Credit: © Araldo de Luca/CORBIS

The developing mouse embryo is an exquisite molecular sculpture. Early on, digits and organs are shaped by a programmed cell death that bears the characteristic hallmarks of apoptosis. Paradoxically, however, organ sculpting is normal (albeit delayed) in embryos missing the usual apoptotic suspects such as caspase-9 and Apaf-1. One explanation is that the embryo is shaped by a second, caspase-independent pathway, and a report in Nature now supports this idea.

A little over two years ago, Guido Kroemer, Josef Penninger and collaborators cloned and characterized apoptosis-inducing factor (AIF), a mitochondrial factor released in response to death-triggering stimuli. They showed at the time that microinjection of recombinant AIF to the cytoplasm of cells led to apoptotic responses, such as the exposure of phosphatidylserine on the cell surface.

Kroemer, Penninger and collaborators now implicate AIF in developmental cell death by showing that genetic inactivation of AIF abolishes death in a model of early embryogenesis. To do this, they first used homologous recombination to disrupt the aif gene in mouse embryonic stem (ES) cells. Then, to mimic the conditions of early development, the authors cultured aggregates of the AIF-deficient ES cells in the absence of leukaemia inhibitory factor and feeder cells. Under these conditions the ES cells develop into embryoid bodies — multicellular aggregates containing an outer layer of endodermal cells and a solid core of undifferentiated ectodermal cells. Cells within this inner core then normally undergo a wave of programmed death in a process known as cavitation.

How does the lack of AIF affect cavitation? The authors found that, in embryoid bodies derived from the AIF-deficient ES cells, cavitation was completely blocked. Then, by checking for the incorporation of 5-bromodeoxyuridine, they confirmed that this apparent block to cell death was not due simply to increased cell proliferation.

During cavitation in wild-type embryoid bodies, caspase-3 — a downstream effector of caspase-9/Apaf-1 — is usually activated. To check the involvement of caspases in cavitation, Kroemer, Penninger and co-workers studied embryoid bodies derived from caspase-9−/− and apaf-1−/− ES cells. In both cases, although caspase-3 was no longer activated, cavitation proceeded as normal. The inference, then, is that the AIF-dependent cell death seen in cavitation is independent of caspase-dependent death pathways.

The idea of a second pathway is a controversial one. But the authors tentatively speculate that AIF may “represent a pathway of apoptosis that predates the caspase pathway”. In support of this idea, they say, is the fact that AIF homologues have been identified in all metazoan phyla, whereas there is, as yet, no evidence for caspases in plants or in unicellular organisms that can nonetheless undergo programmed cell death.