Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Acute lung injury is a severe respiratory syndrome associated with numerous precipitating factors and has a poor prognosis (40 to 60% mortality). Histopathology reveals surfactant disruption, epithelial perturbation, and sepsis, either as initiating factors or as secondary complications, which in turn increase the expression of cytokines that sequester and activate inflammatory cells (1). Concomitant release of reactive oxygen and nitrogen species subsequently modulates endothelial function. Together these events orchestrate the principal clinical manifestations of the syndrome: pulmonary edema, cellular infiltration, and airway collapse (2).

Diverse animal models for the study of acute lung injury have been developed, including its induction by hyperoxia, trauma, sepsis, embolism, lung contusion, endotoxin, cytokines, oleic acid, and ozone (3). Previously, we found that particulate nickel sulfate (NiSO₄) exposure also models acute lung injury in mice as evidenced by perivascular distension, epithelial damage, alveolar congestion, hemorrhage, neutrophilic infiltration, and pulmonary edema (6). Nickel is found in cigarette smoke, urban air particulate matter (7), and several occupational environments (8), and acts as a potent respiratory irritant (9, 10), asthmagen (11), and initiator of lung injury (12). Although the insults used in these models are diverse, acute lung injury appears to be a shared final outcome.

Despite extensive research since the initial description of acute lung injury over 30 yr ago, questions remain about the basic pathogenic mechanisms and their relationship to therapeutic strategies. A number of factors have been associated with the development of acute lung injury. However, the interplay between these factors and the mechanisms involved has remained less clear. We proposed that assessing the temporal and functional relationships of differentially expressed genes after pulmonary insult would reveal novel interactions in acute lung injury. Specifically, 8,734 sequence-verified murine complementary DNAs (cDNAs) were analyzed for temporal changes and functional relationships throughout the initiation and progression of acute lung injury in mice at five times ranging from 3 to 96 h of exposure to particulate NiSO₄.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Experimental Design

To examine differential gene expression during the initiation and progression of acute lung injury, messenger RNA (mRNA) levels were assessed in mouse lung after exposure to particulate NiSO₄ for 0 (control), 3, 8, 24, 48, or 96 h. Lung polyadenylated mRNA was isolated, reverse transcribed, and fluorescently labeled. Samples from exposed mice (Cy5 labeled) were competitively hybridized against samples from unexposed, control mice (Cy3 labeled) to microarrays containing murine cDNAs. To assess the temporal pattern of gene expression, genes were clustered according to similarities in expression with time. To evaluate the pathophysiology of changes in gene expression, genes were categorized according to function. Changes in the expression of five selected genes were also analyzed by conventional assays (Northern blot and S1 nuclease protection).

Exposure

Mice (C57BL/6 strain; 6 to 8 wk) were purchased from Jackson Laboratory (Bar Harbor, ME), housed in our facility for  1 wk before exposure, and placed in stainless steel cages inside a 0.32-m³, stainless steel inhalation chamber. NiSO₄ aerosol was generated from 50 mM NiSO₄ hexahydrate (Sigma, St. Louis, MO) using a modified Collison three-jet nebulizer (3.5 liters/min) (BGI Incorporated, Waltham, MA) placed inside a glass tube (24 mm inner diameter). The particle size was determined using a differential mobility analyzer consisting of an electrostatic classifier (model 3071A; Thermo-Systems, Inc., St. Paul, MN), condensation nucleus counter (model 3022A; Thermo-Systems, Inc.), and scanning mobility particle size fast-scanning software (Thermo-Systems, Inc.). The particulate aerosol had a mass median aerodynamic diameter (MMAD) of 0.22 µm and a geometric standard deviation of 1.85. The chamber nickel concentration was 110 ± 26 µg/m³ using the dimethylglyoxime method (13). Samples of the chamber atmosphere were collected with two midget impingers (Ace Glass, Inc., Vineland, NJ) placed in series, each containing 10 ml distilled water. Collected samples were mixed with a solution containing 1 M hydrochloric acid, 0.2 M bromine, 12 M ammonium hydroxide (Fisher, Fair Lawn, NJ), 1% dimethylglyoxime (Sigma), and 95% ethanol. Absorbance was measured at 445 nm (model DU-64; Beckman, Fullerton, CA). During NiSO₄ exposures, mice were supplied with food and water. At selected times, mice were removed, anesthetized with sodium pentobarbital (Nembutal, 5 mg intraperitoneal; Abbott, Chicago, IL), exsanguinated via incision of the dorsal aorta, lobes of lung removed, and immediately frozen in liquid nitrogen.

RNA Preparation

Total RNA was isolated from homogenized lung (3 ml TRIZOL; Life Technologies, Gaithersburg, MD; Tissumizer; Tekmar, Co., Cincinnati, OH) and recovered by acid phenol extraction. Polyadenylated mRNA was isolated by binding to poly-T oligonucleotides covalently linked to polystyrene-latex particles and then eluted with low salt buffer (Qiagen, Inc., Valencia, CA).

Fluorescent Labeling of Probes

Polyadenylated mRNA was pooled from three mice at selected times throughout NiSO₄ exposure and labeled with 5' Cy5-labeled, random 9-mers during reverse transcription. Polyadenylated mRNA was pooled from three unexposed, control mice and labeled with 5' Cy3-labeled, random 9-mers during reverse transcription (Operon Technologies, Inc., Alameda, CA). For each reverse transcription reaction, 200 ng polyadenylated mRNA was incubated for 2 h at 37°C with 200 U Moloney murine leukemia virus (MMLV) reverse transcriptase (Life Technologies), 4 mM dithiothreitol (DTT), 1 U RNase inhibitor (Ambion, Austin, TX), 0.5 mM deoxynucleotide triphosphates, and 2 µg labeled 9-mers in 25 µl volume with enzyme buffer supplied by the manufacturer. The reaction was terminated by a 5-min incubation at 85°C. For cohybridization, paired Cy3 and Cy5 reactions were combined and purified with a TE-30 column (Clonetech, Palo Alto, CA), brought to 90 µl with dH₂O, and precipitated with 2 µl of 1 mg/ml glycogen, 60 µl of 5 M ammonium acetate, and 300 µl ethanol. After centrifugation, the pellet was resuspended in 24 µl of hybridization buffer (5× saline citrate, 0.2% sodium dodecyl sulfate, 1 mM DTT).

Microarray Preparation

Known genes were selected for the microarray by the manufacturer (Incyte Pharmaceuticals, Inc., Palo Alto, CA) from The Institute for Genomic Research's Mus.ET database of full-length mouse transcript sequences. Nonannotated expressed sequence tags (ESTs) were selected from GenBank's mouse database. These nonannotated ESTs were generated by reverse transcription of polyadenylated mRNA isolated from murine tissues and the sequences were not identified with any known genes. All the cDNAs on the microarray were generated from sequence-verified libraries of clones and are between 500 and 5,000 bp in length, averaging 1,000 bp. Sequences used for the microarray were polymerase chain reaction products purified by gel filtration with Sephacryl-400 (Amersham Pharmacia Biotech, Inc., Piscataway, NJ), dried, and resuspended in dH₂O. Each cDNA was fixed to the surface of modified glass slides and arraying was performed by robotics (Incyte Pharmaceuticals, Inc.). The microarray was then washed three times in dH₂O at room temperature, treated with 0.2% I-Block (Tropix, Bedford, MA) dissolved in 1× Dulbecco's phosphate-buffered saline (Life Technologies) at 60°C for 30 min, and rinsed in 0.2% sodium dodecyl sulfate (SDS) for 2 min followed by three 1-min washes in dH₂O.

Hybridization

Each paired Cy3 and Cy5 probe solution was resuspended by incubating at 65°C for 5 min with mixing, applied to an array, covered with a 22-mm² glass coverslip, and placed in a sealed chamber. After hybridization at 60°C for 6.5 h, slides were washed in three consecutive solutions of decreasing ionic strength. Both Cy3 and Cy5 channels were simultaneously scanned with independent lasers at 10 µm resolution. The detected fluorescent light was optically filtered and photon multiplier tubes used to translate photons into an analog electrical signal. To adjust for differences in probe labeling efficiency, balance coefficients (ratio of total Cy3 to total Cy5 fluorescence signal) were derived for each microarray. The balance coefficients were 1.05, 1.14, 0.83, 0.88, 0.74 for the 3, 8, 24, 48, and 96 h microarrays, respectively. The differential expression level (ratio of Cy5 to Cy3 signal when Cy5 > Cy3; ratio of Cy3 to Cy5 signal when Cy3 > Cy5) for each cDNA was multiplied to the balance coefficient of each microarray, generating a balanced differential expression value for each cDNA. The criteria for inclusion of a cDNA in the analysis were that the fluorescent signal from the cDNA exceeded a signal to background ratio of 2.5 and that the cDNA covered its grid location on the microarray > 40%.

Temporal Clustering

Data were sorted according to the criteria that the balanced differential expression value of a gene at two or more times was  1.8 or  1.8 (Microsoft Access; Microsoft, Redmond, WA). This reduced the data set to a group of 100 genes. The gene name and balanced differential expression value for each of these 100 genes was saved in a tab delimited format (Microsoft Excel) and imported into Gene Cluster (http://rana.stanford.edu/software) (14). Data were clustered using the gene similarity metric (uncentered correlation). To view the cluster results, the clustered data (cdt file extension) was imported into Tree View (http://rana. stanford.edu/software). Within Tree View, balanced differential expression values were assigned pseudocolors; cells representing increased expression levels relative to control were colored red and those representing decreased expression levels relative to control were colored green. Black was assigned to missing values.

Homology Search

To determine the level of homology of nonannotated ESTs with known genes, the accession number from each nonannotated gene in the cluster group of 100 was entered into BLAST 2.0 (http://www.ncbi.nlm.nih.gov/) using blastn (nucleotide) and against the nonredundant (nr) database. When the BLAST search displayed homology with a known gene (sequence alignment score > 200), the nonannotated EST was considered highly homologous to that known gene.

Functional Categorization

To assess functional characteristics, differentially expressed genes were categorized according to function using the hierarchy component of GEMTools (Incyte Pharmaceuticals, Inc.). A gene group of the clustered 100 genes was created within GEMTools. This gene group was searched against the default hierarchies of GEMTools.

Northern Blotting and S1 Nuclease Protection Assays

To assess further increased expression of metallothionein-1 and heme oxygenase-1 as observed with the microarray analysis, Northern blot analysis was performed. Using a method previously described (15), mouse metallothionein-1 and rat heme oxygenase-1 cDNAs were used as the templates for [³²P]-labeled riboprobes (hybridization buffer: 3× SET, 0.1% SDS, 0.2% Ficoll, 0.2% polyvinylpyrrolidine, 0.2% bovine serum albumin, 250 mg/ml yeast transfer RNA; 65°C, 16 h). Blots were washed in 1× saline sodium citrate, 0.1% SDS followed by 0.3× SSC, 0.1% SDS at 65°C, and quantified by phosphorimaging (ImageQuant, Molecular Dynamics PhosphorImager, Sunnyvale, CA). Each blot was subsequently probed using a [-³²P]-labeled oligonucleotide to 28S ribosomal RNA (rRNA) (hybridization buffer: 5× SET, 0.1% Ficoll, 0.1% polyvinylpyrrolidine, 0.1% bovine serum albumin, 0.5% SDS, 0.05 mg/ml denatured salmon sperm DNA, 0.004 mg unlabeled 28S rRNA oligonucleotide; 42°C; 3 h [16]) (1× SET is 150 mM NaCl, 30 mM Tris-HCl, pH 7.8, 2 mM ethylenediaminetetraacetic acid). Blots were washed in 2× SET, 0.1% SDS at room temperature followed by 0.5× SET, 0.1% SDS at 42°C, and quantified by phosphorimaging.

To assess further the decrease in surfactant protein gene expression observed with the microarray analysis, S1 nuclease protection assays were performed. Using a method described previously (17), S1 probes specific for murine surfactant protein (SP)-A, SP-B, SP-C, and ribosomal protein L32 (L32) were linearized, end-labeled with [-³²P]adenosine triphosphate (ATP), combined, and hybridized (16 h, 55°C) with the same total lung RNA used for microarray analysis. Single-stranded regions were digested away from protected fragments by S1 nuclease (110 U; Life Technologies) in the presence of excess unlabeled salmon sperm DNA (1 h, room temperature). The protected fragments were electrophoresed through 6% polyacrylamide gels containing 8 M urea and quantitated by phosphorimaging (ImageQuant, Molecular Dynamics PhosphorImager).

Statistical Analysis

The threshold of significance used in this study was based on the determined probability (P < 0.005) for a gene displaying differential expression greater than 1.74-fold (Incyte Pharmaceuticals, Inc.). The probability of differential expression  1.8-fold occurring twice by chance is  1:40,000 (P  0.000025). For Northern blots, samples were normalized to 28S rRNA and for S1 nuclease protection assays, samples were normalized to L32. Means were compared by one-way analysis of variance followed by Dunnet's method for multiple comparisons (P < 0.05).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

To determine the extent to which genes were differentially expressed during the initiation and progression of acute lung injury, we first assessed gene expression by microarray analysis. Of the > 8,700 genes analyzed at each exposure time, few genes changed in their level of expression relative to control. However, as lung injury progressed, the number and the magnitude of changes increased (Figure 1). At 3 h, 17 genes were differentially expressed (three increased and 14 decreased). The largest increase was three times that of control and the largest decrease was 4-fold lower than control. In contrast, at 96 h, 255 genes were differentially expressed (125 increased and 130 decreased). The maximal increase at 96 h was 13-fold greater than control and the maximal decrease was 33-fold lower than control.

View larger version (28K): [in this window] [in a new window]   Figure 1.   Nickel-induced acute lung injury results in differential gene expression. Polyadenylated mRNA from the mouse lung was isolated after NiSO4 (110 µg/m3, 0.2 µm MMAD) exposure for 3, 8, 24, 48, or 96 h and labeled during reverse transcription with the fluorescent dye Cy5. These samples were competitively hybridized with Cy3-labeled cDNA from the lungs of unexposed, control mice to microarrays of 8,734 murine cDNAs. After hybridization of Cy3 and Cy5 samples, scans were acquired for each fluorophore at each target cDNA element. Each point represents the Cy3 versus Cy5 signal values for each cDNA element, and slanted lines delineate fold changes.

To assess the temporal pattern of gene expression, genes were grouped according to similarities in expression with time. Genes differentially expressed relative to control at two or more times ( 1.8-fold, 100 genes) were clustered into four major temporal groups (Figure 2). Group I contained genes that tended to increase consistently throughout exposure to nickel, including metallothionein-1, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), apolipoprotein E, lactotransferrin, small proline rich protein-1a, and galectin-3. Group II contained genes primarily displaying a delayed increase, including secretory leukoprotease inhibitor and lysyl oxidase. Group III contained genes that tended to decrease in expression in response to nickel and most were nonannotated ESTs. Known genes included deubiquitinating enzyme-1 and L-selectin. Group IV contained genes that showed a delayed decrease in expression, including SP-C and Clara cell secretory protein. Of the 100 genes that were grouped into temporal clusters, 53 were known genes, 15 were ESTs highly homologous to known genes, and 32 were ESTs with nominal homology to known genes.

View larger version (43K): [in this window] [in a new window]   View larger version (42K): [in this window] [in a new window]   Figure 2.   Differentially expressed genes clustered into four major temporal groups. Mice were exposed to particulate NiSO4 (110 µg/m3, 0.2 µm MMAD) and lung cDNA analyzed by microarrays containing 8,734 murine cDNA elements. Genes with two or more changes relative to control (balanced differential  1.8-fold) were clustered using the gene similarity metric as described by Eisen (14). Cells representing increased gene expression relative to control are colored red and those representing decreased expression relative to control are colored green. Black is assigned to missing values. The scale displays relative levels of expression. Large groups displaying similar expression patterns are designated I to IV. Only ESTs designated as highly similar to known genes are annotated. *Nonannotated EST found by BLAST search (nonredundant GenBank, EMBL, DDBJ, and PDB sequences) to be highly homologous to known gene.

To evaluate the pathophysiology of changes in gene expression, genes within the four temporal groups were organized into functional categories (Table 1). Genes sharing common temporal patterns often shared common functions. Temporal Group I genes, including GAPDH and apolipoprotein E, are in the category of metabolism. Other Group I genes, including galectin-3 and signal transducer and activator of transcription-3, are in the category of immunity. Group II genes, including heme oxygenase-1 and lysyl oxidase, are in the category of oxidoreductases. Group I and II genes, including secretory leukoprotease inhibitor and spi2 proteinase inhibitor, are in the category of protein modification and maintenance. Group IV genes, including SP-C, Clara cell secretory protein, and cytochrome P450 2f2, are in the category of cell-specific genes.

By microarray analysis, metallothionein-1 and heme oxygenase-1 expression increased in response to nickel-induced injury. To examine these increases further, we performed Northern blot analysis. The microarray and Northern blot analyses displayed similar fold changes and temporal patterns for both of these genes (Figure 3). Metallothionein-1 expression increased by 3 h (microarray: 2.5-fold; Northern blot: 2.3 ± 0.4-fold) and continued to increase through 8 (microarray: 4.2-fold; Northern blot: 5.3 ± 0.8-fold), 24 (microarray: 6.2-fold; Northern blot: 11.4 ± 3.4-fold), and 48 h (microarray: 6.5-fold; Northern blot: 9.8 ± 2.3-fold). Metallothionein-1 expression reached its highest level by 96 h (microarray: 13.1-fold; Northern blot: 11.6 ± 0.7-fold). Heme oxygenase-1 increased in expression at 8 h (microarray: 2.6-fold; Northern blot: 3.8 ± 0.6-fold) and again at 96 h (microarray: 4.8-fold; Northern blot: 3.0 ± 1.3-fold). (This cDNA on the 96-h microarray covered 37% of its grid, failing to meet the criteria of 40% coverage; however, the balanced differential value is included in this manuscript because of verification by Northern blot analysis.)

View larger version (14K): [in this window] [in a new window]   Figure 3.   Comparison by microarray and Northern blot analyses of two genes that increased in mice exposed to particulate NiSO4 (110 µg/m3, 0.2 µm MMAD). (A) Lung mRNA was analyzed by cDNA microarrays and balanced differential expression values calculated. (B) Lung mRNA was analyzed by Northern blots. Mouse metallothionein-1 (MT-1) and rat heme oxygenase-1 (HO-1) cDNAs were used as the templates for [32P]-labeled riboprobes (hybridization: 65°C, 16 h). Each blot was subsequently probed using a [-32P]- labeled oligonucleotide to 28S rRNA (hybridization: 42°C, 3 h). Levels of metallothionein-1 and heme oxygenase-1 were quantified by PhosphorImaging, normalized to 28S rRNA, and are presented as fold of control (± standard error of the mean [SEM] of three samples).

Having found decreased expression in surfactant proteins by the microarray analysis, we further examined this differential expression by performing S1 nuclease protection assays for SP-A, SP-B, and SP-C. The microarray and the S1 (Figure 4) analyses displayed a similar trend: expression of surfactant proteins decreased with progression of lung injury. As observed with the microarray analysis, SP-B and SP-C expression decreased by 24 h relative to control (microarray: SP-B, 3-fold; SP-C, 6-fold; S1 analysis: SP-B, 7-fold; SP-C, 6-fold). Gene expression continued to decrease throughout 48 (microarray: SP-B, 5-fold; SP-C, 30-fold; S1 analysis: SP-B, 14-fold; SP-C, 50-fold) and 96 h (microarray: SP-B, 4-fold; SP-C, 33-fold; S1 analysis: SP-B, 13-fold; SP-C, 100-fold). SP-A also decreased after 24 h of exposure. Although quantification of the results was not identical, differential expression patterns were similar between assays.

View larger version (15K): [in this window] [in a new window]   Figure 4.   Comparison of decreases in genes constitutively expressed in the lung observed with nickel-induced acute lung injury. Mice were exposed to particulate NiSO4 (110 µg/m3, 0.2 µm MMAD). (A) Lung mRNA was analyzed by cDNA microarrays and balanced differential expression values calculated. Clara cell secretory protein, CCSP. (B) Lung mRNA was analyzed by S1 nuclease protection assays. DNA probes specific for murine SP-A (solid circles), SP-B (open circles), SP-C (solid triangles), and L32 were linearized, end-labeled with [-32P]- ATP, combined, and hybridized (3 µg, 55°C, 16 h) with total lung RNA used for microarray analysis. Single-stranded regions were digested away from protected, double-stranded fragments by S1 nuclease. The protected fragments were then denatured, electrophoresed, and quantified by PhosphorImaging. Levels of SP-A, SP-B, and SP-C were normalized to L32 quantified by PhosphorImaging and are presented as percent of control (± SEM of three samples).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

After selected times of NiSO₄ exposure, differential gene expression in mouse lung progressed with acute injury. The number of significant changes demonstrates the complex pathophysiology of acute lung injury induced by nickel. By clustering the genes according to similarities in expression patterns, four groupings emerged. Group I contains genes that tended to increase consistently throughout nickel-induced acute lung injury. Group II contains genes that primarily display a delayed increase. Group III contains genes that tended to decrease throughout the response to injury, and Group IV contains genes that display a delayed decrease. These groups of genes sharing common temporal patterns of expression likewise often shared common functions. This study revealed the expression patterns of genes previously associated with acute lung injury in relationship to one another and to genes not previously associated with acute lung injury. Thus, novel interactions emerged from this study that incorporated the observations of previous investigators as well as newly made observations and provide a more complete understanding of acute lung injury.

By assessing the temporal and functional relationships of differentially expressed genes, our analyses revealed the progression of the pulmonary response to injury. Temporal Group I contained genes that tended to increase in expression throughout nickel-induced acute lung injury (Figure 2). A portion of these genes previously had been found to increase in the lung after nickel exposure, including metallothionein-1 and lactotransferrin. Metallothionein-1, which increased more than any other gene measured (> 13-fold at 96 h; Figures 2 and 3), is a thiol-rich protein that increases in lung injury induced by nickel (18), cigarette smoke (19), and ozone (20), and can protect against metal-induced oxidative damage to critical macromolecules (21). Like metallothionein-1, the observed increase in lactotransferrin, a metal binding protein, may likewise provide antioxidant protection (22).

Another gene within Group I previously found to increase in expression with nickel exposure is GAPDH (23). Enolase shares the expression pattern of GAPDH and may also share a common function. The pattern of induction shared by GAPDH and enolase may indicate a common response to changes in oxygen tension. In response to altered oxygen tension, a transactivating factor, hypoxia-inducible factor-1, binds to a consensus sequence within regulatory regions of each of these genes (24, 25). Thus, nickel-induced injury increases the expression of genes associated with hypoxia-induced injury possibly through shared transcriptional regulation in response to diminished oxygen levels. Clustering of each of these genes revealed their relationship to one another during nickel-induced acute lung injury.

The pulmonary response to injury was further indicated by additional genes in Group I, including galectin-3, small proline-rich protein-1a, and apolipoprotein E. Commonly in response to pulmonary injury, type II cells proliferate to replace damaged type I cells and to cover the unprotected alveolar basement membrane (26). The signals initiating this process remain largely unknown. Galectin-3 expression is normally undetectable in type I and II cells, yet with injury, expression increases in both cell types (29). Galectin-3 stimulates cell proliferation (30), suggesting a possible role for this lectin in the replacement of damaged alveolar cells. Similarly, small proline-rich protein-1a, has been associated with squamous cell differentiation in response to injury (31). Another Group I gene, apolipoprotein E, is expressed in alveolar type II cells (32). Lipoproteins stimulate the secretion of surfactant proteins from alveolar type II cells (33) and facilitate lipid transport and metabolism (34). Together, these findings are consistent with an increase in expression of genes involved in epithelial replacement and maintenance of epithelial function.

Genes within Group I also indicated the inflammatory response to acute lung injury. Galectin-3 expression increases with macrophage infiltration (29) and may facilitate cytokine release (37). Apolipoprotein E is likewise indicative of activated macrophage (36, 38). MAL (T-lymphocyte maturation-associated protein) is produced principally in T cells and is associated with T-cell maturation (39). Increased expression of these genes signifies activation of components of the inflammatory response with the induction and progression of nickel-induced acute lung injury. One limitation of our analysis was the paucity of microarray cDNAs involved in the pulmonary inflammatory response, limiting the ability to assess their changes in relationship to other genes in this large-scale assessment.

Genes within Group II primarily displayed a delayed increase, revealing the progression of acute lung injury (Figure 2). Some members of Group II previously had been associated with acute lung injury, including heme oxygenase-1 and secretory leukoprotease inhibitor. Elevated levels of heme oxygenase-1 have been found to be protective against hyperoxia-induced acute lung injury (40). Secretory leukoprotease inhibitor is a local source of antiprotease protection in the lung, inhibits protease destruction of the extracellular matrix (41), and accumulates slowly after treatment with elastase (42). Patients with acute lung injury have elevated concentrations of secretory leukoprotease inhibitor (43). As with secretory leukoprotease inhibitor, lysyl oxidase displayed this temporal pattern and is involved in maintaining the extracellular matrix. For example, lysyl oxidase synthesis can contribute to elastin stabilization and often is concomitant with elastin replacement (44). Elastin levels were likewise increased in response to nickel. Thus, nickel-induced acute lung injury results in increased expression of genes involved in extracellular matrix repair.

Previously, several of the genes within Group IV (Figure 2) had been associated with acute lung injury and this work reveals their relationship in the progression of injury. Clara cell secretory protein and cytochrome P450 2f2, both predominately expressed in Clara cells (45), displayed a delayed decline in expression. Clara cell secretory protein may be critical to antioxidant defenses in the airways, indicated by an elevated response to oxidants in mice deficient in this protein (20). Interestingly, secreted Clara cell secretory protein concentrations are elevated in patients who recover from acute lung injury yet Clara cell secretory protein levels are diminished in those who do not survive (46).

Two other members of Group IV, SP-B and SP-C, were unchanged from high constitutive levels of expression until decreasing at 24 h and continuing to diminish throughout the exposure (Figure 4). Surfactant proteins modulate alveolar surface tension, prevent atelectasis, inactivate reactive oxygen species, and augment host antimicrobial defenses (47). Diminished surfactant-associated protein concentration has been observed in patients with acute lung injury (48), and gene transcription decreases with administration of tumor necrosis factor-alpha, a cytokine often elevated in acute lung injury (17). Our findings reveal that these decreases are a delayed response, occurring in relationship to the altered expression in a number of other genes.

In addition to known genes, ESTs with nominal homology to known genes clustered into each of the four temporal groups. Because these ESTs displayed similar expression patterns as known genes, they may similarly play significant roles in the pulmonary response to injury. Characterization of these sequences may provide insight into their possible roles and open new areas of research on the disruption of lung function during acute lung injury and possible mechanisms of repair.

One of the clearest differences noted in this study was the decreased expression of surfactant proteins. Alteration of surfactant homeostasis is a critical event in the pathophysiology of acute lung injury, and although comprising only a small portion of the surfactant complex, surfactant-associated proteins are critical for life. For example, SP-B- deficient mice succumb to respiratory failure within 20 min after birth (49). Our findings are consistent with the ongoing attempts to treat acute lung injury with exogenous surfactant containing surfactant-associated proteins (28), possibly in combination with antioxidant therapy. The delayed temporal pattern of surfactant decreases also implies that early (before focal atelectasis develops) and sustained administration of exogenous surfactant may be necessary to achieve effective therapy.

In summary, this study revealed the expression patterns of genes previously associated with acute lung injury in relationship to one another and in relationship to genes previously not associated with acute lung injury. Thus, novel interactions emerged from the temporal and functional analyses of differential gene expression during nickel-induced acute lung injury. The most notable changes were increases in the expression of genes involved in protection against oxidant injury, indicators of hypoxia, cell proliferation, and extracellular matrix repair, whereas surfactant gene expression decreased. ESTs with nominal homology to known genes displayed similar patterns of expression to those of genes associated with acute lung injury, suggesting these ESTs might similarly play significant roles in the pulmonary response to injury. Thus, this large-scale assessment of gene expression has revealed the possible interplay of factors contributing to lung injury. These findings should open new approaches for determining the possible mechanisms of pathology and repair during acute lung injury.