Studies in cell lines depleted in known BH3-only Initiators have indicated that Effectors do not always require direct activation, suggesting possible alternative mechanisms of activation40,41. BAK autoactivation has recently been shown biochemically42, but its mechanistic basis has not been elucidated. Here we show that autoactivation involves binding in trans of the BH3 helix α2 of active BAK to the activation groove of dormant BAK. We determined the crystal structure of autoactivated BAK BH3-BAK complex to reveal the molecular recognition mechanism and conformational changes in the protein core that destabilize helix α1. Moreover, we unequivocally elucidate helix α1 destabilization in the crystal structure of directly activated BAK bound to a rationally-designed BH3 ligand, thus demonstrating a common basis for both processes. Furthermore, we find that direct activation cooperates with autoactivation in BAK-mediated poration and apoptotic response.
activation BIM 360 Glue 2011 activation
The BAK BH3 peptide forms a helix bound to the activation groove of BAK, similar to previous BH3-bound complexes of BAK (Fig. 1e)19,27. However, in marked contrast to those previous BAK complexes, autoactivated BAK exhibits rearrangements of a buried electrostatic network with helix α1 at the bottom of the activation groove. In particular, helix α1 contacts with the rest of the domain at the bottom of the activation groove are disrupted compared to apo BAK (Fig. 1f, Supplementary Fig. 2c, d and Supplementary Movie 1). In apo BAK helix α1 electrostatic network is stabilized by 2 hydrogen bonds between α1 R42 and α3 D90 and 2 hydrogen bonds between α1 E46 and α5 R137 (Fig. 1f and Supplementary Fig. 2d). Of the 10 autoactivated BAK here, seven feature 2 hydrogen bonds between α1 R42 and E46 and α2 N86; one has only 1 hydrogen bond between α1 R42 and α2 N86; and two (those with lowest b-factors) show no visible density for this network (Fig. 1f, Supplementary Fig. 2c, d and Supplementary Table 3). The destabilized conformation of the electrostatic network is unique to autoactivated BAK providing the structural basis of BAK triggering by BAK BH3. Comparison between autoactivated BAK and previous BH3-bound BAK structures will be presented later.
Altogether, our in vitro investigations support the mechanism of in trans BAK autoactivation by asymmetric BH3-in-groove dimerization as suggested by the new structure. BH3 and groove mutations impaired in autoactivation respond to direct activation albeit they are less efficient than WT BAK in permeabilizing liposomes.
Structural changes observed in autoactivated BAK prompted us to revisit and further elucidate the mechanism of direct activation, since previous structural studies of BH3-only activators in complex with BAK18,19,27,28 have not led to a robust unifying mechanism for BAK destabilization upon activation by BH3 ligands (see below Fig. 6, Supplementary Figs. 8, 9 and Supplementary Table 3). Previous studies were focused largely on molecular recognition between the BH3 ligand and BAK by probing the ligand, while suggesting that cavities in BAK induced by BH3 ligands are critical in activation but did not test this hypothesis.
Our functional and binding analyses suggest that: hydrophobic pockets (1) and (2) are refractory to binding large hydrophobic residues, hence shallow; all other pockets are malleable and can accommodate large hydrophobic residues; and binding to pockets (3) and (4) promotes BAK activation relative to WT BID BH3. We note that membrane permeabilization is more efficient than binding of activators (Fig. 4e), further strengthening the notion that activation of a minor fraction of total BAK induces robust membrane permeabilization (Fig. 1d and Supplementary Fig. 3d, e).
Based on activity and binding profiles of monosubstituted peptides (Fig. 4), we produced disubstituted M(3)W(5) BID-like BH3 peptide, which exhibits 30-fold higher binding affinity for BAK compared to WT BID BH3, judged by ITC and SPR (Fig. 5a, b and Supplementary Fig. 7a, b). In liposome permeabilization assays this peptide is similarly potent as WT BID BH3 in direct BAK activation according to AUC against peptide dose EC50 analysis (Fig. 5a, c and Supplementary Fig. 7c).
Alignments of BH3 ligand-activated and inactivated BAK complexes reveal that the most consistent changes occur within the buried electrostatic network that stabilizes helix α1 at the bottom of the groove (Fig. 6 and Supplementary Fig. 9). Compared to apo BAK this network is disrupted in activated complexes with BAK BH3 (7m5c), BID-like BH3 (2m5b, 7m5b), and BIM-like BH3 (5vwv, 5vww); re-stabilized in inactivated BAK complex with W(3)W(5) BID BH3 (7m5a); and reinforced via molecular glue stabilization in inactivated complexes with BIM BH3 (PDB 5VWY and 5VWZ; Supplementary Fig. 9). Overall, our analysis establishes the rules of engagement of the activation groove by BH3 ligands to activate and inactivate BAK.
We investigated the effects of mutations in residues involved in the electrostatic network stabilizing helix α1 observed in apo BAK. Mutations R42A, E46A, and R137A are hyperactive compared to WT in liposome permeabilization in the absence of BID BH3 (Fig. 6c and Supplementary Fig. 8b). Faster initial kinetics of liposome permeabilization in these mutants is likely caused by direct disruption of 2 out of 4 hydrogen bonds within the apo helix α1 electrostatic network. Upon activation by BID BH3, BAK WT and mutants permeabilize liposomes similarly (Fig. 6c and Supplementary Fig. 8b). The residues in the electrostatic network are not fully conserved in rodents, chicken, and pig, suggesting alternative mechanisms of helix α1 regulation in these species, but the BH3-ligand-induced helix α1 release mechanism observed for human BAK is predicted in other mammals (Fig. 6d and Supplementary Fig. 8c).
To initiate apoptosis BAK undergoes regulated unfolding adopting elusive membrane-associated conformations that porate mitochondria. Here we elucidate early changes in BAK conformation underlying direct activation by BID and autoactivation in trans (Fig. 7). Direct activation by BH3-only proteins has been investigated structurally yet the mechanism of BAK unfolding remains incompletely defined3,18,19,27,28. We show that autoactivation, a postulated but poorly characterized step in BAK activation, contributes substantially to apoptotic response. Autoactivation cooperates with direct activation to amplify signaling and lower the threshold of BAK required for mitochondrial poration (Supplementary Movie 2). Autoactivation is mechanistically similar to direct activation: binding of exposed BH3 of BAK or activators to the activation groove induces conformational changes in the electrostatic network at the bottom of this groove destabilizing helix α1 to initiate BAK unfolding (Fig. 7 and Supplementary Movie 1). While our reconstituted in vitro and cellular systems with BAK and its activator BID rely on increasing BAK concentration, which has not been reported as a general mechanism of apoptosis initiation, they offer the means of directly investigating BAK in the absence of other BCL-2 proteins (Fig. 7 and Supplementary Movie 2).
Direct BAK activation by BH3-only proteins cooperates with BAK autoactivation in trans to lower BAK threshold and to amplify the response leading to mitochondrial poration (Supplementary Movie 2). BAK autoactivation involves binding the exposed BH3 of active BAK to the activation groove of dormant BAK. Our structural analyses of autoactivated (7m5c) and directly activated (7m5b) BAK reveal BH3 ligand-induced conformational changes in the protein core that promote helix α1 destabilization as the mechanistic basis of BAK activation (inset cartoons). In activated BAK complexes helix α1 electrostatic network is destabilized as it rearranges within the N-bundle (helices α1-α2), which dissociates in the presence of membranes but not in solution. In contrast, in dormant, apo BAK (2ims) and an inactivated BAK complex bound to a rationally-designed BID-like BH3 peptide (7m5a), helix α1 is stabilized by the electrostatic network within the C-bundle (helices α3-α8), which prevents N-bundle dissociation at membranes (Supplementary Movie 1). How active BAK porates membranes is unclear but oligomerization of the BAK α2-α5 core dimers has been postulated. Mechanistic questions not addressed in our study are inset.
In cells with complex repertoires of BCL-2 proteins, BAK activation is also regulated through inhibition by the prosurvival BCL-2 Guardians, which act by sequestration of Initiators (MODE 1) or Effectors (MODE 2)25. Our model may explain the effective activation of dormant BAK upon derepression of MODE 2 (i.e., when active BAK is freed from complexes with prosurvival BCL-2 Guardians by BH3 mimetics); presumably this exposes BAK BH3 which can autoactivate in trans nearby dormant BAK. Autoactivation could explain Effector activation in HCT116 cells depleted in BH3-only Initiators when the prosurvival BCL-2 Guardians are antagonized, although it was speculated that Effectors are activated through a membrane permissive model (i.e., membranes activate Effectors)40,41. The autoactivation mechanism does not exclude this model. Our understanding of the dynamics of BAK interaction with membranes is incompletely understood, and therefore we cannot explain the initiating event leading to BAK autoactivation in vitro or in cells in the absence of BH3-only activators (Fig. 7). 2ff7e9595c
Comments