Remodeling for Demolition

Molecular Cell, Vol. 9, 1–9, January, 2002, Copyright 2002 by Cell Press

Preview

Remodeling for Demolition: Changes in Mitochrondrial Ultrastructure during Apoptosis The release of Cytochrome c from mitochondria is a pivotal event in the apoptotic process, but the precise mechanisms remain controversial. A recent paper proposes that this involves two steps, one involving permeabilization of the mitochondrial outer membrane and a second in which the structural organization of the cristae changes to make Cytochrome c accessible for release. Both events are induced by proapoptotic members of the Bcl-2 family and appear to be separable. For apoptosis researchers, mitochondria are a source of both inspiration and frustration. The knowledge that mitochondria can play important roles in cell death regulation such as the release of Cytochrome c inspires us to understand more about the mechanisms involved. On the other hand, our attempts to elucidate just how proteins escape from mitochondria have been fraught with controversy and contradiction; thus the frustration. Indeed, the stories scientists tell about how Cytochrome c becomes liberated from mitochondria are as diverse as the tales parents tell their young children about how babies are born. The fear for apoptosis researchers is that they may be just as fanciful. In a paper that appears in the January 2002 issue of Developmental Cell, Scorrano and colleagues have examined mitochondria that release Cytochrome c upon treatment with a proapoptotic member of the Bcl-2 family called Bid (Scorrano et al., 2002). Their findings suggest a profound remodeling of mitochondrial structure, where the cristae lose their tightly packed and orderly folded appearance, and the space between the inner and outer membranes of these organelles expands as the matrix is compressed, without precipitating a change in the overall size of the mitochondria. The biggest surprise, however, is that these ultrastructural changes of the inner membrane precede the bulk of Cytochrome c release across the outer membrane, implying that somehow cytosolic proteins (such as Bid and other members of the Bcl-2 family that target the outer membranes of mitochondria) transduce signals into the interior of these organelles. Furthermore, the authors propose a two-compartment model for the intermembrane space (IMS), in which part of the Cytochrome c (ⵑ15%) is immediately available for release, whereas the rest (ⵑ85%) is trapped in the narrow necks of the cristae, requiring inner membrane (IM) remodeling to free it. Before considering the implications of these findings, let us first recognize that more than one mechanism for releasing Cytochrome c and other proteins sequestered in the inner membrane space of mitochondria probably exists. For example, excessive cytosolic Ca2⫹ can trig-

ger opening of a large conductance channel in the IM, known as the permeability transition pore complex (PTPC), which dissipates the electrochemical gradient across the IM and precipitates osmotic swelling of the matrix space, followed by organelle rupture. Swelling and rupture of the outer membrane (OM) represents one modus of Cytochrome c release, and it is probably relevant to cell death in the context of ischemia-reperfusion injury. However, programmed cell death that occurs physiologically and in other situations is not accompanied by organellar swelling, but nevertheless often involves mitochondria. What accounts for the swelling-independent route of Cytochrome c release? Here too there may be more than one mechanism (for reviews, see Green and Reed, 1998; Martinou and Green, 2001; Vander Heiden and Thompson, 1999). In the paper by Scorrano et al., the specific stimulus evaluated was addition of recombinant tBid protein to isolated liver mitochondria. Thus, one must be cautious not to overinterpret the generality of the findings, particularly given that bid knockout mice are overtly normal in phenotype, displaying deficits in cell death only when subjected to certain experimental stresses and that Bid is not expressed in all cells in vivo (Krajewska et al., 2002). In this regard, Bid is a potentially unique proapoptotic member of the Bcl-2 family. Its 3-dimensional structure resembles that of several other pro- and antiapoptotic Bcl-2 family members, which share structural similarity with certain pore-forming bacterial toxins. However, while Bid can form ion channels in synthetic membranes in vitro (Schendel et al., 1999), its ability to induce apoptosis is entirely dependent on the presence of proapoptotic proteins Bax or Bak (Cheng et al., 2001). Moreover, Bid can dimerize with Bax and Bak, via its BH3 domain, inducing these killer proteins to oligomerize in mitochondria membranes, with oligomerization representing an event associated with Cytochrome c release in previous models (reviewed in Korsmeyer et al., 2000). Scorrano et al. show evidence that tBid induces a change in the permeability of the IM of mitochondria, based on release of calcein dye from the matrix, but without triggering overt depolarization. Further, this IM permeability change is suppressible by Cyclosporin A (CsA)—a drug that inhibits a matrix peptidyl proylyl isomerase (PPIase) that is known to affect the PTPC. However, CsA did not impair tBid-induced oligomerization of Bak, thus separating those two functions of tBid. Furthermore, a BH3 mutant of Bid that failed to induce Bak oligomerization remained competent at calcein dye release, and Bak-deficient mitochondria were equally sensitive to this effect of tBid on IM permeability (Scorrano et al., 2002). Importantly, tBid induced striking and rapid morphological changes in the cristae within isolated mitochondria in vitro, which were completely suppressible by CsA but independent of Bak/Bax and which could be dissociated from Cytochrome c release across the OM. Altogether, these findings implicate tBid in a Bak/Baxindependent, CsA-suppressible mechanism affecting changes in the IM permeability and structure, begging

Molecular Cell 2

Figure 1. Mitochondria Remodeling during Apoptosis Two views of how mitochondria release cytochrome c. On the left (one view) is the conventional (albeit controversial) view, in which cytochrome c and other intermembrane space proteins are released either as a result of direct permeabilization of the outer membrane via a “pore” (far left), or through the opening of the permeability transition pore, matrix swelling, and rupture of the outer membrane (center). The right (another view) is that represented by the work of Scorrano et al., who suggest that there are two separable events: the permeabilization of the outer membrane via a pore and the reorganization of the cristae via a permeability transition.

the question of how tBid in the OM induces rapid changes to the IM compartment. Does tBid translocate to the IM? Does it relay signals to the IM via contact with OM proteins, such as components of the PTPC? Further, since the BH3 domain seems to be unimportant for this effect, are the pore-forming regions of tBid required for these effects on IM permeability? Mutagenesis studies would be instructive in gaining a better understanding of the mechanisms involved. Further, since Bid is expendable for developmental cell death and normal tissue homeostasis, what other members of the Bcl-2 family serve similar functions when Bid is absent? Among Bcl-2 family proapoptotics, only Bax, Bak, and Bok are predicted to share structural similarity with Bid. The diverse BH3-only members of the family seem to lack the pore-forming regions, based on structure prediction algorithms. Also at issue is whether all mitochondria are created equal, given that a previous report using tomographic reconstruction of EM images failed to see ultrastructural changes of Xenopus mitochondria under conditions where tBid induced cytochrome c release (von Ahsen et al., 2000). Clearly, differences in the solutions used to support mitochondria could have played a role in this discrepancy, but then how physiological are any of the solutions in which we choose to bathe isolated mitochondria? Scorrano et al. recapitulated parts of their findings using intact cells, showing morphological evidence of cristae reorganization even in cultured cells, including cells lacking both Bax and Bak after treatment. However, time-lapsed video microscopy and rapid filtration techniques have failed to capture evidence of a two-compartment model for Cytochrome c release in intact cells previously (Goldstein et al., 2000; Madesh et al., 2002). Is this a reflection of differences in the mitochondria of different cell types used for the experiments? In addition to Cytochrome c, other proteins such as AIF, Smac, EndoG, and HtrA2 also are released from the IMS during apoptosis. Are these proteins distributed unevenly between subcompartments of the IMS? Evidence that AIF and Smac might be in different compart-

ments than Cytochrome c has indeed been presented, where AIF or Smac reportedly exits mitochondria prior to the bulk of Cytochrome c under some circumstances (Chauhan et al., 2001; Ferri et al., 2000). Since tBid fails to induce Cytochrome c release or apoptosis of cells doubly deficient in Bax and Bak (Cheng et al., 2001), these results of Scorrano et al. also dissociate the mitochondria ultrastructural remodeling from commitment to cell death. Thus, IM remodeling is insufficient for cell death. But, is it necessary? If we accept the idea that roughly 15% of the Cytochrome c in mitochondria can be released without remodeling, then it seems probable that apoptosis could nevertheless occur. But, how much Cytochrome c is sufficient probably differs substantially among cells depending on their levels of Apaf1, Caspases, and Caspase inhibitory IAPs. Still, there must be more to the two-compartment model than remodeling inner membrane structure. Even in hypercondensed mitochondria, the cristae junctions are sufficiently wide that an ⵑ12 kDa protein such as Cytochrome c should be freely diffusable. One wonders whether a network of proteins in the IMS pulls the cristae membranes into tight stacks and presents a barrier to diffusion. Could it be that an essential step in mitochondrial remodeling involves eliminating these hypothetical proteins? Like all good works of science, the paper by Scorrano et al. raises more questions than it provides answers. Seeking answers to those questions should keep apoptosis researchers inspired, frustrated, and probably rather busy.

John C. Reed1 and Douglas R. Green2 1 The Burnham Institute 10901 N. Torrey Pines Road La Jolla, California 92037 2 The La Jolla Institute of Allergy & Immunology 10355 Science Center Drive San Diego, California 92121

Previews 3

Selected Reading Chauhan, D., Hideshima, T., Rosen, S., Reed, J.C., Kharbanda, S., and Anderson, K.C. (2001). J. Biol. Chem. 276, 24453–24456. Cheng, E.H.-Y., Wei, M.C., Weiler, S., Flavell, R.A., Mak, T.W., Lindsten, T., and Korsmeyer, S.J. (2001). Mol. Cell 8, 705–711. Ferri, K.F., Jacotot, E., Blanco, J., Este´, J.A., Zamzami, N., Susin, S.A., Xie, Z., Brothers, G., Reed, J.C., Penninger, J.M., and Kroemer, G. (2000). J. Exp. Med. 192, 1081–1092. Goldstein, J., Waterhouse, N., Juin, P., Evan, G., and Green, D. (2000). Nat. Cell Biol. 2, 156–162.

Krajewska, M., Zapata, J.M., Meinhold-Heerlein, I., Hedayat, H., Monks, A., Shabaik, A., Bubendorf, L., Kallioniemi, O.-P., Kim, H., Reifenberger, G., et al. (2002). Neoplasia, in press. Madesh, M., Antonsson, B., Srinivasula, S.M., Alnemri, E.S., and Hajnoczky, G. (2002). J. Biol. Chem., in press. Martinou, J.-C., and Green, D.R. (2001). Nat. Rev. Mol. Cell. Biol. 2, 63–67. Schendel, S., Azimov, R., Pawlowski, K., Godzik, A., Kagan, B., and Reed, J.C. (1999). J. Biol. Chem. 274, 21932–21936. Scorrano, L., Ashiya, M., Buttle, K., Weiler, S., Oakes, S.A., Mannella, C.A., and Korsmeyer, S.J. (2002). Dev. Cell 2, 55–67.

Green, D.R., and Reed, J.C. (1998). Science 281, 1309–1312.

Vander Heiden, M.G., and Thompson, C.B. (1999). Nat. Cell Biol. 1, E209–E216.

Korsmeyer, S.J., Wei, M.C., Saito, M., Weiler, S., Oh, K.J., and Schlesinger, P.H. (2000). Cell Death Differ. 7, 1166–1173.

von Ahsen, O., Renken, C., Perkins, G., Kluck, R.M., Bossy-Wetzel, E., and Newmeyer, D.D. (2000). J. Cell Biol. 150, 1027–1036.

Caught in the Act: How ATP Binding Triggers Cooperative Conformational Changes in a Molecular Machine

mation. This allosteric effect requires communication between the two rings of the chaperonin. These intraand inter-ring allosteric effects are crucial to the mechanism of GroEL: Cooperative binding of ATP to a GroEL ring promotes subsequent binding of GroES to the same ring. When a substrate protein is present in the ring, ATP binding triggers its release into the encapsulated space underneath GroES, where productive folding takes place. At the same time, binding of ATP triggers events in the opposite ring—release of GroES, encapsulated substrate, and ADP. Subsequently, ATP hydrolysis in the folding-active “cis” ring cycles the chaperonin so that the opposite ring can accept ligands and become folding active. This set of coupled conformational changes creates a “two-stroke engine,” with asymmetry between the two rings as a central feature, and relies on the influence of ATP binding both intra-ring and interring. Examination of available crystal and cryo-EM structures points to large movements of the GroEL apical domains; these domains are circa 100 A˚ from the sites of ATP binding on the opposing ring, and therefore communication of conformational signals occurs over long distances. GroEL in the absence of ligand is postulated to assume a TT allosteric state (subunits in both rings in the tense form, according to MWC nomenclature), using a “nested allosteric” model (Horovitz et al., 2001). ATP binding initially would convert the chaperonin to an RT state (subunits in the ATP-bound ring in the relaxed form) and then, at higher concentrations, to an RR state. ATP binding is thus expected to disrupt interactions that hold an empty ring in a T state, promoting a cooperative conformational rearrangement to an R state. Upon binding of multiple ATPs (Hill coefficient 2.5), the overall GroEL molecule should then be RT. It has heretofore been impossible to capture an image of this state and thus to understand how ATP binding initiates the cooperative intra-ring conformational change. The clever strategy of the new study is one that been successful in the past—use of a mutant GroEL (D398A, in this case) that is defective in ATP hydrolysis. This mutant binds ATP in a normal manner, sufficient to support substrate and GroES binding and productive folding, arguing that it is not structurally perturbed from the native chaperonin. Strikingly, cryo-EM images of GroEL(D398)-ATP indeed show distinct asymmetry between the two rings,

A paper recently published in Cell describes ATP-triggered conformational changes in the GroEL folding machine deciphered by use of cryo-electron microscopy, molecular engineering, and X-ray crystallographic data. Mechanistically crucial allosteric effects of ATP binding arise from rearrangement of interdomain electrostatic contacts. In this era of molecular machines (Alberts, 1998), surprisingly few examples have been examined in adequate detail to understand their machine-like workings. The source of energy to drive most molecular machines derives from the nucleotide triphosphates ATP or GTP, yet we lack an understanding of the structural details by which this energy is converted into functional molecular rearrangements. In many cases, binding of ATP or GTP allosterically triggers domain movements at distances greater than 100 A˚ (Schnitzer, 2001). A major challenge is to relate the highly detailed information we obtain from static X-ray crystal structures to the dynamic steps involved in nucleotide-triggered domain rearrangements. In the December 28, 2001 issue of Cell, the Saibil and Horwich laboratories (Ranson et al., 2001) describe synergistic use of cryo-electron microscopy, molecular engineering, and X-ray crystallographic data to decipher intimate molecular details of the response of GroEL to binding of ATP. Their work provides tantalizing structural hints about how ATP binding triggers the cycle of GroEL-facilitated protein folding. ATP binding has two distinct allosteric effects on GroEL function, as shown by extensive work from many laboratories (for reviews, see Sigler et al., 1998; Horovitz et al., 2001; Thirumalai and Lorimer, 2001). First, binding of ATP is positively cooperative within one ring of the two-ring tetradecameric structure. This intra-ring cooperativity depends upon subunit-subunit communication, which must be modulated by nucleotide binding. Secondly, binding of ATP to one ring shows negative cooperativity with the other ring, thus favoring an asymmetric overall confor-