The major heat shock protein, Hsp70, is an effective inhibitor of apoptosis. To study its mechanism of action, we created tumor cell lines with altered Hsp70 levels. The expression levels of Hsp70 in the cells obtained correlated well with their survival following treatments with tumor necrosis factor, staurosporine and doxorubicin. Surprisingly, the surviving Hsp70‐expressing cells responded to the apoptotic stimuli by activation of stress‐activated protein kinases, generation of free radicals, early disruption of mitochondrial transmembrane potential, release of cytochrome c from mitochondria and activation of caspase‐3‐like proteases in a manner essentially similar to that of the dying cells with low Hsp70 levels. However, Hsp70 inhibited late caspase‐dependent events such as activation of cytosolic phospholipase A2 and changes in nuclear morphology. Furthermore, Hsp70 conferred significant protection against cell death induced by enforced expression of caspase‐3. Thus, Hsp70 rescues cells from apoptosis later in the death signaling pathway than any known anti‐apoptotic protein, making it a tempting target for therapeutic interventions.

The transcriptional activity of c-Jun is augmented through phosphorylation at two sites by a c-Jun amino-terminal kinase (JNK). All cells express two distinct JNK activities, 46 and 55 kD in size. It is not clear which of them is the more important c-Jun kinase and how they specifically recognize c-Jun. The 46-kD form of JNK was identified as a new member of the MAP kinase group of signal-transducing enzymes, JNK1. Here, we report the molecular cloning of the 55-kD form of JNK, JNK2, which exhibits 83% identity and similar regulation to JNK1. Despite this close similarity, the two JNKs differ greatly in their ability to interact with c-Jun. JNK2 binds c-Jun approximately 25 times more efficiently than JNK1, and as a result has a lower Km toward c-Jun than JNK1. The structural basis for this difference was investigated and traced to a small beta-strand-like region near the catalytic pocket of the enzyme. Modeling suggests that this region is solvent exposed and therefore is likely to serve as a docking site that increases the effective concentration of c-Jun near JNK2. These results explain how two closely related MAP kinases can differ in their ability to recognize specific substrates and thereby elicit different biological responses.