The development of a 3D-printed in vitro integrated oro-pharyngeal air–liquid interface cellular throat model for drug transport

To simulate the deposition of drugs in the oro-pharynx region, several in vitro models are available such as the United States Pharmacopeia-Induction Port (USP-IP) throat and the Virginia Commonwealth University (VCU) models. However, currently, there is no such in vitro model that incorporates a biological barrier to elucidate drug transport across the pharyngeal cells. Cellular models such as in vitro air–liquid interface (ALI) models of human respiratory epithelial cell lines are extensively used to study drug transport. To date, no studies have yet been performed to optimise the ALI culture conditions of the human pharyngeal cell line Detroit 562 and determine whether it could be used for drug transport. Therefore, this study aimed to develop a novel 3D-printed throat model integrated with an ALI cellular model of Detroit 562 cells and optimise the culture conditions to investigate whether the combined model could be used to study drug transport, using Lidocaine as a model drug. Differentiating characteristics specific to airway epithelia were assessed using 3 seeding densities (30,000, 60,000, and 80,000 cells/well (c/w), respectively) over 21 days. The results showed that Detroit 562 cells completely differentiates on day 18 of ALI for both 60,000 and 80,000 c/w with significant mucus production, showing response to bacterial and viral stimuli and development of functional tight junctions and Lidocaine transport with no significant differences observed between the ALI models with the 2 cell seeding densities. Results showed the suitability of the Low density (60,000 c/w or 1.8 × 105 cells/cm2) ALI model to study drug transport. Importantly, the developed novel 3D-printed throat model integrated with our optimised in vitro Detroit 562 ALI model showed transport of Lidocaine throat spray. Overall, the study highlights the potential of the novel 3D-printed bio-throat integrated model as a promising in vitro system to investigate the transport of inhalable drug therapies targeted at the oro-pharyngeal region.


Introduction
The airway epithelium lining the respiratory tract is a continuous cellular layer of different cell types, starting from the nasopharynx to the alveoli, forming a protective barrier between the inhaled air and the underlying mucosal tissue [1]. Drugs and toxins must cross this barrier to reach their target site and elicit a therapeutic or noxious effect. In vitro air-liquid interface (ALI) of human respiratory epithelial cell lines is an invaluable tool that phenotypically mimics the in vivo airway epithelium, extensively used to study drug transport, and evaluate the fate of inhaled drugs and toxins. These models can be used to predict immune responses and therapeutic efficacy owing to their ease of culture, genetic homogeneity, and greater reproducibility [2,3]. These in vitro ALI models are established when cells are seeded on the apical side of a semi-permeable support membrane, allowing nutrients from the basal media to pass through the membrane. Upon reaching confluency, the culture medium from the apical layer is removed, enabling cell growth and differentiation on the apical, air-exposed surface [4,5]. Thus, cells grown in ALI conditions undergo differentiation and polarisation, forming junctional complexes that recapitulate the key in vivo molecular and structural characteristics of the cells present in the airway epithelium [6].
The ability of the cells to form polarised layers of confluent cells and consequently the 'tightness' of the epithelium is one of the key parameters that determine the suitability of an in vitro cell model to be used for drug transport and other biopharmaceutical studies [7]. Polarised epithelial cells are connected via junctional complexes comprising of tight junctions and adheren junctions that enable cell-cell adhesion and barrier integrity [8]. These tight junctions form a paracellular barrier that permits the movement of ions and small molecules across the airway epithelium, thus limiting the paracellular transport of drugs [9,10].
Currently, there are several therapies including local anaesthetics, antibiotics, and mucolytic agents that are targeted at the oro-pharyngeal region for various therapeutic and prophylactic purposes. However, there is a lack of an established model that could be used to study these therapies targeted at the oro-pharyngeal region. Lidocaine, a local anaesthetic used before surgical interventions, is commercially available as a throat spray. As the Detroit 562 cell line represents the pharynx, these cells could be utilised to investigate the deposition of such therapies targeted in the oropharyngeal region to understand the therapeutic activity.
Several studies have utilised the human pharyngeal cell line Detroit 562 under ALI and in liquid-covered cultures to investigate immune responses to bacterial colonisation and innate signalling and regulatory pathways [11][12][13]. However, to date, no studies have been performed to systematically characterise and optimise the ALI culture condition (seeding density and culture time) of the Detroit 562 cell line and further determine whether these ALI models could be used to study drug transport.
Realistic physical mouth-throat (MT) models have been developed to predict the in vivo drug deposition in the oropharyngeal region for drug product development and bioequivalence assessment of inhalation drug products [14]. The USP induction port simulating the MT region is the recommended attachment for impaction studies to determine drug deposition patterns by the United States Pharmacopeia. However, as the USP port is a metallic tube with a 90° bend, the design of the port does not include the special geometrical features that are fundamental to accurately predicting the in vivo drug deposition in the throat region. Consequently, several MT models have been developed to address this issue [15][16][17][18]. Among these models, the MT models developed by Byron's group at Virginia Commonwealth University (VCU), widely known as the VCU models, have shown enhanced in vitro-in vivo correlation for several marketed inhalation products compared to the other MT models [19][20][21].
Currently, there are no mouth-throat models that incorporate an in vitro cellular model to study the deposition of aerosol drugs targeted at the oro-pharyngeal region and in turn drug transport across the epithelial layers of the pharyngeal cells for greater physiological relevance. Furthermore, there is no consensus or guidelines for the flow rate or spray angle used for testing such throat spray formulations using the available compendial models. Therefore, our study first aims to determine the appropriate culture method required to develop tight junctions and establish the epithelial barrier properties of the in vitro ALI model of the Detroit 562 cell line and to further investigate whether the developed model could be used to study drug transport, using Lidocaine as a model drug. Next, we focused on modifying the current medium-sized VCU throat model by introducing the novel concept of incorporating pharyngeal cells (Detroit 562 ALI model) within the model. The overall aim of the study is to develop an in vitro throat model incorporating a cellular model representative of the oropharyngeal region to investigate drug deposition and transport of throat sprays across the cellular layers, using Lidocaine throat spray as a model drug. This will be accompanied by studying the impact of flow rates and spray angle on aerosol deposition in the oropharyngeal region for the testing of such formulations.

Cell lines
Detroit 562 (immortalised epithelial human pharyngeal cells-carcinoma derived) were purchased from the American Type Culture Collection (VA, USA). All cells were maintained in minimum essential cell growth medium (MEM)) supplemented with FBS (10%) and L-glutamine (1%) and incubated at 37 °C under 5% CO 2 . All experiments were conducted between passage numbers 51-62 for the Detroit 562 cell line.

Air-liquid interface culture of Detroit 562 cells
To establish an in vitro air-liquid interface (ALI) model of Detroit 562 cells, Transwell cell culture inserts (0.33 cm 2 , polyester terephthalate (PET) membrane, 0.4-µm pore size) (Corning Costar, USA) were used as previously described [11]. To determine the appropriate seeding density for Detroit 562 cell line, three different seeding densities were chosen: 30,000 cells/well (c/w) (0.9 × 10 5 cells/cm 2 ), 60,000 c/w (1.8 × 10 5 cells/cm 2 ), and 80,000 c/w (2.4 × 10 5 cells/ cm 2 ). Briefly, Detroit 562 cells were seeded within the apical chamber in MEM media supplemented with 10% v/v FBS and 1% L-glutamine (known as differentiation media) and the same media was added to the basolateral chamber. The cells were incubated at 37 °C with 5% CO 2 for 24 h until confluency was achieved. To initiate ALI conditions, media in the apical chamber were removed after 24 h indicating day 0 and the cells were maintained under ALI conditions for 21 days. Differentiation media in the basolateral chamber were replaced every 2 days. All experiments were performed on day 7, 14, 18, and 21 post ALI induction.

Transepithelial electrical resistance
Transepithelial electrical resistance (TEER) was measured as described previously [4,22]. Briefly, pre-warmed media were added to the apical chamber and allowed to equilibrate for 30 min at 37 °C under 5% CO 2 . TEER was measured using EVOM2 ® epithelial voltohmmeter (World Precision Instruments, USA) attached to STX-2 chopstick electrodes for the ALI cultures, corrected by subtracting the blank inserts, and multiplied by the area of the Transwell inserts (0.33 cm 2 ) and six measurements were taken per Transwell.

Sodium fluorescein permeability assay
Tight junction functionality and paracellular permeability of Detroit 562 ALI cultures were determined using the sodium fluorescein permeability assay [4,23]. Sodium fluorescein (2.5 mg/mL) (Sigma-Aldrich, Sydney, Australia) was added to the apical chamber and pre-warmed Hanks' balanced salt solution (HBSS) was added to the basolateral chamber. Transwells were incubated for 4 h at 37 °C with 5% CO 2 , with basolateral samples (100 µL) collected and replaced with fresh HBSS after every 30 min for the first 2 h and then every hour for the final 2 h to measure the rate of transport (flux) of the sodium fluorescein from the apical chamber to the basolateral chamber. For analysis, the collected basolateral samples were diluted (1:20) and fluorescence was measured using the SpectraMax M2 plate reader (excitation: 485 nm; emission: 538 nm). The permeation coefficient (P app ) was calculated according to Eq. 1.
where dQ/dT represents the flux of sodium fluorescein (µg/s) across the membrane, C 0 is the initial donor concentration (µg/mL), and A is the surface area (cm 2 ).

Live and dead cell staining
LIVE/DEAD ® Viability/Cytotoxicity Kit (Molecular Probes) and the Hoechst stain (Sigma-Aldrich) were used to stain the ALI cultured cell layers. The assay was performed as per the manufacturer's instructions and according to a previously published report [4]. Briefly, the cell layer was washed 3 times with pre-warmed PBS and 2 μM calcein AM and 4 μM ethidium homodimer-1 (EthD-1) were added to the apical compartment. Cells were incubated with the solution for 30 min in the dark at room temperature. Cells were incubated with Hoechst (1:10000) for 10 min to stain the nucleus. The Transwell membranes containing the cell layer were excised and mounted on a glass microscope slide for analysis. Cells were imaged using a Nikon ECLIPSE Ti inverted microscope controlled by the NIS Elements software (Nikon) and equipped with the APO Fluor 20 × air objective. Images were captured using a CoolSNAP ES2 high-resolution digital camera (Photometrics). Ten images of different fields of view were taken per Transwell membrane. (1)

Mucus production
Mucus production of the ALI cultures was characterised by staining the glycoproteins (mucins) with alcian blue as previously described [4,22]. Briefly, cell layers were washed twice with prewarmed PBS and fixed using 4% paraformaldehyde (v/v) for 15 min. Subsequently, the cells were washed thrice with PBS and Alcian blue (1% w/v Alcian blue in 3% v/v acetic acid/water at pH 2.5) was added to the apical chamber and incubated for 20 min. The cell layer was washed up to 10 times with PBS to remove excess Alcian blue and allowed to air-dry for 3 h at room temperature. The Transwell membranes containing the cell layer were excised and mounted on a glass microscope slide for analysis. Mucus staining was imaged using an Olympus BX61 microscope (Olympus) equipped with an Olympus DP71 camera and a 20 × air objective. Ten different fields of view were captured per Transwell membrane. Images were analysed using Image J software (NIH).

Evaluation of cytokine production and inflammatory responses
Lipopolysaccharide (LPS) from Escherichia coli (Sigma-Aldrich, Sydney, Australia) was resuspended at 10 µg/ mL in differentiation media (MEM with 10% FBS and 1% L-glutamine) and added to the basolateral chamber to stimulate the cells to model bacterial infection [24]. The cells were stimulated with polyinosinic-polycytidylic acid (poly (I:C)) (10 µg/mL, Sigma-Aldrich, Sydney, Australia) by resuspending in the differentiation media to model viral infection [13]. Cells were then incubated at 37 °C under 5% CO 2 for 24 and 48 h and untreated cells served as the control. After treatment, samples were collected from the basolateral culture medium for subsequent analysis of IL-6, IL-8, and IL-1β cytokine production using an enzyme-linked immunosorbent assay (ELISA) kit (BD OptEIA, BD Biosciences, USA) according to the manufacturer's instructions.

Immunofluorescence
The presence of tight junctions in the ALI cultures was visualised by immunolabeling tight junction proteins zonula occludens-1 (ZO-1) as previously described [4]. Cell layers were washed 3 times with PBS and fixed with 4% paraformaldehyde (v/v) for 15 min. The cell layers were then washed 3 times with PBS and permeabilised following a 10 min incubation with 0.2% Triton X-100 (v/v) and then blocked and quenched with 10% normal goat serum (v/v) (Invitrogen, Australia) and 0.3 M glycine (Sigma-Aldrich, Australia) respectively, and incubated for 1 h at room temperature. The primary antibody, ZO-1 rabbit polyclonal (Abcam, UK) (1:200) was incubated overnight at 4 °C. The next day, the cell layers were washed 3 times with PBS and incubated with goat anti-rabbit Alexa Fluor ® 488 (Life Technologies, Thermofisher, Australia) (1:500) for 2 h at room temperature. Cell layers were then counterstained with DAPI (Sigma-Aldrich, Australia) (1:10000) and incubated for 30 min at room temperature. Finally, Transwell membranes containing the cell layer were excised and mounted using FluoroSave mounting media (Merck Millipore, Australia) on a glass microscope slide for analysis.
Cells were imaged using a confocal microscope (Nikon Eclipse Ti) equipped with a Plan Apo VC 60 × oil objective. Images were taken using the resonant scanner at a step size of 0.31 µm, 512 × 512 pixels with an average line scan of 16. For cells immunolabelled with FITC-tagged secondary antibodies, the 488-nm laser was set to 3.5% with a smart gain of 55 V and an offset of − 3%. For nuclei excitation, the 405-nm laser was again set to 3% with the smart gain and offset set to 50 V and − 1% respectively.

Lidocaine transport study
To investigate whether the developed ALI model of the Detroit 562 cell line could be used to study drug transport, a study was conducted using Lidocaine as a model drug as previously reported [4,23]. Lidocaine transport across the Detroit 562 ALI cultures was conducted on day 18 post ALI formation. Lidocaine was dissolved in ethanol to produce the stock solution and then further diluted in HBSS to prepare a 20 µg/mL Lidocaine (0.1% ethanol in final Lidocaine solution) to be used for the transport study. Lidocaine solution was added to the apical chamber and HBSS was added to the basolateral chamber. Samples (100 µL) were taken from the basolateral chamber every 30 min for the first 2 h and then every hour for the final 2 h, with samples being replaced by fresh, warm HBSS. After the 4 h assay, the apical chamber was washed twice with HBSS to collect any residual drug using a pipette (denoted as On) and the cell layer was then scraped from the insert membrane and lysed using CelLytic™ buffer (Invitrogen, Australia) to quantify the amount of drug inside the cells (denoted as Cellular). TEER measurements were performed before and after the transport, study to check whether drug deposition altered the epithelial barrier integrity of Detroit 562 ALI culture models. All the samples were subsequently analysed using high-performance liquid chromatography (HPLC) using the quantification method described in the next section.

HPLC quantification method for Lidocaine
All Lidocaine samples were analysed using a high-performance liquid chromatography (HPLC) system equipped with SPD-20A UV-Vis detector, an LC-20AT liquid chromatograph, a SIL-20A HT autosampler (Shimadzu) and a Kinetex C-18 column (250 × 4.6 mm, 5 µm, Phenomenex, Torrance, USA), according to a validated method [25]. The mobile phase was a mixture of acetonitrile: phosphate buffer (26:74 (v/v) with pH 5.5 adjusted using sodium hydroxide (Sigma-Aldrich). Samples were analysed at 230 nm at a flow rate of 1.0 mL/ min and an injection volume of 10 µL. Linearity was obtained between 0.2 and 100 µg/mL (R 2 = 0.99) with a retention time of 8 min.

The development of a 3D-printed throat model incorporated with cells
To evaluate drug deposition and transport of inhaled drugs targeted at the oropharyngeal region, a realistic and more physiologically relevant throat model was designed and developed to include the integration of cells for enhanced in vitro-in vivo correlation. A computer-aided (CAD) design of the medium-sized Virginia Commonwealth University (VCU) throat model was prepared by AutoCAD ® (version 23, USA). The design was modified to connect two separate lower and upper pieces and insertion of two Snapwell inserts (denoted as Upper and lower Snapwells respectively) in which cells grown in ALI conditions could be incorporated for subsequent deposition and transport studies (Fig. 1). The prepared 3D design was then 3D-printed using clear photopolymer resin (FLGPCL02, Formlabs Inc., USA) by stereolithography (SLA), using Form 2 (Formlabs Inc., USA).

In vitro aerosol deposition using USP-IP and the 3D-printed throat models
Deposition profiles of the Lidocaine spray targeted to the throat were determined using the European Pharmacopeia Apparatus E, Next Generation Impactor (NGI) (Copley Instruments Ltd) fitted with the USP stainless steel 90° induction port (US-IP model), as specified in European To optimise the conditions for studying the throat deposition of Lidocaine spray, the experiment was performed at two different angles of spraying (45° and 90°) with 3 different flow rates of 0, 15, and 30 L/min representing no airflow, light breathing and normal breathing condition respectively. Briefly, the NGI was connected to a high-capacity vacuum pump, and the flow rate was set using a flow meter (Model 4040, TSI Precision Measurement instruments). Lidocaine throat spray was primed by firing 5 shots to waste and weighed before each shot. The distance between the spray nozzle and the throat was measured at 7 cm for each NGI experiment to ensure that the spray is aimed primarily toward the throat and not in the oral cavity. The Lidocaine throat spray was attached to the impactor via an airtight adaptor with an actuation time of 4 s. Following the completion of the delivered dose, all components of the NGI (actuator, adaptor, IP, stages 1-7 and micro-orifice collector (MOC)) were washed with acetonitrile: phosphate buffer (26:74 (v/v), transferred to volumetric flasks and sonicated for 10 min. Samples were then filtered (0.45 μm, PTFE) and Lidocaine was quantified using HPLC. Subsequently, the flow rate that resulted in maximum Lidocaine deposition in the throat region was used to determine drug deposition using our novel 3D-printed VCU model at both 45 and 90° angles. All the samples were subsequently analysed using a High-Performance Liquid Chromatography (HPLC) quantification method for Lidocaine described in the previous section.

Transport of lidocaine throat spray using the 3D-printed VCU model integrated with Detroit 562 ALI culture
To investigate whether the developed novel 3D-printed VCU model integrated with the ALI model of Detroit 562 cells grown on Snapwells could be used to study drug transport of Lidocaine throat spray, a transport experiment was conducted over a 4-h period. Prior to conducting the transport study, optimisation of the number of shots of the throat spray on the cellular layers of the Detroit 562 ALI culture was performed at the optimised flow rate of 30L/min and an angle of 45° using 1 shot and 3 shots of Lidocaine spray. Using the optimised conditions, deposition, and transport of Lidocaine spray across the Detroit 562 cells grown in ALI conditions on Snapwell inserts placed within the 3D-printed VCU throat model were studied. The Snapwells with the Detroit 562 cells were placed in the lower and upper part of the 3D throat model and hence referred to as lower and upper Snapwells respectively. Lidocaine throat spray was attached to the impactor via an airtight adaptor and one shot was fired into the 3D throat at an angle of 45°. After deposition, the Snapwell inserts were removed from the 3D-printed throat and transferred into culture plates with 2 mL of fresh HBSS added into the basal chamber. Samples (200 µL) were taken from the basolateral chamber every 30 min for the first 2 h and then every hour for the final 2 h, with samples being replaced by fresh, warm HBSS. The same method of transport study was followed as described in the previous section to determine the amount of drug transported during and after the 4 h period (Transported) and to evaluate the amount of drug present inside the cells (IN) and remaining on the cells (ON). Additionally, to determine whether drug deposition and transport study altered the epithelial barrier integrity of the Detroit 562 cells, sodium fluorescein permeability assay was conducted on untreated cells that served as control and on treated cells following Lidocaine deposition and post 4 h transport study as described earlier in the previous section.

Statistical analysis
All results are expressed as mean ± standard error of the mean (SEM) of at least three biological replicates. Statistical software, GraphPad Prism (version 8.2.1) was used to test for significance using one-way or two-way ANOVA for each experiment. Significance was determined as p < 0.05.

Tight junction formation on day 18 of the ALI culture period
To predict the time for functional tight junctions to develop in the ALI for Detroit 562 cell line and determine the optimum seeding density, TEER measurements and permeability of the known paracellular marker flu-Na were tested on days 7, 14, 18, and 21 of the ALI culture periods for all 3 densities (30,000 c/w, 60,000 c/w, and 80,000 c/w). TEER values significantly increased from day 7 to 18 with no significant changes in TEER values between day 18 and 21 for all the 3 densities, thus indicating progressive tight junction formation until day 18 of the ALI culture period (Fig. 2A). Correspondingly, a significant decrease in apparent permeability (P app ) of flu-Na was observed on day 18 compared to day 7 and 14 for 60,000 c/w and 80,000 c/w ALI models and plateaued till day 21, but not for 30,000 c/w (Fig. 2B). These results suggest that the cell layers have developed functional tight junctions in ALI culture at day 18 for 60,000 c/w and 80,000 c/w ALI models. Therefore, these two densities, termed low and high density, respectively, were used for the subsequent experiments as 30,000 c/w ALI models would require a longer time to develop tight junctions and attain epithelial barrier integrity and an extended culture period may induce greater cell death.

Abundance of live cells on day 18 of the ALI culture period
A cell viability assay was performed to verify cell survival over the 18-day ALI culture period as extended culture periods can induce cell death and compromise the integrity of cellular barrier properties [26]. The cells cultured in ALI conditions were stained with calcein AM and EthD-1 to determine the live and dead cells, respectively (Fig. 3). The fluorescent micrographs show live cells (green), dead cells (red), and nuclei (blue). Calcein AM dye only penetrates through the cellular boundaries of the live cells, staining them as green thus indicating intact cellular membranes, while EthD-1 stains the cells that have lost their membrane integrity as red, indicating the presence of dead cells. Detroit 562 cells seeded at low density showed an abundance of live cells on day 7 (Fig. 3A) that was found to be like day 18 ALI cultures (Fig. 3B). Similar results were observed for high-density ALI cultures (Fig. 3C, D), suggesting that the cells are viable following 18 days of ALI culture and can be used for experimentation up until 18 days of ALI culture for both densities. However, the number of dead cells observed on day 18 for low density (Fig. 2B) is notably lower compared to high density (Fig. 3D), suggesting the feasibility of the use of low-density ALI cultures.

Mucus production increases throughout 18 days of the ALI culture period
To determine the extent of differentiation of the Detroit 562 cell line over the 18-day culture period in ALI, mucus production was assessed. Alcian blue was used as a marker of mucus production as it stains the glycoproteins present in mucus, producing a blue colour. Mucus production significantly increased for both the densities of the Detroit 562 cell line (Fig. 4) during the 18 days ALI culture period, suggesting continued differentiation throughout this period. Day 1 Alcian blue staining of ALI models of the Detroit 562 cell lines showed indistinct, weak blue staining in the micrographs for both Low and High density (Fig. 4A, B). Comparatively, the Alcian blue staining increased with greater blue intensity observed in distinct patches on day 18 for both the densities (Fig. 4C, D), suggesting the cells continued differentiation until 18 days of the ALI culture period for Detroit 562 cells. Therefore, the increase in mucus production following 18 days of ALI indicates that the Detroit 562 cell line has a mucus-secreting phenotype that continues to differentiate until day 18 of the ALI culture period. These results are in accordance with previous reports [11,12].

Significant IL-8 production by the Detroit 562 cells in response to LPS and poly (I:C)
To determine whether the ALI model of the Detroit 562 cell line could be used to study responses to bacterial and Fig. 4 Mucus production of Detroit 562 ALI cultures on day 1 of ALI culture period for A low and B high density ALI cultures compared to Day 18 for C low-and D high-density ALI cultures (scale bar = 0.5 mm) viral infection, as the throat is often the primary site for these kinds of infections, the Detroit 562 cells on day 18 of the ALI culture period were stimulated with the bacterial endotoxin lipopolysaccharide to mimic bacterial infection [24] and the viral component poly (I:C) to simulate viral infection [13]. No significant increase in IL-6 and IL-1β production (data not shown) compared to control (unstimulated cells) was observed in the ALI models for both low and high densities. On the contrary, a significant increase in IL-8 production was observed in the LPS-stimulated cells post 24 and 48 h of stimulation (Fig. 5A) compared to control (p < 0.0001 respectively) for both low-and highdensity ALI models, suggesting that these models could be used to mimic bacterial infection. As no significant differences in IL-8 production were observed between 24 and 48 h of LPS stimulation for both densities (Fig. 5A), 24 h of LPS stimulation can be considered sufficient to model bacterial infection for these ALI models.
Like the IL-8 cytokine secretion profile of LPS-stimulated Detroit 562 cells (Fig. 5A), IL-8 production by the poly: IC stimulated Detroit 562 cells following the same pattern (Fig. 5B). poly (I:C)-mediated IL-8 secretion was significantly increased at both low and high densities of the Detroit 562 cell line post 24 and 48 h of stimulation (Fig. 5B). Comparable concentrations were observed between the two densities at both time points, agreeing with previously published reports [27,28], reinforcing the finding that the ALI model could be used to model viral infection post 24 h of poly (I:C) stimulation. Additionally, no significant changes in the TEER measurements (Fig. 5C) and P app values (Fig. 5D) of the LPS and poly (I:C) stimulated cells were found compared to the control for both the densities, thus indicating that the epithelial barrier integrity of the Detroit 562 cells was maintained post-stimulation. These results further confirm the suitability of the day 18 Detroit 562 ALI model to simulate both bacterial and viral infection.

Confirmation of tight junction formation
To corroborate the development of tight junctions of the ALI models on day 18 for the Detroit 562 cell line, the ALI models were visualised by immunolabelling the cells with markers of tight junction proteins zonula occludens (ZO-1) over the 18 days of ALI culture period. Z-projections of the confocal micrographs show ZO-1 (green), and nuclei (blue) and the merged images are presented in Fig. 5A and B. Punctate formation of ZO-1-positive tight junctions was only visible on day 18 of the ALI culture period (data not shown for other days 7 and 14) (Fig. 6A, B). ZO-1-positive tight junctions were observed along cell-cell contacts throughout the cellular layers for both the low and high densities of the Detroit 562 cell line (Fig. 6A, B, shown by the white arrows), similar to previously published results [29]. Overall, the visualisation of the tight junction formation of the day 18 Detroit 562 ALI cultures for both densities (Fig. 6A, B) corroborates the earlier findings of the significant increase in TEER measurements of these ALI models on day 18 compared to day 7 ( Fig. 2A) and the corresponding decrease in P app of the paracellular marker (Fig. 2B). Taken together, these results suggest complete differentiation of the Detroit 562 cell line at day 18 of the ALI culture period and attainment of epithelial barrier integrity, irrespective of the seeding densities used and fitness of these in vitro models to investigate drug transport.

Lidocaine transport across the Detroit 562 ALI models
A transport study was conducted to determine if the differentiated ALI models of Detroit 562 cells could be used to study drug transport, using Lidocaine as a model drug on day 18 post ALI. The cumulative mass of Lidocaine in the basal samples increased with time for both low-and high-density ALI cultures (Fig. 7A) with no significant differences, suggesting that Lidocaine had been transported through both the ALI cultures. Notably, no Lidocaine was found inside the cells (shown as cellular, Fig. 7B), suggesting that Lidocaine may have been transported using the paracellular pathway.
Seeding density is a key factor that influences the formation of tight junctions and higher densities may negatively impact the transport of certain drugs [30]. The percentage of the total mass of Lidocaine transported after 4 h across the low-and high-density ALI cultures were not significantly different (low: 47.3 ± 2.36%, high: 47.53 ± 1.67%) (Fig. 6B), thus indicating the suitability of the low-density ALI culture of Detroit 562 cell line as an in vitro model of drug transport. Furthermore, no significant changes in TEER values (Fig. 7C) and apparent permeability (P app ) values (Fig. 7D) were observed between pre and post Lidocaine transport study for both densities, confirming that Lidocaine deposition and transport did not alter the epithelial barrier integrity of both low-and high-density ALI cultures. Comparative analysis of the apparent permeability (P app ) of flu-Na and Lidocaine in the ALI cultures on day 18 showed that Lidocaine was transported at a significantly higher rate than the flu-Na molecule for both low-and high-density ALI cultures (p < 0.0001 respectively, Fig. 7E). The significantly increased transport of Lidocaine cannot be attributed to charge as both the Lidocaine and flu-Na molecules are negatively charged. Thus, the increased permeability of Lidocaine may be due to the difference in the molecular weight of the two molecules, with a significantly greater number of the Lidocaine molecules (MW 234.3 g/mol) able to permeate via the tight junctions compared to flu-Na (MW 376.3 g/mol). Overall, the results suggest the involvement of the paracellular pathway in the transport of Lidocaine across the Detroit 562 ALI cultures, but this needs to be further explored.
Altogether, the study indicates that the Detroit 562 cell line completely differentiates at day 18 in ALI and the suitability of the low-density ALI culture (60,000 cells/well or 1.8 × 10 5 cells/cm 2 ) as a representative in vitro air-liquid interface cellular model to study drug transport on day 18 of the ALI culture period.

Optimisation of the experimental parameters for the novel 3D-printed throat model
The experimental parameters (specifically the flow rate and angle at which the Lidocaine throat spray was positioned for each shot) required to study drug deposition in the throat region were determined using the currently used USP-IP throat model. Lidocaine deposition was detected in the mouthpiece, throat, and Stage 1 (shown as S1, Fig. 8A, B) for all the 3 flow rates (0, 15, and 30 L/min) at both 45° (Fig. 8A) and 90° (Fig. 8B) but was not found in the lower stages of the NGI (stages 2-7, data not shown). No significant differences in mass of drug deposited in the throat region were observed between the three flow rates (0, 15, 30 L/ min) at 45° angle (shown as throat, Fig. 8A) and at 90° angle (Fig. 8B). Therefore, a flow rate of 30 L/min was chosen as A Cumulative mass of Lidocaine transported over 4 h period at lowdensity and high-density cells of Detroit 562 ALI models, B percentage of the total mass of Lidocaine transported across the ALI cultures (shown as transported), remaining on the apical layers (on) and inside the cells (cellular) in low-and high-density cells at the end of the experiment (4 h), C TEER measurements, and D Flu-Na permeability of ALI cultures before drug deposition (control) and after 4 h of Lidocaine transport (shown as post transport). E Apparent permeability (Papp) comparison between flu-Na and Lidocaine across ALI cultures (n = 3, mean ± SEM) (****p < 0.0001) using two-way ANOVA with Tukey's post test) the optimum flow rate for subsequent studies as it represents normal breathing conditions. Furthermore, since there were no significant differences in the Lidocaine mass deposited in the throat region between the two angles − 45° and 90° at the optimised flow rate of 30 L/min (Table 1), the 45° angle was chosen for the subsequent experiments as it represents the position at which the throat spray is usually held by the users.
Lidocaine deposition studies were then performed using the physiologically representative 3D-printed throat model at the optimised flow rate of 30 L/min and an angle of 45º. The results showed a significant increase in Lidocaine deposition in the throat region of the 3D-printed throat model in comparison to the mouthpiece and stage 1 (Fig. 9), with no drug detected in the lower stages of the NGI (stages 2-7, data not shown), thus showing the suitability of the throat model to study drug deposition. A comparison of Lidocaine deposition between the USP-IP throat model and the developed 3D-printed VCU model further showed no significant differences in drug deposition between the two throat models (Table 2). Overall, these results show that the developed 3D-printed throat model could be used to study Lidocaine deposition at a flow rate of 30 L/min and an angle of 45°.

Transport study of Lidocaine using the novel 3D-printed throat model integrated with an in vitrocellular model
Before conducting the transport study, the number of shots delivered from the Lidocaine throat spray to the cells incorporated within the 3-D throat model required to achieve a suitable amount that would be above the limit of analytical detection was optimised using empty Snapwell inserts (Fig. 10A). The optimisation of the number of shots was also performed as a greater number of shots may change the epithelial barrier integrity of the cells, compromising L/min) (N = 3, mean ± SEM) **p < 0.01, ***p < 0.001, ****p < 0.0001 (using two-way ANOVA with Tukey's post test) Table 1 Comparison of Lidocaine mass deposition in the USP-IP throat region using the simulated respiratory tract (Next-Generation Impactor (NGI) between 45° and 90° angles at optimised flow rate of 30 L/min (N = 3, mean ± SEM) ns no significance (using one-way ANOVA with Tukey's post test) Flow rate (L/min) Mass deposited (mg) at 45°M ass deposited (mg) at 90°S tatistical Analysis 30 20.9 ± 0.7 17.9 ± 0.5 ns Fig. 9 Lidocaine deposition in the mouthpiece, 3D-printed throat model and stage 1 (S1) using the simulated respiratory tract (nextgeneration impactor (NGI) at the optimised flow rate of 30 L/min and 45° angle (N = 3, mean ± SEM) ****p < 0.0001 (using one-way ANOVA with Tukey's post test) the cell viability. The number of shots chosen to perform the transport study was 1 shot as it is likely to have less impact on the cellular layers of the Detroit 562 ALI model in comparison to 3 shots and thus will enable the cells to retain their epithelial barrier integrity after drug deposition and during the 4 h transport study.
Next, the completely differentiated in vitro ALI cultures of Detroit 562 cells grown on Snapwells were placed inside the 3D-printed throat model (one Snapwell in the upper position referred to as Upper and the other Snapwell in the lower position referred to as Lower). Then, one shot of Lidocaine spray was fired onto the throat model at a flow rate of 30 L/min −1 using an angle of 45° to study drug transport across the developed in vitro throat model integrated with the cellular model. The cumulative mass of Lidocaine transported across the cellular layers of the Detroit 562 ALI cultures significantly increased with time for upper Snapwell (shown as white circles, Fig. 10B) from 0.5 h until 2 h (p < 0.001, Fig. 10B) and then plateaued until 4 h (Fig. 10B), suggesting that Lidocaine was transported across the ALI cultures. Comparatively, no significant differences were observed for the cumulative mass of Lidocaine transported for the lower Snapwell ing on the apical layers (on) and inside the cells (cellular) of the lower and higher Snapwells at the end of the experiment (4 h), and D Flu-Na permeability of ALI cultures before drug deposition (control) and post 4 h of Lidocaine transport (n = 3, mean ± SEM) (*p < 0.05, **p < 0.01, ****p < 0.0001) (shown as black circles, Fig. 10B) between 0.5 h and 2 h and the remaining of the transport study. Furthermore, the mass of Lidocaine transported after 4 h in the upper Snapwell (76.5 ± 5.7 µg) was significantly higher compared to the lower Snapwell (8.24 µg ± 6.45 µg). This data correlates with the results of the optimisation experiment for the number of shots that showed significantly decreased Lidocaine deposition for the lower Snapwell compared to the upper Snapwell (Fig. 10A) indicating that the position of the Snapwells affects drug deposition and transport. Importantly, a significantly higher percentage of Lidocaine was transported in case of the aerosol spray at a significantly greater rate (Fig. 9C, D) compared to the Lidocaine drug solution (Fig. 7B, E) at the end of the 4-h transport study. These results show higher drug transport of the aerosol spray compared to the drug solution using the 3D-printed throat model, thus highlighting the feasibility of the developed novel 3D-printed throat model to study drug transport using Lidocaine aerosol spray. Furthermore, a significantly lower percentage of drug was found inside the cells compared to the percentage of drug transported (Fig. 9C), corroborating with the findings of the transport study of Lidocaine solution (Fig. 7B), suggesting that Lidocaine may have followed the paracellular pathway. However, the primary route of transport for Lidocaine remains unclear and needs to be further investigated.
To determine whether Lidocaine drug deposition and 4 h of transport study altered the epithelial barrier integrity of the Detroit 562 ALI culture, a sodium fluorescein permeability assay was performed on the ALI cultures post transport and compared to untreated cells (control). No significant differences in apparent permeability of Flu-Na (P app ) were found between the lower Snapwell and the control (Fig. 10D), suggesting that drug deposition did not affect the epithelial barrier integrity of the cells in the lower Snapwell. However, a significantly higher flu-Na P app value was observed for the upper Snapwell in comparison to control (untreated cells) (p < 0.01, Fig. 10D). This could be due to significantly higher drug deposition in the upper Snapwell that led to the opening up of the tight junctions by Lidocaine, thus increasing Flu-Na permeability and in turn paracellular flux as reported in a recent study [31]. Overall, the study showed that the novel 3D-printed throat model incorporating the in vitro cellular model could be used to investigate drug transport of therapies targeted at the throat region.

Conclusion
The present study indicated the suitability of the novel 3D-printed throat model incorporating Detroit 562 cells grown in the ALI configuration as a representative in vitro throat model to investigate drug transport of therapies targeted at the oro-pharyngeal region. This study has shown that the Detroit 562 cells completely differentiate on day 18 of ALI at an optimised density of 60,000 cells/well (1.8 × 10 5 cells/cm 2 ) with significant mucus production, showing response in IL-8 production to LPS and poly (I:C) stimulus and development of functional tight junctions. Importantly, the developed cell integrated throat model showed Lidocaine transport, possibly via the paracellular pathway, suggesting the suitability of the novel in vitro model to study drug transport. Overall, the study highlights the potential of the novel 3D-printed throat model integrated with the optimised cellular model of the Detroit 562 cell line as a promising in vitro model to investigate the transport of inhaled drug therapies and predict the therapeutic efficacy of drug-aerosol particles targeting the throat region. Future studies may involve an extension of the current setup towards a complex co-culture system of primary epithelial cells of throat origin for an enhanced in vitro-in vivo correlation.