Quantification of Cigarette Smoke Particle Deposition In Vitro Using a Triplicate Quartz Crystal Microbalance Exposure Chamber

There are a variety of smoke exposure systems available to the tobacco industry and respiratory toxicology research groups, each with their own way of diluting/delivering smoke to cell cultures. Thus a simple technique to measure dose in vitro needs to be utilised. Dosimetry—assessment of dose—is a key element in linking the biological effects of smoke generated by various exposure systems. Microbalance technology is presented as a dosimetry tool and a way of measuring whole smoke dose. Described here is a new tool to quantify diluted smoke particulate deposition in vitro. The triplicate quartz crystal microbalance (QCM) chamber measured real-time deposition of smoke at a range of dilutions 1 : 5–1 : 400 (smoke : air). Mass was read in triplicate by 3 identical QCMs installed into one in vitro exposure chamber, each in the location in which a cell culture would be exposed to smoke at the air-liquid interface. This resulted in quantification of deposited particulate matter in the range 0.21–28.00 μg/cm2. Results demonstrated that the QCM could discriminate mass between dilutions and was able to give information of regional deposition where cell cultures would usually be exposed within the chamber. Our aim is to use the QCM to support the preclinical (in vitro) evaluation of tobacco products.


Introduction
Modelling human disease processes in vitro is important for our understanding of the risks associated with human exposure to known and unknown inhaled chemicals or toxicants. ese in vitro models can be used for mechanisticbased research and/or to assess potential harm of consumer goods, such as household products, cosmetics, or tobacco products. Although it can be argued that in vitro models have limitations in human physiological relevance, it is believed that these models have potential to reduce the �nancial and ethical burden on in vivo animal testing.
Looking speci�cally at the toxicological assessment of tobacco smoke, there are a number of in vitro models of toxicity using whole smoke exposure systems already in use [1][2][3][4][5][6][7][8][9]. Exposure system setups may differ greatly but their function and purpose is shared. A smoking machine/robot is used to generate and/or dilute mainstream whole smoke; this is delivered to an exposure chamber/module containing a simple or complex in vitro model of the lung at the airliquid interface (ALI) so that a biological endpoint (usually related to one of the major smoking related diseases) can be assessed. As with the diversity of exposure systems and setups available, the same can be said of the in vitro model and endpoint testing. ese may range from something as simple as a cytotoxicity assessment using a continuous cell line [1,2,8] through to complex endpoints assessing intracellular markers [9], or utilise more sophisticated cell cultures such as primary cell lines [10,11], coculture systems, 3D-tissue models, or whole lung slices [3]. Currently, there are no de�ned regulatory protocols for whole smoke exposure systems, but these are being developed to support human in vitro models of disease. Aforementioned, there are a vast number of whole smoke exposure systems described in the literature, either commercially available [1][2][3][4][5][6][7][8][9] or one-off inhouse setups [10,12,13]. Additionally, tobacco smoke is a concentrated and complex mix of at least 5,600 chemicals and toxicants found across two phases, the particulate (tar) and vapour phase [14]. us, assessing smoke dose is challenging. Consequently, when presenting whole smoke dose-response data, authors variously describe "dose" in many different ways: as a percentage of smoke; a fraction of smoke; ratios of smoke to air; puff number; total number of cigarettes smoked; total exposure of micrograms per culture insert; a �ow rate of mixing air and vacuum applied to a smoke dilutor, all depending on the machine being used to generate and dilute the smoke [15]. Hence there is a need to accurately quantify particle and/or chemical deposition in our in vitro systems, such that comparisons of biological endpoints between different systems can be achieved with improved precision and accuracy both within and ultimately between laboratories. is is of increasing importance to scientists and regulators as it will allow consistent interpretation of results and quick cross-comparison of biological endpoints for de�ned smoke doses [16,17]. Dosimetry (the assessment/measurement/quanti�cation of smoke dose) is therefore a key element that can unite the biological effects of whole smoke generated by various and diverse exposure systems [3].
ere are small numbers of chemicals or markers which can be quanti�ed to assess tobacco smoke dosimetry; most of these dosimetry measurements assess the particulate phase due to the challenges of measuring individual components in the vapour phase, especially at higher smoke dilutions (lower concentrations) [15]. Chemicals/markers have been selected historically due to their facile quanti�cation and include, but are not limited to: carbon monoxide [5] and oxides of nitrogen (NO ) which are in the gas phase; solanesol [18] and nicotine which are particulate markers; carbonyls which are split between phases; particulate matter as a whole either via gravimetric or chemical methods [7,15].
Adamson et al. [15] reported a gravimetric method of particulate deposition quanti�cation using a single quart� crystal microbalance (QCM) which robustly measured whole smoke particulate dose. e QCM is a sensitive gravimetric balance capable of measuring and detecting changes in mass of thin oscillating adherent �lms, within the nanogram range [19][20][21][22]. Previously, Adamson et al. [15] presented a new application (of an existing microbalance technology) of a single QCM unit in a chamber to assess the real-time deposition of tobacco smoke in vitro with veri�cation by a chemical �uorescence method for smoke particle deposition. We now present a new, unique, and characterised dosimetry tool where the exposure chamber base accommodates 3 identical QCM units (Figure 1), termed the triplicate QCM chamber or 3-in-1 QCM chamber. Primarily, the study objectives were to assess/give an indication of regional deposition within the chamber (uniformity of particle deposition to cell cultures) where the previous single unit QCM could not show us this, and to increase robustness through the use of higher replicate numbers per run (from 1 to 3). For users of this exposure chamber, both sets of additional information are new and useful and could not have been provided by the single unit QCM. In this study we investigated the ability of the triplicate QCM to detect mass differences of whole smoke dilutions from 1 : 5 to 1 : 400 (smoke : air, volume : volume), for 30 minutes/run. e range 1 : 5-1 : 400 was selected as * F 1: British American Tobacco's standard exposure chamber used for in vitro exposures to whole smoke at the air-liquid interface was modi�ed to accommodate the 3 QCM units. e picture shows a top view of the 3-in-1 QCM exposure chamber base (crystal ø = 2.54 cm; cell support insert ø = 2.4 cm). QCM position 1 is proximal to the passive exhaust port * . QCM positions 2 and 3 are distal to the exhaust port and behind position 1 on the right and le, respectively. this represents the RM20S dilutions used in the laboratory to generate a biological dose response using robust in vitro primary and continuous cell cultures [1,7,9].
Our results demonstrated that the QCM was able to discriminate mass between each dilution and was able to give information of regional deposition where ALI cell culture inserts would usually be at positions 1, 2, and 3 in the chamber ( Figure 2). Furthermore, these results showed absolute agreement with the single unit QCM data previously reported [15]. Overall, the integrated triplicate QCM tool delivered robust, real-time, quantitative whole smoke mass measurements at nanogram levels and demonstrated an achievable dose response.

Whole Cigarette Smoke Generation.
Reference cigarettes (3R4F, 9.4 mg pack tar) (University of Kentucky, Lexington, KY, USA) were used for all experiments. Whole cigarette smoke was generated and diluted using two identical Borgwaldt RM20S smoking machines (Borgwaldt-kc, Hamburg, Germany) within the same laboratory, as previously described [1,15]. e smoking machines and individual syringes from each machine were used interchangeably and at random and statistical analysis of the data thereaer showed that this had no effect on mass values obtained at the same dilution (data not shown). Five smoke dilutions were programmed as a ratio of smoke to air-1 : 5, 1 : 10, 1 : 25, 1 : 100, and 1 : 400 (smoke : air, volume : volume)-and 5 replicate experiments were conducted per dilution ( 5). For all experiments, the machine smoked for 30 minute duration at ISO 4387 : 2000 smoking regime (35 mL puff over 2 seconds, once a minute, to a de�ned number of � puffs/cigarette). Aer the last cigarette was extinguished, the QCMs were le to record real-time data until all residual smoke in the chamber had deposited and mass values were observed to plateau and stabilise, usually taking an additional 10 minutes.
To prove deposition was due to smoke particulate alone, a Cambridge �lter pad (CFP) (Borgwaldt-kc, Hamburg, Germany) was installed inline of smoke generation just prior to entry to the chamber for an additional single run of 1 hour (twice the duration of a standard run). As CFPs effectively trap 99.9% particulate matter [14,23], anything detected by the QCM would therefore give an indication of nonparticulate mass activity within the exposure chamber during a smoke run.

e QCM Exposure Chamber
Module. e previously described in vitro whole smoke exposure chamber manufactured for BAT by Curbridge Engineering (Southampton, UK) [7] had its base adapted to symmetrically house 3 identical commercially available QCM microbalance units (Vitrocell Systems GmbH, Waldkirch, Germany). e QCM housing units were installed with 5 MHz AT cut quartz crystals held between two Au/Cr polished electrodes, 1 inch (2.5 cm) diameter as described by Mülhopt et al. [19] (Figure 1). e QCM read at a resolution of 10 nanogram/cm 2 /second [15].
Before smoke exposure, the 3-in-1 QCM chamber was sealed and acclimatised in an incubator at 37 ∘ C to ensure quartz crystal stability. Quartz crystal stability was reached by zeroing the soware before exposure until the baseline showed negligible dri (zero point stability) of less than 20 ng/cm 2 over a duration of a few minutes. is took 5-10 minutes of zeroing prior to and between exposures in a stable environment. During longer periods of stability where the crystal was zeroed over a period of a few hours at constant environmental conditions, we were able to observe absolute zero point stability (0.00±0.01 g/cm 2 ) for 5 minutes prior to smoke exposure. During the deposition phase, the QCM recorded mass every 2 seconds for the 30 minute smoke exposure and 10 minute plateau phase, reporting as mass per unit area. Cell culture media was not included in the chamber for these mass measurements therefore media-in and media-out ports were blocked airtight to stop smoke leaking. Aer whole smoke exposure and between consecutive runs, quartz crystals were cleaned in situ (within the chamber, still screwed into their housing units). e crystal's surface was wiped with a so lint-free tissue and 70% ethanol, and then polished to a shine. Crystals were not removed/replaced between readings; however, these should be replaced if broken or if their surface is severely scratched. Crystal stabilisation time was greatly reduced from up to 60 minutes to around 10 minutes when the crystals were cleaned in situ.

Statistics.
Data were reported as a mean ± standard deviation. e individual value plot of QCM-detected mass (Figure 3(a)), the multi-vari chart (Figure 3(b)), and the interaction plot comparing both QCM devices ( Figure 4) were created using MINITAB v.16 statistical soware. Main effects plots to check experimental variables (not shown) and a one-way ANOVA test using Tukey method to assess differences between QCM positions and QCM devices were also determined using MINITAB. All residual plots for all graphs generated by MINITAB were checked to ensure the quality of the data obtained. Real-time traces of deposited mass ( Figure 5) were made using Microso Excel.

Results
e 3-in-1 QCM exposure chamber was able to record particulate mass in a dose-dependent manor in the dilution range 1 : 5-1 : 400 (smoke : air, v/v) (Figure 3(a)). Mean deposited mass ranged from 0.21 g/cm 2 (210 ng/cm 2 ) ±0.06 g/cm 2 at the most dilute dose of 1 : 400, up to 28.00 g/cm 2 (27,998 ng/cm 2 ) ±2.25 g/cm 2 at the most concentrated dose of 1 : 5. is mass range was obtained from the means of all three QCM positions per dose ( /position). A one-way analysis of variance (ANOVA) test showed that, as expected, there were signi�cant differences between dilutions ( ); however, 1 : 100 and 1 : 400 dilutions had grouped con�dence intervals indicating that overall mass values were in the same range at these high dilutions. e data were presented by individual QCM position so that the regional deposition around the chamber could be assessed (Figure 3(a)). QCM position 1 is most proximal to the passive exhaust port where smoke exits the chamber, whereas positions 2 and 3 are paired distal to the port, right and le, respectively (Figures 1 and 2). To assess the distribution of particulate deposition around the chamber positions at all the dilutions tested, a multi-vari chart was produced (Figure 3(b)). is chart demonstrates the uniformity of particle deposition in the biological exposure range, showing the mean deposited mass of the three QCM positions at all �ve dilutions tested. ere were no signi�cant differences between QCM positions 1, 2, and 3 for all dilutions tested-1 : 5 ( 2 ), 1 : 10 ( ), 1 : 25 ( 2 ), 1 : 100 ( ), and 1 : 400 ( ). e dose-response data generated from this triplicate QCM tool was compared with the aforementioned single unit QCM [15], and good parity was observed between them with no statistically signi�cant difference between the different tools at all dilutions tested (Figure 4). For dilutions 1 : 10-1 : 400 deposition was highly comparable between the two tools ( 5%) ( Table 1). At the most concentrated dilution of 1 : 5, the difference in detected deposited mass between the tools was 8.4% (Table 1). Considering the entire range tested, the difference between both tools at all dilutions was <10% which we would consider to be an acceptable difference (�t for purpose).
Finally, a Cambridge �lter pad (CFP) was placed inline of smoke generation to occlude particulate entering the chamber-the purpose was to assess semivolatile deposition or potentially the behavior of other gases. Figure 5 shows a real-time trace of particulate detection during an extended smoke exposure of 1 hour at the highest smoke concentration of 1 : 5 (smoke : air, v/v) with a CFP inline. e trace showed that mass increased over time, but this increase was nominal considering the high tar level of 3R4F cigarettes (9.4 mg) and the high concentration of whole smoke chosen (1 : 5). ere was an increase of 78 ng/cm 2 with the CFP inline compared with particulate matter deposition of 27,998 ng/cm 2 for BioMed Research International 5 T 1: Mean deposited mass values obtained from the single unit QCM [15] ( ) compared to the 3-in-1 QCM unit ( ); the percentage difference between mean mass detected by the two tools; the values indicating no signi�cant difference between tools at all dilutions tested. For the single unit, the values were the mean of 5 repeat experiments with the QCM (in position 2 of the chamber) [15]; for the triplicate QCM the values were the mean of 5 repeat experiments and 3 QCM positions per repeat ( ).
Mean deposited mass ( g/cm 2 ) ± SD Dilution (1 : ) 5  diluted whole smoke at the same dilution of 1 : 5. Furthermore, aer 1 hour of smoking the overall mass reached was <200 ng/cm 2 (mean of the 3 positions was 156 ng/cm 2 ). Whilst the initial chart would suggest that negligible mass was detected and that the traces were �at lined, however, when the scale was adjusted ( Figure 5 pullout) a repeated and conserved saw-tooth pattern was observed, indicative (at least in time) of some transient mass deposition, probably constituents of diluted vapour phase or water vapour entering the chamber. In addition, a further inline CFP measurement at the higher smoke dilution of 1 : 400 showed a similar sawtooth mass deposition pattern of negligible cumulative mass (not reported). e single "n-shaped" unit observed here would suggest a small increase, plateau, and decrease in mass as diluted vapour phase smoke was delivered to the chamber. is could be an indication of semivolatile deposition on the QCM surface followed by evaporation as air�ow through the chamber changed per puff.
As noted, cumulative mass over 60 minutes with the CFP inline was 156 ng/cm 2 /hr. If the sum of the transient peaks is calculated with subsequent reevaporation ignored ( Figure 5), the cumulative mass was approximately 620 ng/cm 2 /hr. is amount of detected mass is too low to be water vapour or vapour phase in its totality. e repeat pattern does suggest some components of the vapour phase but it would be difficult to identify semivapour phase revolatilising here without additional chemistry analysis. Rodgman and Perfetti [24] cite an example cigarette of generating 7.5 mg vapour phase and 17.4 mg NFDPM (nicotine free dry particulate matter). If this proportionality is maintained for the 3R4F cigarette, ISO yields of 9.4 mg NFDPM, 0.87 mg water (measured), and 4.1 mg vapour phase (estimated) and calculating orders of magnitude and the contribution of particulate, vapour, and water from a 3R4F cigarette, we would expect to see about 14,000 ng for total vapour phase and about 2,800 ng for water. us what we have identi�ed on the QCM with the particulate phase occluded is too low to be either water or total vapour.

Discussion
In this study we present a new and unique tool (Figure 1) to quantify diluted whole smoke particulate matter deposition in vitro. is tool measured real-time deposition at a range of dilutions 1 : 5-1 : 400 (smoke : air, v/v) resulting in quanti�cation of deposited particulate matter in the range 0.21-28.00 g/cm 2 , most dilute to most concentrated smoke dilution (Figure 3(a)). e technology was able to detect deposited mass with such resolution that it could identify puff-by-puff pro�les, probably of some vapour phase constituents alone (when particles were occluded with a �lter pad) ( Figure 5). For the �rst time, mass values were read in triplicate by 3 identical QCM units installed into the exposure chamber, each in exactly the same location as the cell culture inserts would be exposed to whole smoke at the ALI ( Figure  2). is information is particularly useful to users of the chamber as it demonstrates uniformity, with no difference in particle deposition to the 3 positions where culture inserts would be exposed to smoke.
To further assess the sensitivity of the QCM technology, a Cambridge �lter pad (CFP) was placed inline of smoke generation to effectively occlude the particulate fraction of whole smoke. e hypothesis here was that no particulate  mass would be detected and anything being detected would be derived from the diluted gas/vapour phase delivered to the chamber. As predicted, nominal mass was detected in the range <200 ng/cm 2 in 1 hour; however, a distinct and repeated saw-tooth pattern was observed in the real-time trace ( Figure 5). Although we are yet to identify what is being deposited on the QCM surface when the CFP is inline, we predict this could be an indication of some vapour phase components condensing and evaporating as air�ow through the chamber changes per puff. Elution, chemical analysis, and quanti�cation will help us potentially identify what part of the vapour phase is deposited on the exposed QCM surface, and we plan to do this in subsequent studies. ere are unique challenges not only in the identi�cation of smoke components but also their quantitation [14]. Of the particulate matter which is being quanti�ed by the QCM, grossly it approximates 16% water, 6% nicotine, and 78% "tar" (NFDPM), of which the major components of NFDPM are alcohols (20%), acids (17%), and aldehydes and ketones (14%) [14]. Sophisticated analytical tools, such as gas-chromatography mass spectrometry (GC/MS), are required to further identify these compounds. e challenge persists when it is estimated that components representing less than 1 mg of the particulate phase remain unidenti�ed, averaging in the low nanogram and picogram levels [14]. At the ALI cell exposure surface, the particles depositing will have a soluble and an insoluble fraction. erefore there will be species available to the thin surface liquid which lines the cell monolayer. ere is little direct information published on soluble particle components and cellular interactions at the ALI, but recent in-house data suggests that a soluble fraction is 40%-90% of the particle droplet. Lastly and as previously discussed, vapour phase components are largely overlooked by this sampling method; Figure 5 demonstrates the minor quantities of material being deposited when the diluted smoke passes through a CFP.
We previously published a study demonstrating the use of a single unit QCM in the same exposure chamber [15]. Based on that data obtained, and usefulness of the tool, we further developed the triplicate chamber to support in vitro testing and con�rm uniformity of particle deposition around the chamber, where the single unit QCM device could not show us this. As well as allowing us to assess potential positional deposition around the exposure chamber, this expanded tool allowed us to increase dosimetry replicate number per exposure. Within each dilution tested, there were no statistically signi�cant differences in deposition around the 3 positions in the chamber (Figure 3(b)) with values for all dilutions greater than 0.1.
Data obtained from the triplicate QCM exposure chamber compared to that obtained from the previously published single unit device [15] con�rmed they were performing equally (Figure 4). At dilutions greater than 1 : 10, the results obtained from the two tools differed by 5% or less in terms of quanti�ed mean deposited mass; this is to be expected as interrun variability. At 1 : 5 dilution, the difference was slightly higher at 8.4% (Table 1). But overall, these differences were all <10% (a range of 6.1% across the dilution range tested) which we would deem to be acceptable.
We have presented a new study demonstrating an additional/expanded, novel, and simple tool used to quantify tobacco smoke deposition in vitro in real time. e QCM technology itself is not new and in the past it has been used for such activities as environmental monitoring [21,25], biological applications [26,27], and inhalation toxicity assessment of (nontobacco) aerosols and engineered nanoparticles in vitro [19,28,29]. However, the utilisation and combination of these technologies for assessment of whole smoke exposure in vitro is novel. Certainly, QCM technology has many potential applications in the �eld of inhalation toxicology, not just tobacco smoke.
As discussed previously, there are many different setups available to industry and other respiratory toxicology research groups; in this study we used the Borgwaldt RM20S to dilute and deliver mainstream smoke to the triplicate QCM. Because so many different smoking machine and exposure systems are used, it is clear there needs to be a simple and aligned method of measuring dose in vitro; this is especially the case with the current switch from liquid to ALI exposures of aerosols in vitro [16]. Dosimetry is therefore a key element that can link the biological effects of tobacco smoke generated by various exposure systems [3]. As dose measurements with a QCM allow real-time, fast, and accurate determination of the cell deposited dose [28], we believe microbalance technology offers a way forward, not only for the assessment of whole smoke but also for other inhalable toxicants [30]. In terms of human smokers, this tool can be used to generate deposition values in vitro consistent with values measured in human smokers. For example, in this study QCMs detected deposition of diluted particulate matter in vitro in the range 0.21-28.00 g/cm 2 . For context, McAughey et al. [31] estimated daily deposited lung doses in the order of 40-100 g/cm 2 in the extrathoracic region, 1.0-2.0 g/cm 2 in the bronchial/bronchiolar region, and 0.1-0.2 g/cm 2 in the alveolar-interstitial region, based on human smoking data for lung retention and regional deposition. ese human values in the lower range 0.1-2.0 g/cm 2 (bronchial-alveolar) further align with our in vitro exposure system: primarily it is cells of the bronchial epithelium or alveolar (A549) cells which would be used in these biological models of disease and exposed at the ALI in this chamber. e 3-in-1 tool described here was compared to the previously published single QCM chamber ( Figure 4). Currently, neither of the these QCM chambers can accommodate cell cultures at the same time as deposition quanti�cation (the 3-in-1 cannot house cells as there is no space, and yet the single QCM has not been tested concurrently with media and/or cell cultures). However, this is something we will look at in future studies. us the operational advantage of the triplicate QCM versus the single QCM chamber is to increase replicate number from one to three during a single smoke exposure. is would be useful for deposition dose-range studies of different cigarettes types, for example. To obtain directly comparable chemistry deposition data alongside QCM deposition, the single QCM chamber should be used as positions 1 and 3 are available to house nude cell support inserts. Furthermore, if cell cultures could be supported on inserts for a whole smoke exposure without basal media, then this single unit tool could be used to obtain deposition data at the same time as a biological response. As the data have shown here, there is no statistically signi�cant difference between the tools, thus both QCM chambers have applicability depending on the design of the experiment and the desired outcome.

Conclusion
In this study, the QCM successfully determined deposited smoke mass generated from the Borgwaldt RM20S Smoking Machine, but it can easily be adapted to assess smoke mass produced from other commercially available smoking machines. Consequently, we predict that this tool may help align smoke exposure technologies. Additionally, the 3-in-1 QCM exposure chamber, although designed to quantify deposited smoke mass in vitro, could potentially be used to assess other types of aerosol delivery to in vitro cultures, such as manufactured particles and �bres, some aerosolised cosmetics, household products, or pesticides. Certainly aerosols similar to tobacco whole smoke which are submicron liquid spherical droplets would be easy to detect. Hence the scope of this tool is vast. e 3-in-1 QCM also offers a robust and efficient alternative to traditional chemistry methods or supports such methods and delivers additional bene�ts of particulate quanti�cation: (1) it is single person operated; (2) no other resource or reagents are required; (3) data is generated fast, in real time and because of this; (4) it can be used as a QC device to rapidly assess the status of a smoke exposure or identify issues with machine smoking and dilution.
Until now, there have been few (if any) reliable and accurate methods of determining particulate dose delivered to the exposure chamber in real time. us biological dose-response data has been presented as the machine's programmed ratio of smoke to air, with the expectation that smoke dilutions delivered from the smoking robot are robust and repeatable. In the case of the Borgwaldt RM20S, precision of smoke dilution and delivery has been previously investigated and proven to be reliable by a number of physicochemical and analytical methods, and this has been published [1,5]. e addition of this QCM work further supports our con�dence in the machine's ability to dilute and deliver smoke reliably.
Finally, we propose to continue to use these QCM tools to support the preclinical in vitro evaluation of tobacco products and accurately quantify particle dose delivery to cell cultures. In addition, this tool has been shown to align generated deposition values in vitro consistent with values measured in human smokers.