Recent work (Huang et al. 1999; Chandel et al. 2000) suggests that the hypoxic stabilisation and activation of the transcription factor hypoxia inducible factor-1α (HIF-1α) is dependent upon the generation of reactive oxygen and nitrogen species (ROS, RNS). As these may be generated during both hypoxic and hyperoxic excursions, we surmised that there must be fundamental differences in the induction kinetics, oxidant signalling species, or redox buffering system which favours activation of the HIF-1α pathway and hypoxic gene expression. This study aimed to determine the temporal and spatial characteristics of O₂ signal generation which may differentiate between hypoxic and hyperoxic conditions.

Human lung A549 cells stably expressing a HIF-1α-driven green fluorescent protein (GFP) reporter gene (pHRE-d2EGFP; translate half-life = 2 h, K_a,O2 = 50 mmHg) were maintained at P_O2 values of 23 or 142 mmHg or were exposed to activating (142-23 mmHg) or deactivating (23-142 mmHg) O₂ shifts (&Dgr;P_O2) for 8 h. H₂O₂ generation was determined in parallel by incubation of cells with 10 mM dichlorofluorescein-diacetate for the same period of time. H₂O₂ production was 3- to 5-fold higher in cells exposed to severe steady-state hypoxia (< 0.5 mmHg) or either activating or deactivating &Dgr;P_O2 in comparison with those cultured at 23, 100 or 142 mmHg. A549 cultures lacking a mitochondrial genome (ρ^o) revealed that non-mitochondrial H₂O₂ production was proportionately greatest in cells maintained at either P_O2 < 0.5 or 23 mmHg, but was a minor component of total H₂O₂ produced during &Dgr;P_O2 in any direction. The induction of HIF-1α-driven GFP expression was 4-fold greater at < 0.5 and 23 mmHg than that found at 142 mmHg in cells maintained at either oxygen tension. During a moderate activating stimulus (142-23 mmHg), GFP expression was maximal at 6 h, returning to lower steady-state values at 24-48 h. This coincided with a transient increase in HIF-1α nuclear abundance, which was maximal 3 h following the shift and was coincident with a significant (P < 0.01, n = 4, ANOVA with post hoc Tukey’s HSD) transient oxidation of the glutathione redox shuttle. A moderate deactivating stimulus (23-142 mmHg) resulted in a steady decline in GFP expression, a loss of nuclear HIF-1α within 3 h which also coincided with a transiently oxidised glutathione redox shuttle. To differentiate between whole cell (glutathione mediated) and localised redox shuttling in the activation pathway for HIF-1α, we examined the potential for thioredoxin (TRX) to bind the HIF-1α transcriptional complex. Immunoprecipitation studies demonstrated a hypoxia-dependent association of HIF-1α protein with oxidised TRX within the nucleus, which was dispersed on a moderate hyperoxic shift (23-142 mmHg).

We propose that hypoxic and hyperoxic signals are differentiated from one another at the level of HIF-1β activation by localised redox regulation of the transcriptional complex involving the binding and oxidation of TRX.

This work was supported by the Medical Research Council (S.C.L.).