For 10 years, our laboratory has been studying the biochemical pathway of apoptosis in human cells. We are particularly interested in apoptosis executed by mitochondria-initiated activation of a group of intracellular proteases, the caspases, which exist in living cells as inactive zymogens. Caspases are activated when cells undergo apoptosis. Our previous work identified the mitochondrial protein cytochrome c as a key caspase activator. When released from mitochondria to cytosol, cytochrome c triggers a five-step chain reaction that activates the majority of intracellular caspases.

We observed previously that the tumor-suppressor protein PHAPI (putative HLA-associated protein I) promotes caspase activation. But the mechanism through which PHAPI works was unknown. Simply using recombinant proteins of the known caspase activation factors failed to reconstitute the PHAPI effect in vitro. To identify the missing factors, we applied classical biochemical fractionation and affinity purification procedures. We found that two more protein factors, which were not known to play a role in cytochrome c–mediated caspase activation, are necessary and sufficient to reconstitute the PHAPI response. Knocking down one of the proteins with RNA interference (RNAi) technology rendered cells lacking this protein resistant to apoptosis and caspase activation induced by ultraviolet light. We found that this protein facilitates nucleotide exchange to Apaf-1 after the original Apaf-1-bound dATP is hydrolyzed by cytochrome c. When the nucleotide exchange fails to occur, more functional apoptosome formation is promoted, instead of aggregation.

The caspase activation initiated by mitochondria is controlled by the Bcl-2 family of proteins. One family member, Mcl-1, prevents apoptosis and works upstream of other members. Mcl-1 is degraded during apoptosis induced by a variety of agents, and its disappearance from cells is necessary for apoptosis to proceed. We have isolated an Mcl-1 ubiquitin ligase E3 (Mule) that mediates the degradation of Mcl-1, and we have identified a BH3 domain. BH3 domains are used by the Bcl-2 family members to interact with each other. In the case of Mule, the BH3 domain is used to bind Mcl-1.

Mule turns out to be a multifunctional enzyme. In addition to mediating the degradation of Mcl-1, Mule also mediates the degradation of tumor-suppressor protein p53 and is regulated by another tumor-suppressor protein, ARF (reported by Wei Gu's group, Columbia University). We recently mapped the p53- and ARF-binding regions of Mule and discovered that their binding sites overlap and are mutually exclusive but distinct from Mule's BH3 domain. The binding of ARF to Mule moves Mule from the cytosol, where Mcl-1 is, to the nuclei. We are now studying how the recognition of different substrates by Mule is regulated.

We have continued our biochemical studies of RNAi, a form of post-transcriptional gene silencing whereby double-stranded RNA (dsRNA) molecules trigger the sequence-specific degradation of cognate messenger RNA (mRNA). The biochemical pathway of RNAi consists of two sequential steps: first, long dsRNA molecules are cleaved into 21- to 23-nucleotide fragments, the small interfering RNAs (siRNAs). Second, the siRNA is incorporated into a multisubunit nuclease complex—RNA-induced silencing complex (RISC)—and functions as a guide RNA to direct the RISC-mediated sequence-specific mRNA degradation. The mRNA-cleavage step of RNAi is mediated by Argonaute-2 (Ago-2), an endonuclease within the RISC. Ago-2 uses one strand of the siRNA duplex as a guide to find mRNAs containing complementary sequences and cleaves the phosphodiester backbone at a specific site measured from the guide strand's 5' end.

We focused on a specific question: How does double-stranded siRNA become single stranded, so that it can recognize the cognate mRNA through base pairing. We found that both strands of siRNA are loaded onto Ago-2 protein in Drosophila S2 cell extracts. The antiguide strand behaves as an RISC substrate and is cleaved by Ago-2. This cleavage event is important for the removal of the antiguide strand from Ago-2 protein and the activation of RISC.