BEAS-2B cells were maintained at 37°C in an atmosphere of 5% CO₂ in Airway Epithelial Cell Growth Medium (AEGM), that consisted of AEGM (Promocell, C-21060) complemented with SupplementMix (Promocell, C-39165) containing Bovine Pituitary Extract 0.004 mL/mL, epidermal growth factor (10 ng/mL), insulin (5 μg/mL), Hydrocortisone (0.5 μg/mL), Epinephrine (0.5 μg/mL), Triiodo-L-thyronine (6.7 ng/mL); holo-Transferrin (10 μg/mL), Retinoic acid (0.1 ng/mL). Sub-cultivation was performed twice a week in T175 cell culture flasks with a cell seeding density of 8.7E5 and 5E5 cells per flask when cultivated over 3 and 4 days respectively.

Cells were seeded on cell culture inserts designed to be used with microscopic analysis technologies (Millicell Cell culture inserts: Millipore PICM01250). In brief: Cells were seeded using 400 µL of a cell suspension with 3.5E5 cells/mL per cell culture insert. The loaded cell culture inserts were placed into 24-well plates filled with 250 µL AEGM medium. Cells were incubated overnight (appr. 18 h) at 37°C and 5% CO₂ to allow adherence and growth on the cell culture insert membrane. On the next day just before exposure, the medium from the apical compartment of the inserts was removed and the inserts were transferred into new 24-well plates with 250 µL HEPES buffered Dulbecco’s minimal essential medium per well. A 24-well plate with the inserts was placed into one of the exposure chambers of the Smoke Aerosol Exposure In Vitro System [SAEIVS, (Wieczorek et al., 2023)]. Smoke/aerosol exposure was executed as described in Table 1. Following exposure, the plate with the inserts was removed from the SAEIVS.

Cell culture inserts with the treated cells were transferred into new 24-well plates filled with 250 µL AEGM medium. Pre-warmed AEGM medium (400 µL) was added to the cells in the apical compartment of the inserts. Cells were allowed to recover and kept in the incubator for 24 h at 37°C and 5% CO₂. Medium was removed from both compartments and 200 µL 4% formaldehyde in PBS was added to the apical part of the insert for a 15-min fixation step at room temperature.

To compare the effective concentrations of each product type, a study was designed to assess both the number of puffs delivered to the cells (BEAS-2B) and the amount of nicotine delivered at the cell surface. Separate experiments were adopted using glass plates of the same surface areas as the Millipore inserts (and therefore the cell layer surface) and the same puffing parameters and diluted/undiluted smoke/aerosol from the same product variants. For cell exposure, cigarette smoke (1R6F) was diluted 1:5 (smoke: fresh humidified filtered air) and 1:1 (aerosol: fresh humidified filtered air) for the HTP. The aim of this experiment was to determine the number of puffs required for each product to cause a minimum effective concentration (MEC) for the c-jun cellular marker, measured using high-content screening (HCS) and to determine the number of puffs required to deliver nicotine to the cell surface (using glass plates to collect the deposited material). The marker c-jun was chosen as this was the most sensitive of a panel of selected cellular stress markers used with the HCS technique in-house. The aim was to be as conservative as possible, by using the most sensitive cellular marker using our experience with HCS. The MEC is defined as the lowest effective concentration outside of the background negative control range. The MEC had to exceed the calculated background [=3x median absolute deviation (mad)], that was determined by linear or non-linear regression (decided by the best curve fit). For comparative MEC calculation, the average 3x mad of all treatments for c-jun was used as target value. The Phosphorylation of c-jun (p-c-jun) defines a cellular stress marker which is involved in several signal pathways including proliferation, apoptosis, survival, tumorigenesis, and tissue morphogenesis (Schmeck et al., 2006; Dreij et al., 2010).

Previous work has assessed the effect of adding cells (V79 Chinese Hamster lung fibroblast cells and BEAS-2B) to the surface of the glass and compared to the cell-free glass surface. The deposition of nicotine on both surfaces was comparable and so clean glass slides were subsequently used (Wieczorek et al., 2023).

Nicotine was quantified using LC-MS/MS method (Internal Standard: Nicotine-d4). Nicotine trapped in cell media and PBS was measured directly without any further sample preparation. Whereas the nicotine trapped on the surface of the glass disc was eluted with isopropanol prior to final measurement.

Nicotine quantification of isopropanol extracts from exposed glass discs and basal medium was conducted using LC-MS/MS (AB Sciex API 6500 QTRAP (SCIEX, Darmstadt, Germany)). For analysis, medium samples were diluted 1:100 with MilliQ water/MeOH (1:1) and 1:1 in the autosampler with the internal standard solution in methanol. A Gemini NX-C18 column (110 Å, 100 × 2.0 mm, 3 µm) (Phenomenex, Aschaffenburg, Germany) was used for the liquid chromatography (oven temperature, 55°C), sample injection volume was 5 µL and the autosampler temperature was 5°C. The eluent gradient was applied according to the following: 0min: 2% B (methanol)/98% A (0.05% acetic acid in water) (flow rate: 400 μL/min); 1.2 min: 65% B/35% A (400 µ/min); 1.5 min: 95% B/5% A (400 µ/min); 2.5 min: 98% B/2% A (400 µ/min); 3.0 min: 98% B/2% A (400 µ/min). The following conditions were used for the mass spectrometry: Ion spray voltage: 4500V, Ion source temperature: 500°C, MRM: 163/132 quantification; 163/106 qualifier. The isopropanol samples were diluted 1:100 with MilliQ water/MeOH (1:1) and without autosampler dilution because internal standards were present in samples extraction and calibration solution. All other measurement parameters were the same as above.

To trigger the p-c-jun response in BEAS-2B cells, the cells were exposed to fresh smoke/aerosol from 1R6F/Pulze used with iD sticks (HTP). For details of the Pulze device and the iD sticks please see Chapman et al. (2023). The BEAS-2B cells were pre-grown on microporous membranes of dedicated cell culture inserts and supplied with medium apically and basally. For exposure to fresh smoke from the Reference Cigarette 1R6F and aerosol from Pulze and iD the cells were switched to ALI conditions (i.e., apical medium was aspirated) where nutrification of the cells is achieved from the lower compartment below the insert-containing well only. The use of the ALI exposure is key. In vitro exposure systems that deliver aerosols to the surface of human cells cultured at ALI are of particular importance being the most physiologically relevant exposure route, and highly preferable to using submerged cell lines (Upadhyay and Palmberg, 2018). Upadhyay and Palmberg (2018) identified several factors influencing the successful development of ALI-models including the choice of cell line (preferably human), the source of any primary cells, and the use of co-culture systems consisting of multiple cell types (Clippinger et al., 2016). The use of an ALI system is also considered to be a feasible alternative approach and can be used to implement the “3 Rs principle” replacement, reduction, and refinement of animal usage—in conducting pulmonary toxicity studies (Upadhyay and Palmberg, 2018). The apical surface of the inserts with the cells were exposed directly to fresh aerosol and smoke in the in-house Smoke and Aerosol Exposure In vitro System (SAEIVS). This is an in-house built system to enable the delivery of whole aerosols directly to cells at the ALI, being able to deliver different dilutions of aerosol/smoke to two separate cell exposure chambers in under 10 s of generation [for more details and characterization of the system see Figure 1; Wieczorek et al. (2023)]. Following exposure to the aerosol/smoke the cells were covered with recovery medium and were incubated for 24 h before fixation and subsequent antibody staining for p-c-jun.

The study was composed of two parts.

1 - The quantification of phosphorylated c-jun protein in nuclei of BEAS-2B cells after direct exposure to diluted fresh smoke of 1R6F Reference cigarette and to undiluted fresh aerosol of Pulze and iD sticks at the ALI using the SAEIVS. The aim of this part of the study was to determine the p-c-jun MEC for both test articles on a puff basis by means of HCS technology.

Following the determination of MEC and finalization of nicotine dosimetry, the data were combined to determine the MEC on a nicotine basis. To this end the MEC_puff from the HCS approach was used to determine the MEC_nic by using the MEC_puff in the equation of the nicotine dosimetry.

The absorption, distribution, metabolism, and excretion (ADME) properties of nicotine have been previously studied in different species (Benowitz et al., 2009a; Benowitz et al., 2009b). Nicotine is metabolized quickly in the liver, primarily by cytochrome P450s (CYP2B6 and CYP2E1) and has relatively low plasma protein binding (Benowitz et al., 2009a; Benowitz et al., 2009b). Large venous blood nicotine concentrations are produced within minutes after nicotine inhalation as the lungs offer a large surface area for absorption and a favorable dissolution pH of 7.4. Several PBPK models have been developed for nicotine (Plowchalk et al., 1992; Robinson et al., 1992; Teeguarden et al., 2013; Haghnegahdar et al., 2018).

The PBPK model developed here has perfusion-limited compartments for liver and lung and lumped compartment for the remaining tissues (rest of the body). Inhalation of aerosols is a complex process, and PBPK models with regional lung compartments are more descriptive. Here, we use a multicompartment respiratory tract model. Based on anatomical location and function, the respiratory tract is divided into three regions: upper respiratory tract (URT), trachea-bronchial region (TB) and alveolar region. There is also a pulmonary compartment which represents the gas exchange region of the lung. To include processes that transport the absorbed nicotine across the anterior respiratory tract compartments, these regions are further divided into a two-layered substructure: an epithelial cell layer with mucus, and a submucosal tissue layer. The submucosal tissue layer has blood perfusion and clears the absorbed nicotine from the airway compartments. Consistent with common practice, the tissue compartments are well-mixed reservoirs. Exposure is characterized in each lung (inhalation) compartment based on calculated deposition rates. Elimination occurred from both the liver and the blood compartment and was represented by hepatic and renal clearance, respectively. The model structure is shown schematically in Figure 2.

The physiologically-based biokinetic (PBBK) model was developed in Berkeley Madonna software (version 10.1.3; University of California, Berkeley, CA; www.berkeleymadonna.com). Model equations are in Supplemental 1.

Parameters used in the current model are summarized in Table 2. Tissue blood flows are from our ScitoVation database for an adult male (internal database not shown), except for the blood flow to the upper respiratory tract (URT) that was set according to Campbell et al. (2015) and the blood flow to the alveolar region that was set according to Butler (1992). The ScitoVation database includes age-specific physiological parameters for body weight (BW), cardiac output, tissue weights (volumes), and tissue blood flows, and are adapted from published life-stage models (Wu et al., 2015; Song et al., 2016; Ruark et al., 2017).

Volumes were set to values from the ScitoVation database except for the lung volume. Tissue volumes for the three respiratory-tract compartments were estimated by multiplying the appropriate surface area with the tissue thickness. The surface area of the three lung regions was estimated using a standard lung model (MPPD) that quantifies airway length and diameter on a generation-by-generation basis. Epithelial thickness was obtained from Sarangapani et al. (2002b). The submucosal thickness was assumed to be approximately twice the epithelium thickness, based on histological sectioning (Matthew Bogdanffy, personal communication).

Chemical-specific parameters such as hepatic intrinsic clearance and renal clearance were obtained from previously published PBPK model from Robinson et al. (1992) and Yamazaki et al. (2010). Robinson et al. (1992) estimated a hepatic clearance of 1.09 L/min, using a PBPK model calibrated with in vivo human data.

Tissue: blood partition coefficients (PC) are defined as the ratio of the concentration of a test chemical in two media (i.e., tissue and plasma), once equilibrium is reached. PCs are important determinants of the disposition of chemicals in different tissues. Liver PC was from Teeguarden et al. (2013). The other PCs (the rest of the body and lung compartments) were fitted to the in vivo human data. Plasma protein binding of approximately 5% has been reported by Robinson et al. (1992).

Nicotine permeability was taken from literature. There was some variability in reported values: 1.0E-4 cm/s from Gowadia and Dunn-Rankin (2010), 1.28E-4 cm/s from Waddell and Marlowe (1976), 2.5E-5 cm/s and 1.14E-5 cm/s from a QSAR model used by Symcyp. We opted to use the QSAR value as it was giving a better fit with the in vivo PK data. The nicotine permeability was multiplied by the tissue thickness in the experiment to derive a diffusion coefficient in cm²/s.

The fraction of particles deposited in each of the lung compartments was estimated using the MPPD model. To validate the model, we simulated individuals with deep breathing at rest with a short breath hold. Most of the studies used to validate the model did not provide direct information about the scenario of exposure. In most of the studies, volunteers were at rest, i.e., reading quietly, working, or engaging with social media. Based on smoking behavior found in different publications (Pichelstorfer et al., 2016; McEwan et al., 2019), a deep breath at resting and a short breath hold seemed to be the more appropriate scenario to reflect the human in vivo data used to validate the model.

The fraction of vapor inhaled for cigarettes was fitted to the data. Here the particle size diameter, the count median diameter (CMD) was set to 163 nm with a geometric standard deviation (GSD) of 1.44 for cigarettes based on Fuoco et al. (2014) and Sosnowski and Kramek-Romanowska (2016). These values were used for the calibration and validation of the model.

The performance of the model was first evaluated using the in vivo pharmacokinetic (PK) data (plasma concentrations between 0–4 h post exposure) from McEwan et al. (2019) where 48 healthy subjects smoked a single assigned test cigarette and had blood drawn on the morning of each visit. In the afternoon, the smoking behavior assessment was carried out with a single use of the same test product.

The model was also validated using 2 other in vivo PK datasets from Picavet et al. (2016), and Digard et al. (2013).

Picavet et al. (2016) assessed the PK of nicotine after a single use of cigarette in 28 healthy smokers. The cigarettes assessed in this study were non-menthol, manufactured, commercially available cigarettes, with a maximum ISO yield of 1 mg nicotine per cigarette. The pharmacokinetics of nicotine were measured on the days of single use. The first blood sample was collected within 15 min before a single use of the allocated product in the morning, and then at 2, 4, 6, 8, 10, 15, 30, 45, and 60 min, and at 3, 4, 6, 9, 12 and 24 h.

Digard et al. (2013) conducted a study with 20 healthy cigarette users. A cigarette was smoked, and blood samples taken at intervals over 120 min. The subjects were instructed to smoke the cigarette naturally according to their usual smoking behavior for 5 min or until the cigarette had been smoked to 30 mm from the mouth end (if this occurred in less than 5 min), at which point the cigarette was extinguished.

All the puffing scenario and dose (1 mg nicotine) used in each of these studies were the same with a puff duration of 2.3 s, puff interval of 30 s, puff volume of 69.5 mL and number of puffs per cigarettes of 14. Puff duration, number of puffs and puff volume for cigarette were based on McEwan et al. (2019). The puff interval was assumed to be 30 s (Bernstein, 2004).

The dose of nicotine in one cigarette was assumed to be 1 mg (the amount varies between brands but is usually between 0.7 and 1 mg) (EU, 2014).

To evaluate the relative impact of each of the model parameters on nicotine plasma maximal concentration (C_max) and area under the curve (AUC), a sensitivity analysis was performed. The sensitivity coefficient (SC) was calculated according to the equation below (Teeguarden et al., 2005). SC = Fractional change in model output / Fractional change in parameter

Each parameter was individually increased by 1% of its original value, with the other parameters held constant. The larger the absolute value of the sensitivity coefficient, the more important the parameter. A normalized sensitivity coefficient of 1 represents a 1:1 relationship between the change in the parameter and the internal dose metric of choice. A negative SC indicates the given parameter influences the dose metric in an inverse direction. The SCs are grouped into one of three categories: namely, high (absolute SC values greater than or equal to 0.5), medium (absolute SC values greater than or equal to 0.2 but less than 0.5), or low (absolute SC values greater than or equal to 0.1 but less than 0.2), according to the IPCS guideline (WHO, 2010).

Figure 8 shows the results of the performance of the model when evaluated using the in vivo PK data for nicotine from McEwan et al. (2019). A qualitative evaluation of the agreement between experimental plasma concentration and simulations was conducted through visual inspection of the time-course, where good agreement is generally defined as simulations falling within a factor of two of the data (EPA, 2006). In addition, IPCS (2010) guidance on “Principles of Characterizing and Applying PBPK Models in Risk Assessment” states that “In PBPK modeling, predictions that are, on average, within a factor of 2 of the experimental data have frequently been considered adequate”.

Monte Carlo (MC) analysis was also conducted to investigate the population variability. Only the sensitive parameters from the sensitivity analysis were varied. Partition coefficients, body weight, breathing rates, and metabolic constants were simulated as log normally distributed; and cardiac output, blood flows, tissue volumes and parameters related to puffing scenario (puff volume, puff duration, puff interval) were simulated as normally distributed. The coefficients of variation (CV) for partition coefficients were 30%, a CV of 22% and 16% were used for the body weight and for the blood flow to the liver, respectively (Price et al., 2003), while a CV of 30% was assumed for the remaining model parameters. The distributions were truncated at 2 Standard deviations (Clewell and Clewell, 2008) to ensure physiological plausibility. Parameters distribution can be found in Supplementary Table S1. Monte-Carlo simulations were performed with 100 iterations to perform population-level simulations. Mean or median plasma concentration as well as the 5th and 95th percentiles are graphically represented in Figures 8–10. For IVIVE, Monte-Carlo simulations were performed with 1,000 iterations to perform population-level simulations.

The performance of the model was also evaluated using in vivo nicotine PK studies with cigarettes (Digard et al., 2013; Picavet et al., 2016). Simulations results can be found in Figures 9, 10. The PBPK model predictions were consistent with the nicotine concentrations in plasma in adult humans.

The sensitivity of model parameters was calculated with the human PBPK model for the plasma maximum concentration (C_max) and area under the curve (AUC) after low and high inhalation doses of nicotine (1.2 and 12 mg). Results from the normalized sensitivity analysis are in Supplementary Figure S1. The most sensitive parameters are the ones related to inhaled dose: puff duration, puff interval, breathing rate, tidal volume, puff volume and fraction of vapor. The analysis showed that physiological parameters, including body weight, cardiac output and blood flow to the liver are medium to highly sensitive. The parameter for particle deposition in the pulmonary region is also a medium-sensitivity parameter. Note that hepatic metabolism parameters showed no sensitivity for plasma C_max but medium sensitivity for the plasma AUC. This does not mean that metabolism is not influential in plasma C_max, but indicates that metabolism is so efficient, i.e., CL_int_in vivo very much exceeds liver plasma flow and a 1% increase in the parameter value would not make a difference, as it is largely limited by liver plasma flow, which is one of the most sensitive parameters.

The uncertainty of a model reflects the level of confidence in model predictions. The structure of the model is based on previously published models (Plowchalk et al., 1992; Robinson et al., 1992) where we included a more complex description of the lung to be able to link a dosimetry model (MPPD). The vapor fraction for cigarette was fitted to the data. For vapor absorption, we assumed that vapor would be completely absorbed in the deep lung. A more complex description of vapor absorption along the respiratory tract may provide simulations more similar to the observed data.