inducing genes (Ziauddin and Sabatini, 2001). This would allow high-thoughput
screening for cell-death inducers, an aim that is also pursued with the original screen
(see below).
While the correlation and the selection strategy are focused on specific external
stimuli for apoptosis, in the screening assay the transfected genes themselves induce
apoptosis. What consequences does this have for the kind of genes that can be expected
from this assay? For one, genes can be isolated whose endogenous counterparts
mediate a genuine proapoptotic stimulus: upon their inactivation, less apoptosis is
induced. As such a reduced cell death can lead to tumorigenesis, they are also
candidates for tumor suppressors. In fact, two of the genes that have previously not
been linked to apoptosis are only recently discovered tumor-suppressor genes
(Albayrak et al., submitted; Schoenfeld et al., submitted). The cellular counterparts
of the second group of genes from the screen might not themselves transmit a
proapoptotic signal. Nevertheless, when overexpressed, they activate and therefore
define sensors that mediate induction of cell death. These sensors can be expected to
be protein complexes, as apoptosis seems to be induced—at least in most cases—by
direct protein-protein interactions (see Chapter 5). To this class belong also those
genes that are upregulated or constitutively active in diseases, and that therefore
constitute potential drug targets (such as ANT-1 in DCM). Two genes whose
dominant alleles are responsible for degenerative diseases have been uncovered by the
screen (Albayrak et al., submitted). As the expression of the genes isolated by this
screen is sufficient for apoptosis induction, it is possible to detect genes outside the
main apoptosis-signaling pathways. Hence, a subset of the genes might induce
apoptosis only under specific circumstances. An example is NDF, which acquires its
proapoptotic activity when it is activated as an oncogene. This category also includes
genes that are operative only in certain cell lines, such as the transformed tumor cells
used in the screen. Some examples of tumor-specific apoptosis-inducing genes have
been described previously, such as apoptin (Danen-Van Oorschot et al., 1997), mda-7
(Su et al., 1998), and the TRAIL receptor (Walczak et al., 1999).
Even though the screen still leads to individual proapoptotic genes, the longterm
goal of this work is the same as with other assays for apoptosis genes: to establish a
network of functionally interacting proteins—defined by the isolated genes— that
eventually leads to the activation of caspases and to apoptosis induction. In the case
of the screen, this concept calls for a complete coverage of apoptosis-inducing genes.
Such a compendium of apoptosis inducers would indicate the sensors involved. In
addition, the sheer number of genes from the screen makes it imperative to add
additional information (such as expression data, disease-relevance, and sequence
comparisons). An inventory of apoptosis genes would thereby allow researchers to
focus on those that can be easiest integrated into a known physiologic or pathologic
context in which their signal becomes relevant to induction of cell death. To
implement a saturation screen, the assay has already gone through different stages to
increase the throughput (Figure 2). While first done by isolating plasmid DNA in
single reaction tubes (Grimm and Leder, 1997), it can now be performed in 96-well
microtiter plates (Neudecker and Grimm, 2000). This allows workers to reduce the
CELL-CULTURE SYSTEMS IN APOPTOSIS 209