RING-H2 type ubiquitin ligase EL5 is involved in root development through the maintenance of cell viability in rice


(fax + 81 29 838 7073; e-mail ynishi@affrc.go.jp).


Rice EL5 is an ATL family gene characterized by a transmembrane domain at the N-terminal and a RING-H2 finger domain (RFD), which exhibits ubiquitin ligase (E3) activity. To elucidate the physiological roles of EL5, we analyzed transgenic rice plants overexpressing mutant EL5 proteins that are impaired in E3 activity to various degrees. Plants expressing EL5C153A and EL5W165A, which encode an inactive E3, showed a rootless phenotype accompanied by cell death in root primordia, and those expressing EL5V162A, with moderately impaired E3 activity, formed short crown roots with necrotic lateral roots. The dominant-negative phenotype was specifically observed in root meristems where EL5 is expressed, and not recovered by exogenous auxin. When wild-type EL5 was transcriptionally overexpressed, the EL5 protein was barely detected by Western blotting. Neither treatment with a proteasome inhibitor nor mutation of the sole lysine residue, a potential target of ubiquitination, resulted in increased EL5 accumulation, whereas mutations in the RFD led to increased EL5 accumulation. The stabilized EL5 without the RFD was localized in the plasma membrane. Deletion of the transmembrane domain prevented the EL5 from localizing in the membrane and from exerting an inhibitory effect on root formation. Deletion of the C-terminal region also neutralized the negative effect. We concluded that EL5 plays a major role as a membrane-anchored E3 for the maintenance of cell viability after the initiation of root primordial formation. In addition, we propose that EL5 is an unstable protein, of which degradation is regulated by the RFD in a proteasome-independent manner.


Covalent attachment of ubiquitin, a highly conserved 76-amino acid polypeptide, is a post-translational modification system of proteins. In general, ubiquitination is accomplished by three enzymes: E1, the ubiquitin-activating enzyme that forms the E1-ubiquitin thioester; E2, the ubiquitin-conjugating enzyme from which the ubiquitin thioester is transferred; E3, the ubiquitin ligase that mediates the formation of the isopeptide linkage between the C-terminal glycine of ubiquitin and the internal lysine residue of the substrate. A polyubiquitin chain is subsequently formed by the addition of multiple ubiquitin monomers through the lysine 48 (K48) residue within the ubiquitin, and is then recognized by the 26S proteasome where the polyubiquitinated protein is degraded. In addition, ubiquitination is involved in plasma membrane (PM) protein internalization, followed by degradation in lysosomes, and regulates the activities of specific proteins (Fang and Weissman, 2004).

E3s act as either single or multisubunits; the single-subunit E3s are considered to possess an E2-interacting region and a substrate-recognition region. They are divided into four classes according to the characteristic domains: RING-zinc-finger domain (RFD), PHD finger, U-Box and HECT (Fang and Weissman, 2004). The RFD is defined by the consensus sequence C-X2-C-X9–39-C-X1–3-H-X2–3-(C/H)-X2-C-X4–48-C-X2-C, where X is any amino acid and the number of X residues varies in different fingers. The RFD is further divided into two subclasses based on a cysteine (RING-HC) or histidine (RING-H2) at the fifth residue of the consensus motif that binds two zinc atoms in a cross-brace structure (Saurin et al., 1996). Almost all known RING- and some HECT-type E3s are capable of mediating their own ubiquitination (Fang and Weissman, 2004).

In plants, protein degradation by the ubiquitin/proteasome system is important for various vital processes including development, cell division, hormonal responses, and biotic and abiotic stress responses (Smalle and Vierstra, 2004). More than 5% of the predicted proteome in Arabidopsis thaliana, for example, is postulated to be involved in the ubiquitination/26S proteasome pathway, and about 1300 genes are predicted to encode the E3 components (Smalle and Vierstra, 2004). Database searches have identified 469 RFD-containing proteins from the Arabidopsis genome (Stone et al., 2005) and 218 cDNA clones in the rice full-length cDNA database (Katoh et al., 2005). Some RFD-containing proteins are demonstrated to act as E3 with diverse physiological roles, such as SINAT5 and AIP2 in auxin- and abscisic acid (ABA)-signaling, respectively (Xie et al., 2002; Zhang et al., 2005), COP1 and CIP8 in light signaling (Hardtke et al., 2002; Osterlund et al., 2000), HOS1 in cold signaling (Dong et al., 2006), and RIN2 and RIN3 in disease resistance gene-specific hypersensitive response (Kawasaki et al., 2005).

The EL5 (Elicitor 5) gene is transiently induced 30 min after treatment with N-acetylchitoheptaose (Takai et al., 2001). N-Acetylchitooligosaccharide is recognized by a receptor in the PM (Kaku et al., 2006), and causes various cellular responses including generation of reactive oxygen species (ROS) and alteration of defense-related gene expression in many plants (Akimoto-Tomiyama et al., 2003; Shibuya and Minami, 2001). EL5 is structurally related to the ATL family first reported in Arabidopsis (Martínez-García et al., 1996). The ATL family proteins share five characteristic regions: a predicted transmembrane domain (region I), a region rich in basic amino acids (region II), a conserved region among the ATL family (region III), the RFD (region IV) and a highly diverse region in the C-terminal (region V) (Salinas-Mondragón et al., 1999). The ATL family genes are widely distributed in plant species. A genome-wide search for the ATL family identified 80 and 121 genes in Arabidopsis and rice, respectively (Serrano et al., 2006). To date, some ATL family genes are thought to play a role in disease resistance. For example, Arabidopsis ATL2 is induced by chitin and cellulase treatment (Salinas-Mondragón et al., 1999). In Arabidopsis eca mutants that constitutively express the ATL2 gene, basal expression levels of pathogenesis-related genes and salicylic acid (SA)- and jasmonic acid (JA)-responsive genes are also increased (Serrano and Guzmán, 2004). A T-DNA insertional mutant of Arabidopsis ATL9, which also responds to a chitin elicitor, shows increased susceptibility to powdery mildew disease (Ramonell et al., 2005). Moreover, expression of ACRE132, a tobacco ATL family gene, is induced in Avr9- and Cf9-mediated defense systems (Durrant et al., 2000). On the other hand, overexpression of MsRH2-1 of alfalfa, which shows sequence similarity to Arabidopsis ATL4, causes abnormal morphology, implying the participation in developmental processes (Karlowski and Hirsch, 2003). However, physiological roles of most ATL family members are obscure, and their molecular mechanisms are unknown because in vivo characterization of these proteins has not yet been performed.

We previously demonstrated that the RFD of EL5 (residues 129–181) is a binding domain for E2 and essential for E3 activity (Katoh et al., 2003; Takai et al., 2002). The RFD of EL5 fused to maltose-binding protein facilitates self-polyubiquitination in vitro when incubated with ubiquitin, recombinant mouse E1 and human UbcH4/5A (E2), or its rice homologue OsUBC5b, the transcription of which is upregulated by N-acetylchitoheptaose elicitor as well as EL5 (Takai et al., 2002). The three-dimensional structure of EL5 RFD was determined by NMR spectroscopy (Katoh et al., 2003). An EL5 RFD/OsUBC5b NMR titration experiment and an in vitro ubiquitination assay using amino acid replaced RFDs identified unambiguously the amino acid residues involved in OsUBC5b recognition (Katoh et al., 2005). These mutated RFDs show varying degrees of decreased E3 activity depending on the degree of contribution of the replaced amino acid residue to the interaction with E2 (Katoh et al., 2005).

In this study, we produced transgenic rice plants overexpressing the mutated EL5 genes carrying variously impaired E3 activity to prevent the function of the endogenous EL5. We found that mutations in the RFD of EL5 confer a dominant-negative phenotype by the regulation of cell death in root development, indicating that EL5 functions as E3 in rice root systems. EL5 is expressed in the root-forming region, and is a highly unstable protein because of the presence of the RFD. EL5 may interact with putative target(s) through its C-terminal region in the PM.


Overexpression of mutated EL5 gene arrests root growth in an E3-activity-dependent manner

To reveal physiological functions of EL5, we produced transgenic rice plants overexpressing the EL5 gene using the enhanced CaMV 35S promoter (Figure 1a) or suppressing EL5 by RNAi using the maize ubiquitin-1 promoter (data not shown). These plants did not show any obvious phenotypic changes. Overexpression of an inactivated SINAT5, a negative regulator of auxin signaling, or AIP2, a negative regulator of ABA signaling, which contains a single mutation in the zinc-chelating motif of the RFD, causes dominant-negative effects (i.e. hypersensitivity to auxin or ABA, respectively; Xie et al., 2002; Zhang et al., 2005). Substitution of the conserved Cys153 in the RFD of EL5 with serine (Takai et al., 2002) or alanine (EL5C153A) (Katoh et al., 2005) results in loss of its E3 activity in vitro. We therefore introduced the EL5C153A into rice plants and observed their phenotypes. Typical results are shown in Figure 1. Growth of callus and transformation efficiency appeared normal; however, about 68% of the regenerated shoots showed a rootless phenotype, about 13% formed short and brown abnormal roots, and the rest appeared normal with vigorous rooting, in which the EL5C153A gene was not expressed (data not shown). The drastic rootless phenotype observed in the plants overexpressing EL5C153A suggested that EL5 is involved in crown root (adventitious root) formation.

Figure 1.

 Overexpression of mutated EL5 encoding impaired E3 activity affects root formation.
(a) Phenotypes of the regenerated shoots cultured on a hormone-free MS medium for 2 weeks [GUS as control and EL5(wt)] and 1 month (EL5C153A and EL5W165A). The inserted photograph shows a longer-cultured shoot base transformed with EL5W165A producing tillers. Bars = 1 cm.
(b) Degrees of root formation of about 50 independent callus lines introduced with GUS (control), EL5(wt) and mutated EL5 genes. E3 activities deduced from in vitro ubiquitination assay (Katoh et al., 2005) are indicated.

NMR titration experiments identified amino acid residues in the RFD of EL5 important for the interaction with E2 (Katoh et al., 2003). In vitro E3 activity assay of the mutated RFDs revealed that W165A completely loses E3 activity, whereas L138A (strongly) and V162A (moderately) impair the E3 activity (Katoh et al., 2005). To investigate the relationship between the E3 activity of EL5 and rooting during callus regeneration, we introduced these three mutated EL5 genes into rice plants. As shown in Figure 1b, there was a good correlation between the expected E3 activities of the introduced gene products and the degree of root formation. Most plants overexpressing EL5W165A showed a rootless phenotype as observed in the EL5C153A plants. On the other hand, plants overexpressing EL5L138A tended to form brown spots in the shoot base where crown roots are formed (Figure 2a), and those with EL5V162A showed a moderate phenotype, such as short crown roots with necrotic lateral roots (Figure 2b).

Figure 2.

 Immature termination of crown and lateral root development in transgenic rice plants expressing mutated EL5 genes.
(a) Basal region of the shoot expressing EL5L138A. (b) Basal region of the shoot expressing EL5V162A. (c–f) Cross sections at shoot base of the regenerated plants with an empty vector (c and e) and EL5W165A (d and f). Arrowheads indicate crown root primordia. (g and h) Medial longitudinal sections at a lateral root primordium of a crown root of vector control and the EL5V162A plant, respectively. (i and j) Cross sections at a lateral root primordium of a crown root of vector control and the EL5V162A plant, respectively. All the plant samples were grown on a hormone-free MS medium. Scale bars = 100 μm.

To analyze internal structures of the rootless plants, we prepared thin sections of the basal region of the regenerated shoot. Crown root primordia were differentiated in the plants expressing EL5C153A and EL5W165A, but they did not fully develop, and regions containing the initial cells and root cap shrank and showed necrosis (Figure 2d,f; Figure S1). It seemed that the size of the primordial cells enlarged and the number of cells decreased compared with the normal primordia. Furthermore, plants expressing EL5V162A, which had reduced E3 activity, often formed short and brown crown roots with relatively many lateral root primordia that also showed necrosis (Figure 2b,h,j). On the other hand, the aerial parts of these rootless plants were observed as normal with regard to the morphology of shoot apical meristem (SAM) and vascular bundles in the culture bottle (Figure S2). The expression pattern of the homeobox gene OSH1 (Sato et al., 1996) was also indistinguishable from that of the control plant (Figure S2).

EL5 is highly expressed in the root-forming region

We analyzed the expression of EL5 in rice seedlings by Northern blot and in situ RNA hybridization. We divided 9-day-old seedlings into three parts: elongated shoot, basal region of shoot and root where crown roots are actively formed, and elongated root. Consistent with the phenotype, EL5 was mainly expressed in the basal region, and barely detected in the elongated shoot and root (Figure 3a). It is well known that plant hormones, especially auxin, are involved in the root development. Thus, the responses of EL5 expression to various hormones were examined. As shown in Figure 3a, treatment of auxin, cytokinin and JA increased the EL5 mRNA level in the basal region. In the root, treatment of cytokinin and JA, but not auxin, drastically induced EL5 expression. In the elongated shoot, none of the tested hormones except JA induced EL5 expression.

Figure 3.

 Tissue specificity and hormonal activation of EL5 expression.
(a) Rice seedlings were treated with the indicated hormones, and total RNA from each sample was analyzed by Northern blotting. rRNAs from each sample stained with ethidium bromide are displayed at the bottom. Con, control plants treated with water or 10 mm MES buffer (pH 6.0); ABA, abscisic acid; BAP, 6-benzylaminopurine; GA3, gibberellin; JA, jasmonic acid; Et, ethephon; SA, salicylic acid; IAA, indole-3-acetic acid; BL, brassinolide.
(b) In situ RNA hybridization of EL5. Cross sections were prepared from the shoot base of a 9-day-old rice plant treated with JA and hybridized with sense (left) or antisense (right) probe. Scale bars = 100 μm.

Signals of EL5 expression in rice seedlings were merely visible by in situ RNA hybridization, even though the Northern blot hybridization gave strong signals, as shown in Figure 3a. Thus, we analyzed the shoot base treated with JA and obtained signals in the cortex of the crown root primordia (Figure 3b), where cell death occurs in rice plants expressing mutated EL5.

Transgenic callus overexpressing mutated EL5 gene is morphologically auxin-irresponsive and similar to control callus treated with excess cytokinin

Auxin and cytokinin have central, but opposite, regulatory functions in root development. To ascertain whether the loss of EL5 function causes altered responses to these hormones, we compared the growth of callus transformed with an empty vector (control), EL5W165A and EL5Δ4-GFP, which lacks the RFD and is fused with GFP, on the regeneration medium with various quantities of auxin and cytokinin. As shown in Figure 4, on the medium containing 20 mg l−1 kinetin, root elongation in control callus tended to be inhibited, and some shoots showed a rootless phenotype. For the medium containing 2 mg l−1α-naphthalene acetic acid (NAA) and no kinetin, control callus regenerated vigorous roots but less shoots with suppressed growth. On the other hand, callus overexpressing EL5W165A (Figure 4) and EL5Δ4-GFP (Figure S3) did not regenerate roots in any hormonal combinations and retained relatively high shoot-forming ability on the medium containing 2 mg l−1 NAA and no kinetin.

Figure 4.

 Overexpression of mutated EL5 alters responses to auxin and cytokinin during regeneration.
(a) Effects of auxin and cytokinin on the differentiation of callus cells. Calli transformed with only vector as control (left) and EL5W165A (right) were cultured on the regeneration medium with different α-naphthalene acetic acid and kinetin contents. The numbers of calli regenerating shoot with root (white bars), root only (gray bars), and shoot only (black bars) are presented as percentages of the total numbers of calli placed on the medium. The total numbers of calli counted are indicated in parenthesis. The average scores from two independent lines are plotted. The examination was repeated three times and typical results are shown.
(b) Representative phenotypes of the regenerating calli described in (a).

EL5 is a highly unstable protein

We conducted Western blot analysis to characterize the EL5 protein. EL5 was not detected in the non-transformed callus, shoot base where the level of mRNA for EL5 is high (data not shown), and callus overexpressing EL5(wt) (Figure 5a), suggesting that EL5 is rapidly degraded. On the other hand, immunoreactive bands were detected in the callus transformed with EL5C153A and EL5W165A, which lack the E3 activity. Furthermore, an intense band was detected in the cell overexpressing EL5Δ4 that lacks the entire RFD (Figure 5a). These results imply that the level of EL5 is self-regulated through E3 activity.

Figure 5.

 Mutations in RING-H2 finger domain of EL5 enhance its stability.
(a) Stability of EL5 protein. Total extracts from transgenic callus with only vector as control (V), EL5(wt), EL5C153A, EL5W165A and EL5Δ4 were subjected to Western blotting with anti-EL5 antibody. Arrows indicate bands corresponding to EL5C153A and EL5W165A (upper arrow) and EL5Δ4 (lower arrow).
(b) Effects of MG132 treatment on EL5 stability. Callus cells transformed with EL5(wt) and EL5W165A were treated with 50 μm MG132 (+) and DMSO (−). Total extracts were subjected to Western blotting with anti-EL5 antibody. An arrow indicates EL5W165A.
(c) Effect of K122A mutation on EL5 stability. Total extracts from callus cells transformed with indicated genes were analyzed by Western blotting using anti-EL5 antibody. An arrow indicates the mutated EL5.
Major bands on the gel stained with Coomassie brilliant blue (CBB) are displayed to show the loading equality.

To elucidate whether the instability is mediated by self-ubiquitination, followed by degradation in the 26S proteasome, we examined the effect of MG132, a proteasome inhibitor (Figure 5b). EL5W165A accumulation after MG132 treatment indicates that MG132 was incorporated and acted in callus cells; it implies a mechanism that degrades abnormal or foreign proteins by the 26S proteasome. On the other hand, treatment with MG132 did not cause the accumulation of the EL5(wt) protein. EL5 contains only one lysine residue (K122) and the substitution of this lysine with alanine did not affect the stability observed by Western blotting (Figure 5c). These results indicate that the instability of EL5 is not caused by the ubiquitination at K122 and subsequent degradation in the 26S proteasome system.

EL5 functions at the plasma membrane

It was inferred that EL5 is a membrane protein because of a transmembrane domain in the N-terminal (region I). As the level of wild-type EL5 protein is extremely low, as mentioned above, we examined its cellular localization using the mutated EL5 that was stabilized because of the lack of the RFD (region IV), and was fused with GFP (Figure 6a). The introduced EL5-GFP fusion proteins were monitored by Western blot analysis and microscopic observation of GFP fluorescence. As shown in Figure 6b,c, EL5Δ4-GFP was detected in the microsomal fraction (MF), and fluorescence of GFP was observed at the PM. Transformed callus expressing EL5Δ14-GFP, which lacks the transmembrane domain (region I) in addition to the RFD, showed GFP fluorescence in cytosol, and immunoreactive bands were mainly detected in the water-soluble fraction (SF), suggesting that region I anchors EL5 to the PM. EL5Δ4-GFP transformants showed a rootless phenotype, whereas EL5Δ14-GFP transformants regenerated root normally (Figure 6d). These results strongly suggested that EL5 functions in the PM.

Figure 6.

 EL5 without E3 activity is localized in the plasma membrane.
(a) Schematic structure of EL5 and constructs used for cellular localization analysis. Dotted lines indicate the deleted regions. Numbers indicate amino acid position of borders of each region. TM, transmembrane domain; B, region rich in basic amino acids; con, conserved region among the ATL family; RING, RING-H2 finger domain; C-terminal, highly diverse region among the ATL family.
(b) Western blot analysis of microsomal fraction and water-soluble fraction from callus cells with anti-EL5 and anti-GFP antibodies. Δ4, EL5Δ4-GFP; Δ24, EL5Δ24-GFP; Δ14, EL5Δ14-GFP. Asterisks indicate non-specific bands.
(c) Fluorescent images of a small clump of the suspension-cultured cells expressing EL5Δ4-GFP, EL5Δ24-GFP and EL5Δ14-GFP.
(d) Degrees of root formation of about 50 independent callus lines introduced with each construct shown on the left.

We also analyzed transformants expressing EL5Δ24-GFP, which lacks the basic amino acid rich region (region II), in addition to the RFD. Immunoreactive bands were detected in the MF, and GFP fluorescence was observed at the PM as well as EL5Δ4-GFP callus (Figure 6b,c). However, the transformants expressing EL5Δ24-GFP regenerated normal roots in contrast to the EL5Δ4-GFP callus (Figure 6d), implying that region II functions to determine a membrane topology of EL5 based on a positive-inside rule (von Heijne and Gavel, 1988). This hypothesis is supported by an algorithmic analysis for topology prediction using SPLIT 4.0 (http://split.pmfst.hr/split/4/), THUMBUP (http://sparks.informatics.iupui.edu/Softwares-Services_files/thumbup.htm) and PSORT (http://www.psort.org/). They predicted that full-length EL5 is a type-I membrane protein, whereas EL5 without region II is a type-II membrane protein (C-terminal is outside the cell).

Region V is necessary to confer the dominant-negative phenotype

We assumed that the C-terminal region (region V) neighboring the RFD participates in substrate recognition because C-terminal sequences of the ATL families are highly diverse and are likely to confer substrate specificity, although these regions contain no significant motifs for protein–protein interactions. Although plants overexpressing EL5C153A and EL5W165A showed a rootless phenotype (Figure 1), calli overexpressing EL5C153AΔ5 and EL5W165AΔ5, which lack region V, regenerated normal roots as well as the EL5Δ5 plants (Figure 7). We assessed root formation of the transformants that express region V either in the PM (EL5W165AΔ2, EL5Δ4 and EL5Δ34) or in cytosol (EL5W165AΔ1 and EL5Δ1234). Based on the results shown in Figure 6, EL5W165AΔ2 was predicted to be located in the PM with the extracellular C-terminal. Accumulation of the mutated EL5 proteins was confirmed by Western blot analysis or microscopic observation of GFP fluorescence (Figure 5; Figure S4). As shown in Figure 7, only mutants containing regions I, II and V together conferred the dominant-negative phenotype, suggesting that a target of EL5 is membrane-anchored or membrane-associated cytoplasmic protein and interacts with region V.

Figure 7.

 Role of domains in the dominant-negative phenotype.
Constructs used for the root-formation assay are displayed on the left. Gray boxes with an asterisk indicate the RING-H2 finger domain without E3 activity; dotted lines indicate the deleted regions. Degrees of root formation of about 50 independent callus lines introduced with each construct are shown on the right. TM, transmembrane domain; B, region rich in basic amino acids; con, conserved region among the ATL family; RING, RING-H2 finger domain; C-terminal, highly diverse region among the ATL family.


Previous structural and biochemical studies have demonstrated that rice EL5 is a single-subunit E3 containing the RFD that interacts with E2 of an Ubc4/5 subfamily (Katoh et al., 2003; Takai et al., 2002). We aimed to elucidate the physiological functions of EL5 using transgenic rice plants. Rice seedlings overexpressing the wild-type EL5 did not show obvious phenotypic changes, probably because of a mechanism to downregulate the level of EL5 protein (discussed later) and/or other limiting factor(s) for EL5 functioning. We did not find phenotypic changes in about 60 independent RNAi transformants, probably because only very small quantities of protein are needed and the reduction of EL5 possibly results in the increased stability of protein that is produced, or the knock-down of EL5 may be compensated by other proteins.

There are several reports of transgenic plants in which overexpression of the inactive E3 causes dominant-negative effects (Xie et al., 2002; Zhang et al., 2005). One explanation for the effect is that the excessive inactive E3 competitively inhibits the endogenous E3 function by trapping target protein(s). We therefore produced transgenic rice plants overexpressing mutant EL5 proteins, which are impaired in E3 activity, and consequently we would expect the inhibition of endogenous EL5 function(s). There was a good correlation between E3 activity of the introduced EL5 and root formation (Figure 1b). Transgenic rice cells accumulating mutant EL5 (mEL5) differentiated shoots but not roots. The dominant-negative phenotype is not the result of non-specific disturbance by the overexpression, because accumulation of the mutant protein without region V and ectopic accumulation of region V resulted in normal regeneration of root (Figure 7; Figure S4). It is also unlikely that the ectopic expression of mEL5 by CaMV 35S promoter caused the rootless phenotype, because the deficient root formation was also observed when EL5W165A was driven by its native promoter (Y. Nishizawa, H. Nakamura and H. Ichikawa, unpublished data). The dominant-negative phenotypes strongly suggest that mEL5s inhibit the activity of the whole complex in which EL5 is involved, and this activity is crucial for the development of crown and lateral root in rice. The root-specific phenotypic change suggests that the target of EL5 exists or interacts with EL5 only in roots. As transgenic Arabidopsis introduced with EL5(wt) or EL5C153A matured normally (H. Ichikawa and Y. Nishizawa, unpublished data), EL5 probably fails to recognize the Arabidopsis analogue of its target, or it may function in the rice-specific root system.

Model for role of EL5 in root development

Although many mutants and genes associated with root development have been reported in Arabidopsis, few have been reported in rice. Rice mutants, crl1, crl2 and arl1 are defective in crown root formation (Inukai et al., 2001, 2005; Liu et al., 2005). The crl1 and arl1 mutants do not form crown root primordia. Crl1 encodes a member of the plant-specific AS2/LOB protein family and functions as a mediator linking the negative transcription factors AUX/IAAs to the initiation of crown and lateral root development (Inukai et al., 2005). ARL1 is also an auxin-responsive AS2/LOB protein involved in auxin-mediated initiation of crown root primordia (Liu et al., 2005). The crl2 mutant arrests root growth after normal primordial development, but does not show any necrosis, as seen in mEL5 plants, and exhibits severe morphological abnormality in shoots (Inukai et al., 2001). Overexpression of the rice WUSCHEL-type homeobox gene, QHB, results in a crown rootless phenotype because of lack of the primordia formation, and morphologically affects aerial parts including SAM (Kamiya et al., 2003). On the other hand, in mEL5 plants, initiation of root primordia and differentiation of tissues constituting root apical meristem (RAM) occur while the cell proliferation is impaired and cell death is caused, strongly suggesting that EL5 preferentially acts in the root development after the initiation of root primordia through maintaining viability of the meristematic cells. Thus, EL5 participates in the different steps of root formation to CRL1, CRL2, ARL1 and QHB.

It has been suggested that auxin is required in at least two steps in lateral root development in Arabidopsis: (1) initiation of cell division in the pericycle, and (2) promotion of cell division and maintenance of cell viability in the developing lateral root (Celenza et al., 1995). The mechanism underlying the formation of the crown root in rice partly shares that of the lateral root in Arabidopsis (Kamiya et al., 2003). Thus, EL5 may be involved in the latter step in the downstream of auxin action. This idea is supported by the observation that exogenously supplied auxin failed to rescue the rootless phenotype (Figure 4).

Unlike auxin, cytokinin promotes shoot development but inhibits root elongation by reducing the number of dividing cells and the size of root meristem (Beemster and Baskin, 2000). Treatment of rice root with kinetin inhibits root growth accompanied by browning, especially in RAM (Figure S5). We also found that, on the cytokinin-excess medium, shoots regenerated from the control calli formed brown crown roots without lateral roots and showed rootless phenotypes occasionally (Figure 4). The similarity between the cytokinin action on rice root and the inhibitory effects of mEL5s, and the upregulation of EL5 expression by cytokinin, imply that EL5 is involved in the cytokinin action in root cells. Crown roots of mEL5 plants cultured in hormone-free medium differentiate relatively many lateral root primordia near the root tip (Figure 2b), which is inconsistent with the phenotype of the cytokinin-treated control plants. Cytokinin is synthesized in the root cap and contributes root apical dominance, and inhibits differentiation of lateral root primordia in the elongation zone (Aloni et al., 2006). We assume that a loss of the root cap function resulting from cell death caused the ectopic formation of lateral root primordia in mEL5 plants.

JA treatment markedly induced EL5 expression in root, as did 6-benzylaminopurine (BAP) (Figure 3). JA and methyl JA negatively affect primary root growth as well as cytokinin. However, the biological relevance of JA in root development remains unclear. In rice, treatment with 100 μm JA results in brownish roots (Michéet al., 2006), although a low JA concentration (≤ 3 μm) increases the number of crown roots (Moons et al., 1997). Transcriptome analyses of Arabidopsis and maize root tips revealed that genes encoding steps in the JA biosynthesis pathway were upregulated in the root cap, suggesting that JA has certain functions in the root cap (Birnbaum et al., 2003; Jiang et al., 2006). On the other hand, the role of JA in the regulation of cell death during biotic- and abiotic-stress is well-characterized (Overmyer et al., 2003). As EL5 was originally identified as a chitin elicitor-responsive gene (Takai et al., 2001), and EL5 expression was induced by JA treatment in shoot (Figure 3), EL5 might contribute to the mechanisms shared by the maintenance of RAM and defense response.

The Arabidopsis root meristemless 1 (rml1)/cadmium sensitive2 mutant (Vernoux et al., 2000) cannot sustain post-embryonic root growth because of complete inhibition of cell division. The RML1 gene encodes γ-glutamylcysteine synthase, which catalyses the first step in the biosynthetic pathway for glutathione (GSH); exogenously applied GSH rescues the rml1 phenotype. As a significant function of GSH is to act as a scavenger of ROS and as a redox buffer (Noctor and Foyer, 1998), the rml1 mutant suggests the significance of redox regulation in RAM. Recent insights into auxin action indicate that polar transport of auxin causes a highly oxidized environment in the quiescent center (QC), which is essential for maintaining RAM (Jiang and Feldman, 2005; Jiang et al., 2003). Moreover, not only auxin but also cytokinin and JA exert various cellular responses through the production of ROS or nitric oxide (Orozco-Cardenas and Ryan, 1999; Pagnussat et al., 2002; Tun et al., 2001). Constitutive oxidization would cause cellular damage and death; thus, some mechanisms to protect cells around QC should exist in RAM. EL5 could function in the context of the redox homeostasis because severe necrosis was observed centered on the QC of crown root primordia in mEL5 plants (Figure 2; Figure S1).

Taken together, we hypothesize a model that EL5 critically contributes to root development by mediating the degradation of protein(s) generated in root cells after the actions of auxin, cytokinin and probably JA, and that the degradation of the target is essential to avoid meristematic cell death leading to the immature termination of the root (Figure 8).

Figure 8.

 Hypothetical model for the role of EL5 in root development.
X represents a putative target of EL5, which is postulated to be degraded after the ubiquitination in this model. Arrows with an asterisk indicate upregulation of EL5 transcripts in root. QC, quiescent center; RAM, root apical meristem.

Rapid degradation of EL5

We found that EL5 is a highly unstable protein and that the RFD regulates the stability (Figure 5a). Almost all known RING-type E3s are capable of mediating their own ubiquitination in vitro (Fang and Weissman, 2004). In some cases, the self-ubiquitination results in rapid degradation in vivo and the biological relevance has been shown. For example, ubiquitin ligase Mdm2 mediates the downregulation of the tumor suppressor p53 via the RFD to prevent cell death. Mdm2 also regulates its level by self-ubiquitination, and in unstressed cells, binding with ubiquitin-specific protease HAUSP and Daxx inhibits the self-ubiquitination. DNA damaging leads to disassociation of HAUSP and Daxx from Mdm2, which promotes the Mdm2 self-ubiquitination followed by proteasome-dependent degradation. As a result, p53 accumulates and executes apoptosis (Tang et al., 2006). Similarly, anti-apoptosis protein XIAP binds caspases in the N-terminal BIR domain, and its RFD facilitates ubiquitination and degradation of the caspases to inhibit cell death. In response to apoptosis stimuli, Smac/DIABLO protein binds to the BIR domain, and enhances XIAP self-ubiquitination and its proteasome-dependent degradation, which leads to the accumulation of the caspase followed by apoptosis (Lee and Peter, 2003). Thus, the cis- and trans-E3 activities involving the proteasome-dependent protein degradation have opposing effects on cell fate.

The EL5 instability was dependent on its E2-interacting ability, like Mdm2 and XIAP, but it was not a result of the ubiquitination at K122 and the subsequent degradation in the 26S proteasome (Figure 5b,c). For most PM proteins, such as the growth hormone receptors, degradation occurs in the lysosome rather than proteasome because the proteasome itself is not capable of removing proteins from the membrane (Govers et al., 1999). As EL5 is a PM protein, its degradation might occur in lysosome. The mechanism and biological relevance of the rapid degradation of EL5 needs to be addressed.

In conclusion, our data indicate that EL5 plays a major role as an E3 for the maintenance of cell viability in RAM. Unlike other RING-type E3s reported in animals, stability of EL5 is not regulated by self-ubiquitination. At present, we are unaware of the fate of the protein(s) that probably interacts with region V of EL5 and is ubiquitinated around the PM. To reveal the molecular mechanisms of the EL5 action in root development, it is essential to identify the target protein(s) and clarify how EL5 is regulated.

Experimental procedures

Plant materials and hormonal treatments

Rice (Oryza sativa L. japonica cv. Nipponbare) seeds were sterilized in an antiformin solution (in which the available chlorine is about 0.5%) for 20 min and imbibed in sterilized water for 2 days at 25°C in the dark. The seeds were germinated on a mesh floating in the culture bottle with tap water and grown at 28°C under continuous light in a growth chamber for 7 days. To examine the effect of various hormones, 9-day-old seedlings were soaked in a solution of 100 μm ABA, 50 μm BAP, 50 μm gibberellin A3, 100 μm JA, 200 μm ethephon, 100 μm SA, 100 μm indole-3-acetic acid and 1 μm brassinolide for 6 h. The treated samples were harvested after being divided into three parts: shoot cut 1 cm above the seed; 2 cm of basal region including 1 cm each of shoot and root; and elongated root longer than 1 cm. These were then immediately frozen in liquid nitrogen and kept at −80°C until isolation of the total RNA.

RNA isolation and Northern blotting

Total RNA was isolated from the three parts of the seedlings using the Qiagen RNeasy plant mini kit (Qiagen, http://www.qiagen.com). Total RNA (10 μg) was separated in denaturing 1.5% agarose gel, transferred to nylon membranes (Amersham Hybond N; Amersham, http://www.amersham.com), and hybridized with a digoxigenin (DIG)-labeled probe. For the preparation of the DIG-labeled probe, a part of the EL5 cDNA (accession number AB045120) including 3′-UTR (corresponding to the region of nt. 529–1181) was subcloned into pGEM-T vector (Promega, http://www.promega.com). After linearization of the construct by NcoI digestion, DIG-labeled cRNA was synthesized by SP6 RNA polymerase using DIG RNA labeling kit (Roche Diagnostics, http://www.roche.com) with RNase inhibitor (Toyobo, http://www.toyobo.co.jp/e/), and then the riboprobe was purified by LiCl precipitation. Hybridization was performed in DIG Easy Hyb (Roche Diagnostics) at 68°C overnight. The membrane was treated with CSPD (Roche Diagnostics) using DIG Luminescent Detection Kit (Roche Diagnostics) and then exposed to X-ray film (Kodak Biomax; Kodak, http://www.kodak.com).

Plasmid construction and rice transformation

The coding region of EL5 was amplified from the EL5 cDNA clone by PCR using primers 5′-CACCTCTAGAGTCGATTATTATGGTGCGGG-3′ and 5′-GGGATATCTCAATTCCGGACATGCGCGGCG-3′. The purified PCR product was cloned into pENTR/SD/D-TOPO using TOPO-cloning technology (Invitrogen, http://www.invitrogen.com) to produce pENTR-EL5. Derivative genes possessing a single point mutation were created by site-directed mutagenesis using the QuickChange site-directed mutagenesis kit (Stratagene, http://www.stratagene.com) with pENTR-EL5 as a template. Then, the DNA fragments were prepared from the pENTR clones by XbaI and EcoRV digestion and inserted into pBI333-EN4. This binary vector contains a hygromycin phosphotransferase gene cassette as a selection marker, an enhanced promoter (EN4) derived from CaMV 35S promoter, and a nopaline synthase terminator in the T-DNA region (Nishizawa et al., 1999), and was used for the constitutive-overexpression of foreign genes.

To create internal deletion constructs as well as sGFP(S65T) fusion constructs, a PCR-amplified N-terminal fragment with an XbaI site (5′ side) and an XhoI or NheI site (3′ side) was ligated into the XbaI and KpnI sites of pBluescript SK + (Stratagene) together with a PCR-amplified C-terminal fragment with an XhoI or NheI site (5′ side) and a KpnI site (3′ side). PCR-primers used for the deletion constructs are listed in Table S1. Then, the DNA fragments digested by XbaI and KpnI were cloned into pBI333-EN4. A sGFP(S65T) clone (Niwa et al., 1999) was kindly provided by Dr Y. Niwa from University of Shizuoka and used as a PCR template. Sequences of the all PCR-amplified constructs were verified by DNA sequencing.

The constructed binary vectors were transferred into A. tumefaciens strain EHA105 (Hood et al., 1986) by electroporation. Transformation of rice (cv. Nipponbare) was performed as described previously (Itoh et al., 2003), except 12.5 mg l−1 meropen (Dainippon Sumitomo Pharma Co., Ltd., http://www.ds-pharma.co.jp/english/) was added instead of carbenicillin to eliminate A. tumefaciens in the N6D selection medium.

Root formation assay

About 60 independent callus lines per introduced plasmid construct were cultured on the regeneration medium containing 2 μg l−1α-naphthalene acetic acid, 2 mg l−1 kinetin, 50 mg l−1 hygromycin and 12.5 mg l−1 meropen for about 3 weeks with one medium change. Then, the lines regenerating shoots were transferred to the hormone-free MS medium containing 35 mg l−1 hygromycin and 12.5 mg l−1 meropen. After 1 week, shoots were transferred to the fresh medium without hygromycin (about eight shoots per line) and cultured for an additional 1 week. Then root formation was observed for every shoot and categorized into three types: shoot with vigorous roots (normal phenotype); shoot with abnormal roots; and shoot without roots (rootless phenotype). It was observed that shoots showing normal phenotype also included transformants that were hygromycin resistant but did not express the foreign EL5 gene. Regenerated shoots transformed with a control vector rarely showed the rootless phenotype. Therefore, if one line contained both normal and rootless shoots, we scored the phenotype of the line as rootless.

Hormonal response test of transgenic callus

Rice callus was transformed with pBI333-EL5W165A, pBI333-EL5Δ4-GFP and pBI333 (control). Hygromycin-resistant callus cells were repeatedly selected from a single resistant callus three times on the N6D selection medium to obtain homogeneous callus lines. Callus lines that highly expressed the introduced EL5 derivatives were selected by Western blot analysis and GFP fluorescence, and were cultured on the regeneration medium containing various combinations of NAA (0, 0.02 and 2 mg l−1) and kinetin (0, 0.2, 2 and 20 mg l−1). After 2 weeks, calli were transferred to the fresh medium and incubated for another 2 weeks. Regeneration was evaluated by counting the number of each callus categorized by regenerating both shoot and root, root only, and shoot only, whether the regenerated shoots and roots were morphologically normal or abnormal.

Preparation of the microsome-rich fraction

Microsome was prepared from freshly harvested cultured cells. Tissue (300 mg FW) was ground in 1 ml of extraction buffer (PBS containing 10% glycerol and 5 mm DTT) with Protease Inhibitor Cocktail (Nacalai Tesque Inc., http://www.nacalai.co.jp/en/) and centrifuged at 1000 g for 15 min at 4°C. The supernatant was centrifuged again at 13 000 g for 15 min at 4°C to remove cell debris, and then an MF was precipitated by centrifugation at 80 000 g for 1 h at 4°C. The supernatant was reserved as a water-SF and the pellet was next treated with 100 mm sodium carbonate (pH 11.5) for 30 min at 4°C to remove peripheral proteins from microsome according to the method described by Nicol et al. (1998). After this alkaline treatment, microsome was precipitated by centrifugation at 80 000 g for 1 h at 4°C; then, the pellet was rinsed with the extraction buffer and finally dissolved in SDS sample buffer for Western blot analysis.

Protein extraction and Western blotting

To prepare the EL5-specific antibody, the C-terminal region (peptide 212–325) was cloned into pET32 Escherichia coli expression vector (Merck Biosciences, http://www.merckbiosciences.com). The recombinant protein with a thioredoxin tag prepared as described previously (Katoh et al., 2003) was used to immunize a rabbit. The rabbit anti-serum was purified using thioredoxin-fixed column to remove antibodies for thioredoxin, and then the EL5-specific antibody was purified using an antigen-fixed column. Rabbit polyclonal antibody to GFP was purchased from BD Bioscience (Clontech, http://www.clontech.com).

Rice cultured cell lines were homogenized with 3 × SDS sample buffer (3% SDS, 15 mm Tris–HCl, pH 6.3, 6% glycerol and 3%β-mercaptoethanol). Samples were boiled for 5 min and centrifuged at 15 000 g for 15 min at 25°C after doubling the dilution. Protein in the supernatant was quantitated with the Bio-Rad Protein dye reagent (Bio-Rad, http://www.clontech.com) using bovine serum albumin as the standard, and 20 μg of protein per lane was separated in 12.5% SDS-polyacrylamide gel. Protein transfer and immunodetection were conducted as described previously (Nishizawa et al., 2003).

To prevent the 26S proteasome activity, calli were submerged in N6D medium containing 50 μm MG132 (stock solution, 10 mm in DMSO; Sigma-Aldrich, http://www.sigmaaldrich.com) for 6 h before homogenization.

Histochemical analysis

To prepare paraffin sections, regenerated shoots were fixed at 4°C in FAA solution (formalin:glacial acetic acid:50% ethanol, 1:1:18) overnight. Fixed samples were dehydrated with graded ethanol series and substituted with xylene. After embedding these samples in Paraplast Plus (Kendall, http://www.kendallhealthcare.com), paraffin blocks were sectioned to 10-μm thickness using a rotary microtome. Sections were observed using a light microscope without staining and after staining with 0.1% toluidine blue. To analyze cellular localization of EL5, small clumps of cells expressing EL5-GFP fusion protein were prepared by culturing the transgenic callus in liquid N6D and were observed under the fluorescence microscope (OLYMPUS AX70; Olympus, http://www.olympus-global.com).

In situ RNA hybridization and immunological detection of the hybridized riboprobes were performed according to the method of Kouchi and Hata (1993), with some modifications. RNA probes were prepared from linearized plasmids with DIG-11-UTP (Roche Diagnostics). A clone of rice homeotic gene OSH1 was kindly supplied by Dr M. Matsuoka of Nagoya University. Tissues were fixed in 4% (w/v) paraformaldehyde and 0.25% glutaraldehyde in 50 mm sodium phosphate buffer (pH 7.2) overnight at 4°C, dehydrated through a graded ethanol series and then a t-butanol series, and finally embedded in Paraplast Plus (Kendall). Microtome sections (8–10-μm thick) were hybridized with the DIG-labeled probes at 52°C overnight. After washing, the slides were soaked with buffer 1 (150 mm NaCl, 100 mm Tris–HCl, pH 7.5) and then incubated for 30 min in 50% normal rabbit serum in buffer 1 containing 1% Tween 20 followed by incubation with the diluted anti-DIG alkaline phosphatase conjugate (1:1000) in buffer 1 containing 0.1% BSA for 1–2 h.


We thank Iain Wilson for critically reading the manuscript, Hiroyuki Hirano and Shinichiro Sawa for comments on phenotypic analysis, Ryoji Takai for purifying anti-EL5 antibody, Hidemitsu Nakamura for cloning EL5 promoter, and Osamu Ishizaki-Nishizawa for helpful discussion. We also thank Emi Nakajima, Hiroko Kurano, Kyoko Iwasaki and Kyoko Irie for daily technical assistance. This work was supported by a grant (Rice Genome Project, IP-4003) from the Ministry of Agriculture, Forestry and Fisheries of Japan, and by a grant from the Program for Promotion of Basic Research Activities for Innovative Biosciences.