Research Article

EIN2, a Bifunctional Transducer of Ethylene and Stress Responses in Arabidopsis

See allHide authors and affiliations

Science  25 Jun 1999:
Vol. 284, Issue 5423, pp. 2148-2152
DOI: 10.1126/science.284.5423.2148


Ethylene regulates plant growth, development, and responsiveness to a variety of stresses. Cloning of the Arabidopsis EIN2 gene identifies a central component of the ethylene signaling pathway. The amino-terminal integral membrane domain of EIN2 shows similarity to the disease-related Nramp family of metal-ion transporters. Expression of the EIN2 CEND is sufficient to constitutively activate ethylene responses and restores responsiveness to jasmonic acid and paraquat-induced oxygen radicals to mutant plants. EIN2 is thus recognized as a molecular link between previously distinct hormone response pathways. Plants may use a combinatorial mechanism for assessing various stresses by enlisting a common set of signaling molecules.

The plant hormone ethylene (C2H4) is a regulator of a variety of developmental and stress responses in plants including seed germination, cell elongation, cell fate, sex determination, fruit ripening, flower senescence, leaf abscission, defense against pathogens, and responses to mechanical trauma (1). The effect of ethylene on dark-grown seedlings (known as the triple response) has been studied extensively to investigate the role of ethylene in cell growth and stress responses (2–4). In Arabidopsis, the triple response consists of radial swelling of the hypocotyl, exaggeration of the apical hook, and inhibition of hypocotyl and root elongation (2, 4). Mutant loci with an ethylene-insensitive phenotype include etr1 (2),etr2 (5), ein2,ein3 (3, 6),ein5/ain1 (4, 7),ein4, ein6, and eir1(4). Mutants that overproduce ethylene (eto1, eto2, and eto3), constitutively activate the ethylene signaling pathway (ctr1) (8), or fail to form an exaggerated apical hook (hls1) have also been identified (3). A genetic model has been proposed where many of these genes act in a linear pathway in which EIN2 is a central component (4,9).

Cloning and characterization of genes in the ethylene pathway have provided insight into the molecular mechanisms that underlie the plant's response. Ethylene is perceived by a family of receptors (ETR1, ERS, ETR2, EIN4, and ERS2) that are similar to bacterial two-component histidine kinases (10). The receptors activate a Raf-like protein kinase, CTR1, which negatively regulates downstream ethylene signaling events (8). In animals, Raf kinases typically act through a mitogen-activated protein (MAP) kinase phosphorylation cascade (11). Thus, the ethylene signal transduction pathway has features of both the bacterial two-component system and eukaryotic MAP kinase pathways. A similar combination of signaling components is found in the osmolarity response system in yeast (12). Downstream components in the ethylene pathway include the EIN3/EIL (EIN3-LIKE) family of transcription factors (13). Genetic data indicate that EIN2 mediates an essential step in the signal propagation between the CTR1 and EIN3/EIL (4, 6).

Mutations in the EIN2 locus confer insensitivity to both exogenous and endogenous ethylene (Fig. 1) (4, 14). Previously, five ein2 mutant alleles had been identified (3, 4), and by further mutant screens, we extended the number of alleles to 25. With the exception ofein2-9, all ein2 alleles showed complete insensitivity to ethylene at the morphological, physiological, and molecular levels (Fig. 1). ein2 mutants have also been recovered in screens for Arabidopsis mutants resistant to auxin transport inhibitors (15), cytokinins (16), or abscisic acid (17) and in screens for delayed senescence (18). It is intriguing, however, that none of the other ethylene-insensitive loci were recovered in these screens. In this regard, the unique characteristics of EIN2 may be explained by the fact that EIN2 is the only gene known whose loss-of-function mutations lead to complete ethylene insensitivity (4, 14). Alternatively, it is possible that EIN2 mediates cross-talk between several hormone signaling pathways.

Figure 1

Ethylene insensitivity of ein2mutant plants. (A) Comparison of 3-day-old etiolated wild-type and ein2-5 seedlings grown in hydrocarbon-free air or 10 parts per million of ethylene. (B) Comparison of wild-type and ein2-5 adult plants grown in hydrocarbon-free air or in the continuous presence of 1 ppm of ethylene. (C) Comparison of the ethylene-mediated induction of At-GST2 in wild-type and ein2-5 3-day-old etiolated seedlings. Northern blots were performed with 20 μg of total RNA per lane and sequentially probed. Plant growth and ethylene treatment were carried out as described (8).

Cloning EIN2

We cloned the EIN2 gene using a map-based approach (Fig. 2). The ein2-1 allele was mapped to the top of chromosome 5, between the ATCTR1 and ubq6121 loci (4) (Fig. 2A). A yeast artificial chromosome (YAC) contig was assembled by hybridization of the markers to several libraries (Fig. 2A). Restriction fragment length polymorphisms were identified for several of the YACs and cosmids in the region (19) and used to position EIN2 between yUP2G11L and ubq6121 (Fig. 2A).

Figure 2

Cloning, sequence, and membrane localization of EIN2. (A) Map of theEIN2 gene relative to the chromosome walk. Molecular markers in the region and the number of recombinations between the marker and the EIN2 locus are shown. The cosmid g3715 identified a polymorphism between the genomes of ein2-12and wild type. PgEE1.2 represents the Eco RI g3715 subclone that was absent in the ein2-12 mutant genomic DNA. A genomic clone (λgE2) containing the EIN2 gene was identified with this subclone as a probe. The EIN2 gene (AF141202) is schematically represented; the seven introns are indicated by gaps. (B) Sequence alignments of the first 497 amino acids of EIN2 and full-length At-Nramp2 (AF141204), DCT1/mouse Nramp2 (AF008439), human Nramp1 (D50403), Malvolio (U23948), and yeast SMF1 (Z74864) proteins (41). The 12 putative transmembrane domains are highlighted by a line above the sequences, which shows the orientation relative to the membrane (arrows point inward). Identities and similarities among the different proteins are indicated by black and gray boxes, respectively. (C) Schematic representation of the predicted structure of EIN2 and identified changes in ein2 mutant alleles. The predicted transmembrane helices are shown as cylinders through the lipid bilayer. Charged residues within the transmembrane regions are indicated ( + or −). The position of the putative coiled-coil is indicated by the coil after the 12th transmembrane domain. The positions of mutations in ein2 alleles are indicated. (D) EIN2 is an integral membrane protein. (Top and middle panels) A transcript encoding the NH2-terminal 560 amino acids of EIN2 was translated in vitro in the presence or absence of canine pancreatic microsomes. Yeast α-factor and luciferase were used as controls. Translated proteins were extracted with 1 M potassium chloride or sodium carbonate (pH 11.5). Proteins were analyzed by SDS–polyacrylamide gel electrophoresis (PAGE). (Bottom panel) In vitro–translated EIN2 NH2-terminus and glycosylated α-factor were treated with Proteinase K (0.1 mg/ml) in the presence or absence of 0.5% Triton X-100 or deoxycholate (0.5 mg/ml) and then resolved by SDS-PAGE. P, pellet; S, supernatant.

Genomic clones from the EIN2 region were hybridized toein2 allele blots, and polymorphisms were identified inein2-6 and ein2-12 (19). Using a polymorphic 1.2-kb Eco RI fragment as a probe, we identified genomic and cDNA clones (20). The pcE2.17 cDNA was 4747 bases in length with a single, large open reading frame encoding a predicted protein of 1294 amino acids (GenBank accession number AF141203) that spanned the polymorphic 1.2-kb Eco RI fragment (Fig. 2, A and B). The pcE2.17 cDNA detected a single 4.8-kb transcript, indicating that the cDNA was near full-length. Low-stringency hybridization of genomic DNA with the pcE2.17 cDNA indicated that EIN2 is a single-copy gene in Arabidopsis. Mutations in EIN2 cause phenotypes in roots, leaves, inflorescence stems, and flowers. Northern (RNA) analysis revealed that these tissues expressed the pcE2.17 transcript and that the steady-state mRNA level was unaffected by treatment with ethylene. Examination of the mRNA expression inein2 alleles revealed a reduction in the levels of this transcript, suggesting that pcE2.17 encodes the EIN2 mRNA. Confirmation of the identity of EIN2 was provided by examination of the sequences of ein2 alleles (21) and by genetic complementation (22). Twelve of the sequencedein2 alleles predicted a premature termination of the protein (Fig. 2). In addition, the mutations in ein2-6and ein2-12 result in in-frame deletions of 12 and 24 base pairs, respectively, although these mutations confer complete ethylene insensitivity. The least severe allele, ein2-9, contained a substitution of His-1143 to Pro (Fig. 2). For complementation testing, ein2-5:pcE2.17 plants were generated by in planta vacuum transformation (22). The ethylene insensitivity of ein2-5 was abrogated by the ectopic expression of the EIN2 cDNA; ethylene responsiveness was at least partially restored in 60 out of 100 independent transformed lines. We therefore conclude that pcE2.17 corresponds to the EIN2 cDNA.

The Predicted Protein

Comparison of the EIN2 cDNA (AF141203) and genomic (AF141202) sequences revealed that the coding region of EIN2is interrupted by seven introns (Fig. 2A). EIN2 encodes a polypeptide of 1294 amino acids with a molecular mass of 141 kD and dimorphic structure. The NH2-terminal 461 amino acids of EIN2 contain regions of extreme hydrophobicity, whereas the COOH-terminal 833 amino acids are predominately hydrophilic. The protein sequence predicts 12 transmembrane helices, all within the NH2-terminal domain (Fig. 2, B and C). The NH2and COOH ends of the protein are predicted to lie on the same side of the membrane. The COILS algorithm (23) indicated an amphipathic helix in the EIN2 COOH tail (Ser-485 to Leu-515) adjacent to the 12th transmembrane domain (Fig. 2C). The presence of a coiled-coil–forming helix suggests that this region of the COOH tail may be a site for protein-protein interactions.

EIN2 does not contain obvious consensus sequences for protein sorting, although its hydrophobic nature suggests that it is an integral membrane protein. We tested this possibility by in vitro translating EIN2 with canine pancreatic microsomes. Proteins translated in the presence of microsomes were treated under conditions that extract nonintegral membrane proteins, and then membrane fractions were sedimented by ultracentrifugation. An EIN2 protein fragment containing the first 560 amino acids was found to partition with the membrane fraction, whereas luciferase, a nonmembrane-associated protein, did not (Fig. 2D, top panel). Yeast α-factor was translocated into the lumen of intact microsomes (24) (Fig. 2D). After treatment of membranes under conditions that release the microsomal lumen contents, the EIN2 NH2-terminus, but not the yeast α-factor, was found in the microsomal membrane fraction. Thus, the hydrophobic portion of EIN2 protein behaves as an integral membrane protein. Consistent with the lack of a signal sequence or any consensus glycosylation sequence within the first 560 amino acids, the size of the EIN2 protein was not altered by the presence of membranes (Fig. 2D, top and middle panels). The topology of EIN2 within the microsomes was examined by protease digestion (Fig. 2D, bottom panel). The membrane-associated NH2-terminus of EIN2 was not protected from proteinase digestion by the lipid bilayer. However, the lumen-localized glycosylated α-factor remained intact until the bilayer was disrupted (Fig. 2D, bottom panel). These results are consistent with a topology prediction where the NH2-terminal portion of the EIN2 protein is integrated within the membrane. Attempts to determine the cellular location of endogenous EIN2 protein with either EIN2-specific antibodies, epitope-tagged EIN2 proteins, or EIN2::GFP (green fluorescent protein) fusion proteins were unsuccessful.

Database searches revealed that EIN2 contains sequence similarity (21% identity) to the Nramp family of proteins (Fig. 2B); the region of similarity between EIN2 and the Nramp proteins is restricted to the NH2-terminal hydrophobic domain (EIN2 residues 1 to 480). Nramp-related proteins have been found in all organisms from bacteria to humans (25), and several of these function as transporters of a variety of divalent cations (26). Mutations in members of this family cause a variety of phenotypes, including hypersensitivity to bacterial infection (mouse Nramp1), anemia [mouse and rat Nramp2 (25)], and alteration of taste behavior [Drosophila Malvolio (27)]. In addition, we identified anArabidopsis gene more closely related to mouseNramp1 than EIN2 (Fig. 2B). Unlike mutations inEIN2, plants containing a transferred DNA (T-DNA) insertion mutation in At-Nramp2 (28) showed normal responsiveness to ethylene.

Physiological studies indicated that ethylene perception may require a transition metal such as copper or zinc (29). Indeed, copper is needed for the reversible binding of ethylene to a protein receptor (30). Although a copper transporter necessary for ethylene receptor function has been recently identified (31), its similarity to Nramp proteins suggested that EIN2 may also function as a metal transporter. To test this hypothesis, we examined transport activity by expressing EIN2 (or truncated forms) in several heterologous systems, includingXenopus oocytes, yeast [mif mutant (32)], and baculovirus-infected insect cells. Although EIN2 expression was detected, no metal-transporting capacity was observed.

Ethylene Response Signaling

Overexpression of downstream positive regulators in the ethylene pathway, such as EIN3/EIL1 (6) or ERF1 (13), leads to constitutive ethylene response phenotypes. In contrast, none of the plants that expressed the full-length EIN2 protein or NH2-terminal hydrophobic Nramp-related domain (amino acids 1 to 480) showed constitutive ethylene responses or hypersensitivity to ethylene. Transgenic plants that expressed the COOH end of EIN2 (CEND) in the ein2-5 genetic background, however, showed constitutive ethylene response phenotypes as adults (Fig. 3A). ein2-5:CEND plants produced small rosettes, and their flowers showed morphological characteristics similar to those of the constitutive ethylene response mutant ctr1 (Fig. 3A), as the gynoecium protruded through the petals before the bud opened and only the later flowers were fertile and produced viable seeds (8). Under long-day conditions, ein2-5:CEND transgenic plants flowered at the same time as wild type—1 week earlier than the parentalein2-5 plants.

Figure 3

Activation of ethylene responses by overexpression of the COOH end (CEND) of EIN2. (A) Comparison ofein2-5, ctr1, and ein2-5:CEND adult plants and flowers. Flowers of the same age from the first inflorescence are shown for the different backgrounds. (B) Comparison of 5-day-old seedlings germinated in constant light on water-agar plates (no salts or sugars added) containing no ACC (upper panels) or 50 μM 1-aminocyclopropane-1-carboxylic acid (lower panels). (C) Expression of ethylene-inducible genes in wild type, ein2-5, and three independent ein2-5:CEND transgenic lines. Northern blots were performed with 20 μg of total RNA per lane from adult plants treated for 48 hours with hydrocarbon-free air or 10 ppm of ethylene, and sequentially probed. Induction of defensin (PDF1.2) mRNA by paraquat (D) or methyl jasmonate (MeJA) (E), in wild-type,ein2-5, and ein2-5:CEND transgenic plants. Leaf samples were collected 48 hours after spraying with paraquat (25 μM) or MeJA [45 μM in 0.1% (v/v) ethanol]. Northern blots were performed with 20 μg of total RNA per lane and sequentially probed.

We further examined the basis of the phenotypes observed inein2-5:CEND plants using another bioassay for ethylene response. When seeds are germinated and grown in the light in medium devoid of nutrients, ethylene can promote elongation in the hypocotyl (33). Under these growth conditions, the hypocotyl of ethylene-insensitive mutant seedlings was shorter than that of ethylene-treated wild-type or ctr1 mutants (33) (Fig. 3B). The hypocotyl of ein2-5:CEND seedlings was longer than that of ein2-5 seedlings and similar to that ofctr1 seedlings (Fig. 3B). These results suggest that the phenotypes observed in ein2-5:CEND plants are mediated by constitutive activation of the ethylene response. Northern analysis revealed that ein2-5:CEND plants constitutively expressed mRNAs for all ethylene-regulated genes tested (At-GST2,basic-chitinase, and At-EBP), confirming the role of this domain of EIN2 in ethylene signaling at the molecular level. Moreover, gene expression was unaffected by ethylene treatment (Fig. 3C), indicating that the NH2-terminus of EIN2 is necessary for regulation by the hormone. The fact that overexpression of CEND, but not full-length EIN2, confers constitutive ethylene responses also implies that the Nramp-like domain may control CEND function.

Components of both the ethylene and jasmonate signaling pathways are required for induction of the pathogen-responsive genePDF1.2 (34). Mutations that do not completely block the response to ethylene (such as etr1-3 orein3), however, respond normally to jasmonate or the superoxide anion-producing compound paraquat (34). We tested whether the signal generated by the CEND of EIN2 was sufficient to alleviate the ethylene signaling requirement for the induction of PDF1.2 by jasmonate or paraquat. Whereas ein2-5plants were completely unresponsive, ein2 plants overexpressing the EIN2 CEND were able to respond normally to treatment with paraquat (Fig. 3D) or jasmonate (Fig. 3E) and induce PDF1.2 gene expression. Thus, the signal generated by the CEND of EIN2 restores jasmonate and paraquat responsiveness in an ethylene-insensitive plant.

Although genetic studies demonstrate a requirement for an intact CEND sequence for ethylene-mediated activation of the triple response in etiolated seedlings, expression of EIN2 CEND domain inein2-5 was not sufficient to induce the triple response; the seedlings remained completely insensitive to ethylene (35). Thus, the Nramp-like domain of EIN2 is necessary for ethylene-mediated effects on skotomorphogenesis, seedling development in the absence of light. Overexpression of CEND may be unable to evoke the triple response in etiolated seedlings because of an insufficient threshold amount of active CEND or of a CEND-interacting protein in etiolated seedlings.

Order of Gene Action

Because EIN3 has been proposed to act after EIN2 in the ethylene pathway (6), we analyzed the effects of EIN2 CEND overexpression in ein3-1 plants. From 100 independentein3:CEND lines examined, two lines that showed the most severe ctr1-like phenotypes were characterized. Comparison of the effects of EIN2 CEND expression in ein2-5 andein3-1 plants indicates that although the levels of transgene expression were similar (Fig. 4B), reduction in the rosette size was less profound in ein3-1 and comparable to that observed in the ein3/ctr1 double mutant (Fig. 4A). In addition, thectr1-like flower morphology was suppressed in theein3:CEND plants (Fig. 4A). Thus, ectopic expression of the COOH end of EIN2 is sufficient to activate downstream events of the pathway in an EIN3-dependent manner.

Figure 4

Requirement of EIN3 for EIN2 CEND constitutive ethylene response phenotypes. (A) Phenotypes of 4week-old plants and flowers of the indicated genotypes. (B) Comparison of the levels of mRNA expression of the EIN2-CEND transgene in the lines shown in (A) and (B).

Comparisons to Other Pathways

Identification and analysis of EIN2, a regulator of ethylene signaling in Arabidopsis, has uncovered an unexpected feature in the ethylene response pathway. Whereas several of the previously identified components in this pathway have equivalents in other eukaryotic signaling cascades (that is, ETR1/SLN1 and CTR1/Raf-1), EIN2 has no such counterpart. At the amino acid sequence level, EIN2 seems to be structurally unique. Nevertheless, proteins with “functional equivalence” to EIN2 may be found in other regulatory cascades. Indeed, there appears to be a parallel between EIN2 and the yeast glucose sensors Snf3 and Rgt2 (36, 37). Like EIN2, these proteins contain two distinct domains. Their hydrophobic NH2-termini show similarity to glucose transporters but they are thought to act as glucose sensors, not transporters (36,37). They also possess hydrophilic COOH-termini that when overexpressed can partially activate the glucose response in the mutant yeast cells (37). Presumably like Snf3, EIN2 CEND may interact with downstream components in the pathway, while its hydrophobic NH2-terminus may function in sensing an unknown signal. The requirement of the NH2-terminal domain for ethylene responsiveness in etiolated seedlings and the sufficiency of its CEND for activation of the pathway imply that EIN2 acts as a bifunctional signal transducer. The similarity of the EIN2 NH2-terminus to the Nramp proteins suggests that this domain may be involved in “sensing” a divalent cation, inferring the existence of a second messenger in the central part of the ethylene signal transduction pathway. Alternatively, this domain may simply anchor the protein to the membrane, bringing the CEND domain to the site of interaction with other ethylene pathway components, such as ETR1 and CTR1. The presence of a coiled-coil motif in the EIN2 CEND suggests that information transfer may involve interaction between proteins.

  • * These authors contributed equally to this work and are listed alphabetically.

  • To whom correspondence should be addressed. E-mail: jecker{at}


Stay Connected to Science

Navigate This Article