Cellular damage induces death with the highly conserved features of apoptosis. How do disparate types of damage (DNA breaks, inhibition of microtubules, denaturation of proteins etc) all signal to some point of integration to initiate conserved biochemical processes? We have studied the activation of constitutively expressed but latent pro-apoptotic proteins by disparate damage signals. Irrespective of the type of damage, a step-wise change in epitope availability occurs on the pro-apoptotic protein Bak, in all cell types studied. The first change is at its N-terminus, where a cryptic epitope is rapidly exposed after cytotoxins, irrespective of damage type. We presume this is the release of a masking protein that maintains Bak in a latent state. At this time Bak remains bound to Bcl-xL and a cryptic N-terminal epitope of Bax remains closed. There are no indications of apoptosis,. A second change then occurs to expose a central Bak BH1 homology domain epitope. Bak is now freed from binding Bcl-xL, the N-terminus of Bax becomes exposed and the cells activate caspases. At no time is Bak cleaved proteolytically. We suggest that, in a stepwise process, latent pro-apoptotic proteins are activated to expose the BH3 domain necessary for initiating apoptosis. The BH3 domain of the anti-apoptotic Bcl-xL appears to be exposed when the BH4 domain is cleaved by proteolysis, rendering it pro-apoptotic. Using a concentric double-barrelled patch clamp pipette, electrical recordings from mitochondria membranes in the intact squid giant axon pre-synaptic terminal were made. Addition, at neutral pH, of full length Bcl-xL protein induced only modest electrical activity over background whereas DN76 Bcl-xL, resembling the BH4 cleaved protein, rapidly induced electrical activity with unitary openings corresponding to conductances between 0.028 and 3.8 nS. DN76 Bcl-xL with a 3 aminoacid truncation in its BH3 domain had no activity. DN76 Bcl-xL did not stimulate electrical activity of molluscan plasma membranes. We suggest that pro-apoptotic proteins are normally latent and require signals to expose the BH3 domain. Our data suggest that it is the BH3 domain and not the pore forming region with two alpha helical domains, believed to lie between domains BH1 and BH2, which is responsible for channel activity. The electrical activity observed is compatible with the opening of the VDAC.