Structural insights into CED-3 activation

Abstract

In Caenorhabditis elegans (C. elegans), onset of programmed cell death is marked with the activation of CED-3, a process that requires assembly of the CED-4 apoptosome. Activated CED-3 forms a holoenzyme with the CED-4 apoptosome to cleave a wide range of substrates, leading to irreversible cell death. Despite decades of investigations, the underlying mechanism of CED-4-facilitated CED-3 activation remains elusive. Here, we report cryo-EM structures of the CED-4 apoptosome and three distinct CED-4/CED-3 complexes that mimic different activation stages for CED-3. In addition to the previously reported octamer in crystal structures, CED-4, alone or in complex with CED-3, exists in multiple oligomeric states. Supported by biochemical analyses, we show that the conserved CARD-CARD interaction promotes CED-3 activation, and initiation of programmed cell death is regulated by the dynamic organization of the CED-4 apoptosome.

Figure 1.. Cryo-EM structures of the CED-4 apoptosome.

(A) Four proteins—EGL-1, CED-9, CED-4, CED-3—act in a linear fashion to control the onset of PCD in C. elegans. (B) Different oligomerization states of the CED-4 apoptosome in high- and low-salt buffers. Shown here are representative chromatograms of the CED-4 apoptosome eluted from SEC pre-equilibrated in high- (blue) or low-(red) salt buffers. Peak fractions were loaded to SDS–PAGE and visualized via Coomassie blue staining. (C) Four cryo-EM maps of the CED-4 apoptosome, including a hexamer (grey), a heptamer (yellow), and two octamers (cyan and slate) were obtained. CED-4 CARDs are colored in orange and brown to indicate different layers. (D) Two perpendicular views of the overall structure of the hexameric CED-4 apoptosome. The two protomers CED-4a and CED-4b are colored slate and limon, respectively. (E) The CED-4 hexamer loses C4 symmetry. The hexamer cannot accommodate a fourth CED-4 dimer (grey). Potential steric clash with a fourth CED-4 dimer is labeled by the magenta oval. (F) CED-4 CARDs have two stacking modes in different oligomers. Group I stacking only exists in octamer 2. Group II stacking, which has not been observed previously, is found in three oligomeric states. Group II stacking is likely to be maintained by salt bridges between helices H6 and H4 (orange rectangle).

Figure S1.. Sample preparation of CED-4 and CED-4/CED-3 complexes for cryo-EM analyses.

(A) Complexes were prepared in vitro by incubating recombinantly expressed protein components followed by SEC. Gel-filtration results of the CED-4 apoptosome, the CED-4/CED-3_CARD complex, the CED-4/CED-3 catalytic complex, and the holoenzyme are shown in blue, green, magenta, and orange, respectively (left panel). Peak fractions were applied to SDS–PAGE and Coomassie blue staining (right panel). (B) Cryo-EM micrographs of the CED-4 apoptosome in high- or low-salt buffers. Related to Fig 1B.

Figure S2.. Data processing for the CED-4 apoptosome.

A flowchart of cryo-EM data processing for the CED-4 apoptosome. Please refer to methods for details.

Figure S3.. FSC curves of the CED-4 apoptosome and CED-4/CED-3 complexes.

The average resolutions of the reconstructions were determined on the basis of the FSC 0.143 value, including the hexameric CED-4 apoptosome, the heptameric CED-4/CED-3 catalytic complex, the octameric CED-4/CED-3_CARD complex, and the octameric holoenzyme.

Figure S4.. Group I stacking of CED-4 CARDs is more stable than group II stacking.

(A) Group I and group II stacking of CED-4 CARDs are shown in left and right panels, respectively. Unlike group II stacking, group I stacking involves multiple interfaces among one CED-4b CARD and two CED-4a CARDs. (B) CARD rings formed by group II stacking are flexible in cryo-EM maps. Resolution at these regions is relatively low.

Figure S5.. Data processing for the CED-4/CED-3 catalytic complex.

A flowchart of cryo-EM data processing for the CED-4/CED-3 catalytic complex. Please refer to methods for details.

Figure 2.. Cryo-EM structure of a heptameric CED-4 in the presence of the catalytic domain of CED-3.

(A) Representative 2D class averages of the complex. A globular density in the hutch is indicated by a red arrow. (B) Two perpendicular views of the overall structure of a heptameric CED-4/CED-3 catalytic complex. CED-4a and CED-4b are colored slate and limon, respectively. CED-4c, whose CARD is invisible, is colored pink. The catalytic domain of CED-3 cannot be modelled owing to the low resolution. (C) The heptameric CED-4 NOD and Apaf-1 NOD are structurally conserved. NODs from the heptameric CED-4 (slate) and the Apaf-1 apoptosome (orange, PDB code: 3JBT) are aligned. (D) The oligomerization interfaces of heptameric and octameric (grey) CED-4 NODs are similar.

Figure 3.. Cryo-EM structure of an octameric holoenzyme that was assembled by CED-4 and the CARD and catalytic domain of CED-3.

(A) Representative 2D class averages of the holoenzyme. The globular density, which may correspond to the catalytic domain of CED-3, in the hutch is indicated by red arrows. (B) Two perpendicular views of the overall structure of the holoenzyme. CED-4 NODs are colored light grey. Upper and lower CED-4 CARD rings are colored limon and slate, respectively. CED-3 CARD ring is in light orange. (C) Stacking of the CED-4 CARD and CED-3 CARD rings. The CARD–CARD interface is depicted in detail in the inset. Side chains of putative interacting residues are labeled.

Figure S6.. Data processing for the CED-4/CED-3 CARD complex.

A flowchart of cryo-EM data processing for the CED-4/CED-3_CARD complex. Please refer to methods for details.

Figure S7.. Data processing for the holoenzyme.

A flowchart of cryo-EM data processing for the holoenzyme. Please refer to methods for details.

Figure 4.. Biochemical validation of key residues that mediate the CARD–CARD interaction between CED-4 and CED-3.

(A) CED-3 mutations that may compromise the interface disrupted formation of the CED-4/CED-3_CARD complex. Indicated CED-3 CARDs, WT or mutants, were incubated with the CED-4 apoptosome. After SEC, indicated fractions were collected for SDS–PAGE and Coomassie blue staining. (B) CARD–CARD interaction is required for CED-3 activation. Biotin-labeled CED-3 variants were in vitro translated for half an hour before CED-4 or buffer was added. After another half an hour, the mixture was applied to SDS–PAGE and visualized through Western blotting against biotin. CED-4 was used at a concentration of 100 nM (monomer). (C) Conserved CARD–CARD interface. Shown here is the sequence alignment of CARDs from human caspase-9, CED-3 from C. elegans, and Dronc from Drosophila. Conserved residues are colored red. Key residues that mediate the CARD–CARD interactions between the initiator caspases and the apoptosomes are shaded brown and pointed by red arrows.

Figure 5.. The CED-4 apoptosome regulates both CED-3 activation and its protease activity.

(A) Dual modulation of CED-3 activation by CED-4 in a concentration-dependent manner. WT CED-3 was generated by in vitro translation. The CED-4 apoptosome used in the assay was 1,000, 250, 125, 62, 31, 16 nM (monomer). (B) The CED-4 apoptosome–activated CED-3 acquires enhanced protease activity. Biotin-labeled CED-3 R7D was used as the substrate in the cleavage assay. All CED-3 variants were generated by in vitro translation.

Figure 6.. A working model for the regulation of CED-3 activation and activity.

Most of the CED-4 apoptosome molecules are pre-mature hexamers, which may represent a potential regulatory step. CED-3 activation relies on both CED-4_CARD/CED-3_ CARD and the CED-4/CED-3 catalytic domain interaction. CED-3 zymogens bind to the octameric CED-4, within which it is activated and exhibits enhanced activity. Activated CED-3 cleaves other CED-3 zymogens, separating the CARD domain and the catalytic domain. The released catalytic domain and CARD form a heptameric or octameric complex with CED-4 and the protease activity is elevated.