the original rheostat model, both anti- and proapoptotic BH proteins form
homodimers or associate as mixed heterodimers (Korsmeyer et al., 1993). As the
initial biologic activity of Bax was shown to inhibit Bcl-2-mediated survival, Bcl-2:
Bax heterodimers were postulated as the molecular basis for the Bax death-promoting
activity, as a means of inhibiting an innate Bcl-2 survival function. This ordering of
action/reaction has fallen out of fashion, primarily due to two developments. Recent
experimental models are more focused on the direct killing activities of proapoptotic
BH proteins, studied with inducible or transient expression systems (Korsmeyer et
al., 1999; Antonsson, 2001). This co nceptio n o f Bax as a di rec t mediator of c ell death,
rather than a regulator, has also been reinforced by the demonstration that
endogenous Bax is a predominantly cytosolic protein that translocates to a
mitochondrial site of action once an apoptotic stimulus is delivered (Hsu et al., 1997b;
Wolter et al., 1997). Finally, the discovery that BH proteins have pore-forming
activities, predicted by the solution of the Bcl-xL x-ray crystallographic structure, has
been more easily integrated with an inherent death-inducing activity of the
proapoptotic BH members than a survival function (Muchmore et al., 1996).
Whichever direction the action/reaction axis is pointing, mutational studies have
tended to support the functional importance of heterodimeric associations of pro—
and antiapoptotic proteins. The BH family is recognized by the conserved homology
domains BH1–4. Certain mutations in the antiapoptotic BH members, such as the
BH1 domain Gly^138 Ala in Bcl-2 and Gly^159 Ala in Bcl-xL, disrupt associations with
Bax and result in strong loss of survival function (Yin et al., 1994; Yang et al., 1995).
Similarly, substitution of hydrophobic amino acids in the Bax BH3 domain with
charged amino acids, including the Gly67Arg mutation identified in a leukemic cell
line (Meijerink et al., 1998), eliminate heterodimeric interactions with Bcl-xL and
Bcl-2 and reduce cell-death activity (Wang, K. et al., 1998).
A ratchet was thrown into this model by the demonstration that cytosolic BH
proteins could be converted from monomers to dimers with detergents. Youle and
colleagues reported that Bax associations with self, Bcl-2, and Bcl-xL could result from
the presence of nonionic detergents, such as Triton X-100, used to prepare cell extracts
(Hsu and Youle, 1998). Furthermore, Bax dimers were not detected in this study
with the membrane-inserted form of endogenous Bax found in apoptotic cells, using
whole-cell cross-linking methods or membrane extraction in 1% CHAPS detergent,
which did not induce artifactual Bax dimerizations.
This study runs counter to numerous reports demonstrating the presence of Bax
hetero—and homodimers, in isolated mitochondria treated with cross-linkers (Gross
et al., 1998; Desagher et al., 1999; Antonsson et al., 2000; Makin et al., 2001), and
by fluorescence resonance energy transfer experiments in intact cells (Mahajan et al.,
1998; Degterev et al., 2001). As with other features of these enigmatic proteins, while
there is little doubt that dimeric and perhaps oligomeric associations occur under
some circumstances, it remains unresolved whether the core and physiologically
relevant functions of BH proteins depend on partnering. Indicative of the confusion
on this point, there have been reports of mutations of both Bcl-xL and Bax proteins
MAKING SENSE OF THE BCL-2 FAMILY OF APOPTOSIS REGULATORS 51