INTRODUCTION

TOP ABSTRACT INTRODUCTION GENERATION OF A HIGH-DENSITY... cDNA MICROARRAY IMAGING AND... PROFILING HYDROGEN PEROXIDE-... CONCLUSION REFERENCES

Keywords: airways; gene expression; microarray; oxidant injury; tobacco smoke

Large-scale study of the pattern of expression of thousands of genes is a hallmark of the transition from "structural" to "functional" genomics, in which knowing the complete DNA sequence of the genome will prove only the first step in understanding how specific genes contribute to normal and disturbed function. The next step, probably an even more challenging one, is to sort out how the biological function of these genes and the manner of their expression contribute to various biological processes. Central to the advancement of functional genomics is the development of high-density DNA microarray technology (1), which is able to profile simultaneously the pattern of expression of thousands of genes and that is also able to sort out the association between patterns of differential gene expression, specific biological events, and cellular processes.

Several DNA microarray systems are available. These include oligonucleotide arrays based on the "gene chip" concept (7, 10, 11) and developed by Affimatrix, and high-density DNA arrays on a glass slide (1) or on nylon membrane (9). All of these systems take the advantage of the high-density concept made possible by the efficiency of nucleotide hybridization kinetics. With computer-assisted software programs, images of the hybridization intensity of thousands of DNA dots on these gene chips can be quantified and clustered into various gene expression patterns (12). We previously developed a DNA microarray system based on spotting DNA on a positively charged nylon membrane (9). The reason for choosing nylon membrane over glass surface is that much greater amounts of DNA can be spotted on a nylon membrane than on a glass surface. Thus, hybridization kinetics is not limited by the amount of target DNA on the membrane (15).

The conducting airways are constantly exposed to various air pollutants, including those of the environmental oxidants tobacco smoke and ozone. As the first line of pulmonary defense, the airway epithelium is responsible for maintaining the mucociliary clearance that traps and removes various inhaled air particulate and infectious agents (16). However, this airway function is frequently impaired after an exposure to environmental oxidant pollutants. This impairment may permit more damage to the airway epithelium and to the distal lung by environmental air pollutants and the development of various pulmonary diseases (19). There is strong evidence to support the notion that the differential response by airway epithelial cells to these environmental insults is a reflection of differential gene expression produced by these cells. A functional genomic approach at the expression level will be able to examine the global change in gene expression by epithelial cells in response to environmental oxidants. As proof of concept, a high-density DNA microarray membrane was developed and used to profile differential gene expression in a human bronchial epithelial cell line in response to hydrogen peroxide and smoke treatments.

	GENERATION OF A HIGH-DENSITY DNA MICROARRAY MEMBRANE

TOP ABSTRACT INTRODUCTION GENERATION OF A HIGH-DENSITY... cDNA MICROARRAY IMAGING AND... PROFILING HYDROGEN PEROXIDE-... CONCLUSION REFERENCES

Various arrayers have been produced by commercial companies. These arrayers are able to robotically spot DNA at high density on a glass slide or on a nylon membrane. Most of the target DNAs are either from expressed sequence tag (EST) clones of the IMAGE (Integrated Molecular Analysis of Genomes and their Expression) Consortium, which was built to share cDNA clones and to make derived sequence, mapping, and expression data publicly available. Currently, ~ 45,000 UniEST clones with sequence verification are available from Research Genetics (Huntsville, AL). Our Uniclone project selected 9,600 nonredundant EST clones from this sequence-verified pool for our initial study. Most of these EST clones are known genes. A database file in a Microsoft Access form has been established for these clones and the database was linked to internet genome databases (e.g., NHLBI, Swiss Prot, and GeneCards) to update gene definitions. cDNA inserts in these EST clones were amplified by polymerase chain reaction (PCR) as described previously (16). The amplified target DNAs were then spotted at high density onto a nylon membrane with an "arrayer" robotic spotting machine. The Arrayer-02 or -03 (Wittech, Taipei, Taiwan), was capable of spotting DNA on an area smaller than 75 µm in diameter, with 100-150 µm between spots. The size of the DNA microarray membrane for 9,600 DNA spots was 1.8 × 2.7 cm.

	cDNA MICROARRAY IMAGING AND DATA PROCESSING

TOP ABSTRACT INTRODUCTION GENERATION OF A HIGH-DENSITY... cDNA MICROARRAY IMAGING AND... PROFILING HYDROGEN PEROXIDE-... CONCLUSION REFERENCES

The process of hybridization and detection of genes on filter membrane involved the standard molecular biology techniques of Northern and Southern blot hybridization analysis. For the DNA microarray membrane, a colorimetry detection method has been developed (9). This method was based on well-established quantitation methods traditionally used in quantitative protein measurement, such as enzyme immunoassay or enzyme-linked immunosorbent assay (23). cDNA probes labeled with digoxigenin- or biotin-dUTP were hybridized with the membrane. The hybridization results were then developed with a single or dual color through the enzyme-substrate reaction of color-forming enzymes. The dual chromogens were generated by first treating the membrane with 1 ml of 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal) substrate containing 1.2 mM X-Gal, 1 mM MgCl₂, 3 mM K₃Fe(CN)₆, and 3 mM K₄Fe(CN)₆ in 1× TBS buffer (10 mM Tris-HCl, [pH 7.4], 150 mM NaCl, 0.3% bovine serum albumin) for 45 min at 37° C. The membrane was then briefly rinsed with deionized water and subsequently stained with 1 ml of Fast Red TR/naphthol AS-MX substrate (Pierce, Rockford, IL) for alkaline phosphatase reaction. The color development reactions were then stopped by 1× phosphate-buffered saline (PBS) containing 20 mM EDTA.

We estimated that each 75-µm-diameter DNA spot on nylon membrane contained more than 10 ng of DNA, which corresponded to ~ 10⁹ molecules per spot (assuming a 1,000-bp DNA insert). This amount of target DNA was sufficient to carry out an intensity-based first-order kinetics hybridization of cDNA probes generated from mRNA templates, which was about 10⁶ molecules per reaction per template. The amount of target DNA on a glass surface is much smaller than that on a nylon membrane. Thus, the quantitation of the microarray data on glass surface is a ratio-based, competitive hybridization with two fluorescent cDNA probes (1).

After color development, the image of each DNA dot was digitized by a scan on a high resolution flat-bed scanner (UMAX [Fremont, CA] MagicScan at 3,000 dpi). These digitized images were separated into cyan, magenta, and yellow colors (Figure 1). On the basis of the way the human eye perceives color, a color could be described by three components: hue, saturation, and brightness. A slight change in any of these three components resulted in a perceivable difference. By utilizing true-color signals, the digitized image could be reasonably quantified by the GenePix Pro software program (Axon Instruments, Foster City, CA) or by a computer program written in-house (16). To quantify the expression levels of known genes in a cell by using single-color detection, the color yielded by the enzymatic reaction was converted to gray levels. In addition, different amounts of six plant gene mRNAs were included in the mammalian mRNA mixture in cDNA probe preparation during reverse transcription. Hybridization intensities of these plant gene DNA spots in the microarray DNA membrane served as internal references that were then used to quantify the level of gene expression (DataExtract; Figure 1). On the basis of this approach, we have demonstrated a sensitivity of one or two copies per cell (under hybridization conditions with mRNA from 1 million cells) for this colorimetry detection approach. This sensitivity was comparable to those of fluorescent probes used on a glass slide surface (15). After DataExtract (Figure 1), data mining usually followed. Data mining is the stage at which various computational analysis and graphical techniques are used to reduce large amounts of information into manageable categories or "clusters" in order to demonstrate complex relationships in a readily understood fashion (cluster analysis; Figure 1). We used a commercial software program, GeneSpring (Silicon Genetics Products, Redwood City, CA), for our analysis. This software program includes both hierarchical cluster analysis (12) and the Genecluster system (26). For the former cluster analysis, the data take the form of a multiconditional expression intensity matrix, whereas the latter method combines assessments of the reliability of expression measurements with a statistical test of expression profiles.

View larger version (39K): [in this window] [in a new window]   Figure 1.   Flow chart of cDNA microarray image and data processing. Dual-color microarray membrane was first scanned in a flat-bed scanner for image digitization. Using the Adobe Photoshop software program, the image was separated into cyan, magenta, and yellow. The GenePix Pro equivalent software program was used to quantify the color intensity (DataExtract). After normalization and filter for low signal to noise (S/N) and high variance, the data were further clustered by various software programs, such as hierarchical cluster analysis (12) and the gene cluster method (26).

	PROFILING HYDROGEN PEROXIDE- AND SMOKE-INDUCIBLE GENE EXPRESSION BY DNA MICROARRAY MEMBRANE

TOP ABSTRACT INTRODUCTION GENERATION OF A HIGH-DENSITY... cDNA MICROARRAY IMAGING AND... PROFILING HYDROGEN PEROXIDE-... CONCLUSION REFERENCES

As proof of principle, a papillomavirus-immortalized human bronchial epithelial cell line, HBE1 (22), was exposed to hydrogen peroxide and mainstream smoke. Grown under an air- liquid interface, cultures of this cell line express mucous cell differentiation activity with expression of mucin synthesis and secretion, and mucin gene messages (27). Under this type of culture condition, mRNAs were isolated from these cultures at various times (0, 1, 3, and 48 h) after hydrogen peroxide (0.1 mM) exposure and used for microarray analysis. We have observed an immediate induction (within 1 h) of bcl-2 and mdm-2 messages in cells after hydrogen peroxide (0.1 mM) treatment (28). The induction was persistent throughout the time course of the study (up to 48 h). No other genes in the microarray membrane in this study had such a dramatically and lasting response. These results were further confirmed by a hierarchical cluster analysis of these microarray membranes. The expression of these genes was the earliest to be induced by hydrogen peroxide. The study of mRNA samples obtained from cultures at different times after hydrogen peroxide treatment will further define the inducibility of these genes by this oxidant insult.

Using a dual-color approach, the effects of tobacco smoke on HBE1 cells were investigated on a microarray membrane containing 9,600 EST cloned DNAs (28). Both biotin- and digoxigenin-labeled cDNA probes were prepared from the mRNA of cells untreated and treated with smoke. After hybridization and color development, the cDNA molecules labeled with biotin yielded a blue chromogen and the cDNA molecules labeled with digoxigenin appeared red. The majority of spots appear purple, indicating that the level of expression of these genes was not affected by smoke. However, some spots exhibit more distinctive colors, more toward red or blue, an indication of differential expression of genes after smoke exposure. After digitizing the image and quantifying these color spots (see Figure 1), the distinctly red DNA clones with a red/blue ratio greater than 3 were initially selected as smoke-induced genes. A similar approach was also carried out in hydrogen peroxide-treated cells. Twenty-two genes were viewed as inducible genes and 14 of these genes were further confirmed by Northern blot hybridization, and all were commonly elevated in smoke- and hydrogen peroxide-exposed cells (Table 1) (28). The rest of the differentially expressed genes (eight) could be confirmed by a semiquantitative reverse transcriptase (RT)-PCR method due to low abundance of these messages. A time course Northern blot hybridization study was carried out to elucidate the time course of induction by smoke (28). Our findings suggested three phases of gene expression induction (Figure 2). The first phase occurs immediately after the smoke and/or hydrogen peroxide treatment. The genes induced are bcl-2, mdm-2, and mitogen-activated protein (MAP) kinase phosphatase 1 (MAPK-1). The second phase included genes induced by smoke and hydrogen peroxide exposure, such as those for various stress proteins and ubiquitin, which appeared to be induced in cells 5 h after exposure. The third phase involves induction of genes related to oxidant metabolism and cell tissue remodeling. The expression of these genes appeared 10 h after smoke exposure (28).

View larger version (26K): [in this window] [in a new window]   Figure 2.   A summary of three phases of gene induction in the human bronchial eptihelial cell line HBE1 by oxidants (28). On the basis of time course Northern blot analyses, three different phases of gene expression induction were proposed to have occurred in cells after hydrogen peroxide and smoke exposure. Phase I was an immediate event, which might determine cell survival and induced gene expression. Phase II genes were induced at a later time and they might be involved in the early repair of cell injury. Phase III was a late event and genes induced in this phase might be involved in the regulation of cell remodeling and adaptation. Ferritin L = ferritin light chain; deHase = dehydrogenase; HSP = heat shock protein.

	CONCLUSION

TOP ABSTRACT INTRODUCTION GENERATION OF A HIGH-DENSITY... cDNA MICROARRAY IMAGING AND... PROFILING HYDROGEN PEROXIDE-... CONCLUSION REFERENCES

We have developed a high-throughput microarray system on nylon membrane that can be used to profile differential patterns of gene expression. The sensitivity of detection is one or two copies per cell, and the reactivity is linear to the level of expression (9, 15). The system is useful for simultaneous quantitation of the expression of multiple genes by a single hybridization step. The information generated from this approach is superior to that obtained by Northern blot hybridization, which is one gene per hybridization. What needs to be further improved in the system, in addition to improving the quantitative analysis and standardization, and cluster analysis, is the development of bioinformative technology that can integrate various array data and pattern recognition with biological processes. Such a development will fulfill the promise of functional genomics and provide integrated information regarding the association between various molecular events and cellular processes.

Using the microarray membrane generated in our laboratory, we examined the profile of gene expression patterns associated with hydrogen peroxide- and smoke-induced injury and repair in a human bronchial epithelial cell line, HBE1 (28). Northern blot and time course studies were used to verify the expression pattern of these inducible genes. From these studies, three phases of gene induction were detected (Figure 2). The first is an immediate phase, in which inducible genes were seen within 1 h of smoke exposure. These genes included MKP-1, mdm-2, and bcl-2. Except for MKP-1, the expression of these genes was not transient, and their induction was still seen 24 h after exposure. Interestingly, these genes are involved in the prevention of cell apoptosis. The MKP-1 protein is an important negative regulator of mitogen-activated protein kinase (MAPK) pathways, especially the JNK/SAPK and p38 kinase pathway (29). The second phase of gene induction occurred later. This induction proved to be transient, and the expression level was back to normal 15 h after smoke exposure. The representative genes in this phase were HSP60 (32), HSP70 (32), HSP90a (32), and ubiquitin p62. Most of the functions of these genes are involved in repairing denatured proteins. Stress proteins are important as chaperones for the proper folding of macromolecules, whereas ubiquitin protein can tag nonrepairable molecules for protease degradation. The third phase of gene induction occurred about 10 h after smoke exposure. The genes included glutaredoxin, light chain of ferritin, dihydrodiol dehydrogenase, MMP-1, and SPRR1B (35). Except for MMP-1 and SPRR1B, most of the genes induced in this phase are involved in the metabolism of oxidants. MMP-1 is responsible for a variety of tissue remodeling functions, whereas SPRR1B is a squamous cell differentiation marker (36) under the control of AP-1 transcription factor (37). Thus, an early but transient expression of MKP-1 may block the expression of this differentiation marker. Further study is needed to elucidate the significance of such a complex array of gene expression by bronchial epithelial cells after oxidant injury.