The curious case of how mimicking physiological complexity in in vitro models of the human respiratory system influences the inflammatory responses. A preliminary study focused on gold nanoparticles

Environmental and biomedical nanoparticles can pose potential health risks to the human respiratory system by inducing severe lung inflammation. The aim of this case study is to present a comparison of the inflammatory response in four in vitro models of the human lung epithelium, differing by composition and/or culturing substrates, when exposed to gold nanoparticles (AuNPs). Three in vitro models of lung adenocarcinoma (A549) cells and a commercially available three‐dimensional (3D) culture (MucilAir™) were tested. The models were exposed to AuNPs for 3, 6, and 24 h. AuNPs internalisation was investigated by confocal, electron microscopy, and Raman spectroscopy. Enzyme‐Linked Immuno‐Sorbent Assay (ELISA) was used for quantifying the secretion of the inflammatory mediator Interleukin‐6 (IL‐6) following exposure to AuNPs. Finally, a microfluidic approach was developed in‐house to investigate whether pro‐inflammatory mediators present in supernatants harvested from the AuNPs‐treated cell cultures could trigger monocyte activation. Our results demonstrated that AuNPs were internalised only in submerged cultures grown on glass substrates. Nevertheless, AuNPs internalisation did not trigger a significant IL‐6 secretion. Significant amounts of IL‐6 were secreted by AuNPs‐treated mono‐cultures grown on Transwell™ inserts, triggering monocyte activation in dynamic microfluidic experiments. AuNPs did not induce IL‐6 secretion in co‐cultures and MucilAir™ models, although supernatants harvested from co‐cultures triggered monocyte activation. Our case study demonstrates that in vitro complexity, as well as culturing substrates, deeply influence the detectable cellular responses to nanoparticles, and advocate for the adoption of more advanced tissue‐mimetic cultures of the human respiratory system for nanomaterials testing.


Introduction
In the last decades, nanotechnology has found application in many medical and industrial fields, raising enthusiasm but also concerns associated with the fact that nanoparticles (NPs) could show toxic effects. Consumers and workers are exposed to NPs from various sources. While biomedical NPs (e.g., drug nanocarriers (Lee, Loo, Traini, & Young, 2015)) are actively introduced into the human body (Bregoli et al., 2016), NPs embedded in commercial goods can be released into the environment accidentally at any stage of the product life cycle, thus coming into contact with human organs and tissues. Hence, nanomaterials risk assessment is considered as an essential prerequisite for any NPs application (Bregoli et al., 2016;Zarogoulidis, Giraleli, & Karamanos, 2012).
Oxidative stress, excessive lung inflammation, and subsequent pulmonary fibrosis are thought to be the key mechanisms driving NPs-induced local adverse responses in the human respiratory system (Duffin, Tran, Brown, Stone, & Donaldson, 2007;Li, Muralikrishnan, et al., 2010;Mohamud et al., 2014;Muhlfeld et al., 2008;Schinwald et al., 2012;Walling & Lau, 2014). Such inflammatory mechanisms have been thoroughly investigated and elucidated. To date, it is commonly accepted that the NPs physico-chemical properties (such as composition, size, surface charge, and agglomeration) (Braakhuis, Park, Gosens, De Jong, & Cassee, 2014;Kim et al., 2016;Kreyling, Semmler-Behnke, Takenaka, & Moller, 2013;Madl, Plummer, Carosino, & Pinkerton, 2014;Rotoli et al., 2015), as well as the timing (acute vs. chronic) and route (e.g., intratracheal or intranasal) of administration (Landsiedel, Sauer, Ma-Hock, Schnekenburger, & Wiemann, 2014;Morimoto et al., 2016), strongly influence the lung inflammatory response to NPs (Mohamud et al., 2014). Most of these paradigms have been extrapolated from the data originating from animal studies, which are widely used for evaluating inflammatory responses in inhalation toxicology studies. The use of animal models is however constantly raising ethical concerns and research costs. In addition, animal models do not comprehensively mimic the human body, and this entirely holds true in relation to the histology of the human respiratory system (Gordon et al., 2015;Hayes & Bakand, 2014;Pauluhn & Mohr, 2000). Rodents are the main animal model used in inhalation toxicology studies (Pauluhn, 2003); yet, the pulmonary anatomy of these animals significantly differs from that of humans (Phalen, Oldham, & Wolff, 2008). Major differences can be observed in the upper airways: humans have a simple nasal anatomy, whereas rats have a highly complex set of ethmoid turbinates (Frohlich & Salar-Behzadi, 2014). Mice, hamsters, rabbits, and dogs share a similar morphology. In addition, there are significant differences in the distribution of epithelial cell populations along the respiratory tract and in the histology of the connective tissue. Finally, although intratracheal instillation is widely accepted as a useful method of animal exposure in nanosafety research Morimoto et al., 2016), it produces a NPs lung deposition that is heterogeneous and far from physiological (Oberdorster, 2010). Lung inflammation is indeed affected by the architecture of the respiratory system and by NPs deposition pattern. Thus, uncertainties are looming over the inflammation paradigms extrapolated from animal inhalation studies (Landsiedel et al., 2014). Because of these reasons, replacing animal models with more tissue-mimetic in vitro models of the human respiratory system is attracting the interest of the nanosafety and nanomedicine research communities. Various studies and scientific reports have aimed at offering innovative solutions in this context (Leong & Ng, 2014;Rothen-Rutishauser, Blank, Muhlfeld and Gehr, 2008;Zhang & Khademhosseini, 2015). However, the authors believe that knowledge gaps still exist on how to select the most physiologically representative in vitro testing models capable to appropriately imitate inflammatory responses caused by inhaled NPs. Our case study aims at providing an interesting insight into the in vitro parameters that should be taken into account during such selection process.
AuNPs were selected as a subject of this case study, because they are a candidate of extreme interest for the nanomedicine community (Ashraf et al., 2016;Bregoli et al., 2016). The application of gold nanomaterials as drug carriers for the treatment of lung adenocarcinoma (Movia et al., 2014) and pancreatic cancer (Spadavecchia et al., 2016) has also been recently reported by some of the authors. Although it has been demonstrated that NPs formed by inorganic metals generally induce inflammation in the lung (Mohamud et al., 2014), AuNPs seem to constitute an exception from this rule. Inhaled AuNPs are in fact often associated with inflammation inhibition (Jacobsen et al., 2009), as well as with poor translocation from the lung to distant organs (Sung et al., 2011), making them perfect candidates as drug carrier for lung diseases treatment. A significant AuNPs-induced downregulation of the inflammatory signals has also been detected in other organs and cell models (Chen et al., 2013;Khan, Abdelhalim, Alhomida, & Al-Ayed, M. S., 2013;Selim, Abd-Elhakim, & Al-Ayadhi, 2015;Sumbayev et al., 2013;Uchiyama et al., 2014;Villiers, Freitas, Couderc, Villiers, & Marche, 2010). This could potentially open novel therapeutic options for the use of AuNPs as anti-inflammatory agents. Notably, all these effects have been found to be strongly influenced by preexisting health and inflammatory conditions in the lungs (Hussain et al., 2011;Jacobsen et al., 2009); whereas, agglomeration of inhaled AuNPs (Balasubramanian et al., 2013;Gosens et al., 2010) or their interaction with lung surfactant proteins (e.g., surfactant protein D) (Schleh et al., 2013) do not play any role in defining the anti-or pro-inflammatory action of AuNPs in the lungs. Although in general, the scientific literature seems to demonstrate that AuNPs inhibit inflammation, controversial reports can still be found. For example, when monitoring the influence of AuNPs on in vitro phagocyte functions in the lung, both an increased and an unaffected cytokine secretion have been detected (Frohlich, 2015). Although according to Chen et al. no pulmonary inflammation was detectable after short exposure to AuNPs (Chen, Hung, Liau, & Huang, 2009), it has been reported that long-term (90 days) exposure promoted alveolar inflammation in rats (Sung et al., 2011). Other studies indicated that 10-50 nm AuNPs triggered lung emphysema and subsequent animals' death when intraperitoneally injected into mice (Chen et al., 2009). Notably, AuNPsinduced lung inflammation has been found to be dependent on surface chemistry. For example, a proinflammatory response has been reported for epithelial airway models exposed to AuNPs coated with bovine serum albumin (Rothen-Rutishauser, Muhlfeld, Blank, Musso, & Gehr, 2007). Although this and other important AuNPs physico-chemical parameters play a key role in defining the inflammatory effect of such nanomaterials (Mohamud et al., 2014;Seydoux et al., 2016;Tian et al., 2015), the authors believe that one of the main reasons behind the controversial results presented in the scientific literature might originate in the experimental models used.
With this working hypothesis in mind, the in vitro cultures used in our case study differed in complexity (e.g., mono-culture vs. co-culture of multiple cell types; immortalized cells vs. reconstituted tissue; submerged vs. Air-Liquid Interface culturing conditions) or culturing substrate (glass vs. polyethylene terephthalate (PET) Transwell ™ inserts). Our aim was to understand how such differences could affect the detectable inflammation triggered by AuNPs. Inflammatory responses were quantified by Interleukin-6 (IL-6) secretion in Enzyme-Linked Immuno-Sorbent Assay (ELISA) and by monocyte activation in a microfluidic set-up developed in our centre.
Interestingly, such responses were not associated with AuNPs cellular internalisation, while they were deeply influenced by the culturing substrate used. AuNPs pro-inflammatory action was also inversely correlated to the complexity of the in vitro model tested.

Cell culture
Human adenocarcinoma cells (A549 cell line), human lung fibroblasts (MRC-5 cell line), and human monocytic leukaemia cells (THP-1 cell line) were obtained from the American Tissue Culture Collection (LG Standards, Teddington, Middlesex, UK). A549 and THP-1 cells were cultured in Dulbecco's Modified Eagle Medium (DMEM), while MRC-5 cells were cultured in Modified Eagle Medium (MEM). Cell culture media (Gibco, Invitrogen, Bio-Sciences Ltd, Ireland) were supplemented with 1% penicillin/streptomycin (Gibco, Invitrogen, Bio-Sciences Ltd, Dublin, Ireland) and 10% Fetal Bovine Serum (Sigma-Aldrich, Dublin, Ireland). Cells were incubated at 37°C and 5% CO 2 . The passage number of the A549 cells was restricted between 7 and 20, whereas for MRC-5 cells and THP-1 cells the used passage number ranged between 5 and 15, and 8-10, respectively. For cell seeding, cells were detached from cell-culture flask substrate with TryplE ™ (Gibco, Invitrogen, Bio-Sciences Ltd, Dublin, Ireland), centrifuged, counted using a Countess ™ Automated Cell Counter (Invitrogen, Bio-Sciences Ltd, Dublin, Ireland) and diluted in the supplemented media at concentration appropriate for each experiment. The seeding concentration of A549 cells was kept constant among all cell models (1.2 × 10 4 cells/cm 2 ). MRC-5 cells were diluted in the supplemented media at concentration of 4 × 10 5 cells/mL. For the monocyte activation assay, THP-1 cells were used at concentration of 1.6 × 10 6 cells/mL.
A schematic representation of the in vitro cell culture models used in our study is reported in the Supporting Information (Fig. S1). Submerged cell culture model A549 cells were seeded on unmodified glass coverslips (diameter: 13 mm) placed in 24-well plates (final volume/well: 100 μL/well of cell suspension and 600 μL/ well of fresh media), following a protocol previously reported (McIntyre et al., 2016;Verma et al., 2012). Cells were incubated for 24 h at 37°C (5% CO 2 ) to allow cell attachment to the glass substrate.
Mono-cultures on Transwell ™ membranes A549 cells were seeded on Transwell ™ Permeable Supports with PET membrane inserts of 6.5 mm of diameter (growth area: 0.33 cm 2 ) and pore size of 0.4 μm (Corning Costar, VWR International, Dublin, Ireland) accordingly to supplier's protocol. Briefly, fresh medium was first added to the wells of a 24-well plate (600 μL), then the Transwell ™ inserts were inserted into the wells and plates incubated for 2/3 h at 37°C to favour cell attachment and growth. Cells were then added to the apical compartment of the Transwell ™ inserts (final volume: 100 μL/insert), and incubated at 37°C and 5% CO 2 for 24 h to allow cell attachment.
Co-culture models on Transwell ™ membranes A co-culture model was formed by co-culturing A549 cells in the presence of an excess of fibroblasts (MRC-5 cells) on the apical side of Transwell ™ Permeable Supports with PET membrane inserts of 6.5 mm of diameter (growth area: 0.33 cm 2 ) and pore size of 0.4 μm (Corning Costar, VWR International, Dublin Ireland). Firstly, fresh supplemented MEM medium was added to the well (600 μL) of a 24-well plate, the Transwell ™ inserts added, and plates incubated for 2/3 h at 37°C to favour cell attachment and growth. A total of 100 μL/insert of the MRC-5 cell suspension was then added to the apical compartment (final seeding concentration: 1.2 × 10 5 cells/cm 2 ); cells were incubated for 48 h to allow cell attachment and spreading. After 48 h, the media in the basolateral compartment was replaced with fresh MEM media (600 μL/well), and A549 cells were seeded on the top of the MRC-5 cells in DMEM media (final volume: 100 μL/insert). The seeding A549 : MRC-5 ratio was 1 : 10. Prior to seeding, A549 cells were stained with 20 mM Cell Tracker ™ Green CMFDA (Invitrogen, Bio-Sciences Ltd, Dublin, Ireland) for 45 min at 37 0 C and 5% CO 2 to allow their identification in co-culture models. Plates were incubated for 24 h to allow the successful formation of the co-culture models.
MucilAir ™ cultures A commercial three-dimensional (3D) human airway epithelial culture model (called MucilAir ™ ) was kindly supplied by Epithelix Sarl (Geneva, Switzerland). MucilAir ™ models are characterised by a pseudostratified columnar epithelium presenting beating cilia and mucus production. A representative video of the beating cilia is reported in the Supporting Information (Video S1). The MucilAir ™ model mimics The curious case of in vitro complexity Dania Movia et al.
the upper respiratory tract structure of the human lung, including basal, goblet, and ciliated cells. MucilAir ™ cultures used in his study were originated from primary human cells isolated from the human nasal cavity of a human donor. According to the supplier's certificate of analysis, the donor for the cultures was a 40-year old Caucasian woman, with no pathology reported (batch number: MD047101). MucilAir ™ cultures were supplied on Corning Costar Transwell ™ Permeable Supports with PET membrane inserts of 6.5 mm of diameter and pore size of 0.4 μm. MucilAir ™ models were cultured at the Air-Liquid Interface (ALI) in 24-well plate at 37°C and 5% CO 2 with a specific culture medium (MucilAir ™ Culture Medium, Epithelix Sarl, Geneva, Switzerland). The MucilAir ™ apical side was washed prior to AuNPs exposure, according to the supplier's recommendations, to remove the mucus produced by goblet cells overtime.
In vitro exposure to gold nanoparticles and cellular internalisation Synthesis and properties of gold nanoparticles Gold nanoparticles (Ø (TEM) = 12.5 ± 1.0 nm; hydrodynamic diameter (DLS) in deionized water = 14.5 ± 5 nm) were synthesised in sterile water according to a previously published protocol (Sivaraman, Kumar, & Santhanam, 2011). Briefly, 0.25mL of 0.254 mM of HAuCl 4 (Sigma-Aldrich, Dublin, Ireland) was added to a 24.75 mL boiling solution of trisodium citrate (Sigma-Aldrich, Dublin, Ireland) in sterile water at a trisodium citrate to HAuCl 4 molar ratio of 5 : 1. The synthesised particles were left to cool, washed over a nanoporous filter, and re-suspended in sterile water. AuNPs characterisation by Transmission Electron Microscopy (TEM) is reported in Figure S2 in the Supporting Information. AuNPs were tested for endotoxin presence: the chromogenic Pierce® LAL Chromogenic Endotoxin Quantitation kit (Thermo Scientific, Dublin, Ireland) did not detect any significant amount of endotoxin in the AuNPs sample used in this study (Fig. S3 in the Supporting Information). The AuNPs aqueous suspension was diluted in supplemented cell culture medium for further testing.

Exposure to gold nanoparticles
The AuNPs doses tested were kept comparable among the in vitro models under investigation. In order to achieve this, the different growth area between the two culture substrates used (coverslips = 1.76 cm 2 ; Transwell ™ membranes = 0.33 cm 2 ) was kept into account. Cell cultures were exposed to AuNPs at a nontoxic particle dose equal to 0.06 μg/cm 2 . Submerged cultures were exposed to a AuNPs dispersion at Au metal concentration of 13 pM (corresponding to 0.2 μg/mL). A total of 500 μL were added to each well. Cell monocultures and co-cultures grown on Transwell ™ membranes were exposed to AuNPs by adding 100 μL/insert of AuNPs dispersion (13 pM) to the apical compartment. MucilAir ™ cultures were exposed to 0.06 μg/cm 2 by adding 30 μL of AuNPs dispersion (44 pM) in saline (0.9% NaCl + 1.25 mM CaCl 2 + 10 mM HEPES (N-2-hydroxyethylpiperazine-N-2ethane sulfonic acid)) to the apical compartment of the Transwell ™ inserts. Cell models were exposed to AuNPs for 3, 6 and 24 h.

Immunocytochemistry and Laser Scanning Confocal Microscopy
After exposure to AuNPs, cell cultures were fixed with 3.7% paraformaldehyde (Sigma-Aldrich, Ireland) for 10 min at ambient temperature and permeabilized with 0.1% Triton-X (Sigma-Aldrich, Dublin, Ireland) for 3 min. MucilAir ™ cultures was fixed with 3.7% paraformaldehyde for 30 min at ambient temperature and permeabilized with 0.5% Triton X for 2 h. Bovine serum albumin (Sigma-Aldrich, Dublin, Ireland) at 1% in Phosphate-Buffered Saline (PBS) (Fisher Scientific Ireland, Dublin, Ireland) was used to block unspecific staining. A 30 min incubation time was used in all the cell models, excluding MucilAir ™ cultures that were blocked for 2 h. Cells were stained with Hoechst 33342 (Invitrogen, Bio-Sciences Ltd, Dublin, Ireland) for nuclei, rhodamine phalloidin (Invitrogen, Bio-Sciences Ltd, Dublin, Ireland) for F-actin filaments and mouse anti-α tubulin Alexa 488 (Invitrogen, OR) for tubulin microtubules; co-cultures were exclusively stained with Hoechst 33342 and rhodamine phalloidin. Cell culture models were incubated with the staining solution for 2 h at ambient temperature. After multiple washes with PBS, coverslips and Transwell ™ inserts were mounted on glass slides with transparent mounting medium (VECTASHIELD, Vector Laboratories Inc., CAD,USA) and sealed. AuNPs cellular internalisation was evaluated by Laser Scanning Confocal Microscopy (LSCM). LSCM imaging and analysis was carried out with a ZEISS 510 Meta confocal microscope equipped with a Zeiss LSM 5 software (Carl Zeiss Microscopy GmbH, Jena, Germany). AuNPs were imaged in reflectance mode The curious case of in vitro complexity Dania Movia et al.
at λ exc = 561 nm, as previously reported (Movia et al., 2014). AuNPs are well known to scatter light strongly (Aslan, Lakowicz, & Geddes, 2005 Raman spectroscopy and dark-field microscopy of submerged cultures Untreated and AuNPs-exposed (24 h) submerged cultures of A549 cells on glass substrates were examined by Raman spectroscopy. Cell cultures were fixed at ambient temperature in 2.5% glutaraldehyde (Sigma-Aldrich, Dublin, Ireland) in PBS for 10 min and then washed multiple times with PBS. Raman spectroscopy measurements were carried out by means of a Xplora spectrometer (Horiba Scientifics, NJ, USA) at excitation wavelength of 660 nm at ambient temperature. A 100× objective was used with a numerical aperture of 0.9 in backscattering configuration. The achieved spectral resolution was close to 2/cm. As a control, the Raman spectrum of AuNPs in aqueous solution was also acquired.

Transmission Electron Microscopy of biological specimens
Transmission Electron Microscopy images of ultrathin sections of the tested cell models were examined. After exposure to AuNPs for 24 h, cell cultures were fixed at ambient temperature in 2.5% glutaraldehyde, and specimens were prepared as previously described (Movia et al., 2014). After mounting on 300-mesh Cu grids, these sections were stained with uranyl acetate and lead citrate and finally imaged (Tecnai Transmission Electron Microscope, FEI, Oregon, USA).

Cytokines secretion
The production of the pro-inflammatory cytokine IL-6 was evaluated and quantified by Sandwich ELISA (Human IL-6 ELISA MAX ™ Standard Sets, BioLegend, MSC, Dublin, Ireland). Some of the authors have previously demonstrated that A549 cells can secrete IL-6 in response to exposure to nanomaterials with inflammatory potential (Mohamed et al., 2013). Similarly, IL-6 has demonstrated to be a predictive marker of inflammatory response to respiratory sensitizers in in vitro reconstituted 3D human lung epithelial (MucilAir ™ ) models (Huang, Wiszniewski, Constant, & Roggen, 2013). Supernatants harvested from in vitro models exposed to AuNPs at a particle dose equal to 0.06 μg/cm 2 for 3, 6, and 24 h were tested. Supernatants of negative controls (untreated cell models; NT) were also tested for comparison. For cell cultures grown on Transwell ™ Permeable Supports, the concentration of IL-6 is reported only for the supernatants isolated from the apical compartment, as no detectable levels of IL-6 could be found in the supernatants isolated from the basolateral compartment (Fig. S4). This result suggests that there was a polarised IL-6 secretion by the epithelial cells into the apical compartment. This has been previously reported by other research groups for various human lung cell lines (including A549 cells) grown on Transwell ™ membranes (Carolan, Mower, & Casale, 1997;Chow et al., 2010;Sun, Wu, Sun, & Huang, 2008). In order to include a positive control (PT) in the experimental design, the in vitro models were exposed to lipopolysaccharides (LPS) (Sigma-Aldrich, Dublin, Ireland) at concentration of 200 ng/mL for 6 h and the respective supernatants then tested for IL-6 content. LPS is known to trigger inflammatory responses (Abate, Alghaithy, Parton, Jones, & Jackson, 2010) and IL-6 secretion (Relja et al., 2014;Xie et al., 2009) in A549 cells. ELISAs were carried out according to the manufacturer's manual and samples tested in duplicate. The Epoch microplate reader (Biotek, Mason Technology Ltd, Dublin, Ireland) was used to detect the optical density at 450 nm for each well, and the determined values were corrected by subtracting the optical aberration of the 96-well plastic plate at 570 nm; the means of the resulted values were calculated and calibrated against a standard curve. Separating the used AuNPs from the extracted supernatants is a complicated procedure. In order to account for potential optical interference of AuNPs with the ELISA read-outs, the IL-6 standard was dissolved in the assay diluent (as for manufacturer's protocol) or in assay diluent spiked with an AuNPs in supplemented DMEM media at concentration of 13 pM. This allowed evaluating whether AuNPs present in the tested supernatants might interfere with the ELISA. It was found that the particles affected the IL-6 ELISA readout, quenching the absorbance signal, as showed by the difference in the resulting calibration curves reported in Figure S5 in the Supporting Information. Hence, cytokines concentrations in supernatants were extrapolated considering the two ELISA calibration curves and the presence or not of AuNPs in the samples tested.
The curious case of in vitro complexity Dania Movia et al.

Monocyte adhesion assay in dynamic microfluidic environment acting as inflammation recruitment model
The response of human monocytes (THP-1 cells) to pro-inflammatory signals secreted by AuNPs-treated in vitro models was analysed qualitatively and quantitatively in a microfluidic environment. Vena8 ™ biochips (Cellix Ltd, Dublin, Ireland) were used (Konya, Peinhaupt, & Heinemann, 2014;Munir, Rainger, Nash, & Mcgettrick, 2015;Robinson et al., 2009). The biochip includes eight micro-channels, which are 28.00, 0.80, and 0.12 mm in length, width, and height, respectively. The micro-channels were coated with recombinant human Vascular Cell Adhesion Molecule-1 (rhVCAM-1) (Sigma-Aldrich, Dublin, Ireland) at concentration of 20 μg/mL in PBS (final volume: 10 μL/channel) one day prior the assay and incubated at 4°C overnight. THP-1 cells at concentration of 1.6 × 10 6 cells/mL were mixed with the supernatants harvested from untreated or AuNPs-treated in vitro models at ratio of 1 : 1 and incubated at 37°C and 5% CO 2 for 30 min to allow THP-1 cell activation processes. In the case of cultures grown on Transwell ™ inserts, the supernatants collected from the apical compartments were used. The final pool of supernatants tested were harvested from: untreated mono-cultures and co-cultures grown on Transwell ™ inserts and MucilAir ™ models, and the respective AuNPs-treated models exposed at concentration of 0.06 μg/cm 2 for 24 h. As a PT for the assay, THP-1 cells were mixed at a ratio of 1 : 1 with recombinant human Monocyte Chemoattractant Protein-1 (rhMCP-1) (Sigma-Aldrich, Dublin, Ireland) at concentration of 1 μg/mL and incubated at 37°C and 5% CO 2 for 30 min. Untreated THP-1 cells were included in the experimental design as negative control (NT). Further controls were used to test the robustness of the chosen experimental set-up, such as the use of uncoated channels. The experiments under dynamic conditions were performed based on slightly adapted protocols publicly available from Cellix Ltd and that have been used in a number of studies (Choi et al., 2008;Dominical et al., 2015;Ferkau et al., 2013;Long et al., 2004;Munir et al., 2015;Nissinen et al., 2010;Robinson, Kashanin, O'Dowd, Williams, & Walsh, 2008;Wu, Mitchell, & Walsh, 2005). Briefly, after washing the channel with 30 μL DMEM media, THP-1 cells were introduced at shear stress of 0.03 dyne/cm 2 . A neMESYS ™ syringe pump (Cetoni GmbH, Korbussen, Germany) controlled by the neMESYS UserInterface software was used. Cells flow was monitored with a NIKON TE 300 Eclipse epifluorescence microscope (20× objective lens). Brightfield images were continuously acquired at rate of 1 frame every 10 ms for a minute (for a total of 360 frames) by QImaging software (Media cybernetics, Cambridge, UK). Representative videos of the experiments carried out in the dynamic microfluidic environment are reported as Supporting Information (Video S2-S10). When visioning the supporting material, attention should be focused on the cells adhering to the channels: these cells are activated monocyte, and their number correlates with the inflammogenic stimulus. Every sample was run in duplicate on two different rhVCAM-1 coated channels and imaged at three adjacent locations. Every sample was perfused for a total time of 6 min. For quantitative analysis, every video recorded was analysed, and adherent cells present in every chosen location were counted manually.

Statistical analysis
Graph-Pad Prism (Graph-Pad Software Inc., La Jolla, CA, USA) was used to carry out the statistical analysis. A P value < 0.05 was considered statistically significant. The statistical tests used are specified in the corresponding figure caption.

Results
Cytokine secretion and monocyte activation in a dynamic microfluidic environment acting as inflammation recruitment model Changes in the levels of the pro-inflammatory cytokine IL-6 were quantified as an indicator of an acute inflammatory response in the various in vitro models to AuNPs exposure (Figs. 1A and S6). IL-6 is known to be produced at the site of inflammation and plays a key role in the first acute phase of the inflammatory response (Gabay, 2006). No endotoxin contamination was found in the AuNPs sample (Fig. S3 in the Supporting Information); thus, IL-6 secretion levels could be directly linked to the cell interactions with the nanomaterial. Supernatants harvested from those in vitro models that showed high IL-6 production were then utilised for testing monocytes activation and adhesion under dynamic microfluidic environment. Pro-inflammatory and anti-inflammatory signals, together with other physical and chemical cues, do in fact work in a "yin and yang" mode. This makes particularly difficult, if not impossible, to use ELISA to predict which signal (pro-inflammatory or  on glass substrates, mono-and co-cultures grown on Transwell ™ inserts, and in MucilAir ™ models. Secretion levels were normalised to each corresponding negative control (NT) set to a value equal to 1. The in vitro models were exposed to AuNPs at a concentration of 0.06 μg/ cm 2 for 3, 6, and 24 h. Untreated models (NT) and cell cultures exposed to lipopolysaccharides (200 ng/mL; 6 h) as a positive control were also tested for IL-6 secretion for comparison. Data are reported as mean ± standard error of the mean (n tests > 3 and n replicates = 2). The symbols (*) and (***) indicate a significant difference (P value < 0.05 and P < 0.001, respectively) as compared with the corresponding NT (two-way ANOVA followed by Bonferroni post-test). (B) Table highlighting significant differences in IL-6 secretion among the four in vitro models tested (two-way ANOVA followed by Bonferroni post-test). (C-L) Monocyte adhesion assay in a dynamic microfluidic environment. anti-inflammatory) will prevail, triggering (or not) inflammation through monocytes activation. Thus, a microfluidic set-up was optimised in our centre for this study, in order to mimic the monocytes activation and recruitment from the bloodstream to the lung epithelium in response to inflammatory signals. In in vivo conditions, monocytes circulating in the bloodstream play a key role in lung inflammation (Suzuki, Chow, & Downey, 2008): in response to an inflammogenic stimulus, these cells in fact activate and transmigrate from the bloodstream to the site of inflammation across the blood vessel wall Holt, 2005). Transmigration begins with rolling along endothelial cells, followed by firm adhesion to the vessels walls through VCAM-1 interaction and diapedesis across the endothelium towards the harmed tissue (Gerhardt & Ley, 2015). Various bioactive molecules control monocytes activation in the lung, including proinflammatory cytokines and chemokines secreted by epithelial cells and acting as chemoattractants for monocytes recruitment into the tissue (Shi & Pamer, 2011;Turner, Nedjai, Hurst, & Pennington, 2014). Thus, this assay was carried out with the aim to evaluate if human monocytes (THP-1 cells) would be activated in response to the pro-inflammatory signalling molecules secreted by the AuNPs-treated in vitro models, thus providing more physiological information on the in vitro parameters influencing the AuNPs-induced inflammation detected. In our experimental set-up, physiological flow conditions, similar to those found in human capillaries in the alveoli, were used. rhVCAM-1 was used as biomarker coating the microfluidic substrate and allowing for monocytes arrest and adhesion following activation. The number of THP-1 cells attached to the rhVCAM-1 coated channels represented a quantitative indication of the monocytes activation in response to the soluble signalling molecules and proinflammatory mediators secreted by the in vitro models. The assay controls confirmed the validity of the set-up developed in house (Fig. 1C,E-G). The positive control (rhMCP-1 treated THP-1 cells) showed in fact an increase (although not significant) in the number of adherent cells as compared with the untreated THP-1 cells (NT), while almost no cells adhered to the microchannels surface in the absence of rhVCAM-1 coating. rhMCP-1 is known to induce monocytes activation (Ashida, Arai, Yamasaki, & Kita, 2001;Cambien, Pomeranz, Millet, Rossi, & Schmid-Alliana, 2001).
Our data clearly evidence that IL-6 secretion levels are significantly higher for submerged mono-cultures grown on Transwell ™ inserts than in models grown on glass substrates (Figs. 1B and S6). LPS (6 h) did not seem to induce any significant increase in IL-6 secretion, suggesting that a longer exposure time or higher concentration might be necessary for this molecule to trigger an inflammatory response of the in vitro models tested. Overall, submerged cell models did not show any significant IL-6 secretion at the three time points tested, as compared with NT (Fig. 1A). On the contrary, AuNPs-triggered IL-6 production was time-dependent in mono-cultures grown on Transwell ™ inserts, and a significant increase in IL-6 secretion was detected following 24 h exposure. Because these two in vitro models shared the same cell type and cell seeding concentration, our results suggest that the different culturing substrates might have influenced the proinflammatory response of A549 cells to AuNPs. Incubation of THP-1 cells in supernatants harvested from mono-cultures grown on Transwell ™ inserts and exposed to AuNPs also triggered a significant increase in monocyte adhesion in a dynamic microfluidic environment (Fig. 1D,H-I).
IL-6 basal levels in untreated cultures increased with the increasing complexity of the in vitro model tested (Fig. S6). For example, the presence of lung fibroblasts in the co-cultures induced an increase in IL-6 levels in the negative control (NT) as compared with submerged mono-cultures grown on glass or on Transwell ™ inserts. Similarly, MucilAir ™ models, the most complex in vitro model tested in this study, showed the highest IL-6 expression in the negative control among all in vitro cultures, with almost a four-fold increase in IL-6 production as compared with submerged cultures on glass.
In vitro complexity was inversely proportional to the ability of AuNPs to trigger inflammation. No significant increase in IL-6 secretion could in fact be detected in co-cultures grown on Transwell ™ inserts or in MucilAir ™ models exposed to AuNPs at any of the time-points tested (Fig. 1A,B). Similarly, incubation of THP-1 cells in supernatants isolated from MucilAir ™ models exposed to AuNPs did not cause any significant increase in monocyte adhesion in the dynamic microfluidic environment, when compared with the monocytes activation induced by incubation with supernatants harvested from untreated MucilAir ™ models (NT) (Fig. 1D,L-M). In contrast, supernatants collected from co-cultures grown on Transwell ™ inserts (i.e., in vitro The curious case of in vitro complexity Dania Movia et al. models formed by multiple cell types, comparable culturing substrate but a less complex structure than MucilAir ™ models) triggered an enhanced monocyte activation and adhesion when the cultures were exposed to AuNPs (Fig. 1D,J-K).

Cellular internalisation of gold nanoparticles
Following exposure to AuNPs for 3, 6, and 24 h, in vitro models were immunostained and imaged by LSCM.
Analysis of z-stack LSCM images (Figs. 2 and 3) demonstrated that AuNPs did not trigger any evident change in cell morphology and/or in the cytoskeleton organisation in any of the in vitro models tested, thus confirming that, at the concentration tested (0.06 μg/ cm 2 ), AuNPs did not cause acute cytotoxicity. Laser Scanning Confocal Microscopy imaging and analysis ( Fig. 2A), Raman spectroscopy (Fig. 2B), and dark-field microscopy ( Fig.  2C) successfully  demonstrated that AuNPs were internalised in A549 cells grown in submerged conditions on glass substrates at all the time points tested. In detail, AuNPs could be found in the cell cytoplasm, scattered among the αtubulin filaments forming the cells cytoskeleton ( Fig. 2A). As expected, untreated cells did not exhibit any significant Raman signal; whereas, strong Raman bands, associable to the characteristics Raman shifts of AuNPs, were detected in AuNPs-treated cells (Fig. 2B). Finally, dark-field microscopy (Fig. 2C) demonstrated that a large number of AuNPs were localised within the cell body.
In contrast, no evident AuNPs internalisation could be detected by LSCM imaging and analysis in monocultures and co-cultures grown on Transwell ™ inserts, and MucilAir ™ models at any of the time points tested (Fig. 3). Such results were confirmed by TEM imaging (Fig. 4), which was carried out as a tool to confirm the absence of internalised AuNPs into cultures grown on Transwell ™ inserts. TEM images did not show in fact any clear indication of AuNPs internalisation in these three in vitro models. Notably, AuNPs could be seen at the cells surface in co-cultures grown on Transwell ™ inserts (Figs. 3B and 4B), where up to 4 layers of cells could be distinctly recognised by TEM (Fig. 4B), confirming the successful formation of the co-culture model. AuNPs were also detected by LSCM on the top of the MucilAir ™ microtissues (Figs. 3C and 4C), trapped among the cilia structures, which were clearly identifiable by LSCM and TEM.

Discussion
Lung epithelium is the first line of defence against the inhaled NPs. Currently, in vitro models of the human lung epithelium used for NPs risk assessment can vary in their complexity level: from monolayers of immortalized cell lines, through primary cells, to the revolutionary 3D-cell cultures. A detailed summary of these models can be found in a recently published review (Gordon et al., 2015). Nevertheless, the majority of the risk assessment studies on inhaled NPs is performed on submerged monocultures grown on glass or plastic substrates. Several publications have demonstrated that co-cultures of multiple cell types have a strong influence on the observed cellular responses to inhaled NPs Clift et al., 2014;Lehmann et al., 2011;Muller et al., 2010;Rothen-Rutishauser, Mueller, et al., 2008;Rothen-Rutishauser et al., 2007;Stoehr et al., 2015). Thus, in our case study, four different in vitro models of the lung epithelium were tested in order to quantify whether the in vitro complexity and culturing substrates can affect the detectable pulmonary inflammatory response to AuNPs. Such models were as follows: submerged cell cultures of human lung epithelial (A549) cells grown on the glass substrates; A549 cells grown on the Transwell ™ membranes as mono-cultures or co-cultures with human lung fibroblasts (MRC-5 cells); and MucilAir ™ models, which are reconstituted tissues derived from human donors and cultured in ALI conditions on Transwell ™ inserts. MucilAir ™ cultures were selected as representative models of the upper airways epithelium, while cultures of A549 cells were used to mimic the alveolar epithelium. MucilAir ™ models feature a number of unique advantages over cell cultures formed by immortalized cells (such as the A549 cell line). These advantages include fully differentiated epithelium (including basal, goblet, and ciliated cells), a mucus layer covering the epithelial cells, cilia beating, functional tight junction formation, and preservation of homeostatic state up to 1 year (Huang, Wiszniewski, & Constant, 2011). Thanks to these unique features, MucilAir ™ models have been successfully used to discriminate between substance with low and high absorption in humans (Reus et al., 2014), to distinguish the respiratory sensitizers from dermal sensitizers (Huang et al., 2013), and they also have been partially explored for NPs interactions studies (Beaver et al., 2009;Frieke Kuper et al., 2015). MucilAir ™ models are composed by primary human cells isolated from the nasal cavity, the trachea or the bronchus. Because of the physiology of the human respiratory system, however, the primary region where most inhaled NPs deposit is the alveoli (Rogueda & Traini, 2007). Unfortunately, in vitro modelling of the human alveolar epithelium is characterised by serious limitations, as primary cells sourcing and culture maintenance are extremely difficult, while human immortalized alveolar epithelial cell lines available are limited to Type-II cells. The A549 cell line belongs to such category. A549 cells are capable of forming distinct epithelial monolayers, characterised by high confluence and mucin expression; however, these cells do not express functional tight junctions (Forbes & Ehrhardt, 2005). Nevertheless, because of the limitations listed earlier regarding the in vitro modelling of the human alveolar epithelium, A549 cells are an accepted model for nanomaterials risk assessment studies (Paur et al., 2011, De Souza The curious case of in vitro complexity Dania Movia et al. The curious case of in vitro complexity Dania Movia et al. Carvalho, Daum, & Lehr, 2014). In our case study, A549 cells were also co-cultured with human fibroblasts (MRC-5 cells), in an attempt to extend the simulation of the alveolar structure to other cellular components, such as the connective tissue cells. Co-cultures composed by at least two different cell types, one of which being stromal cells, have the advantage to allow for evaluating in vivo cell-cell interactions (Miki et al., 2012). Several studies demonstrated in fact that fibroblasts do influence the sensitivity of in vitro airway culture models to stimulating or injury events (Sacco et al., 2004), drugs treatment (Jastrzebska Jedrych, Grabowska-Jadach, Chudy, Dybko, & Brzozka, 2012), and nanomaterials exposure (Chang, Chang, Hwang, & Kong, 2007;Singh, Movia, Mahfoud, Volkov, & Prina-Mello, 2013). A cell substrate compatible with ALI culturing conditions (i.e., the Transwell ™ inserts) was also included in our study, with the aim of evaluating the influence of such inserts on epithelial cells response to NPs.
To serve the experimental purposes, a non-toxic AuNPs dose was used in our case study. AuNPs dose was estimated based on our previous study (Movia et al., 2014) in a way that AuNPs could achieve cellular internalisation without causing acute cytotoxicity. No significant change in cell morphology, indicative of potential AuNPs cytotoxicity, could be detected in A549 cells cultured on glass and exposed to AuNPs at concentration of 0.06 μg/cm 2 for 24 h (Figs. 2A and S7 in the Supporting Information). Similarly, no cytoskeleton remodelling could be noticed in monocultures and co-cultures grown on Transwell ™ inserts, or in MucilAir ™ models, exposed to AuNPs over 24 h (Fig. 3). This confirmed that AuNPs did not trigger acute cytotoxicity at this dose.
In our case study, quantification of IL-6 secretion was used to understand whether AuNPs did induce an inflammatory response. Epithelial cells, in fact, are known to be the major source of various proinflammatory mediators in response to the impairment of lung epithelium homeostasis (Mohamud et al., 2014). It has been reported that pro-inflammatory mediators (e.g., Tumour Necrosis Factor-alpha, IL-6 and InterLeukin-8) are released by such cells when NPs interact with lung epithelial cells (Napierska et al., 2012). The quantification of the levels of proinflammatory cytokine IL-6 in response to AuNPs exposure was also accompanied by the careful investigation of the cellular internalisation of such nanomaterial overtime by various techniques. No correlation was found between IL-6 levels and AuNPs internalisation into the epithelial cell layer. In detail, where AuNPs cellular uptake was found (i.e., submerged cultures on glass), no significant IL-6 secretion could be detected; whereas, AuNPs-treated mono-cultures models grown on Transwell ™ inserts were characterised by no AuNPs internalisation (Fig. 3A) but significant IL-6 secretion (Fig. 1A) following 24 h exposure. Similar results have been recently reported in a study on polyethylene glycol-coated AuNPs in a monocyte-derived dendritic cells model (Fytianos et al., 2015). Such study showed that, although a limited AuNPs cellular uptake could be observed, a significant release of pro-inflammatory mediators was induced by NPs exposure. Our data find also agreement to other scientific reports, which describe that A549 cells cultured on Transwell ™ inserts in ALI conditions respond more readily to insults as compared with submerged cultures grown on glass (Frohlich et al., 2013;Lenz et al., 2013). Based on our experimental observations, we suggest that the increased sensitivity of in vitro models grown on Transwell ™ inserts is imputable to the culturing substrate used and not to the ALI condition.
In parallel, it should be noted that monocytes activation was not be directly correlated to IL-6 secretion levels. Although co-culture of A549 cells with MRC-5 cells decreased the sensitivity of the epithelium, resulting in a complete loss of IL-6 secretion in response to AuNPs exposure (Fig. 1A), significant monocyte activation and adhesion could be detected after 24 h exposure (Fig. 1D). This suggested that other proinflammatory mediators might have been secreted by the co-culture model and in particular by its fibroblasts component. In agreement with our hypothesis, a previously published study on MRC-5 cells (as a monoculture) reported that these cells could produce various pro-inflammatory cytokines in response to AuNPs exposure (Li, Hartono, et al., 2010). Literature data show also that in a triple co-culture model grown on Transwell ™ inserts and including A549 cells but not MRC-5 cells, no protein induction of pro-inflammatory cytokines (including IL-6) could be detected in response to exposure to AuNPs of comparable size (Brandenberger et al., 2010). When analysing the IL-6 secretion levels of MucilAir ™ models and the monocyte activation triggered by supernatants harvested from such cultures, it was obvious once again, that in vitro complexity was inversely proportional to inflammation. Neither any significant IL-6 could be detected (Fig. 1A), The curious case of in vitro complexity Dania Movia et al. nor monocyte activation (Fig. 1C), after AuNPs exposure for up to 24 h. Interestingly, in MucilAir ™ models AuNPs were found trapped on the cilia structure (Figs. 3C and 4C). The lack of inflammatory response in MucilAir ™ models may be linked therefore to the fact that the beating cilia present in this model together with the remaining mucus, act as a physical barrier to AuNPs penetration and interaction with epithelial cells, thus alleviating inflammation.
Finally, no correlation could be demonstrated between the lack of AuNPs internalisation and monocytes activation. In detail, in co-cultures and MucilAir ™ models grown on Transwell ™ membranes, neither AuNPs internalisation was detected ( Fig. 3B and C), nor monocytes activation (Fig. 1C). Thus, we suggest that the monocytes activation detected when incubating THP-1 cells with supernatants harvested from mono-cultures grown on Transwell ™ membranes, was not triggered by the direct contact of the immune cells with the not-internalised AuNPs still present in solution.

Conclusions
Based on our case study, we can conclude that, in vitro cell models and culturing substrates do indeed affect the AuNPs cellular internalisation and certainly influence the inflammatory response of human epithelial cells and epithelial reconstituted tissue. AuNPs internalisation was in fact achieved only in submerged cultures grown on glass substrates, and it did not correspond to the highest inflammatory response detected. Curiously, the more complex the cell model/culture was, the less intense and sensitive the inflammatory responses became. For example, no significant secretion of IL-6 was detected in supernatants harvested from AuNPs-treated cocultures grown on Transwell ™ inserts; however, significant monocytes activation was induced by the signalling molecules secreted in such models. We hypothesise that such molecules were most likely secreted by fibroblasts. On the contrary, neither IL-6 secretion nor monocyte activation could be detected when MucilAir ™ models, formed by a complex, reconstituted, pseudostratified respiratory epithelium, were exposed to AuNPs.
In conclusion, we have presented a first-case report aiming at describing the in vitro parameters that should be taken into account when selecting the most appropriate testing model for inhaled NPs. Our case study suggests that the validation of in vitro models capable of integrating more cellular components and of mimicking specific aspects of the human tissue of interest is highly needed. The authors believe that further research efforts should focus on expanding the pool of in vitro models, cell lines, and exposure timepoints tested, as well as on carrying out experiments evaluating the recruitment of other circulating immune cells (e.g., neutrophils) in response to AuNPs exposure and in vitro-to-in vivo correlation studies.

CONFLICT OF INTEREST
None declared.

ACKNOWLEGMENTS
The authors would like to thank Dr. Samuel Constant (Epithelix Sàrl, Switzerland) for kindly providing the MucilAir ™ models used in this study, Dr. Dimitri Scholz and Dr. Julie Kennedy (Conway Institute, University College Dublin, Ireland) for TEM samples processing and imaging, and Dr. Alan P. Bell for the technical support with He-ion microscopy imaging. This work has been partially funded by the Irish Research Council Government of Ireland Fellowship to DM, the MSc Molecular Medicine programme of Trinity College Dublin (Ireland), and the EU FP7 NANoREG project (grant agreement 310584) towards the partial support of the work of LDC.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article.   Figure S3. Endotoxin units (EU) detected in AuNPs dispersion in water by means of Pierce® LAL Chromogenic Endotoxin Quantitation kit (Thermo Scientific, Ireland). No endotoxin contamination could be detected in the AuNPs sample. The AuNPs sample was tested at concentration comparable to that tested in in vitro models. The assay was carried out as for manufacturer's protocol, and absorbance was measured at 405 nm to detect the yellow chromogenic color developed in the presence of endotoxin. Dilutions of standards and AuNPs were prepared in endotoxin-free water. The AuNPs sample was tested with and without (blank) the presence of LAL chromogenic substrate to determine any inference in the assay due to the optical properties of the nanomaterial itself. Blank values were subtracted from the readings of the AuNPs sample to accurately determine the endotoxin contamination. Results are reported as average (nreplicates = 2) ± standard deviation. Lower limit of detection of the assay = 0.1 EU/ml; upper limit of detection = 1 EU/ml, as declared by manufacturer. Figure S4. Representative set of data showing the changes in IL-6 secretion by untreated (NT) or AuNPstreated co-cultures grown on Transwell ™ inserts. IL-6 content was quantified in both apical and basolateral compartments of the cultures. While a significant production of such pro-inflammatory cytokine could be detected in the apical compartment of the cultures tested, no significant amount of IL-6 were found in the basolateral compartment. Cell cultures were exposed to AuNPs (0.06 μg/cm2) for 3, 6 and 24 h. Data are reported as mean ± standard deviation. Figure S5. Representative calibration curve deriving from an IL-6 standard dissolved in assay diluent with (in blue; r2 = 0.99) or without (in black; r2 = 0.98) the addition of the AuNPs dispersion (13 pM). Quenching of the absorbance signal is evident in the presence of AuNPs. Figure S6. IL-6 secretion in submerged cultures grown on glass substrates, mono-and co-cultures grown on Transwell ™™ inserts, and in MucilAir ™ models. Secretion levels are not normalized. The in vitro models were exposed to AuNPs at a concentration of 0.06 μg/cm2 for 3, 6 and 24 h. Untreated models (negative control or NT) and cell cultures exposed to LPS (200 ng/ml; 6 h) as a positive control (PT) were also tested for IL-6 secretion for comparison. Data are reported as mean ± standard error of the mean (ntests > 3 and nreplicates = 2). The symbols (*) and (***) indicate a significant difference (p value < 0.05 and p < 0.001, respectively) as compared to the corresponding NT (two-way ANOVA followed by Bonferroni post-test). Figure S7. Representative Helium Ion Microscopy (HIM) images of A549 cells cultured on the glass substrates in submerged conditions and imaged by a Zeiss Orion Plus He-ion microscope (Carl Zeiss, Oberkochen, Germany): (A) untreated (NT) or (B-C) exposed to AuNPs at concentration equal to 0.06 μg/cm2 for 24 h. Specimens were prepared and imaged as previously described (Movia et al., 2014). Scale bars: (A-B) 10 μm; (C) 5 μm.
Video S1. Video of a representative region of a MucilAir ™ culture, showing the characteristic cilia beating (40× objective lens; 360 frames; 8 ms interval). Video S2. Representative video of THP-1 cells flowing into uncoated channels (flow direction: right to left; 20× objective lens; 360 frames; 8 ms interval). No significant monocyte adhesion to the substrate can be noticed. Video S3. Representative video of THP-1 cells flowing into rhVCAM-1 coated channels after incubation for 30 min in media (flow direction: right to left; 20× objective lens; 360 frames; 8 ms interval). Video S4. Representative video of THP-1 cells flowing into rhVCAM-1 coated channels after exposure to rhMCP-1 (flow direction: right to left; 20× objective lens; 360 frames; 8 ms interval). Some monocytes adhere to the channel substrate, demonstrating activation. Video S5. Representative video of THP-1 cells flowing into rhVCAM-1 coated channels after 30 min incubation in media harvested from untreated (NT) mono-cultures grown on Transwell ™ inserts (flow direction: right to left; 20× objective lens; 360 frames; 8 ms interval).
The curious case of in vitro complexity Dania Movia et al.
Video S6. Representative video of THP-1 cells flowing into rhVCAM-1 coated channels after 30 min incubation in media harvested from mono-cultures grown on Transwell ™ inserts exposed to AuNPs for 24 h (flow direction: right to left; 20× objective lens; 360 frames; 8 ms interval). Monocytes activation is evident. Video S7. Representative video of THP-1 cells flowing into rhVCAM-1 coated channels after 30 min incubation in media harvested from untreated (NT) co-cultures grown on Transwell ™ inserts (flow direction: right to left; 20× objective lens; 360 frames; 8 ms interval). Video S8. Representative video of THP-1 cells flowing into rhVCAM-1 coated channels after 30 min incubation with media harvested from co-cultures grown on Transwell ™ inserts exposed to AuNPs for 24 h (flow direction: right to left; 20× objective lens; 360 frames; 8 ms interval). The activated monocytes rolling and adhering to the channel substrate can be easily distinguished. Video S9. Representative video of THP-1 cells flowing into rhVCAM-1 coated channels after 30 min incubation with media harvested from untreated (NT) MucilAir ™ cultures (flow direction: right to left; 20× objective lens; 360 frames; 8 ms interval). Video S10. Representative video of THP-1 cells flowing into rhVCAM-1 coated channels after 30 min incubation with media harvested from MucilAir ™ cultures exposed to AuNPs for 24 h (flow direction: right to left; 20× objective lens; 360 frames; 8 ms interval).