Sheppard Lab Gene Expression Array Data Site

Abstract - The molecular mechanisms of pulmonary fibrosis are poorly understood. We have used oligonucleotide arrays to analyze the gene expression programs that underlie pulmonary fibrosis in response to bleomycin, a drug that causes lung inflammation and fibrosis, in two strains of susceptible mice (129 and C57bl/6). We then compared the gene expression patterns in these mice with 129 mice carrying a null mutation in the epithelial-restricted integrin b6 subunit (b6-/-) which develop inflammation but are protected from pulmonary fibrosis. Cluster analysis identified two distinct groups of genes involved in the inflammatory and fibrotic responses. Analysis of gene expression at multiple time points after bleomycin administration revealed sequential induction of subsets of genes that characterize each response. The availability of this comprehensive data set should accelerate the development of more effective strategies for intervention at the various stages in the development of fibrotic diseases of the lungs and other organs.

Pulmonary fibrosis is a progressive and largely untreatable group of disorders that affects up to 100,000 people in the United States . Current therapies, aimed principally at inhibiting lung inflammation that often precedes fibrosis, are effective only in a minority of affected individuals, and there are currently no proven therapies targeting the fibrotic process itself. Much of the information regarding the development of pulmonary fibrosis has been acquired with a well characterized animal model in which fibrosis is induced by a single intratracheal administration of the cytotoxic drug, bleomycin. Bleomycin, like in most of the conditions associated with pulmonary fibrosis, initially induces lung inflammation that is followed by a progressive destruction of the normal lung architecture. Previous studies of gene expression in bleomycin induced fibrosis have provided some clues to pathophysiology but have generally been limited to analysis of one or a few genes, and therefore have provided little information about the coordinated pattern of gene expression involved in this response.

To understand the molecular bases for pulmonary fibrosis, in more detail, we have analyzed changes in gene expression in response to bleomycin using oligonucleotide microarrays that permit a simultaneous and quantitative measurement of the expression of thousands of genes. A comprehensive profile of the pulmonary response to bleomycin was generated with arrays containing probe sets for approximately 6000 murine genes and expressed sequence tags (ESTs). Changes in gene expression were monitored after bleomycin treatment of wild type C57bl/ 6 mice, 129 mice and 129 mice homozygous for a null mutation of the integrin b6 subunit gene (b6-/-). We have previously shown that b6-/- mice develop inflammation, but do not develop fibrosis in response to bleomycin. In this report, data obtained with the three strains of mice has allowed us to distinguish genes involved in the inflammatory response to bleomycin (induced in all 3 lines) from genes that specifically contribute to the fibrotic response (induced to a lesser degree in b6-/- mice). Analysis of the time course of induction of these genes has identified sequential genetic programs that characterize bleomycin-induced inflammation and fibrosis.

Female C57bl/6 mice (approximately 10 weeks old) were anesthetized with methoxyfluorane using a 24-gauge intubation needle inserted into the trachea via the oral cavity. 50 ml of bleomycin (0.08 units in 0.9% saline, Sigma) or saline was slowly injected. Mice were sacrificed 2, 5, 7 or 14 days after bleomycin or 2 days after saline injection and their lungs frozen in liquid nitrogen. Total RNA was isolated from pooled lungs of six mice per time point and used to prepare twice purified poly A mRNA (Oligotex, Qiagen) for use as template for double stranded cDNA synthesis.

Aliquots of the hybridization murine cRNA mixtures (10 mg cRNA in 200 ml hybridization mix) were hybridized to a Mu6500 and each Genechipâ array was washed and scanned (Hewlett Packard, GeneArray^TM scanner G2500A) according to procedures developed by manufacturer (Affymetrix, Santa Clara, CA).

RT-PCR was performed as described. Briefly, total RNA was isolated and reverse-transcribed (Superscript II, Gibco BRL). Aliquots of each sample were diluted serially and the abundance of Hypoxanthine Phosphoribosyl Transferase (HPRT) among the samples of interest was normalized to identical concentrations of a 450 bp modified HPRT insert in a polycompetitor plasmid (pLOC, gift of Dr. Richard Locksley) added to each reaction (5 ng/ml). Primers for the amplification of a 650 base fragment of MME were (5’ to 3’) AGCATCTTAGAGCAGTGCCC (forward) and ATGTTGGTGGCTGGACTCCC (reverse) . Samples were subjected to 30 rounds of PCR at an annealing temperature of 60 o C. PCR products were separated in 2.5% agarose gel by electrophoresis and visualized with ethidium bromide staining.

To identify patterns of gene expression in response to bleomycin, we performed cluster analysis as previously described using mean values of the average difference intensities for each gene. We chose 470 genes that had a mean intensity of >100 arbitrary units in at least one condition and differed in intensity by at least two fold in at least one pairwise comparison. We then performed cluster analysis treating each possible pairwise comparison of mean values as a single data point. With this approach, we hoped to identify groups of genes that were regulated in a coordinate fashion, both at baseline and after treatment with bleomycin. Cluster analysis of these genes is shown in Figure 1.

One large cluster of genes (cluster B, Figure 1 and 2A) was expressed at similar levels at baseline in wild type and b6-/- mice (Figure 1 and 2A, KO/WTBs). It was dramatically induced by bleomycin in wild type mice (Figure 1 and 2A, WT 7d/Bs and WT 14d/Bs), but was induced to a lesser degree in b6-/- mice (Figure 1 and 2A, KO 7d/Bs and KO 14d/Bs). These genes could thus provide clues to critical steps in the development of fibrosis. The cluster was composed of 66 genes that included 16 proteins involved in the formation of the extracellular matrix, 10 proteins involved in regulation of cellular responses to the extracellular matrix, and 3 genes known to be induced by DNA damage (GADD45, CDKI 1 and DDB2). Quantitative comparison among these genes identified a small sub-group, including osteopontin, tenascin-C, tropoelastin, and the stress-induced protein, heme-oxygenase (HO) which were most dramatically induced by bleomycin. These same genes were the most different between wild type and b6-/- mice after bleomycin treatment (Figure 2B). Expression of 78% of these genes was also increased in C57bl/6 mice, including each of the 4 genes described above (Figure 2C).

Because of the presumed significant role of TGFb in fibrosis and the presence of multiple TGFb-inducible genes in cluster B, we compared the expression of a group of 53 genes known to be induced by TGFb 14 days after bleomycin treatment in wild type and b6-/- mice (Figure 2B). The overwhelming majority of TGFb-inducible genes were expressed at higher levels after bleomycin in wild type mice than in b6-/- mice. Upregulation of these TGFb-inducible genes also occurred in C57bl/ 6 mice (Figure 2C). Despite this dramatic evidence of TGFb effect, increases in TGFb1, b2 and b3 mRNA by bleomycin were not detectable in any of the mouse strains (data not shown).

Another cluster (Cluster A, Figure 1 and 3A), included 63 genes that are expressed at higher levels at baseline in b6-/- mice (Figure 1 and 3A, KO/WT Bs), and are also induced by bleomycin. At both 7 and 14 days after bleomycin, the absolute level of expression of most of the genes in this cluster was the same or higher in b6-/- mice as in wild type mice (Figure 1 and 3A, KO/WT 7d, KO/WT 14d). This cluster is dominated by genes known to be involved in inflammatory responses and include complement components (5), serine proteases (4), serum amyloid proteins (3), chemokines and chemokine receptors (4), genes restricted to leukocytes (9), and others associated with inflammation (6). In this cluster of genes 83% were also induced in C57bl/6 mice.

To determine whether the increase in leukocyte related genes could be explained entirely by a difference in baseline cellular composition of the lungs of wild type and b6-/- mice, we conservatively identified genes whose expression is largely restricted to leukocytes, and compared their baseline expression in wild type and b6-/- mice (Figure 3C). A vast majority of these genes were increased four-fold or less in b6-/- mice, consistent with the 2-4 fold increase in numbers of macrophages, lymphocytes, eosinophils and neutrophils that we have previously described for these mice. However, one gene, the macrophage-restricted metalloproteinase, macrophage metalloelastase (MME) was increased more than 20 fold. RT-PCR of RNA from whole lung and alveolar macrophages confirmed that this dramatic induction reflected a true increase in mRNA expression rather than simply a difference in cell type (Figure 3D). Four other members of this cluster, the acute phase reactant serum amyloid 3 (SAA3), lipocalin 2 - a neutrophil gelatinase associated protein that is expressed also in carcinomas and normal epithelial tissues, BRP 39 - a cartilage glycoprotein, and a C-C chemokine, C10, were also among the genes induced most dramatically in b6-/- mice (Figure 3B).

Cluster analysis also demonstrated that protection of b6-/- mice from bleomycin-induced pulmonary fibrosis was not simply due to blunting of all cellular responses to the drug. One cluster of genes (Cluster C, Figure 1) was expressed at similar levels in wild type and b6-/- mice at baseline and induced to a similar degree by bleomycin. Another cluster (Cluster D, Figure 1) was preferentially induced by bleomycin in b6-/- mice.

To describe the time course of gene induction by bleomycin, data from the detailed time course experiments performed in C57bl/6 mice were analyzed. Overall, 63% of the genes whose expression level was increased after bleomycin in wild type 129 mice were also seen at an increased level in at least one time point after bleomycin in C57bl/6 mice. For the genes in clusters A and B, this number was substantially higher amounting to 78% of the genes in cluster A and 83% of the genes in cluster B. The temporal patterns of expression observed for subsets of genes in each of these clusters are shown in Figure 4. In each of these clusters we were able to identify groups of genes with similar temporal patterns of expression. In some cases, the similar temporal pattern was closely linked by a known function or transcriptional regulation. For example, 5 of the 8 genes from cluster A that were maximally induced at the earliest time point studied (Figure 4A) are known components of interferon-activated signaling including STAT-1 and 4 interferon-inducible genes that are all downstream of STAT1 activation (IFI 1-8d, IFI 6-16, ISG15 and IRF-7). Together, the observed temporal pattern of expression of the genes in cluster A indicates that in response to bleomycin, the earliest action includes activation of an interferon response pathway and the serum amyloid proteins (Figure 4A). This is followed by an induction of the C-C chemokines, MCP-1 and MCP-3 (Figure 4B), a delayed activation of complement factor genes and cathepsins (Figure 4C) and an even later induction of the chemokines C-10 and MIP-1g (Figure 4D).

Cluster B also included coordinately expressed subsets of genes with distinct temporal patterns. All of the extracellular matrix components in this cluster were induced progressively throughout the 14-day period examined (Figure 4H), The matrix-degrading proteinases and their inhibitors were also expressed coordinately, with an early peak at 5 days and a later peak at 14 days (Figure 4G). The genes in cluster B that peaked at the earliest time point (two days - Figure 4E) contained 2 genes, GADD45 and p21, whose expression has previously been shown to be coordinately regulated, as for example in response to p53 and/or DNA damage. Interestingly, TGFb-inducible genes were spread among all of the temporal clusters, suggesting that factors other than the activation of TGFb participate in controlling the patterns of observed expression.

An additional insight from this study was the large number of genes whose expression was decreased after treatment with bleomycin (Figure 1). Genes that are expressed at decreased levels during induction of a disease model could be as important in understanding pathogenesis as genes that are induced, and these genes have been previously largely ignored. Genes with diverse functions are represented in this group, and it is not immediately apparent how these functions are biologically linked. However, the pattern of decreased gene expression was also conserved between strains, with 71% of the genes inhibited in 129 mice also inhibited in C57bl/ 6 mice.

The results of this study identify a large number of changes in pulmonary gene expression induced by bleomycin. From cluster analysis of expression patterns in wild type and b6-/- 129 mice, we have identified distinct groups of genes that are induced in association with the inflammatory and fibrotic responses to this drug. The majority of genes in both clusters were also induced by bleomycin in C57bl/6 mice. In this strain, a detailed time course analysis revealed that the large number of genes in each of these clusters could be grouped into a series of smaller clusters that were each expressed with a distinct temporal pattern. These findings provide a framework for designing interventions that could prevent the development or progression of fibrosis at various stages of disease development.

One of the most informative clusters of genes identified was cluster B. Based on its pattern of expression i.e. little difference between wild type and b6-/- mice at baseline (Figure 1 and 2A, KO/WT Bs), but preferential induction by bleomycin in wild type mice (Figure 1 and 2A, WT 7d/Bs and WT 14d/Bs Vs KO 7d/Bs and KO 14d/Bs); this cluster is likely to contain genes that play a critical role in the fibrotic response to bleomycin. The functional properties of many genes in this cluster are known, including most of the extracellular matrix proteins within the 470 genes subjected to cluster analysis, and genes involved in modification of the matrix and in cellular responses to the matrix. This provides further evidence that these genes are critical to the development of bleomycin-induced pulmonary fibrosis. Interestingly, of the 5 genes induced early after bleomycin in this cluster (Figure 4E), two are well-characterized genes known to be induced by DNA damage i.e. GADD 45 and p21 . These genes demonstrated a unique pattern of early induction and then a persistent expression (Figure 4E) in C57bl/6 mice, and maximal at the earliest time point examined in wild type 129 mice. In contrast, these genes were minimally induced in b6^-/- 129 mice, suggesting that the expression of these genes may be important in initiating the cascade of events that leads to pulmonary fibrosis. Our observation that most of the known TGFb-inducible genes present on the array set are preferentially induced in wild type 129 and C57bl/6 mice compared with b6-/- mice (Figure 2, B and C) supports our recent finding that the integrin avb6 itself plays a role in the activation of latent TGFb1 complexes. However, the up-regulation of TGFb-inducible genes occurred with several distinct temporal patterns, an indication that factors other than activation of TGFb must be involved. These observations also indicate a limitation of the global analysis of gene expression as a method for identifying proteins that play critical roles in biological responses. This method cannot directly identify proteins that participate in biological responses solely or principally as a result of post-translational modification. Nonetheless, by simultaneously analyzing the expression patterns of large numbers of genes it is possible to detect the downstream effects of such events and possibly identify the proteins and pathways involved.

Global analysis of gene expression applied to complex multicellular organs or organisms provides a general expression pattern of the tissue but is limited in distinguishing changes in transcriptional regulation from changes in cellular composition of the organ being studied. A related limitation is the inability to ascribe changes in gene expression to events in any particular cell type. These limitations are relevant to the findings in this report. For example, we know that at baseline the lungs of b6-/- mice contain increased numbers of macrophages, lymphocytes, neutrophils and eosinophils and that bleomycin induces significant recruitment of each of these cell types to the lung. Thus, it is likely that some of the genes that are differentially expressed between wild type and b6-/- mice or between baseline and after treatment with bleomycin represent differences in cellular composition rather than differences in transcriptional regulation. However, one advantage of simultaneously analyzing the expression patterns of large numbers of genes is that the genes analyzed will include several that are restricted in their expression to specific cells. The expression of a group of cellularly restricted genes can thus be used to estimate changes in cellular content. Thus, for example, by comparing values obtained for a large number of leukocyte-restricted genes in wild type and b6-/- mice (Figure 3B and C), we have identified at least one gene, macrophage metalloelastase (MME), that was clearly induced out of proportion to any difference that could be attributed simply to a difference in macrophage number.

Tissue fibrosis represents a final common consequence for a large variety of disease processes in many organs. It is a common cause of organ dysfunction and a major cause of morbidity and mortality. Despite the medical significance of fibrosis, there are currently no approved treatments to specifically target the fibrotic process itself. This may in part be the result of the inherent difficulty in dissecting out the fibrotic response from the accompanying inflammatory response. Our approach has compared the gene expression patterns in mice that develop pulmonary fibrosis in response to bleomycin (wild type 129) with mice from the same genetic background that do not (b6-/- 129). This approach has allowed us to identify a group of genes that are likely to be directly relevant to the fibrotic process. The results presented in this paper and the availability of our large data set on the internet should provide additional insight into the disease process and accelerate the development of effective and specific interventions for the treatment of fibrosis of the lungs and other organs.

Phone: (650) 852-3190                            Phone 972-3-5302147, 972-3-9793430