Arachidonic Acid Metabolites in Self-collected Biospecimens Following Campfire Exposure: Exploring Non-invasive Biomarkers of Wildfire Health Effects

Abstract

Climate change has contributed to increased frequency and intensity of wildfire. Studying its acute effects is limited due to unpredictable nature of wildfire occurrence, which necessitates readily deployable techniques to collect biospecimens. To identify biomarkers of wildfire’s acute effects, we conducted this exploratory study in eight healthy campers (four men and four women) who self-collected nasal fluid, urine, saliva, and skin wipes at different time points before, during, and after 4-hour exposure to wood smoke in a camping event. Concentrations of black carbon in the air and polycyclic aromatic hydrocarbons in participants’ silicone wristbands were significantly elevated during the exposure session. Among 30 arachidonic acid metabolites measured, lipoxygenase metabolites were more abundant in nasal fluid and saliva, whereas cyclooxygenase and non-enzymatic metabolites were more abundant in urine. We observed drastic increases, at 8 hours following the exposure, in urinary levels of PGE2 (398%) and 15-keto-PGF2α (191%) (FDR<10%), with greater increases in men (FDR < 0.01%) than in women. No significant changes were observed for other metabolites in urine or the other biospecimens. Our results suggest urinary PGE2 and 15-keto-PGF2α as promising biomarkers reflecting pathophysiologic (likely sex-dependent) changes induced by short-term exposure to wildfire.

Graphical Abstract

INTRODUCTION

Accelerated by climate change, nowadays wildfires with adverse impacts on regional air quality have become a near-annual occurrence in North America.¹ In the United States, wildfires have become the single largest source of fine particle matters (PM_2.5) accounting for up to 50% of PM_2.5 emissions in Western regions.² Epidemiological studies has associated wildfire episodes with respiratory and cardiovascular illnesses,^3–5 which is mechanistically supported by animal and human studies with controlled exposure to wood smoke.^6–8 However, in real-world settings there is a lack of sensitive biomarkers to detect early pathophysiologic responses to wildfire exposure, hampering timely assessment of the resulting health risk.

Inflammation plays central roles in air pollution’s adverse cardiopulmonary effects.^9,10 We recently found that circulating arachidonic acid (ARA) metabolites, on the causal pathways to pro-inflammatory and pro-oxidative effects,^11,12 were more sensitive than traditional inflammatory and oxidative stress biomarkers in response to traffic-originated air pollution exposure.¹² However, conventional methods to collect venous blood requires a certified phlebotomist, which is often not feasible in an attempt to capture the acute impact of a wildfire that can be hardly predicted for its occurrence. Examining the short-term effects of wildfire requires biomarkers that can be obtained using simpler and timely methods. Self-collected biospecimens, which have become increasingly acceptable since the COVID-19 pandemic,¹³ may offer an opportunity to assess wildfire exposure and health effects.

We conducted an exploratory study in healthy campers who self-collected different types of non-invasive biospecimens before, during, and after 4-hour exposure to wood smoke during a camping event. Wood smoke exposure was verified by monitoring black carbon in the ambient air and polycyclic aromatic hydrocarbons (PAHs) in silicone wristbands. We measured 30 ARA metabolites in different biospecimens to identify sensitive biomarkers that can be potentially used to detect pathophysiologic response to wood smoke exposure.

METHODS AND MATERIALS

Study Design and Participants.

Study participants were recruited from students or staff of Duke University who were (a) non-smokers; (b) at least18 years old; (c) free of cardiopulmonary diseases during the previous 6 months; and (d) agreeing to be around a campfire for four hours in a camping trip. Written informed consent was obtained from all study participants. The study protocol was approved by the Institutional Review Board of Duke University.

Each participant was instructed to sit near the campfire for 4 hours during the camping trips designed for this study. The campfire was primarily fueled by a mixed hardwood firewood bundle purchased at local grocery stores, with an addition of dead leaves and branches collected near the camping site to mimic the combustion condition of a wildfire. Each participant wore a micro-aethalometer (AE51, AethLabs, CA, USA) for real-time monitoring of black carbon during the exposure session as well as ~30 minutes before and after the session (background levels). Two micro-aethalometers used in this study were synchronized before the field measurements, and the readings show good agreement (R²=0.96, Figure S1). Participants also wore a precleaned silicone wristband (24hourwristbands.com, Houston, TX, USA) for 4 hours before, during, and after the exposure session. Methods used for cleaning and storing the wristband prior to deployment can be found in a prior publication.¹⁴ Participants were advised not to eat any barbeque or fried food within two days before and after the camping event. Our research staff accompanied all the study participants during the 4-hour exposure session to aid in exposure assessment and biospecimen collection if needed.

Self-Collection of Biospecimens.

Four types of biospecimens (i.e., nasal fluid, urine, saliva, and skin wipe) were collected at different time points before, during, and after the exposure to campfire (Figure S2). Before the camping event, participants were provided with detailed instructions on biospecimen collection procedures and a sampling kit consisting of biospecimen collection materials and a handout outlining biospecimen collection timeline and procedures. The biospecimen self-collection was supervised by our research staff at baseline and during the campfire exposure, while the collection at other time points was unsupervised.

Nasal fluid was collected using a mixed cellulose ester sampling strip (HAWG 047S6, Millipore Sigma, MO, USA) as described previously.¹⁵ Saliva was collected using Oragene saliva kits as described previously.¹⁵ Skin wipe was collected using pre-cleaned sterilized cotton pads wetted with ethanol as described previously.¹⁶ An area of 100 cm² on participant’s right forearm was sampled before exposure while the corresponding area on left forearm was sampled after the exposure. All biospecimens were stored in a cooler filled with ice upon the collection and transported to the laboratory. Nasal strip and skin wipe were extract with 600 μL deionized water and 1 mL methanol, respectively. The extract of nasal strip and skin wipe, and saliva and urine samples were stored at −80 Celsius until laboratory analysis.

Laboratory Analysis.

We used a targeted lipidomic method to measure the concentrations of 30 ARA metabolites (Figure S3), and eicosapentaenoic and docosahexaenoic acids.¹⁷ Briefly, nasal fluid, saliva, or skin wipe extract were mixed with ice-cold methanol and ascorbic acid for protein precipitation. Urine was incubated with β-glucuronidase-arylsulfatase and ascorbic acid for 16 hours at 37 Celsius to hydrolyze conjugated metabolites. Then we spiked all samples with a cocktail of internal standards and acetate buffer (pH=5.5). The mixture was loaded to a preconditioned Bond Elute Certify II cartridge and eluted with methanol containing 0.2% acetic acid. The eluent was evaporated under nitrogen gas to dryness, concentrated with methanol, and injected into a HPLC-MS/MS (TSQ Quantum Access Max, Thermo Fisher Scientific, MA, USA). A Kinetex 2.3 μm C18 column (100 × 2.1 mm, Phenomenex, CA, USA) was used. The mobile phase and gradient program are shown in Table S1. The parameters used for the MS analysis and method performance for each analyte are shown in Table S2.

To account for a potential influence of circadian rhythms, we measured urinary concentrations of 6-sulfatoxymelatonin (aMT6s), a major metabolite of melatonin, using HPLC-MS/MS as described previously.¹⁸ We measured urinary protein levels using a Bradford assay (Thermo Fisher Scientific, MA, USA.) to screen for semen or blood contamination. Urinary specific gravity was measured with a refractometer for data normalization of urinary biomarkers. Urinary creatinine was also measured using a commercial kit (Cayman Chemical, MI, USA). PAHs were extracted from silicone wristband using organic solvents and measured using GC-MS with isotopically labeled internal standards as described previously.¹⁴

Statistical Analysis.

We conducted statistical analysis only for measurements with detection rates > 50%. Undetected analytes were assigned with a concentration of 50% of the limit of detection. Between-group differences were examined using the student’s t-test (two-independent samples or paired as appropriate). Correlations were examined using the Pearson correlation test. To examine biomarker changes after campfire exposure, we used mixed-effect models in which time of the measurements was categorical fixed effect and within-participant correlation was controlled using random intercepts. The model was adjusted for the ARA concentrations to account for the influence of ARA biological availability on the metabolite concentration. We explore the sex-difference in biomarker responses by introducing an interaction term between time and sex into the model. For each model, the residual was carefully examined for normal distribution and biomarker concentrations were logarithmic transformed as appropriate. We computed the confident intervals for the effect size without multiple comparison adjustment. Nevertheless, we used the Bonferroni method to adjust p-value to control the overall false discovery rate (FDR) for 29 biomarkers included in the analysis. For urinary ARA metabolites (n=22), we further conducted a nonnegative sparse principal component analysis of the z-score to identify key metabolic pathways. The temporal changes of component scores were examined using the mixed-effect model described above.

Three sensitivity analyses were conducted to evaluate the influence of factors other than campfire exposure on the ARA metabolites, including (1) adding urinary aMT6s levels as a covariate to account for the influence of circadian rhythms; (2) adding urinary protein concentrations as a covariate to control urine contamination by semen or blood; and (3) using creatinine instead of specific gravity to normalize the ARA metabolites. All statistical analysis was performed with the statistical software R with lme4, lmeTest, and nsprcomp packages (www.r-project.org).

RESULTS

Demographic Characteristics.

The average (standard deviation) age and body mass index (BMI) of the 8 healthy adults (four men and four women) were 34.8 (7.2) years and 23.3 (4.8) kg/m², respectively. There was no significant difference in age between men and women (p=0.62, Table S3). The BMI of women was significantly lower than men (p=0.027).

Exposure Levels.

The average black carbon concentrations ranged from 2.6 to 27.3 μg/m³ for eight participants during the exposure session and were markedly higher than the background levels (Figure 1). We detected nine PAHs (i.e., acenaphthylene, acenaphthene, anthracene, fluorene, phenanthrene, fluoranthene, pyrene, benzo [α]anthracene, and benzo [e]pyrene) in all wristbands while benzo[c]phenanthrene, benzo[α]pyrene, and chrysene were detected in only one wristband and therefore were excluded. Concentrations of nine PAHs during the exposure session were 1.6–11.3 folds higher than background levels (i.e., before + after), with greater differences observed for semi-volatile PAHs (Figure S4). The benzo[α]pyrene equivalent concentrations during the exposure session were 1.44-fold higher than background levels and positively correlated with the black carbon levels (Figure 1). There was no difference in black carbon or PAHs levels during the exposure session between the men and women (Table S3).

Figure 1. Comparison of black carbon (Panel A) and PAHs (Panel B) concentrations with and without campfire exposure, as well as the correlation between black carbon and PAHs levels during the 4-hour exposure session.

*: p<0.05 tested by paired t-test.

ARA Metabolites in Different Biospecimen.

We compared detection rates and profiles of ARA metabolites by biospecimen type (Figure S5). In nasal fluid, high detection rates and concentrations were observed for ARA and its lipoxygenase (ALOX) metabolites (e.g., 12- and 15-HETE). Likewise, detection rates and concentrations of 12-HETE and ARA were also high in saliva. In contrast, urine had high abundance of cyclooxygenase (COX)-mediated (e.g., PGF2α) and non-enzymatically (e.g., 8,12-iso-iPF2α-VI) oxidative ARA metabolites, while the detection rate for ARA was relatively low. We also detected several ARA metabolites derived from ALOX (e.g., LTB4) and cytochrome P450 (CYP, e.g., 8,9-DHET) in urine. In skin wipe, we observed low detection rates and levels of most ARA metabolites. Relatively higher detection rates and levels were observed for 12-HETE, 15-HETE, and ARA.

Effects of Campfire Exposure on ARA Metabolites.

We observed no significant changes in any ARA metabolite following campfire exposure at 5% FDR levels. Nevertheless, if we relax the FDR threshold from 5% to 10%, there were significant increases in urinary levels of PGE2 and 15-keto-PGF2α 8 hours after the cessation of exposure (Figure 2). The increases in PGE2 and 15-keto-PGF2α were positively correlated with exposure to black carbon and semi-volatile PAHs (Pearson r= 0.18 ~ 0.49, Figure S6), respectively, although the corrections were not statistically significant due to a small sample size. Among men, there were drastic increases in PGE2 (823%, 95%CI: 555% to 1092%) and 15-keto-PGF2α (392%, 95%CI: 256% to 528%) at 0.01% FDR levels. Likewise, the PGJ2 concentrations also increased by 164% (95%CI: 59% to 271%) among male participants with FDR <10% (Figure S7). The increase of these metabolites remains robust in three sensitivity analyses (Figures S8–S10). Among women, no changes in PGE2, 15-keto-PGF2α or PGJ2 levels were observed (Figures 2 and S6). No significant changes were observed for other metabolites in the urine (Figure S6) or other biospecimen (Figures S11 and S12).

Figure 2. Change of PGE2 (Panels A-C) and 15-keto-PGF2α (Panels E-G) levels in the urine before, during, and after 4-hour exposure to campfire.

*: FDR < 5% tested by mixed-effect models in which metabolites levels were modeled as functions of interactions between time and sex, ARA concentrations, and random intercept of participants.

We conducted nonnegative sparse principal component analyses of 22 urinary ARA metabolites to identify ARA metabolic pathways. Four principal components (PC) jointly explained 70.4% of the total variance (Figure S13). PC1 has moderate loadings for all ARA metabolites, which may reflect variations in ARA bioavailability and urinary hydration. PC2 had high loadings of PGE2, 15-keto-PGF2α, and PGJ2, likely representing the COX pathways. PC3 was correlated with all the four DHETs and may indicate the CYP pathways. PC4 had high loadings of 12-HETE and lipoxin A4, likely indicative of the ALOX-mediated inflammation resolution pathways. In consistency with the metabolite-specific analyses, the PC2 score was drastically increased 8 hours after the exposure in a sex-dependent manner. No significant temporal changes were observed for other PCs.

DISCUSSSION

By characterizing ARA metabolites in self-collected biospecimens after campfire exposure, this study aimed to identify non-invasive biomarkers to detect early health effects of short-term exposure to wildfire. Drastic increases in urinary PGE2, 15-keto-PGF2α, and PGJ2 levels were observed at 8 hours following the cessation of exposure, suggesting the promising use of urinary COX metabolites as sensitive biomarkers for wildfire studies. Furthermore, male and female participants presented differential changes in urinary COX metabolites, suggesting that the effects of wildfire exposure on ARA metabolism might be sex-dependent.

The profile of ARA metabolites in four types of self-collected biospecimen was markedly different. High abundance of 12- and 15-HETEs with low levels of 5-HETE were observed in nasal fluid, which is consistent with high ALOX15 and ALOX12 but low ALOX5 expression in airway epithelium.¹⁹ Likewise, high expression of ALOX12 in the esophagus and skin²⁰ explained the relatively higher levels of salivary 12-HETE and the increased likelihood of detecting 12-HETE in skin wipes. These results suggested that ARA metabolites in the nasal fluid, saliva, and skin wipes may reflect local eicosanoid biosynthesis. In contrast, urinary ARA metabolites reflect the total amount of eicosanoid excreted from the systemic circulation, including those originating from tissues, which has been reported previously.²¹ Given increased concentrations of COX instead of ALOX metabolites after campfire exposure, nasal or salivary ARA metabolites might be less sensitive to wildfire exposure due to the lack of COX activity in the nasal and oral cavities..

It is well documented that the up-regulation of COX pathway has been implicated in adverse cardiopulmonary effects after short-term exposure to environmental stimulus.^22–24 Recent studies in allergen-sensitized adults found increased urinary levels of COX metabolites within 24 hours after exposures to diesel exhaust and/or allergens, which mediated the allergen-induced airway impairment.²³ In our previous studies, short-term exposure to ozone and to nitrated-PAHs were independently associated with increased urinary levels of 11-dehydro-thromboxane B2 (11-dhTXB2), indicative of systemic pro-thrombotic and pro-oxidative changes.^25,26 In controlled exposure studies, heathy adults exposed to biodiesel exhaust for 1 hour exhibit increased levels of PGE2 in bronchoalveolar lavage and PGF2α in plasma 6 hours after the exposure.^27,28 Herein, we observed increased levels of COX metabolites in urine collected 8 hours after exposure to wood smoke, suggesting the possible relevance of this pathway to short-term wildfire exposure.

Despite in a small sample size, we observed that male participants had stronger COX metabolite responses to campfire exposure, which was among the first evidence suggesting sex-dependent effects of air pollution on ARA metabolism. Consistently, a recent randomized trial of 39 healthy young adults also found stronger inflammatory effects in nasal lavage of male participants following wood smoke exposure.²⁹ Previous animal studies have observed sex-dependent effects of ozone exposure on pulmonary eicosanoid biology.³⁰ Different from our findings, ozone induced greater productions of COX metabolites among female mice, together with increased ARA and DHA concentrations and their metabolites from ALOX pathways and unchanged expression of COX-2, ALOX5, and ALOX12/15.³⁰ These results suggested that the sex-dependent response may be caused by differences in ARA and DHA bioavailability. In contrast, we observed increased COX metabolites among male participants, without significant changes in other pathways or ARA bioavailability. Thus, multiple mechanisms may contribute to the sex-specific polluted-induced effects on ARA metabolism, which warrants further investigations.

This study has several limitations. First, the sample size was small, making the study exploratory in nature especially for sex-stratified analysis. Second, this single-arm study design cannot effectively control potential confounders. Although we have addressed the potential influence of circadian rhythms²⁴ and semen/blood contamination³¹ using urinary aMT6s and protein concentrations, we cannot exclude the possible effects of other factors on ARA metabolism. Third, this study only examined the effects of a 4-hour exposure and may not be extrapolated to longer-term exposure conditions. Previous animal model studies demonstrated different eicosanoid responses to longer term air pollution exposure which is driven by upregulated ALOX-pathways.^32,33 Our previous studies also found that exposure to combustion-originated air pollution for 6–8 weeks increased ALOX metabolites but decreased COX metabolites.³⁰ The identification of biomarkers to detect longer-term or repetitive exposure to wildfires require anther study. Lastly, this study did not evaluate the specificity of these biomarkers to woodsmoke exposure. Although our results suggested good sensitivity of urinary PGE2 and 15-keto-PGF2α for the detection of wood smoke exposure, further studies are needed to evaluate the specificity of these biomarkers in real-world settings with the co-presence of wildfires and other pollution sources.

Synopsis:

Urinary cyclooxygenase metabolites of arachidonic acid are promising biomarkers to detect sex-dependent effects of short-term wildfire exposure.