Transcriptional Adaptation to Cystic Fibrosis Transmembrane Conductance Regulator Deficiency*

Cystic fibrosis, the most commonly inherited lethal pulmonary disorder in Caucasians, is caused by mutations in the cystic fibrosis transmembrane conductance regulator gene (CFTR). To identify genomic responses to the presence or absence of CFTR in pulmonary tissues in vivo, microarray analyses of lung mRNAs were performed on whole lung tissue from mice lacking (CFTR(−)) or expressing mouse CFTR (CFTR(+)). Whereas the histology of lungs from CFTR(−) and CFTR(+) mice was indistinguishable, statistically significant increases in the relative abundance of 29 and decreases in 25 RNAs were identified by RNA microarray analysis. Of RNAs whose expression was consistently altered by the absence of CFTR, functional classes of genes influencing gene transcription, inflammation, intracellular trafficking, signal transduction, and ion transport were identified. RNAs encoding the transcription factor CCAAT enhancer-binding protein (CEBP) δ and interleukin (IL) 1β, both known to regulate CFTR expression, were induced, perhaps indicating adaptation to the lack of CFTR. RNAs mediating lung inflammation including calgranulin-S100 family members, IL-1β and IL-4, were increased. Likewise, expression of several membrane transport proteins that interact directly with CFTR were increased, suggesting that CFTR-protein complexes initiate genomic responses. Absence of CFTR influenced the expression of genes modulating diverse pulmonary cell functions that may ameliorate or contribute to the pathogenesis of CF.

lation, recurrent infections, and excessive inflammation in the lung. Whereas the pathogenesis of CF is not fully understood, abnormalities in cyclic AMP-dependent chloride secretion and excessive sodium reuptake by epithelial cells related to CFTR deficiency are thought to alter fluid homeostasis at the airway surface liquid leading to its dehydration, impaired mucociliary clearance, and infection (see Ref. 3 for review). Because the elucidation of the primary structure of CFTR, a myriad of functions and numerous interactions with other cellular proteins have been ascribed to CFTR. Thus, in addition to the role of CFTR in the regulation of cAMP-dependent chloride transport, this protein may play pleotropic roles in many cellular processes by interacting with the cytoskeleton, membrane transport proteins, as well as receptors, protein routing and degradation machinery (2). A number of studies support the concept that the excessive inflammatory responses occur in the CF lung, but the mechanisms underlying these abnormalities have not been clarified. Changes in levels of IL-8 and other proteins mediating inflammatory signaling including NFB and iNOS have been associated with CF, in the presence or absence of infection, raising the possibility that abnormalities in CFTR may constitutively alter pathways mediating inflammation (4 -6).
In the lung, CFTR is distributed primarily in apical regions of airway and submucosal gland epithelial cells (7). Abundance and cellular sites of expression of CFTR are strongly influenced by developmental, spatial, and humoral factors, supporting the concept that the expression and function of CFTR are regulated at both transcriptional and post-transcriptional levels. Despite extensive study, the precise role of CFTR in the pathogenesis of CF disease remains poorly understood. At the clinical level, severity of CF disease is highly variable even among individuals bearing identical mutations, supporting the concept that environmental and hereditary factors may influence the severity of the disorder (2). These clinical observations, and observations demonstrating strain differences in the severity of CF phenotype after CFTR gene targeting or mutation in mice (8), support the concept that the expression of CFTR and its function in cellular processes may be influenced by many genes or pathways intensifying or mollifying CF disease in various organs. Morbidity and mortality in patients with CF is strongly associated with pulmonary disease caused by mucous accumulation, inflammation, and infection; however, deletion of CFTR mice does not cause significant pulmonary disease suggesting that expression of alternative channels or other complementary genes maintains pulmonary homeostasis in the mouse. Whereas numerous in vitro and in vivo models have been developed for study of CFTR, analysis of genomic responses to the presence or absence of CFTR are complicated by heterogeneity of cell models, and culture conditions that may influence cell function and gene expression independently of CFTR. Direct RNA analysis of pulmonary tissue from humans with CF is complicated by the nearly ubiquitous, severe pulmonary infections that may secondarily modify cellular responses and gene expression, complicating identification of responses to CFTR in vivo.
In the present study, we undertook experiments to identify RNAs influenced by the presence and absence of CFTR in vivo, seeking to identify genes and pathways that interact with or compensate for the CFTR to maintain pulmonary function. In this study, stereotypic genomic responses to the lack of CFTR were observed in pulmonary tissues in the absence of infection or disease.

MATERIALS AND METHODS
Transgenic mice bearing a null mutation in CFTR (CFTR(Ϫ)), generally succumb to intestinal disease in the weanling period (9). To generate healthy mice deficient in CFTR, the human CFTR cDNA was expressed in the intestinal epithelium under control of the intestinal fatty acid-binding protein gene promoter (iFABP), fully correcting small intestinal pathology and supporting normal postnatal survival of CFTR(Ϫ) mice (10). The iFABP-hCFTR, mCFTR(Ϫ) mice have been maintained in a mixed FVB/N, C57BL/6 background without evidence of gastrointestinal or pulmonary disease for nearly a decade in our laboratory. Histological andbiochemical studies identified no overt pathology in lung tissue from these mice compared with CFTR expressing littermate controls (10,11). Mice were maintained in filtered microisolator cages. Sentinal mice were free of mouse pathogens. Lungs of adult iFABP-hCFTR, mCFTR(Ϫ), and control mice were free of bacterial pathogens or colonization as assessed by quantitative culture of lung homogenates on blood agar plates.
RNA Microarray and Data Analysis-Total RNA was subjected to reverse transcription using oligo(dT) with T7 promoter sequences attached, followed by second strand cDNA synthesis. Antisera cRNA was then amplified and biotinylated using T7 RNA polymerase, prior to hybridization to the Affymetrix genechip mouse U74aV2 using the Affymetrix recommended protocol (12,13). Affymetrix MicroArray Suite version 5.0 was used to scan and quantitate the genechips using default scan settings. Intensity data were collected from each chip, scaled to a target intensity of 1500, and the results were analyzed using both MicroArray Suite and GeneSpring 5.0 (Silicon Genetics, Inc., Redwood City, CA). cDNAs were hybridized to U74aV2 chips (Affymetrix Inc.). Hybridization data were normalized in a CCAAT enhancer-binding protein two-step process to remove or minimize systemic sources of variation at both chip and gene level. Specifically, each chip was normalized to the distribution of all genes on the chip to control for variation between samples. Each RNA from mCFTR(Ϫ) mice was normalized to its specific control (i.e. sex and age-matched mCFTR(ϩ) littermates). Data were further transformed into log ratio for analysis and symmetry of distribution. Changes in mRNAs were identified by the combination of a distribution analysis (JMP4, SAS Institute, Inc.), and the Welch analysis of variance. Outlier box and quartile box plots were used to identify outliers with the definition of up-outlier Ͼ upper quartile ϩ 1.5 (interquartile range), and the down-outlier Ͻ lower quartile Ϫ 1.5 (interquartile range). Significant changes were calculated by the Welch t test at p Ͻ 0.05. Adjusted p values were calculated by Westfall and Young permutation for correction of false positives (GeneSpring 4.2.1, Silicon Genetics). Comparisons between genotype and age groups were performed using one-way analysis of variance. To identify genes that were differentially expressed because of CFTR genotype regardless of age, hierarchical, and k-means clustering were used to identify consistent changes in gene expression in response to the lack of CFTR at all three time points. Candidate RNAs were further filtered on the basis of reproducibility and absolute intensity. Mean, standard deviation, and coefficient variation were calculated for each replicate. Replicates with coefficient variations Ͼ 50% were deleted from analysis. Genes whose expression was below the level of detection were eliminated as experimental noise.
Pathway and Literature Analysis-Selected genes were subjected to intensive search to identify biological function and associated regulatory pathways. A U74Av2 annotation data base with system identifiers was constructed for all the array elements and their associated Gen-Bank TM accession numbers. Gene description, functional categories, biological processes, molecular functions, cellular components, protein domain, and literature information were identified. Information resources included NetAffy (www.affymetrix.com), Source Search genome-www5.stanford.edu/cgi-bin/SMD/source/), BLAST NCBI, Locus Link, mouse-human homolog search (www.ncbi.nlm.nih.gov), and Gene Ontology data base (www.godatabase.org/chi-bin/go.cgi). Differentially expressed genes were classified into functional categories based on the gene ontology definition. To determine which functional category is overrepresented in the selected gene list, the binomial probability was calculated for each category using the entire U74Av2 (contains 12488 mice genes) as a reference data set. The binomial probability is defined by the following.
It returns the probability of getting k successes of n trials if the probability is p in the given population (U74Av2). Potential protein/protein or protein/DNA interactions were identified using the published literature information.
Lung Histology and Immunohistochemistry-Lungs from postnatal animals were inflation fixed with 4% paraformaldehyde at 25 cm H 2 O pressure via a tracheal cannula. Lung tissue was processed according to standard methods and embedded in paraffin. Procedures for immunostaining were previously described (14). Rabbit monoclonal antibody against the 110-kDa Mac-3 antigen was used at 1:40,000 to identify alveolar macrophages (BD Pharmingen, San Diego, CA).
Numbers and histology of alveolar macrophages were not altered by mCFTR.
Identification of Gene Responses to the Lung CFTR Deletion-To identify genes responsive to CFTR, lung RNAs from iFABP-hCFTR, mCFTR(Ϫ), iFABP-hCFTR, mCFTR(ϩ), mCFTR(Ϫ) and mCFTR(ϩ) littermates at the age of 3, 6, or 11 weeks of age were compared. Microarray analyses were performed in duplicate from RNA isolated at 3 and 6 weeks of age. Data from 10 Affymetrix Murine Genome U74Av2 chips were normalized and statistical differences between CFTR-deficient (CFTR(Ϫ)) and control (CFTR(ϩ)) mice were identified. Differences related to age were identified by outlier analysis and/or unpaired t test. After normalization, normal distributions were observed in the intensity data from lung tissue obtained at all ages. Lung RNA data from 3-week-old mCFTR(ϩ) and mCFTR(Ϫ) mice (lacking the iFABP-hCFTR transgene) were similarly distributed to those bearing the FABP-hCFTR gene and were, therefore, included in the analysis.
To identify RNAs that were differentially expressed in response to CFTR regardless of age, mCFTR(Ϫ) and mCFTR(ϩ) data were separated into two groups. The log-ratio distribution and outlier plot of the combined data set are represented by Fig. 2. A total of 1977 outliers were identified from 12442 genes/expressed sequence tags analyzed. The abundance of 848 RNAs was increased; 1129 were decreased. Welch t test together with the Westfall and Young step-down permutation further narrowed the number of differentially expressed RNAs to 315. Hierarchical clustering was used to visualize and classify the data set (Fig. 3). Data are shown in a two-dimensional matrix to identify groups of genes with similar expression patterns and show remarkably ordered gene expression profiles of 315 selected genes. On the chip level (top dendrogram) RNAs influenced by CFTR formed two distinct groups. Within each group, the samples collected from age-matched pairs were more closely related than those from different ages, suggesting that age also influenced gene expression. At the RNA level (the dendrogram at the left side), genes were clearly separated into two major groups: those mRNAs increased or decreased in mCFTR(Ϫ) mice. Genes were further filtered for the consistency of differences in expression levels across all time points (coefficient variation Ͻ 50%) and for their absolute intensity above 243 (90% of genes called absent by Affymetrix software, Ͻ243 for this data set). Additional filters reduced the number of RNAs to 54, of which 29 were consistently increased and 25 were decreased in mCFTR(Ϫ) compared with their mCFTR(ϩ) littermates (Tables I and II). The expression profiles of these 54 genes are shown in Fig. 4, demonstrating consistent patterns of expression of the CFTR-responsive RNAs regardless of age. Differentially expressed genes were further classified according to their known or predicted functions. Each gene was annotated and assigned to a functional category. To simplify the calculation, we assumed that genes in each category could be fit to a binomial distribution. The binomial probability was calculated for each category using the entire U74Av2 as reference data set. "Inflammatory Response" was the most represented category of those RNAs increased in mCFTR(Ϫ) mice. Among RNAs whose abundance was increased by the lack of CFTR, those influencing inflammation, transcription, and transport were most highly represented and consisted of a group of functional categories quite distinct from those whose expression was decreased in mCFTR(Ϫ) mice (Tables I-III).
The potential influence of the FABP-hCFTR transgene on RNA expression was also assessed using the Welch t test at the three ages. Differentially regulated RNAs identified in analysis of gastrointestinal-corrected mice were similarly affected in mCFTR(Ϫ) mice, demonstrating a lack of effect of iFABP-hCFTR on this subset of genes. Genes whose expression was independently altered by the iFABP-hCFTR transgene included 7 RNAs decreased and 11 increased. Differences in their levels of expression were modest (less than 1.5-fold) ( Table IV).
Validation of Selected mRNAs-To validate the responsive RNAs identified by microarray analyses, real time RT-PCR was performed. mRNA levels were normalized using ␤-actin or glyceraldehyde-3-phosphate dehydrogenase. Kir 4.2 (Kcnj15), CEBP␦, TNF-AIP-3, and Grin 2d mRNA were significantly increased in CFTR(Ϫ) mice compared with control littermates (Fig. 5). As expected, murine CFTR was not detectable by RT-PCR in mCFTR(Ϫ) mice, nor was hCFTR mRNA detected in lung from the iFABP-hCFTR bearing mice. DISCUSSION The absence of CFTR caused stereotypic changes in gene expression in the lungs of CFTR(Ϫ) mice in the absence of detectable infection or inflammation. Cellular responses to CFTR included enhanced expression of transcription factors and signaling pathways known to influence CFTR gene expression (IL-1␤ and CEBP␦). Likewise, expression of RNAs modulating inflammation, ion transport, protein trafficking, and degradation were altered, indicating that a number of cellular pathways may compensate for or mediate CFTR-dependent functions that, in turn, may maintain normal pulmonary homeostasis in the CFTR(Ϫ) mice or alternatively, influence CF phenotype following response to pathogens. The absence of CFTR initiates reproducible changes in lung gene expression that may modify inflammation, host defense, and other cellular functions, that likely contribute to or mollify the pathogenesis of pulmonary disease in CF.
Because strain is known to influence pulmonary findings in CF mice (8), and the transgenic mice presently studied were generated in a mixed FVB/N, C57 Bl/6 background, lung RNA was compared from sex-matched littermates at various ages. An extensive data set was utilized to identify CFTR-dependent changes in gene expression that were present throughout development and in the absence of inflammation, seeking to identify pathways influenced primarily by CFTR, rather than sex, strain, age, or other secondary phenomena. As expected, expression of mCFTR was not detected in the gastrointestinalcorrected mCFTR(Ϫ) mice, consistent with previous findings (10). Histologic analyses, performed presently and previously, demonstrated no structural abnormalities, infection or inflammation in the lungs of these mice as maintained in our vivarium (10, 11), supporting the concept that changes in gene expression were related to CFTR and not to age or lung disease. Analysis of arrays prepared from pairs of mCFTR(Ϫ) mice and mCFTR(ϩ) littermates (those lacking the iFABP-hCFTR transgene), confirmed the microarray findings, demonstrating both the lack of mCFTR mRNA in lungs of the mCFTR(Ϫ) and that RNA changes were independent of the iFABP-hCFTR transgene.
Expression of Genes Modulating CFTR-Expression of a number of genes known to influence CFTR expression was enhanced in the lungs of CFTR(Ϫ) mice, including CEBP␦ and IL-1␤ suggesting that CFTR cells responded by enhancing levels of RNAs encoding transcription factors or pathways that may compensate for the lack of CFTR. IL-1␤ increased CFTR gene transcription in epithelial cells in vitro (15) and cis-acting elements binding CEBP␦ (to CCAAT enhancer sites) and c-Fos (to AP-1 elements) are present in the promoter-enhancer regions of the mouse CFTR gene (16,17). Previous studies demonstrated that CEBP␦ directly enhanced CFTR gene transcription in vitro (18), thus, the increased CEBP␦ expression may represent a potential compensatory response to CFTR deficiency that, in turn, may influence expression of genes unrelated to CFTR. Expression of c-fos was also increased in the mCFTR(Ϫ) mice. Whereas AP-1 sites have been identified in the hCFTR promoter, the precise role of c-fos in regulation of CFTR has not been clarified; although treatment of various cell types with phorbol esters decreased CFTR gene transcription in vitro (18). Because CEBP␦ and c-fos activate or inhibit the expression of numerous genes that share cis-active elements with the CFTR gene, changes in their activity may broadly influence gene expression, perhaps inadvertently linking CFTR deficiency to the expression of genes whose activities are not directly related to CFTR protein function. RNA encoding TNF-AIP-3, a zinc finger transcription factor, was also increased in the CFTR(Ϫ) mice, further linking transcription responses to the lack of CFTR. Of considerable interest, TNF-AIP-3 RNA was induced by either IL-1␤ or TNF. TNF-AIP-3 inhibited NFB translocation in vitro, and may represent a compensatory response to the increased expression of IL-1␤ seen in the CFTR(Ϫ) mice (19). JAK-3, nuclear receptor subfamily 2, and interferon regulatory factor-1 RNAs were decreased in CFTR(Ϫ) mice. These RNAs encode transcription factors that regulate various pathways involved in inflammation and may, therefore, represent responses to the proinflammatory proteins induced in the CFTR(Ϫ) mice, in essence, being secondarily responsive to initial compensatory changes.
Enhanced Expression of Genes Modulating Inflammation-Genes involved in inflammation were overly represented among the RNAs induced and were distinct from those that were decreased in the lungs from CFTR(Ϫ) mice (Table III). Most prominent among genes whose expression was increased were a family of calcium-binding proteins termed the calgranulins (S100A8, S100A9, and calbindin D9K). S100A8 is expressed primarily by macrophages and monocytes and its expression is enhanced by various cytokines including TNF-␣, IL-1␤, and interferon ␥ (20,21). This family of peptides is expressed by various cell types and share potent chemoattractant activities, stimulating inflammatory cell trafficking. Increased expression of S100A8 (calgranulin A) was previously demonstrated in alveolar macrophages from CFTR mutant mice, the authors suggesting that increased S100A8 may contribute to the enhanced inflammatory responses seen in the absence of CFTR (22). Expression of S100A8 by alveolar macrophages was induced by TNF-␣, interferon ␥ and IL-1␤, mediated at least in part by AP-1-dependent pathways (20). Chitinase A mRNA was also consistently increased in the lungs of CFTR(Ϫ) mice. The acidic chitinase A is a small peptide containing a mammalian lectin domain that binds complex carbohydrates on surfaces of microbial pathogens including fungi. Like the S100 family of proteins, chitinase family members are also expressed by alveolar macrophages in the lung (23). Thus, decreased mCFTR initiated changes in expression of gene expressed primarily in alveolar macrophages. It remains unclear whether these responses are mediated by direct effects of CFTR in alveolar macrophages or by altered cellular signaling initiated by the lack of CFTR in epithelial or other pulmonary cells that secondarily alters macrophage activity. Intense neutrophilic infiltration and an increased IL-8 production are strongly associated with CF lung disease in humans (2). Increased IL-8 and neutrophilic infiltrates were observed in bronchoalveolar lavage fluid from CF patients in the absence of documented pulmonary infection (4). Although it remains possible that antecedent, but resolved, infections may have contributed to the increased inflammation observed in the CF, these observations support the concept that CF is associated with increased susceptibility to pulmonary inflammation. In the present study, IL-1␤, IL-4, and CSF-1 receptor RNAs were increased and each may contribute to the proinflammatory milieu. IL-1␤ enhances CFTR gene transcription, induces inflammation, and is known to stimulate production of the S100-calgranulins, perhaps indicating a network of genes influenced by CFTR through IL-1␤. IL-4 is a potent inflammatory mediator that enhances inflammation and mucous production in airway epithelia. Transgenic animals expressing IL-4 or animal models in which IL-4 is induced developed severe goblet cell hyperplasia, increased mucous production, and inflammatory cell infiltrates (24), findings typically found in patients with cystic fibrosis. CSF-3r RNA encoding a receptor that mediates monocytic cell migration, proliferation, and activity in response to CSF-3 (G-CSF) was increased 3-4-fold in the CFTR(Ϫ) mice (25). Thus, taken together, expression of a number of genes; many influenced by IL-1␤ and mediating inflammation, were induced in the lungs of CFTR(Ϫ) mice.
Despite the increased expression of proinflammatory molecules, there is no evidence of inflammation in the lung of the CFTR(Ϫ) mice perhaps indicating that normal homeostasis is maintained by the complex responses of the lung to the lack of CFTR. The presence of stereotypic changes in expression of many genes suggests that the presence of a single ameliorating gene, for example, an alternative chloride channel, does not fully explain the physiologic adjustment of the lung the CFTR(Ϫ) mice. At present, it is unclear whether these proinflammatory responses are secondary to changes in the expression and function of CFTR in the epithelial cells that, in turn, modulate cell signaling and cytokine production in the lung. Alternatively, the absence of CFTR in the alveolar macrophages may alter expression of genes mediating inflammation in those cells. It is of considerable interest that changes in RNAs modifying inflammation were altered in the lungs of mCFTR(Ϫ) mice in the absence of detectable bacterial infection or inflammation, supporting the concept that the transcriptional adjustment to CFTR deficiency suffices to maintain normal pulmonary homeostasis in the mouse in vivo. Alternatively, the levels of expression of the proinflammatory molecules may not be adequate to cause histologically detectable inflammation. It remains unclear whether these adjustments in gene expression may, in turn, render the CFTR(Ϫ) mice susceptible to inflammation following infection or injury.
Changes in NFB-and TNF-␣-dependent Pathways-RNAs encoding a number of proteins involved in TNF signaling and NFB activation were also induced in the CFTR(Ϫ) mice. The abundance of TNF-AIP-3 mRNA, a zinc finger transcription protein, a protein whose expression is induced by both TNF-␣ and IL-1␤, was increased in the CFTR(Ϫ) mice. TNF-AIP-3 inhibits NFB activity at target genes (19) and may represent a response to the proinflammatory milieu established in the CFTR(Ϫ) lung. PEG-3, a protein that regulates the induction of NFB following TNF stimulation, was also increased, providing for the support for transcriptional relationships between CFTR deficiency and activity of NFB (26). Recent studies of Schroeder et al. (27) support the concept that the CFTR is required for regulation of NFB, serving as a pattern-associated molecular recognition molecule, following pulmonary exposure to Pseudomonas aeruginosa. In that study, NFB activation and its nuclear trafficking were deficient in CF cells. Taken together with the observation that IL-1␤-mediated transcription of CFTR gene transcription is dependent upon NFB, this important pathway mediating inflammation appears to be influenced by CFTR.
In the present microarray analysis, CFTR deficiency was associated with increased claudin 8 and decreased gap junction connexin 37. Decreased expression of gap junction proteins (Cx43, 40, -37, and -32) and decreased gap junction communication were observed in various in vitro cell systems after exposure to proinflammatory cytokines (28). The observed changes in RNAs mediating cell adhesion and increased ex- FIG. 4. Expression profile chart (A) and hierarchical clustering (B) of 54 selected RNAs that were consistently altered in response to the lack of CFTR regardless of age. Hierarchical clustering was performed by UPGMA. Data were normalized using Trimmed Mean and Z-score calculations. In the profile chart, data were normalized by pairwise control. The y axis is normalized intensity (log scale) and the x axis represents experimental ages. Red lines represent the profiles of lung RNAs increased in CFTR(Ϫ) mice. Green lines represent the profiles of down-regulated RNAs.
pression of IL-1␤ seen in the CFTR(Ϫ) mice are consistent with findings that in the absence of CFTR, IL-1␤ and TNF-␣ failed to inhibit cell communication via gap junctions. Previous studies demonstrated that CFTR is required for the uncoupling of gap functions between epithelial cells during inflammation, a process that may restrict the spread of pathogens or signaling among adjacent cells. It will be of interest to test whether cell adhesion mediated by the claudins is also linked to the pathogenesis of CF.
Changes in Protein Degradation Pathways-Several genes involved in protein degradation were altered in the absence of CFTR compared with normal. Proteosome 26 S subunit (PSMC3) is the major proteolytic component of the ubiquitindependent proteosome. Proteosome 26 S regulates degradation of proteins influencing cell cycle, oncogenesis, transcription, and immunity, including CFTR itself (29,30). The proteosome is composed of two subcomplexes, the 20 S proteosome and PA700. PSMC3 expression was modestly, but significantly increased (1.5-fold) in the absence of CFTR. In contrast, PA28␥ (PSME3), an activator of the 20 S proteosome, was decreased 1.6-fold. These observations are consistent with previous findings that proinflammatory cytokines, including TNF-␣ or interferon-␥ increased expression of the 26 S proteosome and its activators PA28 ␣ and ␤ (31), whereas expression of PA28 ␥ was decreased by interferon-␥ (32). RNA encoding adaptor protein complex AP-2, ␣1 subunit (Ap2a1), a protein involved in the formation of intracellular transport vesicles was also decreased in the absence of CFTR. CFTR co-precipitates with ␣-adaptin (33). Recent studies demonstrated that a C-terminal domain of CFTR binds to the AP-2 adaptor complex to form clathrin-coated vesicles that mediates CFTR internalization (34). ADP-ribosylation factor 5 belongs to a family of GTPbinding proteins that play important roles in the control of membrane trafficking, including formation of secretory vesicles at the trans-Golgi network, endosomal and vesicle-plasma membrane fusion (35). Recent studies support the concept that CFTR regulates endosomal fusion and vesicular trafficking (36,37) indicating potential relationships between CFTR and the actions of ADP-ribosylation factor 5. Another gene in this functional category is represented by kinesin3␣, an mRNA that was decreased in the lungs of CFTR(Ϫ) mice.
Transport Proteins Influenced by CFTR-RNAs encoding several transmembrane transport proteins and receptors were also altered in the lungs of CFTR(Ϫ) mice, including solute carrier 38 (member 4), the potassium inwardly rectifying channel (Kir 4.2 or Kcnj 15), the glutamate receptor (Grin 2d), the naturietic peptide receptor 3 (Npr-3), and the ␤3-adrenergic receptor (ADRB-3). Thus, expression of a number of membrane transport proteins was influenced by CFTR, perhaps representing compensatory responses to defects in CFTR-mediated transport activity. Kir 4.2 and Grin 2d RNAs were increased 2-3-fold in CFTR(Ϫ) mice. Kir 4.2 is expressed in respiratory epithelial cells at sites similar to that of CFTR (38). Kir 4.2 regulates cation transport upon which chloride transport via CFTR or other chloride channels may be influenced. Surprisingly, there is evidence for interactions between CFTR and Kir family members because they are both known to bind via PDZ binding domains through interactions with channel interacting PDZ domain protein (39). Likewise, there is precedence for PDZ-dependent interactions among glutamate receptors, CFTR, and Kir family members (40). Thus, the lack of CFTR enhanced the expression of a number of membrane proteins that may interact with CFTR via PDZ domains, perhaps indicating that CFTR-protein complexes may initiate changes in gene expression. These findings support the concept that CFTR interacts with numerous membrane transport proteins and do not support a model in which activity of an alternative Cl Ϫ transporter alone suffices to compensate for the lack of CFTR in pulmonary cells in the CFTR(Ϫ) mice.
Regulation of Cell Receptors by CFTR-Natriuretic peptide receptor C (Npr-3 or NprC) was increased more than 2-fold in CFTR(Ϫ) mice. Natriuretic peptides comprise a family of 3 structurally related molecules: atrial (ANP), brain (BNP), and C-type (CNP) whose functions are cGMP-dependent (41,42). Among them, CNP increased ciliary beat, mucociliary clearance in airway epithelial cells, and activated CFTR-dependent chloride transport (42). Natriuretic peptides regulate cytokinestimulated NO production via the binding of Npr-3 (43). Because deficient NO production was observed in respiratory epithelial cells of the iFABP-hCFTR, CFTR(Ϫ) mice, the increased expression of Npr-3 in CFTR(Ϫ) mice may represent a compensatory response influencing airway clearance and nitric oxide production via cGMP (44). In contrast to the transport/ receptors that were induced, ADRB3, a G protein-coupled transmembrane protein mediating cAMP production was decreased in CFTR(Ϫ) mice. ADRB3 is co-expressed with CFTR in airway epithelium (45) and may be functionally coupled to CFTR via cAMP-independent pathways (46). ␤2-Adrenergic receptors directly interact with CFTR via the Na(ϩ)/H(ϩ) exchanger regulatory factor to form a signal transduction complex (47). Co-regulation of ADRB3 and CFTR may indicate that these proteins interact closely at both structural and functional levels, a finding that may be linked to the important role of ␤-adrenergic stimulation and cAMP in the activation of chloride transport mediated by CFTR.
Altered Expression of RNAs Encoding CFTR Interacting Proteins-Surprisingly, analysis of the RNA influenced by CFTR identified a number of proteins that directly or indirectly interact with CFTR via protein-protein interactions. This list included proteins involved in protein trafficking and degradation (proteosome 26 S and PA28 subunits and ␣-adaptin), ion transport (Kcnj15 and Grin 2d), and receptors (Npr-3 and ADRB3). Interaction of CFTR with many proteins occurs via PDZ binding domains that mediate protein-protein complex formation. The finding that expression of CFTR interacting proteins is altered in the lungs of CFTR(Ϫ) mice, suggests that CFTR influences networks of signaling and transport activities in the cell and that cells respond to CFTR deficiency via transcriptional responses to CFTR-protein complexes, rather than to CFTR per se.
Are Changes in Gene Expression Bystander Effects?-Whereas genomic responses may compensate for the lack of CFTR mRNA and represent compensation for CFTR function, some responses may be secondary and mediated by pathways not directly related to the action of CFTR per se. Alterations in Maintenance of pulmonary homeostasis in the mCFTR(Ϫ) mouse was associated with complex adaptive responses in gene expression. CFTR influenced RNAs encoding transcription factors, ion channels, membrane receptors, cytokines, and intracellular trafficking proteins. Finally, CFTR altered the expression of a number of proteins that interact with CFTR via protein-protein interactions perhaps representing transcriptional responses to functions mediated by CFTR(Ϫ) protein FIG. 6. Molecular pathways and networks influenced by CFTR. A model is proposed by which the lack of CFTR initiates expression of genes that are known to enhance CFTR expression IL-1␤ and CEBP␦, secondarily modulating inflammation (1,2). Increased expression of cytokines, IL-4 and IL-1␤, and the cytokine receptor, CSF-3R, may influence inflammation (2) as well as expression of genes modulating protein trafficking, degradation, and cell-cell communication (3)(4)(5). Abundance of RNA of chemoattractant peptides of the calgranulin family were enhanced, perhaps contributing to a proinflammatory environment within the lung (2). CFTR altered the expression of RNAs encoding transport proteins and membrane receptors that likely interact with CFTR via PDZ domains (6,7) supporting the concept that pulmonary cells respond to the lack of CFTR or CFTR complexes by regulating the expression of CFTR partners. Alterations in pathways regulating second messengers, including cAMP, via the ␤3-adrenergic receptor (Adrb3), were observed (7). Absence of CFTR also influenced expression of genes mediating endocytosis, membrane recycling, and regulated secretion (4,8,9).  complexes (Fig. 6). The diversity of genes whose expression was altered by CFTR support the concept that, in addition to regulation of Cl Ϫ transport, CFTR plays diverse roles in multiple cellular functions. The present findings support the hypothesis that pulmonary homeostasis in the CFTR(Ϫ) mouse is maintained by complex genomic responses to the lack of CFTR rather than by the action of a single alternative Cl Ϫ channel. Finally, the genes and pathways identified in this study provide new links between CFTR and cellular processes that may influence the pathogenesis of CF lung disease.