HTRA

Maintenance of a protein in a properly folded state is important for many of its functions. Cells therefore contain molecules, such as chaperones and proteases, to refold or rid themselves of misfolded and aggregated proteins. Chaperones recognize and bind proteins in nonnative states by interacting with their surface-exposed hydrophobic groups, holding the proteins in such a way that they may refold. Proteins that cannot be recovered are targeted for proteolytic destruction. Although chaperones and proteases carry out antagonistic functions, the substrates that they bind and act on are similar, and it is likely that they are coordinately regulated. For example, the bacterial proteins ClpP and ClpA (or X) form heterooligomeric complexes containing both protease (ClpP) and chaperone (ClpA or X) functions . Another remarkable example is the HtrA (high-temperature requirement) family of stress-induced proteins, where both activities exist within a single protein and the switch from protease to chaperone is regulated in a temperature-dependent manner.

In multicellular organisms, programmed cell death (apoptosis) is used to remove excess, damaged, or infected cells. The key effector proteases of apoptosis are caspases which remain in a latent form in healthy cells. While caspase activation can be regulated by members of the Bcl-2 family, once activated, caspases can be controlled by binding directly to members of the inhibitor of apoptosis (IAP) family of proteins. The IAPs can themselves be antagonized by proapoptotic proteins that bind to them, displacing the active caspases. For example, upon receipt of an apoptotic signal, the mitochondrial protein DIABLO/Smac is released into the cytosol where it binds to XIAP, thereby displacing processed caspases 9 and 3. DIABLO contains a conserved N-terminal motif (AVPI), similar to that found in Drosophila IAP antagonists, which is required for binding to IAPs. Recently, screens for other regulators of IAPs identified another mitochondrial protein, HtrA2, a mammalian protein that bears the serine protease and PDZ domain of its bacterial counterpart [3, 4, 5, 6 and 7]. Upon release from the mitochondrial intermembrane space, HtrA2 interacts with XIAP in a similar way to DIABLO. However, for maximal induction of apoptosis, overexpressed HtrA2 requires both its protease activity and its AVP amino terminus.

So how can HtrA act as a protease, chaperone, and regulator of apoptosis? Two recent papers describing the crystal structures of the bacterial periplasmic protein, DegP [8] and the mammalian homolog, HtrA2 [9] go some way toward elucidating the mechanism by which they might function. As predicted, each monomer of the E. coli HtrA, DegP, consists of three distinct domains, an N-terminal protease domain and two C-terminal PDZ domains. Three monomers come together, mediated by extensive contacts between the protease domains to form a trimeric ring, and this then forms a hexameric structure by staggered association of two trimers (see Figure 1, panel A). Two distinct hexameric molecules of DegP are apparent in the asymmetric unit, one in an “open form” and the other in a “closed form.” The primary contacts between the trimers are three pillars that are formed by interaction of the two N-terminal β strands from opposing monomers (see Figure, panel A). In the closed form shown in the figure, an additional set of interactions between the trimers occurs via the PDZ domains. However, these interactions involve flexible parts of the structure, and the hexamer is best described as a “dimer of trimers.”



















The molecules were aligned on the protease domain. The left and right images are related by a 90° rotation about the horizontal axis for both molecules.

(A) Structure of the “closed” DegP hexamer (Protein Data Bank code 1KY9). One trimer (orange, yellow, and gray) is shown as a surface representation while in the second trimer, two monomers (green and blue) are shown as a surface and the third as a ribbon. The protease domain (red), PDZ1 domain (purple), and PDZ2 (light pink) are colored separately; the catalytic serine is obscured in these views. The pillar is labeled P and the N and C termini are labeled N and C, respectively.

(B) Structure of the HtrA2 trimer (Protein Data Bank code 1LCY). Two monomers are shown as a surface representation. The third monomer is shown as a ribbon. The protease domain (red) and the PDZ domain (light pink) are colored separately; the site of the catalytic serine is shown (S) and the N and C termini are labeled accordingly.


In the structure of DegP solved by Krojer et al. [8], the critical serine in the catalytic triad was mutated to alanine in order to prevent autoproteolysis, and the structure was solved at 4°C to lock it in the chaperone conformation. As a result, although the general location of the active site is similar to that observed for other serine proteases, the catalytic triad is in an inactive conformation and access to the active site is precluded by a loop from an opposing monomer. A large conformational change is required before substrates can access the active site. This conformational change may be regulated in part by the PDZ domains. The PDZ domains are highly flexible and appear to act as gatekeepers of the catalytic site. In the chaperone conformation solved here, it is expected that partially unfolded substrates will be fed into the central cavity through lateral pores that occur in the open conformation. Once inside, substrates will interact with hydrophobic residues in the protease domain. Unlike other chaperones, ATP does not drive binding and release, and the mechanism by which this cycle is regulated in DegP is unknown. Perhaps rigid body movement of the PDZ domains is involved.

Shi and his colleagues have solved the crystal structures of a number of important proteins involved in apoptosis in recent years. In the latest paper, by Li et al. [9], they now report the structure of mature HtrA2, the mammalian counterpart of DegP. As predicted, this structure is homologous to that of bacterial HtrAs; in particular, the protease domain is highly conserved and the catalytic triad has a similar position between the two lobes of the protease domain (see Figure, panel B). Furthermore, access to the catalytic site is blocked by the PDZ domain and like DegP, the structure solved here is inactive. The regions of HtrA2 involved in trimer formation also adopt a very similar conformation to that seen in DegP, and the trimeric structure is conserved (see Figure, panel B). However, HtrA2 has several deletions when compared to DegP; for example, the N-terminal β strands that form the pillars in DegP are truncated and the first PDZ domain has been deleted. As a consequence of these deletions, HtrA2 only assembles into trimers and does not form hexamers. What is the consequence of no longer being a hexamer? The buried nature of the protease active site in DegP is likely to be important for its duality of function. Does loss of an enclosed cage mean that HtrA2 can no longer function as a chaperone? This is an important question that will require further studies. Other features of the HtrA2 structure are more revealing. For instance, the N terminus of HtrA2 is accessible from the top of the trimer, and if HtrA2 were released from the mitochondria it would be free to interact with IAPs in a manner similar to that seen for DIABLO [10], although it is uncertain whether all HtrA2 monomers would interact with IAPs. The structure of the PDZ domain in HtrA2 is also remarkable because, like bacterial PDZ domains, it is circularly permuted relative to mammalian PDZ domains. This likely reflects the bacterial ancestry of HtrA2.

Do PDZ domains hold the key to regulating HtrA proteases? PDZ domains are important in signal transduction and cellular organization. They bind an array of target proteins through small C-terminal motifs and play roles in the transport, localization, and assembly of these proteins [11]. What is their role in the HtrA proteases? In the cases of DegP and HtrA2, their role is probably not in the assembly of the oligomer, but in the recognition of short motifs for delivery to the proteolytic or chaperone activities of the protease domain. They also appear to act as modulators of activity. Deletion of the PDZ domains from DegP removed the catalytic activity but not the chaperone activity [2], whereas in HtrA2 deletion of the PDZ domain activated the protease [9]. To further understand the role of these domains, structures of complexes between HtrA proteins and substrates are required.

In conclusion, the crystal structures of DegP and HtrA2 have elegantly revealed the structural differences and similarities between these two proteins. However, many questions about their function remain unresolved. What is the function of HtrA2 in mitochondria? Does it perform the same role as DegP in E. coli? Is HtrA2 important for apoptotic function in vivo? It will be interesting to see whether HtrA2 knockout animals have a phenotype that resembles a mitochondrialopathy, suggesting that it is in the mitochondria that it has its primary role, or whether their phenotype resembles those of apaf-1 or caspase 9 knockouts, suggesting its primary role is to regulate cell death. HtrA2 can clearly interact with IAPs but the significance of this interaction is uncertain, as the protease activity is also required to fully induce apoptosis. Identification of the targets of the protease might provide clues as to the importance of HtrA2. Questions about the regulation of DegP also remain unresolved, although the structures allow hypotheses to be proposed and tested.

Serine protease HTRA2, mitochondrial is an enzyme that in humans is encoded by the HTRA2 gene. This gene encodes a serine protease. HtrA2, also known as Omi, is a mitochondrially-located serine protease. HtrA2 can be released from the mitochondria during apoptosis and uses its four most N-terminal amino acids to mimic a caspase and be recruited by IAP caspase inhibitors such as XIAP and CIAP1/2. Once bound, the serine protease cleaves the IAP, reducing the cell's inhibition to caspase activation. Additionally, HtrA2 has a PDZ domain, though little is known about its ability to bind PDZ binding motif peptides. HtrA2 has recently been identified as a gene related to Parkinson's disease. Mutations in Htra2 have been found in patients suffering from Parkinson's disease. Additionally, mice lacking HtrA2 have a parkinsonian phenotype. This suggests that HtrA2 is linked to Parkinson's disease progression in humans and mice.

HtrA2 shows similarities with DegS, a bacterial protease present in the periplasm of gram-negative bacteria. Structurally, HtrA2 is a trimeric molecule with central protease domains and carboxy-terminal PDZ domains.