Characterization of inducible nitric oxide synthase expression in human airway epithelium.

Nitric oxide is an important mediator of inflammatory responses in the lung and a key regulator of pulmonary vascular and bronchomotor tone. We have shown that the inducible nitric oxide synthase (iNOS) isoform is continuously expressed in human airway epithelium at mRNA and protein/activity levels in vivo. However, removal of epithelial cells from the in vivo airway environment resulted in rapid loss of iNOS expression, which suggested that expression is dependent upon conditions and/or factors present in the airway. To investigate the mechanisms responsible for maintenance of expression in vivo, we evaluated regulation of iNOS expression in primary human airway epithelial cells. Interferon-gamma (IFN-gamma) was sufficient for induction of iNOS in primary human airway epithelial cells (HAEC) in vitro, and interleukin-4 (IL-4) potentiated the expression through stabilization of iNOS mRNA. The IFN-gamma/IL-4-induced iNOS expression in HAEC was delayed in onset and prolonged with expression up to 1 week. Furthermore, transfer of overlying culture media [conditioned media (CM)] to other HAEC led to iNOS induction. Interestingly, IFN-gamma/IL-4 induction of iNOS was dependent on new protein synthesis, whereas CM induction of iNOS was not. IFN-gamma and IL-4 activated signal transducers and activators of transcription (STAT1 and STAT6) in HAEC, but CM transfer to HAEC produced even higher levels of STAT1 activation than achieved by direct addition of cytokines. Thus, IFN-gamma/IL-4, which occurs in human lung lining fluid, led to iNOS expression in human airway epithelium through production of soluble mediators and stabilization of mRNA.


Introduction
Nitric oxide ('NO) has many proposed biological functions in the lung, including modifying airway tone, regulating pulmonary vascular tone, stimulating mucin secretion, modulating mucociliary clearance through effects on ciliary beat frequency, and immune surveillance including tumoricidal and bactericidal effects (1). Three types of nitric oxide synthases (NOS; EC 1.14.13.39), the enzymes responsible for endogenous 'NO production, have been described in human cells. Two are constitutive isoforms (neuronal NOS and endothelial NOS) with activities regulated by intracellular calcium/calmodulin producing picomolar levels of NO (2). A third form of NOS (iNOS), which produces nanomolar levels of NO, is induced by microbes, microbial products, or cytokines and is independent of calcium for activity (2,3). Previously, immunostaining of the lung indicated that iNOS was the prominent isoform expressed in human airway epithelium in vivo (4,5).
However, overproduction of NO by iNOS has been proposed as one mechanism of amplifying and/or perpetuating lung inflammation and injury. In this context, endogenous overproduction of NO has been implicated in the pathogenesis of asbestos-related carcinogenesis (6), inflammatory disorders in the lung, such as asthma, bronchiectasis, and chronic obstructive pulmonary diseases (7,8), and in sepsis-related acute lung injury (8). The mechanism of injury by 'NO is likely mediated through its reaction with superoxide to form peroxynitrite, a strong oxidant that can react with biologic molecules resulting in major pathologic consequences (8). For example, 'NO in association with superoxide plays a pathogenic role in the lung injury from influenza viral pneumonia (9). In addition, 'NO and oxidants contribute to the epithelial injury and shedding described in the asthmatic airway (7). Similarly, the human immunodeficiency virus structural envelope glycoprotein gpl20 induces iNOS expression and increased formation of an 'NO radical, which leads to cytotoxic effects (10). The 'NO radical is involved in the hydroxylation of deoxyguanosine in DNA, which leads to 8-hydroxydeoxyguanosine, a potentially carcinogenic DNA lesion (6). Thus the carcinogenicity of asbestos, a naturally occurring mineral fiber that induces iNOS and oxidant formation in lung cells, may be due in part to the formation of *NO radicals (6).
With the knowledge that 'NO plays an important role in both normal lung processes and in the pathogenesis of lung diseases, we have characterized the expression of iNOS in human airway epithelial cells (HAEC) in vivo and in vitro.
iNOS Expression in Human Airway Epithelium in Vivo Initially, we used a reverse transcriptionpolymerase chain reaction (PCR) cloning strategy to clone human epithelial NOS. HAEC were obtained through bronchoscopy with a flexible fiberoptic bronchoscope from normal nonsmoking volunteers with no history of lung disease and on no medications. A contiguous cDNA sequence of 3966 bp compiled from sequences of the five overlapping PCR products demonstrated a 3459 bp open reading frame encoding a peptide of 1153 amino acids (11). Comparison of this respiratory Environmental Health Perspectives * Vol 106, Supplement 5 * October 1998 lllilillillilla epithelial cDNA with human iNOS cDNA sequences from hepatocyte, two types of chondrocytes, and a colorectal adenocarcinoma cell line (12)(13)(14)(15) showed that except for two sequence discrepancies resulting in one amino acid change, our respiratory epithelial NOS was in agreement with the known iNOS sequences (Genbank accession U20141).
To quantitate expression of iNOS in human airway epithelium, we performed Northern blot analyses of total RNA from HAEC freshly obtained by bronchoscopic brushing of normal nonsmoking volunteers. Abundant levels of iNOS gene transcript were demonstrated as a prominent signal at 4.5 kb in all HAEC samples using a 32P-labeled iNOS cDNA probe (pCCF21) ( Figure IA) (11). In contrast, constitutive endothelial type NOS was not detected in HAEC by Northern blot analysis using a full-length human endothelial NOS cDNA probe (16). Northern blot analysis of normal human airway epithelium from bronchi obtained at surgical lung resections also demonstrated abundant continuous expression of iNOS ( Figure IA). In contrast to the high levels of iNOS seen in the airway epithelial cells, Northern blot analysis of total RNA immediately extracted from human alveolar macrophages obtained by bronchoalveolar lavage (BAL) of normal nonsmoking volunteers or peripheral human lung tissues demonstrated no iNOS expression ( Figure IA) (11). Thus, airway epithelium of normal nonsmoking individuals uniquely expressed the iNOS mRNA at high levels under basal unstimulated conditions in vivo.
To determine whether the abundant iNOS mRNA expression was paralleled by similarly abundant iNOS enzyme content in airway epithelial cells, we evaluated iNOS protein expression and function. Western blot analyses using a monoclonal anti-iNOS antibody demonstrated a protein band in airway epithelial cell lysates from normal volunteers that was similar in size to iNOS induced in the mouse macrophage cell line (RAW264.7) with interferon-y (IFN-y) and lipopolysaccharide (LPS) ( Figure 1B) (11). In addition, NOS enzyme activity was quantitated in the HAEC cell lysates by measuring the conversion of 14C-L-arginine to 14C-Lcitrulline in the presence of saturating concentrations of the enzyme's cofactors (11). Positive control of mouse macrophage lysate from RAW264.7 stimulated with IFN-,y and LPS converted L-arginine to L-citrulline at 6 ± 1 nmol/min/mg cell lysate, which is similar to previously reported levels for these cells (2). In the lysate from the freshly obtained unstimulated HAEC, L-arginine was converted to L-citrulline at 1.1 ± 0.3 nmol/min/mg cell lysate (11). These levels of nanomolar conversion of substrate are in agreement with the known activity of iNOS and are an order of magnitude above what would be expected to be produced by the traditional constitutive isoforms of NOS (2). These results confirm that iNOS is abundantly and continuously expressed at the mRNA and protein levels in human airway epithelium in vivo.

Loss of iNOS Expression in Airway Epithelial Cells ex Vivo
Although iNOS expression was present at high levels in vivo, expression was lost ex vivo. Airway epithelial cells obtained by bronchoscopic brushing of normal nonsmoking volunteers were divided into three aliquots for culture in serum-free Lechner and LaVeck media (LHC8, Biofluids, Rockville, MD) for 0, 8, and 24 hr, and RNA was extracted to determine iNOS expression by Northern blot analysis (11). The iNOS expression decreased in culture in a time-dependent fashion in agreement with the previously reported mRNA halflife of 4 hr (17); by 24 hr of culture iNOS mRNA was not detectable by Northern blot analysis ( Figure 2) (11).

iNOS Gene Induction in Vitro
Previous in vitro studies have demonstrated that, similar to other human and murine cell lines, iNOS is induced in human lung epithelial cell lines and primary HAEC by cytokine combinations including IFN-y, interleukin (IL)-10 (IL-1 P), and tumor necrosis factor a (TNF-a) (18,19). Thus, using primary HAEC in culture, we investigated the mechanisms regulating iNOS expression in human respiratory epithelium. As  TNF-a. Although IFN-y alone induced iNOS mRNA expression in HAEC, the levels of iNOS mRNA were less than levels induced by IFN-y and IL-4 ( Figure 3) (20). However, IL-4 alone did not induce iNOS expression, indicating potentiation of IFN-y induction of iNOS by . The time course of iNOS induction in HAEC was delayed and prolonged after IFN-'y and IL-4 stimulation ( Figure 4A) (20). This is in contrast to other primary cultured cells such as hepatocytes, in which iNOS induction occurs in response to cytokines as early as 2 hr, with peak levels at 4 to 6 hr (21). This combination of cytokines may be especially relevant to in vivo iNOS expression, as human lungs contained IFN-y and IL-4 in epithelial lining fluid in vivo at levels that would lead to iNOS induction (IL-4 = 9 ± 5 pg/ml BAL fluid and IFN-y= 33 ± 11 pg/ml BAL fluid; mean ± SE, n = 13) (20).

Soluble Mediators Maintain iNOS Expression in HAEC in Response to a Single Stimulation Event
Our results showed that iNOS mRNA levels in HAEC were maximal at 24 hr and present up to 7 days after a single initiating stimulation with IFN-y and IL-4 ( Figure  4A). Interestingly, persistence of iNOS expression in HAEC in culture was dependent on maintaining the overlying tissue culture media. HAEC were cultured in LHC8 on plates precoated with coating media containing 29 pg/ml collagen (vitrogen from Collagen Corp, Palo Alto, CA), 10 pg/ml bovine serum albumin (Biofluids, Rockville, MD), and 10 pg/ml fibronectin (Calbiochem, La Jolla, CA) (20). After stimulation with IFN-,y and IL-4 and culture for 6 days in this media (20), the cellfree overlying tissue culture media is referred to as the conditioned media (CM).
Transfer of the CM to other HAEC in culture induced iNOS gene expression, whereas transfer of CM to A549 or transformed bronchial epithelial cells (BEAS2B) did not lead to iNOS induction ( Figure  4B) (20). Despite minimal levels of IFN-,y remaining in CM, CM transfer to other HAEC resulted in levels of iNOS induction similar to those found with the addition of fresh IFN-,y and IL-4 ( Figure 4B) (20).  Furthermore, wash-off experiments confirmed that soluble mediators in the CM were responsible for maintenance of iNOS expression. Following 1-or 24-hr incubation of HAEC with IFN-,y and IL-4, the cells were vigorously washed with HEPESbuffered saline and cultured in fresh media (FM) without cytokines for 47 or 24 hr, respectively. Northern blot analysis showed that iNOS expression in the HAEC exposed to IFN-,y and IL-4 for 1 hr was 90 ± 10% of the iNOS mRNA levels in HAEC cultured with IFN-,y and IL-4 continuously for 2 days ( Figure 5) (20). However, the levels of iNOS mRNA in the HAEC were markedly decreased to 15 ± 5% of the positive control if media was removed after 24-hr culture ( Figure 5) (20). These results indicated that IFN-y and IL-4-induced iNOS mRNA expression in HAEC was dependent upon soluble mediator(s). The majority of the mediator(s)' synthesis and/or secretion appeared to occur 1 to 24 hr after IFN-y and IL-4 stimulation.

Protein Synthesis Dependence of iNOS mRNA Induction in HAEC
To determine if the mediators leading to transferable and prolonged iNOS expression in HAEC were dependent on new protein synthesis, we stimulated HAEC with IFN-y and IL-4 in the presen (10 pg/ml); the level were analyzed by North the presence of cycl mRNA levels in stimi decreased to 38 ± 3% c out protein synthesis i (20). In contrast, in mRNA in HAEC by decreased by the pres imide, with iNOS mR] stimulated with CM cycloheximide 150 ± 31 in HAEC stimulatec ( Figure 6) (20).

Electrophoretic P Shift Assays
IFN-y has been identil activating iNOS gene t. ous cells through the transduction pathways interferon-inducible ger ulatory factor-1 (IRF identified as essential f in murine macrophages duction for IFN-y is phosphorylation of S mediates its effects at b STAT6 activation (28) for STATs in the mous promoter have been de case of the mouse iN( tional activation is def elements (29). In gene to cell surface receptc that are transmitted to 1 to the rapid inductic interferon-stimulated genes in the absence iNOS _1 -1 ._:.s..   (30). However, other secondary ce of cycloheximide response genes are induced as a delayed s of iNOS mRNA response to IFN-y and require new protein tern blot analysis. In synthesis, such as major histocompatibility oheximide, iNOS complex class 11 (31,32). The mechanisms alated HAEC were involved in the IFN-y-delayed secondary )f the control withresponse gene expression are less well nhibitor ( Figure 6) understood (31,32). Induction of iNOS in duction of iNOS HAEC by IFN-y displays characteristics of the CM was not a secondary response gene, such as delayed sence of cyclohextime to expression, enhancement of induc-NA levels in HLAEC tion by additional cytokines (TNF-ac, ILin the presence of 1 P, , and requirement of new protein 0% of iNOS levels synthesis. In contrast, the induction of d with CM alone iNOS in HAEC by CM did not require new protein synthesis, indicating that the Mobility process of iNOS induction by CM is not a secondary effect and is different than the IL-4/IFN-y-stimulated iNOS induction. fied as a key factor Possible mechanisms considered for the ranscription in vari-persistent iNOS mRNA expression in Janus OAK) signal HAEC included continued signaling by (26). Further, the cytokines from cell surface receptors or ne, human IFN reginhibition of intracellular signal decay, -1), has also been which would lead to sustained activation of or iNOS activation STATI (26,32). Either of these processes s (27). Signal trans-could lead to continuing iNOS gene through tyrosine expression. To investigate the involvement ,TAT1 (26). IL-4 of the JAK-STAT pathway in our system, east in part through electrophoretic mobility shift assays ). Binding elements (EMSA) with whole cell extract from ;e and human iNOS HAEC stimulated with IFN-,y (100 U/ml) escribed, and in the and IL-4 (10 ng/ml) for 15 min, 8 hr, and DS gene, transcrip-6 days were performed using the IFN-y Pendent upon these activation site oligonucleotide (5'-tcga eral, IFN-y binding GCCTGATTTCCCCGAAATGACGGCrs initiates signals 3') corresponding to the IRF-1 promoter the nucleus and lead from base pairs -130 to -106 sequence relrn of a number of ative to transcription start point (IRF-1 primary response GAS) (26). The results indicated a single of de novo protein major binding complex was induced using the IRF-1 GAS (Figure 7). This binding complex was confirmed to contain STAT1 by supershifting with antihuman STATI antibodies ( Figure 7). In contrast, no complexes were detected upon EMSA, but as evidenced by supershift using STAT6 antibodies, only a minority of the protein binding to the IL-4 GAS element was STAT6 (20). Rather, STAT1 was the major protein binding to the IL-4 GAS element, as might be expected as the IL-4 GAS element contains the DNA consensus element for STATI binding (29).
To determine if CM also activated STAT DNA binding, we performed the EMSA on the whole cell extract of HAEC treated with conditioned media. This resulted in high levels of STATI activation ( Figure 7). Interestingly, this complex was more prominent than the complex detected in HAEC stimulated directly with IFN-,y and IL-4. Moreover, CM resulted in a prominent binding complex of three bands using IL-4 GAS as the probe. Supershift analyses demonstrated that the predominant protein binding to the IL-4 GAS was STATI, with only a very low level of STAT6 detectable in the binding complex. The presence of three distinct bands using the IL-4 GAS probe was best appreciated following supershift of the STATI ( Figure  7, lane 11). The activation of three distinct protein-DNA complexes by CM suggests homoand heterodimeric complexing of STAT proteins (33)(34)(35). Supershift analyses indicated that the fastest moving complex contained STAT 1 homodimers and the slowest moving complex homodimers of STAT6, with the intermediate complex representing a heterodimeric complex of STATs. In addition, IL-4 alone induced only the faint upper band complex (20). These results supported the finding that continued signaling through cell surface receptors was responsible for ongoing iNOS gene expression over time in HAEC.

Summary
These results demonstrate the complex nature of the regulation of iNOS gene expression in airway epithelial cells. Our findings indicate that IFN-y is sufficient to induce iNOS in human airway epithelium, with potentiation of iNOS expression by IL-4 through prolongation of the half-life of the mRNA. Importantly, this combination of cytokines is biologically relevant to expression of iNOS in airway epithelium in vivo, as IL-4 and IFN-y were detected in fluid lining the lung epithelium sampled from healthy volunteers. On the basis of our findings, we propose a mechanism for iNOS induction and maintenance in human airway epithelium in which IFN-y and IL-4 in the epithelial lining fluid lead to epithelial production of soluble mediator(s) that stimulate the airway epithelium in an autocrine and paracrine fashion to express iNOS.