---
pmid: '17586653'
title: Ubiquitin lysine 63 chain forming ligases regulate apical dominance in Arabidopsis.
authors:
- Yin XJ
- Volk S
- Ljung K
- Mehlmer N
- Dolezal K
- Ditengou F
- Hanano S
- Davis SJ
- Schmelzer E
- Sandberg G
- Teige M
- Palme K
- Pickart C
- Bachmair A
journal: Plant Cell
year: '2007'
full_text_available: true
full_text_extraction_method: html
pmcid: PMC1955712
doi: 10.1105/tpc.107.052035
---

# Ubiquitin lysine 63 chain forming ligases regulate apical dominance in Arabidopsis.
**Authors:** Yin XJ, Volk S, Ljung K, Mehlmer N, Dolezal K, Ditengou F, Hanano S, Davis SJ, Schmelzer E, Sandberg G, Teige M, Palme K, Pickart C, Bachmair A
**Journal:** Plant Cell (2007)
**DOI:** [10.1105/tpc.107.052035](https://doi.org/10.1105/tpc.107.052035)
**PMC:** [PMC1955712](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1955712/)

## Abstract

1. Plant Cell. 2007 Jun;19(6):1898-911. doi: 10.1105/tpc.107.052035. Epub 2007
Jun  22.

Ubiquitin lysine 63 chain forming ligases regulate apical dominance in 
Arabidopsis.

Yin XJ(1), Volk S, Ljung K, Mehlmer N, Dolezal K, Ditengou F, Hanano S, Davis 
SJ, Schmelzer E, Sandberg G, Teige M, Palme K, Pickart C, Bachmair A.

Author information:
(1)Department of Plant Developmental Biology, Max Planck Institute for Plant 
Breeding Research, Cologne, Germany.

Lys-63-linked multiubiquitin chains play important roles in signal transduction 
in yeast and in mammals, but the functions for this type of chain in plants 
remain to be defined. The RING domain protein RGLG2 (for RING domain Ligase2) 
from Arabidopsis thaliana can be N-terminally myristoylated and localizes to the 
plasma membrane. It can form Lys-63-linked multiubiquitin chains in an in vitro 
reaction. RGLG2 has overlapping functions with its closest sequelog, RGLG1, and 
single mutants in either gene are inconspicuous. rglg1 rglg2 double mutant 
plants exhibit loss of apical dominance and altered phyllotaxy, two traits 
critically influenced by the plant hormone auxin. Auxin and cytokinin levels are 
changed, and the plants show a decreased response to exogenously added auxin. 
Changes in the abundance of PIN family auxin transport proteins and synthetic 
lethality with a mutation in the auxin transport regulator BIG suggest that the 
directional flow of auxin is modulated by RGLG activity. Modification of 
proteins by Lys-63-linked multiubiquitin chains is thus important for 
hormone-regulated, basic plant architecture.

DOI: 10.1105/tpc.107.052035
PMCID: PMC1955712
PMID: 17586653 [Indexed for MEDLINE]

## Full Text

Abstract

Lys-63–linked multiubiquitin chains play important roles in signal transduction in yeast and in mammals, but the functions for this type of chain in plants remain to be defined. The RING domain protein RGLG2 (for RING domain Ligase2) from Arabidopsis thaliana can be N-terminally myristoylated and localizes to the plasma membrane. It can form Lys-63–linked multiubiquitin chains in an in vitro reaction. RGLG2 has overlapping functions with its closest sequelog, RGLG1, and single mutants in either gene are inconspicuous. rglg1 rglg2 double mutant plants exhibit loss of apical dominance and altered phyllotaxy, two traits critically influenced by the plant hormone auxin. Auxin and cytokinin levels are changed, and the plants show a decreased response to exogenously added auxin. Changes in the abundance of PIN family auxin transport proteins and synthetic lethality with a mutation in the auxin transport regulator BIG suggest that the directional flow of auxin is modulated by RGLG activity. Modification of proteins by Lys-63–linked multiubiquitin chains is thus important for hormone-regulated, basic plant architecture.

INTRODUCTION

Ubiquitylation, the posttranslational attachment of ubiquitin to substrate proteins, proceeds through an enzymatic cascade of activation and transfer, which is conserved in all eukaryotes. Ubiquitin-activating enzyme (UBA or E1) activates ubiquitin by thioester formation between a catalytic Cys residue of E1 and the C-terminal Gly of ubiquitin. Activated ubiquitin is transferred to a ubiquitin-conjugating enzyme (UBC or E2), which associates with a ubiquitin ligase (E3). The E2-E3 complex confers substrate specificity and catalyzes isopeptide bond formation between the ɛ-amino group of internal Lys residues in the substrate and the ubiquitin C terminus. Ubiquitin ligases often contain so-called RING domains, which serve as docking sites for E2 association ( Pickart, 2001 ). Comparison of yeast, animal, and plant ubiquitylation enzymes suggests orthology relationships among UBCs but identifies few clear orthologs among the numerous ligase proteins ( Bachmair et al., 2001 ; Kosarev et al., 2002 ; Smalle and Vierstra, 2004 ; Kraft et al., 2005 ; Stone et al., 2005 ).

Ubiquitin can itself be a substrate for ubiquitylation; it has seven Lys residues that can serve as sites of modification ( Peng et al., 2003 ; Pickart and Fushman, 2004 ; Hochstrasser, 2006 ). The most prominent chain type is formed by linkage of ubiquitin to Lys-48 of another ubiquitin moiety. Lys-48–linked chains target substrates for recognition and degradation by the proteasome. By contrast, Lys-63–linked chains confer a different, nonproteolytic fate onto a substrate. For instance, ubiquitin Lys-63 chains are involved in DNA repair ( Hoege et al., 2002 ) and in receptor signaling ( Chen, 2005 ) in animals. The formation of Lys-63–linked chains requires a heterodimeric E2 enzyme, formed by canonical E2 Ubc13 and another UBC fold protein, called Uev1A (for Ubiquitin enzyme variant 1A) in mammals and Mms2 (for Methyl methane sulfonate–sensitive2) in budding yeast ( Hofmann and Pickart, 1999 ).

In plants, ubiquitylation is involved in many aspects of development and stress response ( Schwechheimer and Schwager, 2004 ; Smalle and Vierstra, 2004 ; Lechner et al., 2006 ; Dreher and Callis, 2007 ). Of particular interest is the contribution to hormonal regulation. Response to the growth regulator auxin ( Woodward and Bartel, 2005 ; Leyser, 2006 ; Quint and Gray, 2006 ; Teale et al., 2006 ; Dreher and Callis, 2007 ) requires the activity of an SCF (for Skp1, Cullin, F-box receptor)–type ubiquitin ligase ( Dharmasiri and Estelle, 2004 ; Lechner et al., 2006 ; Parry and Estelle, 2006 ). Elements of auxin transport ( Blakeslee et al., 2005 ; Paponov et al., 2005 ; Fleming, 2006 ; Kramer and Bennett, 2006 ) may depend on ubiquitin's role in endocytosis ( Dupré et al., 2004 ; d'Azzo et al., 2005 ; Clague and Urbé, 2006 ).

Apical dominance, the growth-inhibitory effect of the main shoot on the initials of side shoots, is a classical trait governed by auxin (for recent reviews, see Leyser, 2005 , 2006 ; Dun et al., 2006 ). Most mutants with decreased auxin content or sensitivity also display reduced apical dominance and therefore have a more bushy stature than wild-type plants. Basipetal transport of auxin from the main shoot is considered to be essential for apical dominance. Directional transport of auxin is also involved in many other differentiation processes, such as vascular tissue formation, and is a focus of intense research ( Teale et al., 2006 ). It is mediated by both influx and efflux systems to channel auxin from cell to cell in a directional manner. Efflux depends on the group of PIN-FORMED (PIN) proteins, which determine auxin flow direction by their asymmetric distribution in the plasma membrane ( Paponov et al., 2005 ; Kramer and Bennett, 2006 ).

In this work, we describe an Arabidopsis thaliana mutant that essentially lacks apical dominance and displays changes in auxin and cytokinin signaling. The mutant has defects in two sequence-related, membrane-associated RING domain proteins, RGLG1 and RGLG2 (for RING domain ligases 1 and 2). RGLG2 is shown to catalyze ubiquitin Lys-63 chain formation.

Inactivation of Both RGLG1 and RGLG2 Results in the Abolishment of Apical Dominance and the Alteration of Normal Leaf Phyllotaxy

T-DNA insertion alleles in the genes RGLG1 (SALK_011892) and RGLG2 (SALK_062384) were characterized from populations of transformed lines ( Alonso et al., 2003 ). Supplemental Figure 3 online presents a characterization of the mutant alleles and indicates that neither mutant produced full-length RGLG transcript. Together with the findings (presented below) that both insertion alleles were recessive, these data strongly suggested that these alleles display a loss of function. Plants homozygous for either the rglg1 or rglg2 knockout mutation were phenotypically similar to wild-type plants (data not shown), supporting the notion that both genes perform identical functions. However, double mutant combinations encompassing rglg1 and rglg2 insertion alleles showed remarkable developmental alterations.

Plants carrying the homozygous rglg1 mutation in the hemizygous rglg2 / RGLG2 background ( Figure 5B , left plant) were only slightly smaller than wild-type plants ( Figure 5A , left plant), whereas the reverse combination ( rglg1 / RGLG1 rglg2 / rglg2 ) displayed reduced apical dominance ( Figure 5B , right plant). The stronger effect of the loss of both RGLG2 copies and one RGLG1 copy, compared with the reverse combination, is consistent with a redundant function but a higher expression level of RGLG2 compared with RGLG1 . The homozygous rglg1 rglg2 double mutant showed a complete loss of apical dominance, lacking normal dominance of the central apical meristem, and the development of a central inflorescence ( Figure 5C ). All double mutants grown to maturity had >30 shoots of comparable length, suggesting that the loss of apical dominance trait has 100% penetrance. By contrast, the rglg1 rglg2 double mutant expressing RGLG2-GFP under the control of the CaMV 35S promoter looked similar to the wild-type plant ( Figure 5A , right plant versus left plant). rglg1 rglg2 plants containing an RGLG1-GFP expression construct also looked like wild-type plants (data not shown), demonstrating the functional equivalence of both proteins.

Figures 5E, 5G, and 5H to 5J show different stages in the rglg1 rglg2 double mutant life cycle. Although the overall growth of young rglg1 rglg2 seedlings closely resembled that of wild-type plants ( Figure 5D , wild-type, versus Figure 5E , mutant), the early rosette leaves were more elongated ( Figure 5F , wild type, versus Figures 4G and 4H , mutant) and darker ( Figures 5H and 5I ). These early rosette leaves senesce after ∼4 to 6 weeks of growth (visible in Figure 5J ). The first emerging shoots were short and terminated by one or few flowers yielding fertile siliques ( Figure 5I ). Later developing shoots were longer ( Figures 5C and 5J ). Another difference between the wild type and the rglg1 rglg2 mutant was leaf shape and architecture. Mutant leaves were more oblong and often asymmetric and had reduced numbers of veins ( Figure 5K , wild-type leaf; Figure 5L , mutant leaves). Figures 5M and 5N show sections through the shoot meristem region, demonstrating characteristic alterations in phyllotaxy. Whereas leaf initials in the wild type formed a spiral, with an ∼137.5° angle between leaf initials ( Figure 5M ), the rglg1 rglg2 mutant formed two initials opposite to each other and then switched by 90° to form the next pair of opposing leaves ( Figure 5N ). Thus, rglg1 rglg2 plants often had a decussate pattern of leaf emergence, instead of the spiral pattern of wild-type Arabidopsis . Additional phenotypes of aerial plant parts, in particular an increased cell size, are shown in Supplemental Figure 4 online (differences in cell size are also visible in Figures 5M versus 5N ).

Abolishment of the development of the primary shoot meristem and changes in leaf phyllotaxy in the rglg1 rglg 2 double mutant coincided with changes in the expression of genes acting in the control of apical stem cell maintenance. We observed a significant change in the level of WUSCHEL ( WUS ; At2g17950) mRNA, which was approximately fourfold higher in the mutants ( Figure 6 ). However, mRNA levels of other genes of the WUS regulon, including CLAVATA1 (CLV1; encoded by At1g75820) and CLV2 (At1g65380), were unchanged. CLV3 (At2g27250) ( Williams and Fletcher, 2005 ; Kondo et al., 2006 ), the gene encoding a peptide ligand of the CLV1/2 membrane receptor, was elevated by more than twofold. Its increased expression would actually downregulate WUS in the wild type, which was obviously not the case in rglg1 rglg2 . On the other hand, WUS negatively regulates the abundance of the SHOOT MERISTEMLESS ( STM ; At1g62360) transcript in the wild type, which was also the case in the double mutant ( Figure 6 ). These measurements suggested changes in signal transduction between CLV3 perception at the membrane and WUS expression. A mechanistic resolution of the role of RGLGs in the regulation of stem cell maintenance and differentiation in the apical meristem remains a task for further studies. As documented in Supplemental Figure 5 online, rglg1 rglg2 mutants also showed impaired circadian clock function.

Roots of wild-type and mutant plants were similar. We found a moderate (14%) reduction in the main root length of plantlets grown on plates (in one experiment, root length at 4 d after germination was 1.65 ± 0.05 cm for the wild type and 1.42 ± 0.04 cm for the double mutant). Reduced root growth in the presence of the auxin naphthylacetic acid (NAA) at 85 nM concentration maintained the percentile difference, and root curvature in a standard gravitropism assay was apparently unchanged (data not shown).

DISCUSSION

Unlike chains formed by linkage of the ubiquitin C terminus to Lys-48 of another ubiquitin moiety, which mediate proteasomal turnover, those formed by linkage via Lys-63 do not generally lead to substrate degradation. Instead, ubiquitin Lys-63 chains may have roles in cellular signaling. In this work, we present the characterization of RGLG1 and RGLG2, two membrane-associated ubiquitin Lys-63 chain–forming enzymes from Arabidopsis .

Double mutant Arabidopsis plants rglg1 rglg2 differ slightly from wild-type plants at the early seedling stage ( Figures 5D versus 5E ), indicating that the genes play only a minor role in early development. This may result from the fact that Arabidopsis has three additional RGLG2 -related genes ( Stone et al., 2005 ). Postembryonal development of rglg1 rglg2 plants, however, differs significantly from that of wild-type plants in a variety of traits. Many of these traits are known to be critically influenced by the plant growth regulator auxin (reviewed in Woodward and Bartel, 2005 ; Leyser, 2006 ; Quint and Gray, 2006 ; Teale et al., 2006 ). In particular, auxin influences apical dominance ( Leyser, 2005 , 2006 ), phyllotaxy ( Reinhardt et al., 2003 ; Jönsson et al., 2006 ; Smith et al., 2006 ), leaf venation ( Fukuda, 2004 ), and circadian rhythms ( Hanano et al., 2006 ), all of which differ in the rglg1 rglg2 mutants. To establish how auxin responses and concentrations are changed in mutant plants, transcription of hormone-responsive genes and hormone concentrations were determined. rglg1 rglg2 plants have lower auxin (IAA) concentrations ( Figure 8 ). Likewise, auxin-responsive genes display lower expression levels ( Figure 7 ). In contrast with the wild type, the mutant also does not adapt transcription of auxin-responsive genes after exogenous application of auxin ( Figures 7F to 7L ; data not shown).

The best understood module of auxin response regulates transcription. The SCF-TIR1 pathway consists of a soluble auxin receptor, TIR1 ( Dharmasiri et al., 2005 ; Kempinski and Leyser, 2005 ; Tan et al., 2007 ), which binds to transcriptional regulators (AUX/IAA proteins) in an auxin-dependent way to mediate their turnover through ubiquitylation by ubiquitin ligase SCF TIR1 (reviewed in Dharmasiri and Estelle 2004 ; Parry and Estelle 2006 ). We made combinations of two SCF-TIR1 pathway mutants, axr1-12 and axr3-1 , with rglg1 rglg2 mutants ( Figure 9 ). Phenotypic analysis shows that mutations in RGLG1/2 are additive with, or partially suppress the effects of, mutations in the SCF-TIR1 pathway. The most plausible interpretation of this result is that RGLG1 and RGLG2 proteins are primarily elements involved in adjusting intracellular auxin levels, whereas the SCF-TIR1 pathway is responsible for transcriptional responses to the intracellular auxin concentration. Therefore, RGLG1 and RGLG2 define hormone regulatory elements distinct from the SCF-TIR1 pathway. This interpretation is supported by the fact that all known elements of the SCF-TIR1 pathway are soluble, whereas RGLG1/2 are membrane-associated ( Figures 3 and 4 ). By contrast, anthocyanin accumulation and the death of the triple mutant tir3-1 rglg1 rglg2 at later stages of development indicate that the respective proteins impinge on the same process(es) ( Figure 9 ). The large membrane-associated protein, BIG/TIR3, was described as a regulator of polar auxin transport ( Gil et al., 2001 ).

Although mutations in BIG are pleiotropic ( Kanyuka et al., 2003 ; Desgagné-Penix et al., 2005 ; López-Bucio et al., 2005 ; Kasajima et al., 2007 ), many phenotypes can be linked to auxin transport defects. In particular, the tir3-1 mutation was shown to reduce GA sensitivity ( Desgagné-Penix et al., 2005 ). However, a link was previously found between auxin transport and GA signaling ( Fu and Harberd, 2003 ). Furthermore, GA defects are not normally linked to cell death, so that we provisionally attribute the lethality of the triple mutant rglg1 rglg2 tir3-1 to auxin transport defects, and not to interference with GA hormonal pathways by either tir3-1 or rglg1 rglg2 mutations.

Interestingly, a number of results can be explained most easily by the assumption that rglg1 rglg2 mutants are compromised in auxin transport. First, auxin concentration, which was measured separately in shoots and roots ( Figure 8 ), is more significantly decreased in roots, which are a sink tissue for auxin, and at least partially depend on auxin import from shoots. Second, pharmacological inhibition of auxin transport has a lower effect on transcription in rglg1 rglg2 mutants than in wild-type plants ( Figures 7F to 7L ), which may be expected if the auxin transport capacity is already lowered by the mutations in RGLG1 and RGLG2. Third, triple mutants with inactive RGLG1, RGLG2, and BIG die during somatic development ( Figures 9F to 9I ), which may indicate that auxin transport needs at least one of the accessory factors, RGLG1, RGLG2, or BIG, in order to allow plant survival. Fourth, an influence of RGLG proteins on auxin transport capacity is also suggested by the finding that a protein fusion between the auxin transport component PIN1 and GFP, expressed from the PIN1 promoter on a transgene, has reduced abundance in mutant plants ( Figure 10 ). The related PIN2 protein, however, seems only marginally affected ( Figure 10 ). PIN family auxin transport proteins are regulated by a network of currently incompletely defined feedback loops that include the assessment of auxin concentration ( Vieten et al., 2005 ; Sauer et al., 2006 ). The contribution of RGLGs to PIN1 regulation may, at least in part, be based on direct physical contact, because RGLG2 and PIN1 can interact in the yeast two-hybrid system (see Supplemental Figure 8 online).

The rglg1 rglg2 mutants also have higher cytokinin concentrations ( Figure 8 ; see Supplemental Figures 6 and 7 online), and cytokinin-responsive genes display a higher expression than in wild-type plants ( Figure 7 ). Furthermore, the upregulation of the gene WUS ( Figure 6 ), which encodes a critical component of a stem cell maintenance regulatory loop in the apical meristem region, may be a consequence of the cytokinin responsiveness of the WUS-STM regulon ( Mizuno, 2004 ; Leibfried et al., 2005 ). Similarly, suppression of the shoot phenotype of the axr3-1 mutation by rglg1 rglg2 in the triple mutant ( Figure 9 ) may be connected to increased cytokinin in the rglg double mutant background, because many phenotypes of the axr3-1 mutation can be partially suppressed by exogenous supply of cytokinin to axr3-1 single mutants ( Leyser et al., 1996 ). The changes in cytokinin could be either a direct result of the absence of RGLG1 and RGLG2 in the mutant or an indirect consequence of differences in auxin concentration. This is because there is a crosstalk between auxin and cytokinin, so that mutants in auxin response can also have changes in cytokinin ( Nordström et al., 2004b ; Rashotte et al., 2005 ; Tanaka et al., 2006 ).

How might RGLG1/2 exert their functions? Interaction partners, protein structure, and in vitro activity of these proteins hint at a possible mode of action. Both proteins interact in a yeast two-hybrid assay with UBC35, a plant homolog of the yeast ubiquitin conjugation enzyme Ubc13. The latter protein forms a heterodimer with Mms2 to catalyze the formation of ubiquitin Lys-63–linked chains ( Hofmann and Pickart, 1999 ). In the presence of ubiquitin, ubiquitin-activating enzyme, UBC35, and the Arabidopsis Mms2 homolog MMZ2, RGLG2 enhances the formation of Lys-63–linked ubiquitin chains ( Figure 2 ).

An interesting facet of RGLG activity is that, in the yeast two-hybrid system, UBC35 does not interact with the full-length RGLG2 protein but only with a fragment that contains the RING domain ( Figure 1 ). We interpret this finding to indicate that, in vivo, interaction of RGLG2 with UBC35/MMZ2, and therefore enzyme activity, is regulated by additional proteins. The recent finding that another yeast two-hybrid binding partner of RGLG2 can stimulate its in vitro activity is consistent with this conjecture (X.-J. Yin and A. Bachmair, unpublished data).

While we cannot rule out that, together with other UBC enzymes, RGLGs may form ubiquitin linkage types other than via Lys-63, the data presented clearly show that RLGL2 ubiquitylation in the presence of UBC35 and MMZ2 is specific for ubiquitin Lys-63 linkages ( Figure 2 ; see Supplemental Figure 1 online). The robustness of the in vitro reaction, and yeast two hybrid data ( Figure 1 ), furthermore suggest that Lys-63 chain formation is the major activity of these ubiquitin ligases. Also consistent with this interpretation is the finding that Arabidopsis mutants in both MMZ1 and MMZ2, UBC cofactors essential for Lys-63 chain formation, display the similar phenotype of reduced apical dominance (X.-J. Yin and A. Bachmair, unpublished data), implying ubiquitin Lys-63 chain formation in the regulation of plant shape.

RGLG1 and RGLG2 have a copine domain (also called a von Willebrand factor type A domain) positioned approximately in the middle of the protein ( Figure 1A ). Copine domains can exist in two alternative conformations, and conformational switches have been investigated at the genetic and structural levels ( Mould et al., 2003 ; Barton et al., 2004 ). Therefore, a conformational switch could be part of RGLG activity.

Ubiquitin Lys-63 chain–forming enzymes that operate with or in membrane-associated protein complexes have been identified in several regulatory contexts in animals and in yeast. For instance, some membrane proteins require ubiquitin Lys-63 chain attachment, instead of the usual monoubiquitylation, for endocytosis and/or recycling (reviewed in Dupré et al., 2004 ; d'Azzo et al., 2005; Clague and Urbé, 2006 ). Furthermore, signaling from membrane receptors may require ubiquitin Lys-63 chain formation. The cascade from the interleukin-1 receptor to the transcriptional activator nuclear factor-κB is best characterized and involves autoubiquitylation of a Lys-63 chain–specific ubiquitin ligase, TRAF6 ( Chen, 2005 ; Krappmann and Scheidereit, 2005 ; Lamothe et al., 2007 ). It remains to be determined whether RGLG1/2 have a role in the internalization of membrane proteins, in signal transduction, or in other processes, such as the modification of membrane protein activity.

This work highlights the role of two ubiquitin ligases as membrane-associated elements of hormone adjustment in plants. One of these ligases was shown to form ubiquitin Lys-63 chains. The ligases are not part of the known SCF-TIR1 pathway and influence the levels of the plant hormones auxin and cytokinin. The membrane localization, triple mutant analysis, and additional results suggest that they may influence auxin transport.