ARTICLE

TOP ARTICLE LITERATURE CITED

Two genes in the newly complete genome sequence of the cyanobacterium Anabaena sp. strain PCC 7120 encode homologs of the ethylene hormone receptors of higher plants, and reanalysis of an ethylene-binding protein in the cyanobacterium Synechocystis sp. strain PCC 6803 reveals a third homolog. Here, we present analyses of these sequences and discuss the implications regarding the ancestry of ethylene receptor genes in higher plants.

Ethylene gas, one of the major plant hormones, influences many aspects of growth and development throughout the plant life cycle. In higher plants, ethylene is perceived by a family of receptors with similarity to two-component signaling proteins (Bleecker, 1999; Chang and Stadler, 2001). The typical two-component system consists of a sensor His protein kinase and response regulator protein (Fig. 1, top). These components, in various configurations, are widely used by bacteria in regulating responses to diverse stimuli (Parkinson and Kofoid, 1992). Several examples of this system have also been found in plants and fungi (Urao et al., 2001; Chang and Stewart, 1998), and include the recently discovered cytokinin hormone receptor in plants (Inoue et al., 2001). It is unclear whether the ethylene receptors signal by a typical two-component signaling mechanism, which involves His autophosphorylation followed by transfer of the phosphate to an Asp residue (Fig. 1, top). It is interesting that the pathway for ethylene signaling in higher plants appears to have evolved from distinct types of signaling components (Bleecker, 1999). That is, the ethylene receptors somehow transmit their signal to the CTR1 Raf-related Ser/Thr protein kinase for which there are animal but not bacterial homologs (Kieber et al., 1993).

View larger version (26K): [in this window] [in a new window]   Figure 1.   Domain conservation between plant ethylene receptors and cyanobacterial genes. Top, Schematic depiction of the two-component system that links diverse input and output functions via a His-to-Asp phosphorelay (Parkinson and Kofoid, 1992). Bottom, Two prototypical plant ethylene receptors, based on Arabidopsis ETR1 (Chang et al., 1993) and ERS1 (Hua et al., 1995; both subfamily 1 ethylene receptors), shown to scale with the three putative cyanobacterial ethylene receptors. The ethylene-binding domain (EB) was identified based on Rodriguez et al. (1999); numbers below each EB domain report the negative log of the E value obtained using PSI-BLAST (Altschul et al., 1997) with amino acids 1 through 120 of Arabidopsis ETR1 as the query sequence. Other domains were identified by the National Center for Biotechnology Information (NCBI) CD search (http://www.ncbi.nih.nlm.gov; Altschul et al., 1997). Conserved domains are designated GAF (cGMP phosphodiesterases, adenylate cylcases, and Fhla), PP (PAS [Per, ARNT, Sim] followed by PAC), and CC (coiled coil). The His kinase contains the pfam00512/smart00388 domain (indicated by H) and the pfam02518/smart00387 domain, which includes the motifs N, G1, F, and G2 (Parkinson and Kofoid, 1992; indicated by vertical bars). The receiver domain (pfam00072/smart00448) is denoted by D. The slr1213-encoded response regulator output domain (orange, unlabeled) is related to HTH-AraC. Numbers below these domains report the negative log of the E value obtained for each domain using the CD search. In the case of slr1212 (*), the H region of the His kinase could not be detected by CD search, but was matched with numerous bona fide two-component His kinases using PSI-BLAST (E = 2 × 1021) and was aligned to Arabidopsis ETR1 using LALIGN (Huang and Miller, 1991; E = 6 × 105).

The ethylene receptors consist of an amino-terminal ethylene-binding domain, which is linked by a GAF domain (Aravind and Ponting, 1997) to a two-component His kinase domain (Fig. 1). Some of the ethylene receptors also have a carboxyl-terminal receiver domain. The ethylene-binding domain includes three membrane-spanning regions and binds ethylene using a copper cofactor (Schaller and Bleecker, 1995; Rodriguez et al., 1999). This signature domain of the receptors was thought to be exclusive to higher plants until a homologous sequence, slr1212, was discovered in the cyanobacterium Synechocystis (Kaneko et al., 1996). The ability of the slr1212-encoded product to bind ethylene was later confirmed (Rodriguez et al., 1999), but any connection to a signaling function has remained unknown, despite weak similarity to His kinases currently noted with an uncertain E value of 0.32 on Cyanobase (http://www.kazusa.or.jp/cyanobase/).

Now, in the recently completed genome sequence of a second cyanobacterium, Anabaena (Kaneko et al., 2001; http://www.kazusa.or.jp/cyano/Anabaena/), we have identified two genes (all0182 and alr4716) that each encode domains characteristic of plant ethylene receptors, including the ethylene-binding and His kinase domains (Figs. 1 and 2). The predicted Anabaena proteins have approximately 40% amino acid identity with plant ethylene receptors in the ethylene-binding domain and 35% identity (for all0182) or 26% identity (for alr4716) in the His kinase domain. As in some of the plant ethylene receptors, the all0182-encoded protein has a carboxyl-terminal receiver domain that shares up to 32% amino acid identity with those of Arabidopsis ethylene receptors, including the conserved phosphorylation site.

View larger version (45K): [in this window] [in a new window]   Figure 2.   Amino acid alignments among three Arabidopsis ethylene receptors (EIN4 [subfamily 2; Hua et al., 1998], ERS1 [Hua et al., 1995], and ETR1 [Chang et al., 1993; both subfamily 1]), the four cyanobacterial proteins with putative ethylene-binding domains (slr1212, all0182, alr4716, and all4896), and two representatives of the slr1212 His kinase subfamily (all1280 and all2095). A, Ethylene-binding domain (Rodriguez et al., 1999). Putative transmembrane segments (Rodriguez et al., 1999) are underlined. B, CC domain found immediately after the cyanobacterial ethylene-binding domains. C, GAF domain (Aravind and Ponting, 1997). The alignment shown extends beyond the GAF domain itself, which ends at amino acid 315 of ETR1. D, His kinase domain. The alignment of proteins encoded by slr1212, all1280, and all2095 is broken out separately after the G1 motif to emphasize the conservation of a distinct set of amino acids. E, Receiver domain. An asterisk below the alignments indicates amino acids conserved among all sequences aligned; a colon indicates positions where at least four of six or five of seven are identical or all are similar. (These annotations exclude the more distantly related all4896-encoded ethylene-binding domain and the degenerate EIN4 His kinase domain.) Known motifs (Parkinson and Kofoid, 1992; Bleecker, 1999) are indicated above the alignments in red, and the amino acids matching these motifs are also indicated in red. In the motifs, O is (I,M,L,V),  is (D,E,N,Q), * is (A,G,P,S,T), and + is (H,K,R). Alignments were constructed by hand, incorporating output from CD search and PSI-BLAST (Altschul et al., 1997).

Upon reexamination of Synechocystis slr1212, we found compelling evidence that it too encodes a His protein kinase related to plant ethylene receptors (Figs. 1 and 2). When the translated slr1212 sequence was used as a query in a PSI-BLAST search (Altschul et al., 1997) of the entire NCBI nonredundant protein database, the top matches (E <10¹⁰⁰) were all His kinases, including plant ethylene receptors. Of the five motifs characteristic of His protein kinases (Parkinson and Kofoid, 1992), all but the F motif and part of the G2 motif are present in the predicted slr1212 protein (Fig. 1). In the bacterial His kinase CheA, amino acids in the F and G2 motifs are essential for activity, but are not essential for ATP binding and are believed to function primarily by allowing appropriate conformational changes during catalysis (Bilwes et al., 2001; Hirschman et al., 2001). There are examples of His kinases with proven activity that lack these motifs (Parkinson and Kofoid, 1992). Several dozen putative bacterial His kinases, including three from Anabaena, are quite similar to that encoded by slr1212 in the region following the G1 motif. A standard BLAST search with residues 801-844 of the slr1212-encoded product retrieved 45 protein sequences with at least 30% identity in this region, and Gly-813 (bold and red in Fig. 2D) is conserved in every one of these alignments. This, together with the conservation of other motifs, including the site of autophosphorylation on His, suggests that the slr1212 protein belongs to a variant subfamily of functional His kinases in which a distinct set of conserved amino acids replaces the F and G2 motifs (e.g. Fig. 2D).

In fact, slr1212 appears to lie within a two-component signaling operon in Synechocystis because the gene residing just 49 bp downstream of slr1212 in the Synechocystis genome (slr1213) encodes a response regulator, which is the typical downstream two-component partner of sensor kinases (Figs. 1, 2). Potential two-component operons are similarly observed for two of the Anabaena His kinase gene homologs in the slr1212 subfamily described above (all1280 and all2095). The output domain of the slr1213-encoded response regulator appears to be an HTH-AraC DNA-binding domain. This DNA-binding domain is conceivably regulated by two-component phosphorylation on the conserved Asp residue in the receiver domain, representing a putative operon that regulates gene expression directly in response to ethylene. It remains to be seen whether cyanobacteria use ethylene (or perhaps a related molecule) as a signal, and whether they also synthesize ethylene.

All of the cyanobacterial ethylene-binding domains, like those in subfamily 1 plant ethylene receptors (Bleecker, 1999; Chang and Stadler, 2001), lack the hydrophobic 21-amino acid amino-terminal extension characteristic of the subfamily 2 plant ethylene receptors (represented by EIN4 in Fig. 2; only subfamily 1 receptors are shown in Fig. 1). The function of the hydrophobic extension in subfamily 2 plant ethylene receptors is unknown. Other domain differences between the cyanobacterial proteins and the plant ethylene receptors can be seen in Figure 1. For one, the cyanobacterial ethylene-binding domains are immediately followed by a novel 40-amino acid coiled coil (CC) domain, which is found in many bacterial two-component His kinases (PSI-BLAST; data not shown), but appears to be absent in the plant ethylene receptors (Figs. 1 and 2). In addition, the all0182 product contains the structurally related PAS/PAC (PP) domain (Ponting and Aravind, 1997) in place of the GAF domain linking the ethylene-binding domain and the His kinase. The slr1212 product has two PP domains as well as a GAF domain, whereas the all4716 product has only the CC domain. The PP domain, the CC, and the GAF domain might all carry out similar functions, such as receptor multimerization or propagation of a conformational change in response to the signal (Galperin et al., 2001).

The combination of GAF and His kinase-related domains is particularly common in the two complete cyanobacterial genomes (Ohmori et al., 2001), with 46 examples in Anabaena and 12 in Synechocystis. The precise arrangement of a GAF domain immediately adjacent to the histidine kinase, found in the plant ethylene receptors, occurs 22 times in the Anabaena genome, but does not occur at all in Escherichia coli. Although this arrangement is not found in either of the putative Anabaena ethylene receptors or in slr1212, the wide variety of signaling domain combinations in cyanobacterial proteins (Ohmori et al., 2001; Galperin et al., 2001) suggests that domain shuffling among two-component signaling proteins is frequent on an evolutionary timescale. A third possible ethylene-binding domain of Anabaena (30% identity; E = 0.5) is encoded by all4896 if the amino terminus is extended by 67 amino acids beyond that which is annotated. The invariant C and H residues thought to chelate copper (Rodriguez et al., 1999; Fig. 2A) are present. This putative protein does not contain a histidine kinase domain, but does contain a GGDEF domain and weak similarity to PAS and GAF domains. The GGDEF domain is often found associated with bacterial two-component signaling domains (Galperin et al., 2001).

It is significant that ethylene receptor sequences have not been found in any organisms other than a few microbes and higher plants.² Apart from cyanobacteria and plants, we have detected three additional examples of ethylene-binding domain sequences in unfinished sequences from three bacterial species: Magnetospirillum magnetotacticum ( proteobacteria), Methylococcus capsulatus ( proteobacteria), and Cytophaga hutchinsonii (Chlorobium-flavobacteria/green sulfur group). Each encodes a complete ethylene-binding domain and adjacent coiled coil region associated with an incomplete sequence containing one or more domains typical of histidine kinases. These three bacterial species are substantially diverged one another and, notably, ethylene receptor genes are lacking in the complete genomes that are available for all three of these taxonomic groups. In fact, with the exception of the two cyanobacterial genomes discussed above, no ethylene receptor sequences are found in any of the 70 fully sequenced microbial genomes. Although we cannot exclude the possibility that the ethylene receptor genes were assembled independently in plants and bacteria, the observed distribution of ethylene receptor sequences is most consistent with a cyanobacterial (plastid) origin of plant ethylene receptors and rare horizontal transfer of ethylene receptor genes from cyanobacteria into diverse bacterial lineages. Whatever the evolutionary history of ethylene receptors within bacteria, these data suggest that functional ethylene receptors were most likely inherited by plants through the plastid lineage. We note that approximately 3% of nuclear protein-coding Arabidopsis genes are more similar to a gene from cyanobacteria than to genes from any other nonplant species, and this has been attributed to the transfer of many genes from the ancestral plastid genome to the plant nuclear genome (Abdallah et al., 2000; Rujan and Martin, 2001; The Arabidopsis Genome Initiative, 2000).

Two other plant receptors that are derived from the two-component system are phytochrome and the cytokinin hormone receptor. Phytochrome has close homologs in cyanobacteria. A phytochrome homolog in Synechocystis was shown to be a light-regulated two-component sensor kinase capable of phosphorylating a response regulator protein encoded within the same operon (Elich and Chory, 1997; Hughes et al., 1997; Yeh et al., 1997). Similar phytochrome operon sequences (aphA/alr3157 plus alr3158 and aphB/all2899 plus all2898) are found in the Anabaena genome (Kaneko et al., 2001). Although the signaling mechanism of higher plant phytochrome appears to have diverged from the two-component system, the structural and functional links to "prototypical" counterparts in cyanobacteria are quite compelling. Ethylene receptors and phytochromes are the only plant proteins known to contain GAF- and His kinase-related domains. Cytokinin receptors have not been described in cyanobacteria, but weak similarity has been noted between these receptors and putative ligand-binding domains with a wide phylogenetic distribution, including cyanobacterial proteins with two-component signaling domains (Anantharaman and Aravind, 2001; Mougel and Zhulin, 2001).

We conclude that the ancestral progenitor of plant ethylene receptors is likely to be a cyanobacterial ethylene receptor. The finding of intact cyanobacterial homologs to more than one key receptor of modern plants suggests that the ancient transfer of plastid genes to plants was of profound importance in shaping the fundamental character of higher plants.