Figures
Abstract
The distribution of lung disease induced by inhaled cigarette smoke is complex, depending on many factors. With the knowledge that the small airway epithelium (SAE) is the earliest site of smoking-induced lung disease, and that the SAE gene expression is likely sensitive to inhaled cigarette smoke, we compared upper vs. lower lobe gene expression in the SAE within the same cigarette smokers to determine if the gene expression patterns were similar or different. Active smokers (n = 11) with early evidence of smoking-induced lung disease (normal spirometry but low diffusing capacity) underwent bronchoscopy and brushing of the upper and lower lobe SAE in order to compare upper vs lower lobe genome-wide and smoking-responsive gene expression by microarray. Cluster and principal component analysis demonstrated that, for each individual, the expression of the known SAE smoking-responsive genes were highly correlated in upper and lower lobe pairs, although, as expected, there were differences in the smoking-induced changes in gene expression from individual to individual. These observations support the concept that the heterogeneity observed among smokers in the anatomic distribution of smoking-induced disease are not secondary to the topographic differences in the effects of cigarette smoke on the airway epithelium.
Citation: Harvey B-G, Strulovici-Barel Y, Vincent TL, Mezey JG, Raviram R, Gordon C, et al. (2013) High Correlation of the Response of Upper and Lower Lobe Small Airway Epithelium to Smoking. PLoS ONE 8(9): e72669. https://doi.org/10.1371/journal.pone.0072669
Editor: Neeraj Vij, Johns Hopkins School of Medicine, United States of America
Received: April 18, 2012; Accepted: July 18, 2013; Published: September 9, 2013
Copyright: © 2013 Harvey et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the National Institutes of Health P50 HL084936, and UL1-RR024143; B-GH is supported, in part, by the JP Smith Clinical Scholar Award. AET was supported, in part, by National Institutes of Health, K23 HL103837. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
The human airways are complex, dichotomous branching structures with up to 223 branches from the trachea to the alveoli [1]. Extensive morphologic and, more recently, imaging analysis has emphasized regional heterogeneity in number, length, width and branching angles of the airways in the upper vs lower lung zones [2], [3]. This, together with upper vs lower lobe differences in blood flow, effects of gravity and differential effects of pressure-volume relationships modulate the differences in volume and flow of inspired air to the upper vs lower lobes [4]–[6]. These topographic upper vs lower lobe differences are relevant to how gases, xenobiotics and particulates suspended in inhaled air impact the airways to varying degree, depending on their size, charge and other properties [7], [8].
The classic example of these concepts relates to cigarette smoke, a complex mixture of gases and particulates [9], [10]. While the smoking-induced lung diseases emphysema and lung cancer exhibit anatomic heterogeneity [2], [3], [11]–[13], it is not clear whether this is secondary to the skewed anatomic distribution of inhaled cigarette smoke or whether this reflects increased susceptibility of some anatomic regions to these diseases [14]. The dispersion of cigarette smoke within the lung is very complex, dependent on individual inhalation parameters, breath holding, the properties of the individual components of cigarette smoke and the differences in lung structure and function among smokers [7], [15]–[17].
Based on the concept that the small airway epithelium (SAE) is the earliest site of smoking-induced lung disease [18]–[21], we have approached the question of topographic distribution of cigarette smoke vs topographic susceptibility by comparing the patterns of gene expression in the SAE of the upper vs lower lobes in the same cigarette smokers. Studies from our laboratory [22]–[26] and others [27], [28] have shown that cigarette smoking induces marked changes in the airway epithelial gene expression program of clinically healthy individuals. From our prior studies of occasional smokers and smokers exposed to second hand smoke, we know that the SAE is exquisitely sensitive to cigarette smoke, in that if nicotine or its derivatives can be detected in urine at any level, the SAE will exhibit up- and down-regulation of smoking-related genes [25]. In that context, the SAE is a sensitive “canary” that detects cigarette smoke and responds with changes in gene expression that are proportional to the extent of exposure to cigarette smoke up to a certain level of smoking, at which point the SAE gene expression response to smoking “saturates”, i.e., for active smokers, additional smoking does not enhance the response [25]. However, we also know that for active smokers, the response to smoking is variable among individuals in regard to the number of small airway epithelial genes responding to active cigarette smoking, i.e., while smokers as a group have changes in the SAE expression of hundreds of genes, some smokers have little response to smoking, while others respond with marked changes in gene expression. Thus, each smoker needs to be assessed individually.
With this background, we have used the sensitivity of the SAE to cigarette smoke to ask: what is the topographic impact of cigarette smoking on the upper vs lower lobe SAE in humans who smoke? To accomplish this, we assessed the impact of cigarette smoking on genome-wide and smoking-responsive gene expression in the SAE of the upper vs lower lobes of the same individuals, all with early manifestations of smoking-induced disease evidenced by normal spirometry, but low diffusing capacity (DLCO), i.e., smokers considered “normal” by the standard criteria for chronic obstructive pulmonary disease (COPD) [29], but with some evidence of smoking-related abnormal lung function.
Methods
Study Population
The protocol and informed consent for this study were approved by the Weill Cornell Medical College Institutional Review Board. Current smokers were recruited from the general population in New York City and written informed consent obtained. Individuals were evaluated at the Weill Cornell NIH Clinical and Translational Science Center and the Department of Genetic Medicine Clinical Research Facility. Each had a normal history, physical exam, complete blood count, coagulation studies, liver function tests, urine studies, chest X-ray, EKG and pulmonary function tests, except for an isolated reduction in DLCO (<80% predicted). All were negative for HIV1 and had normal α1-antitrypsin levels (see Methods S1 for full inclusion/exclusion criteria).
Chest Computed Tomography Scans
Chest high resolution computed tomography (HRCT) scans were used to determine the percentage of lung affected by emphysema in each of the subjects. The lung was divided into quartiles by lung volume, and the top and bottom quartiles were compared for % emphysema at −950 Hounsfield Units (HU). See Methods S1 for details regarding quantification of emphysema in the scans.
Upper and Lower Lobe Small Airway Epithelial Sampling
SAE (10th to 12th generation) was collected using flexible bronchoscopy as previously described [23]. The right lower lobe was brushed first, immediately followed by brushing of the right upper lobe. Cells from the two sources were processed separately (see Methods S1 for details). RNA was hybridized on Affymetrix HG-U133 Plus 2.0 arrays (Affymetrix Inc., Santa Clara, CA) with probes for >54,000 genome-wide transcripts, using Affymetrix protocols, hardware and software [30]. Microarray quality was verified by GAPDH 3′ to 5′ probe sets <3.0 and multi-chip normalization scaling factor <10.0 [31]. The raw data is publically available at the Gene Expression Omnibus (GEO) site (http://www.ncbi.nlm.nih.gov/geo/), accession number GSE26307.
Assessment of Gene Expression
Data were normalized using the MAS5 algorithm (Microarray Suite Version 5, GeneSpring version 7.3, Affymetrix) per chip to the median expression value of each sample and per gene to the median expression value of each gene across all samples. Genome-wide analysis was used to compare the expression in the upper vs lower lobe of the same individual. We performed a paired limma test (linear model for microarray data) to identify differentially expressed genes between upper and lower lobe samples [32]. Limma fits a linear model to the expression data of each gene and uses an empirical Bayes approach to shrink the sample variance towards a common estimate. This has the advantage of making the analysis more stable and allows for more robust inference, even for experiments with a small number of samples. In addition, in order to reduce the number of false-negatives due to the low number of samples, we applied a “q-value” method that allows control of the false-discovery rate by controlling for p value distribution [33]–[35]. As an additional statistical test, we applied the SMVar (Structural Model for Variances) method, which also allows implementation of a moderated t-test for paired data. SMVar assumes a linear mixed model on the logarithm of the residual variances to be a function of both gene and condition, therefore allowing us to estimate gene-specific residual variances across the entire dataset [34], [36].
Expression values were compared using 3 lists of probe sets: (1) smoking-responsive probe sets (“smoking-responsive list”, n = 529 probe sets, using the small airway epithelial smoking-responsive gene list of Strulovici-Barel et al [25], corresponding to 372 unique genes); (2) 100 genome-wide random gene lists of 529 probe sets present in at least 20% of the upper or lower lobe samples, smoking-responsive list excluded (“100 random lists”); (3) based on the subjects in this study, we identified smoking-responsive probe sets significantly differentially expressed in smokers with early emphysema vs healthy nonsmokers, n = 228 probe sets (comparing 11 smokers with early emphysema vs 60 healthy nonsmokers (see Wang et al. [37] for nonsmoker inclusion criteria and demographics), using all probe sets present in at least 20% of the smoker or nonsmoker samples; significantly expressed probe sets were selected using the same criteria used to create the smoking-responsive list, i.e., probe sets with fold-change ≥1.5 between the groups, p<0.01 with Benjamini-Hochberg correction). These genes were functionally annotated using the NetAffx Analysis Center (www.affymetrix.com) to retrieve Gene Ontology (GO) annotations.
Statistical Analyses
Response of the transcriptomes of the upper and lower SAE was assessed using univariate and multivariate approaches (see Methods S1 for details).
Results
Small airway epithelial samples were obtained from the upper and the lower lobes of 11 current smokers, all with normal spirometry, but low DLCO (Table I, Figures 1, 2 A,B). All collected samples were comparable in number and purity of the recovered cells. There was no difference in the % ciliated cells recovered, but there were significantly more secretory and undifferentiated columnar cells and less basal cells in the upper vs lower lobe (p<0.01, both comparisons, Figure 1C). Quantification of the % emphysema present in the upper vs the lower quartiles of the chest HRCT scans for each individual showed overall no significant differences between the upper lobe and the lower lobe % emphysema (p>0.8, Figure 2 C). However, as has been observed by others assessing early disease [12], [13], [38], [39], there was heterogeneity in anatomic distribution among the subjects, with some with upper lobe dominance, others with lower lobe dominance and others with similar upper and lower lobe % emphysema.
A, B. Fluoroscopy of the upper and lower lobes showing the position of the brush used for sampling (arrows). C. Epithelial cell types in the upper vs lower lobes. Shown is data (mean ± standard error) based on counting 500 cells in cytocentrifuge preparations of upper vs lower lobe small airway epithelial samples. p values for paired t-test between upper and lower lobes are shown.
A. Forced expiratory volume in 1 sec (FEV1), forced vital capacity (FVC), and diffusing capacity (DLCO), all as % predicted. B. Ratio of FEV1/FVC as % observed. C. HRCT quantification of emphysema by −950 Hounsfield Units (HU) divided into top (upper lung zones) and bottom (lower lung zones) quartiles by lung volume.
Upper vs Lower Lobe Small Airway Gene Expression
Genome-wide analysis was used to compare the gene expression of the SAE in the upper vs the lower lobe. The expression levels of all probe sets present in at least 20% of the upper or lower lobe samples (31,883 probe sets) were compared using a paired t-test. After applying a multiple test correction (Benjamini-Hochberg), the paired t-test analysis did not return any significant differences for any probe sets (not shown). In order to compare the genes responding to cigarette smoking in the SAE of the upper vs lower lobe, the expression levels of the 529 probe sets in the smoking-responsive list [25] and the 228 probe sets significantly differently expressed between the smokers with low DLCO samples and healthy nonsmoker controls were compared using a paired t-test. After applying a multiple test correction (Benjamini-Hochberg), there were no significant differences in the probe sets for either list (not shown). These results were further validated with analysis using limma and SMVar differential gene expression methods.
A hierarchical clustering analysis of the 529 smoking-responsive list, using either Pearson or Spearman correlation measures, showed that every pair of smokers (i.e., 11 of 11) clustered together (Figure 3). To assess the significance of returning this many cluster pairs for a set of 529 probe sets, we compared this to the number of cluster pairs for each of the 100 random lists. Using the Pearson or Spearman correlation, none of the random lists produced 11 of 11 pairings (p<0.02, both correlations). These data support the concept that, for the smoking-responsive genes, a sufficient amount of smoke reaches the upper and lower lobes to induce a high correlation of expression of the smoking-responsive genes. Given that the cilia cell population is not significantly different in the upper vs lower lobe, which might drive the upper-lower lobe sample pairing, we confirmed that the smoking-responsive gene list did not consist of any cilia-related genes [40]–[42]; therefore, the perfect upper-lower pairing is solely due to the highly correlated SAE gene expression pattern of the smoking-responsive probe sets.
Shown is a cluster analysis of small airway epithelium 529 known smoking-responsive probe sets corresponding to 372 genes [25]. Note that the smoking-responsive gene expression of the upper vs lower lobes for each of the 11 subjects clusters together.
Visual inspection of the first three principal components of the PCA of the smoking-responsive list (Figure 4A) did not explain the high level of pair clustering in the hierarchical analysis, indicating that the pairing was being driven by more variation than captured in the first three components. To assess whether this was the case for the 529 probe sets in this smoking-responsive list, the distance in multivariate probe space between sample pairs was quantified by the length of the vector connecting the upper and lower lobe samples for each individual. The median of these 11 vector lengths for the smoking-responsive list was then compared to the medians of upper and lower sample pairs for 100 random lists (Figure 4B). This analysis showed that the median vector length between each upper and lower lobe pair for the smoking-responsive genes was far less than the median vector length when considering the 100 random lists. Thus, the cluster pairing was being driven by individual smoker overall distance in multivariate probe space and not just the first three principal component dimensions. Together, the cluster and the PCA distance in multivariate probe space support the concept that, for each smoker, the SAE of the upper and lower lobes is responding in a highly correlated way to inhaled cigarette smoke.
A. Principal component analysis (PCA) of the 529 probe sets of the smoking responsive list. Red = upper lobe samples; blue = lower lobe samples. The line connecting the red and blue data points connects the upper and lower lobe samples from the same individual. B. Plot of the distribution of the median distance between the upper and lower lobe samples for each individual in the space of 529 probe sets in the smoking- responsive list as measured by vector length (red circle) compared to the distribution of median distances between upper and lower lobe samples for 100 sets of 529 probe sets selected at random from the genome-wide list of probe sets present in at least 20% of the samples, minus the 529 smoking-responsive list (boxplot: middle bar is overall median, box contains the inner quartiles, and the whiskers, the 0.95 quartiles).
Discussion
The distribution of inhaled cigarette smoke is complex, dependent on a variety of physical, anatomic and physiologic factors, as well as the smoking habits of the individual smokers [4]–[8]. Based on the knowledge that smoking-induced lung disease starts in the SAE [18]–[21], and that smoking induces changes in gene expression in the SAE [23]–[26], we compared the patterns of gene expression in the upper vs lower lobes of the SAE of the same cigarette smokers. Interestingly, though there might be proteins preferentially inactivated by smoke in the upper lobe alone, the expression of SAE smoking-responsive genes was remarkably highly correlated in the upper vs lower lobes within each smoker, significantly more so than random genome-wide gene expression. This is consistent with the data showing that the SAE is exquisitely sensitive to cigarette smoke at low levels, with proportional increases in smoking-responsive gene expression up to a plateau, where active smokers saturate the small airway epithelial gene expression responses to cigarette smoke [25]. The data suggests that, independent of the complex factors that govern cigarette smoke distribution within the lung, the responses of the SAE to cigarette smoking in humans are highly correlated in the upper and lower lobes, and thus topographic cigarette smoke distribution is not responsible for the topographic differences in smoking-induced disease.
Airflow in the Human Lung
Studies using radioactive tracers, inhaled particulates, and chest CT scans have shown that distribution of ventilation in the lung is very complex, affected by differences in airway structure, shape and weight of the lung, gravity, posture, and gradients of pleural pressure [43]. In the upright position at rest, the transpulmonary pressure is less at the base and thus the lung at the base is less expanded compared to the apex. Consequently, the lung at the base has greater compliance, and greater relative ventilation. Studies with radioactive gases show that with greater inspiratory flow rate, the regional differences are less than with very slow inspiration [44]. Measurements with xenon CT scans show that ventilation is greater in the central lung and less in the periphery [45]. The effects of gravity on the distribution of ventilation are complicated, with the relative volume of lung in different regions competing with gravity-dependent changes [43]. In addition to the influence of gravity, many studies document that airway architecture also plays a major role in the distribution of air during respiration [4]–[6], [46]–[49]. For example, while branching of airways is invariably dichotomous, it is asymmetric, with the diameter of the 2 branches often unequal, with consequent different flow rates in the 2 branches [43].
Distribution of Cigarette Smoke
The distribution of the inhaled >4,000 gases, xenobiotics and particulates in cigarette smoke with 1014 oxidants/puff is complex, dependent on inhalation parameters of volume and flow, breath holding, the aerodynamic properties of the individual components of smoke and the complex structure and function of the airways in the different regions of the lung [7], [9], [10], [12], [14]. Of the particulates suspended in cigarette smoke, 10 to 90% will be deposited in the airways, depending on aerodynamic diameters [8]. A higher % of particulates are retained in the airways in smokers compared to nonsmokers [8].
The particle size distribution of main stream cigarette smoke has a median diameter of 0.18 to 0.34 µm [14]. Puffing flow rate and duration, interpuff interval and filters all affect the distribution of particles in cigarette smoke [7], [14]. On the average among smokers, there are 8 to 16 puffs/cigarette, 18 to 64 sec between puffs, 232 to 414 sec duration/cigarette, 1.6 to 2.4 sec puff duration, 21 to 66 ml puff volume, 28 to 40 ml/sec peak flow and 413 to 918 ml inhalation volume [17]. There are also differences in the inhalation patterns of smokers [14] as well as gender differences in cigarette puff volume and post-puff inspiratory volume [16]. The consequence of this variability is that there is a wide variation of the topographic distribution of inhaled cigarette smoke within the lung [50]. Other studies found a uniform distribution of ventilation during cigarette smoking [51], [52]. A pathology study assessing the small airways in the upper vs lower lobe of smokers with and without emphysema found no difference in inflammation, fibrosis, muscle hypertrophy, squamous metaplasia, intraluminal macrophages and pigment deposition, suggesting a uniform distribution of cigarette smoking [3].
Topographic Disease in Cigarette Smokers
Lung cancer associated with cigarette smoking tends to be in the upper lung zones, with an upper to lower ratio of 2.5 to 1 [11], [53]. It has been hypothesized that upper lobe predominance is secondary to distribution of cigarette smoke or that cigarette smoke toxins/carcinogens may persist longer in the upper lung zones due to relatively less ventilation or less efficient lymphatic clearance in the upper zones [2]. Another theory is that the upper lobe predominance of lung cancer is due to less efficient delivery of protective substances via the circulation to the upper vs lower lung zones [2].
In regard to non-malignant smoking-induced airway and alveolar disease, the data regarding regional differences in disease location is not as clear as with cancer. Morphologic studies demonstrate that the earliest lung abnormalities associated with smoking are in the SAE [18]–[21]. Quantitative morphologic studies have shown that the small airway disease associated with smoking is similar in the upper vs lower lung zones [3]. This is consistent with the observations in the present study that the smoking-induced changes in gene expression in the SAE are more highly correlated than random genes in the upper and lower lung zones.
In the later stages of smoking-related non-malignant lung disease, the data is variable, but suggests there is significant heterogeneity in the anatomic distribution of disease. For example, in autopsy cases representing late stage of disease, quantitative morphometry demonstrated upper lobe predominance [38]. In chest X-ray assessment of emphysema, the classic concept is that there is upper lobe predominance, but this conclusion may be, in part, because emphysema is more obvious in the upper lung zones because there is less tissue in the upper lobes [54].
The application of chest CT technology to this issue suggested no differences in the average % emphysema among different lung zones, but recognized there was heterogeneity among subjects, with some with upper lung predominance and others with primarily lower zone emphysema [13]. Consistent with the concept of heterogeneity in the anatomic location of emphysema, others have found homogenously distributed emphysema to be more common than upper or lower zone dominant emphysema [39]. In the National Emphysema Treatment Trial, emphysema was approximately 2 to 1 more prominent in the upper lobes [55]. In contrast, in the Dutch-Belgian Lung Cancer Screening Trial, for 72% of subjects, the topographic distribution was 2.5 to 1 more prominent in the lower lung zones, with 28% of subjects having a more even distribution [15]. More recently, de Torres et al [12], showed that 50% of the tested subjects has homogenously distributed emphysema; these individuals had milder disease than those with upper lung zone predominance. The reasons for the differences in these studies are likely complex, but the overall concept is that there is significant heterogeneity in the anatomic distribution of smoking-related emphysema, with the early disease more homogenous, and the later stages of the disease developing anatomic heterogeneity, with a subset with upper lobe predominance, others with lower zone predominance, and some remaining heterogeneous. This conclusion is consistent with the HRCT data in the present study which showed that, for smokers with mild disease, quantitative HRCT assessment shows heterogeneity in anatomic distribution.
Topographic Homogeneity of Airway Gene Expression in Cigarette Smokers
Despite the difference in the cell population recovered from the upper vs lower lobes in the present study, we observed that the expression of smoking-responsive genes in the SAE of the upper and lower lobes of smokers with mild lung disease is highly correlated. This is consistent with the data of Hackett et al [56] showing that, other than for a few “rapid response” genes that differentially respond to the timing of sample recovery, there was a similar pattern of gene expression in the right vs left lower lobe large airway epithelium. The observation that there are no topographic differences in lung gene expression in cigarette smokers was also described in a study of whole lung samples of nonsmokers, ex-smokers and current smokers, concluding that variance among individuals was far greater than between lobes of the same individuals [57].
Supporting Information
Methods S1.
Supplemental Methods and References.
https://doi.org/10.1371/journal.pone.0072669.s001
(DOC)
Acknowledgments
We thank D. Dang and M. Teater for help with the analyses; and N Mohamed for help in preparing this manuscript.
Author Contributions
Conceived and designed the experiments: B-GH RGC. Performed the experiments: YS-B BG-H CG AC. Analyzed the data: YS-B BG-H AET RR JS JGM RGC TLV. Contributed reagents/materials/analysis tools: BG-H RGC JGM. Wrote the paper: B-GH YS-B JGM RGC. Literature search: YS-B CG RGC AC AS.
References
- 1.
Weibel ER (1997) Design of airways and blood vessels considered as branching trees. In: Crystal RG, West JB, Weibel ER, Barnes PJ, editors. The Lung: Scientific Foundations. Philadelphia: Lippincott-Raven. pp. 1061–1071.
- 2. Lee BW, Wain JC, Kelsey KT, Wiencke JK, Christiani DC (1998) Association between diet and lung cancer location. Am J Respir Crit Care Med 158: 1197–1203.
- 3. Wright JL, Wiggs BJ, Hogg JC (1984) Airway disease in upper and lower lobes in lungs of patients with and without emphysema. Thorax 39: 282–285.
- 4. Glenny RW, McKinney S, Robertson HT (1997) Spatial pattern of pulmonary blood flow distribution is stable over days. J Appl Physiol 82: 902–907.
- 5. Glenny RW, Lamm WJ, Bernard SL, An D, Chornuk M, et al. (2000) Selected contribution: redistribution of pulmonary perfusion during weightlessness and increased gravity. J Appl Physiol 89: 1239–1248.
- 6. Glenny RW, Bernard SL, Robertson HT (2000) Pulmonary blood flow remains fractal down to the level of gas exchange. J Appl Physiol 89: 742–748.
- 7. Kane DB, Asgharian B, Price OT, Rostami A, Oldham MJ (2010) Effect of smoking parameters on the particle size distribution and predicted airway deposition of mainstream cigarette smoke. Inhal Toxicol 22: 199–209.
- 8. Lippmann M, Yeates DB, Albert RE (1980) Deposition, retention, and clearance of inhaled particles. Br J Ind Med 37: 337–362.
- 9. Hecht SS (2003) Tobacco carcinogens, their biomarkers and tobacco-induced cancer. Nat Rev Cancer 3: 733–744.
- 10. MacNee W (2000) Oxidants/antioxidants and COPD. Chest 117: 303S–317S.
- 11. Byers TE, Vena JE, Rzepka TF (1984) Predilection of lung cancer for the upper lobes: an epidemiologic inquiry. J Natl Cancer Inst 72: 1271–1275.
- 12. de Torres JP, Bastarrika G, Zagaceta J, Mendiguren RS, Alcaide AB, et al. (2011) Emphysema presence, severity and distribution has little impact on the clinical presentation of a cohort of patients with mild to moderate COPD. Chest 139: 36–42.
- 13. Saitoh T, Koba H, Shijubo N, Tanaka H, Sugaya F (2000) Lobar distribution of emphysema in computed tomographic densitometric analysis. Invest Radiol 35: 235–243.
- 14. Bernstein D (2004) A review of the influence of particle size, puff volume, and inhalation pattern on the deposition of cigarette smoke particles in the respiratory tract. Inhal Toxicol 16: 675–689.
- 15. Gietema HA, Zanen P, Schilham A, van Ginneken B, van Klaveren RJ, et al. (2010) Distribution of emphysema in heavy smokers: impact on pulmonary function. Respir Med 104: 76–82.
- 16. Nil R, Woodson PP, Battig K (1986) Smoking behaviour and personality patterns of smokers with low and high CO absorption. Clin Sci (Lond) 71: 595–603.
- 17. Surgeon General (1988) The health consequences of smoking: Nicotine addiction. A report of the Surgeon General
- 18. Cosio M, Ghezzo H, Hogg JC, Corbin R, Loveland M, et al. (1978) The relations between structural changes in small airways and pulmonary-function tests. N Engl J Med 298: 1277–1281.
- 19. Cosio MG, Hale KA, Niewoehner DE (1980) Morphologic and morphometric effects of prolonged cigarette smoking on the small airways. Am Rev Respir Dis 122: 265–21.
- 20. Hogg JC, Macklem PT, Thurlbeck WM (1968) Site and nature of airway obstruction in chronic obstructive lung disease. N Engl J Med 278: 1355–1360.
- 21. Hogg JC, Chu F, Utokaparch S, Woods R, Elliott WM, et al. (2004) The nature of small-airway obstruction in chronic obstructive pulmonary disease. N Engl J Med 350: 2645–2653.
- 22. Harvey B-G, O'Connor TP, Salit J, Raman T, Crystal RG (2008) Differences in gene expression of upper vs lower lobe small airway epithelium in individuals with an early emphysema phenotype and predominant upper lobe emphysema. Am J Respir Crit Care Med 177: A960.
- 23. Harvey BG, Heguy A, Leopold PL, Carolan BJ, Ferris B, et al. (2007) Modification of gene expression of the small airway epithelium in response to cigarette smoking. J Mol Med (Berl) 85: 39–53.
- 24. Hubner RH, Schwartz JD, De Bishnu P, Ferris B, Omberg L, et al. (2009) Coordinate control of expression of Nrf2-modulated genes in the human small airway epithelium is highly responsive to cigarette smoking. Mol Med 15: 203–219.
- 25. Strulovici-Barel Y, Omberg L, O'Mahony M, Gordon C, Hollmann C, et al. (2010) Threshold of biologic responses of the small airway epithelium to low levels of tobacco smoke. Am J Respir Crit Care Med 182: 1524–1532.
- 26. Tilley AE, Harvey BG, Heguy A, Hackett NR, Wang R, et al. (2009) Down-regulation of the notch pathway in human airway epithelium in association with smoking and chronic obstructive pulmonary disease. Am J Respir Crit Care Med 179: 457–466.
- 27. Beane J, Sebastiani P, Liu G, Brody JS, Lenburg ME, et al. (2007) Reversible and permanent effects of tobacco smoke exposure on airway epithelial gene expression. Genome Biol 8: R201.
- 28. Steiling K, Kadar AY, Bergerat A, Flanigon J, Sridhar S, et al. (2009) Comparison of proteomic and transcriptomic profiles in the bronchial airway epithelium of current and never smokers. PLoS One 4: e5043.
- 29. Rabe KF, Hurd S, Anzueto A, Barnes PJ, Buist SA, et al. (2007) Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: GOLD executive summary. Am J Respir Crit Care Med 176: 532–555.
- 30. Tumor Analysis Best Practices Working Group (2004) Expression profiling–best practices for data generation and interpretation in clinical trials. Nat Rev Genet 5: 229–237.
- 31. Raman T, O'Connor TP, Hackett NR, Wang W, Harvey BG, et al. (2009) Quality control in microarray assessment of gene expression in human airway epithelium. BMC Genomics 10: 493.
- 32.
Smyth GK (2005) Limma: Linear models for microarray data. In: Gentleman R, Carey V, Dudoit S, Irizarry R, Huber W, editors. Bioinformatics and Computational Biology Soluntions Using R and Bioconductor. New York: Springer. pp. 397–420.
- 33. Storey JD, Tibshirani R (2003) Statistical significance for genomewide studies. Proc Natl Acad Sci U S A 100: 9440–9445.
- 34. Jeanmougin M, de Reynies A, Marisa L, Paccard C, Nuel G, et al. (2010) Should we abandon the t-test in the analysis of gene expression microarray data: a comparison of variance modeling strategies. PLoS One 5: e12336.
- 35. Smyth GK (2004) Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 3: Article3.
- 36. Jaffrezic F, Marot G, Degrelle S, Hue I, Foulley JL (2007) A structural mixed model for variances in differential gene expression studies. Genet Res 89: 19–25.
- 37. Wang G, Xu Z, Wang R, Al-Hijji M, Salit J, et al. (2012) Genes associated with MUC5AC expression in small airway epithelium of human smokers and non-smokers. BMC Med Genomics 5: 21.
- 38. Saito K, Thurlbeck WM (1995) Measurement of emphysema in autopsy lungs, with emphasis on interlobar differences. Am J Respir Crit Care Med 151: 1373–1376.
- 39. Stavngaard T, Shaker SB, Dirksen A (2006) Quantitative assessment of emphysema distribution in smokers and patients with alpha1-antitrypsin deficiency. Respir Med 100: 94–100.
- 40. Avidor-Reiss T, Maer AM, Koundakjian E, Polyanovsky A, Keil T, et al. (2004) Decoding cilia function: defining specialized genes required for compartmentalized cilia biogenesis. Cell 117: 527–539.
- 41. Gherman A, Davis EE, Katsanis N (2006) The ciliary proteome database: an integrated community resource for the genetic and functional dissection of cilia. Nat Genet 38: 961–962.
- 42. Ostrowski LE, Blackburn K, Radde KM, Moyer MB, Schlatzer DM, et al. (2002) A proteomic analysis of human cilia: identification of novel components. Mol Cell Proteomics 1: 451–465.
- 43. Galvin I, Drummond GB, Nirmalan M (2007) Distribution of blood flow and ventilation in the lung: gravity is not the only factor. Br J Anaesth 98: 420–428.
- 44. Bake B, Wood L, Murphy B, Macklem PT, Milic-Emili J (1974) Effect of inspiratory flow rate on regional distribution of inspired gas. J Appl Physiol 37: 8–17.
- 45. Robertson HT, Kreck TC, Krueger MA (2005) The spatial and temporal heterogeneity of regional ventilation: comparison of measurements by two high-resolution methods. Respir Physiol Neurobiol 148: 85–95.
- 46. Glenny RW (1992) Spatial correlation of regional pulmonary perfusion. J Appl Physiol 72: 2378–2386.
- 47. Hlastala MP, Bernard SL, Erickson HH, Fedde MR, Gaughan EM, et al. (1996) Pulmonary blood flow distribution in standing horses is not dominated by gravity. J Appl Physiol 81: 1051–1061.
- 48. Hlastala MP, Chornuk MA, Self DA, Kallas HJ, Burns JW, et al. (1998) Pulmonary blood flow redistribution by increased gravitational force. J Appl Physiol 84: 1278–1288.
- 49. Hlastala MP, Glenny RW (1999) Vascular structure determines pulmonary blood flow distribution. News Physiol Sci 14: 182–186.
- 50. Jaques PA, Kim CS (2000) Measurement of total lung deposition of inhaled ultrafine particles in healthy men and women. Inhal Toxicol 12: 715–731.
- 51. Pearson MG, Chamberlain MJ, Morgan WK, Vinitski S (1985) Regional deposition of particles in the lung during cigarette smoking in humans. J Appl Physiol 59: 1828–1833.
- 52. Taylor DR, Reid WD, Pare PD, Fleetham JA (1988) Cigarette smoke inhalation patterns and bronchial reactivity. Thorax 43: 65–70.
- 53. Huhti E, Saloheimo M, Sutinen S, Reinila A (1983) Does the location of lung cancer affect its prognosis? Eur J Respir Dis 64: 460–465.
- 54. Hughes JA, Hutchison DC, Bellamy D, Dowd DE, Ryan KC, et al. (1982) Annual decline of lung function in pulmonary emphysema: influence of radiological distribution. Thorax 37: 32–37.
- 55. DeMeo DL, Hersh CP, Hoffman EA, Litonjua AA, Lazarus R, et al. (2007) Genetic determinants of emphysema distribution in the national emphysema treatment trial. Am J Respir Crit Care Med 176: 42–48.
- 56. Hackett NR, Heguy A, Harvey BG, O'Connor TP, Luettich K, et al. (2003) Variability of antioxidant-related gene expression in the airway epithelium of cigarette smokers. Am J Respir Cell Mol Biol 29: 331–343.
- 57. Gruber MP, Coldren CD, Woolum MD, Cosgrove GP, Zeng C, et al. (2006) Human lung project: evaluating variance of gene expression in the human lung. Am J Respir Cell Mol Biol 35: 65–71.