Development of LC-HRMS Untargeted Analysis Methods for Nasal Epithelial Lining Fluid Exposomics

Abstract

Background:

The nasal mucosa, as a primary site of entry for inhaled substances, contains both inhaled xenobiotic and endogenous biomarkers. Nasal mucosa can be non-invasively sampled (nasal epithelial lining fluid “NELF”) and analyzed for biological mediators. However, methods for untargeted analysis of compounds inhaled and/or retained in the nasal mucosa are needed.

Objectives:

This study aimed to develop a high resolution LC-MS untargeted method to analyze collected NELF. Profiling of compounds in NELF samples will also provide baseline data for future comparative studies to reference.

Methods:

Extracted NELF analytes were injected to LC-ESI-MS. After spectrum processing, an in-house library provided annotations with high confidence, while more tentative annotation proposals were obtained via ChemSpider database matching.

Results:

The established method successfully detected unique molecular signatures within NELF. Baseline profiling of 27 samples detected 2,002 unknown molecules, with 77 and 463 proposed structures by our in-house library and Chemspider matching. High confidence annotations revealed common metabolites and tentative annotations implied various environmental exposure biomarkers are also present in NELF.

Significance:

The experimental pipeline for analyzing NELF samples serves as simple and robust method applicable for future studies to characterize identities/effects of inhaled substances and metabolites retained in the nasal mucosa.

Graphical Abstract

Introduction

Inhalation serves as one of the major exposure routes to xenobiotics, allergens, pathogens, and other substances that can affect human health.¹ Inhaled chemicals that pose a public health risk include environmental and industrial air pollution, secondhand smoke, and e-cigarette usage.^2–5 Many studies investigating health effects of inhaled chemicals have focused on either understanding the chemical composition of the contaminated air,^{6, 7} or biomarkers in conventional biospecimens such as blood and urine.^{8, 9} However, we propose that studying the primary exposure site (i.e., nasal cavity) will provide the most precise information on the biological impact of inhaled substances. It will allow a narrowing of focus to compounds that are retained in the airway after exposure and insight into the impact of these compounds on respiratory health effects.

Noninvasive biological sampling methods have been favored in the past decade due to ease of sample collection, cost effectiveness, and ethical advantages.¹⁰ However, most large studies focus on systemic sampling methods, including blood and urine, which may not best represent respiratory specific outcomes associated with inhalation-based exposures. A method that is gaining popularity due to its non-invasive, low cost, and quick nature is the collection of nasal epithelial lining fluid (NELF) to sample the nasal mucosa.¹¹ NELF sampling utilizes a synthetic matrix to absorb and retain the nasal mucosa. The use of absorptive matrix permits concentrated collection of NELF and is thought to improve reproducibility, allow serial sampling, avoids dilution effects, and facilitate more flexible sample processing and storage protocols compared to the conventional nasal lavage method.^{11, 12} This method has been demonstrated to facilitate the analysis of protein, gene expression, microbiome, and presence of viral pathogens.^{11, 13, 14} These studies not only identify molecular-level alterations of the nasal physiology under environmental compound or pathogen exposure, but also show wide research potential for using NELF in respiratory health studies. These improvements in collecting nasal mucosa may also promote increased interest utilizing respiratory samples to evaluate biomarkers of respiratory health in large cohort studies. An additional method that may further increase the utility of this sample collection method in future exposomics studies include the establishment of a method to characterize chemical and metabolic compounds contained in NELF.

High resolution mass spectrometry (HRMS) is a primary method for untargeted analysis for chemical and metabolic compounds in exposomics research. HRMS, as well as its corresponding methods in instrumental operation and data processing, have immensely progressed our ability in characterizing compounds involved in metabolic activities and environmental contaminants. Untargeted analysis in HRMS is defined as an unbiased screening over wide m/z range after (simple) sample preparation to allow profiling of representative compounds constituting the analytes,^{15, 16} and can be applied to characterize molecules in the nasal mucosa. To our knowledge, NELF has not previously been evaluated utilizing this method, presenting a novel analytical approach. To fill this gap, here we describe a new protocol for analyzing NELF collected on an absorptive matrix via untargeted HRMS. We collected NELF in healthy participants and describe the results of optimized untargeted HRMS including compound annotation and biological pathway associations.

The goal of this study is to develop a robust LC-HRMS based untargeted analysis to characterize compounds in a absorptive matrix containing NELF samples. We believe establishment of such method promotes future research in respiratory health using similar sampling procedure. Specific goals of the method development include a) assurance of detecting biologically related molecules (rather than compounds included in the absorptive matrices), b) projection of baseline metabolic profile of NELF for future reference, and c) discussion of future research possibilities and potentials using this method in health studies that uses NELF samples.

Materials/Subjects and Methods

Chemicals and Materials

Chemicals and reagents, unless otherwise specified, were purchased from Sigma-Aldrich (St. Louis, MO). [D₂]-indole propionic acid and [D₃]-tryptophan were bought from CDN Isotopes (Pointe-Claire, Quebec, Canada). Optimal LC-MS grade water, methanol, acetonitrile, and formic acid were obtained from Thermo Fisher Scientific (Rockford, IL). Leukosorb medium (P/N: BSP0669) was purchased from by Pall Life Sciences (Port Washington, NY) and were cut to fit the nasal passages using a steel die cutter from Ellison Education (Lake Forest, CA) to 6mm × 44mm dimensions.

Subject Recruitment

Healthy adult participants aged 19–45 were enrolled in the study. Exclusion criteria for the study included current symptoms of allergic rhinitis, asthma, a forced expiratory volume of less than 75% predicted, physician diagnosed chronic cardiorespiratory condition, recent nasal surgery, bleeding disorders, immunodeficiency, and tobacco use. The protocol was approved by the University of North Carolina Institutional Review Board.

Nasal Epithelial Lining Fluid Collection

NELF was collected as described previously and illustrated in Figure 1A.¹¹ Nares were randomly selected for sampling, either right or left side. The nare selected was briefly moistened with 0.1 mL of 0.9% sterile, normal saline solution. Cut Leukosorb strips were inserted into the nare at the inferior nasal turbinate along the anterior side until the indicator mark reached the base of the nare. A padded nose clip was applied to maintain the strip in the nose and to ensure maximum surface area of the strip with the nasal mucosa. The strip remained in the nare for 2 minutes. Strips containing NELF were then removed from the nostril, placed in microcentrifuge tubes, and frozen at −20 °C until analysis.

Figure 1.

Schematic illustration of the analytical pipeline of profiling components in nasal epithelial lining fluid retained on strips. (A) Sampling of nasal mucosa by collecting nasal epithelial lining fluid on absorptive strip. (B) Sample preparation, analysis, and data interpretation. Our pipeline provided simple and verified method in sample pretreatment, instrumental analysis, and initial data processing for future subsequent applications including characterizing differently expressed exposure/effect biomarkers. Sampling schematic created at BioRender.com.

Sample Pretreatment

NELF samples were thawed to room temperature, then each strip was fully presoaked with 100 μL water with gentle shaking (100 rpm) for 30 min. Subsequently, 400 μL of organic solvent (acetonitrile) was added to the strip. It should be noted that acetonitrile was used for final analysis, but methanol was used in a preliminary test to compare extraction efficiency. Constituents of NELF were extracted by incubating the strips at 4°C with vigorous shaking (280 rpm) for 24 hour after a 20-minute-ultrasonication. We also tested 3 hr, 6 hr, 48 hr, and 72 hr extraction times, which are all within the standard range from protein precipitation and metabolomic sample preparation in other studies.^{17, 18} The samples that were incubated for 24, 48, or 72 hr did not differ in feature number (# Features in Volunteer NELF - # Features in Filter Blank) and all resulted in 125% and 111% more than the 3 hr and 6 hr samples, respectively. Proteins and the strip were precipitated and removed by centrifugation at 15,000 × g for 10 minutes. The supernatants were dried using a SpeedVac^® and reconstituted with 80 μL of 2% acetonitrile in water. The processed biological samples were each aliquoted for analyses and a portion of each of the samples was pooled to be used as a positive quality control (QC) sample. Strip blanks and procedure blanks are respectively defined as unused strip and no strip followed by the exact same extraction procedures, which were prepared in parallel for quality assurance.

LC-ESI-MS Analysis

LC-MS analysis was applied on a high-resolution accurate mass spectrometry–based platform as previously reported.^{19, 20} The instrumentation included a Thermo Fisher Scientific Vanquish UHPLC coupled to a Q Exactive mass spectrometer with a heated electrospray ionization source (HESI) as the interface. Processed NELF analytes were injected (10 μL) into a Waters Acquity UPLC HSS T3 (reverse phase C18, 100 Å, 1.8 μm, 2.1 mm × 100 mm) analytical column controlled at 40 °C, with the mobile phase composed of water (A) and acetonitrile (B) both with 0.1% formic acid at a flow rate of 0.4 mL/min. The 15-min-gradient for chromatographic separation was set as the following: 2% B from 0 to 1 min; 2% to 15% B from 1 to 3 min; 15% to 50% B from 3 to 6 min; 50% to 98% B from 6 to 7.5 min; 98% B held from 7.5 to 11.5 min; 98% to 2% B from 11.5 to 11.6 min; and 2% B held from 11.6 to 15 min for a final re-equilibration. Mass spectrometer analyses were conducted in positive mode with the sheath gas, auxiliary gas, and sweep gas set to flow rates of 50, 13, and 3 L/min, respectively. The spray voltage was set to 3.50 kV. Capillary and auxiliary gas heating temperature were respectively controlled to 263°C and 425°C. Under the full scan of m/z ranging from 70 to 1000, the resolution was set to 70,000 FWHM (m/z 200) with automatic gain control and maximal injection times set to 2×10⁵ and 50 msec, respectively.

All NELF samples were processed within one LC-MS batch and the order of injection was randomized in a blocked fashion that allows spotted injections of procedure blanks, strip blanks, a standard mixture composed of [D₃]-tryptophan, [D₅]-glutamic acid, and [D₂]-indole propionic acid each in 500 nM (10 μL injection), and 6 quality controls comprised of pooled aliquots of each of the biological samples. Mass accuracies, retention time stabilities, signal strength of the standards, as well as total signal summed in samples were used to ensure instrumental performance throughout the analysis.

Data Processing

Data obtained from LC-MS analysis (in format of *.RAW) were imported to Compound Discoverer (CD) ver 3.2 (Thermo Fisher Scientific) to identify m/z-retention time features, align peaks among all samples, assort features into (unknown) compounds, and structural annotation.^{21, 22} Procedure blanks and strip blanks were categorized as “blank” to identify background signals. A pre-installed data processing workflow with minor modification was applied (See Additional File 1). Isotopic patterns, adduct formulations, and other spectrum analyses were performed to assort the features into a list of unknown compounds. These queries were calculated for their most accurate chemical composition, and matched across our in-house LC-MS database and the ChemSpider chemical structural database. The in-house library was composed by testing authenticated standards of 438 common metabolites, drugs, environmental exposures, which were matched in parallel for accurate mass (within 1.53 ppm, Figure S5) and retention time (within ± 0.252 min., Figure S6) to the unknown compound list to provide confident structural proposals (Additional File 2). The retention time tolerance (± 0.252 min.) for in-house library matching and mass tolerance (± 1.53 ppm) for both library and ChemSpider matching was derived from the 95% confidence interval of precision testing by repeated injections of authentic standards. A criterion of 50% detected ratio in the NELF samples was applied to select representative and meaningful features, and non-detected signals were permuted as one tenth of the lowest detected signal among all features. After the automatic processing of CD, manual inspection of the alignment, annotation, and background classification was practiced to ensure data quality. Signal area feature-wised were scaled with the total useful signal in each sample as previously described ²³.

Data Interpretation

Annotated chemical structures, either from our in-house library or the ChemSpider database was matched for multiple identifiers in PubChem, CAS registry, HMDB, KEGG, and ChEBi databases to validate, cross-reference, and maximize capability to link biochemical pathways. MetaboAnalyst 5.0 and MetaMapp were used for pathway enrichment analysis and chemical/biochemical similarity network derivation.^{24, 25} Cytoscape 3.8.2. was used to illustrate network files exported from MetaMapp.

Results

Development of Untargeted Profiling for Compounds in Nasal Epithelial Lining Fluid

The non-invasive sampling of nasal mucosa described here can provide insight on both inhaled exogenous exposures and local (i.e., nasal and respiratory system) endogenous metabolites. We aimed to establish untargeted LC-MS based method to profile metabolic and exposure biomarkers from collected NELF. The developed experimental pipeline is illustrated in Figure 1B. Four organic solvents (acetonitrile and methanol, each with and without 0.1% formic acid acidification) were tested for compound extraction (n = 2 for each solvent). Acetonitrile extraction resulted in the highest feature detection in LC-MS analysis (~ 1.3–1.8 fold-increase than other three solvents, data not shown), and was used subsequently for analysis for the NELF samples.

In the same preliminary test, we also compared signals in NELF samples, strip blanks (i.e., unused Leukosorb strips with same dimensions followed by same extraction protocol), and procedure blanks (i.e., same procedure but without any strip). Figure 2A shows signal distribution of the samples and blanks, and the wider violins for NELF samples implied successful compound extraction from NELF. The abundance of base intensity peaks for NELF samples also indicate that compounds attributed to the NELF were successfully detected after pretreatment (Figure 2B). To identify compounds that are abundant and prevalent in human NELF, we further evaluated the NELF samples in an analytical pipeline. Exploratory observation indicate that expected variation exists among subjects, but that the NELF samples also contained significantly different profiles compared to strip blanks, confirming successful extraction and analysis protocols (Figure 2C).

Figure 2.

Feature detection in nasal epithelial lining fluid samples. (A) Violin plot of signal distribution of nasal epithelial lining fluid samples (pooled quality control: purple; single representative sample: orange), strip blanks (blue), and procedure blanks (gray). (B) Distribution of base peak intensity in unused strips (upper panel, blue line) and in pooled NELF quality control (lower panel, red line). (C) Principal component analysis (PCA) of signal distribution for nasal epithelial lining fluid samples (n = 27, orange dots), pooled quality control (n = 6, purple dots) and strip blanks (n = 6, blue dots). Strip blanks overlap and n is indicated in the figure with an arrow.

Metabolomic/Exposomic Profiling of Nasal Epithelial Lining Fluid

Abundant and representative LC-MS features were defined in this methodological study with the criteria of 50% or higher detected prevalence among samples. NELF samples from 27 subjects resulted in 2,002 unknown features with distinctive retention times and calculated monoisotopic mass, which is illustrated in Figure 3 with their respective prevalence and signal strength. Throughout the analysis, a quality control sample (n = 6) and a mixture of standards (n = 7) was routinely injected for LC-MS analysis to examine instrumental stability, and observations showed little fluctuation in mass accuracy, signal strength, and retention time (Figure S2). The 2,002 unknown features were detected in 50% or more of the biological samples analyzed and were also detected in the majority of the quality control injections (100% in 5 QC samples and 89% in 6 QC samples), similar to criteria for feature selection described previously.^{26, 27} For this study, total ion scaling was performed to account for systemic instrumental variation, though the total ion intensity maintained stable among samples (Figure S3).

Figure 3.

Distribution of the 2,002 abundant and representative compounds (clustered features) in nasal epithelial lining fluid samples (n = 27) at different prevalences (point colors from blue to coral represent low to high detected ratio) and signal strengths (point size). Identical compounds with differing isotopes, adducts, or analytical properties (e.g., dimers) were combined into one compound in subsequent annotation queries. The signal strength of each compound (point size) is represented by the strongest ion detected of that compound.

All of the 2,002 unknown features included one or more corresponding ions (due to adduct heterogeneity, dimer formation, etc.) in the LC-MS analysis, and the signal strengths described herein were respectively represented by the strongest ions. The detected ratio was evenly spread across a range of 50% to 100% (Figure S4), and we believe that these unknown features were comprised of common endogenous metabolites including amino acids and membrane lipids, as well as ubiquitous environmental exposures such as allergens, air pollution, and other xenobiotics. To address this hypothesis, we further attempt to decipher the chemical composition and structures of these unknown features to map the metabolic activity and exposure profiles sampled from NELF.

Compound annotation and biological interpretation

A total of 2,002 unknown features attributed to human NELF were queried for chemical composition as well as proposed structures using our in-house LC-MS database (438 compounds analyzed in same instrumental method, see Additional File 2) and the ChemSpider chemical structure database, which integrated multiple repositories including HMDB, KEGG, and ChEBI. Among all queries, 1,802 unknown features were successfully assigned to a top-ranked chemical composition (e.g., C₁₀H₁₃N), and 540 of the 1,802 formulas were annotated to a specific structure. Annotations by our in-house library (77/540) corresponded to higher confidence of proposed chemical structure, and structures identified by matching the ChemSpider database (463/540) provided tentative and possible annotations.

Both relationships in biochemical pathways and structural similarity provide clue as to the repertoire of compounds in the nasal mucosa. The 540 compounds characterized were imported to MetaMapp and MetaboAnalyst 5.0 to perform cluster and enrichment analyses (Figure 4). The clusters in Figure 4A indicate strong structural similarity, and the three large and dense clusters corresponded to common amino acids, their derivatives, and mono/poly-benzene structures. The common amino acids annotated, including arginine, histidine, lysine, etc., were also evident in the enrichment analysis, as their biosynthetic pathway showed over-representation (Figure 4B). Various common nucleosides/nucleotides (e.g., cytidine, thymine, uracil) were also detected (cluster of “pyrimidine and purine derivatives” in Figure 4A) and over-represented (Figure 4B). The top 25 enriched pathways include biosynthesis/metabolism/biodegradation of the common amino acids (e.g., phenylalanine, tyrosine, tryptophan), sugars (e.g., galactose), nucleic acid monomers (e.g., pyrimidine, purine), and lipids (e.g., sphingosine).

Figure 4.

Compounds in nasal epithelial lining fluid annotated by both in-house library and ChemSpider chemical structure database. (A) Networks of structural similarities (Tanimoto) and (KEGG) biochemical pathways calculated by MetaMapp. Detected proportion between 50% and 100% are shown in a scale of green to yellow to red, respectively. KEGG pathways are shown with solid lines, while Tanimoto structural similarity is shown with dotted lines. Node size indicates signal strength of compound on LC-MS. In-house database annotations are shown with circles and ChemSpider annotations are shown with squares. Clusters are labeled according to structural similarity. (B) Top 25 pathways enriched in MetaboAnalyst 5.0 analysis. The enrichment ratio (ratio between number of annotation and expectation) and p-value are shown on the x-axis and by point size, respectively.

Annotation by our in-house library, though restricted to the limited number of compounds archived, provided additional confidence to the proposed structures; therefore, they were in higher priority in annotation compared to Chemspider matching. However, it should be noted that the 77 in-house library matched structures were all consistent to respective ChemSpider annotations, which suggests that annotations with ChemSpider provide reasonable, though tentative, results. The 77 in-house library annotated compounds were analyzed for structural similarity and biochemical pathway enrichment (Figure 5). Enrichment analysis indicated that these compounds were over-represented with metabolites associated with the biosynthesis, metabolism, and biodegradation of amino acids (e.g., phenylalanine, arginine), redox related metabolites (e.g., nicotinate, nicotinamide, glutathione) and other essential molecules in endogenous cellular processes (Figure 5B).

Figure 5.

Compounds in nasal epithelial lining fluid annotated only by in-house library. (A) Network of structural similarities and (KEGG) biochemical pathways calculated by MetaMapp. Detected proportion between 50% and 100% are shown in a scale of green to yellow to red, respectively. KEGG pathways are shown with solid lines, while Tanimoto structural similarity is shown with dotted lines. Node size indicates signal strength of compound on LC-MS. Clusters are labeled according to structural similarity. (B) Top 25 enriched pathways identified via MetaboAnalyst 5.0. The enrichment ratio (ratio between number of annotation and expectation) and p-value are shown on the x-axis and by point size, respectively.

Discussion

Rationale of method development

Our study addresses the current lack of methods for untargeted chemical analysis of nasal biospecimens and specifically address the utility of this method using an absorbent matrix-based NELF sampling method. The verified method we developed is composed of NELF collection, pretreatment of samples, LC-MS instrumental analysis, and data processing (Figure 1). The application of absorptive membrane carrier to collect biospecimen has advantages of concentrating analytes, sampling ease, storage efficiency, etc.^{11, 28}

This method parallels the utilization of dried blood spots (DBS) as tool to collect, inventory, and analyze blood on a similar synthetic matrix-based platform. DBS has been frequently utilized in qualitative and quantitative biological applications, and several studies have evaluated the use of similar untargeted metabolomic techniques to those described here for exposomic evaluation of blood samples ^{18, 29–31}. For instance, Petrick et al. (2017) concluded that 80% (volume) acetonitrile in water provided optimal compound extraction and instrumental sensitivity among the other tested solvents. In another study, water presoaking before addition of organic solvent resulted in higher compound recovery.³² Since only very limited studies are available for LC-MS analysis in nasal mucosal samples ³³, and none were available using the absorbent matrix-based collection, method optimization utilized in DBS studies were critical to the our establishment of our NELF sample pretreatment protocol.

Our preliminary data showed that more features were detected by presoaking the absorptive matrix and extracting with acetonitrile. Additionally, our data also show that the majority of these features were associated with the NELF biospecimen, rather than the absorptive matrix (Figure 2B). The six unused absorptive matrices (strip blanks), with the same pretreatment, showed nearly identical signal profile (Figure 2A and 2C), which demonstrated that the low and consistent background contributed by the absorptive matrix did not conceal signals from the NELF analytes. The further analysis of 27 human NELF samples successfully detected molecules contained within NELF (Figure 3), which provided a pioneering profiling of exogenous and endogenous compounds contained within NELF that may be of utility in understanding changes in the respiratory mucosa due to exogenous chemical, medicinal, or pathogen exposures.

Confidence and interpretation in compound annotation

CD is a relatively new commercial tool that facilitates data processing in HRMS-based untargeted analyses, and has been validated and applied in several studies analyzing the metabolome,^{22, 34} exposome,³⁵ and environmental chemicals.^{36, 37} After harmonizing feature signals across samples, annotation of chemical structures was completed by matching structures in our in-house library and the ChemSpider database. Our database was comprised of common endogenous metabolites, typical drugs, and some environmental xenobiotics (Additional File 2), and was also used in our previous metabolomic studies.^{19, 20} Structures were proposed by matching in our library and qualified as “Level 1 (highest)” confidence using three orthogonal data tested with authentic standards: monoisotopic mass (high resolution), retention time in the same chromatographic conditions, and isotopic pattern. Confidence levels were utilized as defined by the Metabolomic Standard Initiative, ranging from 1–5, Level 1 indicating highest quality.³⁸ Constituents of Level 1 molecules in NELF were enriched amino acids, nucleic acid (nucleoside/nucleotide), carnitines, and their derivatives (Figure 4), which were expected and consistent with prior targeted chemical profiling in nasal biospecimens.³³ We believe that these confident annotations and their association with prior targeted methods, together support that our untargeted HRMS analysis is will likely capture exposomic profiles in NELF samples. The more tentative structural proposals by matching ChemSpider served as “Level 4” annotations based on Metabolomic Standard Initiative established standards ³⁸. The broad inclusion of the ChemSpider universe provided a larger compound library than our in-house database and supported the exploration of exogenous chemicals present in NELF, such as the possible presence of barbiturate pesticides, polyethylene molecules, polyaromatic hydrocarbons, and other exposures (Figure 3). These tentative annotations on environmental contaminants suggested high potential of our method in capturing some inhaled xenobiotics that may be captured in the nasal mucosa. As the nose is a primary site of xenobiotic exposures, chemical identified may be less attenuated by metabolism, resulting in higher concentrations of detected primary exposure chemicals, rather than metabolites, as is common with exposome studies using blood as the primary substrate.³⁹ With these pilot exposomic profiles, data-driven hypotheses can be developed to further explore the utility of NELF biomarkers in a variety of studies and support additional MS/MS data to improve annotation confidence.

Limitations and future directions

Our objective in the development of these methodologies is to establish a robust pipeline to profile unknown molecules in NELF samples. However, there are some limitations. Specifically, the absence of MS/MS spectrum limited our ability to make annotations with confidence higher than Level 4. Also, only the versatile positive mode on C18 analytical column was applied in our LC-MS system for this study. To strengthen future studies utilizing this pipeline, we recommend collecting LC-MS data in multiple modes and via MS/MS, specifically for unknown compounds that differ in levels between/among study groups. Finally, we utilized CD. There are various open source tools that have similar functions to CD that could also be utilized, such as aligning signals across samples (e.g., XCMS, MS-DIAL) and proposing structures (e.g., MS-FINDER), with varying accuracies and sensitivities.^{34, 40} Nevertheless, some inhaled xenobiotics are volatile or vaporizable thus instruments such as gas chromatography may be more favorable.

Significance and conclusion

Efforts in precision medicine have amplified the needs for tissue specific assessment of biomarkers to understand innate variability within the human population, as well as in response to drug treatments and environmental exposures. It has also highlighted the need for big data approaches to understand the role of the exposome in human health effects. In order to advance the understanding of the exposome on respiratory health, here we evaluate utilizing a novel, quick, non-invasive sampling method to assess exogenous chemical exposure and endogenous metabolites using untargeted LC-HRMS. We establish a simple and robust method to extract and detect compounds, as well as perform compound annotation in NELF samples. Application of this method in 27 subjects revealed common metabolites and potential exposures to various xenobiotics, such as environmental pesticides. The success of this untargeted analytical approach supports its utility in future toxicological, pharmacological, and respiratory health studies to understand the identities and influences of inhaled substances.

Impact Statement.

The nasal mucosa contains exogenous and endogenous compounds. The development of an untargeted analysis is necessary to characterize the nasal exposome by deciphering the identity and influence of inhaled compounds on nasal mucosal biology. This study established a high resolution LC-MS based untargeted analysis of non-invasively collected nasal epithelial lining fluid. Baseline profiling of the nasal mucosa (n = 27) suggests the presence of environmental pollutants, along with detection of endogenous metabolites. Our results show high potential for the analytical pipeline to facilitate future respiratory health studies involving inhaled pollutants or pharmaceutical compounds and their effects on respiratory biology.

Data Availability

Data are available from the corresponding author upon reasonable request.