Gene expression profiling in inflammatory airway disease associated with elevated adenosine

Suman K. Banerjee, Hays W. J. Young, Jonathan B. Volmer, Michael R. Blackburn


Adenosine has been implicated as a modulator of inflammatory processes central to asthma. However, the molecular mechanisms involved are poorly understood. We used Atlas mouse cDNA arrays to analyze differential gene expression in association with lung inflammation resulting from elevated adenosine in adenosine deaminase (ADA)-deficient mice. We report that of the 1,176 genes on the array, the expression patterns of 280 genes were consistently altered. Of these genes, the steady-state levels of 93 genes were upregulated and 29 were downregulated. We also show that lowering adenosine levels with ADA enzyme therapy has striking effects on gene expression that may be associated with resolution of pulmonary eosinophilia. In addition, we confirmed the nucleic acid and protein expression of vascular endothelial growth factor and monocyte chemoattractant protein-3, two candidate genes that may be regulated by adenosine. In conclusion, high-throughput profiling of gene expression by cDNA array hybridization has provided an overview of critical regulatory genes involved in airway inflammation in ADA-deficient mice. These mice will serve as a useful in vivo model for characterizing molecular mechanisms of adenosine-mediated lung damage.

  • complementary deoxyribonucleic acid
  • adenosine deaminase
  • vascular endothelial growth factor
  • monocyte chemoattractant protein-3

asthma is an inflammatory disease of the lung characterized by acute nonspecific airway hyperresponsiveness in association with chronic pulmonary inflammation (43). The disease affects ∼10% of children and 6% of adults in the United States alone, and its incidence is increasing at an alarming rate (47). It is now commonly accepted that the underlying pathophysiological aspect of bronchial asthma is airway inflammation that correlates with the severity of the disease and contributes to the development of airway hyperresponsiveness (23). Eosinophils have emerged as a major inflammatory cell type in asthma, and an increase in eosinophils is often observed in the lungs of asthmatic patients (39). Defining the signaling pathways that mediate lung eosinophilia will greatly enhance our understanding of this disease and provide important avenues for therapeutic intervention.

Adenosine is a purinergic signaling nucleoside that has been implicated in the pathogenesis of asthma (reviewed in Ref. 24). Adenosine functions as an intercellular signaling molecule by engaging G protein-coupled receptors on the surface of target cells (35). Clinical evidence has shown elevated adenosine levels in bronchoalveolar lavage fluid collected from asthmatic patients (13). Also, bronchoconstriction with inhaled adenosine in individuals suffering from asthma (9), altered adenosine receptor expression in patients with airway inflammation (45), and the therapeutic benefits of theophylline, an adenosine receptor antagonist (1), have been demonstrated. In addition, there are many in vitro studies that implicate adenosine as a modulator of the inflammatory processes that are central to asthma (24). These include the ability of adenosine to influence mast cell (31), eosinophil (45), macrophage, and epithelial cell (25) function. Most of these effects are mediated through the A2B or A3 adenosine receptors, depending on the cell type and species examined (24). Despite this evidence, a causative link between adenosine signaling and lung inflammation is unclear, and this may be attributed in part to the lack of well-developed in vivo models for the study of adenosine signaling in asthma.

To study the influence of adenosine on lung inflammation in mice, we used a two-stage genetic engineering strategy to generate mice deficient in the purine catabolic enzyme adenosine deaminase (ADA) (3). ADA plays a critical role in controlling the concentration of adenosine in cells and tissues, thereby affecting many areas of intercellular signaling (5). In the absence of ADA, the uncontrolled elevation of adenosine in vivo unleashes a variety of signaling cascades, allowing one to analyze the phenotypic and metabolic consequences of ADA deficiency. ADA-deficient mice develop a combined immunodeficiency that has been linked to profound disturbances in purine metabolism (4). In addition to immunodeficiency, ADA-deficient mice develop many of the histopathological and biochemical features seen in asthmatic patients, including lung eosinophilia, elevated IgE, activation of alveolar macrophages, and mucus hypersecretion (6). These mice fail to thrive and die from respiratory insufficiency by 3 wk of age (3, 6). The impaired pulmonary physiology and pathological changes observed in ADA-deficient mice are strongly correlated with the direct metabolic consequences of ADA deficiency (6). Lowering adenosine and 2′-deoxyadenosine levels with ADA enzyme therapy decreases lung eosinophilia, attenuates mucus production, and resolves many of the observed lung histopathologies. Hence the ability to correlate pulmonary features with disturbances in the concentrations of ADA substrates clearly provides evidence that perturbations in the signaling pathways accessed by these substrates are involved. It is also of interest that some ADA-deficient patients have elevated levels of IgE, eosinophilia, and an increased incidence of asthma (28,38). This makes the ADA-deficient mouse model an attractive in vivo model for the study of the specific roles of adenosine signaling in the mediation of the fundamental events that underlie the inflammatory process of allergic asthma.

Comparing the expression patterns of several thousand genes in both normal and ADA-deficient mice will considerably enhance the ability to identify and characterize biological roles for adenosine-regulated genes in mediating pulmonary insufficiency. A promising approach for simultaneously analyzing multiple gene expression patterns is the synthesis of cDNA probes from mRNA populations prepared from cells and/or tissues and subsequent hybridization to nucleic acid arrays (22, 37). This kind of broad-scale expression profiling, also known as cDNA microarray expression profiling, allows one to efficiently explore interrelationships among genes, providing insights into how gene expression results in a complex phenotype. Such expression profiling may serve as a powerful diagnostic tool for the classification and characterization of diseases.

In this study, we used the Atlas mouse cDNA expression array (version 1.2) for high-throughput monitoring of gene expression in normal and ADA-deficient lungs. This study has allowed us to distinguish the genes involved in several biological pathways in the lung and also to identify genes that may potentially be regulated by adenosine. When gene expression in normal lungs was compared with that in inflamed ADA-deficient lungs, only 280 genes were consistently altered. Of these genes, the steady-state levels of 29 were downregulated and 93 were upregulated. This study also showed that lowering adenosine levels with ADA enzyme therapy has striking effects on gene expression that may be associated with the resolution of pulmonary eosinophilia.


Transgenic mice.

ADA-deficient mice were generated and genotyped as described previously (3, 44). Control mice were either wild type [(+/+)] or heterozygous for the null Ada allele [(m1/+)] because there was no phenotype seen in heterozygous animals (3). All mice were housed in cages equipped with microisolator lids and maintained under strict containment protocols.

Tissue specimens.

Lungs were harvested from 18-day-old ADA-deficient mice and corresponding control mice. The effects of ADA enzyme therapy were monitored by injecting 18-day control or ADA-deficient mice intramuscularly with a single dose of 10 μl (2.5 U) of polyethylene glycol (PEG)-modified ADA (PEG-ADA) (2, 6). PEG-ADA, also known as Adagen, was graciously provided by Enzon (Piscataway, NJ). All treatment periods were 72 h. Lung halves that were used for RNA isolation were immediately snap-frozen in liquid nitrogen. The other lung halves, which were used for histopathological studies, were infused with 0.25–0.5 ml of fixative (4% paraformaldehyde in PBS). The infused lungs were placed in fixative overnight at 4°C, rinsed in PBS, dehydrated, and embedded in paraffin according to standard techniques. Sections (5 μm) were collected on microscope slides and stained with hematoxylin and eosin (Shandon-Lipshaw) according to the manufacturer's instructions. Lungs from PEG-ADA-treated mice were collected and processed in a similar manner 72 h after PEG-ADA treatment.

RNA isolation and cDNA probe synthesis.

Total RNA was isolated from whole lung tissue with the TRIzol reagent from GIBCO-BRL (Life Technologies, Grand Island, NY). For array analysis, 3 mice/group were analyzed. For quantitative RT-PCR, 6 mice/group were analyzed. Reagents from the Atlas pure total RNA labeling system (Clontech, Palo Alto, CA) were used to treat (with DNase) total RNA to eliminate potential genomic DNA contamination. This was followed by poly(A)+ RNA enrichment with streptavidin-coated magnetic beads and biotinylated oligo(dT). For first-strand cDNA synthesis, poly(A)+ RNA bound to 6 μl of resuspended beads was mixed with 1 μl of coding sequences primer mix (0.02 μM; Clontech). The RNA and primer mixture was incubated at 65°C for 2 min and then at 50°C for 2 min. The following was added to each reaction: 4 μl of 5× reaction buffer, 2 μl of 10× dNTP mix (5 mM each dCTP, dATP, dGTP, and dTTP), 0.5 μl of dithiothreitol (100 mM), 2 μl of Moloney murine leukemia virus reverse transcriptase (100 U/μl), and 5 μl of [α-32P]dATP (10 μCi/μl, 3,000 Ci/mmol; Amersham Pharmacia Biotech, Arlington Heights, IL). After incubation for 25 min at 50°C, the reaction was terminated, the labeled cDNA probes were purified, and label incorporation was assessed by scintillation counting.

cDNA array hybridization.

High-throughput gene expression profiling between ADA-deficient, PEG-ADA-treated ADA-deficient, and control mice was performed with Atlas mouse cDNA expression arrays (Clontech). Each array was spotted with cDNA fragments representing 1,176 known genes including several housekeeping genes and positive and negative controls. Radiolabeled probes were denatured and then added to separate 5-ml aliquots of ExpressHyb hybridization solution (Clontech) containing 100 μg/ml of heat-denatured sheared salmon testes DNA (Sigma) to attain a final probe concentration of ∼6 × 106 counts/min. Hybridization-cDNA probe solutions were applied to prehybridized Atlas arrays (30 min at 68°C in ExpressHyb-salmon testes DNA in the absence of labeled probe) and hybridized overnight at 68°C. After hybridization, membranes were washed three times with 200 ml of 2× saline-sodium citrate (SSC)-1% SDS solution at 68°C for 30 min followed by one wash in 200 ml of 0.1× SSC-0.5% SDS at 68°C. Membranes were then rinsed with 200 ml of 2× SSC for 5 min by continuous agitation at room temperature and exposed to X-ray film at −70°C for various lengths of time (6–72 h) to ensure linear comparison ranges for both strongly and weakly expressed genes.

Quantification of gene expression and analysis of results.

Hybridized Atlas arrays were visualized and quantitated with Atlas Image (version 1.01; Clontech). Calculated adjusted intensities (absolute intensity − background intensity) correlated linearly with the concentration of target mRNAs present in the total mRNA population. For assessing differences in gene expression between mRNA populations, the intensity value of each known gene was normalized to the averaged intensity values of designated housekeeping genes (β-actin and hypoxanthine phosphoribosyltransferase) that have been shown not to change between control and experimental samples. Some of the genes that could not be analyzed with Atlas Image because of expression levels being above the linear range were quantitated independently with Image-Pro Plus (version 4.0; Media Cybernetics, Silver Spring, MD). Only genes that showed an average multiple of induction or reduction of at least 1 SD (≥1.5-fold) above mean expression levels were considered differentially expressed. This level was chosen to help maintain relevant correlations and observations within and between different arrays and experiments consistent. Because we calculated the multiple of induction or reduction ratios by using the mean values of the average differences of several mice, a change of <1.5-fold was not considered substantial.

Immunohistochemical localization of monocyte chemoattractant protein-3 and vascular endothelial growth factor.

Paraffin-embedded tissues were sectioned (5 μm), exposed to two changes of Histoclear, and rehydrated in a series of graded alcohols to water. Antigen unmasking was performed before monocyte chemoattractant protein (MCP)-3 localization with target retrieval solution according to the manufacturer's guidelines (DAKO, Carpinteria, CA). Endogenous biotin activity was blocked with avidin and biotin (biotin blocking kit; DAKO), and endogenous peroxidase activity was blocked by incubation in 0.3% hydrogen peroxide for 5–10 min. Immunohistochemistry (IHC) for MCP-3 and blocking procedures was performed according to the manufacturer's guidelines with a goat IgG VECTASTAIN Elite ABC kit (Vector Laboratories, Burlingame, CA). MCP-3 localization was performed by incubating slides for 30 min at room temperature with a 1:4 dilution of goat anti-mouse MARC/MCP-3 antibody (R&D Systems, Minneapolis, MN) as the primary antibody. Negative controls consisted of sections incubated with rabbit serum only. For vascular endothelial growth factor (VEGF) IHC, lung sections were incubated with either a 1:200 dilution of murine IgG2 monoclonal antibody that recognized all isoforms of VEGF (Santa Cruz Biotechnology, Santa Cruz, CA) or a 1:200 dilution of a nonspecific anti-mouse IgG2 control antibody (DAKO) for 1 h. After incubation with appropriate biotinylated secondary antibodies, the slides were incubated with either streptavidin-peroxidase enzyme conjugate (DAKO) for 15 min for VEGF IHC or with avidin-biotinylated peroxidase complex (Vector Laboratories) for 30 min for MCP-3 IHC. The slides were developed with 3,3-diaminobenzidine tetrachloride (DAKO) for 7–10 min, dehydrated, and mounted with Permount.

Quantitative real-time RT-PCR.

Quantitative real-time RT-PCR was performed with the 7700 sequence detector (Applied Biosystems, Foster City, CA) (22). Specific quantitative assays for VEGF and MCP-3 were developed with Primer Express software (Applied Biosystems) following the recommended guidelines based on sequences from GenBank. The sequences of all the oligonucleotides used are given in Table1. VEGF primers and probes were designed to recognize all isoforms of VEGF. Total RNA was isolated from whole lung tissue with the TRIzol reagent from GIBCO-BRL followed by DNase treatment to eliminate potential genomic DNA contamination. This was followed by cDNA synthesis and real-time PCR with established protocols (12). The resulting data were analyzed with SDS software (Applied Biosystems, Foster City, CA), with TAMRA as the reference dye. The final data were normalized to β-actin and are presented as the molecules of transcript per molecules of β-actin × 100 (%β-actin). Results are expressed as means ± SE.

View this table:
Table 1.

Primer pairs and internal probe sequences for MCP-3 and VEGF used for real-time RT-PCR


Gene expression profiles between control and ADA-deficient lungs.

Gene expression in lung tissue isolated from 18-day control and ADA-deficient mice was characterized by examining the expression of 1,176 known regulatory genes with Atlas mouse cDNA expression arrays (version 1.2). Experiments were performed with two different sets of Atlas array membrane lots to enhance detection of genes that show only modest differential expression between control and ADA-deficient mice. This also helped to minimize the number of spuriously identified, differentially expressed genes. The array membranes were hybridized with cDNA probes synthesized from mRNA isolated from control and ADA-deficient lung tissue. The hybridization results of a typical experiment are shown in Fig. 1, A(control) and B (ADA deficient). Of the 1,176 genes, an average of 365 genes were expressed in each experiment. Only 280 of these genes were consistently expressed between experiments and used for data analysis. Interestingly, there were always fewer genes expressed in control compared with ADA-deficient mice. Yet, importantly, when different lung mRNA populations from control mice were compared by scatterplot analysis, the profiles and concentration levels of the expressed genes represented in each mRNA population revealed a tight distribution pattern along the diagonal “line of identity” (Fig. 2 A). This ensured the integrity of mRNA isolation, probe preparation, hybridization conditions, and data analysis. Similar analysis of lung mRNA isolated from ADA-deficient mice showed reproducibility (Fig.2 B). Scatterplot analysis of gene expression data between control and ADA-deficient mice (Fig. 2 C) revealed a much wider distribution pattern. Although the majority of expressed genes lay relatively close to the diagonal line of identity, the steady-state levels of 29 were downregulated and 93 were upregulated. Only those genes (gene and/or protein name and GenBank accession nos.) that exhibited an average upregulation or downregulation of at least 1 SD (≥1.5-fold) above mean expression across all experiments are listed as differentially expressed (Table 2). These changes in gene expression are highly reproducible and represent changes in the expression of a variety of molecular markers including transcription factors, cell surface antigens, cell cycle regulators, and cell adhesion receptors. The complete grouping of genes into functional clusters was provided by Clontech's functional database. These findings demonstrate that gene expression patterns in ADA-deficient lungs, which have characteristically elevated adenosine levels, differ from expression patterns in control lungs.

Fig. 1.

cDNA array images of the expression pattern of genes in 18-day control (A) and 18-day adenosine deaminase (ADA)-deficient (B) lungs. Differential hybridization of identical Atlas mouse cDNA expression arrays was performed. The 1,176 cDNAs of known regulatory genes that survey major biological pathways are spotted on each membrane in 6 panels and grouped according to function. Nine putative housekeeping genes and negative controls are located in the bottom row. A complete list of gene names and their locations is available at Clontech's Web page ( atlasinfo/–1&display=genelist& subapp=atlasinfo).

Fig. 2.

Scatterplot analysis and comparison of log-transformed expression data for 2 different mRNA populations isolated from lungs of control (A) and ADA-deficient (B) mice. Differential gene expression is represented by scatterplot analysis of averaged log-transformed expression data for 2 different mRNA populations isolated from control and ADA-deficient lungs (C). Each point represents the expression level of an individual gene within both mRNA populations. Solid line, predicted line of identity; the distance that a point lies along this line denotes its level of expression, particularly in relationship to other genes. The perpendicular distance of a point away from the diagonal line represents differential expression of a gene between the 2 mRNA populations. Distribution pattern of most points in control (A) and ADA-deficient (B) mice lies within the predicted range of a 2-fold or less difference in gene expression (dashed lines). The distribution patterns in C show a wider scatter. Many points lie outside the predicted range of 2-fold (dashed lines), representing significant changes in gene expression between the 2 mRNA populations. Monocyte chemoattractant protein (MCP)-3 and fibroblast growth factor (FGF)-4 are examples of genes in which expression levels changed by a factor of >7-fold.

View this table:
Table 2.

Differentially expressed genes between 18-day control and ADA-deficient murine lungs

PEG-ADA treatment has been used to effectively lower adenosine levels (2). To identify those genes directly affected by the lowering of adenosine levels, ADA-deficient mice were treated with PEG-ADA and gene expression patterns were analyzed. Genes that were differentially regulated by PEG-ADA treatment by >1 SD above or below mean expression are listed in Table 2. No differences in gene expression were observed between control mice and age-matched control mice treated with PEG-ADA (data not shown). The array technology has thus provided insights into differential gene expression and potential regulation by adenosine in ADA-deficient lungs.

Gene-specific real-time RT-PCR analysis.

We used real-time RT-PCR to confirm the differential expression of VEGF and MCP-3 transcripts. VEGF and MCP-3 expression exemplify genes that were affected by PEG-ADA treatment and can be taken as examples to confirm the reliability of Atlas hybridization. Furthermore, there is in vitro evidence to suggest that adenosine signaling can regulate VEGF (34) and MCP-3 (27) expression. The expression of VEGF, which regulates vascular permeability and endothelial function, was decreased 1.9-fold in ADA-deficient lungs. After PEG-ADA treatment in ADA-deficient mice, we observed an increase in the expression of VEGF by a factor of 2.4-fold. As shown in Fig.3 A, VEGF expression analysis by real-time RT-PCR was similar to that seen in the Atlas array. VEGF expression was significantly decreased in ADA-deficient lungs compared with expression in control lungs, and a significant increase in VEGF expression was seen after PEG-ADA treatment in ADA-deficient mice compared with that in untreated ADA-deficient mice. No significant difference in VEGF expression was observed between control and ADA-deficient mice treated with PEG-ADA.

Fig. 3.

Real-time RT-PCR analysis of vascular endothelial growth factor (VEGF; A) and MCP-3 (B) transcripts. Relative VEGF and MCP-3 mRNA transcript levels were calculated by dividing VEGF levels by β-actin levels measured in the same RNA preparations. VEGF transcripts were significantly reduced in 18-day ADA-deficient (ADA-def) lungs (n = 6) compared with expression of VEGF in control lungs (*P = 0.006 by Student'st-test). Polyethylene glycol (PEG)-ADA treatment for 72 h significantly enhanced VEGF expression in ADA-def lungs (ADA-def + PEG-ADA; n = 6; # P = 0.005 by Student's t-test). No significant difference in transcript expression was observed between control and ADA-def + PEG-ADA lungs. B: MCP-3 transcripts were significantly elevated in ADA-def lungs (n = 6) compared with expression of MCP-3 in control lungs. *P = 0.006 by Student's t-test. PEG-ADA treatment for 72 h significantly decreased MCP-3 expression in ADA-def lungs (n = 6; # P = 0.008 by Student's t-test). No significant difference in transcript expression was observed between control and ADA-def + PEG-ADA lungs.

MCP-3 is a C-C chemokine that can regulate chemotaxis of inflammatory cells. According to Atlas hybridization results, the expression of MCP-3 increased in ADA-deficient lungs, and expression decreased after PEG-ADA treatment. Real-time RT-PCR analysis of MCP-3 transcript expression (Fig. 3 B) revealed a significant increase in expression in ADA-deficient lungs. After PEG-ADA treatment in ADA-deficient mice, we observed a significant reduction in MCP-3 expression. This real-time RT-PCR expression pattern of VEGF and MCP-3 mimics the pattern seen in the Atlas array, thus confirming results of the differential Atlas array analysis. These findings suggest that elevated adenosine can differentially regulate VEGF and MCP-3 expression in the lung.

Protein localization of VEGF and MCP-3 by IHC.

Once differential transcript expression of VEGF and MCP-3 was confirmed, we sought to localize protein expression with IHC staining. We found that the airway epithelial expression of VEGF (Fig.4) and MCP-3 (Fig.5) was regulated by PEG-ADA treatment in ADA-deficient mice. ADA-deficient lungs exhibited reduced staining for VEGF in the airway epithelium (Fig. 4 B) compared with intense staining in the airways of control lungs (Fig. 4 A). PEG-ADA treatment in ADA-deficient mice enhanced VEGF expression in the airway epithelium (Fig. 4 C). VEGF staining was also detected in type II alveolar epithelial cells, blood vessels, alveolar macrophages, and smooth muscle and pulmonary parenchyma; however, relative levels of staining intensity did not change after PEG-ADA treatments. No staining was observed in sections incubated with control antibody (Fig. 4 D). These findings for VEGF expression further confirmed our results as obtained by array analysis and real-time RT-PCR. Similarly, we found that MCP-3 expression was predominantly increased in the airway epithelium of ADA-deficient mice (Fig. 5 B). There was uniform staining of almost all bronchial epithelial cells. Control mice (Fig. 5 A) and PEG-ADA-treated ADA-deficient mice (Fig. 5 C) showed reduced MCP-3 protein in the airway epithelium. Staining for MCP-3 was also observed in mononuclear cells of the lung parenchyma in all sections; however, the relative staining intensity did not change after PEG-ADA treatments. This pattern of MCP-3 expression corroborated our findings with array analysis and real-time RT-PCR. These findings suggest that the metabolic consequences of ADA deficiency can regulate the expression of VEGF and MCP-3 in the airway epithelium.

Fig. 4.

Immunohistochemical staining of lung tissue for VEGF. Sections were obtained from lungs of 18-day-old control mice and ADA-def mice with and without PEG-ADA treatment. Specimens were processed for immunohistochemical localization of VEGF with a mouse monoclonal (IgG2a) VEGF antibody. A: intense staining for VEGF in the airway epithelium (arrows) of control lungs. B: 18-day ADA-def lung showing reduced staining in airway epithelium.C: 18-day ADA-def + PEG-ADA lung showing staining in the airway epithelium. D: control lung tissue treated with negative control antibody in lieu of anti-VEGF. Bar, 50 μm.

Fig. 5.

Immunohistochemical staining of lung tissue for MCP-3. Sections were obtained from lungs of 18-day-old control and 18-day-old ADA-def mice with and without PEG-ADA treatment. Specimens were processed for immunohistochemical localization of MCP-3 with a goat anti-mouse MCP-3 polyclonal antibody. A: staining in control lungs showing MCP-3 expression along the airway epithelium (arrow). B: 18-day ADA-def lung showing intense staining for MCP-3 in airway epithelium. C: 18-day ADA-def + PEG-ADA lung showing MCP-3 expression along airway epithelium similar to that seen in control tissue. D: ADA-deficient lung treated with control serum. Bar, 50 μm.

ADA enzyme therapy results in decreased expression of several genes regulating pulmonary inflammation in ADA-deficient mice.

Eighteen-day-old ADA-deficient mice developed severe pulmonary inflammation characterized by the accumulation of enlarged and foamy macrophages, eosinophils, and multinucleated giant cells (6). Eosinophils were found in the interstitium and luminal spaces throughout the lung, with high concentrations around the bronchioles and pulmonary blood vessels (Fig.6, B and C). PEG-ADA treatment rapidly reversed the eosinophilia seen in the 18-day-old ADA-deficient mice (Fig. 6 D) in association with the lowering of adenosine levels (6), suggesting that adenosine may regulate airway eosinophilia. Hence treatment with PEG-ADA had pronounced effects on airway inflammation.

Fig. 6.

PEG-ADA treatment for 72 h reverses airway eosinophilia in ADA-def mice. A: hematoxylin and eosin (H&E)-stained 18-day-old control lung. B: low-power magnification of H&E-stained 18-day-old ADA-def lung. Arrow, areas of inflammation; arrowhead, presence of activated foamy macrophages. C: high magnification of H&E-stained ADA-def lung demonstrating eosinophil infiltration (arrow) and multinucleate giant cells (arrowhead). Note the pronounced enlargement of alveolar spaces (AS). D: low-power magnification of H&E-stained lung from 18-day-old ADA-def + PEG-ADA mouse. E: high magnification of 18-day-old ADA-def + PEG-ADA mouse lung. Note the characteristic absence of eosinophils. Enlarged AS and macrophages (arrow) persisted after 72 h of PEG-ADA treatment. Bars, 100 μm.

An example of a cluster of genes known to modulate inflammation and cell-cell adhesion was identified (Table 2) and is represented in Fig.7, A and B, respectively. MCP-3, CD32, urokinase-type plasminogen activator (uPA), osteopontin, fibronectin, and P-selectin are known genes that can mediate eosinophil trafficking. The expression of these genes was elevated in ADA-deficient lungs, and PEG-ADA treatment decreased their expression in conjunction with decreased lung eosinophilia. Transforming growth factor-β2, integrin β1, and intercellular adhesion molecule-1 mediate macrophage trafficking. Macrophage numbers were unaffected by 72 h of PEG-ADA treatment (6) (Fig. 6 E), and, coincidentally, the expression of these genes was unchanged. The expression of several airway-relevant cysteine proteases (cathepsins A, B, D, H, and L) and antiproteases (cystatins) were also affected by PEG-ADA treatment, suggesting that regulating adenosine levels may also help regulate the expression of these genes that are thought to play a critical role in maintaining alveolar morphology. Thus microarray analysis has identified many genes that may be involved in mediating adenosine-related changes in inflamed ADA-deficient lungs.

Fig. 7.

PEG-ADA treatment regulates expression of several genes modulating pulmonary inflammation (A) and cell adhesion (B). Gene expression in ADA-def and ADA-deficient + PEG-ADA mice are compared with gene expression in control mice. Gene expression in control lungs in this cluster is at the intersection of and axes at zero. Absence of a solid bar next to an open bar indicates expression of that particular gene in ADA-def lung after PEG-ADA treatment was not different from expression in control lungs. CTSD, CTSB1, CTSC, CTSH, and CTSL represent cathepsins D, B, C, H, and L, repectively. SPI2, serine protease inhibitor 2; Cys 3, cystatin C; MCSF, macrophage colony-stimulating factor; TGF, transforming growth factor; UPA, urokinase-type plasminogen activator; ICAM, intercellular adhesion molecule; LFA, leukocyte adhesion protein.


Adenosine has long been thought to play a role in the pathogenesis of allergic airway disease and asthma (9, 13, 30, 45). However, the mechanisms of adenosine-mediated lung damage are poorly understood. In the current study, we used cDNA microarray technology to profile the expression of 1,176 known regulatory genes in trying to uncover some of the regulatory pathways involved in adenosine-mediated lung damage. Our laboratory (3, 6) has used a mouse model in which lung adenosine levels are markedly elevated in association with lung inflammation and damage. The results of this study identify a number of differentially expressed genes that may be regulated by adenosine and hence play a pivotal role in modulating the underlying lung pathology.

ADA enzyme therapy, which is directed at lowering adenosine levels, effectively altered differential gene expression in ADA-deficient lungs in association with attenuating lung eosinophilia. This suggests that the differential gene expression observed in ADA-deficient lungs was the consequence of elevated adenosine levels. This is consistent with past evidence suggesting that adenosine signaling may play an important role in the type of inflammation and tissue damage seen in asthma (24). In humans, 10–100 μM concentrations of adenosine induce bronchoconstriction in a dose-dependent manner in vitro, and similar levels have been demonstrated in bronchoalveolar lavage fluid, suggesting physiological relevance (13). These levels are similar to the 100–125 μM concentration of adenosine measured in whole lungs of ADA-deficient mice (M. R. Blackburn, unpublished observations). These results provide supporting evidence for a mechanism of adenosine-mediated inflammatory events that affect smooth muscle contraction and airway damage. Perhaps the best-characterized influence of adenosine on inflammatory cells is its ability to increase mast cell degranulation (31). This effect is likely mediated through the A2B adenosine receptor in humans (14) and the A3 adenosine receptor in rodents (36). In addition, adenosine signaling can influence the function of eosinophils (45), macrophages (20), and airway epithelial cells (25), all of which play important roles in inflammatory lung diseases such as asthma. Characterizing the pattern of gene expression in inflamed lungs with increased adenosine levels will be important in our attempts to understand the molecular mechanisms of adenosine-mediated lung damage.

Results from our microarray analysis provide us with tentative regulatory pathways that may be involved in the molecular basis for inflammatory cell trafficking in ADA-deficient lungs. Some of the upregulated and downregulated genes in ADA-deficient lungs and their response to PEG-ADA treatment are analyzed here. We have attempted to correlate gene expression pattern with inflammatory changes observed in ADA-deficient lungs. Although associations between increased adenosine and alterations in gene expression can be made from these studies, our results do not establish a cause-and-effect relationship between adenosine signaling and the expression of the identified genes. More in-depth analysis is needed to first quantify the expression of the genes found and then establish that they are directly regulated by adenosine signaling as opposed to being influenced by upstream mediators. Nonetheless, the information from this analysis can allow us to begin to identify clusters of genes involved in the lung inflammation and damage seen in ADA-deficient mice. One such informative cluster of genes, illustrated in Fig. 7, can be associated with the increased pulmonary eosinophilia, macrophage activation, and airway damage observed in ADA-deficient lungs. Several molecules such as uPA and its receptor (CD87), along with B7.2, are genes known to regulate the development of murine allergic asthma by mediating inflammatory cell influx (26, 29). In addition, the role of adhesion molecules intercellular adhesion molecule-1, P-selectin, and leukocyte adhesion protein-1 in mediating eosinophil trafficking in a murine model of pulmonary inflammation is well documented (7, 8). By being able to regulate expression of these molecules with adenosine modulation, we can hypothesize about possible mechanisms of airway eosinophilia in ADA-deficient mice. For example, regulating adenosine levels may be involved in determining interactions of CD44 with osteopontin, collagen, fibronectin, and laminin, all of which are important in promoting macrophage and eosinophil adhesion and migration (17, 18).

Protease-antiprotease imbalances (41) have long been known to mediate inflammatory and morphological changes in airway disease. The cathepsins A–D, H, and L belong to a family of cysteine proteases that have been implicated in alveolar destruction (46). An abnormal alveolar phenotype is a well-characterized anatomic feature in ADA-deficient lungs (6). Here, we show that adenosine levels may regulate expression of cathepsins. Although lowering adenosine levels decreases cathepsin expression, the fact that no changes are seen in alveolar phenotype may be attributed to the permanent nature of these defects at this stage. In support of this, we have shown that the abnormal alveolar phenotype in ADA-deficient mice may be reversed by PEG-ADA treatment starting at birth (2).

Global analysis of gene expression provides a general expression profile of the tissue but is limited in its ability to distinguish changes in transcriptional regulation from changes in cellular composition of the organ being studied. Related to this limitation is the inability to ascribe changes in gene expression to events in any particular cell type. For example, we know that ADA-deficient lungs contain increased numbers of macrophages and eosinophils (2). Thus it is likely that some of the genes that are differentially expressed between ADA-deficient, control, and PEG-ADA-treated ADA-deficient mice represent differences in the cellular composition rather than in transcriptional regulation. This is likely the case with the expression of CD32, which is constitutively expressed on murine eosinophils and is a well-recognized marker for eosinophilia (10). Hence, it is not surprising that PEG-ADA treatment, which results in attenuated lung eosinophilia, is also associated with decreased CD32 expression in ADA-deficient lungs. To distinguish between differential gene expression in ADA-deficient lungs as a result of altered cell populations and the direct regulatory actions of adenosine, we have begun more in-depth analysis to identify genes that are directly regulated by adenosine in the lung. VEGF and MCP-3 gene expression patterns, as elucidated by array analysis, were confirmed by real-time RT-PCR (Fig. 3, A and B). This provided conclusive evidence of the reliability of array hybridization data. Furthermore, immunohistochemical localization of VEGF and MCP-3 illustrated differential expression, predominantly in the airway epithelium (Figs. 4 and 5). Thus the differential expression of MCP-3 and VEGF is not due to increased cell populations but rather reflects a direct modulation of airway epithelial expression by the regulation of adenosine.

VEGF is an endothelial mitogen that increases endothelial permeability and induces endothelial cell growth (33). In allergic airway diseases such as asthma, maintenance of vascular integrity and capillary permeability is critical in preventing interstitial accumulation of inflammatory cells and development of edema, and VEGF has been inversely correlated with asthma severity (11). On the basis of our findings in vivo, we can speculate that alterations of adenosine levels in vivo interfere with VEGF-mediated endothelial cell survival and hence have an impact on vascular integrity. Adenosine is an important regulator of VEGF expression (19, 34); however, its effects are determined by cell type and specific adenosine receptor engagement (19, 34, 40). Preliminary findings in ADA-deficient lungs suggest upregulation of A3 and A2B receptors in the bronchial epithelium (M. R. Blackburn, unpublished observations). These receptors couple to effector systems that regulate intracellular cAMP levels and intracellular Ca2+ mobilization (35). Engagement of these signaling pathways may regulate the expression of VEGF in bronchial epithelial cells. Regulation of VEGF by adenosine in the bronchial epithelium likely exerts significant effects on the observed pulmonary pathology in ADA-deficient mice.

MCP-3 belongs to a family of C-C chemokines that has attracted recent attention because of its diverse role in allergic inflammation (16, 21, 32, 42, 48). Here, we provide evidence that MCP-3 expression is increased in an adenosine-rich environment. This increased expression of MCP-3 may play a proinflammatory role and mediate eosinophil recruitment and chemotaxis in ADA-deficient lungs. Conversely, because posttranslationally modified MCP-3 can serve as a chemokine receptor antagonist that dampens inflammation (32), MCP-3 increases may represent an effort to control the level of inflammation seen in the lungs of ADA-deficient mice. It is of interest that transcriptional and posttranscriptional regulation of MCP-3 expression is mediated by cAMP signaling (27). Because adenosine can regulate cAMP levels by engaging G protein-coupled receptors (15), one can speculate that adenosine signaling may be responsible for the increased MCP-3 expression in the airways of ADA-deficient mice. Further characterization of adenosine receptor expression in the airways of ADA-deficient lungs in association with MCP-3 expression is currently under investigation.

In conclusion, we have demonstrated the utility of cDNA microarrays in analyzing molecular changes in ADA-deficient lungs. It may be deduced from our results that multiple genetic changes and gene products are required for the pathogenesis of airway inflammation in ADA-deficient mice. Such primary findings provide important clues to the communication among genes and contribute to the exploration of potential target genes for possible molecular diagnosis and therapy. A reassuring aspect of our results was the overlap in findings from past studies directed at airway inflammation. Equally important was our finding that adenosine has a significant impact on the regulation of the expression of these genes. Genes that play important roles in common pathways need to be analyzed carefully and studied further to help characterize specific actions of adenosine. In addition, the correlation of increased lung adenosine and asthma and the ability to relieve lung eosinophilia by lowering adenosine levels suggests that ADA enzyme therapy may be beneficial in the treatment of eosinophilic lung inflammation. ADA-deficient mice may thus serve as a useful experimental means to study high-throughput gene expression and differential gene regulation involved in the pathogenesis of asthma.


We thank Enzon for the gracious gift of Adagen. We also thank Gregory Shipley at the Quantitative Genomics Core Laboratory facility at the University of Texas-Houston Medical School for assistance in conducting real-time RT-PCR experiments.


  • This work was supported by National Institutes of Health Grants AI-43572 and HL-61888 (to M. R. Blackburn) and a Junior Investigator Award from the Sandler Family Supporting Foundation (to M. R. Blackburn).

  • Some of the reported studies were presented at the annual meeting of the American Thoracic Society, San Francisco, CA, May 2001, and were published in abstract form (Am J Respir Crit Care Med163: A424, 2001).

  • Address for reprint requests and other correspondence: M. R. Blackburn, Dept. of Biochemistry and Molecular Biology, Univ. of Texas-Houston Medical School, 6431 Fannin St., Houston, TX 77030 (E-mail: Michael.R.Blackburn{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • 10.1152/ajplung.00243.2001


View Abstract