In central neurones, two possible mechanisms account for slow hypoxic inhibition of K+ channels in excised patches: first, exogenous application of reduced moieties suggests the redox mechanism described above. Second, there is evidence for the involvement of an iron-containing but nonhaem protein. This is based on the observation that the effects of hypoxia could be mimicked by agents that chelate metal in metal-containing proteins16. Using canine pulmonary myocytes, Post et al. 31 demonstrated that hypoxia inhibited delayed rectifierlike K+ channels via release of Ca2+ from caffeine-sensitive intracellular Ca2+ stores. Thus, different mechanisms, and evidence to support these mechanisms, exist to account for hypoxic inhibition of K+ channels. However, what remains unresolved (and is of paramount importance to determine) is whether the channels can themselves directly act as O2 sensors32. That is, can they directly respond to low PO2, or does hypoxia require an intermediate O2 sensor that couples to the channel? Direct redox modulation, for example, could be mediated via thiol-containing cysteine residues in the channel protein. Certainly, K+ channels (both α-and ß-subunits) possess cysteine residues on their cytosolic aspect which can undergo oxidation and reduction and so modulate channel inactivation. This has been shown in a recombinant voltage-gated K+ channel33, but in response to reducing agents, and not so far to hypoxia