A lung-on-chip model reveals an essential role for alveolar epithelial cells in controlling bacterial growth during early tuberculosis

Vivek V. Thacker, Neeraj Dhar, Kunal Sharma, Riccardo Barrile, Katia Karalis, John D. McKinney

Preprint posted on 14 June 2020

Article now published in eLife at

Surfactant secreted by alveolar cells protects from Mycobacterium tuberculosis

Selected by Snehal Kadam

Context and background: Tuberculosis is a major healthcare burden, affecting a large proportion of the world (10 million individuals affected worldwide in 2018 (WHO)). A respiratory infection caused by Mycobacterium tuberculosis (Mtb), it begins with host-pathogen interactions at the air-liquid interface in the lung in early stages. Pulmonary surfactant, a mixture of lipids and proteins secreted by alveolar cells, has been traditionally known for its role in reducing surface tension and contributing to overall lung alveolar function [1]. Due to the fatality of surfactant deficiency and complexity of animal models, it has been difficult to probe the effects of surfactant on host-pathogen interactions. Organ-on-chip models combine the simplicity and feasibility of in vitro models with the ability to recapitulate physiologically relevant conditions like in in vivo models. This preprint looks at the role of surfactant released from alveolar epithelial cells using a model that recapitulates the air-liquid interface of lungs on a chip.


Experimental setup: The lung-on-chip model was made using polydimethylsiloxane (PDMS, a silicone-based organic polymer widely used for microfluidics) and had two chambers – air (alveolar) and liquid (vasculature) – separated by a porous membrane (see figure). The alveolar chamber contained alveolar epithelial cells (AECs), and the vascular chamber had endothelial cells. GFP-labelled macrophages were also added to the alveolar chamber, which could transverse the porous membrane to the endothelial side. Use of both AECs and macrophages allows one to study infection of both cell types. The chip was infected with fluorescently labelled Mycobacterium tuberculosis, WT, attenuated or avirulent mutants. Infection was visualized using time-lapse microscopy.

Schematic representation of the lung-on-chip infection model used in this study h1 = 1 mm, h2 = 250 μm, h3 = 800  μm. (Figure reproduced without modification from Figure 1G of the preprint Thacker et al. 2020 under a CC-BY-NC-ND 4.0 International license).

Important Results:

The lung-on-chip model recapitulates infection – This model makes use of two kinds of AECs – freshly isolated AECs which produce normal surfactant levels (NS), or in vitro passaged AECs (6-11 times) which have deficient surfactant levels (DS). These phenotypes were maintained in the chip, thus allowing  the study of endogenous pulmonary surfactant. Addition of Mtb to the chip lead to both cell types getting infected.

Surfactant reduces growth rates of Mtb – The intracellular growth rate of Mtb post infection was monitored over five days. Though highly variable, a pattern emerged with respect to the surfactant conditions. NS conditions reduced growth rates of Mtb, with a larger proportion of cells showing very slow growth (non-growing fraction) when compared to DS conditions. This was evident in both AECs and macrophages. Exogenous addition of a surfactant solution in the DS condition leads to an increase in the non-growing fraction of cells, restoring control of Mtb growth. This indicates that surfactant does play a role in host-pathogen interactions, by protecting the host cells.

An attenuated strain, deficient for ESX-1 type III secretion system, is unable to grow irrespective of surfactant conditions, indicating that ESX-1 is still required in DS conditions. Similarly, another attenuated strain, previously shown to be unable to grow in lungs of mice, showed a similar phenotype in the AECs and macrophages of the chip model. This underscores the importance and ability of the lung-on-chip model to reproduce similar phenotypes as animal models.

Surfactant plays a role in attenuation of growth by removing virulence-associated lipids of Mtb

Surfactant solution was seen to coat the bacteria upon exposure (but this effect was heterogenous). Additionally, sulfoglycolipids and trehalose dimycolate were partially removed from the surface of Mtb cells. These lipids are known to be important for virulence in inducing granulomas in mice and inhibiting cytokine production and immune response [2,3].


What I found interesting about this preprint:

The use of organ-on-chip approach enabled the study of endogenously secreted surfactant from alveolar cells, while maintaining the physiology and microenvironment of the air-liquid interface of lungs. Organ-on-chip models are gaining popularity in the field of infection research due to the ability to dissect host-microbe interactions at higher resolutions compared to animal models. I think this study makes a strong case for the use of such models, reproducing phenotypes and features known from animal models.

This study also shows that surfactant does more than its widely known role of reducing surface tension in the lungs, as it also protects the host again a pulmonary infection-causing pathogen. It would be interesting to see if such a protective role is seen against other pulmonary pathogens as well.


Questions for the authors:

  1. Where do you think the heterogeneity in coating of the Mtb surface with surfactant arises from? Does the correlate with any heterogeneity in the cell surface proteins of Mtb?
  2. Did you look at bacteria that cross over from the porous membrane into the endothelial side?


References/Further Reading:

[1] Torrelles, J. B., & Schlesinger, L. S. (2017). Integrating lung physiology, immunology, and tuberculosis. Trends in microbiology25(8), 688-697.

[2] Hunter, R. L., Olsen, M., Jagannath, C., & Actor, J. K. (2006). Trehalose 6, 6′-dimycolate and lipid in the pathogenesis of caseating granulomas of tuberculosis in mice. The American journal of pathology168(4), 1249-1261.

[3] Blanc, L., Gilleron, M., Prandi, J., Song, O. R., Jang, M. S., Gicquel, B., … & Vercellone, A. (2017). Mycobacterium tuberculosis inhibits human innate immune responses via the production of TLR2 antagonist glycolipids. Proceedings of the National Academy of Sciences114(42), 11205-11210.

Tags: lung-on-chip, mycobacterium, tuberculosis

Posted on: 15 July 2020 , updated on: 16 July 2020


Read preprint (No Ratings Yet)

Author's response

Dr. Vivek Thacker shared

Dear Snehal,

Thank you very much for your interest in our work and we are delighted you have chosen to highlight our pre-print and summarize it in a clear and concise manner. Our vision for this work was to establish the lung-on-chip developed at Emulate Inc as a new model for chronic respiratory infections such as Tuberculosis, and to use this model to make measurements and to perform experiments that cannot be done in an animal model. To that end, we show that it is compatible with time-lapse imaging over many days, which provides information on bacterial growth rates at the very early stages of  infection that would not be provided by traditional colony forming  unit measurements. We also use the modular nature of this model to recreate an extreme surfactant deficiency and directly probe the role of this component in early Tuberculosis. The insights we have gained from this work can now be translated to questions that we can address in a more complete animal model ( e.g. can treatments that improve surfactant levels attenuate bacterial growth in the mouse) or in simpler systems (e.g. how do surfactant lipids interact with the mycobacterial cell surface lipids and proteins).
As you correctly point out, organ-on-chip and organoid approaches are rapidly being applied to all areas of biology. Infection biology is a challenging field to apply these approaches for two reasons. First, there are biosafety considerations and restrictions, which often limit  the degree of experimental manipulation possible. For example, in our study, we were restricted to widefield imaging and did not have access to the excellent core microscopy facilities at EPFL. Two, it is hard to set good benchmarks because the ‘system’ is stressed for the entire time. As an example, cell death could be a consequence of the infection, or the host response, or merely because of experimental errors. We believe these challenges only reinforce the potential impact such models and approaches can have in infection biology and future developments will extend them further by increasing physiological mimicry in terms of physiology and cell types, the throughput, and the range of readouts possible.
The questions you raise are both very interesting and pertinent and will inspire the next stage of inquiry. The direct interaction of surfactant with Mtb lipids probably depends to a large extent on the conformations of the lipids and the biophysics of lipid-lipid interactions. Pulmonary surfactant has been shown to have multiple lamellar phases at the air-liquid interface, and mycobacterial cell surface lipids such as TDM are also either in an extended or relaxed
configuration based on surface tension and chemical composition. The importance of developing a complete biophysical understanding was neatly demonstrated in a paper by Augenstreich et al (PNAS 2019), where they showed that PDIM adopts a conical shape and
accumulates in host lipid bilayers, thus altering host signalling pathways. It may well be that similar interactions occur between surfactant and Mtb lipids,and that a better understanding
would allow us to tune these interactions in a way to render them more host protective.

We did see some instances of infection transferring over to the vascular side via migration of macrophages indicating that this is possible even at a very early stage of infection. However, this was not the outcome for most infected cells. This could be reflective of what occurs in vivo or might be different because the distance between the epithelial and endothelial faces is about 10x larger on-chip than in the alveolus in vivo, and migration is limited to the pores in the membrane. Alternatively, the lack of a proper interstitial space in these chips may alter the migratory profiles of macrophages.


Have your say

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Register here