Activation of an H 2 O 2 -generating NADH Oxidase in Human Lung Fibroblasts by Transforming Growth Factor (cid:98) 1*

The cellular source(s) and mechanisms of generation of reactive oxygen species (ROS) in nonphagocytic cells stimulated by cytokines are unclear. In this study, we demonstrate that transforming growth factor (cid:98) 1 (TGF-(cid:98) 1, 1 ng/ml) induces the release of H 2 O 2 from human lung fibroblasts within 8 h following exposure to this cytokine. Elevation in H 2 O 2 release peaked at 16 h ( (cid:59) 22 pmol/min/10 6 cells) and gradually declined to undetectable levels at 48 h after TGF- (cid:98) 1 treatment. NADH consumption by these cells was stimulated by TGF- (cid:98) 1 while that of NADPH remained unchanged. NADH oxidase ac- tivity as measured by diphenyliodonium (DPI)-inhibit-able NADH consumption in TGF- (cid:98) 1-treated cells fol-lowed a time course similar to that of H 2 O 2 release. DPI, an inhibitor of the NADPH oxidase complex of neutrophils and other flavoproteins, also inhibited the TGF- (cid:98) 1-induced H 2 O 2 production. Inhibitors of other enzymatic systems involving flavoproteins that may be responsible for the production of H 2 O 2 in these cells, including xan- thine oxidase, nitric oxide synthase, and both mitochon-drial and microsomal electron transport systems, failed to inhibit TGF- (cid:98) 1-induced NADH oxidation and H 2 O 2 production. The delay ( (cid:62) 4 h) following TGF- (cid:98) 1 exposure along with the inhibition of this process by cyclohexi- mide and actinomycin D suggest the requirement of new protein synthesis for induction of NADH oxidase activ- ity in TGF- (cid:98) 1-stimulated fibroblasts.

The cellular source(s) and mechanisms of generation of reactive oxygen species (ROS) in nonphagocytic cells stimulated by cytokines are unclear. In this study, we demonstrate that transforming growth factor ␤1 (TGF-␤1, 1 ng/ml) induces the release of H 2 O 2 from human lung fibroblasts within 8 h following exposure to this cytokine. Elevation in H 2 O 2 release peaked at 16 h (ϳ22 pmol/min/10 6 cells) and gradually declined to undetectable levels at 48 h after TGF-␤1 treatment. NADH consumption by these cells was stimulated by TGF-␤1 while that of NADPH remained unchanged. NADH oxidase activity as measured by diphenyliodonium (DPI)-inhibitable NADH consumption in TGF-␤1-treated cells followed a time course similar to that of H 2 O 2 release. DPI, an inhibitor of the NADPH oxidase complex of neutrophils and other flavoproteins, also inhibited the TGF-␤1induced H 2 O 2 production. Inhibitors of other enzymatic systems involving flavoproteins that may be responsible for the production of H 2 O 2 in these cells, including xanthine oxidase, nitric oxide synthase, and both mitochondrial and microsomal electron transport systems, failed to inhibit TGF-␤1-induced NADH oxidation and H 2 O 2 production. The delay (>4 h) following TGF-␤1 exposure along with the inhibition of this process by cycloheximide and actinomycin D suggest the requirement of new protein synthesis for induction of NADH oxidase activity in TGF-␤1-stimulated fibroblasts.
Transforming growth factor ␤ (TGF-␤), 1 a potent fibrogenic cytokine, is overexpressed at sites of active tissue repair and fibrosis such as in idiopathic pulmonary fibrosis (1), glomerulonephritis (2), and vascular restenosis (3). Administration of anti-TGF-␤ antibodies has been shown to suppress the in vivo activity of TGF-␤1 and protect against tissue fibrosis in kidney (4), skin (5), and lung (6). Multiple actions of TGF-␤ on target cells appear to be important in tissue repair and remodeling including its ability to regulate its own production as well as that of other cytokines (7). However, signaling mechanism(s) of this and many other of its actions on target cells are poorly understood.
There is growing recognition that redox-active biomolecules play important roles in cellular signaling, including activation of transcription factors (8). H 2 O 2 , in particular, is a powerful activator of NF-B, a transcription factor that is associated with gene activation of a number of cytokines involved in tissue injury and repair processes (9). Other cytokines involved in these processes, namely TNF-␣ and IL-1, have also been reported to stimulate production of ROS in nonphagocytic cells (10). Although the source of ROS generation in these cells was initially thought to represent a plasma membrane-bound NADPH oxidase similar to that of neutrophils, subsequent studies have demonstrated this enzymatic complex to be structurally and genetically unrelated (11).
The ability of TGF-␤1 to stimulate cellular production of H 2 O 2 , at least in some cell lines, has been well established (12,13). However, the source of this TGF-␤1-stimulated ROS generation has not been determined. We have previously reported that TGF-␤1 stimulates the production of H 2 O 2 in endothelial cells (12). In the current study, we extend this observation of a pro-oxidant effect of TGF-␤1 to human lung fibroblasts. The main purpose of this study was to identify the source(s) of the TGF-␤1-induced H 2 O 2 production in lung fibroblasts and to specifically determine if NAD(P)H oxidoreductase activities might be involved.

EXPERIMENTAL PROCEDURES
Reagents-Porcine platelet-derived TGF-␤1 was obtained from R & D Systems, Minneapolis, MN. DPI was from ICN Biochemicals, Cleveland, OH. All other reagents and inhibitors were obtained from Sigma, unless otherwise stated.
Cell Culture-Normal human fetal lung fibroblasts (IMR-90, Institute for Medical Research, Camden, NJ) were grown in medium consisting of RPMI 1640 (Life Technologies, Inc.) supplemented with 5% fetal calf serum (Hyclone Laboratories, Logan, UT), 100 units/ml penicillin, 100 g/ml streptomycin, and 1.25 g/ml amphotericin B. Cells were plated on 35-mm Petri dishes and incubated at 37°C in 5% CO 2 , 95% air. Medium containing 5% serum was changed every 3 days. After reaching confluence, cells were placed into the quiescent state by reducing the serum content of medium to 0.05% for 2 days prior to stimulation with TGF-␤1 in serum-free medium.
Measurement of H 2 O 2 Release-H 2 O 2 release from cultured fibroblasts into the overlying medium was assayed using a modification of the method of Ruch and co-workers (14). This fluorimetric method is based on the conversion of homovanillic acid, a substituted phenol, to its fluorescent dimer in the presence of H 2 O 2 and horseradish peroxidase. At each time point after exposure to TGF-␤1, all cells were first washed with Dulbecco's phosphate-buffered saline, pH 7.4, and then incubated with a reaction mixture containing 100 M homovanillic acid, 5 units/ml horseradish peroxidase, type VI, and 1 mM HEPES in Hanks' balanced salt solution without phenol red, pH 7.4. This solution was then collected following a 1-h incubation, the pH was adjusted to 10.0 with 0.1 M glycine-NaOH buffer, and fluorescence was measured at excitation and emission wavelengths of 321 and 421, respectively. Linearity of the rate of H 2 O 2 release from control and TGF-␤1-treated cells was established by measuring the amount of H 2 O 2 released at regularly timed intervals over a 2-h period. All incubations of experimental samples were made with control samples containing the reaction mixture alone (i.e. without cells) to correct for any spontaneous dimerization of homovanillic acid. The exact H 2 O 2 concentrations of solutions used to plot standard curves were determined spectrophotometrically at 240 nm * This work was supported by National Institutes of Health Grant HL07053 and by a Massachusetts Thoracic Society Research Award. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
using an extinction coefficient of 43.6 M Ϫ1 ⅐cm Ϫ1 .
Measurement of NADH, NADPH Consumption, and Oxidase Activities-Measurements of NADH and NADPH oxidase activities were made using methods similar to those described by Brightman and co-workers (15). Cells were first washed with RPMI medium without phenol red, pH 7.4, and then incubated with NADH (250 M) or NADPH (250 M) in the same medium for varying time intervals. The rate of NADH or NADPH consumption was monitored by the decrease in absorbance at ϭ 340 nm, using a Hewlett-Packard 8452A diode array spectrophotometer. The absorption extinction coefficient used to calculate the amount of NADH or NADPH consumed was 6.22 mM Ϫ1 ⅐cm Ϫ1 . For measurements of specific oxidase activity, only the DPI-inhibitable rate of consumption of NADH and NADPH was used. This was done by adding DPI (10 M) 30 min prior to the assays for NADH and NADPH consumption. As shown in Fig. 3, TGF-␤1-stimulated NADH consumption was inhibited by DPI. This "DPI-inhibitable" NADH consumption was later used to measure NADH oxidase activity at specific time points after exposure to TGF-␤1. All measurements were expressed in nanomoles of substrate/min/10 6 cells.
Cell Counts-Cell counts were performed concurrently with all of the biochemical measurements described. After removal of the extracellular medium for assay, culture dishes were washed with warm physiological saline, incubated with 1.0 ml of trypsin-EDTA for 2-3 min, and rapidly suspended in solution by pipetting. A 0.2-ml aliquot of cell suspension was then diluted for counting in a model ZM Coulter Counter (Coulter Electronics, Hialeah, FL).
Statistical Analysis-Data from various groups were expressed as means Ϯ S.D. Statistical comparisons were made using the Student's t test for unpaired samples. For studies involving more than two groups, two-way analysis of variance was determined using the Scheffe's test (GB-STAT; Dynamic Microsystems, Silver Spring, MD). Statistical significance in all cases was defined at p Ͻ 0.05.

Effect of TGF-␤1 on H 2 O 2 Release-
The rate of extracellular release of H 2 O 2 from fibroblasts treated with a single dose of TGF-␤1 (1 ng/ml) was measured at regularly timed intervals over a 48-h period. There was no measurable release of H 2 O 2 from unstimulated cells. H 2 O 2 release by TGF-␤1-treated cells was detected at 8 h following exposure, with no demonstrable increase at 4 h (Fig. 1). Measurements at even earlier time points, less than the 4-h exposure time shown (i.e. 1 and 2 h), also failed to demonstrate any H 2 O 2 production (results not shown). The peak rate of H 2 O 2 release was seen at 16 h following treatment with TGF-␤1 prior to a gradual decrease to baseline (undetectable levels) by 48 h.
Effect of TGF-␤1 on NADH and NADPH Consumption-The rates of NADH and NADPH consumption by control and TGF-␤1-stimulated cells were determined at 16 h of exposure to TGF-␤1, corresponding to the peak rate of H 2 O 2 release observed (Fig. 1). As shown in Fig. 2, the rate of NADH consumption by TGF-␤1-treated cells was 2-fold higher than that of control cells (approximately 0.37-0.40 nmol of NADH/min/10 6 cells versus 0.19 nmol of NADH/min/10 6 cells, respectively). The baseline rate of NADPH consumption was almost 4-fold lower (ϳ0.05 nmol of NADPH/min/10 6 cells) than that of NADH, and there was no significant change in the rate of NADPH consumption by cells treated with TGF-␤1.
Effect of DPI in TGF-␤1-induced NADH and NADPH Consumption-The effect of the flavoprotein inhibitor, DPI, on the rate of NADH and NADPH was examined. At a dose that has been previously reported to inhibit cellular flavoprotein oxidases (16), DPI (10 M) inhibited the increase in NADH utilization by TGF-␤1-treated cells. In these experiments, DPI was added 30 min prior to the assay for NADH or NADPH consumption, following a 16-h exposure to TGF-␤1. There was DPI-inhibitable NADH consumption noted in unstimulated (control) cells as well, although this was about half that of TGF-␤1-treated cells (Fig. 3). In contrast, there was no inhibition of NADPH consumption by control or TGF-␤1-treated cells by DPI.
We also examined the possibility that NADH oxidation in the extracellular medium was due to NADH-utilizing enzymes released into the medium of TGF-␤1-treated cells. Conditioned media from control and TGF-␤1-treated cells were assayed for NADH oxidase activity (by measuring DPI-inhibitable NADH consumption) and H 2 O 2 . Both NADH oxidase activity and H 2 O 2 were undetectable in the conditioned media of control and TGF-␤1-treated cells. The rates of DPI-noninhibitable NADH consumption were similar in both groups. The requirement for the fibroblast monolayer to be in direct contact with the extracellular medium for detection of NADH oxidase activity and H 2 O 2 release suggests that these reactions are taking place in or around the surface of fibroblast plasma membranes. Measurement of NADH oxidase activity cannot, therefore, be accounted for by release of enzymes into the extracellular medium.
Effect of Inhibitors on TGF-␤1-induced NADH Consumption and H 2 O 2 Release-Although previously reported as a specific inhibitor of NAD(P)H oxidases (16), DPI and other related compounds are now recognized as relatively nonspecific inhib-

TGF-␤1 Induced NADH Oxidase Activity
itors of flavoproteins (17). Since other flavoprotein-requiring enzymatic systems may be potential sources of ROS generation, we evaluated the effect of more specific inhibitors of these enzymatic systems on NADH consumption and H 2 O 2 release. Inhibitors of nitric oxide synthase (L-N G -nitroarginine methyl ester, 1 mM), xanthine oxidase (allopurinol, 100 M), mitochondrial electron transport (KCN, 1 mM), and microsomal cytochrome P-450 (methoxypsoralen, 250 M) had no significant effect on the NADH consumption or H 2 O 2 release induced by TGF-␤1 (Table I). DPI, on the other hand, produced a marked reduction in TGF-␤1-induced H 2 O 2 release, similar to the effect of this inhibitor on NADH consumption.
Cycloheximide (1 g/ml, an inhibitor of protein synthesis) and actinomycin D (0.05 g/ml, an inhibitor of gene transcription) also produced significant reductions in NADH consumption and H 2 O 2 release, suggesting that new protein synthesis is required for the NADH oxidase activity induced by TGF-␤1 in fibroblasts. Of uncertain significance is the finding of a small amount of H 2 O 2 release by TGF-␤1-treated cells in the presence of cycloheximide and actinomycin D despite complete inhibition of the TGF-␤1-induced increase in NADH consumption by these inhibitors.
We examined the possibility that TGF-␤1-induced NADH oxidation may occur as a result of a peroxidase-catalyzed free radical chain reaction in the presence of H 2 O 2 and a transitional metal ion (the "peroxidase-oxidase" reaction) (18). Addition of superoxide dismutase (Cu,Zn-superoxide dismutase, 100 units/ml) to the reaction mixture did not inhibit TGF-␤1induced NADH oxidation or H 2 O 2 release (Table I), making this reaction unlikely. Nonenzymatic oxidation of NADH by H 2 O 2 was also ruled out by adding H 2 O 2 (1-10 M, concentrations that exceeded experimentally measured levels) directly to the NADH reaction mixture and finding no significant change in absorption at ϭ 340 nm.
Effect of TGF-␤1 on NADH Oxidase Activity-The time course of the TGF-␤1-induced NADH oxidase activity follows a pattern almost identical with that observed with H 2 O 2 release (Figs. 1 and 4). Although we have chosen to report DPI-inhibitable NADH consumption as a more specific measure of NADH oxidase activity, a plot of "total" NADH consumption showed a similar pattern (results not shown). No NADH oxidase activity was detected at 4 h following TGF-␤1 treatment, and peak activity was noted at 16 h (Fig. 4). As with H 2 O 2 release, NADH oxidase activity decreased to control levels by 48 h of exposure to TGF-␤1. DISCUSSION Certain cytokines have the ability to stimulate the production of ROS in nonphagocytic cells (10,12,19). The source(s) of cytokine-stimulated ROS generation, however, have not been clearly identified. We have previously reported on the stimulatory effect of TGF-␤1 on bovine pulmonary artery endothelial cell H 2 O 2 production without being able to identify its source (12). In the present study, we extend this observation to human lung fibroblasts which, unlike endothelial cells, demonstrate no release of H 2 O 2 extracellularly in the unstimulated state but, similar to endothelial cells, can be induced to generate H 2 O 2 by TGF-␤1. Our study suggests that the source of this H 2 O 2 is the 2-electron reduction of O 2 via a novel electron transport system most likely localized in fibroblast plasma membranes. This electron transport system involves a flavoprotein which specifically utilizes NADH as the electron donor. Further, the activation of this flavoprotein oxidase by TGF-␤1 appears to require new protein synthesis.
NADH and NADPH are ubiquitous dinucleotides (each containing adenine and a nicotinamide ring) that function as carriers of "reducing equivalents" and are used as coenzymes in many cellular oxidation-reduction reactions. They do this by the transfer of hydride ion (H Ϫ , 2 electrons plus a proton) from the 4-position of the nicotinamide ring. The only structural difference between NADH and NADPH is the presence of an extra phosphate group at the 2Ј-position of the adenine nucleotide moiety of NADPH. Although this extra phosphate group is far from the active redox region (of the nicotinamide ring) and is of no importance in the hydride ion transfer reaction, it appears to confer specificity for the enzymes to which NADPH can bind as a coenzyme.
Based on our findings of NADH-specific flavoprotein oxidase activity in TGF-␤1-treated fibroblasts, we can construct the following simple diagram The potential for additional "carriers" of electron transport from NADH (donor) to O 2   TGF-␤1 Induced NADH Oxidase Activity membranes, consumption rates of these biomolecules measured in the extracellular medium are likely to represent oxidation by enzyme(s) on the outer aspect of the membrane. We considered the possibility of NADH oxidation by non-flavoprotein oxidases and dehydrogenases or degradation by nonoxidative pathways and, therefore, specifically measured DPI-inhibitable consumption of NADH. DPI, previously thought to be a specific inhibitor of NADPH oxidase, is now recognized as a relatively nonspecific inhibitor of flavoproteins by direct binding (17). It is unlikely that DPI, in the manner used in our studies, could diffuse through the membrane to inhibit intracellular flavoprotein oxidoreductases. However, we have assessed the contributions of intracellular sources of ROS involving flavoproteins with the use of specific inhibitors of these enzyme systems, and our results support the likelihood that they are not involved in TGF-␤1-induced H 2 O 2 generation.
TGF-␤1-induced NADH oxidase activity appears to require new protein synthesis based on the finding of a lag time (Ͼ4 h, Ͻ8 h) following exposure to TGF-␤1 and its inhibition by cycloheximide and actinomycin D. It is unlikely that this activity is mediated directly by TGF-␤1 since this peptide has been shown to bind rapidly to its receptor(s) and is internalized and degraded by lysosomal enzymes within 4 h at 37°C (21). Moreover, assays for NADH consumption and H 2 O 2 release are made after first washing off the medium and without the reintroduction of TGF-␤1 in assay solutions. Together, these observations suggest that binding of TGF-␤1 to its receptor(s) results in the activation of a signaling cascade that results in the induction of H 2 O 2 -generating capacity by the cell via an NADH-specific flavoprotein oxidase. It appears likely that some component of this protein complex, required for enzymatic activity, has to be newly synthesized. Our finding of detectable H 2 O 2 release in TGF-␤1-treated cells in the presence of cycloheximide and actinomycin D, despite an apparent complete inhibition of oxidase activity (by the observed lowering of NADH consumption to control levels, Table I) suggests that the relatively smaller amounts of H 2 O 2 measured under these conditions may be unrelated to NADH oxidase activity. H 2 O 2 detected under conditions when protein synthesis is inhibited might reflect a relative imbalance of oxidant stress and cellular antioxidant capacity since reduced levels of newly formed superoxide dismutase, catalase, and components of the glutathione redox cycle would be expected under these conditions. Moreover, TGF-␤1 alone has been reported to suppress the expression of these antioxidant enzymes in rat hepatocytes (22).
Although the source(s) of cytokine-stimulated ROS have remained largely speculative, some recent studies have attempted to identify the ROS-generating enzymatic systems. Meier et al. (10) have reported on a superoxide-generating NADPH oxidase in human skin fibroblasts stimulated by TNF-␣ and IL-1 that was initially thought to be similar to the NADPH of neutrophils (10), but subsequently has been shown to be structurally and genetically distinct (11). There are several important differences between TNF-␣-and IL-1-stimulated ROS production in nonphagocytic cells (10,19) and the TGF-␤1-induced H 2 O 2 production observed in our studies. First, NADH specificity has not been reported previously. Secondly, the formation of ROS in response to TNF-␣ and IL-1 is immediate, and there is no apparent requirement for new protein synthesis. Thirdly, we are unable to demonstrate O 2 . formation in association with TGF-␤1-induced NADH oxidase activity. Finally, we have not observed similar NADH oxidation or H 2 O 2 production in intact lung fibroblasts in response to TNF-␣ and IL-1 using short incubation (immediate activation) or longer exposure (induction of activity possibly through new protein synthesis, results not shown).
The presence of a growth factor-responsive (epidermal growth factor, insulin, pituitary extract) NADH oxidase which may function as part of a redox system at the plasma membrane has been proposed by Brightman and co-workers (15). In these studies, NADH oxidase activity was measured by monitoring NADH utilization and O 2 consumption, not H 2 O 2 production. In fact, ROS production in association with this activity is apparently lacking. 2 Another recent study has demonstrated the stimulation of NADH and NADPH oxidase activity in smooth muscle cells by angiotensin II (23). NADH oxidoreductase activity has also been shown to be a major source of O 2 . in unstimulated bovine coronary artery endothelium (24). The relationship between the TGF-␤1-induced NADH oxidase activity noted in our studies and similar activities reported in these former studies is, at present, unclear.
The best characterized NADPH-or NADH-utilizing flavoprotein oxidase associated with the formation of ROS is that of phagocytic cells (25). Although there are some obvious similarities such as the involvement of a flavoprotein as part of the functional enzymatic complex and the use of O 2 as the electron acceptor, there are also several differences from TGF-␤1-induced NADH oxidase activity. The most striking difference is the specificity of the oxidase for NADH (and not NADPH). The kinetics of the formation of ROS by TGF-␤1-induced NADH oxidase activity is characterized by the sustained release of low concentrations of H 2 O 2 for several hours from fibroblasts in comparison to the rapid, short-lived release of relatively larger amounts of ROS from the "respiratory burst" of neutrophils. New protein synthesis is not required for the activation of the NADPH oxidase in neutrophils by the stimulants studied.
Although the functional significance of TGF-␤1-induced NADH oxidase activity remains speculative at this time, there is increasing recognition of redox-active biomolecules as regulators of cell function (8,9). The kinetics and lower concentrations of ROS generated by TGF-␤1 as well as the lack of evidence for cell toxicity favor the potential for the H 2 O 2 generated to function as a signaling molecule. The lack of under-standing of post-receptor signaling mechanisms have led investigators to believe that unidentified pathways for TGF-␤ signaling exist (26). Further studies of the relationship of TGF-␤1-induced NADH oxidase activity and H 2 O 2 production to specific known effects of this cytokine on target cells may help to uncover redox-sensitive signaling pathways for the multiple actions of this cytokine.
Victor J. Thannickal and Barry L. Fanburg