Traveling Bax and forth from mitochondria to control apoptosis.

Antiapoptotic Bcl-2 proteins on mitochondria inhibit prodeath proteins, such as Bax, which are found primarily in the cytosol. In this issue, Edlich et al., (2011) show that Bax and Bcl-xL interact on the mitochondrial surface and then retrotranslocate to the cytosol, effectively preventing Bax-induced permeabilization of mitochondria.

Main Text

The apoptotic cell death pathway, in its basic tenents, is conserved among metazoans. During development, apoptosis shapes organs and tissues; during adulthood, it maintains tissue homeostasis. Resistance to apoptosis is a hallmark of cancer, whereas excessive programmed cell death is associated with degenerative diseases. Apoptosis is controlled and amplified through a variety of mechanisms, many of which converge on release of proapoptotic factors from mitochondria (Danial and Korsmeyer, 2004). Edlich et al. (2011) now uncover how regulated trafficking of the proapoptotic protein Bax off mitochondria and into the cytosol prevents mitochondrial permeabilization, reconciling discordant models for how antiapoptotic proteins on the mitochondrial surface inhibit proapoptotic proteins in the cytosol.

The permeabilization of mitochondria is controlled by proteins belonging to the Bcl-2 family, which is composed of anti- and proapoptotic members sharing homology in short stretches of amino acids called Bcl-2 homology (BH) domains. The prosurvival members, including Bcl-2, Bcl-xL, and Mcl-1, contain BH domains 1–4 and antagonize the prodeath members. Two types of pro-death Bcl-2 family members participate in apoptosis. The multidomain proteins, such as Bax and Bak, contain BH domains 1–3 and directly mediate mitochondrial permeabilization, whereas the BH3-only proteins (e.g., Bid, Bim, Bad, etc.) are thought to act as sensors of cellular stress (Danial and Korsmeyer, 2004). Despite clearly defined roles for the Bcl-2 family members, two major questions have remained: how are proapoptotic members activated, and how do the antiapoptotic members inhibit them?

The simplified rheostat model, first proposed in the early 1990s, suggests that pro- and antiapoptotic proteins directly counterbalance each other, with the more abundant protein dictating whether the cell dies or survives (Danial and Korsmeyer, 2004). Complicating this model is the existence of two types of BH3-only proteins: the “activators” that directly activate the proapoptotic multidomains and the “sensitizers” that neutralize the antiapoptotic inhibition of the multidomain proteins (Letai et al., 2002). A modified version of the rheostat model postulates that activator BH3-only proteins act as receptors for multidomain proapoptotic proteins on the mitochondrial outer membrane (OM), whereas antiapoptotic proteins act like a sponge, soaking them off of the mitochondrial surface (Lovell et al., 2008). A mutually exclusive model postulates that the antiapoptotic proteins directly bind and inhibit Bax and Bak, with BH3-only molecules acting as inhibitors of the inhibitors (Youle and Strasser, 2008). This latter model suffered from the lack of evidence that could explain the spatial paradox of this direct inhibition: under normal circumstances, Bax is in the cytosol, whereas Bcl-2 is on the surface of mitochondria. Upon induction of apoptosis, Bax accumulates on the mitochondrial OM, undergoes conformational changes to expose a helical domain, and triggers the release of cytochrome c and other proapoptotic effectors.

To gain insight into the regulation of Bax, Edlich et al., (2011) examine the localization of a mutant version of Bax that is constrained in its cytosolic conformation by engineered disulfide bridges. Surprisingly, this mutant protein, which does not interact with the mitochondrial antiapoptotic protein Bcl-xL, localizes to mitochondria but does not induce apoptosis. Following up on this observation, the authors ask whether Bax normally translocates to the mitochondrial surface. Using fluorescence loss in photobleaching, they monitor the localization of GFP-Bax in cells and find that Bax cycles on and off the mitochondria in healthy cells. The presence of antiapoptotic proteins in the mitochondrial OM is required to constitutively retrotranslocate Bax into the cytosol. Notably, the rate of retrotranslocation is almost doubled in cells overexpressing Bcl-xL and requires the physical interaction between Bcl-xL and Bax. Once in the cytosol, Bax quickly returns to its monomeric form, ready to cycle back to the mitochondrial surface. Edlich and coworkers observe an acceleration of the retrotranslocation rate upon the coexpression of Bcl-xL and Bax. Similarly, the Bcl-2/Bcl-xL inhibitor ABT-737, as well as BH3-only proteins, disrupts the equilibrium between cytosolic and mitochondrial Bax by reducing its retrotranslocation (Figure 1).

Figure 1

Bax Movements in Healthy Cells and during Apoptosis

In healthy cells, inactive Bax continuously cycles between mitochondria and the cytosol. Bax retrotranslocation requires interaction with an antiapoptotic protein (Bcl-xL, Bcl-2, or Mcl1). Together, these two proteins leave the mitochondrial outer membrane (OM). Once in the cytosol, the complex immediately dissociates. The retrotranslocation process is stimulated by the antiapoptotic proteins Bcl-xL, Bcl-2, or Mcl1 and is inhibited by vMIA, ABT-737, and BH3-only proteins. Upon induction of apoptosis, Bax is directly stimulated by activating BH3-only proteins (e.g., Bid, Bim, or Puma; blue arrow) to expose its C-terminal domain and insert in the mitochondrial OM. During this process, Bax exposes a novel N-terminal epitope (6A7), triggering the formation of foci and release of cytochrome c. Neutralizing BH3-only proteins (or small molecule inhibitors; green rectangle) can indirectly activate Bax by binding and inactivating antiapoptotic proteins. Consequently, Bax accumulates on the mitochondrial OM, where it acquires its active conformation.

This new study provides a new paradigm for understanding Bax regulation during apoptosis, but how the two categories of BH3-only proteins antagonize Bax retrotranslocation by Bcl-xL remains an open question. One possibility is that the activators (e.g., Bid and Bim) directly inhibit Bax retrotranslocation by unmasking its α9 helix and anchoring Bax in the mitochondrial OM. Conversely, the sensitizer (e.g., Bad) could obstruct the binding of Bax to Bcl-2. The interplay between proapoptotic and antiapoptotic proteins is part of the normal regulation of the trafficking of Bax in healthy cells. Under nonapoptotic conditions, the equilibrium between the cytosolic and mitochondrial pools of Bax is maintained by its Bcl-xL-dependent retrotranslocation. During apoptosis, however, this balance may be tipped by BH3-only proteins, which increase the insertion of Bax into the mitochondrial OM (Figure 1). Thus, the retrotranslocation model is an important step forward, reconciling two opposing models of how BH3-only, multidomain, and antiapoptotic proteins cooperate to dictate whether a cell lives or dies (e.g., Cheng et al., 2001; Lovell et al., 2008 versus Willis et al., 2005).

In addition to this new model, Edlich and colleagues also find that the weak interaction between Bcl-xL and Bax is sufficient to extract Bcl-xL from the mitochondrial OM and transiently localize it to the cytosol. This highlights an unappreciated conformational change that facilitates the removal of an antiapoptotic protein from the mitochondrial OM. Interestingly, the rate of BcL-xL retrotranslocation increases upon treatment with ABT-737, raising the possibility that inactive Bcl-xL might be predominantly cytosolic. It will be important to verify whether this is a feature common to extramitochondrial antiapoptotics, such as endoplasmic reticulum (ER)-localized Bcl-2/Bcl-xL, and to understand the mechanism by which interaction with Bax inactivates the transmembrane domain of Bcl-xL. Similarly, this study illustrates that inactive Bax preferentially associates with the mitochondrial OM. Interestingly, a fraction of inactive Bax is associated with intracellular membranes, including mitochondria and ER in transformed cells, but not primary hepatocytes (Scorrano et al., 2003). Understanding how oncogenic transformation influences the on and off rates of Bax translocation could open new perspectives to apoptosis in cancer cells.

From a teleological perspective, one might wonder why a cell would maintain a continuous flux of Bax cycling on to the mitochondria and then back to the cytosol. One reason could be to prime cells to rapidly execute death, positioning Bax in a “ready to attack” mode such that minor changes in the kinetics of Bax retrotranslocation can swiftly induce apoptosis. Another possibility is that Bax has additional functions that require its regulated presence on cellular membranes. Interestingly, inactive Bax controls transient membrane events such as mitochondrial fusion (Hoppins et al., 2011) and maintenance of ER Ca2+ levels (Scorrano et al., 2003). In these cases, however, Bax cannot remain on the organelle without becoming active and promoting apoptosis (Nutt et al., 2002; Hoppins et al., 2011). The retrotranslocation model not only provides a paradigm for understanding how apoptosis is regulated, but also provides a platform for integrating the diverse functions of Bax and other Bcl-2 family members.