Genetics of Apoptosis

A point of convergence between receptor-induced apoptosis and that elicited by
different stimuli is the triggering of the mitochondrial permeability pore. A central
role in this process is filled by proteins of the Bcl-2 family, harboring several conserved
motifs that are frequently referred to as the BH1 through BH4 regions. Among the
factors liberated from the mitochondrion is cytochrome c, which binds to and
activates another layer of adapter proteins that recruit downstream caspases. The
domains involved in this process include CARD and the NB-ARC-type ATPases.
The apoptotic signaling pathway can be described by a multilayer model in which
the different layers are populated by different domain types. A recurring mechanism
in this pathway is the activation of downstream enzymes by ‘induced proximity’
(Salvesen and Dixit, 1999). Evolution has heavily exploited the modular nature of
the signaling proteins by designing single-domain inhibitors. This phenomenon
occurs not only in physiologic situations but also in pathologic processes. A deep
understanding of the resulting signaling network is thus a prerequisite of any
pharmacologic intervention targeting apoptosis.

1.1

The domain concept

The concept of protein domains originally comes from the analysis of three-
dimensional protein structures. Most small proteins (<100 residues) have a
‘monolithic’ structure, consisting of various secondary structure elements, folded in
a way that gives rise to a hydrophobic core and hydrophilic solvent-exposed regions.
Larger proteins, by contrast, can follow two different architectural principles: in
addition to forming a larger monolithic structure, they can also consist of several
smaller folding units, the so-called domains. Each of these domains can fold
independently from the rest of the protein and has its own hydrophobic core region.
Most inter-residue contacts are satisfied within the domain, only a few interactions
being found between domains. As a consequence of their autonomous folding
capabilities, domains can typically be excised from their host protein and pasted into
a different context, while maintaining both fold and function. In the course of
evolution, this series of events has happened several times for many domain types.
Evolutionary processes such as exon shuffling, together with the duplication, fusion,
and fission of genes and gene regions, have helped to create the multidomain ‘mosaic’
structure found in many extant proteins.
In the absence of structural information, domains can frequently be detected by
analyzing the protein sequences alone. When comparing two dissimilar sequences
that both have acquired a domain of the same type by shuffling events, the domain
typically appears as a region of localized sequence similarity. Such regions are
frequently referred to as ‘homology domains’. However, a region of local sequence
homology is not necessarily a true homology domain: the possibility that the detected
homolo gy regio n is just the best-co nserved part of two pro teins wi th ove rall homology
has to be excluded first. This question can be answered unambiguously only in those
situations where the boundaries of the putative homology domain are well defined,

72 GENETICS OF APOPTOSIS