BACKGROUND: We recently developed a novel, noninvasive method for sampling nonvolatile material from the distal airways. The method is based on the collection of endogenous particles in exhaled air (PEx). The aim of this study was to characterize the protein composition of PEx and to verify that the origin of PEx is respiratory tract lining fluid (RTLF).
METHOD: Healthy individuals exhaled into the sampling device, which collected PEx onto a silicon plate inside a 3-stage impactor. After their extraction from the plates, PEx proteins were separated by SDS-PAGE and then analyzed by LC-MS. Proteins were identified by searching the International Protein Index human database with the Mascot search engine.
RESULTS: Analysis of the pooled samples identified 124 proteins. A comparison of the identified PEx proteins with published bronchoalveolar lavage (BAL) proteomic data showed a high degree of overlap, with 103 (83%) of the PEx proteins having previously been detected in BAL. The relative abundances of the proteins were estimated according to the Mascot exponentially modified protein abundance index protocol and were in agreement with the expected protein composition of RTLF. No amylase was detected, indicating the absence of saliva protein contamination with our sampling technique.
CONCLUSIONS: Our data strongly support that PEx originate from RTLF and reflect the composition of undiluted RTLF.
Every day the respiratory system is exposed to various air pollutants, allergens, pathogens, and microbes that may cause airway inflammation (1, 2). As a protective interface between the external environment and epithelial cells, the respiratory tract lining fluid (RTLF)3 covers the airways. The main component of RTLF is lung surfactant, which consists of phospholipids (approximately 90%) and proteins (approximately 10%) (3). Alterations in RTLF composition may reflect inflammatory changes in the airways. Changes in the protein composition have been reported under conditions in which the respiratory tract is subjected to stress, e.g., as that caused by asthma and infections (4–6).
Biological changes caused by airway inflammation are difficult to monitor. Currently, the most efficient methods for sampling RTLF are invasive, such as bronchoalveolar lavage (BAL), or semi-invasive, such as induced sputum. These methods are not applicable for studies of large populations, or for repeated sampling and screening methods. Therefore, the noninvasive sampling of exhaled breath condensate (EBC) has become a rapidly growing field in respiratory analysis; however, EBC has a number of serious methodologic difficulties, e.g., correction for dilution and lack of standardization, such that data produced by different groups can vary considerably (7–9).
We have recently developed a novel, noninvasive technique for collecting nonvolatile material from the respiratory system (10). The individuals exhale into the sampling device, which uses a 3-stage impactor to collect endogenous particles in exhaled air (PEx) for subsequent chemical analysis. How PEx are formed is not yet fully understood, but one likely mechanism is rupture of the RTLF film during airway reopening after airway closure (11). In previous work, we have detected phospholipid species, such as phosphatidylcholine and phosphatidylglycerol, that were consistent with an RTLF composition (10). The observed phospholipid composition of PEx was consistent with that of lung surfactant, further supporting RTLF as the origin of PEx.
To evaluate the potential of PEx for respiratory research requires establishing the molecular composition of the exhaled endogenous particles. We present the first proteomic study of PEx. By analyzing the protein composition of PEx, we aim to verify their origin as being the RTLF and to evaluate them as a source for biomarkers of respiratory diseases.
Materials and Methods
CHEMICALS AND REAGENTS
dl-Dithiothreitol ≥99%, iodoacetamide Sigma Ultra, and trifluoroacetic acid reagent grade ≥98% were purchased from Sigma-Aldrich (http://www.sigmaaldrich.com). Acetonitrile HPLC gradient grade was purchased from Fisher Scientific (http://www.fishersci.com). Sequencing-grade trypsin was purchased from Promega (http://www.promega.com). Ammonium bicarbonate Puriss (NH4HCO3) was purchased from Riedel-de Haën (http://www.riedeldehaen.com). PBS (0.14 mol/L NaCl; 0.0027 mol/L KCl; and 0.01 mol/L phosphate buffer; pH 7.4) was purchased from Medicago (http://www.medicago.se). All chemicals and precast gels for the 1-dimensional gel electrophoresis were purchased from Invitrogen (http://www.invitrogen.com). High-purity water was used for all experiments (Milli-Q Plus; Millipore, http://www.millipore.com).
The study included 12 healthy volunteers. These individuals had a mean age of 36 years (range, 28–57 years), and 8 (67%) of the volunteers were female. The study was approved by the local ethics committee of the Sahlgrenska Academy of Gothenburg, and all participants gave their informed consent.
PEx were collected onto a silicon plate with a custom-built sampling device previously described by Almstrand et al. (10). Forced exhalations were performed into the sampling device, where the particles were simultaneously counted by an optical counter and collected inside a 3-stage impactor. Between forced exhalations, the participants inhaled particle-free air tidally through a particle filter.
There were 2 sampling sessions. Pooled samples from 6 healthy individuals were collected in the first session, for a 3000-L total volume of exhaled air. In the second session, pooled samples from 10 healthy individuals were collected, for a total volume of 4400 L.
We collected pooled samples by having the participants take turns exhaling for 25 min (100–200 L of air) onto the same sampling plate. Four of the participants from the first session were included in the second session. All participants rinsed their mouths with purified water and breathed particle-free air for 2 minutes before sampling. Each participant wore a nose clip throughout the sampling procedure. The silicon plates containing the exhaled particles were stored at −20 °C before analysis. For PEx analysis, the proteins were extracted from the silicon plates with NuPAGE sample buffer (Invitrogen) and PBS.
A control sample of room air was collected by drawing room air into the sampling device for 7.5 h. Particles collected in this sample were analyzed with the same procedure used for the pooled samples.
TOTAL PROTEIN CONTENT
The total protein content of PEx was measured with a CBQCA Protein Quantification Kit (Molecular Probes Europe) and a FluoroMax-3 spectrofluorometer (Horiba Jobin Yvon).
Proteins in PEx extracts were separated by SDS-PAGE before tryptic digestion and LC-MS analysis. Proteins were separated under reducing conditions on precast 1-dimensional gels (NuPAGE 4%–12% Bis-Tris Gel) using NuPAGE MES SDS Running Buffer and the XCell SureLock™ Mini-Cell system. SeeBlue® Plus2 Pre-Stained Standard protein mix was used as molecular weight standards, and proteins were stained with the Colloidal Coomassie Blue Staining Kit.
IN-GEL DIGESTION AND PEPTIDE EXTRACTION
The gel from the first sample was cut into 17 slices, including 1 control slice from a blank region of the gel. The slices were cut into 1 mm cubes and destained in acetonitrile/100 mmol/L NH4HCO3 (45:55 volume ratio) for 1 h on a shaker at room temperature. The solution was removed, the gel samples were dehydrated by incubation with 75 μL of ethanol for 5 min, and the ethanol was removed. Proteins were reduced and alkylated by incubation with 10 mmol/L dithiothreitol for 30 min at 37 °C, followed by incubation with 55 mmol/L iodoacetamide for 30 min in the dark at room temperature. After each step, the supernatants were removed, and the samples were dehydrated by incubating them with 75 μL ethanol for 5 min and then removing the ethanol. For trypsinolysis, 15 μL trypsin solution (10 ng/μL trypsin in 50 mmol/L NH4HCO3) was added to the dry gel pieces. After a 30-min incubation at 4 °C, additional volumes of 50 mmol/L NH4HCO3 sufficient to cover the gel pieces were added to the samples, followed by incubation overnight at 37 °C. Peptides were extracted with 5 mL/L trifluoroacetic acid in 2 mmol/L n-octyl-β-d-glucopyranoside for 30 min. The eluates were lyophilized in a SpeedVac (Savant) and stored at −20 °C until LC-MS analysis.
The gel from the second sample was cut into 15 slices. The individual slices were reduced, alkylated, and digested with trypsin as described above with the aid of a Biomek 2000 Laboratory Automation Workstation (Beckman Coulter).
Peptide mixtures were analyzed by reversed-phase nanoscale capillary liquid chromatography coupled to electrospray ionization mass spectrometry. The samples were injected onto a 15-cm reversed-phase column (3-μm ReproSil-Pur C18 AQ media; Dr. Maisch GmbH) packed in an uncoated, fused silica emitter (PicoFrit: 15-cm length, 75-μm inner diameter, 8-μm tip inner diameter; New Objective) according to the procedure of Ishihama et al. (12). Liquid chromatography was performed on a CapLC System (Waters) fitted with a passive flow-splitter to reduce the flow to 150–250 nL/min. The liquid chromatography system was connected to a linear quadrupole ion trap–Fourier transform ion cyclotron resonance (LTQ-FT ICR) mass spectrometer (Thermo Scientific) operated in the data-dependent mode. Survey spectra in the m/z range of 350–2000 were acquired in the ICR cell, and the 5 detected ions with the highest intensities were selected in order of decreasing intensity for fragment ion mass spectrometry (MS2) analysis in the LTQ. Singly charged ions were excluded from the MS2 analysis. Ions within a window of 5 ppm around the m/z values of the ions selected for MS2 analysis were dynamically excluded for a time period of 45 s. The data from the first sample were converted to centroid data with the instrument manufacturer's software. The data from the second sample were stored as profile data, and monoisotopic m/z values were extracted with Mascot Distiller software (Matrix Science).
Proteins were identified with the Mascot software by searching the IPI human database (Mascot version 2.3 and IPI version 3.72, 86 392 sequences). The following settings were used for the searches: mass error tolerance for the precursor ions, 10 ppm for centroid data and 5 ppm for raw data; mass error tolerance for the fragment ions, 0.4 Da; fixed modification, carbamidomethylation; variable modification, methionine oxidation; number of missed cleavage sites, 1; type of instrument, FT-ICR.
A protein was considered identified if it had a minimum of 2 peptides with a Mowse score ≥20. For proteins identified from <2 peptides with Mowse scores >20, peptide fragment spectra were evaluated manually to accept or reject the identification.
TOTAL PROTEIN CONTENT
Particles were collected from 300 L of exhaled air. The amount of total protein in the sample was 100 ng, i.e., 0.33 ng of PEx per liter of exhaled air. The total mass of particles calculated for this sample was 440 ng, which corresponds to a protein content of 23% in PEx. This result is in agreement with the 10% protein content of surfactant previously reported (3). The result from the analysis of total protein content was used to estimate the total protein content in the 2 samples used for protein characterization, according to the volumes of air sampled.
We identified 124 proteins in the 2 PEx samples, 103 (83%) of which have previously been identified in BAL (5, 6). The first sample was collected from 3000 L of exhaled air containing approximately 1 μg total protein; the second sample obtained from 4400 L of exhaled air contained approximately 1.5 μg total protein. We identified 32 proteins in the first sample and 116 in the second sample, with 24 of the proteins being shared by the 2 samples. Among the shared proteins were some of the most prominent blood and BAL proteins: albumin, serotransferrin, surfactant protein A (SP-A), α1-antitrypsin, and immunoglobulins (13). Table 1 lists the identified proteins.
The proteins were sorted according to their cellular component annotations in ProteinCenter™, a Web-based data-interpretation tool (Proxeon). Many of the identified proteins were associated with multiple cellular locations (Table 2), with extracellular proteins constituting the largest group. Ninety-eight (79%) of the 124 identified proteins were annotated as extracellular. Cytoplasm and membrane were other abundantly represented cellular locations, accounting for 63% and 48%, respectively, of the identified proteins. Many of the cytoplasmic and membrane proteins were additionally annotated as extracellular proteins, specifically 74% of the cytoplasmic proteins and 75% of the membrane proteins.
The high number of extracellular proteins in PEx, which constituted almost 80% of the identified proteins, is in strong agreement with our assumption that PEx should contain mainly secreted and extracellular proteins, similar to those of RTLF.
RELATIVE QUANTIFICATION OF THE IDENTIFIED PROTEINS
We used the exponentially modified protein abundance index (emPAI) to obtain an approximate, relative quantification of the identified proteins. The emPAI is a label-free protocol that estimates the relative quantities of the proteins in a mixture from the protein coverage revealed by the peptide matches in a database search (14). The contributions of the 9 most abundant proteins were: albumin, 26%; surfactant proteins, 13% (including SP-A, 12%); various immunoglobulins, 14%; serotransferrin, 4%; Clara cell protein (CC16), 4%; lysozyme C, 2%; proteases and inhibitors, 4% (including α1-antitrypsin, 2%); annexins, 1%; and complement factors, 1%.
PROTEINS IDENTIFIED IN ROOM AIR
Six of the 40 proteins identified in room air were also identified in at least one of the samples, indicating possible contamination of the sample. The identified possible contaminants were: filaggrin 2, glyceraldehyde-3-phosphate dehydrogenase, deleted in malignant brain tumors 1 protein, isoform DPI of desmoplakin, prolactin-inducible protein, and serpin B12. None of the major proteins mentioned above in the PEx samples were identified in room air, however.
For PEx to originate from RTLF, one would expect PEx to show similarities with BAL fluid, which is commonly used for the collection of RTLF. We used data from 2 independent BAL proteomic studies for comparison with the proteins identified in PEx (5, 6). Of the 124 identified proteins, 103 had previously been reported to occur in BAL fluid, a finding that is in good agreement with our assumption that the composition of PEx reflects that of RTLF. The majority of the identified proteins were extracellular proteins. We compared the distribution of the cellular localizations of the proteins identified in PEx with the distributions obtained in other human proteome projects: HUPO Brain, HUPO Plasma, HUPO Urine. PEx showed considerable overrepresentation of extracellular proteins compared with these projects, with 70% of the PEx proteins being extracellular, in contrast to the 20%, 16%, and 55% values obtained in the brain tissue, plasma, and urine projects, respectively. The prominent representation of extracellular proteins is in agreement with our suggestion that PEx mainly contain secreted and extracellular RTLF proteins.
The main component of lower-airway RTLF is lung surfactant, which contains surfactant proteins, SP-A being the most abundant. Our evaluation of the relative distribution of proteins in PEx with the emPAI protocol showed that the most abundant protein was albumin (26% of the total protein content). Owing to the permeability of the endothelial and epithelial barrier of the lungs, we expected a high abundance of major blood proteins. The second most abundant protein was locally produced SP-A, which constituted 12% of the total protein. The high abundance of SP-A indicates that lung surfactant is a major component of PEx, as would be expected for material collected from the lower airways. Because the protein distribution we have presented is based on an approximate quantification, confirmation with an alternative quantification method is required.
The identification of proteins typical of alveolar type II cells and Clara cells (such as SP-A, SP-B, SP-C, and CC16) further supports the hypothesis that PEx is formed in the distal airways. A number of the proteins identified in PEx, such as complement factor B, C1q, C2, C3, and C4, have previously been reported to play a role in allergic airway inflammation. Other identified proteins, such as annexins A1, A2, A3, and A5, have proved to be inhibitors of phospholipase A2. Phospholipase A2 is an enzyme responsible for the release of arachidonic acid, a precursor of inflammatory mediators (eicosanoids). Various proteases and protease inhibitors were among the identified proteins. The identification of proteins related to airway inflammation suggests that PEx measurement will enable monitoring of ongoing inflammatory processes and be of great importance in respiratory research. The complete list of identified proteins presented in Table 1 provides references to published data.
At present, EBC is the only noninvasive commercially available method for sampling material from the airways. A variety of biomarkers, including proteins, have been detected in EBC (15, 16); however, EBC protein studies have been restricted mainly to specific protein assays, e.g., cytokine immunodetection; therefore, we did not include a comparison of EBC and PEx protein compositions in this study.
With the PEx method, the material to be analyzed is sampled from exhaled air similarly to EBC sampling; however, the PEx and EBC sampling methods have substantial methodologic differences (10). EBC collection involves the condensation of exhaled matter (8), and condensation efficiencies are likely to vary, depending on the chemical composition of the exhaled particles. In PEx, the collection of exhaled particles is based on particle size and should be independent of the particle content. Particles or droplets consisting of RTLF are also collected with EBC (16); however, they are more efficiently collected with the PEx method and show no dilution. Dilution is an uncertainty factor for EBC, and is perhaps even more uncertain for BAL, in which adding external fluid to the airways makes it difficult to calculate exact concentrations (e.g., of inflammatory mediators) in the sample (8, 17). The PEx sample is dry when collected. The sample is then extracted into a well-defined volume of suitable buffer for subsequent analysis. During the collection of PEx, the particles are counted and measured according to size, making it possible to calculate the mass of the sample. This feature is important for calculating the content of the various substances in the sample. We estimated that PEx consists of approximately 23% total protein. This finding is in agreement with previous published reports, which have estimated a protein content of 10% in surfactant, which is a major component of RTLF (3).
The ability to collect and count particles simultaneously is important, not only for measuring the total amount of sampled material from the numbers and sizes of the collected particles, but also for studies of the formation of exhaled particles. Particle formation is an important avenue of investigation and should be controlled, especially in comparative studies. In the current study, we used forced exhalations because of the original observation that they increase the formation of particles compared with tidal breathing. Our recent data show, however, that airway reopening after airway closure is the main mechanism of particle formation (11). Airway-closure maneuvers will be used instead of forced exhalations in follow-up comparative studies to improve our control of particle formation among individuals and to increase the amounts of collected material.
Given that PEx are exhaled through the oral cavity, the risk of saliva contamination has to be addressed. Amylase accounts for 60% of the total protein in saliva (18) and therefore is a good marker for the degree of salivary protein contamination. Because amylase is by far the most abundant protein in saliva, detection of salivary proteins thus is possible only when amylase is observed. The absence of amylase among the identified PEx proteins is a strong indication that we had no saliva protein contamination and that all of the proteins identified in our samples originated from the airways. This important feature of the PEx method makes it possible to perform comparative proteomic studies on material sampled exclusively from the airways.
The present work is the first proteomic study of PEx aimed at providing sufficient data for evaluating PEx as a potential source of RTLF proteins. The amounts of analyzed material were rather small, 1–1.5 μg, compared with those of published BAL proteomic studies (5, 6). For this reason, the proteins identified in the current study likely represent the most abundant PEx proteins, and many less abundant proteins are missing. The number of proteins identified showed a large increase, however, when we increased the volume of sampled air. An increase of 50% (3000–4400 L) from sample 1 to sample 2 generated a 260% increase in the number of identified proteins (32 to 116). Improving the PEx sampling efficiency to increase the total amount of collected material should therefore lead to the identification of additional proteins. Our comparison of the 2 protein-identification sets showed that 75% of proteins identified in the first sample were also detected in the second. Given that the proteome is a dynamic system and that the applied mass spectrometry–based protein identification strategy discriminates the detection of low-abundance proteins, the 75% overlap indicates that the major component of the PEx proteome is rather stable in healthy individuals.
A key aspect of mass spectrometry in proteome research that sets it apart from all other analytical techniques is the ability to identify large numbers of proteins in small quantities of biological samples. Limitations of the technique for candidate protein biomarker discovery include the following: (a) There is a strong bias toward the detection of abundant sample components that, considering that the concentrations of proteins in human tissues and body fluids span many orders of magnitude, precludes the detection of many low-abundance proteins. Therefore, improving the identification coverage often requires implementing extensive protein- and peptide-fractionation steps before the mass spectrometry analysis. (b) Relative quantification of individual proteins in different biological samples is time-consuming, and differential analysis is typically limited to relatively small sample sets, features that can lead to high false-discovery rates. Mass spectrometry is therefore most often used in the discovery phase. Any candidate biomarkers discovered then have to be independently confirmed via alternative targeted and quantitative methods (19).
Our data show that the PEx-collection technique should be suitable for comparative mass spectrometry–based proteomic studies that use pooled material from well-defined groups of patients for identifying candidate biomarkers associated with disease-induced alterations of RTLF protein composition.
SUMMARY AND FUTURE PERSPECTIVE
Sampling of the distal airways is important for diagnosis, monitoring, and treatment of distal airway inflammation. The development of alternative noninvasive sampling methods would greatly facilitate our understanding of the biological processes occurring in the distal airways. As a step toward this goal, PEx provide the means for the noninvasive collection of proteins from the respiratory system. Our data strongly support the conclusion that PEx reflect the composition of undiluted RTLF; these results therefore open new possibilities for comparative proteomic studies of disease- or drug-induced alterations in the protein composition of RTLF. Such comparative studies may lead to the discovery of novel inflammatory mediators that have potential as therapeutic targets or biomarkers. Similar studies carried out with BAL collected from healthy and asthmatic individuals have identified 1895 proteins, 10% of which showed different concentrations (6). Validation of the identified candidate biomarkers by alternative targeted approaches (e.g., immunoassays) could lead to the establishment of additional biomarkers applicable for monitoring of individual disease- or drug-induced alterations. Our laboratory has recently detected SP-A in PEx by ELISA (20). The advantage of PEx for monitoring alterations in the distal airways is that PEx, similarly to BAL, are collected from the target organ but that, in contrast to BAL, PEx collection is noninvasive, thus offering possibilities for studies of large populations.
We gratefully acknowledge Ulla Ruetschi, Clinical Neurochemistry Laboratory, University of Gothenburg, for discussion and assistance with experimental design, and we thank Aron Hakonen, Analytical Chemistry, University of Gothenburg, for spectrofluorometric analysis. We also thank personnel from Occupational and Environmental Medicine, University of Gothenburg, for their participation in PEx sampling.
↵3 Nonstandard abbreviations:
- respiratory tract lining fluid;
- bronchoalveolar lavage;
- exhaled breath condensate;
- particles in exhaled air;
- surfactant protein A;
- LTQ-FT ICR,
- linear quadrupole ion trap–Fourier transform ion cyclotron resonance;
- fragment ion mass spectrometry;
- exponentially modified protein abundance index;
- Clara cell protein.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the Disclosures of Potential Conflict of Interest form. Potential conflicts of interest:
Employment or Leadership: None declared.
Consultant or Advisory Role: None declared.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: A.-C. Olin (principal investigator), Swedish Heart-Lung Foundation (254212008) and FORMAS (254212006), administered by the University of Gothenburg.
Expert Testimony: None declared.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.
- Received for publication May 23, 2011.
- Accepted for publication October 11, 2011.
- © 2012 The American Association for Clinical Chemistry