Impact of cigarette versus electronic cigarette aerosol conditioned media on aortic endothelial cells in a microfluidic cardiovascular model

Atherosclerosis is a complex process involving progressive pathological events, including monocyte adhesion to the luminal endothelial surface. We have developed a functional in vitro adhesion assay using BioFlux microfluidic technology to investigate THP-1 (human acute monocytic leukaemia cell) monocyte adhesion to human aortic endothelial cells (HAECs). The effect of whole smoke conditioned media (WSCM) generated from University of Kentucky reference cigarette 3R4F, electronic cigarette vapour conditioned media (eVCM) from an electronic nicotine delivery system (ENDS) product (Vype ePen) and nicotine on monocyte adhesion to HAECs was evaluated. Endothelial monolayers were grown in microfluidic channels and exposed to 0–1500 ng/mL nicotine or nicotine equivalence of WSCM or eVCM for 24 h. Activated THP-1 cells were perfused through the channels and a perfusion, adhesion period and wash cycle performed four times with increasing adhesion period lengths (10, 20, 30 and 40 min). THP-1 cell adhesion was quantified by counting adherent cells. WSCM induced dose-dependent increases in monocyte adhesion compared to vehicle control. No such increases were observed for eVCM or nicotine. Adhesion regulation was linked to increased ICAM-1 protein expression. Staining of ICAM-1 in HAECs and CD11b (MAC-1) in THP-1 cells demonstrated adhesion molecule co-localisation in BioFlux plates. The ICAM-1 adhesion response to WSCM was downregulated by transfecting HAECs with ICAM-1 siRNA. We conclude that the BioFlux system is able to model human monocyte adhesion to primary human endothelial cells in vitro and WSCM drives the greatest increase in monocyte adhesion via a mechanism involving endothelial ICAM-1 expression.

A key early stage in the initiation of atherosclerosis involves circulating monocyte trafficking to the arterial endothelium in response to inflammation. The process of monocyte adhesion comprises upregulation of endothelial cell adhesion molecules such as E-and P-selectin, intercellular adhesion molecule-1 (ICAM-1), and vascular cell adhesion molecule-1 (VCAM-1) in lesion prone areas 13 . Selectins promote rolling along the endothelium and interact with P-selectin glycoprotein ligand-1 (PSGL-1), expressed at significantly higher levels by inflammatory monocytes. Endothelial VCAM-1 binds very late antigen-4 (VLA-4) to mediate slow rolling on inflamed endothelium, facilitating transition between rolling and firm adhesion. Firm monocyte adhesion to the endothelium is mediated by chemokine (C-C motif) ligand 2 (CCL-2) and IL-8 and endothelial ICAM-1. Monocyte arrest and subsequent extravasation is dependent on monocyte-expressed integrins lymphocyte function-associated antigen-1 (LFA-1) and macrophage-1-antigen (Mac-1/CD11b) alongside endothelial ligands ICAM-1 and ICAM-2 13,14 . Once attached, monocytes undergo extravasation, sub-endothelial accumulation and differentiation into macrophages. Over time, these cells give rise to lipid-laden pathogenic foam cells and initiate formation of atherosclerotic plaques 13 .
Cigarette smoke has been shown to induce monocyte activation, increase levels of circulating leukocytes and increase leukocyte recruitment and adhesion to the endothelium 15 . Endothelial cells exposed to aqueous cigarette smoke extract (CSE) express increased levels of pro-inflammatory adhesion molecules ICAM-1, E-selectin, VCAM-1 and CCL2 16 , causing increased monocytic MM6 cell adhesion to HUVECs 17 . Poussin et al. demonstrated a greater TNFα-induced monocyte adhesion response following HUVEC exposure to conditioned media from MM6 cells treated with CSE compared to direct HUVEC exposure to CSE 17 . Further work demonstrated higher concentrations of CSE are required to induce monocyte adhesion to human coronary artery endothelial cells (HCAECs) after direct endothelial cell treatment whereas lower concentrations are required following treatment with monocyte-conditioned media 18 . Kalra et al. demonstrated that adhesion of peripheral blood monocytes to bovine aortic endothelial cells (BAEC) or HUVECs increased following exposure to cigarette smoke condensate (CSC) 19 . A concomitant increase in CD11b expression on the monocyte surface was noted. Treatment with nicotine demonstrated no such effects 19 . HUVEC exposure to CSC increased endothelial ICAM-1 and E-selectin expression. Monocyte-to-endothelial cell adhesion was inhibited by treatment with monoclonal antibodies to CD11b, ICAM-1 and E-selectin 19 . Other work has linked cigarette smoke exposureinduced monocyte-to-endothelial cell adhesion to increased CD11b/ICAM-1 expression but not VCAM-1 or E-selectin expression 20,21 .
One strategy to reduce the harmful side effects of tobacco smoking is the development of next-generation tobacco and nicotine products (NGPs) such as electronic cigarettes (e-cigarettes) and tobacco heating products (THPs). Research by various groups has demonstrated that NGPs produce reduced levels of toxicants compared to cigarettes 22,23 . Recent comparisons between total particulate matter (TPM) generated from a 3R4F cigarette against TPM from a commercial THP, e-cigarette liquid (e-liquid) and aerosol produced by the same e-liquid found that THPs produce negative results in standard genotoxicity tests including the mouse lymphoma assay and in vitro micronucleus assay, whilst 3R4F TPM induced mutations in both assays 24,25 .
A recent review of the effects of e-cigarettes on cardiovascular health detailed several in vitro and in vivo studies which suggest that markers of oxidative stress, inflammation, vascular dysfunction and thrombosis were adversely affected by exposure to e-cigarettes, although to a lesser degree than tobacco smoke exposure 26,27 . A small study of 40 healthy subjects found smoking either a traditional cigarette or an e-cigarette was associated with increased oxidative stress markers and adversely effected brachial artery flow-mediated dilatation (an in vivo test of endothelial function), although e-cigarettes appeared to have lesser impact 28 . An in vitro study also found that incubation of platelets with either tobacco smoke or e-cigarette vapour extracts, led to upregulation of inflammatory markers and increased platelet activation, aggregation and adhesion 29 . However, the effect of NGPs on cardiovascular endothelial cell function, inflammation and monocyte activation still requires much greater investigation.
It has previously been suggested that the most appropriate cardiovascular model for cigarette smoke exposure would incorporate appropriate cardiovascular cell type(s), exposed to whole smoke bubbled media (containing both the particulate and vapour phase constituents) and would recapitulate in vivo vascular physiology by incorporating shear flow 30 . This model must demonstrate a correlation between clinical and in vitro effects of cigarette smoke exposure 30 and as chronic human diseases (such as CVD) manifest over several years to decades, the practical limitations of exposure for any in vitro model must be considered. These same criteria can also be applied to examine exposure of the cardiovascular system to vapour from NGPs.
The BioFlux system is a high content screening platform for running directional flow assays at precisely controlled, automated shear rates combined with high resolution fluorescent microscopy using well plate microfluidic technology. It combines the biological relevance of a laminar flow cell set-up with the convenience and form factor of standard laboratory 24-or 48-well microplates, by embedding micron-scale fluidic channels on the plate bottom. The system is fully integrated and does not require manual control of shear pressure regulation or flow direction as in traditional microfluidic systems. An automated microscope stage allows controlled imaging at pre-defined channel locations supporting experimental reproducibility and consistency. By controlling the liquid flow rate through the fluidic channels (shear flow), the system can simulate a physiologically relevant environment for endothelial cells which depend upon shear flow to maintain homeostasis. Previous studies have examined vascular endothelial cell responses under conditions that partially recapitulate physiological context using BioFlux automated microfluidic technology [31][32][33] . In vitro CVD models have been used in both the absence and presence of shear flow 34 . Cockcroft and colleagues demonstrated an in vitro cardiovascular model that utilised HUVECs and a variety of flow patterns to model TNFα-induced inflammatory responses 35 . Similarly, others have used bespoke in vitro systems to deliver varying rates of shear flow to BAECs [36][37][38] , HUVECs 39 and, more recently HAECs [40][41][42] . High shear flow has an atheroprotective effect on the endothelium and reduces endothelial cell response to cardiovascular risk factors compared to regions exposed to low or disturbed shear www.nature.com/scientificreports/ flow 43 . A previous study found that monocyte-to-endothelial cell adhesion is increased by CSE under low laminar flow but not under high laminar flow conditions 16 .
Here, we evaluate the effects of whole smoke conditioned media (WSCM) generated from University of Kentucky reference cigarette 3R4F, electronic cigarette vapour conditioned media (eVCM) from an electronic nicotine delivery system (ENDS) product (Vype ePen) and nicotine on human monocytic leukaemia (THP-1) cell adhesion to HAECs using the BioFlux microfluidic system. To investigate the mechanism of adhesion, cellular adhesion molecules (ICAM-1, MAC1/CD11b and E-selectin) were examined by immunofluorescence microscopy and western blotting, and subsequent modulation of the adhesion response was performed by transfecting HAECs with ICAM-1 siRNA prior to WSCM, eVCM or nicotine treatment.

Results
To determine the feasibility of using the BioFlux microfluidic device as an appropriate tool for modelling cardiovascular disease-related events in vitro, HAEC monolayers were grown in BioFlux microfluidic channels and their response to pro-inflammatory cytokine, TNFα, assessed ( Fig. 1). Images taken using phase contrast microscopy demonstrated that monocyte adhesion to HAECs was quantifiable in the BioFlux microfluidic system in response to TNFα treatment (Fig. 1a). An increase in monocyte adhesion to HAECs was observed with increasing TNFα concentration and adhesion period length in the BioFlux channels (Fig. 1b).
The impact of endothelial cell exposure to WSCM, eVCM or nicotine on THP-1 cell adhesion was evaluated across a range of non-toxic concentrations (15,150 and 1500 ng/mL n.e. 8 ) for 24 h alongside concurrent vehicle control, and 1 and 10 ng/mL TNFα as positive adhesion controls (Fig. 2). Representative channel images demonstrate THP-1 cell adhesion to HAECs after the final (40 min) adhesion period and wash step (Fig. 2a). THP-1 cell adhesion increased from 7.4-fold at the 10-min adhesion period to 13.8-fold at the 40-min adhesion period in response to HAEC treatment with 1 ng/mL TNFα compared to the concurrent vehicle control (Fig. 2b). THP-1 cell adhesion increased from 8.9-fold at the 10-min adhesion period to 21.7-fold at the 40-min adhesion period in response to HAEC treatment with 10 ng/mL TNFα compared to the concurrent vehicle control (Fig. 2b). WSCM-treated HAECs demonstrated a concentration-related increase in THP-1 cell adhesion after each incubation period compared to concurrent vehicle control-treated HAECs. WSCM (1500 ng/mL n.e.) induced an approximate threefold increase in THP-1 cell adhesion to HAECs after the 40-min incubation period, compared to the concurrent vehicle control (Fig. 2c). HAEC treatment with 15 or 150 ng/mL n.e. WSCM induced an approximate 1.5-fold increase in THP-1 cell adhesion after the 40-min incubation period, compared to the concurrent vehicle control (Fig. 2c).

Scientific Reports
The ICAM-1-mediated THP-1 cell-to-HAEC adhesion response was reproduced in BioFlux microfluidic channels treated with 0 or 1500 ng/mL WSCM, eVCM or nicotine for 24 h and processed for confocal microscopy using antibodies to endothelial ICAM-1 and monocyte-expressed CD11b. This demonstrated upregulation of ICAM-1 and evidence of interaction with adhered THP-1 cells via CD11b after HAEC exposure to TNFα or WSCM but not after exposure to eVCM or nicotine (Fig. 6).

Discussion
Cigarette smoke has been demonstrated to cause endothelial cell dysfunction 4 and upregulation of monocyte adhesion, a key initiating step in the development of atherosclerosis [15][16][17][18][19] . Although NGPs are believed to be a less harmful alternative to smoking traditional cigarettes 23 , there is still more research required to understand the potential consequences of NGP use, particularly on cardiovascular health. Here, we have demonstrated that THP-1 monocyte adhesion to HAECs in BioFlux microfluidic channels was statistically significantly increased after HAEC exposure to WSCM. Increased THP-1 cell adhesion to HAECs was noted across multiple WSCM concentrations (15,150 and 1500 ng/mL n.e.), with increases up to approximately threefold in response to 1500 ng/mL (n.e.) WSCM. We have previously demonstrated that this concentration of Figure 8. Analysis of ICAM-1 involvement in test article-induced monocyte adhesion to HAECs in the BioFlux microfluidic system. Quantification of monocyte adhesion to HAECs over adhesion periods of 10, 20, 30 and 40 min following 72 h transfection with control or ICAM-1 siRNA and 24 h HAEC exposure to (a) nicotine, (b) WSCM or (c) eVCM, with fold change graphs inset and representative brightfield images of BioFlux microfluidic channels captured after the 40 min adhesion period using a Zeiss Axio Observer.Z1 microscope (×10 magnification). Two-way ANOVA with Dunnett's multiple comparisons between all treatment conditions. Error bars denote ± SD (n = 6); *p < 0.05 **p < 0.01 ***p < 0.001. www.nature.com/scientificreports/ WSCM exposure has no effect on HAEC viability 8 . In comparison, a much lower level of THP-1 cell adhesion to HAECs was observed after exposure to eVCM at equivalent nicotine concentrations (up to 1.6-fold), with minimal response detected up to the final 40-min adhesion period. The observed effects were based on multiple independent WSCM and eVCM batches generated in compliance with Good Laboratory Practice. HAEC exposure to nicotine induced no notable increase in THP-1 adhesion. Further experimentation revealed that WSCM upregulates ICAM-1 levels in HAECs but not E-selectin or VCAM-1, measured by immunofluorescence microscopy and Western blot. No significant upregulation of ICAM-1, E-Selectin or VCAM-1 was detected following HAEC exposure to eVCM or nicotine. It should be noted that upon TNFα treatment, E-selectin re-distributed into intense fluorescent regions on the immunofluorescence microscopy images. This resulted in a higher fluorescence reading (arbitrary fluorescence units) compared to vehicle and WSCM treatments where E-Selectin staining was more diffuse. Western blot data indicated there was no change in E-selectin protein levels following 24 h TNFα treatment, thus cellular redistribution was considered to account for differences in the results between the two detection methodologies. Confocal microscopy analysis detected co-localisation between ICAM-1 and monocyte-expressed CD11b. This co-localisation occurred in regions of monocyte-to-endothelial cell adhesion, predominantly involving HAECs with upregulated ICAM-1 expression in response to WSCM or TNFα. The CD11b integrin is known to bind ICAM-1 in response to inflammation 13 . Transfection of HAECs with ICAM-1 siRNA provided further evidence for the role of ICAM-1 in WSCM-induced monocyte adhesion to HAECs. ICAM-1 depletion reduced THP-1 cell adhesion in response to WSCM to levels similar to those observed in the vehicle control. ICAM-1 depletion did not result in any such reductions in THP-1 cell adhesion to HAECs treated with eVCM or nicotine. Thus, WSCM-induced THP-1 cell adhesion to HAECs was preferentially ICAM-1-mediated.

Scientific Reports
Although no controlled shear pressure was applied during the monocyte adhesion periods, cells in the microfluidic channels were subject to gravity flow at all times. In an attempt to model a chronic condition that would otherwise occur over many years in a physiological setting and to model an inflammatory background in which quantifiable monocyte adhesion responses could be observed, we opted to use low shear rates to test the effects of smoke-conditioned media 43 . Previous work has demonstrated that high shear rates can be used to model a 'healthy' vascular environment in the BioFlux system 33 . Briefly, the effects of applied shear flow on primary HAECs in BioFlux microfluidic channels were investigated. Changes in cell and actin alignment in the direction of flow, real-time production of NO and gross cell membrane shape changes in response to physiological shear flow were observed 33 . These commercial systems have a range of potential applications, including within the consumer and pharmaceutical industries, thereby reducing the dependency on animal testing for regulatory safety assessments.
While in vitro CVD models have been used in both the absence and presence of shear flow, it is generally accepted that an ideal, dynamic in vitro CVD model would recapitulate in vivo vascular physiology by incorporating three dimensional (3D) shear flow. A wide range of in vitro microfluidic technologies encompassing 3D cell culture under flow have emerged with varying physiological relevance and complexity 30,33,[44][45][46] . These models can be used to study different aspects of vascular biology and offer varied advantages including: high throughput capacity, ease of setup and creation of a more physiological in vitro environment. They also come with limitations such as complex set-up, low throughput or lower physiological relevance 45,46 . It is important that the vascular model selected to define the biological mechanism of interest is considered for scalability, reproducibility, development potential and feasibility of use in addition to physiological relevance and throughput capacity. For example, the use of venous endothelial cells is less physiologically relevant then aortic cells to the study of atherosclerosis which predominantly occurs in the arteries however venous cells tend to be more widely available and have therefore been more commonly used in such in vitro models. Many labs assess monocyte adhesion in standard cell culture plate format or in plates with microfluidic channels that require a rocking platform to generate flow 45,46 . Our method utilises the commercially available BioFlux system and therefore offers a robust platform with high throughput capacity and reduced inherent variability to assess monocyte-to-endothelial cell adhesion at pre-defined, automated shear flows. Furthermore, in an attempt to more closely model atherosclerosis as a chronic and progressive condition the assay encompasses multiple adhesion periods of increasing length.
We conclude that the BioFlux microfluidic system may be a useful tool for modelling CVD in vitro to assess effects of both cigarette smoke and NGP vapour exposure. Our model encapsulates several of the criteria outlined by Fearon et al. for an appropriate cardiovascular model including a human-derived cardiovascular cell type (HAEC) and exposure to whole smoke or vapour bubbled media (WSCM and eVCM) 30 . The current assay design is relevant to a pathological environment in which shear flow is disturbed (close to zero) 47 . We have previously demonstrated that physiological shear flow induces cellular changes in HAECs in the BioFlux system 33 . To develop the monocyte-to-HAEC adhesion model further and to more accurately reproduce in vivo physiology, incorporation of constant shear flow could be explored. In conclusion, the THP-1 cell adhesion assay can be used to investigate the tobacco risk continuum 48 , as we observe consistently differential responses in monocyteto-HAEC adhesion after exposure to WSCM, eVCM and nicotine.

Methods
Cell culture. Primary Human Aortic Endothelial cells (HAECs), were sourced from Lifeline Cell Technology (Frederick, MD, USA; Catalogue Number FC-0014, Lot Number 00503), and maintained in VascuLife VEGF Endothelial Cell Culture Medium supplemented with VEGF LifeFactors (VECM; Lifeline Cell Technology) 8,33 . Cell stocks were preserved in the vapour phase of liquid nitrogen, initiated from frozen, and maintained in culture until near confluence (approximately 70-90%) 8,33 . Cells were cultured at 37 ± 1 °C, in a humidified atmosphere of 5% (v/v) CO 2 in air, and re-fed with VECM as required. HAECs at passage 5 were used in all experiments 8 www.nature.com/scientificreports/ Monocytic THP-1 cells, obtained from ATCC (Virginia, USA; Catalogue Number TIB-202), were maintained in RPMI 1640 media (Gibco, Billings, MT, USA) supplemented with 10% v/v heat-inactivated fetal bovine serum, penicillin and streptomycin. Cell stocks were preserved in the vapour phase of liquid nitrogen, initiated from frozen and maintained in culture with bi-weekly passage up to passage 30. Cells were cultured at 37 ± 1 °C, in a humidified atmosphere of 5% (v/v) CO 2 in air, and re-fed with supplemented RPMI 1640 as required.
WSCM or eVCM generation. 3R4F Kentucky Reference (3R4F) cigarettes were obtained from the University of Kentucky and stored at < − 10 °C. Prior to smoking, cigarettes were removed from frozen storage and conditioned as individual cigarettes for at least 48 h and no more than 10 days at 22 ± 1 °C and 60 ± 3% relative humidity (ISO 3402, 1999) 8 . 3R4F cigarettes were smoked using a Borgwaldt RM200 smoking machine according to the puffing parameters of ISO 3308, 2012, namely a 35 mL puff every 60 s with a 2 s duration 8 .
Vype ePEN II and Vype blended tobacco ePen cartridges (18 mg/mL nicotine) (British American Tobacco) were stored at 25 ± 5 °C. Prior to puffing, e-cigarettes were held in the laboratory in airtight plastic containers, with caps in place, until exposure commenced. Vype ePEN II e-cigarettes were puffed using a VC10 smoking robot according to the puffing parameters of CORESTA CRM 81, namely a 55 mL square wave puff every 30 s with a 3 s duration.
For each WSCM or eVCM generation, whole smoke from four 3R4F cigarettes or e-vapour from one cartridge was passed through a 30 mL glass impinger containing approximately 6 g of 3 mm glass beads and 20 mL VECM 8 . WSCM or eVCM was filter sterilised under aseptic conditions using a 1000 mL Stericup-VP polyethersulfone Express PLUS, radio-sterilised 0.10 µm filter (Millipore, Sheffield, UK) 8 . WSCM or eVCM was stored frozen in 1 mL aliquots at < − 50 °C prior to use, analysed for nicotine content by gas chromatography with flame ionisation detector (GC-FID) and used within 4 weeks of generation 8 .
BioFlux system. The BioFlux system (Fluxion; Alameda, CA, USA) is comprised of an air compressor and electropneumatic regulators to deliver precisely controlled pressure to a BioFlux plate via a pressure interface 33 . BioFlux plates are in a standard 24-or 48-well plate format with integrated microfluidic channels 33 . The bottom of each microfluidic channel consists of standard 180-μm coverslip glass, which allows microscopic examination at defined viewing windows 33 . Flow channels are connected to input and output wells from which reagents are pushed by pneumatic pressure through the channels 33 .
BioFlux plate preparation and seeding. BioFlux microfluidic channels were coated with 100 µg/mL fibronectin, seeded with HAECs in VECM and incubated overnight under gravity flow conditions at 37 ± 1 °C in a humidified atmosphere of 5% CO 2 in air, as described in Makwana et al. 33 .
BioFlux HAEC treatment. Prior to treatment, brightfield images were taken at 3 fields of view per channel for reference monolayer confluence and morphology. VECM was removed from inlet and outlet wells and appropriate treatments vehicle control (VECM), positive control (1 and 10 ng/mL TNFα, Life Technologies, Altrincham, UK) or eVCM, WSCM or nicotine (Sigma-Aldrich) [0, 15, 150, 1500 ng/mL nicotine equivalent (n.e.)] were added to the inlet wells. HAECs were treated for 24 h in total. Treatments were perfused through the channels at 2 dyn/cm 2 for 5 min to ensure appropriate distribution in the channel, then the plates were incubated under gravity flow conditions at 37 ± 1 °C in a humidified atmosphere of 5% CO 2 in air for the remainder of the 24 h. Six experiments with separate generations of WSCM, eVCM or fresh preparations of nicotine were performed.
BioFlux THP-1 cell adhesion assay. At the end of the incubation period, excess treatment solution was removed from all inlet and outlet wells. A sufficient volume of activated THP-1 cells at 5 × 10 6 cells/mL was added to inlet well B and Hank's Buffered Saline Solution (HBSS) (Gibco) was added to inlet well A. The plate was loaded onto the microscope stage with the BioFlux temperature controller set at 37 °C. THP-1 cells were perfused over the HAEC monolayer at 2 dyn/cm 2 for 30 s followed by 0.5 dyn/cm 2 for 1 min. Cells were incubated at 37 °C under gravity flow conditions for set adhesion periods and then washed with HBSS containing 10 µg/ mL Hoechst 33342 (H33342) (Sigma-Aldrich) at 2 dyn/cm 2 for 1 min, 5 dyn/cm 2 for 1 min and 6 dyn/cm 2 for 1 min. THP-1 cell perfusion, adhesion period and wash cycle was performed four times with increasing adhesion period length (10, 20, 30 and 40 min). Images were acquired using BioFlux Montage software (https ://biofl ux.fluxi onbio .com/monta ge-softw are). Phase contrast images were captured from 3 fields of view per channel pre and post each wash step. After the final wash step, images were captured using the DAPI filter (420-480 nm).
Following the THP-1 adhesion assay cells were fixed in 3.7% paraformaldehyde (Sigma-Aldrich) in HBSS applied through inlet well B at 5 dyn/cm 2 for 10 min at 37 °C. Excess paraformaldehyde was removed by washing in HBSS through inlet well A at 5 dyn/cm 2 for 5 min. Cells were permeabilised in 0.1% Triton-X (Sigma-Aldrich) applied through inlet well B at 5 dyn/cm 2 for 5 min and washed in PBS applied through inlet well A at 5 dyn/ cm 2 for 5 min. Non-specific binding was blocked by applying 5% BSA through the inlet wells at 2 dyn/cm 2 for at least 60 min followed by washing in PBS at 5 dyn/cm 2  www.nature.com/scientificreports/ outlet well at 2 dyn/cm 2 for 2 min and then incubated overnight at room temperature. Appropriate secondary antibodies (anti-rabbit IgG (H+L) F(ab′)2 fragment Alexa Fluor 555 conjugate and anti-mouse IgG (H+L) F(ab′)2 fragment Alexa Fluor 488 conjugate (Life Technologies)) were applied through the outlet well at 2 dyn/cm 2 for 2 min and then incubated at room temperature for 2-3 h. Cells were washed in PBS at 5 dyn/cm 2 for 5 min and imaged at 40× magnification using a Zeiss LSM780 confocal microscope and Zeiss Zen software (https ://www. zeiss .com/micro scopy /int/produ cts/micro scope -softw are/zen.html). Co-localisation was assessed using ImageJ (https ://image j.nih.gov/ij/). Immunofluorescence microscopy. HAECs were seeded into 8-well Ibidi µ-slide chambered coverslips (Thistle Scientific; Glasgow, UK), pre-coated with 0.1% gelatin from porcine skin (Sigma-Aldrich) for 20 min at 37 °C. HAECs were seeded at 1 × 10 5 /mL in VECM and incubated overnight at 37 ± 1 °C, in a humidified atmosphere of 5% (v/v) CO 2 in air. Confluent HAEC monolayers were treated with vehicle control (VECM), positive control (1 and 10 ng/mL TNFα) or eVCM, WSCM or nicotine [0, 15, 150, 1500 ng/mL nicotine equivalence (n.e.)] for 24 h at 37 °C. Six experiments with separate generations of WSCM, eVCM or fresh preparations of nicotine were performed. Cells were fixed in 3.7% paraformaldehyde for 10 min at 37 °C. Excess paraformaldehyde was removed by washing 3 times in PBS. Cells were permeabilised in 0.1% Triton-X for 4 min at room temperature and washed three times in PBS. Non-specific binding was blocked by incubating cells with 5% BSA for at least 60 min at room temperature followed by three PBS washes. Primary antibodies [ICAM-1 1:100, E-selectin 1:100 (R&D Systems)] in 1% BSA were applied to the cells and incubated overnight at room temperature. Secondary antibodies, anti-mouse IgG (H + L) Alexa Fluor 488 conjugate (Life Technologies) and DNA stain, H33342 were applied to the cells and incubated at room temperature for 2-3 h. Cells were washed three times in PBS and images captured with a Zeiss Axio Observer.Z1 microscope connected to a Hamamatsu CCD camera. Images were acquired using BioFlux Montage software. Three random fields of view were captured per well. Images were quantified using ImageJ to calculate fluorescence intensity of each image and divided by the number of cells per image to give average fluorescence per cell. Relative fluorescence was calculated by dividing average fluorescence at each treatment concentration by the vehicle control.