triggering of the mitochondrial permeability pore, cytochrome c is released from the
mitochondrium and binds to the WD-40 repeat region of APAF1 (Zou et al., 1997).
As a consequence, the N-terminal CARD of APAF1 recruits caspase-9 by its CARD-
containing prodomain and thus activates the enzyme by induced proximity
(Srinivasula et al., 1998). Several other proteins with established or likely roles in
apoptosis signaling were found to contain CARD domains, including two IAP-types
apoptosis inhibitors (discussed in Section 4.2) and RIP2/RICK, a protein kinase
closely homologous to RIP, but with a CARD instead of the death domain.
The function of the CARD domain does not seem to be strictly specific for
apoptosis; instead, several CARD-containing proteins appear to have a role in
interleukin-1 signaling and/or NF-κB activation, as in the situation encountered with
the death domains. Nonapoptotic caspases, such as the founding member caspase-1
(ICE), also contain a CARD in their prodomain and are likely to be activated by a
mechanism similar to the other CARD caspases. Caspase-1 is required for the cleavage
and secretion of interleukin-1. The CARD is frequently found in the same proteins
as the pyrin domain discussed in the next section, which has its main role in
inflammatory signal transduction.
In the current public database, 31 different mammalian proteins with CARD
domains are listed (Table 5). Like the related death adapter domains, the CARD
domain spans a region of approximately 90 residues and is located preferentially at
the N—or C-termini of proteins. An alignment of selected CARD domains is shown
in Figure 1. All characterized CARD domains appear to work by mediating the
interaction with other CARD domains, in a way analogous to the death and death-
effector domains. The evolutionary relationship between DDs, DEDs, and CARDs
had originally been proposed through sequence analysis (Hofmann et al., 1997) and
was confirmed by the solution of several three-dimensional structures of CARD
domains. The first CARD structure to become available was that of RAIDD; it
revealed a six-helix bundle similar to that of the other death adapter domains (Chou
et al., 1998). Of particular interest is the structure of an interaction complex involving
the CARD domains of APAF1 and caspase-9 (Qin et al., 1999). Like the structure
of the DD-DD interaction complex, this structure reveals a dimeric interaction mode.
There is, however, a major difference in the interaction surface used, which is
completely different in the two complex structures. Since the two observed
interactions modes are not mutually exclusive, it has been suggested that all classes
of death adapter domains can form trimeric complexes in physiologic situations, using
both types of interactions simultaneously (Weber and Vincenz, 200 1a; 2001b).
84 GENETICS OF APOPTOSIS