|Home | About | Journals | Submit | Contact Us | Français|
As pivotal immune guardians, B cells were found to be directly associated with the onset and development of many smoking-induced diseases. However, the in vivo molecular response of B cells underlying the female cigarette smoking remains unknown. Using the genome-wide Affymetrix HG-133A GeneChip® microarray, we firstly compared the gene expression profiles of peripheral circulating B cells between 39 smoking and 40 non-smoking healthy US white women. A total of 125 differential expressed genes were identified in our study, and 75.2% of them were down-regulated in smokers. We further obtained genotypes of 702 single nucleotide polymorphisms in those promising genes and assessed their associations with smoking status. Using a novel multicriteria evaluation model integrating information from microarray and the association studies, several genes were further revealed to play important roles in the response of smoking, including ICOSLG (CD275, inducible T-cell co-stimulator ligand), TCF3 (E2A immunoglobulin enhancer binding factors E12/E47), VCAM1 (CD106, vascular cell adhesion molecule 1), CCR1 (CD191, chemokine C-C motif receptor 1) and IL13 (interleukin 13). The differential expression of ICOSLG (p = 0.0130) and TCF3 (p = 0.0125) genes between the two groups were confirmed by realtime reverse transcription PCR experiment. Our findings support the functional importance of the identified genes in response to the smoking stimulus. This is the first in vivo genome-wide expression study on B cells at today’s context of high prevalence rate of smoking for women. Our results highlight the potential usage of integrated analyses for unveiling the novel pathogenesis mechanism and emphasized the significance of B cells in the etiology of smoking-induced disease.
Cigarette smoking, the leading cause of death in USA, accounts for ~440,000 deaths and costs more than 167 related billion dollars, annually (Marshall et al. 2006). Smoking is the predominant cause of lung cancer (Alberg et al. 2007; Gandini et al. 2008) and highly associated with the onset and development of cancers, stroke, heart attack, vascular disease, respiratory diseases, and other severe chronic disease, such as chronic obstructive pulmonary disease (COPD) (Bergen and Caporaso 1999).
The components of inhalational smoke enter the body through the pulmonary alveoli and are distributed throughout all tissues by blood via capillary vessel. Thus, the smoking components travel within the blood, which contains immune cells such as B cells, T cells, monocytes, and other active components. These immune guardians, such as B cells, may react to smoking stimulus and exert on other organs and tissues with systematical effects (Buttner et al. 2007). Molecular studies showed that several nicotinic receptors up-regulated in the B cells in the presence of smoking components (Cloez-Tayarani and Changeux 2007; Skok et al. 2003). Long-term exposure to nicotine can induce B cells to decrease the antibody secretion, inhibit the cell proliferation and development, and finally suppress the normal immune function (Skok et al. 2003, 2005, 2007).
As a result, smoking may induce abnormal changes in the immune system and cause diseases via the dysregulation of impaired B cells. Particularly, many studies have demonstrated that smoking can suppress the normal function of B cells (Chan et al. 1990;Geng et al. 1995; Moszczynski et al. 2001; Savage et al. 1991; Yang et al. 1999) and hence increase the risk of COPD and emphysema (Brandsma et al. 2008; van der Strate et al. 2006), asthma (Tsoumakidou et al. 2007), chronic periodontitis (Al-Ghamdi and Anil 2007; Barbour et al. 1997), and inflammatory reaction (Bosken et al. 1992). The healthy consequences of smoking for women are worse than men (Mackay and Amos 2003). Cohort studies showed that female smokers appear to have increased susceptibility to tobacco carcinogens (Henschke et al. 2006). Women are more vulnerable to the cigarette-smoking-induced respiratory symptoms (asthma, chronic bronchitis, wheezing, and/or breathlessness) than men (Langhammer et al. 2000). Smoking affects the fertility of women and causes early menopause (Langhammer et al. 2000). In the USA and the UK, cigarette smoking becomes the single most important preventable cause of premature death in adult women, accounting for at least a third of all deaths in the age from 35 to 69 (Amos 1996).
Microarray has been employed to provide novel clues of the pathogenesis of smoking-induced disease in given specific samples. The peripheral mononuclear white cells (PMWCs), which contain mixed immune cells, had already been used for gene expression studies in response to smoking in vitro (Ryder et al. 2004; van Leeuwen et al. 2005, 2007). A pilot in vivo study further purified the fresh T cells from PMWCs and applied microarray to investigate the role of T cells in male smokers (Buttner et al. 2007). The differentially expressed genes (DEGs) identified by those studies contributed to illuminating the involvement of immune cells in smoking-induced reactions. Although many clinical and cohort studies indicated that women are more sensitive to smoking than man, studies seldom focused on the in vivo molecular responses and interactions of genes for the immune cells underlying the female smoking behavior. This may be because, in many countries (especially in the developing countries), cigarette smoking still tends to be regarded as a major male healthy issue (Mackay and Amos 2003; Amos 1996). One other clinical reason could be because women have a lower rate of death of lung cancer compared with men after diagnosis (Henschke et al. 2006), which appeared to relieve the serious results of smoking in women to a certain extent.
In this study, using the genome-wide Affymetrix HG-133A GeneChip® microarray, we compared the gene expression profiles of peripheral circulating B cells between smoking and non-smoking white women in the USA. Knowledge about the identified DEGs may fill the gaps of existing female smoking studies and further contribute to our understanding of pathogenesis of disease induced by smoking in the context of an increasing prevalence rate of cigarette smoking among female population (Bergen and Caporaso 1999). This is the first genome-wide expression study on in vivo human B cells relating to smoking in females. A novel ICOSLG, TCF3, and vascular cell adhesion molecule 1 (VCAM1) gene network was suggested in B cells for the response of smoking-induced reactions. The results can also be used to accelerate the progress of gender-specific treatment or approaches for effective tobacco control.
The study was approved by the Institutional Review Board, and all the subjects signed informed consent documents before entering the project. All the study subjects were US whites of European origin living in Omaha, NE and its surrounding areas.
We recruited 80 unrelated healthy women age 40–60, comprising 40 smokers and 40 non-smokers. The 40 subjects in each group can be also equally categorized to 20 premenopause and 20 postmenopause women. This recruitment strategy was adopted to avoid the potential effects on B cell gene expression profiling due to different menopausal status. A total number of 79 subjects were finally used for this study. We dropped one individual because the corresponding microarray chip was failed or damaged with abnormal background signal and thus cannot pass the strict quality control of dChip software. Smoking-related data were recorded using a nurse-administered questionnaire, which also included a detailed medication and disease history. Subjects were categorized as “smokers” based on recorded answers of smoking history in the questionnaire. The 39 current smokers were actively smoking 1.0±0.5 pack/day and had a average 31.89±9.37 years (mean ± SD) smoking history with the range from 14 to 48 years. In the nonsmoking group, only one individual (Nsm-52, 56 years old) smoked before when she is 19 and quit at her age of 29 years. After carefully inspecting, the other 39 women never smoked before. All of these 40 subjects were thus defined as “non-smokers.” There were no differences in age (p>0.7) between the smokers (49.37±8.05 years) and non-smokers (49.98±7.92 years). Additional strict exclusion criteria elaborated in Appendix 2 of the Electronic supplementary material were used to minimize other potential effects of non-smoking factors on B cell gene expression changes.
A random sample containing 840 unrelated whites (417 men and 423 women) was identified from our established and expanding genetic repertoire currently containing more than ~10,000 subjects. Similarly, subjects were categorized as “smokers” based on their answer in the questionnaire. A subject who had never smoked was defined as a “non-smoker.” For non-smokers, we excluded those subjects younger than the age of 25. There are 224 and 154 qualified smokers in the male and female subgroup, respectively. We documented the detail subject characteristics in our recent genome-wide smoking association study published elsewhere (Liu et al. 2009a).
Seventy milliliters of blood were drawn from each recruited female. B cell isolation from 70 ml whole blood was performed using a positive isolation method with Dynabeads® CD19 (Pan B) and DETACHaBEAD® CD19 (Dynal Biotech, Lake Success, NY, USA) following the manufacturer’s protocols. B cell purity was assessed by flow cytometry (BD Biosciences, San Jose, CA, USA) with fluorescence-labeled antibodies, PE-CD19 and FITC-CD45. The average purity was 96.3% with <1% deviation.
Total RNA from B cells was extracted using Qiagen RNeasy Mini Kit (Qiagen, Inc., Valencia, CA, USA). Total RNA concentration and integrity were determined by an Agilent 2100 Bioanalyzer (Agilent, Palo Alto, CA, USA). Each RNA sample has an excellent integrity number >9.0 in this study, indicating that RNA degradation due to processing was minimal and negligible. Finally, a 4 µg total RNA/sample was used for the production of complementary RNA (cRNA). The production of cRNA, hybridization, and scanning of the HG-133A GeneChip® were performed according to the manufacturer’s protocol (Affymetrix, Santa Clara, CA, USA), as we did before (Lei et al. 2009; Liu et al. 2005).
Two-step real-time reverse transcription polymerase chain reaction (RT-PCR) was used to verify the differentially expressed genes identified from the analyses of chip experiments. All reactions were run in triplicates for each gene. The first step is RT for synthesis of complementary DNA from total RNA, and the second step is real-time quantitative PCR. All the RT reagents were supplied by Taqman® Reverse Transcription Reagents (Applied Biosystems, Foster City, CA, USA). Real-time quantitative PCR was performed in a 25-µl reaction volume using standard protocols on an Applied Biosystems 7900HT. The experimental procedures were followed as the manufacture’s protocol and elaborated in Appendix 2 of the Electronic supplementary material.
Genomic DNA was extracted from whole human blood. We used the genotyping platform Affymetrix Mapping 250K Nsp and 250K Sty arrays to conduct the single nucleotide polymorphism (SNP) genotyping according to the standard protocol recommended by the manufacture. The SNP density of this platform can cover the whole genome. However, we only choose 702 SNPs, which covered most of our identified DEGs for further candidate gene association (some of the DEGs were not covered by the Affymetrix 500K chips). All the genotyping results has been strictly tested, and only those that passed the quality control will be involved in our later association analysis (Liu et al. 2009a,b,c; Xiong et al. 2009).
We submitted our gene expression profiling to Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/), and the access number was GSE18723. Microarray Suite 5.0 (MAS 5.0, Affymetrix, Santa Clara, CA, USA) software was used to generate array raw data in CEL files. A preselecting procedure adopted by our previous study (Liu et al. 2005) was also used here. We selected a subset of 7,215 probe sets from a total of 22,283 for ~14,500 genes in the HG-133A array. The selection criteria firstly remove those unreliable probes that did not express in B cells (MAS 5.0 generated fluorescence intensities value <150) or overexpress due to non-specific hybridization (>10,000). Then, the second process minimizes the multiple-testing problem by excluding the genes with relatively constant expression value across the samples from further analysis (Liu et al. 2005; Golub et al. 1999). Thus, we actually tested those probes with more biological information (large variance). The approaches employed here can improve the statistical power. Subsequent data analyses were performed on these selected probe sets. The fold change of each probe set was calculated by comparing the mean expression value given by MAS 5.0 in all smokers with the mean value in non-smokers.
We applied multiple analytic approaches including mas5.0 algorithm (Affymetrix, Santa Clara, CA, USA), robust multiarray algorithm (RMA) (Irizarry et al. 2003), the improved GC-content adjusted RMA (GCRMA) (Wu et al. 2004), and dChip (Li and Hung Wong 2001; Li and Wong 2001) for converting and normalizing our raw probe data to gene expression values, respectively. The expression data generated by mas5.0, RMA, GCRMA, and dChip were analyzed respectively by Bioconductor’s Multitest package to identify DEGs between the smoking and the nonsmoking groups using t test. The Benjamini and Hochberg (BH) procedure (Benjamini and Hochberg 1995) was used for multiple-testing adjustment, and the adjusted p<0.05 was used as the significant criterion.
Finally, we set our criteria for identifying the significant candidate genes (probe sets). If a gene was commonly suggested by all four methods (mas5.0, RMA, GCRMA, and dChip) (BH adjusted p < 0.05) and whose expression value was changed by more than ±1.5-fold, we selected this one as the identified DEG. This combination of multiple analytic approaches was used in order to minimize the number of false positives due to potential bias of different algorithms and non-specific hybridization.
According to the similarity in our samples and gene expression, the identified DEGs were further clustered and visualized by Cluster&TreeView software (Eisen et al. 1998), using hierarchical method with mas5.0 normalized data in two dimensions at both the gene and sample levels (Getz et al. 2000; Wu and Dewey 2006). To gain an overall picture of potential functions of the DEGs, we classified the genes according to four organizing principles (biological process, molecular function, cellular component, and the involved pathways) of the gene ontology (GO) database (http://www.geneontology.org/) by Onto-Express (http://vortex.cs.wayne.edu/ontoexpress/) and Pathway-Express (http://vortex.cs.wayne.edu/ontoexpress/).
Briefly, the multicriteria evaluation method can integrate multiple criteria (information) and finally get an optimal decision. We used this model in order to get the optimal and reliable functional gene network, which associate with female smoking. The results of both candidate gene SNP association analysis (p values) and gene functional database network analysis (database scores) were served as multiple criteria and combined into the model. The whole procedures of applying the model were elaborated in Appendix 2 of the Electronic supplementary material.
The association statistical analyses were carried out with SAS software (SAS Institute Inc., Cary, NC, USA). Genotypic association analysis was conducted with logistic regression for the association with smoking status (dependent variables) after testing the potential population stratification. Two methods were employed to crosscheck the population stratification: the Structure 2.2 software (http://pritch.bsd.uchicago.edu/software.html; Pritchard et al. 2000), which used Markov chain Monte Carlo algorithm to cluster individuals into different cryptic subpopulations on the basis of 200 random unlink SNPs data. Then, the genomic control method was used to generate the inflation factor (Devlin and Roeder 1999). Ideally, for a homogeneous population with no stratification, the inflation factor should be equal to or near 1.0. For our total sample, the estimated value was 1.009. Both of the two methods indicated essentially no substructure in our sample.
The effects of covariates like age and/or sex were adjusted by comparing with the corresponding restricted model. In the first original model, both covariates and genotype were included, and in the restricted model, only the effects of the covariates were estimated. We compared these two models to assess the effects of the genotype. The significance of the covariate-adjusted association, which is the difference in likelihood of the two models, was tested using a chi-squared test (Liu et al. 2009a). The detailed whole procedures were elaborated in our smoking association study (Liu et al. 2009a).
To investigate the direct (physical) and/or indirect (functional) interactions among the genes identified, we use the STRING database (http://string.embl.de/) for functional network analysis. The STRING database provided a score for each gene–gene interaction, which is computed as the joint probability of the probabilities from the different evidence channels (protein interaction, fusion, co-expression, text mining, etc.), correcting for the probability of randomly observing an interaction (von Mering et al. 2005). The high database score means that there are more experimental or predicted evidences for gene–gene functional relationship (interaction).
Briefly, the Student’s t test was employed to compare the relative gene expression [i.e. 2−ΔCT,where ΔCT (amplified cycle threshold) = (CTTarget Gene − CTGAPDH control)] between the smoking and non-smoking groups (Liu et al. 2005). For the ICOSLG gene, the subjects Nsm-29, Nsm-58, and Sm-75 were discarded for t test due to the amplification failed. For the TCF3 gene, the subjects Nsm-10, Sm-20, Nsm-29, Nsm-58, Sm-63, Sm-66, Sm-67, Sm-75, and Sm-89 were discarded for the same reason.
On average, 35.16±1.96% of the total of 22,283 probe sets in the array were called “present” for our samples based on the analysis with the MAS 5.0 suite software. According to our criteria (Appendix 1, Table A1 of the Electronic supplementary material), we identified 125 genes differentially expressed between the smoking and non-smoking groups, among which 94 genes (75.2%) were down-regulated (fold change ≤ −1.5) in the smoking sample, and other 31 genes were up-regulated (fold change ≥ +1.5) in smokers.
Figure 1 shows the results of the clustering analyses of the DEGs. Seventy-nine studied subjects were separated into two distinct major clusters by the unsupervised clustering analysis. Only six individuals (7.6%) were incorrectly categorized. Some important genes discussed latter in our study were also marked by rectangle in Fig. 1. We can see similar gene expression pattern among these genes. Particularly, the ICOSLG and VCAM1, the IL13 and MUC5AC, the ARF6 and CCR1 paired with each other and presented a very close distance. Other genes were distributed in many clusters. The TCF3 and the GATA2 genes remained relatively independent.
Figure 2 represents the results of the GO enrichment and pathway enrichment analyses. In the “molecular function” principle, functions of the 125 genes were focused on the protein binding and nucleotide binding. The most significant ontological term was “oxygen transporter activity” with a p < 1.00×10−4, which indicated that smoking may cause potential oxidative stress on the B cells. The p value was provided to determine the significance of probability that the ontological category of genes in the data set is explained by chance alone. In the “biological process” principle, the functions were mainly on immune response. In the “cellular component” principle, the products of those genes were primarily located on plasma membrane and cytoplasma. The pathway enrichment result showed the identified gene enriched in the cytokine–cytokine receptor interaction pathway.
A total of 702 SNPs of our identified DEGs were queried for association analysis. There are 30 and 24 DEGs associated with smoking in total sample and female subgroup, respectively. Nineteen genes among them were significant in both total and female samples (overlapped). Using STRING to further analyze the 125 DEGs, a functional associated network was constructed (Fig. 3). This network includes 40 genes, and 29 of them were DEGs. Information regarding the involved genes with the ontology results was summarized in Table 1 and Appendix 1, Table A2 of the Electronic supplementary material.
The multicriteria evaluation model generated a weighted sum score C for each DEG, based on the association p value and the STRING database score. This score simultaneously measured the significance of DEGs in both association and STRING pathway analysis. Generally, the higher this score, the more significant of DEGs in our combined analyses (Appendix 2 of the Electronic supplementary material). We ranked their C scores from the largest to zero and provided the top 30 results in Appendix 2 of the Electronic supplementary material. Obviously, the genes that associated with smoking and/or actively interacted with others had a high C score, such as IL13, TCF3, GATA2, MYH11, VCAM1, ARF6, GIT2, CCR1, and ICOSLG (Appendix 2, Table A2 of the Electronic supplementary material).
Interestingly, the results showed that VCAM1, CCR1, and GATA2 were highly associated with smoking status in both total and female sample (Table 2). The association of ICOSLG gene was detected only in the female subgroup (Table 2). Other SNPs with known function were marked in Table 2. Many functionally important genes are included in both the network and the meaningful ontology categories (Table 1).
Based on the results of evaluation model and prior known functional relevance of DEGs, we built the potential biological relationships and network for the most promising genes with their immune functions in B cells (Fig. 4). In this network, there are 12 DEGs induced by the smoking stimulation in women, which influence broad aspects including the cell development, the cytokine secretion, the signaling transduction, the chemotactic response, and the B cell lineage commitment.
We also provided the association signals of male subgroup in Table 2 in order to investigate whether the identified DEG can effect on both sexes or not. Interestingly, the significant SNPs (p<0.05) effect on men cannot be detected in women, which indicated that our study probably mainly identified the female-specific results. For each gene, based on only the association signals, the VCAM, CCR1, and CPM may associate with male smoking.
Based on the limited total RNA quantity and primer assay availability, their highly statistical significance of differential expression, their functional importance in our multicriteria evaluation analyses, and their prior known function relevance to smoking response, we selected the ICOSLG and TCF3 genes for replication. The relative expression level was 0.130±0.096 (mean ± SD) in smoking group and 0.219±0.176 in non-smoking group for the ICOSLG gene. For the TCF3 gene, it was 0.425±0.238 and 0.581±0.245, respectively. These two genes were significantly down-regulated in the smoking group with p = 0.0130 (ICOSLG) and 0.0125 (TCF3), respectively.
B cells, as major producers of antibodies in the human body (Parkin and Cohen 2001), may be involve in the development of diseases induced by cigarette smoking (Chan et al. 1990; Tsoumakidou et al. 2007; Kidney et al. 1996; Regius et al. 1990; Vignes et al. 2000). A common hypothesis for the onset of these diseases is that smoking influences the functions of B cells and thus inhibits the production of protective immunoglobulins and suppresses normal immune response. In this study, we identified 125 DEGs, 75.3% of them were down-regulated in the smokers, which indicated a broad suppression of smoking stimulating on B cells. Based on the functional and replication analyses, we suggested a potential novel B cell-mediated pathophysiological mechanism for the etiology of smoking-induced diseases. The down-regulation of genes in the functional network may systematically inhibit different immune functions in B cells (Fig. 4). The potential oxidative stress on the B cells may also play roles in the development of microvascular and small airway complications (Brandsma et al. 2008; Hasnis et al. 2007).
VCAM1 (CD106) is a member of the Ig superfamily and encodes a surface adhesion protein, which mediates cell adhesion and signal transduction (Mebius 2003). At the early stage of B cell development, VCAM1 adheres the pro-B cells to stromal cells, which is one of the important steps for pro-B cells evolving to pre-B cells (Funk et al. 1994). Thus, the down-regulation of VCAM1 in smokers’ B cells revealed a suppression of immune signal transduction and may interrupt the cellular development.
Furthermore, it has been demonstrated that VCAM1 directly interacts with the IL13 gene (Fig. 3; Sironi et al. 1994). In our study, the down-regulation of VCAM1 (fold change = −1.50) in smokers was correlated with the down-regulation of IL13 (fold change = −1.56) as expected. IL13 is a well-known candidate gene for the smoking-exposed COPD, asthma, and other serious respiratory disease (Cozen et al. 2004; Feleszko et al. 2006; Lee et al. 2007; Sadeghnejad et al. 2007, 2008; Zheng et al. 2000). It encodes an immunoregulatory-stimulating cytokine, which involves in several stages of B-cell differentiation and maturation (Cocks et al. 1993; Defrance et al. 1994; McKenzie et al. 1993). The down-regulation of IL13 (fold change = −1.56) may cause dysregulation of major histocompatibility complex class II expression, inhibit the IgE class switching, and interrupt the inflammatory cytokine production (Wynn 2003). We speculated that the suppression would significantly reduce the normal cell number and activity of B cells under the stimulation of smoking.
The down-regulation of ICOSLG in B cells (fold change = −1.68) is a signature of in vivo clusters of B cells and plasma cells in close contact with T cells (Hutloff et al. 2004). Down-regulation of ICOSLG in our study indicated the interaction of B cells with T cells, thus may mediate the T cell-driven B cell activation process and force the formation of memory B cells and plasma cells (secreting the antibodies; Hutloff et al. 2004). In addition, both down-regulation of ICOSLG and IL13 may be critically relevant in inducing an acute immune response and perpetuating inflammation in chronically immune-mediated disorders (Hu et al. 2007).
Notably, another inhibited (fold change = −1.50) gene ARF6 associated with the formation of plasma cell was also identified in our study. ARF6 was responsible for vesicular transport and thus is important for the formation of plasma cell from B cells (Yang et al. 1998). The suppression of ARF6 may influence cell-mediated immune response and differentiation of other related immune cells (e.g., monocyte) (Yang et al. 1998).
Chemokines and their receptors perform the roles of signal transduction and are critical for the recruitment of immune cells to the site of inflammation. CCR1 as one of the chemokine receptor also enhanced the migration and circulation of B cells (Corcione et al. 2002). A high level expression of the CCR1 may improve the chemotactic responses for B cells, such as facilitating their preferential localization in lymphoid tissues (Ehrhardt et al. 2005) and infiltrating to the site of inflammation or infection (Noda et al. 2007). A significant low expression of CCR1 (fold change = −2.10) in our study thus suggested a limited ability of chemotactic responses in smokers. Therefore, one of the possible reasons for the high frequency of chronic inflammations or infections that occurred in smokers may be attributed to the ability of B cells migrating or localizing to the target site was weakened or even damaged.
On the other hand, one of the up-regulated gene (fold change = +2.00) GIT2 (G protein-coupled receptor kinase interacting ArfGAP 2) may impair the motility of B cells because only the down-regulation of GIT2 can enhance the cell motility (Frank et al. 2006).
TCF3 gene, which regulates the development of B cells (Bradney et al. 2003), was also identified to be suppressed in the smokers. TCF3 served as pivotal regulator for different sets of orchestrated and stage-specific genes in lineage commitment and differentiation of B cells (Gisler and Sigvardsson 2002; Greenbaum et al. 2004). The down-regulation of TCF3 gene usually happened in the old B cells or elderly people, which indicated the reduced protection to infectious agents (Frasca et al. 2008). More importantly, down-regulated TCF3 gene associated with the decline in the percentage of memory B cells but an increase in that of naive B cells (Frasca et al. 2008). Therefore, the suppressed TCF3 gene (fold change = −1.56) in smokers indicated a B cell defect or overaged in the immune system.
TCF3 also interacted with GATA binding protein 2 (GATA2) gene (Fig. 3). GATA2 gene is expressed in multipotent progenitors prior to many kinds of cell lineage commitment (Murrell and Green 1995). Induced expression of GATA2 promotes proliferation and differentiation (Briegel et al. 1993). Down-regulated GATA2 with TCF3 provided novel clues for factors regulating the B cell lineage commitment.
Some risk genes of obstruction disease in the respiratory and circulating system were identified here, such as VCAM1 (Cavusoglu et al. 2004), IL13 (Lee et al. 2007), and MUC5AC (Kraft et al. 2008). Smoking can up-regulate the VCAM1 gene in endothelial cells and may cause atherosclerosis due to the vascular cellular overadhesion (Cavusoglu et al. 2004; Cirillo et al. 2007). In addition, smoking can up-regulate IL13 and induce the overexpression of mucus (MUC5AC protein) in the airway epithelial cells of asthma and COPD patients (Nakao et al. 2008).
However, our in vivo study revealed the down-regulation of the VCAM1, IL13, and MUC5AC in smoking-stimulated B cells. This phenomenon observed in women may suggest a different mechanism in B cells, which was different from the one in epithelial cells. Presumably, smoking-stimulated overexpression of VCAM1, IL13, and MUC5AC in endothelial cells surface, which leads to over accumulation of mucus and adhesion protein. On the other hand, smoking inhibited expression of VCAM1, IL13, and MUC5AC in B cells, which interrupted the normal immune signal transduction and response. The effects of suppressed immune function in B cells combined with the accumulated mucus and adhesion proteins in endothelial cells surface may induce the obstruction in the respiratory and circulating system. Interestingly, supportive evidence showed that some smoking-induced diseases may involve interactions between immune cells and vascular endothelium (Cavusoglu et al. 2004). Our study further contributed useful clues for this pathogenesis.
In conclusion, we have characterized the genome-wide expression profile for the first time and, by extrapolation, defined the functional relationship of a specific set of genes from the B cells across a relatively large sample of females with/without smoking habit. Whether this mechanism is female-specific or can affect on both sex still needs further investigation.
We thank Dr. Peng Xiao for coordinating the RT-PCR experiment. This work was partially supported by the LB595 and LB692 grant from the State of Nebraska. The study was also benefited from support from Xi’an Jiaotong University, Shanghai Leading Academic Discipline Project (project number S30501) and the Ministry of Education of China. HWD was partially supported by grants from NIH (P50AR055081, R01AG026564, R01AR050496, RC2DE020756, R01AR057049, and R03TW008221) and Franklin D. Dickson/Missouri Endowment.
Electronic supplementary material The online version of this article (doi:10.1007/s00251-010-0431-6) contains supplementary material, which is available to authorized users.
Feng Pan, The Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, People’s Republic of China. Department of Biomedical Sciences and Osteoporosis Research Center, School of Medicine, Creighton University, Omaha, NE 68131, USA.
Tie-Lin Yang, The Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, People’s Republic of China. Department of Orthopedic Surgery and Basic Medical Science, School of Medicine, University of Missouri-Kansas City, 2411 Holmes Street. Room M3-C03, Kansas City, MO 64108-2792, USA.
Xiang-Ding Chen, Department of Biomedical Sciences and Osteoporosis Research Center, School of Medicine, Creighton University, Omaha, NE 68131, USA. Laboratory of Molecular and Statistical Genetics, College of Life Sciences Hunan Normal University, Changsha, Hunan 410081, People’s Republic of China.
Yuan Chen, The Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, People’s Republic of China. Department of Biomedical Sciences and Osteoporosis Research Center, School of Medicine, Creighton University, Omaha, NE 68131, USA.
Ge Gao, Department of Biomedical Sciences and Osteoporosis Research Center, School of Medicine, Creighton University, Omaha, NE 68131, USA.
Yao-Zhong Liu, Department of Orthopedic Surgery and Basic Medical Science, School of Medicine, University of Missouri-Kansas City, 2411 Holmes Street. Room M3-C03, Kansas City, MO 64108-2792, USA.
Yu-Fang Pei, The Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, People’s Republic of China. Department of Orthopedic Surgery and Basic Medical Science, School of Medicine, University of Missouri-Kansas City, 2411 Holmes Street. Room M3-C03, Kansas City, MO 64108-2792, USA.
Bao-Yong Sha, The Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, People’s Republic of China. Department of Biomedical Sciences and Osteoporosis Research Center, School of Medicine, Creighton University, Omaha, NE 68131, USA.
Yan Jiang, Institute of Systems Science, School of Management, University of Shanghai for Science and Technology, Shanghai 200093, People’s Republic of China.
Chao Xu, Institute of Systems Science, School of Management, University of Shanghai for Science and Technology, Shanghai 200093, People’s Republic of China.
Robert R. Recker, Department of Biomedical Sciences and Osteoporosis Research Center, School of Medicine, Creighton University, Omaha, NE 68131, USA.
Hong-Wen Deng, The Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an, Shaanxi 710049, People’s Republic of China. Department of Orthopedic Surgery and Basic Medical Science, School of Medicine, University of Missouri-Kansas City, 2411 Holmes Street. Room M3-C03, Kansas City, MO 64108-2792, USA. Laboratory of Molecular and Statistical Genetics, College of Life Sciences Hunan Normal University, Changsha, Hunan 410081, People’s Republic of China. Center of Systematic Biomedical Research, University of Shanghai for Science and Technology, Shanghai 200093, People’s Republic of China.