Genomic analysis of the Hsp70 superfamily in Arabidopsis thaliana

Abstract

The Arabidopsis genome contains at least 18 genes encoding members of the 70-kilodalton heat shock protein (Hsp70) family, 14 in the DnaK subfamily and 4 in the Hsp110/SSE subfamily. While the Hsp70s are highly conserved, a phylogenetic analysis including all members of this family in Arabidopsis and in yeast indicates the homology of Hsp70s in the subgroups, such as those predicted to localize in the same subcellular compartment and those similar to the mammalian Hsp110 and Grp170. Gene structure and genome organization suggest duplication in the origin of some genes. The Arabidopsis hsp70s exhibit distinct expression profiles; representative genes of the subgroups are expressed at relatively high levels during specific developmental stages and under thermal stress.

INTRODUCTION

Originally identified as the most abundant proteins produced in response to an elevated temperature, the 70-kDa heat shock proteins (Hsp70s) are now known to be up-regulated during many forms of cellular stress and are essential during normal growth (for early references, see Lindquist and Craig 1988; Vierling 1991; Georgopoulos and Welch 1993; Nover and Scharf 1997). Hsp70s function as molecular chaperones, presumably by protecting proteins against aggregation, based on the property of binding to hydrophobic amino acid residues or surfaces that are exposed by proteins in nonnative states (for reviews, see Craig et al 1994; Boston et al 1996; Hartl 1996; Miernyk 1997; Agashe and Hartl 2000). Successive cycles of binding and release of substrate proteins are coupled to the intrinsic adenosine triphosphatase (ATPase) activity of Hsp70, which usually requires activation by cohort systems, such as the DnaJ-type molecular chaperones. Structurally, Hsp70 comprises an amino (N)–terminal, approximately 45-kDa ATPase domain (Chappell et al 1987; Flaherty et al 1990) and a carboxyl (C)–terminal, approximately 25-kDa peptide-binding domain (Zhu et al 1996); both are highly conserved. Sequence variation among Hsp70s occurs in the extreme N and C ends, where the information for subcellular localization and for intramolecular and intermolecular interactions resides (Gupta and Golding 1993; Boorstein et al 1994; Freeman et al 1995; Hartl 1996; Miernyk 1997). Hsp70s are ubiquitous in all eubacteria and eukaryotes and in a subset of the Archaea (Karlin and Brocchieri 1998; Macario et al 1999). Genomes of these organisms known to date encode multiple Hsp70s. In eukaryote cells, members of the multigene family are localized to distinct subcellular compartments: cytoplasm, plastids, mitochondria, and endoplasmic reticulum (ER). The various subcellular localization imply both functional specificity and phylogenetic divergence (for reviews, see Vierling 1991; Parsell and Lindquist 1993; Boorstein et al 1994).

Hsp70 has long been recognized as one of the most conserved protein families (Gupta and Golding 1993; Boorstein et al 1994). Recently, in eukaryotes, distinct sets of proteins have been identified that are larger and, although similar, more diverse than the conventionally grouped proteins related to DnaK, the Hsp70 in Escherichia coli (reviewed in Easton et al 2000). Included in those proteins are the orthologs of mammalian Hsp110 and yeast SSE proteins that are located in the cytoplasm (Fathallah et al 1993; Foltz et al 1993; Mukai et al 1993; Lee-Yoon et al 1995; Storozhenko et al 1996) and their counterparts in ER, the orthologs of the mammalian Grp170 (Chen et al 1996; Ikeda et al 1997). Based on similarities in structural and functional properties, these proteins are considered parts of the Hsp70 superfamily.

A thorough search of the genome sequence of Arabidopsis thaliana revealed 18 members in the Hsp70 superfamily, including 14 in the DnaK subfamily and 4 in the Hsp110/SSE subfamily. We found that most of these genes are expressed at a basal level during normal growth, whereas some members show enhanced expression under heat stress and at particular developmental stages. Data on genome organization and expression pattern are compiled herein to provide a basis for functional genomics investigations, such as addressing the specificity in regulation and function of members in this multigene family.

DnaK SUBFAMILY

Seven hsp70 genes of the dnaK type have been previously reported for Arabidopsis (see Table 1 for references). Using these sequences and keywords, we searched the genome database and identified 7 additional paralogous genes belonging to the DnaK subfamily. To minimize changes to the existing nomenclature, for ease in organizing the information, and based on the characteristics that are detailed below, we adopted the numbering system Athsp70-x for Arabidopsis hsp70 gene number x, as shown in Table 1.

Table 1.

 Arabidopsis hsp70s

The Athsp70 gene products are assigned to distinct subcellular locations based on the presence of consensus sequences that have been reported and also following the predictions of the PSORT program (Nakai and Kanehisa 1992; Nakai 2000) (Table 1). For examples, the AtHsp70s that contain GPKIEEVD at the extreme C-terminus are predicted to localize in cytoplasm, whereas those with specific targeting signals at the N-terminus are predicted to localize to the respective organelles (see Table 1 and online Fig 5). Most deduced proteins predicted to localize in the cytoplasm, AtHsp70-1∼AtHsp70-5, AtHsp70-14, and AtHsp70-15 also contain consensus nuclear localization signals (Dingswall and Laskey 1991; Rensing and Maier 1994) (Table 1). In Athsp70-13, the C-terminal end of the predicted protein does not contain an ER retention signal and the overall sequences, although highly similar to the other genes coding for the ER form AtHsp70s, show apparent divergence for the last 21 amino acids at the C-terminus (online Fig 5). Examining the sequences, we found that by insertion of a single nucleotide in the corresponding coding region, the C-terminus could contain a similar number of residues, including the presence of the ER retention signal, HDEL. Since the region is rich in repetitive nucleotides, it cannot be excluded that some sequences may have been misread. Therefore, we decided to temporarily assign Athsp70-13 to the ER group (Table 1). On the other hand, Athsp70-18 is most similar to Athsp70-1∼Athsp70-5 in the group of genes coding for the cytoplasmic form (online Fig 5). However, the nucleotide sequences diverge following the appearance of a stop codon such that the predicted protein is truncated at the junction between the structurally resolvable peptide-binding domain and the G/P-rich region (approximately 30 residues). Hence, it does not contain the consensus residues for the cytoplasmic Hsp70s in eukaryotes, EEVD, at the very C-terminal end (online Fig 5). Although we temporarily assigned this gene to the cytoplasmic group, since no expressed sequence tag (EST) was found for this gene and we could not detect its expression, Athsp70-18 might be a pseudogene.

The Hsp70 family in Saccharomyces cerevisiae has been well characterized (Boorstein et al 1994; Rassow et al 1997). We performed phylogenetic analysis of the family members from yeast and Arabidopsis together with DnaKs from E coli and Synechocystis sp. The resulting phylogenetic tree (Fig 1) shows the clustering of yeast and Arabidopsis Hsp70s belonging to the same subcellular location. In general, members of the DnaK and the Hsp110/SSE subfamilies are separated into 2 large groups. The cytoplasmic AtHsp70s and the yeast SSA are derived from the same branch on the phylogenetic tree. There are no obvious Arabidopsis homologs of the yeast SSB proteins. The mitochondrial and chloroplast Hsp70s appear to form a monophylogenetic group, with DnaK of E coli more related to the mitochondrial Hsp70s and the DnaKs of Synechocystis sp. forming a distinct branch together with the Arabidopsis plastid Hsp70s. These results are consistent with the prokaryotic origin of these organelles and in agreement with the notion that orthologous Hsp70s localized to distinct cellular compartments are more conserved evolutionarily than the paralogs in respective genomes (Vierling 1991; Boorstein et al 1994).

Fig 1.

 Phylogenetic tree of Hsp70s from the entire gene families in Arabidopsis and in yeast, with DnaKs from E coli and Synechocystis sp. Constructed with the complete deduced amino acid sequences of the 36 Hsp70s using the DARWIN program (Gonnet et al 1992). With the exception of the yeast mitochondrial SSH1 (*), the DnaK subfamily and the Hsp110/SSE subfamily members are clustered in separate regions of the tree as indicated by a dividing line on the right. The eukaryotic Hsp70s in the DnaK subfamily that are predicted to localize in the same subcellular compartments are clustered, as indicated by square brackets at the right. Accession numbers of the sequences analyzed are as follows: E coli DnaK, P04475; Synechocystis sp. strain PCC6803 DnaK1_Syn, Q55154; DnaK2_Syn, P22358; DnaK3_Syn, P73098; yeast KAR2, M25064; LHS1, S37895; SSA1, P10591; SSA2, P10592; SSA3, S36753; SSA4, B36590; SSB1, CAA31995; SSB2, S50721; SSC1, M27229; SSE1, P32589; SSE2, P32590; SSH1, S44545; ECM10, S50429; and PDR13, S46712. For the accession numbers of AtHsp70-1∼AtHsp70-18, see Table 1

As expected, sequences of Athsp70s are highly conserved. The amino acid sequences of the deduced proteins are very similar, particularly in the ATPase domain, and within the same subcellular groups (Fig 2, online Fig 5, online Table 3). Two cytoplasmic forms, AtHsp70-1 and AtHsp70-2, are more than 95% identical in amino acid sequences and so are 2 chloroplast forms, AtHsp70-6 and AtHsp70-7, and 2 ER forms, AtHsp70-11 and AtHsp70-12 (online Fig 5, online Table 3). Other AtHsp70s also share more than 64% and 57% similarity in the ATPase and peptide-binding domain, respectively. An exception with large sequence variation is found in Athsp70-8, which presumably encodes a chloroplast form that is approximately 25% similar to most other AtHsp70s (Figs 1 and 2, online Fig 5, online Table 3). It would be particularly interesting to understand the phylogenetic origin of this gene and whether any specialization in its function is involved.

Fig 2.

 Consensus domain structure of AtHsp70s. aa, amino acids; D, DnaK subfamily; E, Hsp110/SSE subfamily. Alignment of the deduced amino acid sequences of AtHsp70s is given in online Fig 5. Pairwise similarity and identity of the sequences are compared according to the structural domains and are listed in online Table 3

Hsp110/SSE SUBFAMILY

The Hsp110/SSE orthologs

Although the genes in this subfamily encode proteins with Hsp70 signatures, they are in general not highly conserved at the nucleotide sequence level. We used the deduced protein sequences from the previously identified genes, including those from Arabidopsis, mammals, yeasts, and Caenorhabditis elegans, to search the database and discovered 2 genes that encode proteins homologous to the previously reported Hsp110/SSE of Arabidopsis, Hsp91 (Storozhenko et al 1996) (online Fig 6). As listed in Table 1, the new genes Athsp70-15 and Athsp70-16 encode proteins of smaller sizes, 82 and 85 kDa, respectively. Based on predictions from the PSORT program (Nakai and Kanehisa 1992), these Hsp110/SSE homologs are likely localized to the cytoplasm, whereas a pattern for nuclear targeting, PKKK, is also present in AtHsp70-15 and Hsp91 (renamed AtHsp70-14 herein). AtHsp70-14 and AtHsp70-15 contain 98% identity in the amino acid sequences, whereas AtHsp70-16 is 80% and 55% similar to these proteins in the ATPase and the peptide-binding domains, respectively (Fig 2, online Figs 5 and 6, online Table 3). Phylogenetic analysis shows that these deduced proteins form a distinct group, clustered with the yeast SSEs (Fig 1).

A Grp170 ortholog

Exhaustive searches for a Grp170-like gene revealed 3 coding sequences predicted from the adjacent and the same regions in a BAC clone, ATFCA6. The first predicted gene (protein identification number CAB10440) spans 6013 base pairs (bp) of the genome. It codes for a protein of 912 amino acids homologous to the ATPase domain of Hsp70s, but in its C-terminal region. Downstream of this gene, following a gap of 358 bp, 2 predictions for genes have been made from the same region, which essentially code for the same protein (protein identification numbers CAB46039 and G71433), except the latter was 41 amino acids short in the C-terminus. However, these coding regions are partial, because they lack the ATG codon, and yet the deduced amino acid sequences are similar to the peptide-binding domain of Hsp70s. We then subjected 20-kb nucleotide sequences covering these regions to exon prediction using the GENSCAN program (Burge and Karlin 1997) and identified a new gene, designated Athsp70-17 (Table 1, online Fig 7). The coding sequence starts 4102 bp downstream of the predicted ATG codon in CAB10440, spanning 14 exons (Fig 3, online Fig 7) and encoding 867 amino acids. The deduced protein contains 50% and 62% similarity to the human (ORP150) and yeast (LHS1) Grp170 orthologs, respectively, throughout the entire length, including the ER targeting signal (online Fig 8). In the phylogenetic tree (Fig 1), AtHsp70-17 and the yeast LHS1 form a distinct group, separated from the yeast SSEs and AtHsp70-14∼AtHsp70-16. In the EST database, a complementary DNA clone (APZL40f12) exists that contains sequences identical to the complete open reading frame predicted for Athsp70-17, with additional untranslated regions of 105 bp at the 5′-end (accession number AV529564) and 161 bp at the 3′-end (accession number AV523811). Thus, we have confirmed that the Arabidopsis genome contains at least an ortholog of Grp170. So far we have not found other sequences similar to Athsp70-17. It is an open question whether paralogous genes are present and remain to be discovered.

Fig 3.

 Gene structure of the Arabidopsis hsp70s. Similar exon-intron organization was found in most genes encoding proteins predicted to localize in the same subcellular compartment. The pattern of tripartite HSEs (3 × HSE), 5′-nGAAnnTTCnnGAAn-3′, was searched in 5′ flanking noncoding region (≤1 kb, if present) using the default parameters in MatInspector program version 2.2 (Quandt et al 1995). Clusters of HSEs in other forms exist in the same region of all genes (not shown, see text for details)

GENOME ORGANIZATION

The gene structures of Athsp70s provide further information for their phylogeny. As seen in Figure 3, genes encoding proteins localized to the same cell compartment tend to have similar exon-intron organization. For example, 5 of 6 genes in the cytoplasmic group contain a single intron in a conserved position, whereas the only other gene in this group is intronless. The highly similar pairs Athsp70-6 and Athsp70-7 in the chloroplast group, Athsp70-9 and Athsp70-10 in the mitocondrial group, Athsp70-11 and Athsp70-12 in the ER group, and the group of hsp110/SSE homologs, each has conserved, though not exactly identical, gene structure. These patterns are highly suggestive of a duplication origin for these genes.

The 18 Athsp70s are distributed among all 5 chromosomes. As shown in Figure 4, Athsp70-1 and Athsp70-2 are arranged in tandem on chromosome V, as are Athsp70-14 and Athsp70-15 on chromosome I. These genes likely resulted from local gene duplication. Other genes of the superfamily did not show clustering in the genome, although some are located within regions containing high density of ribosomal protein genes (data not shown). Athsp70-13 on chromosome I and Athsp70-8 on chromosome II are located in conserved positions in the regions of genome that have been reported to be duplicated (Blanc et al 2000). However, the sequences are highly diverged. Athsp70-8 is the most distantly related species among all members of the gene family (Fig 1, online Fig 5, online Table 3). Athsp70-11 and Athsp70-12 are within a region on chromosome V that is duplicated on chromosome IV but with no homologous copies present. Such is also the case for Athsp70-9, which is located in a region of chromosome IV that is duplicated on chromosome II where no hsp70 genes were found. These data might indicate a loss of genes or insertion by transposition in the duplicated region.

Fig 4.

 Chromosome positions of the Arabidopsis hsp70s. Constructed based on the sequence map provided by The Arabidopsis Information Resource (TAIR, http://www.arabidopsis.org/cgi-bin/maps/Schrom). For clarity in the figure, gene names are abbreviated into numbers only, in boldface type, to the left of the chromosomes. Map markers are listed to the right of the chromosomes, with distance (cM) from the top of the chromosome in brackets

In genomic sequences, the 5′-flanking noncoding regions of Athsp70s contain consensus cis-regulatory promoter element for heat shock response (HSE) (Fernandes et al 1994; Schöffl et al 1998). These elements exist in clusters of 3 or more units in tandem or alternating between the complementary strands of genomic DNA. At least 4 such clusters are found within 500-bp upstream sequences in each Athsp70 (data not shown). These features are the signature of heat-inducible genes (reviewed in Fernandes et al 1994; Schöffl et al 1998). Since we found that not all Athsp70s are induced by thermal stress (see below), it will be interesting to experimentally test the specificity of these cis-elements and also to see whether there are enhancer elements involved in the thermal regulation of gene expression. Figure 3 indicates in each Athsp70 the location of tripartite HSEs with the basic pattern 5′-nGAAnnTTCnnGAAn-3′, which were shown to be required for effective binding of trimeric heat shock transcription factors (Fernandes et al 1994; Schöffl et al 1998), as determined with MatInspector program (version 2.2, Quandt et al 1995).

EXPRESSION PATTERN

Since the Athsp70 sequences are highly conserved, we took the approach of relative quantitative reverse transcription–polymerase chain reaction (RT-PCR) to assess the expression pattern of the various genes. From the diverse 3′-untranslated region, gene-specific primer pairs were designed (online Table 4) such that the PCR products are of similar size, approximately 200 bp, which are amplified with similar efficiency among the reactions being compared. The relative expression level of various genes in each given organ was compared and cross-referenced with that of each given gene in a panel of organs. In more than 30 independent runs of reactions, we could not detect the RT-PCR product of Athsp70-18. As previously mentioned, there is no EST for this gene in the database, and the inferred amino acid sequences show that the predicted gene product resembles a truncated version of a regular cytoplasmic Hsp70. Together, these data suggest that Athsp70-18 is a pseudogene.

Table 2 gives a summary of the results from the expression analyses on Athsp70s. In normal growth temperature, the transcripts of Athsp70-1, Athsp70-6, Athsp70-7, and Athsp70-17 are present at relatively high levels in seedlings and in most organs of mature plants, whereas most other genes are expressed at a basal level in the vegetative organs. Athsp70-2, which is genomically arranged in tandem 3′ to Athsp70-1, is expressed at very low or undetectable levels in most organs, yet shows enhanced expression in the siliques. Selected genes in each subcellular group are induced in the mature yellow siliques. Included among these genes are Athsp70-4 and Athsp70-5 for those encoding cytoplasmic proteins, Athsp70-8 for the chloroplasts, Athsp70-10 for the mitochondria, and Athsp70-15 in the hsp110/SSE group.

Table 2.

 Relative expression levels of Arabidopsis hsp70s^a

Thermal stress significantly induces a number of Athsp70s in the seedlings and flowers, but has no or minimal effect on roots, leaves, and yellow siliques of mature plants. For some subcellular groups, there are highly heat-induced species, eg, Athsp70-4 and Athsp70-5 in the cytoplasmic group and Athsp70-8 in the chloroplast group (Tables 1 and 2). The expression of the chloroplast Athsp70-7, the mitochondrial Athsp70-10, the ER Athsp70-11, and the Hsp110/SSE Athsp70-15 are also enhanced substantially by thermal stress in the seedlings. Those that are expressed at higher levels in normal condition, such as the cytoplasmic Athsp70-1, the chloroplast Athsp70-6, and the Grp170 ortholog Athsp70-17, are also heat induced, but to a less extent. The expression profiles of Athsp70s observed here are highly similar to those of genes encoding small Hsps in Arabidopsis and in other plants (Coca et al 1994; DeRocher and Vierling 1994; Coca et al 1996; Waters et al 1996; Wehmeyer et al 1996; Wehmeyer and Vierling 2000). These results indicate a division of labor, in terms of the temporal and spatial regulation of gene expression, among the Athsp70s.

CONCLUSION

The Arabidopsis genome contains at least 18 genes encoding members of the Hsp70 superfamily. These genes have complex regulation patterns, and the protein products are localized to specific subcellular compartments, suggesting distinct functional roles in vivo. In the yeast and human genome, so far 14 and 16 genes have been identified, respectively. It is not surprising to see plants having more hsp70s than other eukaryotes because of the presence of an extra organelle, the plastid, in the cell. The Hsp70s homologous to bacterial DnaK have been recognized as a highly conserved group of proteins in evolution and now are joined by relatively distant members of the Hsp110/SSE type. This ubiquitous superfamily provides a wealth of information for functional genomic studies and for the investigations of comparative genomics and molecular phylogeny.

Fig 5.

 Conservation of the deduced amino acid sequences of the 18 Arabidopsis Hsp70s. Sequences were aligned using GeneDoc program version 2.5.000 (Nicholas and Nicholas 1997). Shades in black, dark gray (with white letters), and light gray (with black letters) indicate the conservation of the residues in 90%, 70%, and 50%, respectively, of the 18 sequences. A total of 76 residues are conserved in all members. In the conserved regions, members of the DnaK subfamily (70-1∼70-13 and 70-18) and of the Hsp110/SSE subfamily (70-14∼70-17) contain distinct sequences. Divergence in the sequences is found in the extreme N- and C-terminus, where members of the same subcellular location contain varied degrees of similarity

Fig 6.

 Conservation of the deduced amino acid sequences of Hsp110/SSE homologs in Arabidopsis (AtHsp70-14∼AtHsp70-16), human (APG-1, O95757), and yeast (SSE1, P32589). Sequences were aligned as in online Fig 5

Fig 7.

 The genomic and encoding sequences of Athsp70-17. The exons and introns were predicted using GENSCAN program (Burge and Karlin 1997) and modified according to the sequences of the EST clone APZL40f12 (accession numbers AV529564 and AV523811)

Fig 8.

Conservation of the deduced amino acid sequences of Grp170 homologs in Arabidopsis (AtHsp70-17), human (ORP150, AAC50947), and yeast (LSH1). Sequences were aligned as in online Fig 5

Table 3.

 Pairwise comparisons among members of the Arabidopsis Hsp70 family^a

Table 4.

 Nucleotide sequences of gene-specific polymerase chain reaction primers