Animal models have undoubtedly been essential for the development of oral inhalation drug products, especially for demonstrating safety as well as, at least for some diseases, also efficacy in preclinical research. But it must be realized that, already in healthy state, animal models hardly reflect the human respiratory tract with regard to the administration and deposition of aerosolized medicines. While forced inhalation or tracheobronchial instillation may still allow to draw some conclusion about pulmonary toxicity, the problem becomes more challenging for efficacy studies. This is especially true for inhalable anti-infective drugs, where the available animal models fail to adequately replicate how such diseases affect the human respiratory tract (Lorenz et al., 2016). This can be attributed to evident species-species variations between humans and model organisms (e.g., in lung anatomy, airway histology, cellular composition of epithelial and sub epithelial compartments) that amongst other reasons eventually slow down the development of orally inhaled drug products (Barnes et al., 2015; Artzy-Schnirman et al., 2019a; Jimenez-Valdes et al., 2020).

In contrast to animal models, complex in vitro models, especially when human based, allow to focus on key elements of underlying (patho-) physiological conditions as observed in the clinic and to model such conditions within a controlled in vitro environment (Carius et al., 2021). Because at present no such predictive in vitro models are available yet, their technological development and subsequent validation represent important and demanding scientific tasks (European Commission Joint Research Centre, 2021). Pulmonary in vitro models thereby profited from utilizing well-based permeable growth supports (e.g., Transwell^®) as a cell culture environment, because these substrates most importantly enable the establishment of air-liquid interface (ALI) conditions as well as polarized differentiation of pulmonary epithelial cells (Lacroix et al., 2018). The easy accessibility to the apical as well as to the basolateral compartment further supports aerosol deposition and/or permeability studies, along with the biophysical measurement of barrier properties via transepithelial electrical resistance (TEER). Various simple pulmonary in vitro models, usually consisting of bronchial or alveolar epithelial cell monocultures, but also complex models comprising pulmonary epi- and/or endothelial cells in co-culture with other cell types like immune cells (e.g., dendritic cells, macrophages or neutrophils) or fibroblasts have been extensively reviewed (Gordon et al., 2015; Hittinger et al., 2015; Hittinger et al., 2017; Ehrmann et al., 2020).

In moving beyond simple Transwell^® cultures that are mainly limited by static culture conditions, organ-on-chip systems have advanced miniaturized biomimetic devices that allow the in vitro culture of human cells under physiological conditions similar to in vivo, including continuous perfusion and mechanical deflection (Tenenbaum-Katan et al., 2018). Lung-on-chip devices may moreover replicate the characteristics of specific regions of the lung (e.g., cell composition, exposure to air (flow), or breathing dynamics) and comprise tissue relevant cell types. Among the earliest efforts, Nalayanda et al. (2009) described the ALI culture of the ATII-like carcinoma cell line A549 under low flow conditions (0.35 μl/min) in a PDMS-based lung-on-chip for up to 3 weeks. Following the pioneering work of Huh et al. (2010), that introduced an alveolar-capillary model able to co-culture pulmonary epithelial cells under ALI and flow conditions together with endothelial cells, both stretched by cyclic mechanical strain, the field rapidly advanced. Punde et al. (2015) followed the concept of a dynamic Transwell®-like device, thereby showing that the implementation of dynamic flow created a concentration gradient that effectively guided the transmigration of fibrocytes in an inflammation lung model. An anatomically inspired true-scale acini-on-chip allowed the investigation of the immune response of alveolar epithelial cells in co-culture with differentiated THP-1 macrophage-like cells to nebulized lipopolysaccharide (LPS) as a surrogate for bacterial infections (Artzy-Schnirman et al., 2019a). Recently, the same group used a branching airway-on-chip platform to realistically mimic the transport and deposition of aerosolized particulate matter and study its cytotoxic effect on normal human bronchial epithelial (NHBE) cells (Elias-Kirma et al., 2020). Other devices worth mentioning enable the displacement of a flexible membrane in a diaphragm-like motion (Stucki et al., 2015) in combination with perfusion (Cei et al., 2020) as well as the membrane-free culture of airway smooth muscle cells in co-culture with epithelial cells embedded within a hydrogel (Humayun et al., 2018) or the culture of human alveolar epithelial cells in a gelatin methacryloyl hydrogel resembling alveoli-like hemispheres in a breathing lung-on-chip (Huang et al., 2021).

Despite such advances, the lung-on-chip models described above are technically cumbersome to manufacture and unfortunately need either sophisticated microfabrication techniques, stemming from academic labs with a high level of bioengineering expertise and equipment, or are too costly to be introduced in a standard bio-/pharmaceutical laboratory. Inspired by these devices, we here provide the blueprint as well as a technical proof of principal study of a novel versatile perfusable platform to assess permeability and barrier function of air exposed pulmonary epithelia (PerfuPul), using the bronchial carcinoma cell line Calu-3. Furthermore, to mimic chronic infections of epithelial cells and to enable repetitive treatment of these cultures in vitro with aerosolized drug products, survival times should be spanning multiple days, ideally even weeks. Hence, we hypothesized that constant perfusion of cell culture medium could not only prolong the survival time of such complex infected co-culture models, by removal of bacterial toxins and virulence factors, but also accelerate and enhance cellular differentiation, as supported by others (Chandorkar et al., 2017). To this end, we identified ALI conditions, aerosol deposition, TEER measurements and especially perfusion as a physiological relevant clearance mechanism as needed prerequisites for the intended infection studies. Our platform fulfills the technical requirements for the intended infection models and can easily be reproduced in any bio-/pharmaceutical laboratory with a moderate time as well as financial investment.

In spite of their physiological advantages, perfusable transport chambers have only rarely found their way into in vitro models of epithelial cell culture, in particular for the lungs (Artzy-Schnirman et al., 2020). Complex designs, mostly encompassing micron-sized rectangular channels, as well as the underlying problems of impaired aerosol deposition and/or complexity of handling by non-experts, might restrict the exploitation of the full potential of lung-on-chip devices as preclinical research tools (Junaid et al., 2017; Ehrmann et al., 2020). Perfusable systems offer particular advantages for studying infectious diseases, as concentration of e.g., bacterial virulence factors can be kept at lower levels by continuous supply and dilution with fresh media. Before this can be studied in more detail, however, it is necessary to ensure to what extent data generated on novel perfusabel setups can be compared to those obtained on established Transwells^® under static conditions (Ainslie et al., 2019). We therefore here first describe the development as well as the characterization of a simple perfusable platform for pulmonary epithelial cell cultures at ALI conditions. This platform maintains the advantages of traditional Transwell®-based systems, but adds the possibility to implement perfusion as a prerequisite to develop longer-lasting in vitro models of chronic pulmonary diseases in the future. In addition, we want to share the protocol needed to produce the platform with a broad scientific community, in order to make this technology available to end-users that might not yet be familiar with microfabrication techniques.

Apart from standardized readouts to assess cell viability or cytotoxicity in vitro, such as the MTT or lactate dehydrogenase (LDH) assay respectively, information from TEER measurements or permeability studies appear extremely helpful for the preclinical evaluation of orally inhaled drug products (Wilson 2000; Hittinger et al., 2017). For such purposes, these models must allow ALI conditions together with the subsequent deposition of aerosols. Although pulmonary epithelial cells grown on Transwell^® inserts meet all of these technical specifications, they are limited to static culture conditions. Recent studies, however, demonstrated that normal human bronchial epithelial (NHBE) cells cultured under ALI conditions and simultaneous perfusion of cell culture medium showed improved barrier properties in comparison to static conditions (Chandorkar et al., 2017; Bovard et al., 2018). Perfusion is, amongst other physiological important mechanical stimuli, also recognized as a central element to be implemented in organ-on-chip systems in general to improve cellular development (Thompson et al., 2020).

The new platform thus was designed to comply with the familiar technical standards and features of Transwell^® inserts, but at the same time also enabling perfusion as a physiological relevant clearance mechanism not present in Transwell®-based culture plates. Major specifcations are: 1) a comparable surface area (24-well based Transwell^®: 0.33 cm²; perfusable platform: 0.28 cm²), 2) the possibility to routinely inspect cell layers under a microscope while maintaining sterile culture conditions, 3) the option to measure TEER, and 4.) the implementation of ALI conditions as well as subsequent aerosol deposition (Figures 1, 3, 5; Supplementary Figure S1). The access to the basolateral compartment via a combination of a channel-based inlet connected to a flexible tubing enables the culturing of cells under static culture conditions, but also adds the possibility to perform experiments under perfusion (Figure 7).

In a side-by-side comparison using Calu-3 cells grown on regular Transwell^® inserts or within the perfusable platform we could show that both, the resistance readings from the two electrodes (chopstick vs. custom-made electrode) within the same Transwell^® insert as well as the measurements performed in the perfusable platform and on Transwell^® inserts under equal experimental conditions agree by good approximation (Figure 3C, D). Notably, this was irrespective of the fact that in the case of the custom electrode, when compared with the chopstick electrode within the same Transwell^®, four separate cables needed to be arranged in each well. The custom-made Ag/AgCl electrode used by us can be connected to existing instrumentation (e.g., EVOM 2) with only little technical effort needed to solder the different silver wires to a 4-cord cable (Figure 3A), as already described for organ-on-chip systems in similar ways (Douville et al., 2010; Ferrell et al., 2010; Huh et al., 2010; Kim et al., 2012; Griep et al., 2013). Most of these studies reported significantly higher TEER values in case of the organ-on-chip systems when compared to results obtained for the same cell models cultured on Transwell^® inserts. However, these higher TEER values could be related to the geometry of the rectangular micron-sized channels within these organ-on-chips, that generate a non-uniformly distributed electrical current, rather than showing a biological effect (Odijk et al., 2015). The perfusable platform reported in the present paper, however, is based on a design encompassing two equally sized wells separated by a circular membrane that has a cell culture area identical to that of a 24-well Transwell^® insert (Figure 1A). Limiting the use of micron-sized channels only to the in- and outlet that enable access to the basolateral well, further generates similarity to the Transwell^®.

Especially during permeability studies within the same lab, TEER value measurements using an EVOM 2 or similar direct current based Voltohmmeters are widely accepted as a valuable non-invasive readout to routinely assess barrier integrity. This holds still true while some more advanced bioelectrical methods (e.g., impedance spectroscopy) might be more precise in determining information about actual barrier integrity (Günzel et al., 2012; Odijk et al., 2015; Srinivasan et al., 2015; Henry et al., 2017).

As already indicated by the stable TEER values during the course of the 7 h transport experiments, the apparent permeability of fluorescein sodium transported under LCC also did not reveal any significant differences between the perfusable platform and the Transwell^® (Figure 6B). Further were the obtained Papp values in this study, in relation to the respective TEER values, consistent with reported Papp values in the literature performed under similar study conditions on Transwell^® inserts using Calu-3 cells (Ehrhardt et al., 2002; Fiegel et al., 2003; Grainger et al., 2006; Haghi et al., 2010). To our knowledge, these historical data from Transwells^® inserts have not been confirmed using any perfusable, microfluidic or lung-on-chip system so far.

To assess barrier integrity by staining for relevant cellular markers, techniques to perform immunofluorescence staining and subsequent confocal laser scanning microscopy or other imaging methods can be used with the perfusable platform in the same way as for Transwell^® inserts. This was demonstrated by the example of the visualization of the proteins Occludin and ZO-1 that are involved in the formation of functional tight junctions (Figure 4). For that purpose, the apical as well as basolateral entity can be separated from the membrane, independently from each other. This procedure allows to flexibly conduct staining and cell fixation methods.

One of the unique features of the perfusable platform as presented here is the possibility to reproducibly deposit pre-metered aerosols from aqueous drug solutions on pulmonary epithelial cells grown at an ALI using a clinically relevant vibrating mesh nebulizer (Aeroneb^® Lab). The removable cover glass, that allows to easily open and close the top of the apical compartment, thereby enables the establishment of ALI conditions after cells have been grown to confluency under LCC. The flexibility to open and close the top of the apical compartment is what creates the possibility to connect a vibrating mesh nebulizer to the perfusable platform by using the custom-made deposition chamber. The custom-made deposition chamber presented here was modified to fit the opening of the apical compartment of the perfusable platform, but mostly retains the design as well as dimensions of the device described by Horstmann et al. (2021) to be used with a single 12-well (12 mm) Transwell^® insert or a single 24-well. Since we extensively discussed the obtained results and characterized the deposition chamber in the work mentioned above, we limited the characterization for its use with the perfusable platform to the experiments described in Figure 5, to ensure that reproducibility of results is consistent with our previous findings. While reproducibility could be maintained in good approximation (5.6% relative standard deviation (deposition on perfusable platform) vs. 4.8% relative standard deviation [12-well Transwell^®)], the deposition efficiency showed a 4-fold reduction [1.06 ± 0.06% (perfusable platform) vs. ∼4% (12-well Transwell^®)]. When all parameters, such as the used nebulizer, the volume to be nebulized in the same concentration and the settling time are kept constant, the deposition efficiency in these chambers is mainly limited by the inner diameter of the outlet (Horstmann et al., 2021). The inner diameter amounts to ∼5 mm in the version for the perfusable platform and ∼11 mm for the 12-well Transwell^® version. A substantial amount of the aerosol mist generated by the vibrating mesh nebulizer thus deposits on the walls of the deposition chamber and therefore is not channeled through the outlet. A similar concept, also including a vibrating mesh nebulizer and a connected chamber, was recently introduced by Cei et al. (2020) in a dynamic in vitro stretch lung model. The outlet of the aerosol chamber in their model was about 20 mm in inner diameter and resulted in an impressive deposition efficiency of about 52% after nebulization.

Although we acknowledge the fact that a deposition efficiency of 1% from the invested dose seems relatively low, the dose delivered per surface area provides better comparability in this case. Based on an equally invested dose of 20 µg (20 µl from a 1 mg/ml solution) which would be nebulized, the delivered dose per surface area would result to ∼0.7 μg/cm² (perfusable platform, 0.28 cm²; deposition efficiency: 1%), ∼0.8 μg/cm² (12-well Transwell^®, 1.12 cm²; deposition efficiency: 4.8%) or ∼2.2 μg/cm² [(Cei et al., 2020), 4.67 cm²; deposition efficiency: 52%]. Ultimately, between different devices the dose delivered per surface area should be used for comparison instead of the isolated deposition efficiency. This considerably mitigates the major differences observed when focusing on the isolated deposition efficiency while enhancing comparability, despite the fact that the delivered dose per surface area is still about three times higher in the device of Cei et al. (2020) when compared to the perfusable platform, since less compound is deposited on the walls of the deposition chamber. The foregoing notwithstanding, although the main share of compound is deposited on the walls of the deposition chamber and not reaching the cellular layer, the quantity of compound invested to obtain meaningful in vitro results with the devices mentioned before is still substantially lower than that normally required for conducting in vivo inhalation studies (Phalen and Mendez 2009).

As demonstrated by the data of Figure 7, transport studies under static as well as perfused conditions were successfully conducted after the reproducible deposition of a pre-metered aerosol from an aqueous fluorescein sodium solution. Interestingly enough, under perfusion we only observed a negligible decrease in TEER values (−2%) after 5 h of transport after ALI deposition of fluorescein sodium under perfusion (Figure 7B) while the difference after 5 h of transport under static conditions amounted to (−26%) when comparing only the mean values. Although this difference was not statistically supported, it could suggest a positive effect of perfusion on barrier properties during the course of the transport experiment in comparison to static conditions, but would need more thorough investigation. The ability to implement perfusion into the perfusable platform, further allowed the comparison of transport of nebulized fluorescein sodium as a model compound under static conditions to transport under perfusion. Although we also could only observe a slight trend towards higher C_max and lower t_max in case of the fluorescein sodium transported under continuous perfusion, we think that if such experiments would be performed under different parameters (shorter sampling intervals, drugs with different permeation, different flow rates etc.) meaningful insights into the transport of experimental drugs intended for inhalation under physiological relevant conditions could be provided.

We are well aware that the features that make up the perfusable platform are built upon the work of other researchers active in the lung-on-chip field. A similar concept comprising a PDMS-based chip which possessed an open apical compartment, thereby allowing the establishment of ALI conditions, which was also able to provide perfusion and TEER measurements was already introduced by Nalayanda et al., in 2009. Although the authors compared their findings systematically to Transwell^® inserts at an ALI using the alveolar type 2-like cell line A549, they did not show the deposition of aerosols. The importance of flow in lung-on-chip systems initially demonstrated by Punde et al. (2015), was used by others to follow the concept of a flowable Transwell^® (Blume et al., 2015; Blume et al., 2017; Chandorkar et al., 2017; Bovard et al., 2018; Schimek et al., 2020). The problem with such systems is that the Transwell^® inserts where the cells are cultured during perfusion need to be taken out of the sophisticated culture devices that provide the perfusion, and are then transferred to separate culture wells or devices before TEER measurements can be performed or aerosols could be deposited. Especially for substances that are rapidly absorbed after aerosol deposition, such delays could aggravate subsequent analysis. Sophisticated aerosol deposition on cells grown at an ALI was described by Artzy-Schnirman et al. (2019a) for a morphologically inspired acinus-on-chip and also recently for a bronchial bifurcation mimic by Elias-Kirma et al. (2020). Unfortunately, these devices were not characterized in terms of TEER measurements nor in the application of perfusion. In addition, the former device needs sophisticated microfabrication during production and the latter device requires a complex set of devices for the generation of aerosols as well as the related physiologically relevant airflows. Researchers that consider implementing lung-on-chip devices for their research or just want to try out if the physiological features provided by these devices, such as perfusion, could benefit their work, are thus facing considerable difficulties, if they do not have the needed technical expertise in micro-fabrication or the financial resources to invest in commercial platforms. Although companies like Alveolix, CN bio, Emulate, Kirkstall or TissUse introduced innovative lung-on-chip devices or perfusable Transwells^®, the material costs of the chips and the related infrastructure needed to operate them could add up to sums which are difficult to finance for academic labs (Kirkstall Ltd 2020; AlveoliX 2021; Chips - TissUse GmbH 2021; CN Bio Innovations 2021; Emulate 2021).

The perfusable platform presented here holds some limitations. The platform is based on a PET membrane, which is cut from Transwell^® inserts and known to be rather rigid and bio-inert in comparison to the extracellular matrix found in vivo (Humayun et al., 2018). The used PDMS is also known to absorb small, hydrophobic molecules (Toepke and Beebe 2006). This technical characteristic can be corrected though, by pre-equilibrating the devices or adjusting for the loss of substance (Jimenez-Valdes et al., 2020). Depending on the skill of the operator but also on the used peristaltic pump, results obtained from the perfusable platform are in the low- to mid-throughput range depending on the extent of the intended studies.

The minimalistic design of the presented perfusable platform “PerfuPul” allows its production by non-experts in most lab environments without the need for specialized equipment. Furthermore, the platform combines TEER measurements, aerosol deposition as well as the implementation of perfusion as a physiological relevant clearance mechanism in a single device that matches the design as well as the quality of results known from Transwell® inserts. The open apical entity not only allows the establishment of ALI conditions but could be also utilized for working with cutting-edge techniques such as 3D bio printing. Moreover, the easy access to the basolateral compartment would permit co-culture studies, for instance, by also growing endothelial cells on the basal side of the membrane. Taking all of these factors in consideration, we are convinced that this perfusable platform will enable the development of novel pulmonary in vitro models especially to study long-term diseases, such as e.g., bacterial lung infections and their treatment by aerosolized drugs and nanoscale carriers therof.