(Received for publication, November 6, 1996, and in revised form, January 23, 1997)
From the European Molecular Biology Laboratory, Meyerhofstrasse 1, D-69117 Heidelberg, Germany and the Department of
Internal Medicine IV, University of Heidelberg,
69115 Heidelberg, Germany
We have studied the responses of iron regulatory
protein-1 (IRP-1) to extra- and intracellular sources of reactive
oxygen intermediates (ROIs). IRP-1 is a cytoplasmic
RNA-binding protein that regulates iron metabolism
following its activation by iron deficiency, nitric oxide, and
administration of H2O2 or antimycin A, an
inhibitor of the respiratory chain (Hentze, M. W., and Kühn, L. C. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 8175-8182). We show that 10 µM
H2O2 suffice for complete IRP-1 activation
within 60 min when H2O2 is generated
extracellularly at steady-state. By contrast, rapid cellular
H2O2 degradation necessitates a 5-10-fold higher bolus dose. To study IRP-1 responses to intracellular oxidative stress, mitochondrial respiration was inhibited with antimycin A (to
generate oxidative stress by leakage of ROIs from complex III), or
catalase was blocked with 3-amino-1,2,4-triazole (to diminish
H2O2 degradation); in parallel,
2,7
-dichlorodihydrofluorescein diacetate was used as a
redox-sensitive probe to monitor intracellular H2O2 levels by fluorescence-activated cell
sorting. Catalase inhibition elevates intracellular
H2O2, but surprisingly does not cause
concomitant IRP-1 activation. Following antimycin A treatment, IRP-1 is
activated, but the activation kinetics lag behind the rapid increase in
detectable intracellular H2O2. IRP-1 is thus
activated both by extra- and intracellular generation of ROIs. While
extracellular H2O2 rapidly activates IRP-1 even
without detectable increases in intracellular H2O2, intracellular
H2O2 elevation is not sufficient for IRP-1 activation. IRP-1 thus represents a novel example of an
H2O2-regulated protein that responds
differentially to alterations of extra- and intracellular
H2O2 levels. Our data also suggest that a
direct attack on the 4Fe-4S cluster of IRP-1 by
H2O2 (or an
H2O2-derived reactive species) represents an
unlikely explanation for IRP-1 activation by oxidative stress.
Reactive oxygen intermediates (ROIs)1
such as superoxide anion (O2) and hydrogen peroxide
(H2O2) are often considered as intracellular biohazards when present in excess, a condition referred to as "oxidative stress" (reviewed in Refs. 1 and 2). ROIs are generated
within cells as byproducts of biological oxidations, including electron
transfer reactions in the respiratory chain. In addition, ROIs are
released from specialized cells to affect target cells from the
outside, for instance during the respiratory burst of phagocytes (3).
Since oxidative stress is cytotoxic, prokaryotic and eukaryotic cells
respond to ROIs by activating protective mechanisms (1, 4, 5).
Apart from their toxic nature, ROIs have recently received much
attention as biological signaling molecules and mediators of gene
regulatory circuits in eukaryotic cells (2, 6). In several signaling
pathways involving cell surface receptors and growth factors, signals
are transmitted via ROIs; for example, ligation of the CD28 surface
receptor in primary T cells results in the production of ROIs, leading
to activation of the transcription factor NF-B and the expression of
interleukin-2 (7). The cytotoxic response to tumor necrosis factor
(8), signaling by transforming growth factor
1 (9), or the
activation of epidermal growth factor receptor by UV irradiation (10)
all appear to involve ROIs as signaling molecules.
Several proteins which respond to ROIs have been identified and characterized. In Escherichia coli, many genes are transcriptionally activated when the bacteria are challenged with H2O2 (reviewed in Ref. 4). These responses are mediated by a central regulatory protein, OxyR, which stimulates transcription of the target genes when activated under prooxidant conditions (11). A cysteine residue, Cys-199, appears to be the critical sensor of oxidative stress and OxyR induction involves Cys-199 oxidation, possibly to sulfonic acid (12). In Bacillus subtilis, the defense against oxidative stress by H2O2 is mediated by mrgA, a DNA-binding protein which is transcriptionally induced by H2O2 (13). Bacterial defense mechanisms against oxidative stress display specificity toward different forms of ROIs and other oxidants (4). The defense against superoxide radical and nitric oxide (NO) is orchestrated in E. coli by SoxR, a transcriptional activator of the SoxS gene, which in turn stimulates the expression of several antioxidant genes (reviewed in Ref. 4). This response to oxidative stress involves an iron-sulfur cluster: while apo-SoxR binds to its target DNA with equal affinity as iron-loaded SoxR, only the SoxR homodimer (which contains a 2Fe-2S cluster per monomer) stimulates transcription (14). An iron-sulfur cluster is also involved in the oxygen regulation of the FNR protein (15). FNR is a transcription factor in E. coli which controls the expression of genes required in anaerobiosis. In oxygen-deprived bacteria, FNR assembles a 4Fe-4S cluster which is required for binding to the target promoters (16). This 4Fe-4S cluster is labile to oxygen and is disassembled under aerobic conditions (17).
Most of the eukaryotic ROI-activated proteins are not exclusively
regulated by oxidative stress, but can respond to other physiological
signals, such as other forms of cellular stress, growth factors, or
cytokines. A common denominator in some of these signaling pathways
appears to be production of ROIs; therefore it is not surprising that
some of these responses can be mimicked by treatment of culture cells
with extracellular H2O2. For instance, the
epidermal growth factor receptor is not only activated by its ligand,
but also by UV irradiation and H2O2 (10). The
stress-induced MAPKAP kinase 2 is activated by cytokines, heat shock
(18, 19), and H2O2 (20). In lymphocytes,
H2O2 activates p56lck kinase mimicking
a yet unidentified protein tyrosine kinase (21). The early response
genes c-fos, c-jun, and egr-1 are
transcriptionally activated under many stress conditions, including
H2O2 (22). The transcription factor NF-B is
activated by a variety of proinflammatory stimuli (reviewed in Ref. 23)
and by H2O2 (24, 25). In most, if not all of
these examples, the transcription factors are not directly altered by
the ROIs, but the ROIs induce a signal transduction mechanism that
leads to changes in transcription factor activities. A novel
stress-response pathway in mammals, which appears to be specific for
oxidative stress and H2O2, involves a homologue
of the Saccharomyces cerevisiae Ste-20 kinase (26).
A regulatory link between oxidative stress and mammalian iron metabolism which is mediated by iron regulatory protein-1 (IRP-1) has recently been identified (27, 28). IRP-1 is a post-transcriptional cytoplasmic regulator of mRNAs that contain iron-responsive elements (IREs) (reviewed in Refs. 29 and 30). It is a bifunctional protein with two mutually exclusive activities: as a holo-protein containing a cubane 4Fe-4S cluster, it is a cytoplasmic aconitase (31), whereas it binds with high affinity to IREs as a cluster-less apoprotein (32). Activation of IRE-binding by IRP-1 is thus associated with a post-translational switch from 4Fe-4S- to apo-IRP-1 (reviewed in Refs. 15 and 33). Cellular iron deficiency and nitric oxide (NO) lead to a slow (8-12 h) activation of IRP-1, while extracellular H2O2 triggers rapid (30-60 min) IRP-1 induction (27, 28, 34). Moreover, when the release of ROIs from the respiratory chain is stimulated pharmacologically with antimycin A, IRP-1 is also activated (34), identifying IRP-1 as a genetic regulatory protein that responds to extra- and intracellular oxidative stress. The mechanism(s) by which oxidative stress induces IRP-1 is still unclear, and different models have been entertained. A direct oxidant attack on the cluster has been considered (35, 36), as has the possibility of a ROI-initiated signaling mechanism that ultimately leads to cluster removal (36). In this report, we specifically explore the effects of extra- and intracellular ROIs on IRP-1, and discuss their implications for the mechanism(s) underlying IRP-1 activation.
Glucose, glucose oxidase (344 units/mg), catalase (38000 units/mg), antimycin A, 3-amino-1,2,4-triazole (ATZ), N-acetyl-L-cysteine, L-buthionine-(S,R)-sulfoximine, luminol, hypochlorite, and H2O2 were from Sigma. A glutathione assay kit for the colorimetric determination of total glutathione levels was purchased from Calbiochem. Murine B6 fibroblasts were grown in Dulbecco's modified Eagle's medium supplemented with 2 mM glutamine, 100 units/ml penicillin, 0.1 ng/ml streptomycin, and 10% fetal calf serum. Treatments with H2O2 and glucose, glucose oxidase, and catalase were performed in serum-free minimal essential medium.
Electrophoretic Mobility Shift Assay (EMSA)EMSAs were performed as described earlier (37) using a radiolabeled human ferritin H-chain IRE probe (38). RNA-protein complex formation was estimated by densitometric scanning of the depicted autoradiographs.
Generation of Steady-state Levels of Extracellular H2O2H2O2
generation by the enzymatic oxidation of glucose with glucose oxidase
is described by dH2O2/dt = kGO (kGO, the rate
constant of glucose oxidase) under conditions of substrate (glucose and dioxygen) saturation. Accumulation of H2O2 can
be controlled by adding appropriate amounts of catalase. Since catalase
is not saturable up to molar H2O2
concentrations, the determination of Km is
impossible. The rate of H2O2 degradation by
catalase linearly depends on H2O2
concentration; thus, dH2O2/dt = kcat*[H2O2] (kcat, the rate constant of catalase).
Therefore, the catalase-mediated H2O2
decomposition follows first-order kinetics (39). By mixing glucose,
glucose oxidase, and catalase, H2O2 generation
reaches steady-state levels when kGO = kcat*[H2O2].
The concentration of H2O2 is then determined by
the ratio kGO/kcat. Based
on these considerations, the amount of glucose oxidase and catalase
required to generate steady-state levels of
H2O2 can be calculated, provided that
kGO and kcat are known.
In routine experiments, the rate constants of glucose oxidase, and
catalase preparations were determined as described in Refs. 40 and 43,
and the appropriate amounts of the enzymes were mixed with 5 mM glucose to yield H2O2
concentrations in the micromolar range. Direct chemiluminescence
measurements of H2O2 at different time points
(for up to 3 h) confirmed the maintenance of steady-state levels
of H2O2 in the calculated concentrations.
A
recently described, sensitive non-enzymatic chemiluminescence assay was
used for the determination of H2O2 (41). In
brief, 500 µl of culture medium were mixed with luminol, and NaOCl
was added by an injection device in the luminometer. The integral of
the luminescence peak was determined over 2 s and
H2O2 concentration was calculated from a
calibration curve. A flow technique was used to adjust the
H2O2 concentration in the glucose/glucose
oxidase/catalase system. Briefly, a solution of glucose, glucose
oxidase, and catalase aspirated by a peristaltic pump (4 ml/min) was
mixed with luminol (104 mol/liter) and hypochlorite
(10
4 mol/liter), continuously added by a perfusion pump
(6 ml/min). This procedure allows monitoring of the actual
H2O2 concentration in real time by measuring
the luminescence emitted. All luminescence measurements were performed
using a luminometer AutoLumat LB 953 (Fa. Berthold, Wildbad,
Germany).
The
method is based on the oxidation of 2,7
-dichlorodihydrofluorescein
diacetate (H2DCF-DA) (Molecular Probes). Oxidation of
intracellularly trapped H2DCF-DA requires removal of the
diacetate group by esterases. Activated H2DCF is converted
by H2O2 and peroxidases to the fluorescent
derivative 2
,7
-dichlorofluorescin (DCF) (42). A stock solution of
H2DCF-DA (10 mM in Me2SO) was
always freshly prepared. Two types of experiments were performed: (i)
intracellular detection of H2O2 generated
intracellularly, either by treatment of cells with antimycin A or with
ATZ. Cells grown in normal medium were treated with 100 µM antimycin A (dissolved in EtOH, freshly prepared) or
50 mM ATZ, and H2DCF-DA was added at 5 µM during the last 30 min of the treatment. (ii)
Intracellular detection of H2O2 supplied
extracellularly, either by direct treatment of cells with a bolus of
H2O2, or by treatment with an
H2O2-generating system (glucose/glucose
oxidase). In this series of experiments, cells were pretreated with
H2DCF-DA (5 µM) for 30 min in normal medium.
Subsequently, the dye was washed away and cells were incubated for 30 min with H2O2, glucose oxidase, or a mixture of
glucose oxidase and catalase. Treatments were performed in serum-free minimal essential medium supplemented with penicillin-streptomycin, glutamate, and 5 mM glucose. Samples were prepared for FACS
analysis as follows: cells were washed twice with ice-cold
phosphate-buffered saline, trypsinized (1 ml of trypsin) for 2-3 min
at 37 °C, and finally suspended in 5 ml of phosphate-buffered
saline.
Data were acquired using a FACScan flow cytometer (Becton Dickinson), supported by a Macintosh computer system and CellQuest (Becton Dickinson) software. The software offers acquisition and analysis tools for plotting, gating, statistical analysis, and reporting. It also allows instrument control. The FACScan is equipped with an air-cooled argon-ion laser fixed at an excitation wavelength of 488 nm. The emitted fluorescence was collected at 530 nm using a narrow band-pass filter. Dead cells and debris were gated out on the Forward and 90° scatter parameters.
Treatment of cultured
cells with 50-100 µM H2O2
results in a rapid activation of IRP-1 within 30-60 min (27, 28). Even when the H2O2 treatments are performed in
iron-free (to avoid the Fenton reaction) and serum-free (to avoid
degradation by serum catalases) media, H2O2
added as a single bolus is rapidly degraded as a function of the
absolute amount of H2O2 and the number of cells
in culture (43, 44), indicating H2O2
degradation by the cells. To maintain tissue culture cells for up to
3 h in the presence of steady-state concentrations of 5-100
µM H2O2, a calibrated enzymatic
H2O2-generating system was used (only the lower
concentrations were tolerated by the cells for 3 h). With
appropriate amounts of glucose, glucose oxidase, and catalase,
generation and degradation of H2O2 reach
equilibrium (see "Experimental Procedures") to maintain steady-state H2O2 levels (43). When 100 µM H2O2 were administered either
as a bolus (Fig. 1A) or maintained at steady
state (Fig. 1B), the kinetics of IRP-1 induction were found
to be similar. This result implies that the threshold
H2O2 concentration for IRP-1 activation must be
exceeded for sufficient time even when H2O2
decays rapidly after bolus administration (Fig. 1A). When 100 µM H2O2 are added as a bolus,
full IRP-1 activation is achieved following a 15-min treatment, media
replacement and 45 min of chase, whereas 10 min treatment and 50 min
chase allow only partial activation (34). Analysis of the
H2O2 decay curve (Fig. 1A) shows
that 10 µM H2O2 are present
during the first 15 min of treatment. The calibrated
H2O2-generating system by glucose, glucose oxidase, and catalase was therefore employed to estimate the threshold steady-state concentration required for IRP-1 induction. Fig. 2 shows an experiment where cells were treated for 5-60
min under conditions where 10 or 5 µM
H2O2 were maintained at steady state, and IRP-1
activity was analyzed by EMSA. To confirm the equal loading of all
lanes, cell extracts were treated with 2% 2-mercaptoethanol, which
activates 4Fe-4S IRP-1 in vitro (45). Treatment with 10 µM H2O2 results in partial
activation (~50%) of IRP-1 after 30 min, and in complete activation
(~5-fold) within 1 h (lanes 6 and 7). On
the contrary, treatment with 5 µM
H2O2 over 1 h (lanes 8-13), or
2 and 3 h (not shown) has no effect on IRP-1. Thus, the
H2O2 concentration which suffices to induce
IRP-1 appears to be ~10 µM, which is 5-10 times less
than previously estimated. As predicted by these results and our recent
findings (34), as little as 10 µM
H2O2 administered to cells at steady state for
15 min followed by a chase in control medium is sufficient to activate
IRP-1 within 1-2 h; under these conditions, IRP-1 activity remains
elevated for up to 4 h (data not shown).
Kinetics of IRP-1 Activation and Intracellular H2O2 Accumulation Following Antimycin A Treatment
Inhibition of respiratory chain complex III by
antimycin A is associated with increased H2O2
leakage (46). We demonstrated previously that antimycin A treatment
activates IRP-1 within 2 h (34), suggesting the possibility of a
causal relationship between increased intracellular
H2O2 levels and IRP-1 activation. To
investigate this possibility, we first analyzed the kinetics of IRP-1
induction by antimycin A. Cells were treated with 100 µM
antimycin A for 0.5, 1, and 2 h, and IRP-1 activity was analyzed by EMSA (Fig. 3A). Complete IRP-1 activation
(~4-5-fold) is observed after treatment for 2 h (lanes
1-4), while even a 90-min incubation does not suffice for IRP-1
activation (not shown). Thus, the effect of antimycin A on IRP-1
requires 60-90 min longer than that of extracellularly administered
H2O2.
To test whether the observed kinetics of antimycin A-mediated IRP-1
activation are explained by a delayed intracellular increase in
H2O2, we employed FACS analysis using the dye
H2DCF-DA, which is thought to be specific for
H2O2 (see "Discussion"). Cells treated with
100 µM antimycin A received 5 µM
H2DCF-DA for the last 30 min of incubation. Increased
intracellular DCF fluorescence is clearly detected as early as 30 min
after antimycin A administration (Fig. 3B). The increase in
intracellular H2O2 concentration then persists
for the duration of the experiment, up to 2 h (Fig.
3B). Consequently, there is a lag phase of at
least 90 min between the emergence of increased DCF fluorescence
(intracellular H2O2) and the activation of
IRP-1 by antimycin A. This contrasts with the activation of IRP-1 by 10 µM extracellular H2O2 within 60 min (Fig. 2), which is not even detectable by increased DCF
fluorescence (see below: Fig. 7).
Millimolar concentrations (30 mM) of the antioxidant
N-acetyl-L-cysteine inhibit IRP-1 activation by
exogenous H2O2 (Fig. 4, compare lanes
5 and 6) (27). By contrast, pretreatment of cells with
30 mM N-acetyl-L-cysteine for 4 h followed by addition of 100 µM antimycin A for another
2 h fails to inhibit IRP-1 induction (Fig. 4, lanes
1-4), although it partially antagonizes the intracellular accumulation of H2O2 over 2 h (Fig.
5). Furthermore, when an excess of purified catalase is
added to the culture medium, IRP-1 activation by antimycin A is not
affected (Fig. 4, lanes 7 and 8), although the
same concentration of catalase completely prevents IRP-1 activation by
the glucose/glucose oxidase system (not shown). These results appear to
exclude the possibility that antimycin A acts by releasing H2O2 into the culture medium for subsequent
activation of IRP-1 by this "extracellular"
H2O2.
We also tested whether extracellular (bolus) H2O2 or antimycin A treatment cause a reduction of cellular glutathione levels by a colorimetric assay to measure total glutathione. No differences were observed compared with untreated control samples, whereas treatment of cells with L-buthionine-(S,R)-sulfoximine (100 µg/ml for 24 h), a glutathione-depleting drug (47), resulted in a 6-fold reduction of total glutathione (data not shown). Likewise, no IRP-1 activation by L-buthionine-(S,R)-sulfoximine was observed, indicating that glutathione depletion is not sufficient for IRP-1 activation (not shown).
Catalase Inhibition Raises Intracellular H2O2 Levels Without Concomitant IRP-1 ActivationCatalase and glutathione peroxidase represent the two
major cellular H2O2-decomposing enzymes, and
pharmacological inhibition of catalase should result in an increase of
intracellular H2O2. Indeed, treatment of cells
for 90 min with 50 mM ATZ, a catalase inhibitor, leads to
70% inhibition of catalase activity (not shown) and to a profound
increase in DCF fluorescence (Fig. 6B).
However, this increase in intracellular H2O2
does not cause a concomitant activation of IRP-1 (Fig. 6A).
As a positive control, cells were treated for 30 min with a bolus of
100 µM H2O2, and IRP-1 activation (~5-fold) is clearly apparent (Fig. 6A).
This result underscores the notion that IRP-1 activation by extracellular H2O2 is not simply mediated by an increase in intracellular H2O2 and, indeed, that increased levels of intracellular H2O2 are not sufficient to activate IRP-1. Does extracellular H2O2-mediated IRP-1 activation involve increases in intracellular H2O2? To investigate this question, we performed the following experiment: B6 fibroblasts were incubated with H2DCF-DA (5 µM), followed by treatment with different sources of exogenous H2O2 for 30 min. Intracellular H2O2 levels were then assessed by FACS (Fig. 7B), and IRP-1 activity was monitored in parallel by EMSA (Fig. 7A). Increasing concentrations of H2O2 generated by glucose/glucose oxidase (without catalase) leads to a substantial increase of intracellular H2O2, and a ~5-fold activation of IRP-1 (Fig. 7, B and A, lanes 1 and 2). Treatment with glucose, glucose oxidase, and catalase calibrated to yield 100 µM steady-state H2O2 is associated with a detectable increase in intracellular H2O2 and a ~5-fold IRP-1 activation (Fig. 7, B, and A, lanes 1 and 3). On the contrary, treatment with 10 µM steady-state H2O2 or bolus addition of 100 µM H2O2 for 30 min result in a ~2.5- or ~5-fold IRP-1 activation, respectively (Fig. 7A, lanes 1, 4, and 5), without a detectable increase in DCF fluorescence (Fig. 7B). Thus, the activation of IRP-1 by extracellular H2O2 requires, at most, increases in intracellular H2O2 which are below the threshold of detection by FACS.
Oxidative stress activates genetic responses in both prokaryotic and eukaryotic cells. In most of these cases, the underlying molecular mechanisms are incompletely understood. In at least three examples, IRP-1 in mammals, and SoxR as well as FNR proteins in bacteria, the oxidative stress-response involves the biochemistry of iron-sulfur clusters. Thus, IRP-1, SoxR, and FNR define a group of proteins that respond to reactive oxygen species by changes in the status of their iron sulfur-clusters. IRP-1 and FNR share an additional common feature: the genetic (nucleic acid-binding) activities of both proteins appear to be regulated by a cluster assembly-disassembly mechanism that is triggered by oxygen-derived reactive species. These similarities have allowed proposal of an attractive model, where the 4Fe-4S clusters of IRP-1 and FNR serve as biosensors that are liable to cluster disassembly by a direct oxidant attack (35, 36).
In this report we have analyzed the responses of IRP-1 to different sources of oxidative stress. First, we have estimated the minimal concentration of extracellular H2O2 which is sufficient for IRP-1 induction. By employing a system that maintains H2O2 at steady state, this threshold is estimated to be ~10 µM (Figs. 1 and 2), which is 5-10 times less than previously found with an H2O2 bolus (27, 28, 34). Second, we find that IRP-1 responds differentially to oxidative stress in the form of extracellular administration of H2O2 compared with the pharmacological stimulation of intracellular H2O2 accumulation.
While treatment of cells with exogenous H2O2
results in the rapid (60 min) activation of IRP-1, several lines of
evidence suggest that elevation of intracellular
H2O2 levels is at least not sufficient for
IRP-1 induction. First, the apparent lag phase of at least 90 min
between the emergence of detectable intracellular H2O2 and IRP-1 activation following antimycin A
treatment. The FACS analysis of DCF fluorescence indicates that
intracellular H2O2 remains elevated between 30 min and 2 h after the addition of antimycin A (Figs. 3B
and 5). Second, treatment of cells with a high concentration of the
antioxidant N-acetyl-L-cysteine does not inhibit
antimycin A-induced IRP-1 activation (Fig. 4), even though it has a
negative effect on intracellular H2O2 levels
(Fig. 5). It should be noted that the same concentration of
N-acetyl-L-cysteine prevents activation of
NF-
B in tumor necrosis factor
-treated mouse fibrosarcoma cells
L929 (48). However, viewed in isolation, the reduction of
H2O2 levels by
N-acetyl-L-cysteine may simply not suffice to
reduce the H2O2 concentration below the
activation threshold. Third, a profound increase in intracellular
H2O2 levels induced by treatment of cells with
the catalase inhibitor ATZ does not activate IRP-1 (Fig. 6).
The degradation kinetics of a 100 µM bolus of H2O2 (Fig. 1) and the finding of an activation threshold of ~10 µM H2O2 (Fig. 2) are in perfect agreement with recent data showing that IRP-1 activation by bolus addition of 100 µM H2O2 can be divided into an early (0-15 min) "induction phase" during which H2O2 has to be present, and a subsequent (15-60 min) "execution phase," which does not require the presence of the effector (34). Our results also suggest that extracellular H2O2 does not activate IRP-1 by increasing the intracellular H2O2 concentration: a bolus of 100 µM H2O2 or steady-state administration of 10 µM H2O2 suffice to activate IRP-1 within 30 min without detectable (at least with the methodology used here) increases in intracellular H2O2 levels (Fig. 7). This contrasts with the much higher but ineffective H2O2 levels after a longer (2 h) administration of ATZ (Fig. 6). These findings are also consistent with the failure of H2O2 to activate IRP-1 in cell extracts (27, 28).
The method applied here to detect intracellular oxidative stress is based on oxidation of H2DCF-DA to fluorescent DCF by cellular peroxidases and H2O2. H2DCF-DA has been reported to display specificity for H2O2 and secondary and tertiary organic peroxides (49), and has been used as a "specific" probe for intracellular H2O2 by numerous investigators and in several cell types. These include macrophages (50), melanocytes (51), osteoblastic cells (9), and primary T lymphocytes (7). The specificity of H2DCF-DA as a tool to monitor intracellular H2O2 has also been addressed in the mouse epidermal cell line JB6 by genetic means: stable transfectants with Cu/Zn superoxide dismutase, which catalyzes conversion of superoxide anion to H2O2, show significantly increased DCF fluorescence, while stable transfectants with catalase, an H2O2-degrading enzyme, display less DCF fluorescence compared with wild type cells (25). Furthermore, in rat vascular smooth muscle cells, which produce increased levels of H2O2 upon stimulation with platelet-derived growth factor as detected by DCF fluorescence, addition of purified catalase to stimulated cells results in a specific, energy-dependent uptake of catalase and the profound reduction of DCF fluorescence (52).
Taken together, the data presented here question one of the plausible models for IRP-1 activation by oxidative stress: a direct chemical attack of H2O2 or H2O2-derived reactive species on the 4Fe-4S cluster of IRP-1. If this were the case, one would expect that intracellular H2O2 elevation would suffice for IRP-1 activation. Our data rather support the notion that extracellular H2O2 constitutes a signal for IRP-1 induction that must be transmitted from the outside to the inside of the cell. The unexpected complexity in the responses to extra- and intracellular oxidative stress raises several questions for future investigations: (i) how does extracellular H2O2 induce IRP-1 activation if not by increasing intracellular H2O2 concentration? (ii) Why does inhibition of catalase with ATZ and the concomitant increase in intracellular H2O2 not lead to IRP-1 activation? (iii) How does the modulation of mitochondrial respiratory chain activity with antimycin A cause IRP-1 activation? Although IRP-1 activation by treatment with H2O2 or antimycin A does not appear to involve marked increases in intracellular H2O2 as a common effector, it is clear from our experiments that IRP-1 responds to mitochondrial-derived oxidative stress. Additional experiments are required to elucidate the underlying mechanism.
Physiological Implications of IRP-1 Responses to Oxidative StressFerrous iron displays strong chemical reactivity toward H2O2 to yield the deleterious hydroxyl radical (Fenton reaction). Fenton chemistry largely contributes to iron toxicity in cells and is implicated in tissue damage and degenerative disorders (53, 54). Therefore, characterization of the mechanisms by which IRP-1, a central regulator of cellular iron metabolism, responds to H2O2 is an important task.
Commonly, little distinction has been made between the effects of
extra- and intracellular H2O2 on cellular
targets, most likely because H2O2 is a
diffusible molecule which readily passes through membranes (2).
However, it is noteworthy that different pathological conditions
confront tissues with elevated H2O2
concentrations derived either extra- or intracellularly. For example,
the oxidative burst of circulating neutrophils and activated
macrophages results in the release of superoxide anions, hypochlorous
acid, NO, and H2O2 (3, 41, 55, 56). Other
non-phagocytosing cells may also release ROIs: stimulation of cultured
human lung fibroblasts with transforming growth factor-1 results in
activation of an NADH oxidase; the generated
H2O2 is released and can be measured extracellularly (57). On the other hand, the large family of mitochondrial disorders is characterized by the increased production of
intracellular H2O2 and other ROIs (58, 59). To
our knowledge, IRP-1 is the first example of a regulatory protein that
actively responds to extra- and intracellular oxidative stress by
different means.
We thank Dr. H. D. Riedel for helpful discussions and Isabel Mohr for technical assistance.