Three nicotine delivery products were used in this study: 1) the 1R6F Kentucky reference cigarette (1R6F) (University of Kentucky, Kentucky, USA); 2) an ENDS product, the myblu pod-based system (Imperial Brands PLC); and 3) a HTP, the Pulze and iD stick heated tobacco system (Imperial Brands PLC). The myblu ENDS was of a tobacco flavour e-liquid variant, with a 1.6% nicotine concentration, and the system was as described by Rudd et al. (2020); the iD stick contained a reconstituted tobacco portion similar to the ‘Regular’ product as previously described and used in Chapman et al. (2022). The iD stick had a target specification of 4.6 mg nicotine/stick.

Aqueous phosphate buffered saline (PBS) (Sigma Aldrich, Germany) extracts of the aerosols generated from the respective products were created for application to the in vitro experimental system using the Vitrocell VC 10S-Type smoking robot. These extracts (termed bubbled PBS (bPBS)) were generated by bubbling the respective experimental products’ aerosols through three in-line impingers, each containing 10 mL PBS, which was then combined into 30 mL stock solutions. 1R6F smoke was generated according to the International Organization for Standardization (ISO, 20778:, 2018) formerly known as Health Canada Intense smoking regime 55 mL puff volume, 2 s puff duration, 30 s puff interval, bell-shaped puff profile, with ventilation blocking; the HTP bPBS was generated using a modified version of this regime, i.e., no ventilation blocking was applied. The ENDS aerosol was generated according to ISO 20768: 2018 (formerly known as the Cooperation Centre for Scientific Research Relative to Tobacco (CORESTA) Recommended Method N°81) (ISO, 20778:, 2018): 55 mL puff volume, 3 s puff duration, 30 s puff interval, square-wave puff profile. The pooled solution concentration for 1R6F = 1.8puffs/mL (54 puffs/30 mL) and the ENDS and HTP products 4.8puffs/mL (144 puffs/30 mL).

This extract was analysed to confirm smoke/aerosol trapping, using nicotine and eight carbonyls (formaldehyde, acetaldehyde, acetone, acrolein, propionaldehyde, crotonaldehyde, 2-butanone, n-butyraldehyde) as markers. These markers were selected based on those detailed by Buratto et al. (2018), and additionally, are present on regulatory HPHC lists (FDA, 2012); detailed methodology of their quantification in the PBS is described by Chapman et al. (2023).

Two cell lines were used for the co-culture model, Human Coronary Artery Endothelial Cells (HCAECs) (PromoCell, C12221) and the human leukaemia monocytic cell line, THP-1, (ATCC, Tib-202). Prior to seeding, HCAECs were thawed and cultured in T-75 flasks (Corning, 431464U) using MV2 cell culture medium (Promocell, C22022) containing supplement mix (Promocell, C39226) and 1% Penicillin/Streptomycin (Sigma Aldrich, P4333), further referred to as MV2 medium. HCAECs were harvested at passage 5 by washing three times with PBS (Thermo Fisher Scientific, 70,013,016) and addition of 0.025% trypsin/ethylenediaminetetraacetic acid (EDTA, Lonza, CC-5012).

THP-1 cells were thawed and cultured upright in T-25 flasks (Corning, 353,009) using Roswell Park Memorial Institute (RPMI) 1,640 medium (Sigma Aldrich, R0883) supplemented with 10% fetal bovine serum (FBS), 1% Penicillin/Streptomycin and 1% GlutaMAX (Gibco, 35,050). Cells were maintained around 2 × 10⁵ - 5 × 10⁵cells/mL during culture.

The OrganoPlate 2-lane 96 (MIMETAS, 9605400B, Oegstgeest, Netherlands), consisting of 96 microfluidic chips with, respectively, 9 mm and 12.2 mm long gel and perfusion channels (400 µm × 220 µm (w x h)), was used in this study. A 4 mg/mL extracellular matrix (ECM) gel, consisting of 100 mM N-(2-Hydroxyethyl)piperazine-N′-(2-ethanesulfonic acid) (HEPES) buffer, (Gibco, 15,630-056), 3.7 mg/mL NaHCO₃ (Sigma Aldrich, S5761) and 5 mg/mL collagen I (Cultrex, 3447-020-01), was prepared and 2 µL was dispensed in the gel inlets of the microfluidic chips. Plates were incubated at 37°C to allow polymerization of the ECM gel after which 50 µL of Hank’s Balanced Salt Solution (HBSS) (Thermo Fisher Scientific, 14175095) was added to the gel inlets to prevent dehydration. In addition, 50 µL of HBSS was added to the observation windows to prevent evaporation and increase optical clarity. Plates were placed in the incubator (37°C, 5% CO₂) overnight until cell seeding.

Harvested HCAEC cells were counted using Trypan Blue (1:1) and the LUNA-II automated cell counter (Logos Biosciences, L40002), spun down (200 x g, 5 min) and resuspended at a density of 1 × 10⁶ cells/mL using MV2 medium. 2µL of cell suspension was dispensed in perfusion inlets of microfluidic chips, followed by 50 µL of MV2 medium. Plates were incubated at 37°C, 5% CO₂ on their side for 2 h to allow cell adherence to the ECM gel. Next, 50 µL MV2 medium was added to perfusion outlets and plates were placed on a rocking interval platform (MIMETAS, MI-OFPR-L) set at 7° with 8-min intervals to induce gravity-driven flow. Figure 1 describes the plate set-up.

Prior to exposure within the OrganoPlate system, cell culture medium was pre-conditioned with the respective test articles at the concentrations detailed in Table 1 using THP-1 cells to model the presence of systemic cellular responses, for example, the generation of cytokines, within the co-culture. The THP-1 cells were thawed and cultured for 48 h prior to being spun down and resuspended at a concentration of 2 × 10⁶ cells/mL for a 2 h exposure. Harvested conditioned medium was then frozen prior to use in the main study.

To understand the effect of pre-conditioning on the outcomes measured, a comparison was also made to the responses in the OrganoPlate where the top concentration of the respective test articles was added directly with no pre-conditioning step.

To initially assess maximal bPBS concentrations that could be applied to the cell cultures (to match pre-conditioning and co-culture exposure concentrations), a range of levels of PBS, including standard concentrations (i.e., 10% or below) and higher concentrations were added to 250 µL of 2 × 10⁶ cells/mL suspension in 48-well plates for 2 h (0, 2.5, 5, 10, 20, 40% PBS). Cell viability was determined using Trypan Blue staining (1:1 with cell suspension) and the LUNA automated cell counter.

Based on this assessment, it was decided that 20% would be the maximal applied concentration for the pre-conditioning step due to greater variability in outcomes following the 40% exposure (Supplementary Figure S1).

An initial dose range finding experiment was carried out to select appropriate, non-cytotoxic concentrations for the study presented here. Based on historical data, 1R6F was expected to be the most potent test article, and this was applied to the model in the first instance to assess cell viability, barrier integrity, oxidative stress, monocyte adhesion and ICAM-1 expression.

Automated phase contrast imaging was performed on regular basis using the ImageXpress XLS Micro HCI system (Molecular Devices, San Jose, CA, USA).

Cell culture medium was replaced by fresh MV2 medium containing 0.5 mg/mL tetramethylrhodamine isothiocyanate (TRITC)-dextran (155kDa, Sigma Aldrich, T1287). First, 10 µL fresh MV2 medium was added to the gel inlets of the chips. Next, MV2 medium containing TRITC-dextran was perfused through the lumen of the endothelial vessel in the top channel (40 µL in the inlet and 30 µL in the outlet). Leakage of the dye from the lumen of the endothelial vessel into the adjacent extracellular matrix compartment was recorded using time-lapse imaging on an ImageXpress XLS Micro high content imaging (HCI) system (7 timepoints, 2 min intervals). Ratio between fluorescent signal in the endothelial vessel compartment (donor) and the ECM compartment (receiver) was calculated using FIJI (version 2 build 1.52d) and used to calculate apparent permeability (P_app) values.

Cell viability was measured using the Cell Counting Kit-8 (Sigma Aldrich, 96992). The water-soluble tetrazolium salt (WST-8) reagent was diluted 1:11 in MV2 medium as recommended by the supplier. Cell culture medium was replaced with diluted WST-8 reagent. Cultures were incubated at 37°C, 5% CO₂ on a rocking interval platform (8 min, 7° angle) for 30 min. After incubation, the plate was placed static for 2 min before absorbance was measured at 450 nm using a multi-well plate reader (TECAN Spark^®, TECAN Life Sciences). Acquired absorbance values were corrected for chip dimensions and corresponding volumes.

Assessment of glutathione (GSH) levels within the cell system was carried out using monochlorobimane (mBCI, Sigma Aldrich, 69899-5 MG), which binds to GSH to produce a fluorescent conjugate with an absorption/emission of 394/490 nm. Endothelial vessels were incubated with fresh MV2 medium containing 125 μg/mL mBCl +1:1000 DRAQ5 (Abcam, ab108410) at 37°C, 5% CO₂ on a rocking interval platform (8 min, 7° angle) for 30 min after which the chips were washed with HBSS. Fresh HBSS was added to the chips and cultures were imaged using an ImageXpress Micro Confocal HCI System (Molecular Devices). Z-series were captured to generate sum projections which were used to quantify the cumulative intensity of the fluorescent signal after background subtraction using FIJI. Number of endothelial nuclei was quantified using built-in features of FIJI; first, a rolling ball background correction was applied to improve the signal-to-noise ratio (Sternberg, 1983) after which an automated thresholding routine was performed followed by particle detection to quantify individual nuclei. Subsequently, the mean fluorescent intensity per nucleus was calculated.

Based on the expected window to observe effects on GSH depletion (i.e., presence of oxidative stress), this analysis was carried out 4 h following first exposure of the OrganoPlate model to the test articles. Sigma Aldrich, SML1083 was applied as a positive control at a concentration of 25 μM for the GSH assay.

To assess monocyte adhesion, fresh monocytes (i.e., not those which had been previously exposed in the generation of the pre-conditioned media) were applied to the OrganoPlate. THP-1 monocytes were suspended at a density of 1 × 10⁵cells/mL in complete RPMI medium containing 0.5 μg/mL Calcein-AM (Life Technologies C3099, Carlsbad, CA, USA). After labelling for 15 min at 37°C, the monocytes were spun down (200 x g, 5 min) and resuspended in fresh MV2 medium at a density of 1 × 10⁵ cells/mL. Meanwhile, nuclei of the endothelial vessels were stained by incubating with 1:2000 Hoechst 33342 (Thermo Fischer Scientific, H3570) for 20 min at 37°C. The endothelial vessels were washed with fresh MV2 medium after which labelled THP-1 monocytes were added apically to the endothelial vessels. Cultures were incubated at 37°C and 5% CO₂ for 15 min on a rocking interval platform (8 min, 7° angle) to allow for monocyte adhesion and cultures were washed twice with HBSS. Cultures were imaged using an ImageXpress Micro Confocal High-Content Imaging System. The number of endothelial nuclei were quantified in FIJI as described before. The number of adhering monocytes was quantified using a nearly identical method as described before, except a minimum size of 10 µm was selected.

Cultures were fixed after the experiment using 3.7% paraformaldehyde (Sigma Aldrich, P6148) in HBSS (Sigma Aldrich, 55037C). After washing twice with HBSS, cultures were permeabilized for 10 min using 0.3% Triton X-100 (Sigma Aldrich, T8787) and incubated with blocking solution containing 2% FBS, 2% bovine serum albumin (BSA; Sigma Aldrich, A2153) and 0.1% Tween-20 (Sigma Aldrich, P9416) for 45 min. Next, cultures were washed with PBS containing 4% FBS and incubated overnight with anti-ICAM-1 primary antibody (1:100; R&D systems, BBA3). Cultures were then washed and incubated with donkey anti-mouse 647 secondary antibody (1:250; Life Technologies, A31571) for 30 min. Hoechst 33342 (1:2000) was used as a nuclear counterstain. All steps were performed at room temperature. For the representative image in Figure 1E, CD31 staining was carried out as above, however, the primary antibody, CD31 was added (1:20, Dako, M0823) followed by secondary antibody, goat anti-mouse 488 (1:250, Life Technologies, AF32723).

Images were acquired with an ImageXpress Micro Confocal High-Content Imaging System. Z-series were acquired, and sum projections were generated for quantification of the fluorescent signal. The mean fluorescent intensity was calculated in FIJI and corrected for the number of endothelial nuclei.

Culture medium was sampled after exposure and frozen at −80°C. Cytokine contents of sampled medium was assessed using a ProCartaPlex Inflammation 20-plex (Thermo Fisher Scientific, EPX200-12185-901). The following analytes were measured: GM-CSF, IFN-α, IFN-ɣ, IL-1α, IL-1β, IL-4, IL-6, IL-8 (CXCL8), IL-10, IL-12p70, IL-13, IL-17A, TNF-α, IP-10 (CXCL10), MCP-1 (CCL2), MIP-1α (CCL3), MIP-1β (CCL4), ICAM-1, CD62E (E-selectin) and CD62P (P-selectin). Medium samples were processed according to supplier’s protocol and cytokine contents were measured using MAGPIX^® (Luminex Corporation, Austin, TX).

Nine chips/test conditions were used across 2 biological replicates (n = 9, N = 2). Data was normalised to respective controls for visualisation, and error bars plotted were standard deviation (SD) about the mean. Statistical significance was determined using a one-way analysis of variance (ANOVA) using GraphPad Prism version 9. Statistical significance is marked as follows: *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001.

A summary of the experimental timeline can be found in Table 2.

To confirm aerosol trapping, nicotine and eight select carbonyls (selected based on a list outlined by Buratto et al., 2018) were quantified within the respective bPBS stock solutions generated for the test articles. Nicotine was also used as a dosimetry marker. Out of the eight carbonyls, only formaldehyde was quantifiable within the ENDS bPBS (at a substantially lower level than measured for 1R6F), with the other carbonyls reported as below the limit of quantification (LOQ). For the HTP, all eight carbonyls were present in the bPBS, however levels were below/substantially below those seen for the 1R6F reference cigarette. Trapped nicotine was highest in the ENDS bPBS, followed by the HTP bPBS, then 1R6F bPBS. Based on the respective test article concentration ranges applied in the subsequent cell culture analyses, the ENDS and HTP bPBS therefore delivered greater nicotine levels than those applied for the 1R6F bPBS (Table 3). When the top concentrations added to the cell cultures were compared, levels of carbonyls were approximately equivalent between 5% 1R6F bPBS and 20% HTP bPBS. However, the 1R6F stock required fewer puffs per mL PBS than the NGP to achieve these final concentrations and upon comparison on a per puff basis, levels of carbonyls in the 1R6F smoke were 5–19 and 70 times higher than for the HTP and the formaldehyde in the ENDS sample respectively.

To determine appropriate concentrations to use in the main in vitro study, 1R6F was assessed as, based on previous studies, it was the most potent test article (Simms et al., 2022) (Figure 2). Additionally, two PBS control conditions were applied (20% and 40%) to elucidate the maximum bPBS stock concentration that could be applied with regards to the two less potent test articles (HTP and ENDS). At the 1R6F concentrations tested, barrier integrity (indicated by apparent permeability) declined significantly compared to PBS control at 1R6F concentrations of 10% and 20%; this effect was also observed at these concentrations upon assessment of cell viability. Dose-dependent increases in oxidative stress (indicated by depletions in GSH levels) were also observed for 1R6F. Up to 10% 1R6F bPBS, there appeared to be a dose-dependent increase in monocyte adhesion, however, due to high variability, this result was not statistically significant; the 20% 1R6F bPBS concentration induced a decrease, however, which may have been linked to reductions in cell viability and barrier integrity. Upon assessment of ICAM-1 expression, there were no significant increases upon application of the test articles, however positive controls TNF-α and ethacrynic acid induced significant increases, confirming the validity of the assay.

To optimise the in vitro system for the application of nicotine product aerosol aqueous extracts, 1R6F reference cigarette smoke bPBS was initially applied to the system. Based on previous studies using endothelial-monocytic co-cultures, pre-conditioning of medium using monocytes exposed to the test articles was applied as the exposure condition within the HCAEC vessel to model indirect/secondary effects on the endothelium (Poussin et al., 2014). Based on the endothelial barrier integrity and cytotoxicity outcomes, it was concluded that 1R6F bPBS should be applied to the pre-conditioning medium up to a concentration of 5% to ensure effects in the subsequent main experiment were not related to excessive cytotoxicity in the in vitro model. The application of higher concentrations of PBS (i.e., 20%, 40%) was also carried out in the pre-study to assess whether addition of historically less potent NGP bPBS test articles (Simms et al., 2022; Chapman et al., 2023) could potentially be applied at higher concentrations without confounding solvent effects. Application of 40% PBS to the cell system did not appear to significantly affect cell viability, however some decrease in barrier integrity was observed compared to 20% PBS. In addition to this, PBS was also applied to THP-1 monocyte cultures for 2 h up to a concentration of 40% to assess the maximal concentration for the medium pre-conditioning step (Supplementary Figure S1). At 40%, although not significantly altered compared to untreated control, there was an increase in variability between replicates. Therefore, the maximal concentration of PBS to be applied for the additional test articles in the main study (HTP and ENDS aerosol bPBS) was selected at 20%.

To increase the model’s human/in vivo relevance, and better represent potential inflammatory responses/atherogenic processes, pre-conditioned medium was generated with THP-1 monocytes for the exposure within the HCAEC vessels (Poussin et al., 2020). To confirm whether the pre-conditioned media played a role in the responses observed, direct bPBS exposure to the HCAEC vessel treatment conditions were also included at the highest respective bPBS concentrations applied for each test article. The direct addition of the test articles only appeared to further decrease GSH levels within the OrganoPlate system 4 h following the start of exposure for the 1R6F and HTP highest concentrations, however this was not the case for the ENDS treatments, which were not different to one another. This could potentially be attributed to the presence of modulators within the pre-conditioned medium regulating some of the endothelial response to the 1R6F and HTP bPBS extracts present, or initial maximal (nominal) exposure to reactive species within the 1R6F/HTP bPBS. This effect, however, was not observed for the ENDS and HTP bPBS in the monocyte adhesion endpoint, although pre-conditioning demonstrated a clear influence on THP-1 adhesion in the 1R6F bPBS-exposed model. Upon comparison of the effects of exposures including pre-conditioned medium to test-article-only exposures on the inflammatory panel readouts, there was a clear dose-dependent influence of pre-conditioned medium within the system, with mainly increases observed for 1R6F and the HTP, and only slight increases or decreases for the ENDS. In contrast, across all test articles, levels of the inflammatory modulators largely decreased for the compound only controls, again with a trend for the greatest changes with 1R6F, then HTP, then ENDS. This may explain some of the differences in responses observed in the GSH/monocyte adhesion endpoints in the presence and absence of pre-conditioning. These results highlight the importance of capturing the interaction of different cell types within in vitro models.

GSH depletion was used as a proxy assessment for the presence of oxidative stress following exposure of the co-culture to test article pre-conditioned media. A reduction in GSH signal correlates with a response to the presence of oxidative species, which may be a direct result of constituents of the test articles or secondary inflammatory responses (Poussin et al., 2016; Yang et al., 2017). As anticipated, 1R6F demonstrated the most potency out of the three test articles, inducing dose-dependent reductions in GSH, and at concentrations four times lower than those inducing a similar response for the HTP test article. In contrast, the ENDS sample did not induce any significant reductions in GSH at the concentrations tested. Oxidative stress can drive the expression of adhesion molecules in the vascular endothelium in the early stages of atherosclerosis, leading to the recruitment and adhesion of monocytes (Björkegren and Lusis, 2022; Ohara and Ito, 2023); the oxidative stress outcomes under pre-conditioned medium treatments indeed appeared to be reflected by the monocyte adhesion assay outcomes for 1R6F and the HTP, however, not for the ENDS, suggesting the role of potential other factors here.

Monocyte adhesion is a key step in the development of atherosclerosis: monocytes attach to the vascular endothelium and migrate into the vessel intima where they can transform into macrophages (Fearon et al., 2013). These macrophages in turn form foamy macrophages following uptake of oxidised LDL, and accumulation of these fatty cells contributes to the formation of plaques (Javadifar et al., 2021). In order to model the early phase of this (monocyte adhesion), FITC labelled THP-1 monocytes were perfused through the vessels for 30 min at the end of the main (24 h) exposure. All three test articles induced apparently dose-dependent increases in monocyte adhesion following exposure, however, in light of replicate variability, this was only significant at the highest dose tested for each respective test article. For the highest concentrations tested, compared to treatments without pre-conditioned medium, there were no significant differences in monocyte adhesion for the ENDS and HTP bPBS, suggesting that exposure of the endothelium to the test article constituents may be the main driver for monocyte adhesion. However, when making this comparison for 1R6F, the presence of pre-conditioned medium was required to induce significant increases in monocyte adhesion, which was not observed in the absence of pre-conditioning. This may be due to the different compositions of 1R6F bPBS and the NGP bPBS samples, which may elicit different response mechanisms, for example, higher levels of nicotine within the NGP samples; this is reflected in the inflammatory marker outcomes. In comparison to similar studies, the magnitude of change in monocyte adhesion differed from that observed by others (Poussin et al., 2020; Ohashi et al., 2023), however, this could be due to methodological differences in detecting and quantifying the cell signal, or differences in chemical fractions applied. For example, Ohashi et al. (2023) utilised total particulate matter (TPM)/aerosol collected mass (ACM), fractions of cigarette smoke/HTP aerosol respectively, whereas this study applied aqueous smoke/aerosol extracts. Furthermore, whilst both are human primary monocyte models, there might be differences between the MM6 cell line used by Poussin et al. (2020), and the THP-1 cell line used in this study in their responses to the test articles.

ICAM-1 is known to be associated with adhesion of monocytes to the vascular endothelium during the atherogenic process (Lawson and Wolf, 2009; Ohara and Ito, 2023). Therefore, to understand potential drivers for monocyte adhesion within the co-culture following exposure to the respective test articles, measurement of endothelial ICAM-1 expression was carried out. For the HTP bPBS, endothelial ICAM-1 expression did not appear to significantly change upon exposure of the test system to the test articles/pre-conditioned medium, indicating that, where monocyte adhesion was observed, ICAM-1 was not a major driver for these effects. Monocyte adhesion is also linked to other adhesion molecules, including VCAM-1 (Edwards and Thomas, 2017), which was not measured in this study, but could warrant assessment in future investigation. The present findings are consistent with those of others’, where ICAM-1 expression was found not to always directly correlate with monocyte adhesion (Poussin et al., 2018); the study additionally found that VCAM-1 expression decreased upon exposure of cardiovascular cells to similar test articles, and proposed other adhesion molecules may be involved (Poussin et al., 2018). Further to this, many other inflammatory mediators, including those measured within the inflammatory panel applied within this study, are linked to monocyte adhesion (Poussin et al., 2014; Kong et al., 2022), and therefore there could be some involvement of these in the responses observed in this study, as discussed in the next section. Interestingly, for the ENDS bPBS, endothelial ICAM-1 expression was significantly increased at the two lower dose levels tested (5% and 10% bPBS), but not for the highest treatment concentration (Figure 5). This could, however, be due to high variability within the test replicates; comparison to the ICAM-1 readout for the ENDS bPBS in the inflammatory marker panel (Figure 6) indicated low ICAM-1 expression at the three concentration levels applied (pre-conditioning condition). 1R6F did induce significant increases in endothelial ICAM-1 expression at the highest concentration tested (5%), however, again, high variability was observed for the outcomes. Therefore, ICAM-1 may be involved in the monocyte adhesion observed for 1R6F, which is supported by the outcomes for the inflammatory panel.

As mentioned above, the three test articles induced distinct dose-dependent inflammatory profiles compared to one another, with 1R6F eliciting the greatest increases in markers, followed by lesser responses to the HTP, and even lesser changes compared to vehicle control for the ENDS test article. There was additionally a clear distinction between readouts for the test article-only treatment compared to the pre-conditioning treatments, highlighting the cross-talk of different cell types in cellular responses to exogenous agents, and the role of secondary responses to exposure. A number of the markers within the panel play a role in processes involved in the early stages of atherosclerosis, including drivers of immune cell recruitment, adhesion and migration and modulators of other stress responses (Libby et al., 2002; Poussin et al., 2014; Fatkhullina et al., 2016). The dose-dependent increases induced by the 1R6F bPBS following the 24 h pre-conditioning were largely consistent across the markers measured, however, IL-4 and IL-10 levels were below detection limits for all three treatment levels. Due to the interplay of different mediators, this downregulation may have been driven by the increased levels and feedback of other cytokines present, or potentially linked to the absence of stimulation of IL-10 by IL-4 (Mitchell et al., 2017), which may represent a lack of presence of driver for this kind of cytokine response. Many of the markers within the panel have known roles in cellular recruitment (E-selectin, P-selectin, ICAM-1, etc.) and additionally drivers of the process such as TNF-α (Warboys et al., 2011; Fatkhullina et al., 2016); whilst there were the clear dose-dependent trends for 1R6F in the pre-conditioned exposure conditions, this was not always the case for the NGP exposures. However, dose-dependent increases in mediators such as TNF-α and ICAM-1 mirror the outcomes in the oxidative stress and monocyte adhesion assays for the HTP. Interestingly, where levels were below detection limits for IL-4 for 1R6F, levels were elevated for all three HTP treatment concentrations, indicating that different inflammatory mechanisms may be underway in response to the discrete product categories. The inflammatory panel revealed greater dose-dependent trends with ICAM-1 than seen in the endothelial assessment, indicating that variability may have driven a lack of significant responses or that there is no correlation between the membrane-bound and the soluble forms of ICAM-1 (Theresa et al., 2014). However, the ICAM-1 outcomes for ENDS was reflected in the inflammatory panel readout, where levels were increased at the lower two test concentrations only. Overall, the panel indicated a complex interplay of inflammatory mediators present in the cell culture medium, which was both dose- and test article-dependent. Additionally, the panel only assessed 20 markers, and others have demonstrated further mediators of atherogenic endpoints in response to nicotine delivery products (Poussin et al., 2018; Poussin et al., 2020), therefore, further characterisation, such as transcriptomics analysis, may be beneficial.

Upon comparison of the three test products applied within this study, the 1R6F reference cigarette, an ENDS and an HTP, there were overall clear distinctions between responses. Firstly, the highest bPBS concentration applied to the in vitro system was 5% for 1R6F due to losses in cell viability and barrier integrity at the concentrations tested above this (10% and 20%) in the initial dose-finding assessments (Figure 2A,B). In contrast, the two NGP products were able to be applied to the test system up to a concentration of 20% and generally this 4x higher bPBS concentration was required to elicit significant responses to the NGP test articles. These outcomes were associated with the relative numbers and levels of carbonyls measured in the respective bPBS samples (Table 3); the relative potencies of the test articles in the in vitro system was 1R6F >> HTP > ENDS. For the ENDS bPBS, levels were below LOQ for seven carbonyls and it contained 26-fold less formaldehyde than the 1R6F stock. Overall, there were substantial reductions in carbonyls within the HTP bPBS compared to 1R6F. Where responses were observed for the NGP, potential drivers may be product/extract dependent. This corresponds to the temperatures used to generate the respective smoke/aerosols: cigarettes burn tobacco whereas the HTP heats tobacco at 345^oC, and a combustion process and generation of the associated chemicals therefore does not take place (Chapman et al., 2023). Furthermore, lower temperatures still are used to heat e-liquids in ENDS (Jaegers et al., 2021; Li et al., 2021), although differences in e-liquid and tobacco compositions will also have an influence on aerosol characteristics. When calculated on a relative nominal carbonyl level applied to cell culture medium within the bPBS, levels were comparable between the 5% 1R6F bPBS and 20% HTP bPBS. This provides a potential explanation, coupled with the higher nicotine concentrations present in the HTP bPBS, for generally similar magnitudes of responses in some of the endpoints measured. Nicotine is known to have toxic in vitro effects at high concentrations (Rudd et al., 2020) and this may also have been a driver for effects in the case of the ENDS bPBS, which demonstrated the highest relative levels of nicotine. However, 1R6F still demonstrated greater potency, suggesting other toxicants may be present within this sample, either direct-acting, or those causing secondary effects; further analysis of all bPBS samples would be the next step to understand the drivers for the relative responses observed to the test articles in more detail. An additional limitation in defining the relative exposures is the difficulty in accurately capturing the levels of each component of the extract due to its potential degradation within the pre-conditioning medium/loss upon removal of the THP-1 cells. It is also of note that nicotine concentrations applied within the in vitro system ranged from 2.3 (1R6F) - 53.64 (ENDS) μg/mL (Table 1), concentrations which greatly exceed the 5–50 ng/mL range observed in the blood plasma of smokers, HTP and EVP users (Benowitz et al., 2009; Jacobson et al., 2020; McDermott et al., 2023) following single product use; it can also be inferred that this kind of difference would be the case for other compounds present within the smoke/aerosols of the products. However, it must be noted that use of multiple nicotine products during the day may increase nicotine exposure and is product and user topography dependent, for example, as seen with ENDS (Jacobson et al., 2020). A recent in vitro to in vivo extrapolation study by Moreau et al. (2024) predicted that due to rapid nicotine elimination, following the use of 10 cigarettes or HTP sticks, a steady state of nicotine is still only approximately double that observed following the use of one product. Therefore, the responses observed in this study are still to supraphysiological levels of nicotine exposure, the study has demonstrated that the co-culture system is responsive to nicotine delivery product extracts, and may act as a useful tool in screening for effects at physiological levels in future. These levels are also regularly applied for in vitro assessments of such products (Czekala et al., 2021; Chapman et al., 2023).

This study has provided valuable insights into the effects of three discrete nicotine delivery products in an in vitro 3D co-culture cardiovascular model. However, this study is not without its limitations. First, whilst the endpoints modelled some processes involved in atherosclerosis development, there are many other factors and key molecular events involved, which develop over a number of years (Libby et al., 2002; Rafeieian-Kopaei et al., 2014). Whilst it would be difficult to fully model in vitro, the investigation of further endpoints associated with atherosclerotic processes may add to the weight of evidence for the tobacco harm reduction potential of the NGP provided by this study. These include further investigation markers involved in monocyte adhesion, assessment of immune cell migration and additionally assessment of the formation of foamy macrophages (Park, 2021; Riddle et al., 2022). Ideally, human primary monocytes would be used for these experiments instead of the THP-1 monocytic cell line since, although a convenient and valuable tool to study monocyte function, there are significant differences between the immortalized cell line and the primary cells they are representing (Bosshart and Heinzelmann, 2016).

Furthermore, whilst the inflammatory panel provided useful insights into the dose-dependent effects of the test articles (and a comparison of both the pre-conditioning and test article-only conditions), some of the markers produced signals above or below the limits of detection, and therefore further optimisation may be required for these analytes.

The study also only assessed one variant of each of the NGP products, so does not account for changes in composition or flavour between products. However, the evidence indicates that each discrete product category generally induces relative responses as observed here: cigarette is the most potent test article, and responses are reduced upon exposure to HTP and further still with ENDS products (McNeill and Munafò, 2013; Zeller, 2019; Simms et al., 2022). An aqueous aerosol extract was selected for application in this study; this is a medium that has been previously utilised for the in vitro assessment of nicotine delivery products, and particularly within these types of models (Poussin et al., 2020; Simms et al., 2020; Czekala et al., 2021) and may model the aqueous soluble fraction of smoke aerosol, which may most likely absorbed through the lung alveolar wall and to be circulating within the blood (Poussin et al., 2014). However, comparison of effects with the addition of filterpad trapped particulate matter/aerosol collected mass may be beneficial to highlight any differences in results with the addition of this fraction of smoke/aerosol. Furthermore, the test articles in the study were not fully characterised in terms of their chemical composition (only select analytes within the bPBS stock solutions were measured) and therefore exposure comparisons were based on the stock solution dilutions rather than actual measurements within the chip. Additionally, the effects of primary constituents of the stock solutions, and not necessarily physiologically relevant metabolites, were assessed here. Therefore, further characterisation of exposure levels and metabolism of the test articles over time and also maintenance of initial exposure levels within the system would be useful to assess outcomes under even more physiologically relevant conditions.