Cobalt and desferrioxamine, like hypoxia, stimulate the production of erythropoietin in HepG2 cells. It is believed that cobalt as well as desferrioxamine interact with the central iron atom of heme proteins by changing their redox state similar to hypoxia. A subsequent decrease of the intracellular H2O2 levels under hypoxia was presumed to be the key event for stimulating erythropoietin production. We therefore investigated whether cobalt and desferrioxamine control the intracellular H2O2 levels that regulate gene expression by interacting with hemeproteins. Deconvolution of light absorption spectra revealed respiratory heme proteins such as cytochrome c, b558 and cytochrome aa3, as well as cytochrome b558, which is a nonrespiratory heme protein found in HepG2 cells. Whereas respiratory heme proteins are located in mitochondria, cytochrome b558 similar to the one described for the neutrophil NADPH oxidase can be visualized in the cell membrane of HepG2 cells by immunohistochemistry. Incubation with cobalt (100 microM/24 hr) interacts predominantly with cytochrome b558 and cytochrome b558. The interaction of cobalt with the respiratory chain results in an increased oxygen consumption of HepG2 cells as revealed by PO2 microelectrode measurements. Desferrioxamine (130 microM/24 hr), however has no influence on the cytochromes. In response to an external application of NADH (1 mM), the membrane bound cytochrome b558 produces two times more O2- than to the external NADPH (1 mM) application. Neither desferrioxamine not cobalt has any influence on the NADH stimulated O2- generation. Incubation with cobalt or with desferrioxamine, however, leads to a decrease of the intracellular H2O2 level as revealed by the dihydrorhodamine 123 technique, perhaps causing the well-known enhanced erythropoietin production. The cobalt-induced H2O2 decrease seems to be caused by an increased activity of the glutathion peroxidase that is also induced under hypoxia. Desferrioxamine, however, leads to an apparent H2O2 decrease only because it seems to inhibit the iron catalyzed reaction of H2O2 with dihydrorhodamine 123, hinting at the occurrence of the Fenton reaction in HepG2 cells. Therefore, it must be determined whether or not degradation products of H2O2 by the Fenton reaction suppress erythropoietin production under normoxia.

We have recently proposed a H2O2-generating b-type cytochrome as part of the cellular oxygen sensor that controls O2-dependent erythropoietin (Epo) production in the human hepatocellular carcinoma cell line HepG2. H2O2 could act as an intracellular signaling molecule because its production in HepG2 cells is strictly dependent on the pericellular PO2. High cellular levels of H2O2 inhibit hypoxia-induced Epo production while low levels-as under hypoxic conditions-allow full expression of the Epo gene. Since cobalt chloride (CoCl2) and the iron chelator desferrioxamine (DSF) both mimic the hypoxic induction of Epo production we studied the influence of CoCl2 and DSF on the formation and on the action of reactive O2-species with respect to Epo production. Both chemicals reduced the H2O2-dependent 123-dihydrorhodamine fluorescence in HepG2 cells. The inhibition of Epo production by exogenous H2O2 was completely antagonized by DSF. This might indicate that H2O2 exerts its inhibition through a Fenton type reaction. On the other hand, NADPH and pyrogallol which stimulate the production of O2- inhibited Epo production. CoCl2 antagonized their effects. From our results we propose different sites of interaction with the putative signaling chain for DSF and CoCl2. While DSF appears to reduce the action of the H2O2 molecule, CoCl2 might act further upstream through the induction of H2O2-scavenger systems or by interfering with its production.

In superfused in vitro rat carotid body, we recorded chemoreceptor discharges and the redox state of cytochromes simultaneously to identify the primary oxygensensing protein controlling transmitter release and electrical activity of the carotid sinus nerve. These parameters were tested under the inf luence of heme ligands such as oxygen, cyanide, 4-(2-aminoethyl)-benzenesulfonyl f luoride, and CO. During stimulation, there was an initial increase in discharge frequency followed by a decline or suppression of activity. Photometric changes lagged and were maintained as nerve activity decreased. Reducing mitochondrial cytochromes by cyanide or prolonged severe hypoxia, suppressed the chemoreceptor discharge. 4-(2-Aminoethyl)-benzenesulfonyl f luoride, a specific inhibitor of the phagocytic cytochrome b 558 , also silenced the chemoreceptors after an initial excitation. CO increased the chemoreceptor discharge under normoxia, an effect inhibited by light, when the cytochromes were not reduced. When the discharges were depressed by severe hypoxia, exposure to light excited the chemoreceptors and the cytochromes were reduced. The rapidity of the chemosensory responses to light and lack of effect on dopamine release from type I cells led us to hypothesize that carotid body type I cells and the apposed nerve endings use different mechanisms for oxygen sensing: the nerve endings generate action potentials in association with membrane heme proteins whereas cytosolic heme proteins signal the redox state, releasing modulators or transmitters from type I cells.