This produces a 10- m-thick layer of uncured PDMS on the silanized PDMS slab (Fig. 3a).

60| When spin-coating is complete, allow the sample to rest for 3 min before transferring it to a work bench.

61| Slowly peel the silanized PDMS block from the glass slide. CRITICAL STEP Avoid unwanted contact with the top surface coated with uncured PDMS and prevent excessive bending of

the slab.

62| Invert the silanized PDMS slab and bring the PDMS-coated surface in contact with a silanized silicon wafer containing an array of microfabricated silicon pillars that are 10 m in diameter and 50 m in height with a center-to-center spacing of 40 m (Fig. 3b).

CRITICAL STEP The silicon wafer needs to be silanized properly before use to prevent detachment or fracture of the microfabricated silicon posts during membrane fabrication.

a SilanizedPDMS

Spin-coated PDMS(uncured)

b

Siliconmaster

c Compression

Siliconmaster

PDMS

Weightd

SilanizedPDMS

PorousPDMS

membrane

Figure 3 | Fabrication of porous PDMS membranes. (a) PDMS is spin-coated on a silanized PDMS slab to form a 10- m-thick film of uncured PDMS. (b,c) Subsequently, the PDMS slab is placed on a silicon wafer patterned with an array of microfabricated pillars (b) and compressed uniformly against the master using weight during PDMS curing (c). The diameter and height of the pillars are 10 m and 50 m, respectively. (d) After complete curing of PDMS, the weight and silicon master are removed to produce a 10- m-thick PDMS membrane with microfabricated through-holes that is reversibly attached to a silanized PDMS surface.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2147

63| Place a glass coverslip on the PDMS block. CRITICAL STEP Push down gently on the coverslip to ensure intimate contact between the slab, the silicon master and

the coverslip.

64| Place a 200 g weight over the entire surface of the coverslip to allow microfabricated pillars to penetrate the uncured PDMS layer on the silanized PDMS slab (Fig. 3c).

65| Leave the sample at room temperature overnight. CRITICAL STEP It is crucial to keep the sample level during overnight curing.

66| Place the entire assembly in a 60 °C oven for 1 h to complete PDMS curing.

67| Remove the sample from the oven, and remove the weight.

68| Carefully remove the coverslip from the back of the silanized PDMS block. Use a scalpel to lift off the coverslip starting from its edges.

CRITICAL STEP This step should be performed very slowly in order to prevent the coverslip from breaking.? TROUBLESHOOTING

69| Take great care to slowly peel the PDMS slab from the silicon wafer. Completion of this step results in a fully cured 10- m-thick PDMS membrane with through-holes on the silanized PDMS slab (Fig. 3d).

70| Place the PDMS slab in a covered Petri dish with the membrane side facing up, in order to prevent contamination of the membrane surface with dust particles.

PAUSE POINT The PDMS membranes can be stored in this manner indefinitely until use.

Alignment and assembly of the lung-on-a-chip TIMING 1.1 d71| Use a microscope to inspect the PDMS membrane at low magnification, and remove any particulates that might have settled on its surface using compressed filtered air.

72| Locate a defect-free space on the membrane without ripped membrane fragments and blocked pores that has large enough surface area to cover the upper microchannels.

73| Treat the membrane surface with plasma by using the corona treater for 1 min and 40 s. CRITICAL STEP Use sweeping motions to achieve uniform surface treatment. Avoid surface overtreatment because

exposure of PDMS to corona for durations exceeding 3 min often results in the generation of surface cracks and failure of bonding.

74| Remove packaging tape from the upper microchannel slab and expose the channel side to plasma, as described in Step 73.

75| Overlay the upper microchannel slab on the PDMS membrane (Fig. 4a). CRITICAL STEP Ensure that the upper microchannels are bonded to the defect-free area of the membrane identified in

Step 72. Press the PDMS slabs to permit close contact between layers and to remove trapped air.

76| Incubate the assembled layers in a 60 °C oven overnight to achieve complete irreversible bonding between the membrane and the upper microchannels.

77| Take out the sample from the oven and cool it for 10 min.

78| Cut along the edges of the upper channel slab by using a scalpel.

79| Gently lift one of the corners of the upper microchannel slab and slowly peel off the channel slab from the silanized PDMS. This process separates the upper microchannel permanently bonded to the membrane from the silanized PDMS surface.

80| Use fine-tip tweezers to remove PDMS membrane from the inlets and outlets of the lower microchannels.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2148 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

81| Treat the exposed PDMS membrane surface with corona for 1 min and 20 s.

82| Peel the packaging tape from the lower microchannel slab, and sweep the corona-discharging electrode of the corona treater over the entire surface on the channel side for 1 min and 20 s.

83| Transfer the corona-treated samples to a microscope, and place the lower microchannel slab attached to a coverglass on a microscope stage.

84| Hold the assembly of the upper channel slab and membrane with thumbs and index fingers, and place it on the lower channel block (Fig. 4b). Use a ×10 or ×4 objective to visually guide the alignment.

CRITICAL STEP During this procedure, align the walls of the upper microchannels to those of the lower microchannels.? TROUBLESHOOTING

85| When alignment is complete, press down the upper PDMS slab to eliminate trapped air pockets and to ensure strong contact between layers.

CRITICAL STEP Do not apply excessive pressure during this step to prevent irreversible collapse of the intervening membrane.

86| Place the assembled device into a 60 °C oven, and incubate it overnight to achieve permanent bonding between the upper and lower microchannels (Fig. 4c).

CRITICAL STEP After the completion of this step, attempt to pull apart the upper and lower PDMS slabs to qualitatively determine the success of device bonding. Incomplete or unsuccessful bonding results in peeling off and separation of the pulled PDMS slabs.

Chemical etching of PDMS membrane in the lung-on-a-chip TIMING 20 min87| Trim 3 mm off the tapered ends of four 200- l pipette tips.

88| Insert the needle electrode of the corona treater into the trimmed pipette tips and treat their interior surfaces with corona for 10 s each.

89| Put packaging tape over the inlet and outlet ports leading to the central microchannels to prevent unwanted etching of the cell culture membrane in these channels caused by spillage of PDMS etchant.

90| Insert the trimmed pipette tips into the inlets and outlets of the upper and lower side vacuum microchannels.

91| Add 150 l of PDMS etchant (see Reagent Setup) to each tip, and use a microscope to ensure that the etchant flows into the microchannels.? TROUBLESHOOTING

92| Allow 5 min to completely etch away the PDMS membrane in the side microchannels (Fig. 4d). Observe the etching process under a microscope.

CRITICAL STEP The etchant dissolves the PDMS membrane at a rate of ~3 m min−1 under the conditions described in this protocol. The channel walls separating the side vacuum channels and central cell culture channels are etched at the same rate,

a b c

d

Lower channelslab

Upper channelslab

PDMSmembrane PDMS

membrane

Side

SilanizedPDMS

Peel

Beforeetching

Afteretching

chambers

Figure 4 | Alignment, bonding and chemical etching of the lung-on-a-chip microdevice. (a) After brief surface treatment with corona, the upper PDMS slab is irreversibly bonded to the membrane. (b) Once bonding is accomplished, the upper microchannel slab is carefully separated from the silanized PDMS block, primed with corona and aligned/bonded to the lower microchannels. (c,d) This results in the production of a fully assembled microfluidic device (c) in which the membrane layers in the vacuum microchannels are etched away to form two hollow side chambers (d).

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2149

and thus their thickness changes from 50 m to ~35 m as a result of etching for 5 min. Continuous etching for over 8 min leads to excessive thinning of these walls, and this often results in the failure of bonding and leakage of etchant into the cell culture microchannels. Conversely, etching completed within 3 min substantially increases the strength of vacuum required to achieve effective membrane stretching. Therefore, we recommended keeping the total etching time between 4 min and 7 min for best results.

93| When the etching process nears completion, aspirate the etchant from the pipette tips and microchannels using vacuum, and completely dry the channel surfaces.

94| Remove the pipette tips to complete the production of an assembled lung-on-a-chip microfluidic device.

Incorporation of fluidic interconnects and sterilization of the lung-on-a-chip TIMING 1.5 h95| Bend the tips of four 18-gauge blunt needles by 90° using pliers and break off the needle at a point near the syringe entry port.

96| Insert the needles into the four inlet ports of the side microchambers.

97| Put small pieces of packaging tape over the outlet ports of the side microchambers. This is to prevent loss of vacuum suction during mechanical stretching.

98| Repeat Step 95 for four 20-gauge blunt needles and fit them into the inlet and outlet ports of the central microchannels.

99| Cut eight pieces of silicone tubing to a length of 5 cm and connect them to the free ends of the needles.

100| Mix 5-min epoxy on a clean surface and apply it around each needle.

101| Wait for 30 min to allow the applied epoxy to set.

102| Completely cure the glue in a 60 °C oven for 30 min.

103| Remove the device from the oven, and allow it to cool to room temperature. CRITICAL STEP To examine the quality and performance of the device, connect the device to the vacuum pump and test

vacuum-assisted membrane stretching after the completion of this step. Leakage of vacuum from the side chamber owing to unsuccessful PDMS bonding results in little or no stretching of the membrane. Alternatively, fill the central cell culture

a

b c

d

e

f150 µm

1,600

µm

1,000

µm

1,600

µm

Upperlayer

Upper layer+

porous membrane

Plasmatreatment

Plasmatreatment

Porousmembrane

Tear offby tweezers

Bond/cure

Bond/cure

Gut-on-a-chip

Vacuumchamber

Vacuumcontroller

Cellchannel

Peel off support

Lower layer

PDMSslab

5 mm

Figure 5 | Microfabrication of gut-on-a-chip. (a) Dimensions of the central cell culture microchannel and side vacuum chambers in the upper layer of the gut-on-a-chip device. The size of microchannels in the lower layer is the same as that in the upper layer. (b–e) Layer-by-layer microfabrication process. The gut-on-a-chip microdevice is composed of the upper layer, a porous membrane and the lower layer. Each layer is sequentially bonded and cured to fabricate the upper (blue) and lower (orange) cell culture microchannels and two lateral vacuum chambers (gray). The porous PDMS membrane in the vacuum chambers is manually torn off by using tweezers to create full-height vacuum chambers. (f) A photograph of the fully assembled gut-on-a-chip microdevice with the upper and lower microchannels filled with blue and red dyes, respectively.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2150 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

channels with deionized water, and check for the leakage of water into the side chambers when vacuum is applied. Nonuniform chemical etching generates variations in the thickness of the vertical walls separating the central cell culture channels from the side vacuum chambers, leading to varying degrees of membrane stretching along the channel length. These undesirable events can be visually detected using a microscope.

104| Place the fully assembled device into the UV-ozone cleaner, and sterilize the device for 40 min.

105| Remove the sterile device from the cleaner and immediately put it in a sterile Petri dish to transfer it to a biosafety cabinet.

CRITICAL STEP Devices should be sterilized in this manner immediately before use.

Microfluidic cell culture in the lung-on-a-chip TIMING 5–21 d106| Introduce 500 l of the ECM coating solution (see Reagent Setup) into both the upper and lower central microfluidic channels, and pinch off tubing to prevent coating solution from leaking out of the device.

CRITICAL STEP Ensure that the entire microchannels are filled with the ECM solution without any trapped air bubbles.? TROUBLESHOOTING

107| Place the microdevice in a humidified 37 °C incubator overnight.

108| Flush the channels with medium to remove residual coating solution, and return the microdevice to the incubator for 2 h.

109| Remove the culture medium from a T75 flask containing 80–90% confluent human alveolar epithelial cells. Wash the cells with Ca2+-free and Mg2+-free PBS twice.

110| Add 1 ml of prewarmed trypsin/EDTA solution (0.05%) to the cell culture flask, and incubate the mixture in a humidified incubator (37 °C, 5% CO2) for 5–7 min or until the cells are released from the growth surface. Count and pellet 250,000 cells by centrifugation at 220g for 5 min.

111| Thoroughly resuspend the epithelial cell pellet in 50 l of serum-containing epithelial culture medium, and load it into a 1-ml syringe.

112| Use clamps to close the inlet and outlet of the lower microchannels, and introduce the concentrated cell suspension into the epithelial compartment through the outlet of the upper microchannel.

CRITICAL STEP Inject the cell suspension solution very slowly and carefully to prevent introduction of air bubbles. Make sure that the needle attached to the syringe containing the cell suspension is tightly fitted into the outlet tubing.? TROUBLESHOOTING

113| Clamp closed the inlet and outlet ports of the upper microchannel to allow the seeded epithelial cells to attach to the membrane surface.

114| Incubate the device at 37 °C for 3–4 h.

115| Inspect the microchannels under a microscope to ensure cell adhesion to the membrane.

116| Connect the device to a syringe pump and perfuse both central microchannels with epithelial culture medium at a volumetric flow rate of 20 l h−1 in a humidified incubator maintained at 37 °C.

117| On the following day, repeat Steps 109–111 to harvest and resuspend primary human pulmonary microvascular endothelial cells.

CRITICAL STEP Before seeding of endothelial cells, ensure that the majority of the pores on the PDMS membrane are covered by the attached epithelial cells to prevent inter-compartment cell migration through the membrane pores during culture.

118| Disconnect the device from a syringe pump and close the upper microchannel.

119| Invert the microdevice and slowly inject the endothelial cell suspension into the outlet port of the lower channel by using the same method described in Step 112.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2151

120| Pinch off tubing connected to the inlet and outlet of the lower channel, and incubate the device, inverted, in a humidified 37 °C incubator for 3–4 h.

121| Once endothelial cell attachment is accomplished, reconnect the device to the pump and perfuse the upper and lower channels with epithelial and endothelial culture medium, respectively, at a volumetric flow rate of 20 l h−1 in a humidified incubator maintained at 37 °C.

122| Allow the cells to grow to confluence over the course of 5–7 d. Add dexamethasone (1 M final concentration) to the epithelial cell culture medium on day 3.

CRITICAL STEP Dexamethasone strengthens the structural integrity of the alveolar epithelial monolayer.? TROUBLESHOOTING

123| Once cells form a confluent monolayer, gently aspirate the epithelial cell culture medium from the upper channel to create an air-liquid interface over the apical surface of the alveolar epithelial cells.? TROUBLESHOOTING

124| Replace the culture medium in the lower channel with a 50/50 mixture of epithelial and endothelial media containing 1 M dexamethasone. Continue microfluidic cell culture in this manner for 15 d.

125| On day 15, apply cyclic strain to the cell populations via the side chambers by using a vacuum pump controlled by the Flexcell Tension System. Apply cyclic strain of 10% at a frequency of 0.2 Hz in a sinusoidal waveform to mimic physiological breathing in the human lung.? TROUBLESHOOTING

Analysis of microengineered tissues TIMING 30 min–1.25 d126| When the microfluidic cell culture is complete, assess the viability of the cultured cells (option A), assess barrier integrity via immunofluorescence staining of intracellular junctions (option B) or measure trans-bilayer electrical resistance (option C).(A) Evaluation of cell viability TIMING 40 min (i) Dilute calcein-AM and ethidium homodimer-1 stock solutions in D-PBS to prepare a working solution containing 2 M

calcein-AM and 4 M ethidium homodimer-1. (ii) Flush D-PBS through the microchannels to wash the cells. (iii) Introduce the dye solution into the microchannels, and keep the microdevice in a cell culture incubator

for 30 min. (iv) Transfer the device to a microscope to visualize and enumerate live and dead cells by using fluorescence microscopy.(B) Immunofluorescence staining of intercellular junctions TIMING 1.25 d (i) Introduce 4% paraformaldehyde into the channels, and incubate the device in a cell culture incubator at room

temperature for 15 min. (ii) Flush the channel with D-PBS to remove the fixative. Permeabilize the cells by introducing 0.3% (vol/vol) Triton X-100

in D-PBS into culture area, and incubate the device at room temperature for 15 min. (iii) Wash the channel with D-PBS. Introduce BSA blocking solution (2% (wt/vol)) to the channels and incubate the device

at room temperature for 2 h. (iv) For staining of endothelial junctions, dilute primary VE-Cadherin antibody in BSA solution at a ratio of 1:100, and

introduce it into the lower channel. Incubate the mixture at 4 °C overnight. To measure negative background signal, use isotype controls at identical concentrations and staining conditions as the target primary antibody.

(v) Remove antibody solution by flushing the channels with D-PBS three times. (vi) Dilute fluorescently labeled goat anti-rabbit antibody in BSA solution (1:100) and introduce it into the lower channel.

Incubate it at room temperature in the dark for 1 h. (vii) Repeat Step 126B(v). The device is now ready for visualization of endothelial adherens junctions by using fluorescence

microscopy. (viii) Repeat Step 126B(i–vii) with fluorescently labeled anti-occludin antibody in the upper microchannel to visualize

tight-junction formation in the epithelial cells. (ix) For staining nuclei, incubate the cells with DAPI antibody at room temperature for 30 min and repeat Step 126B(v) for

visualization. (x) For F-actin staining, repeat Step 126B(i–iii) and incubate the cells with fluorescently labeled phalloidin at room

temperature for 30 min. Repeat Step 126B(v) for visualization.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2152 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

(C) Measurement of trans-bilayer electrical resistance TIMING 30 min (i) Sterilize two Ag/AgCl wire electrodes with 70% ethanol in a biosafety cabinet. (ii) Connect the electrodes to a digital volt-ohm meter by using alligator clips. (iii) Prepare an assembled and sterilized device without cells and inject the culture medium into the upper and lower

culture channels through silicone tubing leading to the inlet ports. CRITICAL STEP Continue injection until the culture medium flows out of the channels, and fill the silicone tubing

connected to the outlet ports. (iv) Insert one of the wire electrodes into the tubing connected to the inlet of the upper microchannel and the other

into the tubing connected to the lower microchannel. Ensure that the electrodes are in contact with culture medium in the tubing.

(v) Allow 1 min before measuring electrical resistance. (vi) Repeat Step 126C(iv,v) with a device with cultured living cells. (vii) Evaluate the net electrical resistance by subtracting the resistance measured in a device without cells from that

obtained from a cell-containing device. (viii) Multiply the net electrical resistance by the total surface area covered by the tissue barrier to calculate trans-bilayer or

trans-epithelial electrical resistance in cm2.

? TROUBLESHOOTINGTroubleshooting advice can be found in Table 1.

TABLE 1 | Troubleshooting table.

Step Problem Possible reasons Solution

68 Fracture of glass coverslip

Attempt to rapidly peel off the cover slip

Use a scalpel or fine-tip tweezers to remove the broken pieces of glass from the PDMS slab

84 Misalignment Failure to match the walls of the microchannels in the upper and lower PDMS slabs

Lift off the upper PDMS slab slowly and repeat Step 84 until complete alignment of the microchannel walls is achieved. Take great care when peeling off the slab, as this may cause the PDMS membrane to tear

91 Lack of etchant flow Hydrophobicity of PDMS channel surfaces

Apply gentle vacuum aspiration to the outlets of the side vacuum microchannels to induce flow. Alternatively, treat the surfaces of the side vacuum channels with corona before etching to render them hydrophilic and to facilitate the introduction of etchant

106, Box 1 (step 35)

Trapped air bubbles Rapid injection of ECM solution into hydrophobic PDMS channels

Connect syringes loaded with ECM solution to the inlet ports of the upper and lower cell culture microchannels and flush the channels by manually pushing the plungers of the syringes until the trapped air bubbles are removed. Alternatively, make the cell culture channels hydrophilic using corona treatment before the injection of ECM solution

112 Air bubbles in the cell culture channels after cell seeding

Trapped air in the syringe needle used for cell seeding or inadvertent injection of air from the syringe after the total volume of cell suspension solution is dispensed

Flush the channels with fresh culture medium to remove the bubbles, and repeat Steps 109–112 to re-seed cells

122 Subconfluent cellular monolayers

Low cell seeding density or cell aggregation

Re-seed the device using the same number of cells and allow the cells to populate the uncovered areas of the membrane over the course of 2 d

123 Re-filling of the upper alveolar channel with culture medium

Defective or immature epithelial monolayers with intercellular gaps

Re-seed the device with epithelial cells and allow 2 d to repopulate the gaps that are not covered by cells

(continued)

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2153

TIMINGFabrication of the lung-on-a-chip: ~3.6 dSteps 1–22, master fabrication: 17 hSteps 23–44, preparation of the upper and lower microchannels: 6.5 hSteps 45–70, membrane fabrication: 1.4 dSteps 71–86, device alignment and assembly: 1.1 dSteps 87–94, chemical etching: 20 minSteps 95–105, installation of interconnects and device sterilization: 1.5 hSteps 106–125, microfluidic cell culture in the lung-on-a-chip: 5–21 dStep 126, analysis of microengineered tissues: 30 min–1.25 dFabrication of the gut-on-a-chip: ~3.5 dSteps 1–22, master fabrication: 17 hBox 1, steps 1–7, preparation of the upper and lower microchannels: 6 hBox 1, steps 8–20, membrane fabrication: 1 dBox 1, steps 21–34, device alignment and assembly, installation of interconnections and sterilization: 1.5 dBox 1, steps 35–53, microfluidic cell culture and microbial coculture: 5–15 dStep 126, analysis of microengineered villi: 30 min−1.25 d

The total time required to fabricate these microdevices can potentially be substantially reduced by baking PDMS at a higher temperature (e.g., 85 °C) to shorten the curing process from 4–5 h to 1 h for the production of the upper and lower micro-channel slabs and porous membranes. This also can eliminate the need for overnight curing after corona treatment to achieve irreversible bonding between PDMS layers and allow for completion of the bonding process within a few hours. Under these conditions, the total fabrication time can be reduced from 3.5 d to 2.1 d.

ANTICIPATED RESULTSThis protocol provides a step-by-step procedure for creating 3D organ-on-chip microfluidic systems that enable one to form one or two closely apposed confluent monolayers of differentiated human cells and to expose them to physiologically relevant mechanical and biochemical cues in polarized microenvironments. The critical steps for manufacturing these microdevices include production of the upper and lower cell culture microchannels (Steps 1–44, Box 1 steps 1–7), fabrication of porous flexible PDMS membranes (Steps 45–70, Box 1 steps 8–20), alignment and irreversible bonding of the upper and lower micro-channels (Steps 71–86, Box 1 steps 21–29; see also Fig. 5), chemical etching of PDMS membranes (for the lung-on-a-chip only, Steps 87–94) and long-term microfluidic culture of human cells (Steps 106–125, Box 1 steps 35–53). We summarize the yields from the critical steps and possible reasons for commonly observed modes of failure in Table 2. On the basis of the results of the fabrication procedures routinely performed by our group, the overall yield is estimated to be ~50–60%. In addition, it is our recommendation to fabricate no more than five devices per person simultaneously to ensure the quality of final products.

In Figure 6 we show an example of a lung-on-a-chip microfluidic device produced by the fabrication protocol described here. Successful completion of the initial steps for channel/membrane fabrication and bonding leads to well-aligned upper and lower microchannels with a microfabricated porous PDMS membrane sandwiched in between that look similar to those shown in Figure 6a. After PDMS etching, the porous membrane in the side channels is no longer visible in the micrograph, indicating that it has been etched away to form hollow chambers (Fig. 6b). This etching process also can leave dark bands along the side walls of the central channels and surface irregularities on the walls of the side chambers that are visible during light microscopy imaging. The fully assembled lung-on-chip is the size of a portable universal serial bus (USB) flash data storage

TABLE 1 | Troubleshooting table (continued).

Step Problem Possible reasons Solution

125 Diminishing levels of cyclic strain

Reduced vacuum strength because of partial or complete blockage of vacuum tubing by liquid formed in the humid environment of cell culture incubators

Install a liquid trap between the device and vacuum pump to prevent the blockage of vacuum tubing

Box 1 (step 51)

Overgrowth of bacteria in co-cultures

High initial seeding density of bacterial cells

Decrease the bacterial cell seeding density. Optimization of initial seeding density is required for different microbial species: as bacteria quickly consume nutrients and oxygen and secrete toxic wastes, the overgrowth of bacteria can impose serious adverse effects on the viability of host cells in microdevices

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2154 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

device, and optical clarity of PDMS makes it possible to resolve the shape and location of individual microchannels embedded in the device with the naked eye (Fig. 6c). Fluidic delivery into and out of the microchannels is enabled by connecting the device with tubing via bent needles inserted into the access ports, as illustrated in Figure 6d.

Human lung epithelial cells and pulmonary microvascular endothelial cells seeded into this device proliferate continuously over the period of 5–7 d to cover the entire surface of the membrane on their respective side (Fig. 7a). Successful formation of these confluent monolayers prevents leakage of culture medium from the lower microchannel into the upper chamber during air-liquid interface culture, facilitating the stable maintenance of the alveolar epithelium at the air-liquid interface for prolonged periods (Fig. 7b). If the epithelial cells are prematurely exposed to air before they are allowed to form a confluent monolayer, the upper alveolar channel can become flooded

TABLE 2 | Yields of critical steps for device fabrication and operation.

Steps Main objectives Yield Common failure modes and possible reasons

1–44, Box 1 (steps 1–7)

Fabrication of the upper and lower cell culture microchannels

~99% PDMS channel slabs with uneven surfaces owing to the failure to keep the curing oven level

45–70, Box 1 (steps 8–20)

Fabrication of thin porous PDMS membranes

~85% Blockage of membrane pores owing to incomplete contact between the silicon master and spin-coated PDMS Membrane rupture during membrane peeling owing to unwanted adhesion of the microfabricated posts on the silicon master to cured PDMS

71–86, Box 1 (steps 21–29)

Alignment and irreversible bonding of the upper and lower PDMS layers and the porous membrane

~80% Misalignment of microchannel walls Incomplete bonding owing to uneven surface treatment with corona Rupture of PDMS membranes during multiple attempts to align the microchannels

87–94 Chemical etching of PDMS membranes

~95% Leakage of chemical etchant into the cell culture channels owing to compromised channel bonding

106–125, Box 1 (steps 35–53)

Microfluidic cell culture ~70% Lack of cell attachment after seeding owing to incomplete membrane surface coating with ECM Cell death and detachment caused by the detrimental mechanical forces generated by air bubbles formed in the cell culture channels as a result of evaporation Detachment and subsequent rolling-up of fully confluent cellular monolayers owing to defective ECM coating Specific to the lung-on-a-chip: unwanted cell injury and death during the withdrawal of culture medium from the upper channel to form an air-liquid interface

Sidechannels

b

dc

a

Sidechannels Side chamber

Device

Tubing

Side chamber

Upper andlowerchannels

Figure 6 | A multilayered 3D microfluidic device for the production of the human breathing lung-on-a-chip. (a) A micrograph of well-aligned upper and lower microchannels separated by a thin porous PDMS membrane in an assembled microdevice. Scale bar, 200 m. (b) Chemical etching of PDMS successfully removes the porous membrane in the side channels and causes the thinning of the PDMS walls between the central microchannels and the side chambers. Scale bar, 200 m. (c) Optical transparency of PDMS permits direct visualization of individual microchannels embedded in the device. In this image, the cell culture microchannels and vacuum chambers are shown in orange and blue, respectively. (d) Tubing is attached to the assembled device by using bent needles inserted into the access ports. The final PDMS microdevice can be placed on a regular microscope stage for imaging.

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2155

with culture medium infiltrating from the lower channel; in this case, cultures need to be maintained longer under flow in both channels before air is introduced into the upper channel. For the entire duration of air-liquid interface culture, the alveolar epithelial cells are fed from their basal side, and this culture configuration effectively maintains its viability (Fig. 7c).

10% linear cyclic strain is applied to the alveolar-capillary barrier for 3 d before the end of air-liquid interface culture to mimic physiological breathing, and this does not compromise barrier integrity54. At the termination of air-liquid interface culture, the epithelial and endothelial cells remain viable and form intact homogeneous cell monolayers that express high levels of intercellular junctional proteins, such as occludin and VE-cadherin (Fig. 7d).

In the gut-on-a-chip, human intestinal epithelial Caco-2 cells cultured on the ECM-coated porous PDMS membrane form a confluent monolayer with structural integrity over 40 h when exposed to low levels of fluid shear stress (0.02 dyne cm − 2) and cyclic peristalsis–like movements (10% strain at 0.15 Hz; Fig. 8a). The Caco-2 monolayer displays polarized columnar morphology within 2 d and then begins to spontaneously form 3D villus-like structures by 5–7 d at a 30 l h − 1 flow rate (Fig. 8b). Confocal fluorescence microscopy reveals that the nuclei (blue) of the gut epithelial cells are aligned along the boundary of each villus and covered by a continuous apical brush border (green) (Fig. 8c). This intestinal villus structure can be maintained up to 2 weeks in the presence of physiological trickling flow and mechanical deformation.

Air

Submerged

a

b

Epithelium Endothelium

Occludin VE-cadherin

c

d

Figure 7 | Production and microfluidic engineering of the alveolar epithelium and microvascular endothelium in the lung-on-a-chip microdevice. (a) Confluent monolayers of human lung epithelial cells and microvascular endothelial cells are formed on the opposite sides of the membrane within 5 d of cell seeding. During this period, both the upper and lower microchannels are continuously perfused with appropriate culture media. This phase-contrast image (right) was taken on day 5. Scale bar, 75 m. (b) A micrograph of the human alveolar epithelial tissue (right) in the air-filled alveolar microchannel on day 10 in air-liquid interface (ALI) culture. When exposed to air, the confluent monolayer of the alveolar epithelial cells serves as a barrier to fluid leakage from the lower vascular microchannel, facilitating the maintenance of ALI culture conditions for extended periods. Scale bar, 75 m. (c) The majority of the epithelial and endothelial cells in the microdevice are maintained highly viable at the completion of ALI culture (day 20), as evidenced by their staining with CellTracker Green CMFDA (epithelial) and CellTracker Red CMTPX (endothelial) live cell–specific dyes. Scale bar, 25 m. (d) ALI culture in our system also leads to the formation of tight (green; anti-occludin antibody conjugated with Alexa Fluor 488) and adherens (red; anti–VE-cadherin antibody conjugated with Alexa Fluor 594) junctions in the epithelial and endothelial cells, respectively. Blue shows nuclear staining. Scale bars, 25 m.

Upper channelGut epithelial

monolayer

Intestinal villi

Lower channel

a b cFigure 8 | Intestinal epithelial cell culture and spontaneous villus morphogenesis in the gut-on-a-chip microdevice. (a,b) Schematics showing transformation of a planar intestinal epithelial monolayer into 3D villus structure (top) and corresponding phase-contrast images of Caco-2 cells cultured in the gut-on-a-chip recorded at 40 h (a) and 140 h (b) after seeding, respectively (bottom). (c) A horizontal cross-sectional view of Caco-2 villi produced by microfluidic cell culture in the gut-on-a-chip for 140 h under peristalsis-like motions and fluid flow. This fluorescence image shows the continuous brush border membrane labeled with F-actin (pseudocolored in green; phalloidin-CF647) covering the nuclei of the intestinal epithelial cells aligned along the boundary of each villus. An inset indicates the vertical location of the cross-section. Scale bar, 25 m.

ACKNOWLEDGMENTS We thank D. Levner and C. Hinojosa for their assistance in preparing the protocols for device fabrication. This work was supported by the Wyss Institute for Biologically Inspired Engineering at Harvard University and grants from US National Institutes of Health (NIH) NIEHS (ES016665-01A1) and the NIH Common Fund (U01 NS073474) through the Division of Program Coordination, Planning, and Strategic Initiatives (DPCPSI), Office of the Director, NIH and the US Food and Drug Administration (FDA). Additional funds were provided by the Defense Advanced Research Projects Agency (DARPA)

under Cooperative Agreement Number W911NF-12-2-0036, and FDA contract HHSF223201310079C. The content of the information does not necessarily reflect the position or the policy of DARPA or the US Government, and no official endorsement should be inferred.

AUTHOR CONTRIBUTIONS D.H. led development of the lung-on-a-chip, performed experiments, analyzed data and prepared the manuscript. H.J.K. led development of the gut-on-a-chip, performed experiments, analyzed data and

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

2156 | VOL.8 NO.11 | 2013 | NATURE PROTOCOLS

contributed to preparation of the manuscript. J.P.F., D.E.S., M.K., A.B. and G.A.H. provided assistance in experiments, data analysis and manuscript preparation. D.E.I. led the organ-on-chip effort, assisted in experimental design and analysis and helped in writing of the manuscript.

COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests.

Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.

1. Huh, D., Hamilton, G.A. & Ingber, D.E. From 3D cell culture to organs-on-chips. Trends Cell Biol. 21, 745–754 (2011).

2. Jang, K., Sato, K., Igawa, K., Chung, U.I. & Kitamori, T. Development of an osteoblast-based 3D continuous-perfusion microfluidic system for drug screening. Anal. Bioanal. Chem. 390, 825–832 (2008).

3. Leclerc, E. et al. Study of osteoblastic cells in a microfluidic environment. Biomaterials 27, 586–595 (2006).

4. O’Neill, A.T., Monteiro-Riviere, N.A. & Walker, G.M. Characterization of microfluidic human epidermal keratinocyte culture. Cytotechnology 56, 197–207 (2008).

5. Chao, P.G. et al. Dynamic osmotic loading of chondrocytes using a novel microfluidic device. J. Biomech 38, 1273–1281 (2005).

6. Baudoin, R., Griscom, L., Monge, M., Legallais, C. & Leclerc, E. Development of a renal microchip for in vitro distal tubule models. Biotechnol. Prog. 23, 1245–1253 (2007).

7. Leclerc, E., Sakai, Y. & Fujii, T. Microfluidic PDMS (polydimethylsiloxane) bioreactor for large-scale culture of hepatocytes. Biotechnol. Prog. 20, 750–755 (2004).

8. Powers, M.J. et al. A microfabricated array bioreactor for perfused 3D liver culture. Biotechnol. Bioeng. 78, 257–269 (2002).

9. Tilles, A.W., Baskaran, H., Roy, P., Yarmush, M.L. & Toner, M. Effects of oxygenation and flow on the viability and function of rat hepatocytes cocultured in a microchannel flat-plate bioreactor. Biotechnol. Bioeng. 73, 379–389 (2001).

10. Shin, M. et al. Endothelialized networks with a vascular geometry in microfabricated poly(dimethyl siloxane). Biomed. Microdevices 6, 269–278 (2004).

11. Huh, D., Torisawa, Y.S., Hamilton, G.A., Kim, H.J. & Ingber, D.E. Microengineered physiological biomimicry: organs-on-chips. Lab Chip 12, 2156–2164 (2012).

12. van der Meer, A.D. & van den Berg, A. Organs-on-chips: breaking the in vitro impasse. Integr. Biol. 4, 461–470 (2012).

13. Sung, J.H. et al. Microfabricated mammalian organ systems and their integration into models of whole animals and humans. Lab Chip 13, 1201–1212 (2013).

14. Moraes, C., Mehta, G., Lesher-Perez, S.C. & Takayama, S. Organs-on-a-chip: a focus on compartmentalized microdevices. Ann. Biomed. Eng. 40, 1211–1227 (2012).

15. Ghaemmaghami, A.M., Hancock, M.J., Harrington, H., Kaji, H. & Khademhosseini, A. Biomimetic tissues on a chip for drug discovery. Drug Discov. Today 17, 173–181 (2012).

16. Esch, M.B., King, T.L. & Shuler, M.L. The role of body-on-a-chip devices in drug and toxicity studies. Annu. Rev. Biomed. Eng. 13, 55–72 (2011).

17. Jang, K.J. et al. Fluid-shear-stress-induced translocation of aquaporin-2 and reorganization of actin cytoskeleton in renal tubular epithelial cells. Integr. Biol. 3, 134–141 (2011).

18. Jang, K.J. & Suh, K.Y. A multi-layer microfluidic device for efficient culture and analysis of renal tubular cells. Lab Chip 10, 36–42 (2010).

19. Jang, K.J. et al. Human kidney proximal tubule-on-a-chip for drug transport and nephrotoxicity assessment. Integr. Biol. 5, 1119–1129 (2013)

20. Carraro, A. et al. In vitro analysis of a hepatic device with intrinsic microvascular-based channels. Biomed. Microdevices 10, 795–805 (2008).

21. Kane, B.J., Zinner, M.J., Yarmush, M.L. & Toner, M. Liver-specific functional studies in a microfluidic array of primary mammalian hepatocytes. Anal. Chem. 78, 4291–4298 (2006).

22. Khetani, S.R. & Bhatia, S.N. Microscale culture of human liver cells for drug development. Nat. Biotechnol. 26, 120–126 (2008).

23. Lee, P.J., Hung, P.J. & Lee, L.P. An artificial liver sinusoid with a microfluidic endothelial-like barrier for primary hepatocyte culture. Biotechnol. Bioeng. 97, 1340–1346 (2007).

24. Nakao, Y., Kimura, H., Sakai, Y. & Fujii, T. Bile canaliculi formation by aligning rat primary hepatocytes in a microfluidic device. Biomicrofluidics 5, 22212 (2011).

25. Griep, L.M. et al. BBB ON CHIP: microfluidic platform to mechanically and biochemically modulate blood-brain barrier function. Biomed. Microdevices 15, 145–150 (2013).

26. Khademhosseini, A. et al. Microfluidic patterning for fabrication of contractile cardiac organoids. Biomed. Microdevices 9, 149–157 (2007).

27. Grosberg, A., Alford, P.W., McCain, M.L. & Parker, K.K. Ensembles of engineered cardiac tissues for physiological and pharmacological study: heart on a chip. Lab Chip 11, 4165–4173 (2011).

28. Feinberg, A.W. et al. Muscular thin films for building actuators and powering devices. Science 317, 1366–1370 (2007).

29. Nawroth, J.C. et al. A tissue-engineered jellyfish with biomimetic propulsion. Nat. Biotechnol. 30, 792–797 (2012).

30. McCain, M.L., Sheehy, S.P., Grosberg, A., Goss, J.A. & Parker, K.K. Recapitulating maladaptive, multiscale remodeling of failing myocardium on a chip. Proc. Natl. Acad. Sci. USA 110, 9770–9775 (2013).

31. Giridharan, G.A. et al. Microfluidic cardiac cell culture model (muCCCM). Anal. Chem. 82, 7581–7587 (2010).

32. Wilson, K., Das, M., Wahl, K.J., Colton, R.J. & Hickman, J. Measurement of contractile stress generated by cultured rat muscle on silicon cantilevers for toxin detection and muscle performance enhancement. PLoS ONE 5, e11042 (2010).

33. Esch, M.B. et al. On chip porous polymer membranes for integration of gastrointestinal tract epithelium with microfluidic ‘body-on-a-chip’ devices. Biomed. Microdevices 14, 895–906 (2012).

34. Imura, Y., Asano, Y., Sato, K. & Yoshimura, E. A microfluidic system to evaluate intestinal absorption. Anal. Sci. 25, 1403–1407 (2009).

35. Kimura, H., Yamamoto, T., Sakai, H., Sakai, Y. & Fujii, T. An integrated microfluidic system for long-term perfusion culture and on-line monitoring of intestinal tissue models. Lab Chip 8, 741–746 (2008).

36. Mahler, G.J., Esch, M.B., Glahn, R.P. & Shuler, M.L. Characterization of a gastrointestinal tract microscale cell culture analog used to predict drug toxicity. Biotechnol. Bioeng. 104, 193–205 (2009).

37. Sung, J.H., Yu, J., Luo, D., Shuler, M.L. & March, J.C. Microscale 3-D hydrogel scaffold for biomimetic gastrointestinal (GI) tract model. Lab Chip 11, 389–392 (2011).

38. Wang, L., Murthy, S.K., Barabino, G.A. & Carrier, R.L. Synergic effects of crypt-like topography and ECM proteins on intestinal cell behavior in collagen based membranes. Biomaterials 31, 7586–7598 (2010).

39. Puleo, C.M., McIntosh Ambrose, W., Takezawa, T., Elisseeff, J. & Wang, T.H. Integration and application of vitrified collagen in multilayered microfluidic devices for corneal microtissue culture. Lab Chip 9, 3221–3227 (2009).

40. Chung, S. et al. Cell migration into scaffolds under co-culture conditions in a microfluidic platform. Lab Chip 9, 269–275 (2009).

41. Shin, Y. et al. Microfluidic assay for simultaneous culture of multiple cell types on surfaces or within hydrogels. Nat. Protoc. 7, 1247–1259 (2012).

42. Sung, K.E. et al. Control of 3-dimensional collagen matrix polymerization for reproducible human mammary fibroblast cell culture in microfluidic devices. Biomaterials 30, 4833–4841 (2009).

43. Sung, K.E. et al. Transition to invasion in breast cancer: a microfluidic in vitro model enables examination of spatial and temporal effects. Integr. Biol. 3, 439–450 (2011).

44. Zervantonakis, I.K., Kothapalli, C.R., Chung, S., Sudo, R. & Kamm, R.D. Microfluidic devices for studying heterotypic cell-cell interactions and tissue specimen cultures under controlled microenvironments. Biomicrofluidics 5, 13406 (2011).

45. Achyuta, A.K. et al. A modular approach to create a neurovascular unit-on-a-chip. Lab Chip 13, 542–553 (2013).

46. Park, J., Koito, H., Li, J. & Han, A. Multi-compartment neuron-glia co-culture platform for localized CNS axon-glia interaction study. Lab Chip 12, 3296–3304 (2012).

47. Prabhakarpandian, B. et al. SyM-BBB: a microfluidic Blood Brain Barrier model. Lab Chip 13, 1093–1101 (2013).

48. Ma, S.H., Lepak, L.A., Hussain, R.J., Shain, W. & Shuler, M.L. An endothelial and astrocyte co-culture model of the blood-brain barrier utilizing an ultra-thin, nanofabricated silicon nitride membrane. Lab Chip 5, 74–85 (2005).

49. Allen, J.W. & Bhatia, S.N. Formation of steady-state oxygen gradients in vitro: application to liver zonation. Biotechnol. Bioeng. 82, 253–262 (2003).

50. Kim, S.H., Kang, J.H., Chung, I.Y. & Chung, B.G. Mucin (MUC5AC) expression by lung epithelial cells cultured in a microfluidic gradient device. Electrophoresis 32, 254–260 (2011).

51. Torisawa, Y.S. et al. Microfluidic platform for chemotaxis in gradients formed by CXCL12 source-sink cells. Integr. Biol. 2, 680–686 (2010).

©20

13 N

atur

e A

mer

ica,

Inc.

All

righ

ts r

eser

ved.

PROTOCOL

NATURE PROTOCOLS | VOL.8 NO.11 | 2013 | 2157

52. Douville, N.J. et al. Combination of fluid and solid mechanical stresses contribute to cell death and detachment in a microfluidic alveolar model. Lab Chip 11, 609–619 (2011).

53. Huh, D. et al. Acoustically detectable cellular-level lung injury induced by fluid mechanical stresses in microfluidic airway systems. Proc. Natl. Acad. Sci. USA 104, 18886–18891 (2007).

54. Huh, D. et al. Reconstituting organ-level lung functions on a chip. Science 328, 1662–1668 (2010).

55. Huh, D. et al. A human disease model of drug toxicity-induced pulmonary edema in a lung-on-a-chip microdevice. Sci. Transl. Med. 4, 159ra147 (2012).

56. Kim, H.J., Huh, D., Hamilton, G. & Ingber, D.E. Human gut-on-a-chip inhabited by microbial flora that experiences intestinal peristalsis-like motions and flow. Lab Chip 12, 2165–2174 (2012).

57. Kim, H.J. & Ingber, D.E. Gut-on-a-Chip microenvironment induces human intestinal cells to undergo villus differentiation. Integr. Biol. 5, 1130–1140 (2013).

58. Ochs, M. et al. The number of alveoli in the human lung. Am. J. Respir. Crit. Care Med. 169, 120–124 (2004).

59. Toepke, M.W. & Beebe, D.J. PDMS absorption of small molecules and consequences in microfluidic applications. Lab Chip 6, 1484–1486 (2006).

60. Wang, J.D., Douville, N.J., Takayama, S. & ElSayed, M. Quantitative analysis of molecular absorption into PDMS microfluidic channels. Ann. Biomed. Eng. 40, 1862–1873 (2012).

61. Gomez-Sjoberg, R., Leyrat, A.A., Houseman, B.T., Shokat, K. & Quake, S.R. Biocompatibility and reduced drug absorption of sol-gel-treated poly(dimethyl siloxane) for microfluidic cell culture applications. Anal. Chem. 82, 8954–8960 (2010).

62. Borysiak, M.D. et al. Simple replica micromolding of biocompatible styrenic elastomers. Lab Chip 13, 2773–2784 (2013).

63. Domansky, K. et al. Clear castable polyurethane elastomer for fabrication of microfluidic devices. Lab Chip 13, 3956–3964 (2013).