Physicochemical and toxicological properties of wood smoke particulate matter as a function of wood species and combustion condition

Abstract

Wood burning is a major source of ambient particulate matter (PM) and has been epidemiologically linked to adverse pulmonary health effects, however the impact of fuel and burning conditions on PM properties has not been investigated systematically. Here, we employed our recently developed integrated methodology to characterize the physicochemical and biological properties of emitted PM as a function of three common hardwoods (oak, cherry, mesquite) and three representative combustion conditions (flaming, smoldering, incomplete). Differences in PM and off-gas emissions (aerosol number/mass concentrations; carbon monoxide; volatile organic compounds) as well as inorganic elemental composition and organic carbon functional content of PM_0.1 were noted between wood types and combustion conditions, although the combustion scenario exerted a stronger influence on the emission profile. More importantly, flaming combustion PM_0.1 from all hardwoods significantly stimulated the promoter activity of Sterile Alpha Motif (SAM) pointed domain containing ETS (E-twenty-six) Transcription Factor (SPDEF) in human embryonic kidney 293 (HEK-293T) cells, a biomarker for mucin gene expression associated with mucus production in pulmonary diseases. However, no bioactivity was observed for smoldering and incomplete combustion, which was likely driven by differences in the organic composition of PM_0.1. Detailed chemical speciation of organic components of wood smoke is warranted to identify the individual compounds that drive specific biological responses.

Graphical Abstract

1. INTRODUCTION

Wood burning is a major contributor to global particulate matter (PM) air pollution due to use of wood in appliances for domestic heating and cooking in high and low-income countries,¹ as well as due to the rising global incidence of wildfires². Inefficient wood burning devices account for ~80% of airborne particle concentrations in ambient air during the winter heating season.³ Air pollution is associated with significant mortality and disease burden in vulnerable populations.^4,5 During the summer, forest fires in the Western U.S. and recently as far north as Canada are a major source of air pollution.^6,7 The World Health Organization (WHO) estimates that roughly 4 million people worldwide die prematurely each year from diseases caused by domestic burning of biomass.⁸ Airborne PM from domestic wood burning may be to blame for 10,000 to 40,000 premature deaths annually in the US.^4,9

Wood burning emissions contain a heterogenous mixture of gaseous and particulate matter which includes chemical species such as oxides of carbon, nitrogen and sulfur, several elemental and ionic species, and hundreds of volatile, semi-volatile and non-volatile organic compounds^10,11 that when inhaled can lead to adverse cardiopulmonary and neurological consequences.^12–14 Additionally, wood smoke exposure is associated with exacerbation of asthma-like symptoms and chronic obstructive pulmonary disorder (COPD), diseases associated with mucus hypersecretion and obstruction.¹⁵ In our previous studies, it was shown that wood smoke is associated with a higher risk for chronic bronchitis symptoms and lower lung function in humans,¹⁶ and that mice exposed to wood smoke had higher levels of smoke-induced inflammation.^17–19 Furthermore, the water-soluble, hydrophilic components of pine smoke such as sugars were found to induce the activity of the promoter of Sterile Alpha Motif (SAM) pointed domain containing ETS (E-twenty-six) Transcription Factor (SPDEF) gene, a transcription factor responsible for increasing mucin gene expression and necessary glycosylating enzymes and thereby the mucous differentiation phenotype of airway epithelial cells.^20,21 Other unidentified organic chemical compounds in wood smoke may also be driving or inhibiting mucin gene expression. Several studies^22–26 have previously characterized the chemical composition of the inhalable fraction of woodburning PM and certain toxicological endpoints, however these studies employed commercially available logwood or pellet stoves/heaters and domestic fireplaces which do not allow for a systematic investigation of the role of different woodburning parameters (e.g., wood species, burning conditions, moisture content) on the physicochemical properties of PM and the factors that affect the biological activity of the PM.²⁷

Here, we present a systematic investigation of physicochemical and toxicological properties of ultrafine wood smoke particles (PM_0.1) emitted from three different hardwood species commonly used as firewood during three different burning scenarios, using our recently developed integrated methodology that allows for well-controlled parametric studies under reproducible conditions of wood burning.²⁸ Emitted PM and gaseous co-pollutants were physicochemically characterized using state-of-the-art instrumentation and analytical methods. To assess the toxicological properties of emitted PM_0.1, a rapid luciferase reporter assay was employed to measure the bioactivity of SPDEF promoter in PM_0.1-exposed human embryonic kidney 293 (HEK-293T) cells. The primary innovation of our research was to provide a link between specific wood species, wood burning scenarios, physicochemical properties of wood smoke aerosols and their potential pulmonary health implications. Although only a single biological endpoint relevant for mucus generation was investigated, the rapid bioassay employed here serves as a reliable screening tool to prioritize future mechanistic investigations of the most bioactive PM exposures for their role in the development of other adverse pulmonary outcomes.

2. MATERIALS AND METHODS

2.1. Wood chips and physicochemical characterization

We chose three species of hardwoods (oak, cherry, mesquite) that are commonly used for domestic heating, cooking and grilling applications due to their high heating value and the medium-to-strong smoky flavor they impart to food.^29–33 Red oak was procured from Home Depot as a ¼ in. × 1 in. × 3 ft. board, which was then chopped into fine wood chips. Pre-made cherry and mesquite chips were procured from Weber (product names: Cherry Firespice, Mesquite Firespice, respectively) on Amazon.com. A commercial moisture meter (Proster Digital Wood Moisture Meter, Proster Trading Limited, Hong Kong) was utilized to measure inherent moisture content (% water on a dry wood mass basis) of the as-obtained wood chips.

Bulk inorganic elemental composition of the wood chips was analyzed by inductively coupled plasma mass spectrometry (ICP-MS) and quantitative concentrations of 55 elements were reported as % weight of element by weight of wood (% w/w). Further ICP-MS details are described in Supporting Information (SI).

2.2. Wood burning exposure platform and combustion conditions

Wood combustion experiments were conducted using the modified Integrated Exposure Generation System (INEXS), a platform previously developed by authors to investigate thermal decomposition behavior of materials,^34–37 and recently adapted and used to assess potential toxicological implications of emissions from burning of pine wood (Fig. S1).²⁸ Briefly, the versatile platform allows for thermal degradation of materials under controlled combustion conditions, i.e., O₂:N₂ ratio in the feed air, heating rate, and final temperature (up to 1200 °C). A variety of state-of-the-art real-time monitoring and time-integrated sampling instrumentation are connected in-line with the platform to allow for sampling and detailed physicochemical and toxicological characterization of the PM and gaseous emissions.

Residential wood combustion for heating and cooking purposes can expose occupants to freshly generated wood smoke that may be produced from different wood types and under different burning conditions. For example, high-temperature flaming conditions can be achieved using a modern pellet stove that is designed for high combustion efficiencies,³⁸ whereas incomplete combustion typically occurs in conventional wood stoves where fuel overloading and insufficient ventilation lead to high fuel-to-oxygen conditions.³⁹ Smoldering may happen when wood logs undergo a slow, low-temperature, sustained burn for hours after the flaming phase of a fireplace.⁴⁰ For this study, the effect of wood species was assessed by comparing the three hardwoods (oak, cherry, mesquite) under the flaming scenario, as most of the recently manufactured wood stoves are EPA-certified and burn fuel efficiently at high temperatures.⁴¹ Since oak is one of the most popular and versatile firewood used for domestic heating and grilling applications due to its easy availability, extremely high heating value and desirable smoke flavor for different foods, it is more likely to encounter a variety of burning scenarios compared to other hardwoods.^29,42 Hence, to investigate the effect of combustion condition, the three real-world scenarios were studied for one hardwood only, i.e., oak (Table S1).

Operational parameters for these combustion scenarios were determined from a literature review of the known thermal decomposition behavior of common hardwood species.⁴³ For flaming, the final temperature was set to 600 °C with a heating rate of 20 °C/min and ambient O₂ concentration (20.9 vol% in air), followed by holding at 600 °C for 15 min in order to ensure complete combustion. For smoldering, the final temperature was set to 300 °C at the same heating rate and O₂ concentration, followed by a continuous burning at 300 °C for 1 h. For incomplete combustion, temperature conditions were the same as flaming except for a much lower O₂ concentration (5 vol% in air). The starting weight of wood chips for combustion was fixed at 50 mg in order to ensure reproducibility of the combustion atmosphere between replicate experiments.

2.3. Real-time monitoring of gaseous emissions

Evolved carbon monoxide (CO) during combustion was monitored using a portable combustion analyzer (Bacharach PCA3, Pittsburgh, PA) and measurements (in ppm) were logged every 10 s for the entire duration of combustion. Gaseous total VOCs (TVOCs) were monitored using a portable photo-ionization detector probe (TVOC probe, TG-502, GrayWolf Sensing Solutions, Shelton, CT), with measurements (in ppb) logged every 10 s.

2.4. Real-time monitoring and physical characterization of emitted PM

Emitted PM during combustion was monitored using the Scanning Mobility Particle Sizer (SMPS) Model 3080 (TSI Inc., Shoreview, MN) for particle number concentration as a function of size (5-300 nm aerosol mobility diameter) and the Aerodynamic Particle Sizer (APS) Model 3321 (TSI Inc., Shoreview, MN) for submicron and micron-sized particles (0.5-20 μm aerodynamic diameter). Due to the high nanoparticle number concentrations, the emitted aerosol was sufficiently diluted (200-fold) using the Rotating Disk Thermodiluter Model 379020A (TSI Inc., Shoreview, MN) before the SMPS measurements, as described previously.³⁴ Measurements were logged every 2 min for SMPS and every 10 s for APS for the entire duration of combustion. The reported SMPS concentrations were adjusted for the dilution factor.

2.5. Time-integrated and size-fractionated PM sampling

Emitted PM was also size-fractionated and collected using the Harvard Compact Cascade Impactor (CCI)⁴⁴ to collect various PM aerodynamic size fractions, i.e., PM_0.1, PM_0.1-2.5, PM_2.5-10 and PM_>10 for physicochemical and toxicological characterization, although only PM_0.1 was characterized in this study.³⁴ PM_0.1 was collected on Teflon filters (PTFE membrane disc filter: 2 µm pore size, 47 mm diameter, Pall Corporation, Port Washington, NY) while the larger size PM were collected on polyurethane foam (PUF) substrates, as described previously.³⁴ For elemental and organic carbon analysis (EC-OC, see below), PM_0.1 was collected on pre-baked quartz filters (Pallflex® Tissuquartz™ filter: 47 mm diameter, Pall Corporation, Port Washington, NY). In addition, gravimetric analysis of the collected PM size fractions was performed using an analytical microbalance (Mettler Toledo, Columbus, OH) to calculate average PM mass concentrations (mg/m³ of air) as a function of aerodynamic size in the emitted aerosol. It is worth noting that since the sampling flow through the CCI was fixed at 30 L/min and the aerosol flow coming from the furnace was 5 L/min, the aerosol was diluted by a factor of 6-fold prior to sampling, which was used to correct the calculated average PM mass concentrations in the aerosol.

2.6. Chemical characterization of PM_0.1

Elemental and organic carbon (EC-OC):

Since PM_0.1 was found to be the most bioactive in our previous study on pine burning emissions,²⁸ we only analyzed this size fraction for chemical and toxicological characterization in this study. PM_0.1 collected on quartz filters was analyzed for its EC-OC content by thermal-optical transmittance (TOT) method based on NIOSH method 5040 using a Lab OC-EC Aerosol Analyzer (Sunset Laboratory Inc., Tigard, OR).⁴⁵ Further details are described in SI.

Bulk inorganic elemental composition:

The total inorganic elemental composition of PM_0.1 was analyzed by preparing PM_0.1 extracts in water (details on preparation in subsequent section) and analyzing for mass concentrations of 55 elements using ICP-MS. Elemental concentrations were reported in ng/L and converted to % w/w of PM_0.1 using the known PM_0.1 mass concentration in suspension.

Organic functional group characterization:

The functional characterization of organic carbon in PM_0.1 was performed by analyzing PM_0.1 aqueous extracts (details on preparation in subsequent section) through one-dimensional proton nuclear magnetic resonance spectroscopy (1-D ¹H-NMR) using the Bruker AVANCE III HD 600 MHz spectrometer (Bruker Scientific LLC, Billerica, MA). The methodological protocol was identical to that previously used to characterize the functional content of different types of organic compound mixtures such as atmospheric aerosols,^46,47 electronic cigarettes,⁴⁸ incinerated nano-enabled thermoplastics,³⁵ and printer-emitted nanoparticles.⁴⁹ The concentration of non-exchangeable functional protons was computed based on the linear relationship between signal intensity and proton concentration using Trimethylsilylpropanoic acid (TSP) as the internal standard.⁵⁰ The following functional groups of non-exchangeable organic hydrogen were quantified: (i) saturated aliphatic hydrogen [H-R] (δ_H 0.6-1.8 ppm); (ii) allylic hydrogen [H-C-C=] (δ_H 1.8–3.2 ppm); (iii) saturated oxygenated hydrogen, including α-hydrogen to hydroxyl, ether and ester group [H-C-O] (δ_H 3.2–4.7 ppm); (iv) acetalic and vinylic hydrogen ([O-CH-O]+[H-C=]) (δ_H 5.0–6.4 ppm); (v) aryl hydrogen [H-Ar] (δ_H 6.5–8.3 ppm), and (vi) carbonyl hydrogen [H-C=O] (δ_H 8.4–12 ppm), using the ACD Spectrus Processor (Version 2013, Advanced Chemistry Development, Inc., Toronto, ON, Canada).

2.7. PM extraction, dispersion preparation, colloidal characterization and dosimetry modeling for bioactivity studies

The collected PM_0.1 samples were extracted into deionized endotoxin-free water using the sampling, extraction, dispersion, and dosimetry (SEDD) methodology previously developed by authors.⁵¹ PM_0.1 stock suspensions in water were prepared at a concentration of 0.5 mg/mL and sonicated to deagglomerate particles and prepare stable particle dispersions, per previous protocols.^52,53 To assess potential contribution of water-soluble substances from PM_0.1 in driving bioactivity, filtered suspensions were prepared via diafiltration of PM_0.1 stock suspensions (14,000 × g for 30 min; Amicon Ultra-0.5 Centrifugal Filter Unit, Nominal molecular weight limit = 3 kDa, MilliporeSigma, Burlington, MA). PM_0.1 stock suspensions and their corresponding filtrates were diluted to 100 μg/mL in Dulbecco’s Modified Eagle Medium (DMEM) (Sigma, Waltham, MA) supplemented with 10% Fetal Bovine Serum (FBS) (Corning, Woodland, CA) and vortexed for 30 s before being used for bioactivity assessment. PM_0.1 dispersions in water and culture medium were characterized for colloidal properties using dynamic light scattering (DLS, Malvern Zetasizer Nano ZS, Malvern Panalytical Inc., Westborough, MA), which provides intensity-averaged hydrodynamic diameter, zeta potential, polydispersity index, and specific conductance of the particles in suspension. To estimate the dose of particles delivered to cells during the in vitro exposure period, measurements of effective density of PM_0.1 in the culture medium were also performed, followed by running a computational simulation of the dosimetry, using previously validated and published protocols.^53,54

2.8. Cell culture

HEK293T cells were purchased from American Type Culture Collection and seeded onto 24-well plates at approximately 50,000 cells per well in 0.5 mL DMEM medium containing 10% FBS and 1% penicillin streptomycin (Corning, Manassas, VA). After 24 h, HEK-293T cells were transfected with a pGL3 plasmid construct containing the SPDEF promoter driving a luciferase reporter gene that was generated as part of a previous study.²⁰ The rationale for selecting this cell line and bioassay is described in the following section.

2.9. SPDEF promoter activation assay

Because the number of wood/combustion/dose conditions to be tested in our study were quite many, we used our previously validated rapid assay system to screen the different wood smoke PM_0.1 for mucin-inducing bioactivity.²¹ SPDEF is a transcription factor regulated by p53 that is sufficient and necessary for development of mucous cell phenotype.^55,56 Therefore, a luciferase reporter construct that is driven by SPDEF gene was used to assess whether genes responsible for mucous differentiation were expressed. Our previous studies have shown that this rapid screening bioassay is highly reliable in replicating the mucin gene expression response observed in differentiated human airway epithelial cells and that the doses used in this assay are also effective in primary human airway cells.^20,21 Although differentiated human airway cultures are a more realistic in vitro model for assessing the toxicology of aerosols, they are not practical for the screening of many samples/conditions because of the long time (~3 weeks) needed for differentiation. Furthermore, when human bronchial epithelial cells are not differentiated on Transwell membranes and grown on air-liquid interface cultures, they do not show mucin gene expression in a reliable manner, thus necessitating the use of alternative in vitro assays for the biological screening of PM. For this assay, the SPDEF promoter construct driving the firefly reporter gene, along with a Renilla-firefly gene-expressing construct as an internal standard to normalize for transfection efficiency, were transfected into HEK-293T cells by mixing with TransIT®-2020 Reagent (Mirus, Carlsbad, CA, USA), and gently adding the DNA complex to culture media in different areas of the well. After 24 h, cells were treated with PM_0.1 dispersions in culture medium and luciferase activity was measured 24 h later. Cells were treated with 10, 100 and 1000 ng/mL concentrations of administered PM_0.1 and their corresponding filtrates to assess dose-response relationships. The rationale for the chosen in vitro administered doses was based on modeling the deposited dose of particles in a human lung exposed to wood smoke from residential wood combustion at a real-world PM concentration of 64 μg/m³ and median particle diameter of 100 nm,^57,58 using the Multiple-Path Particle Dosimetry (MPPD) model.^59–61 Detailed calculations to justify the environmental relevance of the in vitro doses are presented in the SI. Thereafter, enzyme activities for firefly and Renilla luciferase were determined in the same cell lysate using the Dual-Luciferase reporter assay system. After 18 h, cells were washed once with phosphate buffered saline and lysed by rocking for 15 minutes with passive lysis buffer (Promega, Madison, WI). Cell lysates (20 μL) were transferred in 96-well plates and SPDEF promoter luciferase activity measured one minute after addition of 100 μL/well of Luciferase Assay Reagent II (LAR II) and adding Stop & Glo® Reagent using Fluoroskan Ascent (Thermo Fisher Scientific, Vantaa, Finland). Quantitation of luminescent signal was done with SpectraMax iD5 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA) as described previously.²¹

2.10. Statistical analysis

Woodburning experiments were performed in triplicates for each combination of wood species and combustion condition. Real-time aerosol and off-gas data were averaged across replicate experiments at respective timepoints during combustion and plotted with error bars representing the sample standard deviation across replicates, normalizing for 50 mg of starting amount of wood. For aerosol size statistics, the aerosol number-size distributions corresponding to timepoints of maximum particle number concentration in the SMPS and APS were chosen and analyzed in Aerosol Instrument Manager (AIM v9.0, TSI Inc., Shoreview, MN). The PM mass concentrations as a function of aerodynamic size were also averaged from gravimetric analyses across the replicate experiments. The presented chemical characterization data represent averages of analytical replicate measurements (n=3) from EC-OC, ICP-MS and NMR analysis performed on a single PM_0.1 sample (filter or suspension). The biological experiments were done using PM extracts from at least two replicate combustion experiments. For the PM from each replicate experiment, the biological experiments were performed on at least two individual wells for each dose and repeated three times. Grouped biological results for each type of PM/dose were expressed as mean ± standard error of the mean (SEM). Differences among the PM groups were examined by one-way analysis of variance and t-tests using the Prism statistical analysis software (GraphPad Software, San Diego, CA), and by application of Tukey’s test for multiple comparisons. Differences were considered significant at p<0.05.

3. RESULTS

3.1. Physicochemical characterization of hardwoods

The moisture contents of the three procured hardwoods were in a similar range between 8.6 – 11.3 % (Table S1), which is expected for seasoned firewood that should be below the recommended 20% moisture level for clean and high-heat burning.⁶²

Table 1 summarizes total quantitative inorganic elemental concentrations analyzed in different hardwoods, with concentrations of individual elements summarized in Table S2. The total inorganic elemental concentration in oak was 0.064% w/w, whereas that for cherry and mesquite was ~3-fold and ~12-fold higher at 0.203% w/w and 0.735% w/w respectively. The relative elemental abundances of the most dominant elements differed significantly between the hardwoods. While oak contained mostly K (68%), cherry was dominated by both K (39%) and Ca (34%), and mesquite was rich in Ca (60%), in addition to several other non-metals, metalloids, transition and rare-earth elements (Fig. S2).

Table 1.

Chemical composition of the hardwoods and emitted PM_0.1 under various combustion conditions. Data for hardwoods include total inorganic elemental weight concentrations measured in the wood pellets from ICP-MS analysis. Data for PM_0.1 include the total carbon (TC), organic carbon (OC) and elemental carbon (EC) contents measured in PM_0.1 from EC/OC analysis, as well as the total inorganic elemental weight concentrations measured in PM_0.1 along with total elemental enrichment factors (EF) with respect to corresponding wood type enclosed in parentheses (). Concentrations and EFs for individual elements in PM_0.1 are in SI.

	*TC (% w/w)	*OC (% w/w)	*EC (% w/w)	Total elements, % w/w (EF)
Oak	-	-	-	0.0641
Oak Flaming PM0.1	58.18 ± 2.97	58.18 ± 2.94	negligible	0.8416 (13)
Oak Smoldering PM0.1	55.80 ± 2.90	55.80 ± 2.85	negligible	1.255 (20)
Oak Incomplete Combustion PM0.1	65.25 ± 3.33	65.02 ± 3.28	0.2273 ± 0.165	1.597 (25)
Cherry	-	-	-	0.2028
Cherry Flaming PM0.1	61.00 ± 3.16	60.99 ± 3.11	negligible	0.8748 (4)
Mesquite	-	-	-	0.7351
Mesquite Flaming PM0.1	55.07 ± 3.35	55.07 ± 3.05	negligible	2.242 (3)

^*Analytical uncertainties are calculated from EC/OC measurement on a single quartz fiber filter that sampled PM_0.1.

Data not measured is indicated by “-”.

3.2. Gaseous emissions characterization

Effect of wood species:

Peak CO levels of ~500 – 700 ppm were observed during flaming combustion of different hardwoods when temperature was between 420 – 520°C (20 – 25 min), with the highest emissions observed for oak and the least for mesquite (Fig. S3 (A–C)). Peak concentrations of VOCs were reached at a similar time as CO and varied between ~4,000 – 5,000 ppb across hardwoods, with cherry and mesquite emitting the highest and lowest VOC levels respectively (Fig. S4 (A–C)).

Effect of combustion condition:

Marked differences in the maximum CO concentration were observed between various combustion conditions of oak. Highest CO level was observed for flaming (~700 ppm), followed by incomplete combustion (~55 ppm) and smoldering (~4 ppm) (Fig. S3 (A, D–E)). However, for the release of VOCs, incomplete combustion resulted in the highest VOC level (~8,000 ppb), followed by flaming (~5,000 ppb) and smoldering (~250 ppb) (Fig. S4 (A, D–E)). Collectively, these data indicated that oxygen enrichment during combustion (from incomplete to flaming combustion) decreased the production of VOCs while increasing the concentration of CO, which makes sense since the presence of oxygen enables more of the wood fuel to be oxidized to CO, whereas in the deficiency of oxygen, pyrolysis of wood occurs which produces a large concentration of unsaturated byproducts including VOCs, semi-volatile organics, and soot.⁶³

3.3. PM_0.1 emission characteristics

Effect of wood species:

PM emission during the flaming combustion of hardwoods was dominated by nanoparticles, with peak nanoparticle number concentrations ranging between 1 × 10⁷ and 3 × 10⁷ particles/cm³, with oak showing highest emission on average and mesquite the least (Fig. 1 (A–C)). Most nanoparticle emission occurred between 320 – 420°C (15 – 20 min). The geometric mean diameter (GMD) of the emitted nanoparticles was between 38 – 62 nm (Fig. 1 (D–F)). In comparison, PM number concentration in the submicron to micron size regime (0.5 – 20 μm) was much lower (Fig. S5 (A–F)). Oak and cherry emitted a peak of ~6,000 particles/cm³ with a GMD of 1.26 μm, whereas less than 1,000 particles/cm³ were observed for mesquite with a GMD of 0.92 μm.

Fig. 1.

Real-time monitoring of particle number concentration and size distribution of emitted nanoparticles during the combustion of oak (A, D), cherry (B, E), and mesquite (C, F) woods under flaming conditions using a Scanning Mobility Particle Sizer Spectrometer (SMPS) covering particle mobility diameter between 5 – 300 nm. Panels A-C display the total emitted nanoparticle number concentration (PNC) as a function of time (or temperature), whereas panels D-F display the nanoparticle number-size distributions captured at the timepoint of peak nanoparticle emission, for the respective woods.

The emitted PM mass concentrations as a function of aerodynamic size were also dominated by nano-sized fraction (PM_0.1), which accounted for ~36 – 46% of total PM mass, across the hardwoods. Cherry emitted the highest total PM mass concentration (20.1 mg/m³), followed by oak (18.4 mg/m³) and mesquite (8.2 mg/m³) (Fig. S6 (A–C)).

Effect of combustion condition:

Incomplete combustion of oak resulted in the highest emission of nanoparticles (Fig. S7 (A–B), Fig. 1A) as well as submicron to micron-sized particles (Fig. S8 (A–B), Fig. S5A), whereas smoldering produced the lowest particle number concentrations. As for nanoparticles, incomplete combustion generated a maximum of ~9 × 10⁷ particles/cm³, followed by flaming at ~3 × 10⁷ particles/cm³ and smoldering at ~2 × 10⁶ particles/cm³. The GMD of emitted nanoparticles was between 32 – 54 nm (Fig. S7 (C–D), Fig. 1D). For particles between 0.5 – 20 μm, incomplete combustion produced ~10,000 particles/cm³, followed by flaming at ~6,000 particles/cm³ and smoldering at ~2,000 particles/cm³, with similar GMDs between 1.26 – 1.39 μm (Fig. S8 (C–D), Fig. S5D).

In agreement with the particle number concentrations, incomplete combustion was also characterized by the highest total PM mass concentration (22.1 mg/m³), followed by flaming (18.4 mg/m³) and smoldering (1.7 mg/m³), with PM_0.1 constituting the bulk of the total PM mass (~45 – 67%) for all combustion conditions (Fig. S6 (A, D–E)).

3.4. Chemical composition of PM_0.1

Effect of wood species:

Flaming PM_0.1 from the hardwoods consisted of mostly OC, with slightly higher OC levels for cherry (61.0% w/w), followed by oak (58.2% w/w) and mesquite (55.1% w/w) and negligible EC levels for all (Table 1). The total mass concentration of inorganic elements in PM_0.1 was the highest for mesquite (2.2% w/w), followed by similar levels for oak (0.84% w/w) and cherry (0.88% w/w) (Table 1). The most common elements present in wood smoke, i.e., Ca, K, Na and S, constituted 81 – 96% w/w of all inorganic elements in PM_0.1 across the hardwoods, however, relative abundances of elements (% w/w of total elements) differed significantly between the hardwoods (Fig. S9 (A–C)). While both Ca (~30%) and S (~28%) dominated in oak PM_0.1, S alone was the major element for both cherry (~59%) and mesquite (~57%). Besides these dominant elements, oak PM_0.1 had the most abundance among the hardwoods of Al (6.6%), Mg (5.3%) and B (1.6%), whereas Fe (3.2%) was the most abundant in cherry PM_0.1. The concentration enrichment factors of inorganic elements in PM_0.1 with respect to the hardwood also differed significantly between the hardwoods (Table S2). Elements with the highest enrichment factors from oak included Ag, Al, B, Ca, Ce, Cs, Cu, La, Fe, Mg, Mn, Mo, Na, P, Rb, Sr, Ti and Tl, while those for cherry were Cr and S, and for mesquite were Cd, Co, Eu, Li, Ni, Pb, Sb, Sc, Se, Sn, W and Zr. As for organic composition, the flaming PM_0.1 from oak contained the highest total non-exchangeable organic H content (~2938 μmol H / mg PM_0.1), followed by mesquite (~60 μmol H / mg PM_0.1) and cherry (~18 μmol H / mg PM_0.1) (Table S3). Functional groups of lactate, acetate, succinate, and formate were identified in the NMR spectra for all hardwoods under flaming condition, in addition to the most common wood smoke tracer, levoglucosan (Fig. S10). Flaming PM_0.1 from all hardwoods were mostly composed of H-C-O, followed by H-R, H-C-C=, O-CH-O & H-C=, H-Ar, and H-C=O functionalities (Fig. 2 (A–C)). H-C-O was the most abundant in oak PM_0.1 (~71%) followed closely by mesquite (~70%) and the least in cherry (~62%). Abundance levels of H-R, H-C-C= and H-C=O were similar between the hardwoods at relative fractions ranging between 11.5-16.2%, 7.1-10.6% and 0.3-0.4%, respectively. However, cherry PM_0.1 was much more abundant in O-CH-O & H-C= (7.2%) and H-Ar (6.6%) functional groups compared to oak (3.1%, 2.4%, respectively) or mesquite (4.0%, 3.7%, respectively).

Fig. 2.

Organic functional group characterization of sampled wood smoke PM_0.1 particles generated from (A) oak wood flaming combustion, (B) cherry wood flaming combustion, (C) mesquite wood flaming combustion, (D) oak wood smoldering combustion, and (E) oak wood incomplete combustion. Pie charts show the relative abundances of the protons (% by mol of total non-exchangeable functional organic hydrogen) associated with the respective functional groups characterized by 1-D ¹H-NMR spectroscopy. The following functional groups of non-exchangeable organic hydrogen were quantified: (i) saturated aliphatic hydrogen [H-R]; (ii) allylic hydrogen [H-C-C=]; (iii) saturated oxygenated hydrogen, including α-hydrogen to hydroxyl, ether and ester group [H-C-O]; (iv) acetalic and vinylic hydrogen ([O-CH-O]+[H-C=]); (v) aryl hydrogen [H-Ar]; and (vi) carbonyl hydrogen [H-C=O].

Effect of combustion condition:

Incomplete combustion produced the highest concentration of OC (65.0% w/w) in oak PM_0.1, followed by flaming (58.2% w/w) and smoldering (55.8% w/w) (Table 1). Extremely low EC concentration was recorded for incomplete combustion PM_0.1 (0.23% w/w), whereas EC levels were negligible for flaming and smoldering. Incomplete combustion PM_0.1 from oak also contained the highest inorganic elemental concentration of 1.6% w/w, followed by smoldering (1.26% w/w) and flaming PM_0.1 (0.84% w/w) (Table 1). Prominent wood smoke elements, i.e., Ca, K, Na and S, accounted for 81 – 94% w/w of total elemental content of PM_0.1 across combustion scenarios (Fig. S9 (A, D–E)). While oak flaming PM_0.1 was equally abundant in Ca (~30%) and S (~28%), smoldering dominated in Ca (~66%) and incomplete combustion in S (64%). Al (6.6%), Mg (5.3%) and B (1.6%) were more abundant in flaming PM_0.1, whereas Fe (3%) was more prevalent in incomplete combustion PM_0.1. In terms of concentration enrichment in PM_0.1 with respect to pristine oak, elements preferentially enriched during incomplete combustion included Cd, Cr, Cs, Eu, Fe, Li, Mo, Na, S, Sc, Se, Sn, W and Zr, whereas those enriched during smoldering were Ba, Ca, Co, Cu, Mn, P, Pb, Sb and Tl, and during flaming were Ag, Al, B, Ce, La, Mg, Rb, Sr and Ti (Table S2).

In terms of characterization of organic functional groups, the total non-exchangeable organic H concentration was the highest for flaming PM_0.1 (~2938 μmol H / mg PM_0.1), followed by smoldering (~44 μmol H / mg PM_0.1) and incomplete combustion (~30 μmol H / mg PM_0.1) (Table S3). The NMR spectra showed a relatively strong signal of succinate and acetate groups for flaming condition, whereas smoldering and incomplete combustion had a much larger peak for levoglucosan (Fig. S11). All the combustion conditions showed a dominance in the H-C-O region, although oak flaming PM_0.1 had the highest abundance at 71.0%, followed by smoldering and incomplete combustion having a similar H-C-O abundance of ~61% (Fig. 2(A, D–E). Relative abundances of H-R and O-CH-O + H-C= were similar between the combustion conditions at 11.7-17.8% and 2.9-4.6% respectively. However, incomplete combustion PM_0.1 contained a much higher relative fraction of unsaturated aliphatic compounds (H-C-C=, 18.4%) than either flaming (7.1%) or smoldering (8.5%). Both aromatic (H-Ar) and carbonyl (H-C=O) functional groups were significantly enhanced in smoldering PM_0.1 (8.3%, 1.5% respectively) compared to either incomplete combustion (4.0%, 0.6%, respectively) or flaming (2.4%, 0.3%, respectively).

3.5. Colloidal characterization of PM_0.1

The extraction efficiencies of the different PM_0.1 from the Teflon filters were 95% or higher (data not shown). Table S4 presents a detailed colloidal characterization of the various PM_0.1 utilized in cellular experiments, both for dispersions in water and in culture medium (at t = 0 and 24 h). Colloidal parameters of PM_0.1 in culture medium remained relatively stable over 24 h, indicating no significant agglomeration of particles during the in vitro exposure period. In summary, the average hydrodynamic diameter of particles in culture medium across the different conditions was 214 – 525 nm, polydispersity between 0.322 – 0.728 (indicating particles of a wide range of sizes), and a zeta potential between −11.7 and −15.6 mV (indicating fairly stabilized particles). The measured effective densities of PM_0.1 agglomerates in the culture medium were ~1.5 g/cm³ for all conditions (data not shown). In vitro cellular dosimetry simulation of the PM_0.1 particles performed using their measured colloidal parameters and agglomerate effective density yielded a deposited particle mass fraction of up to ~50% of the administered PM_0.1 after 24 h exposure across different conditions (data not presented).

3.6. Bioactivity of PM_0.1

Effect of wood species:

Flaming PM_0.1 from the hardwoods increased SPDEF promoter activity relative to untreated and blank filter controls in a dose-dependent manner, with significant 1.5-2-fold change observed at doses of 100 and 1000 ng/mL (Fig. 3 (A–C)). Cherry showed the most significant increase in bioactivity at 1000 ng/mL (2-fold, p<0.001) (Fig. 3B), followed by oak at 1000 ng/mL (~2-fold, p<0.01) (Fig. 3A) and the least for mesquite at 100 ng/mL (1.75-fold, p<0.05) (Fig. 3C). However, the filtrates of flaming PM_0.1 extracts (containing water-soluble compounds) from the hardwoods did not change SPDEF promoter activity relative to the controls at any dose.

Fig. 3. SPDEF promoter activities of wood smoke PM_0.1.

Wood smoke PM_0.1 size fraction and its dissolved component obtained by filtration of the PM_0.1 aqueous extract were tested at 10, 100 and 1000 ng/mL concentrations for SPDEF promoter luciferase activity in HEK-293 cells for (A) oak wood flaming combustion, (B) cherry wood flaming combustion, (C) mesquite wood flaming combustion, (D) oak wood smoldering combustion, and (E) oak wood incomplete combustion. SPDEF promoter construct in PGL3 basic vector was transfected into HEK-293 cells one day prior to treatment with PM_0.1 or vehicle control from blank PM substrates (PFU) or left untreated (NT). SPDEF promotor activity in the cell lysates was measured by luminometer. For each figure, the mean +/− SEM is graphed, with significant differences at *p<0.05, **p<0.01, and ***p<0.001 from NT control.

Effect of combustion condition:

While flaming PM_0.1 from oak significantly enhanced SPDEF promoter activity at 100 ng/mL (1.5-fold, p<0.05) and 1000 ng/mL (~2-fold, p<0.01), smoldering and incomplete combustion PM_0.1 did not a show a statistically significant increase in bioactivity at any dose compared to the controls (Fig. 3(D–E)). However, the filtrate of smoldering PM_0.1 at 10 ng/mL (but not at the other doses) significantly increased bioactivity by up to 1.2-fold (p<0.01) (Fig. 3D), while the filtrates of incomplete combustion PM_0.1 showed no significant change in promoter activity at any dose compared to the controls.

4. DISCUSSION

Our study demonstrates the key role of hardwood species and combustion scenarios on the physicochemical and toxicological properties of wood smoke particles. The effect of wood type for a given combustion scenario on the physicochemical properties of emissions was only minor, at least for the investigated hardwoods. The PM_0.1 from the flaming combustion of oak, cherry and mesquite were similar in functional composition of organic compounds and significantly upregulated the activity of the SPDEF promoter. However, the type of burning condition strongly influenced the emissions of off-gases (i.e., CO and VOC) and PM in addition to the chemical composition of PM_0.1. Smoldering and incomplete combustion resulted in significant shifts in the relative contributions of certain organic functionalities in the PM_0.1 relative to flaming and did not induce a significant biological response of the SPDEF promoter.

In the following sections, we attempt to explain our in vitro findings in light of the chemical characterization data obtained across the different PM_0.1 and the recent toxicological studies performed by authors and elsewhere on wood smoke aerosols:

4.1. Combustion scenario strongly influences PM_0.1 chemical composition and activity of SPDEF promoter

Similar to findings in this study for oak, our previous studies on pine wood also showed that flaming PM_0.1 was the most potent in increasing the activity of the SPDEF promoter, while PM_0.1 from smoldering and incomplete combustion did not induce any bioactivity.^21,28 The absence of bioactivity of wood smoke from smoldering or incomplete combustion is probably driven by differences in the chemical composition of PM_0.1 between these scenarios and flaming combustion.

Examining the inorganic elemental composition of PM_0.1, smoldering and incomplete combustion enriched the total metallic content in PM_0.1 relative to flaming for both oak (Table 1) and pine,²⁸ suggesting that the increased metal content of PM_0.1 could play a role in suppressing the SPDEF promoter activity. However, this seems unlikely since significant bioactivity was observed for mesquite flaming PM_0.1, which had the highest metallic content among all hardwoods/combustion conditions (2.2% w/w, Table 1). Hence, it appears that differences in the composition of organic compounds between the combustion conditions, which constitute a majority of the PM_0.1 mass (Table 1), are more likely to contribute to observed differences in biological effects.

In support of this hypothesis, NMR data showed that the non-exchangeable organic H concentration of each functional group in PM_0.1 was orders of magnitude higher in the flaming condition than in smoldering or incomplete combustion, for both oak (Table S3) and pine.²⁸ However, the total organic H content of PM_0.1 alone is not able to explain its bioactivity since the total organic H contents of cherry and mesquite flaming PM_0.1 (which were bioactive like oak) were at par with those of smoldering and incomplete combustion PM_0.1 from oak (which were non-bioactive) (Table S3). Thus, it is possible that the observed differences in the relative abundances of organic functional groups in PM_0.1 between the combustion conditions could be driving differences in biological responses, due to changing relative proportions of agonist and antagonist compounds that could activate and inhibit SPDEF promoter activity, respectively. In this study (Fig. 2) and in our previous study on pine,²⁸ PM_0.1 from flaming conditions were dominant in saturated oxygenated aliphatic compounds (H-C-O) and were also bioactive, suggesting that compounds responsible for bioactivity likely contain this functional group. This is supported by our recent investigation that identified various sugars such as levulinate, oxalate and xylitol as active fractions of pine smoke that induced SPDEF and mucin gene (MUC5AC) expression in primary human airway epithelial cells and in mice.²¹ Similarly, a recent study identified the organic chemical, coniferaldehyde (that contains an H-C-O group), in pine smoke that is responsible for stimulation of mucin gene expression in human bronchial cells and in mice.⁶⁴ Although smoldering and incomplete combustion PM_0.1 from oak also contained a major abundance of H-C-O (Fig. 2) albeit lower than in flaming, it is possible that specific organic compounds responsible for bioactivity were no longer present in the PM_0.1 or were at very low concentrations compared to flaming, or the concentrations of inhibitor compounds were elevated relative to the concentrations of agonist compounds during smoldering and incomplete combustion. We indeed found that smoldering and incomplete combustion increased the relative abundances of certain functional groups such as unsaturated allylic chains (H-C-C=), aromatics (H-Ar) and aldehydes (H-C=O) relative to oak flaming PM_0.1 (Fig. 2), that could be exerting inhibitory effects on the SPDEF promoter activity as opposed to the agonist effect of H-C-O, although this hypothesis needs to be investigated in future studies. In a similar pattern, smoldering and incomplete combustion PM_0.1 from pine in our previous study²⁸ had also shown increased relative fractions of H-C-C= and H-Ar compared to flaming, in addition to elevated abundance of H-R, further suggesting probable antagonist behavior of such functional groups on the SPDEF promoter activity.

Previous toxicological studies on inhalable wood smoke particles have also shown marked differences in biological outcomes between various burning conditions for a given wood type.^65–70 However, no trend is apparent on the relative toxicological potential of flaming, smoldering and incomplete combustion since the defining operational parameters of these combustion conditions are inconsistent across studies. Furthermore, toxicity assessment for a given burning scenario and wood type can be different across studies depending on the biological exposure model employed (in vitro or in vivo) and the target endpoint. Several studies have however linked the organic composition of wood smoke particles, especially their content of polycyclic aromatic hydrocarbons (PAH), to various inflammatory, cytotoxic, and genotoxic endpoints.^{22,24–26,71–75} We did not characterize PAH content of PM_0.1 in this study, rather only the content of aromatic fragments (H-Ar) using NMR, which did not seem to correlate with bioactivity since the PM_0.1 with the highest H-Ar abundance (i.e., smoldering of oak, Fig. 2D) did not exhibit any bioactivity.

4.2. Composition of whole PM_0.1 determines bioactivity, rather than water-soluble components alone

It is worth noting that the bioactivity of PM_0.1 under flaming conditions from all hardwoods disappeared when only the water-soluble fractions of PM_0.1 were assessed, showing that the bioactive compounds were strongly associated with the particulate phase and did not readily dissolve in water (Fig. 3). Consistent with this observation, the water-soluble extract of pine flaming PM_0.1 had also shown significantly diminished bioactivity compared to the whole PM_0.1,²⁸ indicating that the PM-bound relatively insoluble organic compounds were biologically more potent than the water-soluble organic substances, at least for the bioassay employed here. However, for the smoldering of oak, the opposite was observed, where the whole PM_0.1 did not show bioactivity while its filtrate at 10 ng/mL significantly elevated SPDEF promoter activity (Fig. 3D). One reasonable explanation for this could be that the filtration of smoldering PM_0.1 extract could have removed the relatively insoluble non-polar compounds such as those containing H-R, H-C-C= and H-Ar that we hypothesized in the previous section as probable inhibitors of bioactivity, thus leading to a slight excess of polar compounds such as those containing H-C-O (probable agonists) in the soluble fraction, which could have caused a small but significant increase in the SPDEF promoter activity. This is further corroborated by our recent investigation of pine smoke,²⁸ where flaming PM_0.1 from the higher-moisture pine showed a significant increase in SPDEF promoter activity when it was filtered compared to the whole PM_0.1, indicating that the insoluble and probably antagonist functional groups (H-R, H-C-C= and H-Ar) were removed during the filtration of PM_0.1.

Thus, from our bioassay results, it is clear that the whole particulate phase of the wood smoke aerosol governs its biological activity rather than its water-soluble organic substances only and that the bioactive compounds are found both in the dissolved and insoluble phase of the PM extract. This finding has important dosimetry implications when evaluating the toxicological effects of wood smoke nanoparticles in an in vitro exposure model, since nanoparticles do not deposit immediately on cells because of the greater effect of Brownian motion diffusion compared to gravitational sedimentation and the low effective density of organic nanoparticle agglomerates in culture medium, unlike soluble compounds which immediately become bioavailable to the cells.⁷⁶ Evaluation of the deposited dose of nanoparticles over the course of the in vitro exposure period is thus critical to accurately characterize dose-response relationships based on actual deposited particle doses. For future in vitro toxicological studies, target cellular exposure metrics (e.g., PM mass per unit surface area) must be adjusted for the deposition mass fraction in order to generate reliable and reproducible dose-response curves across different woods/combustion conditions, as PM_0.1 generated under different conditions may have different settling kinetics in vitro. In our study, since the deposition mass fraction was ~50% for all PM_0.1, comparisons of biological responses could be done reliably between the different PM_0.1 as a result of similar delivered-to-cell doses.

4.3. Comparison of bioactivity between hardwood and softwood smoke PM_0.1

The type of wood used as firewood could either be a hardwood or a softwood depending on the intended application and the local supply availability of firewood species. Hardwoods are much denser compared to softwoods and thus have a higher heating value compared to softwoods per unit volume.⁷⁷ While hardwoods are good for indoor fireplaces, woodburning stoves and grilling due to their relatively slow, hot and clean burning behavior as well as a stronger flavor profile, softwoods burn faster and give off much more smoke due to their high sap and resin content and thus serve well as fire starters and for outdoor campfires.^42,78 As regards bioactivity, the PM_0.1 from the flaming combustion of the hardwoods investigated in this study and from pine (a softwood) in our previous study²⁸ consistently and significantly increased SPDEF promoter activities, although effects were much stronger for pine (2-5-fold change) than any of the hardwoods (1.5-2-fold change) (Fig. 3). Similarly, in a recent study,⁶⁴ pine smoke particles were a more potent inducer of mucin gene expression than mesquite. This is despite the fact that hardwoods exhibited a significantly higher abundance of H-C-O in the flaming PM_0.1 (62 – 71%) (Fig. 2) compared to pine (~54%),²⁸ which we have hypothesized as the bioactive functional group. However, it is possible that the particular species of agonist organic compounds present in pine smoke belonging to the H-C-O group may be more biologically potent than those present in the hardwood smoke, or it could be that the hardwood smoke may contain a greater content of inhibitory compounds that are in low concentrations or absent in pine smoke. For example, hardwood flaming PM_0.1 contained higher relative fractions than pine²⁸ of H-C-C= (7.1-10.6% in hardwoods vs. 4.9% in pine), H-Ar (2.4-6.6% vs. 1.4%), and H-C=O (0.3-0.4% vs. 0%) (Fig. 2), which we have hypothesized as probably inhibitory functional groups in light of our findings. Additionally, it is known that hardwoods and softwoods have different emission profiles of certain classes of organic compounds such as substituted phenols, resin acids, diterpenoids, guaiacols and syringols,^11,79–81 and different relative ratios of sugar anhydrides levoglucosan, mannose and galactosan,⁸² which could have contributed to significant differences in the PM_0.1 biological activity. Hence, it is reasonable to say that the NMR spectral information alone on the relative dominance of possibly agonist and antagonist functional groups is not sufficient to predict and compare the net biological activity of PM across the different woods/combustion conditions. Detailed organic speciation of the target functional groups in PM is needed to identify and quantify individual organic compounds that can then be tested individually or in various combinations for assessment of biological potency in a particular bioassay.

4.4. Conclusions and future research directions

The novelty of our study lies in being able to systematically investigate the role of wood species and burning conditions on potential environmental health implications of inhalable wood smoke particles using an integrated platform with well-controlled parameters of wood combustion. By comprehensively characterizing the physicochemical properties of generated PM using state-of-the-art instrumentation and analytical methods, we attempted to link the PM emission profile to the biological activity of an essential transcription factor involved in the activation of mucus production, which is associated with adverse pulmonary outcomes. Our study provides crucial insight into the wood combustion parameters that strongly influence the organic chemistry of emitted PM_0.1 which then drives its bioactivity. It is worth noting though that the present study characterized the physicochemical properties of freshly generated and sampled PM_0.1 with limited dilution in the air, which is more relevant for assessing exposure profiles in the vicinity of the woodburning source (e.g., indoor wood stoves and fireplaces), rather than atmospherically diluted emissions (e.g., forest fires). Low dilution affects the gas-particle partitioning of the VOCs and semi-volatile organics as the gas phase tends to saturate,⁸³ potentially resulting in the gaseous organics being sampled as particulates that would otherwise not be a part of the atmospheric PM.⁸⁴ Furthermore, the chemical composition of wood burning emissions changes with multiphase physical and chemical processes in the atmosphere (i.e., aging),^85,86 leading to the formation of secondary organic aerosols which could have a significantly different toxicological footprint than pristine emissions.⁸⁷ Thus, future studies should take into consideration such aging transformations in order to better characterize population health risks from emissions traversing in the environment over long distances and time periods such as those from wildfires and controlled agricultural burns.

In terms of limitations of our study, we did not identify specific agonist or antagonist organic compounds in PM_0.1, which would have enabled a deeper understanding of individual chemical pollutants that induce such mucus-related pathways leading to potential development of pulmonary disease. Organic speciation of representative PM from different wood combustion atmospheres using liquid chromatography/gas chromatography–mass spectrometry (LC/GC-MS) approaches is necessary to document concentrations and potencies of individual organic compounds that drive or inhibit activity of mucin-related biomarkers and to predict potential pulmonary health effects from a given PM composition. In addition, bioassays for other adverse pulmonary outcomes should be employed to determine potential health effects of a myriad of other wood smoke components. It is worth noting that the use of submerged cell culture systems for understanding of toxicological effects of aerosols has limitations both in terms of dosimetry and relevance of exposure.^88,89 Thus, for future studies, our versatile exposure generation platform could be coupled with more realistic exposure techniques such as differentiated human airway cultures at the air-liquid interface and whole-body animal inhalation chambers, that allow for a more holistic and accurate toxicological assessment of combustion aerosols.^71,90–92 Such extensive toxicological data derived from well-controlled and realistic human exposure conditions will enable risk assessors and regulators to develop strategies to manage population health risks of wood smoke inhalation, such as disseminating specific recommendations on the choice of firewood and the safe operation of woodburning appliances.

Environmental Implication.

Wood burning releases a complex mixture of hundreds of semi-volatile/volatile organic compounds and has both acute and chronic effects on pulmonary health. The WHO estimates nearly 4 million people die prematurely each year from diseases caused by biomass burning. This work addresses a critical knowledge gap on the link between woodburning parameters (wood type and combustion condition) and human health implications by examining the effect of wood smoke on a biomarker of mucin gene expression implicated in mucus-related respiratory diseases. Such data will enable risk assessors to provide operational recommendations on domestic wood combustion to protect health of vulnerable populations.