Created by Ginny Kim

   Apaf-1 (pdb ID: 1z6t) is a 14 kDa protein found in Homo sapiens with an isoelectric point of 5.95 (1). The secondary structure of Apaf-1 consists of 7% β-sheets and 56% α-helices. Inactive Apaf-1 has one associated ligand, the prosthetic group adenosine diphosphate  (ADP) (3,4,7). It is controversial whether ADP must be exchanged for ATP and/or hydrolyzed before Apaf-1 can form the apoptosome.  (3,4,7,8).

   Apaf-1 contains four identicalsubunits. Each subunit contains: an N-terminal caspase recruitment domain (CARD), a central nucleotide-oligomerization domain (NOD), 15 WD40 repeats at the C-terminal , a three layered α/β domain, and a winged helix domain (WHD) (3,4,7,8).  The first 589 of 1249 amino acid residues are part of CARD and NOD. The remaining 660 amino acids form two β-propellers (WD1 and WD2) (7). NOD can be subdivided into the nucleotide-binding domain (NBD), the helical domain 1 (HD1) and helical domain II (HD2). These domains stack against each other through inter-domain interactions (4). CARD is the binding module for the prodomain of caspase-9 (procaspase-9). NOD has the nucleotide binding site and helps in oligomerization in the apoptosome (7). The WD40 repeats, composed of β-propellers, are the cytochrome c binding site (4,7).

   In 2005, Riedl et al. used the DALI Server to find structural homologues of Apaf-1 (4). The two most similar structures they found were D2-hexamerization domain of N-ethylmaleimide-sensitive fusion protein (NSF) and p97 (4). NSF (pdb ID: 1D2N) is an important ATPase needed for intracellular vesicle fusion in Cricetulus griseus, the Chinese hamster. Compared to Apaf-1, NSF has a DALI  z-score of 9 and a Blast E value of .097 (6,8). This high z-score means that the tertiary structures are similar, but the high E value means that the sequence homology is low. The associated ligands of NSF are phosphoaminophosphonic acid-adenylate ester, glycerol, and magnesium ion. The protein p97 (pdb ID: 1E32-A) is from Mus musculus, the house mouse. The associated ligand of p97 is adenosine triphosphate. Compared to Apaf-1, it’s DALI z-score is 15.9 and a Blast E value is .58 (6,8). Once again, the high z-score means that the tertiary structures are similar, but the high E score means that the sequence homology is low.

   P97, NSF, and Apaf-1 are all part of the AAA+ (ATPases associated with various cellular activities) superfamily of ATPases (4,8). Specifically, Apaf-1 is a representative member of the NOD family of proteins, which belongs to the STAND (Signal Transduction ATPases with Numerous Domains) clade of AAA+ ATPases (4,7). All AAA+ ATPases have a conserved region of α/β folds and HD1. The characteristic feature of this family is a short helical domain following the α/β fold. Apaf-1 is different from other AAA+ ATPases because its WHD coordinates ADP with His 438and Ser 422 giving two hydrogen bonds. His 438 replaces the sensor II of AAA+ATPases and coordinates the β-phosphate group and thus directly links the WHD to the bound nucleotide. These interactions bury ADP deeply within the protein distinguishing Apaf-1 from the rest of the AAA+ ATPases

   Since caspase cleavage is usually irreversible, Apaf-1 is synthesized as a zymogen. As an autoinhibited monomer, Apaf-1 cannot activate caspase-9 (7,4). Inter-domain interactions keep Apaf-1 in a “closed” inactive conformation. Apaf-1 contains ADP and has dATPase activity (4,7). ADP is mostly bound by residues from the NBD-HD1subdomains.  WHD contributes one hydrogen bond between H438 and an oxygen atom of β-phosphate (7). Thebinding pocket is in the inner face of the NBD-HD1 subdomains. ADP may be an organization center that brings CARD, α/β fold, HDI and WHD together and locks Apaf-1 into a closed conformation.  Because of these interactions, ADP is deeply buried and CARD blocks the narrow channel to the solvent (4).

   In Apaf-1, eight direct hydrogen bonds specifically bind ADP.  Two hydrogen bonds are to the adenine base and six are to the phosphate group. In the adenine base, the N1 and N6 are coordinated by the main-chain amide and carbonyl groups of Val 127respectively. The α-phosphate is coordinated by one hydrogen bond from the amide groups of Val 162. The β phosphate is bonded by five hydrogen bonds from: the amide groups of Gly 159, Lys 160, and Ser 161;the side-chain amino group of Lys 160; and the imidazole group of His 438. Gly 159, Lys160, Ser 161 and Val 162 come from the α/β fold; and His 438 comes from the WHD (4).

   Additionally, a few water molecules help mediate hydrogen bonds that keep ADP in place.  The side chain of Arg 129and the carbonyl of Gly 159 make water-mediated hydrogen bonds with the N7 atom of the adenine base. The carbonyl group of Val 125 makes a water-mediated hydrogen bond to adenine base. Furthermore, the carbonyl group of Ser 422 makes a water-mediated hydrogen bond to ribose (4). Some residues stabilize adenine and ribose through van der Waals forces. From the α/β fold, the van der Waals contributing residues are Pro 123, Phe 126, Val 127, Arg 129, Gly 159, and Val 162. From helical domain I (HD1), the van der Waals contributing residues are Ile 294, Pro 321, Leu 322, and Ser 325 (4).

   The key conformational change that primes Apaf-1 for oligomerization with cytochrome c to form the apoptosome is the rotation of the NBD-HD1 subdomain (7, 8,10). WD40 β-propellers are on top of the NOD subdomains HD2 and NBD. β-propellers are a type of β-sheet architecture that  has four to eight blade-shaped β-sheets arranged in a propeller orientation around a central axis. WD2 is above HD2, and WD1 contacts HD2 and NBD. The NBD-HD1 subdomain is held together by β-propeller WD1 and with WHD-HD2 subunit of the NOD respectively. When cytochrome c binds between β-propellers, propeller WD1 rotates out of its resting position and releases the NBD-HD1 subdomain making it free to rotate. β-propeller WD1, NBD and HD1 move as rigid bodies relative to a stiff rod formed by WHD, HD2 and β-propeller WD2.  Its free rotation exposes the oligomerization area and moves CARD so that it can bind caspase-9 to start apoptosis.

   Whether or not hydrolysis of ADP or ATP plays a role in apoptosome formation is still unclear (9).  Riedl et al.’s research from 2005 suggests that hydrolysis may help form the apoptosome. In this study, they deleted the WD40 repeats and crystallized the remaining 586 amino-acid residues. CARD is believed to dock with the prodomain of caspase-9 (procaspase-9) and activate it. In their crystallization, CARD stacks against α/β fold and the WHD. In this formation, procaspase-9 cannot dock to CARD  because of significant steric hindrance between procaspase-9 and the α/β-fold. The researchers propose that ADP-bound Apaf-1 has a closed conformation and that exchange and/or hydrolysis of the nucleotide is necessary before procaspase-9 can dock (4).

   Research published by Reubold et al. in 2009 and this past August 2011 suggests that hydrolysis is not involved in apoptosome formation (3, 7). Their research suggests that the closed conformation of Apaf-1 contains an ADP but that it must be exchanged for an ATP before apoptosome formation.  Unlike Riedl et al., they crystalized the entire protein including the WD40 repeats. Importantly, the inclusion of the WD40 repeats changed the relative location of CARD and the other domains. The WD40 β-propellers were previously thought to cap CARD in autoinhibited Apaf-1 (4,7). Reubold et al.’s crystalline structures show that the β-propellers are actually at a maximum distance away from CARD and CARD is not held in place by a regulatory domain or steric hindrance.  Instead, charged interactions between CARD and NOD keep CARD in the inactive resting location (7). Reubold et al. suggest that rotation of the NBD-HD1 subdomain does not require chemical energy from nucleotide hydrolysis because the pull on the WD1 propeller created by the cytochrome c binding is enough to open Apaf-1 (7). They suggest that ADP must be exchanged with ATP for Apaf-1 to open; however, ATP hydrolysis does not occur because no energy is used. They speculate that when cytochrome c binds to ADP-bound Apaf-1, cytochrome c is not entirely bound by both β-propellers because the NBD-HD1 subdomain in its resting ADP-bound position hinders rotation of WD1. The electrostatic pull between cytochrome c and WD1 finishes rotations when ATP releases the NBD-HD1 subdomain.

   Reubold et al.’s research suggests that a salt bridge forms between the γ-phosphate group of Arg 265 and ATP. They believe this salt bridge makes rotation of the NBD-HD1 subdmain easier.  Arg 265 is at the end of the β-strand parallel to the β-strand preceding the P-loop, the side chain of Arg 265 sticks out into the interface between NBD-HD1 subdomain and WHD. Reubold et al. believe that Arg 265 stabilizes the NBD-HD1 subdomain throughout the activation process. Furthermore, Arg 265 is part of the sensor 1 motif of AAA+ ATPases and senses ATP’s presence by forming a salt bridge to the γ-phosphate group of Arg 265. Changes in local charge distribution in the guanidium group of Arg 265 might help the rotation of the NBD-HD1 subdmain (7).

   When cytochrome c binds to Apaf-1 to form the heptameric apoptosome, it forms a so-called “central hub” of seven Apaf-1 monomers. The NBD of seven Apaf-1 monomers form an inner ring. Seven arms formed by HD2 radiate from the center of the ring (9). There is a regulatory region at the end of each arm that has two β-propellers with cytochrome c in between them. The HD1 and WHD of Apaf-1 form an outer ring that encircles the NBD inner ring. HD2 and WHD of each of the subunits form an extended arm which supports tandem β-propellers in a V-shaped regulatory region. Additionally, CARDs are flexibly linked to their respective NBDs in ground state apoptosomes. This disk may trigger proximity-induced dimerization of procaspase-9 (8,9).

   The Apaf-1•cytochrome c complex (the apopotosome) can be thought of as a platform . The apoptosome looks similar to other rings formed by members of the AAA+ ATPase family, however there are two major differences. Firstly, the apoptosome has an inititator-specific motif (ISM) in the NBD. Secondly, WHD forms extended arms with tandem β-propellers. These differences may reflect the difference in function between AAA+ ATPase and the apoptosome. The apoptosome is a stable platform while most AAA+ ATPases do mechanical work (8).

   When the apopotosome binds with procaspase-9, it forms a holo-apoptosome. Apaf-1 and procaspase-9 CARDS interact to form a disk shaped structure above the central hub. The holo-apoptosome activates executioner procaspases like pc-3 and pc-7. The holo-apoptosome proteolyzes the procaspases and rearranges critical loops to form active sites. Procaspase-9 is an initiator caspase and a monomer in solution; executioner caspases like pc-7 are dimers in solution. By an unknown mechanism, the apoptosome activates procaspase-9 by dimerization (8,9).