Characterization of Arabidopsis mur3 mutations that result in constitutive activation of defence in petioles, but not leaves


  • Jennifer D. Tedman-Jones,

    1. Sainsbury Laboratory, John Innes Centre, Colney Lane, Norwich, Norfolk NR4 7UH, UK
    Search for more papers by this author
    • Present address: Southern Crop Protection and Food Research Centre, Agriculture and Agri-food Canada, 1391 Sandford Street, London, ON N5V 4T3, Canada.

  • Rita Lei,

    1. Sainsbury Laboratory, John Innes Centre, Colney Lane, Norwich, Norfolk NR4 7UH, UK
    Search for more papers by this author
  • Florence Jay,

    1. Sainsbury Laboratory, John Innes Centre, Colney Lane, Norwich, Norfolk NR4 7UH, UK
    Search for more papers by this author
    • Institut de Biologie Molèculaire des Plantes du Centre National de la Recherche Scientifique, 67084 Strasbourg Cedex, France.

  • Georgina Fabro,

    1. Sainsbury Laboratory, John Innes Centre, Colney Lane, Norwich, Norfolk NR4 7UH, UK
    Search for more papers by this author
  • Xuemei Li,

    1. Department of Molecular and Cell Biology, University of Connecticut, Storrs, CT 06269-3125, USA
    Search for more papers by this author
    • §

      Metabolix, Inc., 21 Erie Street, Cambridge, MA 02139, USA.

  • Wolf-Dieter Reiter,

    1. Department of Molecular and Cell Biology, University of Connecticut, Storrs, CT 06269-3125, USA
    Search for more papers by this author
  • Charles Brearley,

    1. School of Biological Sciences, University of East Anglia, Norwich, Norfolk NR4 7TJ, UK
    Search for more papers by this author
  • Jonathan D. G. Jones

    Corresponding author
    1. Sainsbury Laboratory, John Innes Centre, Colney Lane, Norwich, Norfolk NR4 7UH, UK
      (fax +44 1603 450011; e-mail
    Search for more papers by this author

(fax +44 1603 450011; e-mail


A screen was established for mutants in which the plant defence response is de-repressed. The pathogen-inducible isochorismate synthase (ICS1) promoter was fused to firefly luciferase (luc) and a homozygous transgenic line generated in which the ICS1:luc fusion is co-regulated with ICS1. This line was mutagenized and M2 seedlings screened for constitutive ICS1:luc expression (cie). The cie mutants fall into distinct phenotypic classes based on tissue-specific localization of luciferase activity. One mutant, cie1, that shows constitutive luciferase activity specifically in petioles, was chosen for further analysis. In addition to ICS1, PR and other defence-related genes are constitutively expressed in cie1 plants. The cie1 mutant is also characterized by an increased production of conjugated salicylic acid and reactive oxygen intermediates, as well as spontaneous lesion formation, all confined to petiole tissue. Significantly, defences activated in cie1 are sufficient to prevent infection by a virulent isolate of Hyaloperonospora parasitica, and this enhanced resistance response protects petiole tissue alone. Furthermore, cie1-mediated resistance, along with PR gene expression, is abolished in a sid2-1 mutant background, consistent with a requirement for salicylic acid. A positional cloning approach was used to identify cie1, which carries two point mutations in a gene required for cell wall biosynthesis and actin organization, MUR3. A mur3 knockout mutant also resists infection by H. parasitica in its petioles and this phenotype is complemented by transformation with wild-type MUR3. We propose that perturbed cell wall biosynthesis may activate plant defence and provide a rationale for the cie1 and the mur3 knockout phenotypes.


Plants have evolved to recognize invading pathogens and activate defence responses that inhibit pathogen growth and prevent disease. Potential plant pathogens can trigger the host immune system at various stages of ingress: (i) as soon as the pathogen makes physical contact with the host (Hardham et al., 2007; Lipka et al., 2005); (ii) through recognition of general elicitors known as pathogen-associated molecular patterns (PAMPs) (Lotze et al., 2007); and (iii) as late as the point of delivery of pathogen effectors into the infected host cell (Wu et al., 2003). The formation of physical and/or chemical barriers may be sufficient to prevent a particular pathogen species from infecting a particular plant species. Successful pathogens have evolved effector proteins to overcome these defences, infect a host plant and cause disease. In turn, plants have evolved resistance (R) genes whose products trigger a battery of defences upon recognition of an effector encoded by a cognate avirulence (Avr) gene expressed by specific pathogen races. In many cases, R protein-mediated resistance is associated with early production of reactive oxygen intermediates (ROIs), followed by an accumulation of salicylic acid (SA) (reviewed by Jones and Dangl, 2006). Recent studies in Arabidopsis thaliana have also established a role for nitric oxide in mediating plant defence (reviewed by Delledonne, 2005). Reactive oxygen intermediates, NO and SA are all thought to act as signalling molecules that activate additional resistance mechanisms; these include the induction of pathogenesis-related genes and other defence-related genes, as well as the timely activation of a form of programmed cell death known as the hypersensitive response (HR) (reviewed by Nimchuk et al., 2003).

In order to dissect genetically the signalling pathway activated during R protein-mediated defence, various screens have been performed with the model plant Arabidopsis thaliana to identify mutants that are compromised for resistance to avirulent pathogens. Two of these mutants, eds1 (enhanced disease susceptibility) and ndr1 (non-race-specific disease resistance), are fully compromised for resistance mediated by certain R genes (Aarts et al., 1998; Century et al., 1995; Parker et al., 1996). The sgt1b, rar1 (A. thaliana orthologue of barley RAR1: required for Mla-resistance) and pad4 (phytoalexin-deficient) mutants range from full resistance, to partial susceptibility, to complete susceptibility depending on the race of avirulent pathogen tested (Austin et al., 2002; Feys et al., 2001; Glazebrook et al., 1997; Muskett et al., 2002). Two mutants, sid2/eds16 and eds5/sid1 (SA-induction deficient), which are both deficient for pathogen-induced SA production, show partial susceptibility to all avirulent pathogens tested (Nawrath and Metraux, 1999). These results suggest that plant defence is controlled by a branched signalling network rather than a linear pathway.

The SID2 gene encodes an enzyme, isochorismate synthase (ICS1), which is involved in SA biosynthesis (Strawn et al., 2007). ICS1 expression is induced by pathogen stress, and SA levels are directly correlated with ICS1 transcript levels (Wildermuth et al., 2001). Pathogen-inducible SA synthesis is therefore likely to be directly controlled through ICS1 expression. Taken together, these results suggest that ICS1 is tightly controlled by the defence signalling network.

Genetic screens have also been used to identify mutants that show constitutive defence responses in the absence of pathogen stress. Many groups have focused on recessive mutants as these are most likely loss of function mutations in negative regulators that have evolved to suppress the defence response. These mutants are typically characterized by high levels of SA, constitutive expression of PR genes and enhanced resistance to virulent pathogens. Additionally, two E3 ligases have been shown to be required for the defence response; it is hypothesized that upon pathogen recognition, their role is to ubiquitinate negative regulators to promote their destruction in the proteasome and thus de-repress defence mechanisms (González-Lamothe et al., 2006; Yang et al., 2006).

Most forward genetic screens for negative regulators have been focused on identifying mutants that show alterations in the activation and/or control of HR. Mutants have been identified that show spreading lesions after pathogen challenge; these include lsd1 (lesion-simulating disease resistance), acd (accelerated cell death) mutants acd1, acd2 and acd11, and vad1 (vascular-associated cell death); these mutants have been classified as propagative lesion-mimics. The rest of the lsd mutants (lsd2, -3, -4, -5, -6, -7), acd5, acd6, and cpr5 (constitutive expresser of PR genes) all produce visible lesions spontaneously; these mutants are known as initiation lesion-mimics. By contrast, the dnd mutants, dnd1 and dnd2 (defence-no-death), hrl1 (hypersensitive response-like) and hlm1 (HR-like lesion mimic) mutants show only microscopic lesions and actually suppress HR when challenged with an avirulent pathogen (reviewed by Lorrain et al., 2003). Some of the genes identified through these screens encode proteins typically associated with signal transduction: lsd1 encodes a zinc finger protein, and dnd1 and dnd2 both encode cyclic nucleotide-gated ion channels (Clough et al., 2000; Dietrich et al., 1997; Jurkowski et al., 2004). Characterization of lesion-mimics cpr5 and vad1 has revealed links between cell death programs involved in defence and those involved in development and/or senescence (Kirik et al., 2001; Lorrain et al., 2004; Yoshida et al., 2002). Identification of mutants such as cpr5 and vad1 suggests that cell death programs controlling various aspects of plant life might be coordinately regulated.

Other researchers have screened for mutants that constitutively express defence genes but do not show a lesion-mimic phenotype. This strategy may prevent the identification of mutants that are affected in cell death programs unrelated to defence. Four screens have been performed using the following approaches: both cpr1 and snc1 (suppressor of npr1-1 constitutive) were identified based on their constitutive expression of PR2; the cir mutants (constitutively induced resistance) and cim mutants (constitutive immunity) were identified based on their constitutive expression of PR1 (Bowling et al., 1994; Li et al., 2001; Maleck et al., 2002; Murray et al., 2002). None of the cim or cir mutants have yet been cloned. SNC1 was identified as a gain of function mutation in an R gene homologue from a gene cluster that includes the RPP5/RPP4 (resistance to Peronospora parasitica) gene locus. Genetic crosses showed cpr1 to be allelic to the bal locus. The bal mutation involves epigenetic overexpression of an R gene homologue that is also located within the RPP5/RPP4 R gene cluster. Although cpr1 maps to the same R gene cluster, the cpr1-1 mutant was not found to overexpress any members of this R gene family (Stokes and Richards, 2002; Stokes et al., 2002).

Since recessive mutants that show constitutive defence are also typically characterized by high levels of SA, we hypothesized that ICS1 expression itself is under strong negative regulation. ICS1 is therefore a good target for a genetic screen aimed at identifying genes that negatively regulate the defence response. A genetic screen based on ICS1 expression might also yield other interesting insights into the regulation of the SA-controlled branch of the defence signalling network. In the present study we have identified mutants that show constitutive ICS1 expression (cie) using an ICS1:luciferase promoter–reporter gene fusion. This study also describes the cloning and characterization of one of these mutants, cie1, and demonstrates that cie1 is a novel allele of the previously characterized MUR3 gene involved in cell wall synthesis and cytoskeleton organization (Madson et al., 2003; Tamura et al., 2005).


Forward genetic screen for mutants that constitutively express ICS1 using an ICS1:luciferase promoter–reporter gene fusion

The objective of this study was to identify genes that negatively regulate expression of ICS1 in A. thaliana. We carried out a forward genetic screen to identify A. thaliana mutants that constitutively express ICS1. To rapidly screen thousands of mutagenized A. thaliana plants for constitutive ICS1 expression, we generated a construct, ICS1:luc, consisting of the putative ICS1 promoter region fused to the firefly luciferase reporter gene (details in Experimental procedures). The putative ICS1 promoter region used in this study consists of 3.1 kb upstream of the ICS1 translational start site and includes several predicted pathogen-inducible cis-elements (Tedman-Jones, 2004). The ICS1:luc construct was introduced into A. thaliana wild type Columbia (Col-0) by Agrobacterium tumefaciens-mediated transformation. Ten transformed lines were identified and a single line (C5) that showed a strong increase in luciferase activity in response to pathogen challenge was chosen for further analysis and made homozygous.

To determine if the ICS1:luc line (C5) was an accurate reporter of ICS1 expression, a time-course experiment was carried out with Col-0 and Col-0 (ICS1:luc) plants comparing ICS1 and luciferase transcript accumulation after infiltration with an avirulent bacterial pathogen (Figure 1). As expected, ICS1 expression was not affected by insertion of the ICS1:luc transgene. Significantly, ICS1 and luciferase showed a similar pattern of expression in response to Pseudomonas syringae pv. tomato DC3000 carrying AvrRpt2, in Col-0 (ICS1:luc) plants throughout the time-course.

Figure 1.

 Reverse transcriptase-PCR analysis of luciferase transcript accumulation in Arabidopsis thaliana Columbia (Col-0) plants, transformed with ICS1:luc, over the course of an infection with Pseudomonas syringae carrying AvrRpt2.
Abbreviations: hpi, hours post-inoculation; ICS1, isochorismate synthase transcript; LUC, luciferase transcript; ACT, actin transcript; -RT, cDNA synthesis reaction performed without reverse transcriptase on RNA from untreated Col-0 plants; U, cDNA from untreated controls. The experiment was repeated three times with similar results.

Identification of cie mutants

Columbia (ICS1:luc) seed was mutagenized with ethyl methane sulphonate (EMS) and divided into 50 pools consisting of 200 M1 plants per pool. The M2 seeds were allowed to germinate on agar plates. Seedlings aged 10–12 days were sprayed with luciferin and screened for constitutive luciferase activity using a low-light imaging camera. Two hundred and twenty-eight putative cie (constitutive ICS1 expression) mutants were identified from a screen of approximately 45 000 M2 seedlings. The M3 progeny of 20 of these putative cie mutants showed constitutive luciferase activity observed in the original parent and produced viable M4 seed. Interestingly, cie mutants could be grouped into four classes according to the tissue-specificity of their luciferase activity: cie mutants with luciferase activity in their (i) petioles (Figure 2b); (ii) leaves (Figure 2c); (iii) cotyledons (Figure 2d); (iv) meristems (Figure 2e).

Figure 2.

cie mutants show tissue-specific luciferase activity.
(a) Wild-type (ICS1:luc) plants show no luciferase activity under screening conditions.
(b–e) Representative cie mutants showing luciferase activity in different tissues: (b) petioles, (c) leaf tissue, (d) cotyledons, (e) meristem.

Defences activated in the petioles of cie1 plants

cie1, a cie mutant showing luciferase activity in its petioles, was chosen for further characterization (Figure 2b). To exclude the possibility that the constitutive luciferase activity observed in cie1 was due to a mutation in the promoter of the ICS1:luc transgene, cie1 plants were crossed with plants carrying the β-glucuronidase reporter gene fused to the ICS1 promoter (ICS1:gus). cie1 plants carrying the ICS1:gus transgene showed constitutive β-glucuronidase activity in petiole tissue only, thus confirming the constitutive nature and localization of ICS1 expression (Figure 3a). A RT-PCR analysis of cie1 revealed increased levels of ICS1 as well as EDS5, another pathogen-inducible gene involved in SA biosynthesis (Figure 3b). The SA-inducible defence genes PR1 and PR5, also show increased expression in cie1 plants compared with wild-type controls, as well as GST1, a gene normally induced by the oxidative burst preceding the HR. It should be noted that the RT-PCR shown in Figure 3(b) was performed on RNA extracted from whole seedlings. When RT-PCR was performed on RNA extracted from cie1 petiole tissue, a further enhancement of ICS1 and PR1 expression was observed (Figure S1).

Figure 3.

 Defence responses activated in the cie1 mutant.
(a) Wild-type (ICS1:gus) and cie1 plants (ICS1:gus) stained for GUS activity.
(b) Reverse transcriptase-PCR analysis of defence transcripts expressed in cie1 plants: -RT, cDNA synthesis reaction performed on RNA from cpr6 plants without reverse transcriptase; cie1, cie1 cDNA; WT, cDNA from wild-type Columbia (ICS1:luc) plants; cpr6, cDNA from cpr6 mutant plants (positive control for defence transcripts).
(c) Petioles of wild-type and cie1 mutant plants stained for cell death with trypan blue.
(d) sid2 mutation does not rescue the cie1 microlesion phenotype: proportions of wild-type, cie1 and cie1sid2 plants showing microlesions. Graph shows the mean and SEM from three independent experiments where 12–20 sample leaves were stained per genotype per experiment.

Based on the observation that cie1-mediates constitutive expression of both SA- and ROI-inducible genes, wild-type and cie1 plants were assayed for SA and ROIs to determine the extent to which synthesis of these molecules is active in cie1 plants. To test for increased SA production, petioles were removed from leaf blades and the two tissues were assayed for both free and conjugated SA using the protocol developed by Raskin et al. (1989), but with modifications to accommodate small quantities of tissue (see Experimental procedures and Raskin et al., 1989). It was observed that cie1 mutant plants accumulate significantly higher levels of conjugated SA in petiole tissue compared with wild-type plants (Figure S1). Wild-type and cie1 plants were stained for ROIs using 3,3′-diaminobenzidine (DAB). A significantly greater proportion of cie1 plants showed microscopic patches of ROIs compared with wild-type plants (Figure S1).

Small patches of necrotic cells, absent from wild-type controls, were observed in the petioles of most trypan-blue stained cie1 plants (Figure 3c). Thus, cie1 plants appear to activate the defence signalling network to the point of inducing HR. Interestingly, the sid2-1 mutation is not sufficient to rescue this phenotype as cie1sid2 double mutants show the same microlesion phenotype in proportions that are not significantly different from cie1 plants (Figure 3d).

cie1 requires an intact SA signalling pathway for enhanced resistance to a virulent oomycete pathogen

Arabidopsis thaliana mutants that show constitutive activation of a number of defence markers are typically characterized by enhanced resistance to otherwise virulent pathogens (Lorrain et al., 2003). To test if cie1 also confers enhanced resistance, cie1 plants were sprayed with a virulent oomycete pathogen Hyaloperonospora parasitica Noco2. On most wild-type Col-0 (ICS1:luc) plants, H. parasitica is able to spread throughout the entire leaf (Figures 4b and 5). Hyaloperonospora parasitica is able to colonize and spread through the lamina of cie1 leaves; however, it cannot penetrate the petioles of these mutant plants (Figures 4d,e and 5). Thus, defence responses as well as enhanced resistance are confined to the petioles of cie1 plants. The enhanced resistance phenotype observed in cie1 appears to be dependent upon SA signalling, as this phenotype is abolished in cie1sid2 double mutants (Figures 4f and 5). sid2 mutants show enhanced susceptibility to virulent pathogens (Dewdney et al., 2000; Nawrath and Metraux, 1999). A higher proportion of cie1sid2 plants showed infection in leaves and petioles compared with wild-type controls (Figure 5).

Figure 4.

 Phenotypes of wild-type, cie1 and cie1sid2 double homozygotes when challenged with Hyaloperonospora parasitica Noco2. Hyphal growth was visualized by trypan blue staining 6 days after pathogen challenge.
(a–c) Range of phenotypes observed in wild-type Columbia (ICS1:luc) plants (a) plants showing hyphal growth in leaves but not in petioles and (b, c) plants showing infection throughout the leaf including petioles. (d), (e) Phenotype observed in cie1 plants: hyphal growth in leaves but not petioles.
(f) Susceptibility of petioles to infection by H. parasitica isolate Noco2 is restored in cie1sid2 double homozygous mutants.

Figure 5.

cie1 plants require salicylic acid (SA) signalling for enhanced resistance to Hyaloperonospora parasitica infection in petiole tissue.
Two- to 3-week-old plants were sprayed with spores produced by H. parasitica isolate Noco2. Hyphal growth was visualized at 6 days post-inoculation by trypan blue staining. Phenotypes of wild-type (ICS1:luc), cie1 and cie1sid2-1 when challenged with Noco2: grey bars, seedlings showing Noco hyphae spreading in leaves, but not petioles; black bars, seedlings showing Noco hyphae throughout leaves including petioles. Each bar represents the mean and SEM of three experiments. Between 90 and 220 infected leaves were scored for each genotype for each experiment.

cie1 encodes a mutation in MURUS3 (MUR3)

The CIE1 locus was identified by positional cloning. cie1 Col-0 (ICS1:luc) plants were crossed with wild-type Landsberg erecta (Ler) plants to create a mapping population. cie1 was mapped to the bottom arm of chromosome II between markers PHYB and PLS8 (Figure 6a). A population of 1715 F2 progeny was screened and 11 informative recombinants were identified that localized cie1 to a single bacterial artificial chromosome (BAC) clone, F11A3 (Figure 6b). cie1 plants are characterized by short petioles, curled leaves and reduced length of inflorescence stems (Figure 7a and data not shown). SALK knockouts of candidate genes on BAC F11A3 were screened for this morphological phenotype. One SALK line, 141953, was identified that is morphologically similar to cie1. This SALK knockout carries a T-DNA insertion in a gene that was originally named MURUS3 (MUR3) for its involvement in cell wall biosynthesis. Another mutant, cie185, was identified from the cie screen that has the same morphology and the same pattern of luciferase activity as cie1. cie1 was crossed with both cie185 and the SALK mur3 knockout mutant and found to be allelic. DNA sequence analysis indicated that both cie1 and cie185 plants carried point mutations in the MUR3 gene. Furthermore, cie1-mediated luciferase activity and morphological phenotypes revert to wild type in the progeny of crosses between cie1 plants and mur3-knockout lines transformed with the wild-type MUR3 gene (see Experimental procedures).

Figure 6.

 Mapping and identification of cie1.
(a) Genetic map showing the distance between cie1 and markers on the bottom arm of chromosome II.
(b) cie1 was delineated to bacterial artificial chromosome F11A3 using markers C126-XX which were based on known Columbia/Landsberg erecta polymorphisms; numbers between markers represent the number of recombinants identified in the mapping population.
(c) cie1 carries two mutations that result in amino acid substitutions in the MUR3 protein (S470L and T539I). The MUR3 protein is characterized by a transmembrane domain (black box) and a domain with homology to animal exostosins (grey box). Two other mutant alleles of mur3: cie185 (Q457stop) and the SALK knockout (141953) were identified in the course of this study and are shown along with previously identified alleles: mur3-1 and mur3-2 (corresponding locus and alleles identified by Madson et al., 2003); as well as kam1-1, and SALK knockouts kam1-2 (127057) and kam1-3 (141953) (identified by Tamura et al., 2005).
(d) The mur3 SALK knockout mutant (mur3-3) resists infection by H. parasitica Noco2 in petiole tissue and the mur3-3 resistance phenotype is complemented by wild-type MUR3. Two week old plants were infected with Hyaloperonospora parasitica Noco2. Infection structures were visualized with trypan blue staining. Col-0, wild-type Columbia plant; mur3-3, SALK knockout line 141953; comp 13, comp11, comp 3, comp16, mur3-3 complemented with wild-type MUR3 transgenic line 13, line 11, line 3, and line 16, respectively. The bars represent the mean and SEM of two experiments. Between 60 and 100 infected leaves were scored for each genotype for each experiment.

Figure 7.

 Relationship between morphology and defence responses of mur3 mutants.
(a) Morphology of mur3-5 compared with wild-type (ICS1:luc) plants. The other mur3 mutants, as well as the mur3-5sid2 double mutant, are listed in brackets below the image that is similar to their morphology.
(b) Reverse transcriptase-PCR analysis of wild-type plants, mur3 mutants and the mur3-5/sid2-1 double mutant. Lane 1, -RT control (mur3-5 RNA); lane 2, wild-type Columbia (ICS1:luc); lane 3, mur3-5; lane 4, wild-type Columbia; lane 5, mur3-1; lane 6, mur3-2; lane 7, mur3-3 (SALK knockout); lane 8, mur3-5sid2 double homozygote. The RT analysis was repeated three times with similar results.

MUR3 was first identified in a screen for mutants affected in cell wall biosynthesis. The mur3 mutants show structural alterations in the hemicellulose xyloglucan (Reiter et al., 1997). The MUR3 gene was identified by positional cloning and found to encode a protein of 619 amino acids. The predicted protein carries a type II transmembrane domain found in Golgi-localized proteins, as well as a domain that is conserved in a class of animal glycosyltransferases called exostosins (Pfam domain: 03016). In A. thaliana MUR3 functions as a xyloglucan galactosyltransferase (Madson et al., 2003). Animal exostosins function in an analogous fashion, synthesizing heparan sulphate for the extracellular matrix. cie1 carries two mutations in MUR3, an S→L substitution in the putative exostosin catalytic domain, as well as a T→I substitution in the C-terminal region (Figure 6c); cie185 carries a stop codon in the MUR3 exostosin domain. The mur3-1 and mur3-2 mutants carry amino acid substitutions in the exostosin domain of MUR3 (Figure 6c). Another mutant allele of mur3, kam1-1, was identified from a screen for mutants that show disorganized endomembrane systems; the kam1-1 mutant carries a stop codon in the N-terminal region of MUR3. Further analysis showed that kam1-1 mutants were defective in actin organization. Based on the results from these screens, MUR3 is predicted to have at least two cellular functions: cell wall biosynthesis (xyloglucan galactosyltransferase) and organization of the endomembrane system through actin (Madson et al., 2003; Tamura et al., 2005).

Complementation of the mur3-mediated resistance phenotype

The mur3-3/kam1-3 SALK knockout mutant was challenged with H. parasitica Noco2 and showed enhanced resistance in its petioles; an identical phenotype to cie1/mur3-5 (Figure 6d). Four mur3-3 knockout lines carrying transgenic copies of wild-type MUR3 were challenged with Noco2 alongside the mur3-3 mutant and wild-type Col-0 plants. As in the wild-type line, the pathogen successfully colonized petiole tissue of the four MUR3-complemented lines.

mur3 mutants that show altered morphology also show constitutive PR gene expression

While mur3-5 (cie1) and mur3-3 (SALK knockout/kam1-3) show a high level of ICS1 and PR1 gene expression, the mur3-1 and mur3-2 mutants show wild-type levels of ICS1 and PR1 expression (Figure 7b). Furthermore, unlike the mur3 alleles identified in the present study, mur3-1 and mur3-2 are morphologically wild type (Figure 7a). Thus, there appears to be a correlation between the morphology and defence-gene expression among these mur3 mutants.

Defence gene expression and morphology are uncoupled in the cie1sid2 double mutant

The cie1sid2-1 double mutant still shows constitutive luciferase activity (data not shown); however, it does not show constitutive ICS1 expression. In the sid2-1 lines, the ICS1 gene has a stop codon in exon IX (Wildermuth et al., 2001), and we presume that this compromises accumulation of mRNA from this ICS1 allele. PR1 is expressed at wild-type levels in the cie1sid2 double mutant. Interestingly, these double mutants still retain cie1-like morphology (compare Figure 7a). The SA signalling pathway is therefore required for cie1-mediated defence gene expression and enhanced resistance, but not for cie1 morphology.


A novel screen for constitutive activation of plant defence

A forward genetic screen was conducted to identify mutants that constitutively express ICS1 using an ICS1:luc promoter gene fusion. Use of the ICS1:luc line for a genetic screen was validated with a time-course experiment which showed that luciferase expression mirrors ICS1 expression in ICS1:luc plants following infiltration with an avirulent bacterial pathogen (Figure 1). Strikingly, all cie mutants identified from the screen show tissue-specific luciferase activity (Figure 2). cie1 shows luciferase activity only in the petiole tissue of both cotyledons and leaves and this pattern does not change regardless of the developmental stage of the plant. Luciferase activity was monitored at various developmental stages for three other cie mutants. Like cie1, these mutants show constitutive luciferase activity that is confined to a particular part of the plant and does not change as the plant matures (data not shown). Previous screens for mutants that show constitutive expression of defence genes have identified mutants that also have unique patterns of expression. When treated with the SA analogue 2,6-dichloroisonicotinic acid (INA), sni1 (PR2:GUS) mutants show GUS-staining throughout leaves and roots, but reporter gene activity is strongest in their vascular tissues (Li et al., 1999). cpr1 mutants show expression of the PR2:GUS reporter gene throughout adult leaves, but only at the tips of newly emerging leaves and in the petioles of cotyledons (Bowling et al., 1994). cpr6 mutants, which also carry a PR2:GUS transgene, show reporter gene activity throughout cotyledons and older leaves, in the petioles of mature leaves, and not at all in newly emerging leaves (Clarke et al., 1998). Conversely, the tissue-specific defence gene expression of at least four cie mutants examined so far is not developmentally regulated, an observation that has not been described for mutants identified from other screens. The significance of this result is not yet known. We propose three scenarios which might account for the observed results: a different set of genes regulates ICS1 expression for each plant tissue; the same set of genes controls ICS1 throughout the plant, but their activity is modulated by tissue-specific factors; or a combination of these two scenarios. The proposition that defence responses can be controlled by tissue-specific regulators is further substantiated by the identification of the lesion-mimic mutant vad1 which shows propagative HR-like lesions that are confined to the vascular system (Lorrain et al., 2004). The vad1 mutant also shows constitutive expression of defence markers as well as enhanced resistance, and like cie1, these phenotypes are dependent on SA signalling.

cie1 constitutive defence phenotypes are confined to petioles

ICS1, EDS5 and PR1 are all constitutively expressed in cie1 plants. Higher levels of constitutive expression were observed for all defence transcripts in cpr6 compared with cie1 from RNA extracted from whole seedlings (Figure 3b). In cpr6 mutants, PR2, and presumably other defence-related genes, are constitutively expressed not only in petioles but also in stem tissue, cotyledons and older leaves (Clarke et al., 1998). Enhancement of expression of defence-related transcripts in cpr6 as compared with cie1 might therefore be accounted for by the increased proportion of tissue showing this phenotype. Alternatively, the cpr6 mutation may de-repress the defence signalling network to a stronger extent than cie1.

Previous studies have implicated the isochorismate synthase (ICS1/SID2) SA biosynthesis pathway as the major pathway activated during the A. thaliana defence response (Strawn et al., 2007; Wildermuth et al., 2001). The observation that constitutive PR1 expression and resistance to H. parasitica are both abolished in the cie1sid2 double mutant, lacking isochorismate synthase, is consistent with the idea that cie1-mediated defence is also mediated through the ICS1-SA biosynthesis pathway. Constitutive expression of defence genes and enhanced resistance to virulent pathogens are both characteristics of a plant that is activated for SAR (Sticher et al., 1997). Following a localized resistance response SAR is typically activated in all distal tissues providing enhanced resistance to ordinarily virulent pathogens. By contrast, the enhanced resistance phenotype observed in cie1 plants is confined to petiole tissue. Restriction of enhanced resistance to petiole tissue suggests that cie1 plants do not produce the diffusible SAR signal. Although it has been shown that SA is not the SAR signal (Vernooij et al., 1994), it has also been shown that exogenous application of SA to healthy plant tissues is sufficient to induce SAR (White, 1979). The fact that we can detect conjugated SA, but not free SA, suggests that the tissue is producing SA but that it is immediately converted to its conjugated form. In light of the observed accumulation of conjugated SA in cie1 petioles, restriction of SAR-like enhanced resistance to petiole tissue therefore appears contradictory. One possible hypothesis is that insufficient SA is produced in the petioles for SAR to be induced in leaves or cotyledons.

The role of SA signalling in mediating cie1 phenotypes

cie1 is characterized by reduced inflorescence stem length and stunted growth that is most pronounced at the petioles (data not shown). In most cases, constitutive defence mutants show stunted growth, and for many of these mutants wild-type morphology is restored in double mutants that are defective in defence signal transduction (Lorrain et al., 2003 and references therein). cie1 is an exception to this rule in that the cie1sid2 double mutant still shows the same morphology as cie1, despite loss of constitutive PR1 gene expression and enhanced resistance. The observation that cie1sid2-1 plants retain cie1 morphology but allow growth of H. parasitica into petioles is important to note as it implies that lack of H. parasitica infection in cie1 petioles is not simply due to an altered morphology that prevents the pathogen from physically entering this tissue, but is the result of an active process that requires a known defence signalling pathway. Interestingly, cie1sid2 plants still produce spontaneous microlesions, further suggesting that programmed cell death is not sufficient for resistance.

How is defence activated in the cie1 mutant?

Previous forward screens have implicated MUR3 in cell wall biosynthesis and actin organization. cie1 encodes another allele of mur3, designated mur3-5. This observation raises three important questions. Do mur3-5 mutants exhibit any of the cell wall or actin disorganization phenotypes characteristic of other mur3 mutants? If so, do any of these phenotypes contribute to the defence phenotypes observed in mur3-5? Finally, why are defence gene expression and enhanced resistance localized to the petioles of mur3-5 plants? Both mur3-3 (SALK T-DNA allele) and mur3-5 show the same disease resistance phenotypes and morphology. Mur3-5 has a second mutation in the MUR3 gene in addition to the one that it shares with mur3-1; it therefore seems likely that, like mur3-3, it is a complete loss-of-function allele. If this were the case, we would predict that mur3-5 is affected in other mur3-related phenotypes (i.e. actin organization and xyloglucan biosynthesis) to the same extent as mur3-3. Even if mur3-5 is found to show other mur3-related phenotypes it is still unclear how they might activate defence responses. To the best of our knowledge there are no studies that directly link either xyloglucan biosynthesis or actin organization with constitutive defence activation.

In order to access nutrients contained within a host cell, most pathogens must secrete a large variety of hydrolytic enzymes to break down the surrounding cell wall. Plant cell walls are composed of a highly complex mixture of polysaccharides. There is great variation in cell wall composition between species. Cell wall complexity is thought to contribute to defence, because latent signaling molecules, capable of eliciting plant defences, are synthesized and stored within the cell wall only to be released when the wall is being attacked by a pathogen. It is not currently possible to identify individual cell wall polysaccharides that might make one plant species a non-host and one a host to a given pathogen species (reviewed by Vorwerk et al., 2004). The only polysaccharides with a well-characterized role in eliciting plant defences are a group of molecules called oligogalacturonides (OGAs) which are derived from the degradation of pectin, a major constituent of the cell wall. Host-derived molecules such as OGAs have recently been classified as damaged-associated molecular patterns (DAMPs), in order to distinguish them from pathogen-derived PAMPs (Lotze et al., 2007). It is possible that the cie1 mutations disrupt MUR3 function to the extent that a latent signal molecule is released from the cell wall eliciting a defence response. However, MUR3 is expressed ubiquitously (Madson et al., 2003), so why are defence responses only activated in the petioles of cie1 plants? Structural analysis of cell wall material from different A. thaliana tissues revealed that different tissue types also vary greatly in terms of cell wall composition (Richmond and Somerville, 2001). It is tempting to speculate that the unique composition of cell walls in petiole tissue is such that the cie1 mutations lead to the release of a latent signal molecule in petiole tissue alone, either by the production of a petiole-specific signal molecule or differences in the structural integrity that lead to the release of a common signal molecule, or both.

The mechanism by which cie1 activates defence might be revealed by a screen for mutations that suppress cie1-mediated phenotypes. Null mutations in the BON1 gene lead to constitutive activation of defence responses (Yang and Hua, 2004). The bon1-1 phenotype was found to be dependent on an R gene homologue, SNC1, that is present in Col-0 but absent in other A. thaliana accessions. Absence of BON1 protein is therefore thought to activate SNC1-mediated defence responses. BON1 might therefore be interpreted according to the guard hypothesis, whereby R-protein SNC1 monitors BON1 and activates defence when the protein is absent. Conceivably, an R protein might detect perturbations in the cell wall or actin cytoskeleton conditioned by MUR3 loss of function, and then activate defence. By screening for mutations that suppress cie1 we might identify an R gene that has evolved to monitor the MUR3/CIE1 protein or its products. Alternatively, the screen might reveal a gene that has evolved to detect an elicitor, normally produced by pathogen activity, which is also produced as a result of cie1-mediated changes to the cell wall.

In summary, we have successfully identified a unique variety of mutants that show tissue-specific luciferase activity driven by an ICS1:luciferase promoter–reporter gene system. One of the mutants identified from this screen, cie1, shows co-localization of luciferase activity and constitutive defence responses to petiole tissue. cie1 was identified through positional cloning and found to carry two mutations in MUR3, a gene required for cell wall biosynthesis and actin organization. Both cie1 and a mur3 knockout mutant induce the defence system to the point of preventing a virulent oomycete pathogen from entering petioles, despite successful colonization throughout the rest of the leaf. The cie1-mediated enhanced resistance phenotype requires defence responses that are specifically activated through the SA signalling pathway, as mutations in ICS1 abolish this phenotype. These findings have important implications for plant–pathogen interactions occurring at the cell wall interface.

Experimental procedures

Arabidopsis thaliana accessions, mutant lines and growth conditions

Transformations and crosses were performed using either A. thaliana ecotype Col-0 or Ler. cpr6 mutant seeds were obtained from Xinnian Dong (Duke University, Durham, NC, USA). The SALK knockout line 141953 was obtained from the Nottingham Arabidopsis Stock Centre (NASC, Nottingham, UK). Sown seeds were left at 4°C for 2–3 days before being moved to a growth room or glasshouse. Plants grown for the purpose of performing crosses or for stable transformation by A. tumefaciens were kept in a glasshouse. Plants required for pathology tests were grown under a short-day photoperiod (in this case 10 h light/14 h dark) in a standard growth cabinet. The ambient humidity and temperature were maintained at 70% and 22°C, respectively. Seeds that were to be germinated under axenic conditions were first subjected to vapour phase sterilization. Seeds were placed in a vacuum desiccator around a beaker containing 100 ml of a 10% (w/v) solution of sodium hypochlorite. A universal containing 3 ml of a 37% (v/v) solution of hydrochloric acid was placed in the middle of this beaker and a vacuum was applied to the desiccator. The hydrochloric acid was released into the sodium hypochlorite solution by gently rocking the desiccator. Seeds were treated for 4–16 h, removed and left to air for 1 h in a laminar flow hood. Seedlings grown on Germination medium (GM) plates (½ MS salts, 1% agar, no sucrose) were kept in a tissue-culture growth room under a long-day photoperiod (16 h light/8 h dark). The tissue-culture growth room was kept at 22°C.

Hyaloperonospora parasitica infection

Seedlings at 2–3 weeks old were gently sprayed with fresh H. parasitica Noco2 spore suspension (culture was grown for a week) containing 4 × 104 spores ml−1, using a spray gun. Sprayed seedlings were covered with a plastic lid and placed in a reach-in growth cabinet at 16–17°C for 1 week. Trypan blue staining of infected leaves was carried out as described previously (Koch and Slusarenko, 1990). Stained samples were mounted onto microscope slides using 60% (v/v) glycerol and examined using a light microscope (Zeiss, The presence of microlesions in unchallenged plants was also determined by trypan blue staining.

Pseudomonas syringae infection

Liquid P. syringae cultures grown overnight were resuspended in 10 mm MgCl2 to a final concentration of 106 colony-forming units (cfu) ml−1. Seedlings aged 4–5 weeks were infiltrated with P. syringae pv. tomato DC3000 carrying either pVSP61 (empty vector) or pV288 (avrRpt2) (Kunkel et al., 1993 and references therein); using a needle-less syringe. Control plants were infiltrated with 10 mm MgCl2. Treated plants were left in a short-day growth room until it was time to harvest the tissue.

Determination of conjugated and free salicylic acid from small tissue samples

Both conjugated (SAg) and free (SA) salicylic acid were extracted with modifications to the methods of Raskin and Yalpani and their colleagues (Raskin et al., 1989; Yalpani et al., 1991). Leaf blade samples were put into microfuge tubes, weighed, snap frozen in liquid N2 and stored at −80°C until extraction. Salicylic acid and SAg extractions were carried out as follows: 0.1 g of the frozen leaf tissue was ground to a powder with a plastic pestle. Each sample was extracted with 0.5 ml 90% (v/v) methanol with further grinding, sonicated in a water bath for 20 min and centrifuged at 13 000 g for 10 min. For petioles, the tissue was extracted directly into 0.25 ml 90% (v/v) methanol with grinding. Supernatants were transferred to clean microfuge tubes, and the tissue debris were re-extracted with 0.5 ml (leaf tissue) or 0.25 ml (petioles) 90% (v/v) methanol, sonicated for 10 min and centrifuged at 13 000 g for 10 min. The supernatants were pooled, and 15 μl of 150 mm sodium hydroxide was added. The homogenate was dried under N2 (40°C) or with a UniVap (Anachem,, resuspended in 0.5 ml 5% (w/v) trichloroacetic acid, mixed by vortexing, sonicated for 10 min and centrifuged at 12 000 g for 10 min. The supernatant was transferred to a clean tube and partitioned with 1 ml of ethylacetate/cyclopentane/isopropanol (5/4.95/0.05 v/v/v). The sample was briefly centrifuged to separate the phases. The upper organic phase was transferred to a glass vial. The lower aqueous phase was re-extracted with 1 ml of ethylacetate/cyclopentane/isopropanol (5/4.95/0.05 v/v/v) and centrifuged as above. The upper organic phases were pooled, and 60 μl of 0.2 m sodium acetate pH 5 was added. Salicylic acid glucoside was measured after release of SA from the aqueous phase by the addition of 2 μl of concentrated hydrochloric acid and heating at 100°C for 30 min. The sample was cooled to room temperature, and extracted with ethylacetate/cyclopentane/isopropanol as for SA. Samples were dried and resuspended in 1 ml (leaf tissue) or 0.5 ml (petiole tissue) of mobile phase for HPLC analysis (95% solution A, 5% solution B).

Chromatography was performed on a 25 cm × 4.6 mm Shandon Hypersil 5 μm C18-BDS column fitted with a Phenomenex C18 SecurityGuard cartridge, both obtained from Phenomenex ( The column was eluted at 1 ml min−1 with a gradient of decreasing solvent polarity delivered from solvent reservoirs containing: A, 25 mm NaH2PO4, adjusted to pH 2.5 with H3PO4; and B, acetonitrile/methanol/water in a 40/50/10 v/v/v ratio. The gradient was obtained by mixing solutions A and B as follows: time (min), % B; 0,5; 20,80; 21,5; 30,5. Samples of 100 μl were injected for both standards and biological samples.

Both SA and SAg were quantified by reverse phase (RP)-HPLC with tandem UV and fluorescence detection. The UV absorbance of plant extracts was determined at 325 nm on a Merck-Hitachi L-4000 UV detector (, while fluorescence detection was performed on a Jasco FP-920 fluorescence detector ( set at 282 nm excitation, 420 nm emission, with a gain of 100 and a slit width of 18 nm. Detector outputs were processed in the Flo-One for Windows chromatography software of a Canberra-Packard ( A515 series radiochemical flow detector for HPLC. A standard curve for quantification was obtained by injection of 0–50 pmol SA.

Staining for reactive oxygen intermediates

Plant samples were stained in the dark with a 1% solution of diaminobenzidine for 4 h. Plants were destained by boiling in 90% ethanol for 10 min, then transferred to a series of ethanol solutions: 90%; 50%; 25%, then water and finally to a 50% glycerol solution. Whole seedlings or leaves were mounted onto microscope slides and scored using a light microscope (Axioplan, Zeiss).

Generating ICS1:luc and ICS1:gus transgenic lines

The Escherichia coli strain used for cloning experiments and plasmid propagation was DH5α (Hanahan, 1983). A 2.6-kb BamHI/SalI fragment encompassing the ICS1 upstream regulatory region was excised from DNA of BAC F25A4 and cloned into pBluescript (SK−) to yield the plasmid pSLJ13761. Primers were designed that amplified a 1-kb region starting upstream of the SalI site and extending to include the ICS1 translational start site and the ICS1 transit peptide. This fragment was amplified from BAC clone F25A4, digested with SalI/NcoI and cloned into pSLJ13761 to give pSLJ14182. A 1.9-kb NcoI/HindIII fragment from pSLJ3792 (a derivative of pDO432) (Ow et al., 1986) containing the firefly (Photinus pyralis) luciferase cDNA and a nos 3′ terminator was cloned into pSLJ14182 to give pSLJ14391. A 2.5-kb NcoI/HindIII fragment that included the β-glucuronidase (GUS) cDNA and an ocs 3′ terminator, was obtained from digesting pSLJ4B11 (Jones et al., 1992) and cloned into pSLJ14182. This construct was named pSLJ14381. BamHI/HindIII fragments of pSLJ14391 and pSLJ14381 containing the promoter–reporter gene fusions were separately subcloned into a modified version of the binary plant vector pGREEN0229 (Hellens et al., 2000) that contained a pUC-like polylinker to produce clones pSLJ200801 and pSLJ200791, respectively. To create ICS1 promoter–reporter gene constructs without the transit peptide, primers were designed to amplify a fragment that started upstream of the SalI site of the ICS1 promoter (ICS800F: 5′-AACACCCACACGAGAGGAAC-3′) and ended at the ICS1 translational start site (ICS750R: 5′-TGAAGCCATGGCAGAAATTCGTAAAGTG-3′) with an NcoI site designed around the ICS1 translational start site. This PCR product was amplified from BAC clone F25A4, digested with SalI and NcoI and used to replace the 1-kb SalI/NcoI fragment from pSLJ14182 to generate construct p200815. The p200815 clone was sequenced to verify that no mutations had been introduced by Taq DNA polymerase. A 3.1-kb BamHI/NcoI fragment from p200815 was ligated to a 6.3-kb BamHI/NcoI fragment from pSLJ200801 to produce the ICS1:luc construct p200821 (BamHI–ICS1 promoter–NcoI–luciferase–nos 3′terminator–HindIII–pGREEN0229). Similarly, a 3.1-kb BamHI/NcoI fragment from p200815 was ligated to a 6.9-kb BamHI/NcoI fragment from pSLJ200791 (6.9 kb) to yield the ICS1:gus construct p200831 (BamHI–ICS1 promoter–NcoI–β-glucuronidase–ocs 3′ terminator–HindIII–pGREEN0229). The ICS1:luc and ICS1:gus constructs were isolated from E. coli and transformed into the A. tumefaciens strain GV3101 by electroporation (Holsters et al., 1980). Arabidopsis thaliana Col-0 and Ler plants were transformed, by floral dipping (Clough and Bent, 1998), with A. tumefaciens strains carrying ICS1:luc and ICS1:gus, respectively.

RT-PCR analysis

RNA was extracted from 50–200 mg of leaf tissue using Tri-Reagent® (Sigma, as per the manufacturer’s instructions. Two micrograms of total RNA was used in a reverse transcription reaction using Expand reverse transcriptase (Roche Diagnostics, and the reaction was carried out as per the instructions provided by the manufacturer. The following primers were used to analyse expression of the following genes by RT-PCR: Actin: AC1, 5′-ATGGCAGACGGTGAGGATATTCA-3′ and AC2, 5′-GCCTTTGCAATCCACATCTGTTG-3′; ICS1, ICS1F6, 5′-AGTGAATTTGCAGTCGGGAT-3′ and ICS1R6, 5′-AATCGCCTGTAGAGATGTTGT-3′; EDS5, EDS5F2, 5′-TTGGTTGCTCAAAGTGCAAG-3′ and EDS5R2, 5′-AGCCTTGGGAAGATTACGGT-3′; PR1, PR1L, 5′-TCGTCTTTGTAGCTCTTGTAGGTG-3′ and PR1R, 5′-TAGATTCTCGTAATCTCAGCTCT-3′; Luciferase, LUC62F, 5′-GGAGAGCAACTGCATAAGGC-3′ and LUC658R, 5′-AATCTGACGCAGGCAGTTCT-3′; PR5, PR5F, 5′-CGTACAGGCTGCAACTTTGA-3′ and PR5R, 5′-GCGTTGAGGTCAGAGACACA-3′; GST, GSTF, 5′-TGTCGAGCTCAAAGATGGTG-3′ and GSTR, 5′-CCTTGCCAGTTGAGAGAAGG-3′. The following PCR thermocycling program was used for all primer combinations: 94°C for 2 min, X cycles of (94°C for 30 sec, 52°C for 30 sec, 72°C for 1 min), 72°C for 7 min. For each primer combination the number cycles used (X) was optimized to reflect the relative levels of transcript between samples: Actin, = 26 cycles; ICS1, = 24 cycles; EDS5, = 25 cycles; PR1, = 25 cycles; Luciferase, = 27 cycles; PR5, X = 24 cycles; GST, = 23 cycles. The RT-PCR products were visualized by standard gel electrophoresis.

Ethyl methane sulphonate mutagenesis of ICS1:luc seed

Approximately 10 000 ICS1:luc seeds were allowed to soak in a 0.3% (v/v) solution of EMS for 15 h. Seeds were washed three times in sodium thiosulphate (100 mm) for 15 min at a time. The seeds were then washed three times in water for 15 min, and four times for 30 min. Seeds were suspended in a sterile solution of 0.1% (w/v) agar and pipetted onto soil using a sterile Pasteur pipette.

Reporter gene assays

Ten- to 12-day-old M2 (ICS1:luc) seedlings, grown on GM–agar plates, were sprayed with a 1 mm solution of D-luciferin. Seedlings were left for 30 min to take up the luciferin substrate. Seedlings were imaged with a liquid nitrogen-cooled CCD camera (Princeton Instruments, The aperture was set at F-stop 1.2 and each image was taken at a 90-sec exposure. Images were processed using the MetaMorph® software program (Universal Imaging, Two- to 3-week-old seedlings carrying the ICS1:gus construct were stained for β-glucuronidase activity as described previously (Jefferson et al., 1987).

Mapping and identification of cie1

cie1 (ICS1:luc in the Col-0 background) was crossed with A. thaliana ecotype Ler. The F2 progeny of this cross were screened for segregation of constitutive luciferase activity by in vivo imaging. The cie1 mapping population showed a 3:1 segregation −luc:+luc. To establish linkage of the cie1 phenotype with a Col-0 marker, the genomic DNA of +luc plants (15–30 plants) was pooled. A pool consisting of DNA from −luc plants was used for comparison. Primers that had previously been designed to detect simple sequence length polymorphisms (SSLPs) between Ler and Col-0 were used for bulked segregant analysis to get a rough map position for the cie1 locus (Bell and Ecker, 1994; Lukowitz et al., 2000). The PCR amplifications were carried out according to the conditions specified by the authors. Markers used for the genetic map are described in the TAIR website ( Markers used for the physical map of the cie1 locus are based on SSLPs listed in the Cereon Database (now maintained by Monsanto: Sequencing of the MUR3 gene for cie1, cie185 and wild-type Col (ICS1:luc) were carried out using the PerkinElmer Big Dye® kit ( according to the manufacturer’s protocol. Samples were run on an ABI Prism 3700 DNA analyser and evaluated by abi software ( Software in the DNASTAR® package ( was employed to edit and align sequences.

Identification and genetic complementation of the mur3-3 mutant

mur3-3 (SALK_141953) plants were grown in ProMix BX potting mixture (Milikowski, under a 16-h light/8-h dark cycle at 25°C during the light period and 18°C during the dark period. The light intensity was 125 μmol m−2 sec−1, and the relative humidity was 70%. Homozygous mutant lines were identified by PCR, and the mur3 phenotype was confirmed by gas-liquid chromatography of alditol acetates (Reiter et al., 1993). Transformation of mur3-3 plants with the wild-type gene was carried out as described (Madson et al., 2003), and complementation of the mur3 phenotype was confirmed by monosaccharide composition analysis as described above.


The authors gratefully acknowledge the help of F. Robson with imaging, P. Green with SA analysis, as well as M. Smoker, J. Pike and S. Perkins for assistance with plant materials. All sequencing runs were performed by Nigel Hartley and David Baker at the John Innes Centre, UK. Thanks to X. Dong for kindly providing cpr6 seeds. The SALK knockout line 141953 was obtained from the Arabidopsis Stock Centre, Nottingham, UK. We also thank the Monsanto Corp. for access to the collection of Arabidopsis polymorphisms originally produced by Cereon Genomics. This work was funded by the Gatsby Charitable Foundation (JDGJ). as well as National Science Foundation (grant no. IBN–0215535 to W-DR). The authors thank M. Gijzen and D. Qutob for critical reading of the manuscript and helpful suggestions.