INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS AND DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Proteolytic processes in chloroplasts (C) and mitochondria (M) have been a subject of interest for many years, but until recently the identity of the proteases involved has remained obscure (for review, see Adam, 1996; Herrmann, 1996; Andersson and Aro, 1997). Genetic, biochemical, and molecular approaches taken in recent years has led to the identification and characterization of several organellar proteases, all of which are homologs of bacterial proteases best characterized in Escherichia coli (for review, see Clarke, 1999; Adam 2000a, 2000b). Four major families of proteases have been discovered and characterized in Arabidopsis to varying degrees: Clp protease in Cs (Sokolenko et al., 1998; Nakabayashi et al., 1999 and refs. therein) and M (Halperin et al., 2001b), FtsH (Lindahl et al., 1996; Chen et al., 2000) and DegP proteases (Itzhaki et al., 1998; Haussühl et al., 2001) in Cs, and Lon protease in M (Sarria et al., 1998).

Clp proteases comprise a large family of ATP-dependent, Ser-type proteases characterized by separation of the two essential functions to two different polypeptides: a small subunit containing the proteolytic active site (ClpP), and a larger regulatory ATPase subunit (for review, see Gottesman, 1996). Lon protease is the first ATP-dependent protease to be described in E. coli. It is also a Ser-type protease, but unlike Clp proteases, its catalytic and ATPase domains reside within a single polypeptide (for review, see Gottesman, 1996). In contrast to Clp and Lon, FtsH is the only essential ATP-dependent protease in E. coli. It is a membrane-bound metalloprotease that is characterized by an approximately 200-amino acid ATPase domain (AAA motif) and a zinc-binding domain, His-Asp-X-X-His, that serves the catalytic function (for review, see Gottesman, 1996). Like for the Lon protease, the catalytic and ATPase functions of FtsH reside on the same polypeptide. E. coli contains another family of Ser-type proteases known as the Deg proteases of which DegP (or HtrA) was the first described and is currently the best characterized member (Pallen and Wren, 1997). It is a peripheral membrane protein localized to the periplasmic side of the inner membrane (Skorko-Glonek et al., 1997). The other members of the family are DegQ and DegS, also known as HhoA and HhoB, respectively. All three family members share the putative catalytic triad of Ser-type proteases (His-Asp-Ser), and one or two PDZ-like domains at their carboxy-termini (Pallen and Wren, 1997; Ponting, 1997) implicated in protein-protein interactions.

Rapidly accumulating DNA sequences, primarily from Arabidopsis, have revealed that unlike in E. coli, plant genomes code for multiple isomers from each of the major protease families. As more member proteins are discovered and studied, the risk increases of conflicting and misleading nomenclature appearing in the literaturea problem that has already materialized for Clp proteases with similar names being given to different gene products (Clarke, 1999; Nakabayashi et al., 1999). To avoid such confusion in the future and facilitate scientific communication we have searched the Arabidopsis sequence databases and compiled all available data on the different plant organellar proteases and their isomers. The intracellular location and putative processing sites for each isomeric form was predicted using dedicated software, and we propose a standardized nomenclature for future reference to these proteases.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS AND DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Clp Proteases

Six different ClpP genes encoding the proteolytic subunit of Clp protease are found in Arabidopsis (Table I). All ClpP proteins share a common motif, the highly conserved catalytic triad of Ser-type proteases comprising of Ser-His-Asp residues. One of them, ClpP1, is found in the C genome of all higher plants and most eukaryotic algae, and is the only protease encoded within the plastid genome. The remaining five ClpP isomers are nuclear-encoded, on chromosomes I or V. ClpP4 through 6 are targeted to the C stroma (Sokolenko et al., 1998; K. Nakabayashi, unpublished data), and ClpP2 is targeted to M (Halperin et al., 2001b). It should be noted that the current designation of ClpP isomers is somewhat different from previous reports, where different names were given to the same protein. Thus, ClpP1 was previously known as pClpP (Clarke, 1999; Nakabayashi et al., 1999); ClpP3 was previously designated nClpP4 (Clarke, 1999), and vice versa; ClpP5 was known as nClpP1 (Nakabayashi et al., 1999); and ClpP6 was known as nClpP1 (Clarke, 1999).

The Arabidopsis genome contains four other sequences that show homology to ClpP proteins. However, these proteins do not contain the aforementioned catalytic triad, and therefore cannot be considered as bona fide ClpP isomers. Because some of them were previously designated as ClpP, we chose to present them here, although their function and relevance to Clp proteases remains unclear. For their classification we have used the earlier nomenclature of ClpR (Porankiewicz et al., 1999). The previously designated nClpP5 and nClpP2 (Nakabayashi et al., 1999) are now named ClpR1 and ClpR2, respectively. In vitro translation import studies have shown ClpR1 to reside in the C stroma (Nakabayashi et al., 1999), as is also predicted for the other ClpR isomers.

The proteolytic subunit of the Clp protease is dependent on a cognate ATPase for its proteolytic activity. ClpC, a homolog of the E. coli ClpA containing two ATP-binding sites, is found in the C stroma of many different plant species (Moore and Keegstra, 1993; Shanklin et al., 1995; Halperin and Adam, 1996; Nakabayashi et al., 1999). Two near-identical ClpC isomers exist in Arabidopsis (Table II). ClpC is found associated with ClpP in the stroma (Desimone et al., 1997; Sokolenko et al., 1998) in an ATP-dependent manner (Halperin et al., 2001a). Another ATPase containing two ATP-binding domains, ClpD, is also found in the stroma (Kiyosue et al., 1993; Nakashima et al., 1997). Although closely related, ClpD differs from ClpC1 and -2 by specific signature sequences (Schrimer et al., 1996) and by its differential expression characteristics (Nakabayashi et al., 1999). ClpD has previously been referred to as ERD1 (Kiyosue et al., 1993; Nakashima et al., 1997; Nakabayashi et al., 1999) and SAG15 (Lohman et al., 1994).

In E. coli, ClpP can associate with another member of the Clp-ATPase family, ClpX, which contains only a single ATP-binding site as opposed to the two in ClpA, ClpC, and ClpD. As for ClpC, Arabidopsis has two distinct forms of ClpX encoded in chromosome V (Table II). ClpX1 is absent from Cs and is instead localized in M (Halperin et al., 2001b). Similar targeting to M is also predicted for the second ClpX isomer. The localization of ClpX and ClpP2 to M suggests that an active Clp protease is found not only in Cs, but also in M.

Lon Protease

Three genes encoding Lon proteases are found in Arabidopsis (Table III). Lon1 was localized to M (Sarria et al., 1998), matching the mitochondrial localization of a Lon homolog in yeast (Suzuki et al., 1994; van Dyck et al., 1994). A second isomer, Lon2, deduced from a genomic sequence, is predicted by the Predotar program to be chloroplastic, although the ChloroP program fails to detect a corresponding transit peptide. This situation is reversed for a third isomer, Lon3. ChloroP suggests Lon3 is chloroplastic, whereas Predotar predicts it is not chloroplastic or mitochondrial. Immunoblot analysis with an antibody specific to Lon supports the presence of at least one Lon protease in both plant organelles (O. Ostersetzer and Z. Adam, unpublished data), although this observation needs to be further substantiated.

FtsH Protease

Arabidopsis contains nine isomers of the FtsH protease (Table IV). Two of these isomers, FtsH1 and -2 were found in Cs as integral proteins within the thylakoid membrane, with their ATP-binding domain and catalytic zinc-binding site facing the stroma (Lindahl et al., 1996; Chen et al., 2000). Mutations in FtsH2 were linked with a variegated phenotype, having green- and white-sectored leaves (Chen et al., 2000). Reported biochemical activities of FtsH protease include degradation of unassembled or oxidatively damaged membrane proteins (Ostersetzer and Adam, 1997; Lindahl et al., 2000). Four additional homologs, FtsH5 through 8, are predicted as chloroplastic, whereas FtsH3 is predicted to localize in M. The location of FtsH4 cannot be predicted with available programs, although a phylogenetic analysis (O. Ostersetzer and Z. Adam, unpublished data) shows that its closest homolog is one of the three mitochondrial homologs found in yeast (Leonhard et al., 1996). FtsH9 is a partial cDNA whose localization cannot be predicted from the available sequence, although this may be resolved by the eventual identification of its genomic sequence. It should be noted that additional FtsH-like sequences can be found among Arabidopsis genomic sequences, but the lack of a catalytic zinc-binding domain currently excludes them from our nomenclature.

DegP Protease

Thirteen different genes encoding proteins related to DegP, all containing the catalytic triad His-Asp-Ser, are present in the Arabidopsis genome (Table V). Although some of them are more similar to one E. coli protein than to the other (i.e. DegP, DegQ, or DegS), they differ in the number of PDZ domains. Because of this discrepancy we chose to designate all plant members of this protease family as DegP.

DegP1 and -2 are located in C thylakoid membranes (Itzhaki et al., 1998; Peltier et al., 2000; Haussühl et al., 2001). Other members of the family are likely to be expressed because they are represented among Arabidopsis ESTs, but their predicted locations vary from Cs (DegP5 and -8) to M (DegP3, -4, and -10¹²), nuclei (DegP9), cytoplasm (DegP7), and the endoplasmic reticulum or plasma membrane (DegP13).

Another predicted difference within the plant DegP family is the apparent lack of PDZ domains in some of the isomers (i.e. DegP5, -6, -8, -9, -11, and -13). As PDZ domains mediate not only protein-protein interaction, but also influence the catalytic properties of the enzyme, the presence or absence of these domains in the different isomers is an important feature. DegP1-4, -7, -10, and -12 are all predicted to contain a single PDZ domain at their carboxy termini. Thus, the absence of the PDZ domain might have implications on the function of these proteins. It is also interesting to note that E. coli DegP contains two such domains, whereas none of the plant isomers have more than one. However, the functional significance of this difference is not yet clear.

CONCLUDING REMARKS

Up until a few years ago the identity of chloroplastic and mitochondrial proteases was a mystery that defied resolution by existing experimental approaches. In recent years, however, the advent of broad-scale EST/genomic sequencing and functional genomics have revealed the identity and structural characteristics of several distinct families of proteases in higher plants. Also revealed have been the many different isomeric forms that exist in plants for each of these protease families. To date, our understanding of the different protease families remains in its infancy, especially in terms of their molecular structure, substrate specificity, expression patterns, regulation of activity, and physiological roles. Such characterizations are further complicated by the numerous isomers that exist for each of the protease families. For example, it remains unclear whether isomers of the same protease have overlapping activities, and thus a degree of redundancy, or instead have distinct properties such as different substrate specificity. As a consequence, the future challenge to elucidate the roles of proteases in the biogenesis and functions of Cs and M will require taking into consideration the multiple isomers of each.

To address the complexity of C and M proteases and facilitate scientific exchange, a standard nomenclature is urgently required such as the one proposed here derived from the analysis of Arabidopsis sequence data. Our nomenclature has been based on that used for the well-characterized E. coli proteases and has been adapted for the numerous predicted homologs in the nearly completed Arabidopsis genome. We propose that this nomenclature be used as a foundation for defining homologous proteases in other plant species, with extra numbers used if additional homologs than those in Arabidopsis are later discovered.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS AND DISCUSSION MATERIALS AND METHODS LITERATURE CITED

DNA Sequence Analysis

Four families of proteases were analyzed, three of which are ATP-dependent (Clp, Lon, and FtsH) and one is ATP-independent (DegP-like). Peptidases were excluded from this analysis due to their large number and the relative lack of functional information currently available. Identification of family members was done using the BLAST program (Altschul et al., 1997). Protein sequences of members of each family were compared with six-frame translations of DNA sequences available in GenBank (as of October 15, 2000). Although in some cases there were more Arabidopsis homologous sequences than we present here, specific selection criteria were applied for inclusion in each of the families, as detailed in the "Results and Discussion." Protein sequences were further analyzed. Prediction of intracellular locations were performed using the programs TargetP version 1.01 (Emanuelsson et al., 2000) and Predotar version 0.5 (http://www.inra.fr/Internet/Produits/Predotar/). In cases where contradictory predictions arose from TargetP and Predotar, we used the prediction with the highest probability value. Transit peptide-processing sites for C proteins were predicted by ChloroP version 1.01 (Emanuelsson et al., 1999). Each protein entry includes accession number to the corresponding full-length cDNA clone (if available) or EST clone (if available), predicted sizes of the precursor, mature and transit proteins, chromosomal location with its accession number and protein id number, proven or predicted cellular location, and a reference to the relevant publications. For proteins deduced from genomic sequences that are not fully annotated, and hence do not have protein id number, the nucleotides that span the coding sequence are indicated.