This article was written by Marjorie Metzger, and reviewed by Julie Cavallasca and Guilhem Velvé Casquillas.
Marjorie Metzger, master student at EPFL in life sciences, orientation neurosciences and neuroengineering.
Julie Cavallasca, scientific translator, studied languages and translation at Aix-Marseille University and Paris 7 Diderot University. She now works for for Elvesys and the NBIC Valley.
Dr. Guilhem Velvé Casquillas
Dr. Guilhem Velve Casquillas is the CEO of Elvesys and co-founded many innovative companies. He is a former cellular biology researcher at the Curie Institute and in microfluidics.
Recent research breakthroughs in lung-on-chip technology
Development of new therapeutic drugs is decreasing while research costs are rising. A study from the Tufts Center for the Study of Drug Development (USA)  stated an increase of the total capitalized costs of 8,5% per year (regardless of price inflation) with a consequent decreases in success rates. According to Marshall and Rowan, alternative methods can be a solution to find new treatments at lower costs and without the ethically questionable use of animal testing. In particular, organs-on-chip represent an innovative method that could favor new discoveries in health research like specified by Zhang & al . Different micromodels have already been elaborated for muscles, bones, blood brain barrier, gut, liver, as well as multi-organs systems.
It could be extremely interesting to build a human-on-chip (see Fig. 1) that will model the interactions between different organs, but it is also essential to develop simulations of tissue-tissue interfaces and more generally of local organ behavior. Fundamental exchanges happen by ways of the respiration process, especially on the capillary-alveolar interface. Being able to reproduce the lung properties on a micro-device will help to better understand its functions and can also allow to simulate pulmonary diseases.
Mouse models already exist that reproduce partially pulmonary injuries but microfluidics devices like organ-on-chips can model the mechanical and physiological responses of an organ or more specifically of a tissue-tissue interface exclusively. Moreover, access is easier in the case of a microchip model than for a living mouse. Lungs-on-chip could allow a more precise control of the cellular microenvironment in a way closer to the in vivo biochemical context than the other in vitro simulations.
Challenges in the reproduction of pulmonary thrombosis in a lung-on-chip
Huh & al.  used the organ-on-chip principles to represent the alveolar-capillary interface (see Fig. 2). Indeed, the alveolar-capillary gas exchanges are essential to the lung function. Being able to reproduce this lung unit can be helpful for nanotoxicology studies and disease simulations. According to Jain & al.  pulmonary thrombosis is a major cause of mortality in lung diseases like in ALI (acute lung injury). A blood clot (see Fig. 3) forms locally in a pulmonary vessel and can cause a respiratory failure due to the lack of blood flow in a part of the lung.
Research is ongoing to find proper models of thrombus formation in lung vessels with the aim of developing new therapeutic drugs targeting the inflammatory mechanisms underlying thrombi formation. However, the many animal models built to represent the formation of microvascular thrombosis in lungs are limited because they could not be exactly similar to the human lungs. Besides, experimenting on animals is under ethics debate. Research should therefore evolve toward efficient in vitro models. Jain & al. prospect the hypothesis that organs-on-chips can better represent the thrombosis mechanisms than the existing artificial models. Among other advantages, it allows easier reconfiguration and access thanks to its separate compartments.
Technical characteristics of the microdevice
- Alveolar representation
The Wyss Institute develops several pulmonary microchips –. Huh & al.  grew both alveolar epithelial and endothelial cells on each side of a PDMS (polydimethylsiloxane) membrane covered with an extracellular matrix (ECM) (see Fig. 4). In vivo, the ECM provides structural support and segregates tissues, the model should thus integrate them for intercellular communication and to avoid direct contact between cells of interest and the silicone wall. To reproduce at best the alveoli, the epithelium cells were in contact with air whereas the endothelial cells were in contact with a fluid mirroring blood properties. Two lateral chambers were also added to simulate the pressure changes in the alveoli. Applying vacuum to these chambers causes a deformation of the separating membrane and therefore models the expansion-contraction of the alveolus wall due to the physiological breathing movements.
- Pulmonary thrombosis model
Based on previous works on lung-on-a-chip microdevices  , the Wyss laboratory could elaborate an innovating alveolus-on-chip microdevice reproducing a pulmonary thrombosis. They modified the system described before by using primary human lung cells instead of culture cells. They increased the chambers dimensions to have more exposed area. Human vascular endothelial cells from umbilical cord (HUVEC) were also added all over the walls of the bottom chamber to be able to replace the fluid mimicking plasma with real human blood (see Fig. 5). In fact, in vivo, blood is not directly in contact with the extracellular matrix.
- Pulmonary thrombus formation process
Jain & al.  from the Wyss Institute first noticed that coating the four walls with ECM in the alveolus-on-a-chip model avoid platelet adhesion to the walls and thrombi formation. This microdevice can thus adequately represent a healthy pulmonary vessel behavior. To simulate an intravascular thrombosis, the TNF-α (Tumor necrosis factor) was added to the upper chamber. It is an inflammation-involved cytokine that recruits leukocytes and activates platelets, -which triggers thrombi formation . It was measured by Jain & al. that the permeability of lung vessels increases proportionally to the TNF-α concentration. They also measured a raise of ICAM-1 concentration on the endothelium chamber, below which follow by thrombi formation on the endothelial membrane. An in vitro model of endothelial thrombosis formation was therefore obtained by releasing TNF-α in the upper compartment.
Prof. Ingber laboratory team  observed a similar shape and binding dynamic for the thrombosis model-on-a-chip compared to the in vivo laser-induced thrombi formation in a mouse model. The new model established on a microchip also demonstrates another point generally observed in vivo but never seen previously in vitro: the intra-thrombus variations due to a central kernel surrounded by a more unstable region (see Fig. 6).
To better demonstrates the potentiality of organ-on-chip for research, Jain & al  tried to add LPS (lipopolysaccharides) to the lung-on-chip. LPS are endotoxins present on the outer membrane of Gram-negative bacteria. These large molecules are often used to induce pulmonary vascular injury because it promotes leukocytes recruitment  . At the tissue level represented by the presence of only the vascular endothelium, no thrombosis was observed. A contrario, at the organ-level simulated by a chip with both the epithelial and endothelial cells, intravascular platelets aggregates are observed. The gap areas between endothelial cells increase after the LPS treatment. Jain & al.  also found an increase of three specific cytokines following the LPS adding: IL-6 (interleukin), IL-8 and monocyte chemotactic-protein. They noticed similar physiological reactions in vivo: it is possible to increase the induction of lung vascular thrombosis by injecting LPS when the vessels are laser-injured.
Globally, the Wyss Institute demonstrated, using an alveolar microchip, that LPS-induced thrombosis is correlated with an inflammation process mediated by cytokines and induced by interactions between alveolar epithelial cells and vascular endothelial cells.
- Using the alveolus-on-a-chip to evaluate a potential antithrombotic treatment
Once they assessed the lung alveolus-chip as a proper model for thrombosis formation, they used it to test possible drug treatments. They choose to study the effects of PM-2 (Parmodulin-2) that have been shown in both in vivo and in vitro models to improve vascular fluidity and thus decrease the risk of thrombosis formation. Aisiku & al.  distinguished parmodulins, a family of PAR1 (Protease-activated receptor 1) antagonists, and compared them with others antagonists. They noticed that parmodulins reduce vascular inflammation without causing endothelial injury. Based on previous studies, Jain & al. decided to analyze the action of PM-2 in their alveolus chip compared to a mouse model. Effects of PM-2 in both models match. Therefore, they demonstrate that their lung alveolus-on-a-chip model could potentially be used for drug testing.
Further possibilities using a lung-on-chip
To conclude, it has been demonstrated that a lung alveolus-on-a-chip could properly represent the mechanical and physiological reaction of an in vivo alveolus. It allows an interesting real-time visualization of tissue-tissue interface. Moreover, the Wyss Institute  was able to integrate human blood flow to the microdevice to produce a model closer to the living processes. They noticed good correlations between their in vitro model and the real clotting dynamics in pulmonary vessels due to inflammations. Some aspects of the model can still be improved on because it does not consider the breathing movement. It was already done for the alveolus-capillary interface model by Huh & al. . The Wyss Institute also used HUVEC cells from umbilical cord whereas lung endothelial cells have some specificities, even if these endothelial cells are highly similar.
However, they could use the alveolus-on-a-chip to assess the therapeutic potentiality of a molecule. It creates opportunities for future drug assays at a lower cost and an extremely higher speed. The door is also open for the modeling of other pulmonary diseases by the way of the microchip techniques. Benam & al. (2015)  still used this method to develop a small airway-on-a-chip that they employ to mimic the physiological reactions happening during acute asthmatic crisis. Other pulmonary complications could also be analyzed in the future using this simple and efficient technique.
 J. A. DiMasi, H. G. Grabowski, and R. W. Hansen, “Innovation in the pharmaceutical industry: New estimates of R&D costs,” J. Health Econ., vol. 47, pp. 20–33, May 2016.
 B. Zhang and M. Radisic, “Organ-on-a-chip devices advance to market,” Lab Chip, vol. 17, no. 14, pp. 2395–2420, 2017.
 D. Huh, B. D. Matthews, A. Mammoto, M. Montoya-Zavala, H. Y. Hsin, and D. E. Ingber, “Reconstituting Organ-Level Lung Functions on a Chip,” Science, vol. 328, no. 5986, pp. 1662–1668, Jun. 2010.
 A. Jain et al., “Primary Human Lung Alveolus-on-a-chip Model of Intravascular Thrombosis for Assessment of Therapeutics,” Clin. Pharmacol. Ther., Jul. 2017.
 D. Huh et al., “A Human Disease Model of Drug Toxicity–Induced Pulmonary Edema in a Lung-on-a-Chip Microdevice,” Sci. Transl. Med., vol. 4, no. 159, p. 159ra147-159ra147, Nov. 2012.
 P. Pignatelli et al., “Tumor necrosis factor-α as trigger of platelet activation in patients with heart failure,” Blood, vol. 106, no. 6, pp. 1992–1994, Sep. 2005.
 M. Uchiba and K. Okajima, “Antithrombin III (AT III) Prevents LPS-Induced Pulmonary Vascular Injury: Novel Biological Activity of AT III,” Semin. Thromb. Hemost., vol. 23, no. 06, pp. 583–590, Dec. 1997.
 K. Murakami et al., “Activated protein C attenuates endotoxin-induced pulmonary vascular injury by inhibiting activated leukocytes in rats,” Blood, vol. 87, no. 2, pp. 642–647, 1996.
 O. Aisiku et al., “Parmodulins inhibit thrombus formation without inducing endothelial injury caused by vorapaxar,” Blood, vol. 125, no. 12, pp. 1976–1985, Mar. 2015.
 K. H. Benam et al., “Small airway-on-a-chip enables analysis of human lung inflammation and drug responses in vitro,” 2015.
 X. Li, S. Xu, P. He, and Y. Liu, “In Vitro Recapitulation of Functional Microvessels for the Study of Endothelial Shear Response, Nitric Oxide and [Ca2+]i,” PLOS ONE, vol. 10, no. 5, p. e0126797, May 2015.
 S. Rajwal, M. Richards, and M. O’Meara, “The use of recalcified citrated whole blood – a pragmatic approach for thromboelastography in children,” Pediatr. Anesth., vol. 14, no. 8, pp. 656–660, Aug. 2004.
Written by Marjorie Metzger, corrected by Julie Cavallasca, under the supervision of Dr. Guilhem Velve Casquillas