An Arabidopsis homolog of the bacterial peptidoglycan synthesis enzyme MurE has an essential role in chloroplast development

Authors


(fax +81 96 342 3432; e-mail takano@kumamoto-u.ac.jp).

Summary

Enzymes encoded by bacterial MurE genes catalyze the ATP-dependent formation of uridine diphosphate-N-acetylmuramic acid-tripeptide in bacterial peptidoglycan biosynthesis. The Arabidopsis thaliana genome contains one gene with homology to the bacterial MurE:AtMurE. Under normal conditions AtMurE is expressed in leaves and flowers, but not in roots or stems. Sequence-based predictions and analyses of GFP fusions of the N terminus of AtMurE, as well as the full-length protein, suggest that AtMurE localizes to plastids. We identified three T-DNA-tagged and one Ds-tagged mutant alleles of AtMurE in A. thaliana. All four alleles show a white phenotype, and A. thaliana antisense AtMurE lines showed a pale-green phenotype. These results suggest that AtMurE is involved in chloroplast biogenesis. Cells of the mutants were inhibited in thylakoid membrane development. RT-PCR analysis of the mutant lines suggested that the expression of genes that depend on a multisubunit plastid-encoded RNA polymerase was decreased. To analyze the functional relationships between the MurE genes of cyanobacteria, the moss Physcomitrella patens and higher plants, a complementation assay was carried out with a P. patens (Pp) MurE knock-out line, which exhibits a small number of macrochloroplasts per cell. Although the Anabaena MurE, fused with the N-terminal region of PpMurE, complemented the macrochloroplast phenotype in P. patens, transformation with AtMurE did not complement this phenotype. These results suggest that AtMurE is functionally divergent from the bacterial and moss MurE proteins.

Introduction

Peptidoglycans are continuous covalent macromolecular structures that are present on the outside of the cytoplasmic membrane in virtually all eubacteria (van Heijenoort, 2001). These molecules protect bacterial cells from osmotic pressure, maintain the unique shapes of bacterial cells and are involved in cell division. In the bacterial peptidoglycan biosynthesis pathway, the peptide chain of the peptidoglycan is synthesized from the precursor uridine diphosphate-N-acetylmuramic acid (UDP-MurNAc) by the four ATP-dependent ligases MurC, MurD, MurE and MurF, which are in the same Enzyme Commission (EC) class (6.3.2.*) (Figure 1a). The enzymes encoded by the bacterial MurE genes catalyze the formation of UDP-MurNAc-tripeptide. MurE proteins in different organisms are classified as diaminopimelic acid-adding enzymes (6.3.2.13) or l-lysine-adding enzymes (6.3.2.7). The pentapeptides involved in cross-linkage in most cyanobacteria contain the typical Gram-negative bacterial type of diaminopimelic acid, although l-lysine is present in the peptidoglycan of Anabaena cylindrica (Hoiczyk and Hansel, 2000).

Figure 1.

 Characterization of AtMurE.
(a) Bacterial peptidoglycan synthesis pathway and the genes involved (MurA–MurG, MraY and PBP). MurE is a uridine diphosphate-N-acetylmuramic acid (UDP-MurNAc)-tripeptide synthase. GlcNAc represents N-acetylglucosamine.
(b) Comparison of the amino-acid sequence encoded by the Arabidopsis thaliana MurE gene (At) with those of the moss Physcomitrella patens (Pp), the cyanobacterium Anabaena sp 7120 (An) and Escherichia coli (Ec). The full region of MurE except an amino terminal domain is shown. The numbers of the first amino acids in each protein are indicated. The 24 amino acid residues conserved are shown in the bottom line. Amino acids for ATP binding are indicated by underlining. Glycine and proline residues, which play a structural role, are in bold. Three amino acids that interact with the product UDP-MurNAc-tripeptide are boxed. Glutamic acid and lysine for interaction of UDP-MurNAc-dipeptide and ATP are indicated by asterisks. The other five amino acids are for substrate binding. Circles above the sequences represent amino acid substitutions in the AtMurE protein.
(c) Northern-blot analysis with the AtMurE gene as a probe. The methylene blue staining of rRNA bands is shown as a control.

The endosymbiotic theory states that all plastids are derived from a single cyanobacterial ancestor that possessed a cell wall (Rodriguez-Ezpeleta et al., 2005). Therefore, plastids of the glaucocystophyta, red algae, and green plants, including green algae, have evolved as siblings. Although a small group of algae, the glaucocystophytes, has peptidoglycan-armored plastids, it is thought that the peptidoglycan biosynthetic pathway has been lost from the cells of the green-plant lineage. However, it has been reported that independent antibiotics that prevent bacterial peptidoglycan biosynthesis cause the appearance of macrochloroplasts in the moss Physcomitrella patens by inhibiting chloroplast division (Kasten and Reski, 1997; Katayama et al., 2003). We isolated nine homologous P. patens Mur genes that are related to bacterial peptidoglycan biosynthesis, including the MurE gene (Machida et al., 2006). The P. patens MurE protein PpMurE contains a transit peptide and localizes to chloroplasts. Disruption of the PpMurE gene in P. patens results in the appearance of macrochloroplasts, and transformation with the normal gene restores the normal phenotype. These results suggest that the plastidic peptidoglycan synthetic pathway is closely related to plastid division in moss.

The Arabidopsis thaliana genome (Arabidopsis Genome Initiative, 2000) contains five homologous Mur genes: MurE (At1g63680), MraY (At4g18270), MurG (At1g73740) and two Ddl (At3g08830 and At3g08840) genes (Machida et al., 2006). The TargetP program (Emanuelsson et al., 2000) predicts that the A. thaliana MurE (AtMurE), AtMraY and AtMurG proteins have plastid-targeting signals. In A. thaliana, the lack of enzymes upstream of MurE in the peptidoglycan biosynthetic pathway strongly suggests that there is no substrate for MurE in the cell. Moreover, to the best of our knowledge, there is no evidence that the use of β-lactams, such as the carbenicillin used in Agrobacterium-mediated transformation methods, has an effect on plastid division. This information suggests that Mur-like genes are unrelated to peptidoglycan biosynthesis in A. thaliana.

Is the AtMurE gene a relic in the evolution from mosses to higher plants? A genetic approach using various tagged mutant lines offers a useful method for analyzing gene function. In a project to identify genes that are involved in plastid development and biogenesis, the RIKEN Yokohama Institute has identified 38 albino or pale-green (apg) A. thaliana mutants in a collection of 11 000 Ac/Ds transposon-insertion mutant lines (Ito et al., 2002, 2005; Kuromori et al., 2004). We found that one of these apg mutants, named apg13, shows an albino phenotype and has a Ds insertion in the MurE gene. Moreover, in the SeedGenes Project (Tzafrir et al., 2003), one T-DNA insertional line was defined as PDE316 (Pigment Defective Embryo 316), showing white seeds, white embryos and white seedlings. The ratio of mutant seeds in siliques of the PDE316 mutant was estimated at 25.7%, implying that the mutation is single and recessive. These results suggest that the AtMurE gene is related to chloroplast development. The expression of plastid genes encoded by the plastid genome is closely related to the development of proplastids into chloroplasts. In the present study, we analyzed the transcription of plastid genes in the mutant A. thaliana lines. The results show that the plastid-encoded plastid RNA polymerase (PEP)-dependent expression system is deficient in the mutant lines.

Results

Characterization of the AtMurE gene

A gene with similarity to the bacterial MurE gene was identified in the A. thaliana genome (Arabidopsis Genome Initiative, 2000; Machida et al., 2006). RT-PCR analysis showed that the predicted AtMurE protein is composed of 767 amino acids with a molecular mass of 85 kDa. The coding sequence of the cDNA matches perfectly that of the PDE316 cDNA (accession number NM_105045). The protein is predicted to contain the catalytic (pfam01225) and middle (pfam08245) domains, and shows high similarity to the bacterial MurE proteins (Figure 1b). It is reported that 24 amino acid residues are strictly conserved in the bacterial MurE sequences (Gordon et al., 2001). An alignment analysis with AtMurE and PpMurE shows that five amino acid substitutions in the conserved amino acids were found only in AtMurE. Although there is no substitution at the ATP binding site and at the structural amino acids, mutations occur at substrate and product (UDP-MurNAc-tripeptide) binding sites, and at interaction sites between UDP-MurNAc-dipeptide and ATP. This result suggested a change of the MurE function for the AtMurE protein, as discussed below. The phylogenetic tree with the amino acid sequences of MurE from bacteria and plants revealed that plant MurE had a separate phylogenetic position from bacterial ones (Figure S1). RNA gel-blot analysis revealed the accumulation of the AtMurE mRNA in leaves and flowers, but not in roots or stems (Figure 1c). AtMurE gene expression data deposited in public microarray repositories (Zimmermann et al., 2004) indicate a similar expression pattern (Table S1). The highest expression of the AtMurE mRNA was detected in juvenile leaves.

The strong probability of the plastid localization of the AtMurE protein was predicted by TargetP (Emanuelsson et al., 2000), Predotar (Small et al., 2004) and pclr (Schein et al., 2001). To analyze the localization of AtMurE in cells, we constructed plasmids directing the expression of either the full-length AtMurE or only its putative transit peptide fused to GFP. Particle bombardment with either of these plasmids resulted in GFP fluorescence in chloroplasts, corroborating the prediction (Figure 2). These results suggest a stromal localization for the AtMurE protein in chloroplasts, because the AtMurE protein is not predicted to have transmembrane domains.

Figure 2.

 Subcellular localization of the AtMurE–sGFP fusion protein.
Fluorescent microscopic images are shown of cells in Nicotiana tabacum leaves transformed with the sGFP control plasmid, the putative AtMurE transit peptide-sGFP fusion plasmid (AtMurEtp-sGFP) or the AtMurE cDNA-sGFP fusion plasmid (AtMurE-sGFP). GFP fluorescence, chlorophyll autofluorescence and merged images are shown.

Isolation of A. thaliana mutants tagged in the AtMurE gene

To analyze the function of the AtMurE gene in A. thaliana, four lines tagged in the AtMurE gene were used. One of the tagged lines (13-6090-1; atmurE-2) was identified as an apg mutant in a RIKEN Ds-tagged library (Ito et al., 2002, 2005; Kuromori et al., 2004) that was generated using the Activator (Ac)/Dissociation (Ds) system. Also used were one T-DNA-tagged (SALK_126518, designated atmurE-1) and two Ds/Lox T-DNA-tagged (WiscDsLox477-480E17, atmurE-3; WiscDsLox311G11, atmurE-4) mutant alleles of atmurE, all of which contain disruption elements inserted in different positions in the first exon of the AtMurE gene (Figure 3a). All four alleles showed a white-seedling phenotype (Figure 3b), and no differences in the phenotypes in different genetic backgrounds were observed in either seedlings or plants of the atmurE-1, atmurE-3 and atmurE-4 mutants (Columbia strain) and the atmurE-2 mutant (Nossen). These mutant lines also produced seeds that were white at stages when they would normally be green (data not shown). The white seeds were harvested, plated on plates containing 1% sucrose and placed under conditions of normal light, and all produced white seedlings (data not shown). RT-PCR performed using AtMurE-specific primers detected no AtMurE transcripts in the leaves of any of the atmurE mutants (Figure 3c).

Figure 3.

 Characterization of atmurE mutants.
(a) Schematic representation of the genomic region flanking the Ds and T-DNA insertions. Insertion sites for the left border (LB) of atmurE-1, atmurE-3 and atmurE-4, and the H-edge (H) of atmurE-2 are shown. Boxes represent exons of the AtMurE gene. The shaded box represents a putative transit peptide-encoding region. Arrows indicate the position and orientation of the primers used in (c).
(b) Four atmurE mutants showing the white phenotype on medium supplemented with 1% sucrose (right half of each panel). Heterozygous siblings (left) are green.
(c) RT-PCR analysis of AtMurE gene expression in the Columbia (C) and Nossen (N) strains, and in the atmurE mutants (atmurE-1atmurE-4) of Arabidopsis thaliana.
(d) Pale-green phenotype of antisense AtMurE transformants. RT-PCR analysis of the AtMurE gene in the Columbia strain, four wild-type-like and three pale antisense lines is shown.
(e) Chlorophyll contents of the wild-type and atmurE-1 plants.

To confirm a relationship between AtMurE and chloroplast development, we generated transgenic Arabidopsis plants (Columbia) in which an antisense AtMurE gene was overexpressed under the control of a constitutive CaMV 35S promoter (35S:antisense AtMurE). Seeds of the transgenic plants expressing the antisense AtMurE gene were plated on germination medium (GM) plates containing 1% sucrose, and their phenotypes were observed. Three of the seven transformants showed a pale-green phenotype. The levels of AtMurE transcripts in these lines were determined by RT-PCR amplification of the transgene. Transgenic and wild-type Arabidopsis plants (Col) were grown for 3 weeks in lighted growth chambers before being used in the experiments. The RT-PCR results show that a decrease in the AtMurE mRNA level correlates with the pale-green phenotype (Figure 3d). These results strongly suggest that AtMurE acts in chloroplast development.

Characterization of the atmurE mutants

The chlorophyll contents in the atmurE-1 mutant were determined. The quantities of chlorophyll a and b in this line (Figure 3e) were 0.05 ± 0.01 and 0.03 ± 0.01 mg g−1 fresh weight, respectively, much lower (3.1% and 3.8%) than those in wild-type plants (a, 1.61 ± 0.32 mg g−1 fresh weight; b, 0.80 ± 0.15 mg g−1 fresh weight). Next, to investigate the photosystem-II (PSII) activity, we analyzed the Fv/Fm = (FmF0)/Fm value (F0 and Fm are mentioned and maximum chlorophyll a fluorescence of dark-adapted leaves, respectively), which is usually used as an indicator of photoinhibition (Schreiber et al., 1994), by measuring the chlorophyll fluorescence by pulse amplitude-modulated fluorometry. The Fv/Fm of the wild-type plants was determined to be 0.81 ± 0.00 (n = 3), but the Fv/Fm value could not be calculated in homozygous atmurE-1 plants because they produced no measurable chlorophyll fluorescence. These results suggest that the photosynthetic activities of the mutant are severely damaged. Therefore, when atmurE seeds were sown in plates lacking sucrose, they germinated but did not survive (data not shown). Even when atmurE-1 seedlings were grown on medium containing 2% sucrose, no seeds could be harvested, even though the plants produced flowers.

We investigated the plastid morphology in protoplasts of the white leaves of atmurE-1. Staining with the fluorescent DNA-binding dye SYBR Green I showed that the plastid DNA was packed into large dots in the atmurE-1 mutant (Figure 4). In A. thaliana, large plastid nucleoids with large quantities of plastid DNA occupy the stroma in small plastids of the leaf primordia before the development of thylakoids, and the appearance of small nucleoids with small quantities of DNA occurs during the development of the thylakoid system and plastid enlargement (Fujie et al., 1994). Large plastid nucleoids in cells were also observed in white sectors of the leaf variegation mutant yellow variegated2 (var2), suggesting undifferentiation of plastids in the white sectors (Kato et al., 2007). The nucleoids in the atmurE-1 mutant were large bodies that resembled those typically observed in undifferentiated plastids, and in plastids of the white sectors of the var2 mutant. As the leaves contained little chlorophyll, no chlorophyll autofluorescence was detected at the same exposure time used for the photos of wild-type protoplasts. Therefore, an increased exposure time for red fluorescence was used for the mutant.

Figure 4.

 SYBR green staining of protoplast cells from the wild-type (WT) and the atmurE-1 mutant.
Images are merged with those showing chlorophyll autofluorescence. A longer exposure time was used for the chlorophyll autofluorescence of the atmurE mutant in order to detect the low levels of chlorophyll in the mutant. Inserts are photos of higher magnification.

We also investigated plastid morphology using transmission electron microscopy (Figure 5). In leaf primordia of the mutant, there are few thylakoid membranes. While mature chloroplasts of the wild-type plants featured thylakoid membranes, plastids in the mutant leaves had irregular shapes and no mature thylakoid membranes. The mutant cells did not contain giant plastids, like those that are detected in the A. thaliana ftsZ mutants. Along with the abnormalities observed in the atmurE mutant plants, these results indicate that MurE influences chloroplast development in A. thaliana.

Figure 5.

 Electron micrographs of plastids of wild-type and atmurE-1 mutant plants.
Plastids from immature leaves are shown in the upper panels. A mature chloroplast from a wild-type plant and a plastid from an expanded leaf of the mutant are shown in the lower panels.

RT-PCR analysis of chloroplast-encoded genes

The development of chloroplasts from proplastids is closely related to the expression of chloroplast-encoded genes. The plastid transcription-regulation system of A. thaliana depends on the plastid-encoded plastid RNA polymerase PEP and the nuclear-encoded plastid RNA polymerase (NEP) (Shiina et al., 2005). PEP preferentially transcribes photosynthesis-related genes in mature chloroplasts, whereas NEP preferentially transcribes housekeeping genes during early phases of plant development. Plastid genes can be grouped into three classes based on promoter structures (Hajdukiewicz et al., 1997; Ishizaki et al., 2005). Transcription activities of many photosynthesis-related genes depend largely on PEP (class I), whereas genes from the rpoB operon and accD are transcribed exclusively by NEP (class III). Non-photosynthetic housekeeping genes are generally transcribed by both PEP and NEP (class II).

It was recently reported that AtMurE was present in an affinity-purified PEP complex from tobacco chloroplasts (Suzuki et al., 2004), and AtMurE was also identified by TOF-MS analysis as a component of plastid transcriptionally active chromosomes (TACs) isolated from mustard (Sinapis alba) and A. thaliana (Pfalz et al., 2006; see also Discussion). If the AtMurE was related to the PEP transcription system, the number of transcripts transcribed by the PEP complex may decrease in the atmurE mutants. Therefore, we analyzed the quantities of transcripts that encode several chloroplast proteins transcribed by NEP and/or PEP in both wild-type and atmurE mutant plants (Figure 6). The psaA and psbA genes were selected as PEP-dependent genes (class I), accD was chosen as a NEP-dependent gene (class III), and the genes atpB, clpP and ndhB were used as class-II genes. RT-PCR experiments using primer sets specific for each gene showed decreased mRNA accumulation for the class-I gene, and increased accumulation of the class-III gene in the total mRNA pool. For the class-II genes, the accumulation of the ndhB mRNA in the tagged lines was higher than in the wild-type plants, whereas no difference was detected for the atpB or clpP genes. It is known that the ndhB gene is transcribed at a higher rate in the Δrpo mutant than in wild-type plastids (Legan et al., 2002). These results suggest that the chloroplast expression system that utilizes PEP is deficient in the atmurE mutant.

Figure 6.

 RT-PCR analysis of plastid-encoded genes. RNAs isolated from the Columbia (C) and Nossen (N) strains and the atmurE mutants were used.
The psaA and psbA genes were selected as plastid-encoded plastid RNA polymerase (PEP)-dependent genes (class I). For class-II genes, atpB, clpP and ndhB were used. For a nuclear-encoded plastid RNA polymerase (NEP)-dependent class-III gene, the accD gene was chosen. The Actin gene was used as a control.

Complementation assays in the P. patens MurE knock-out line

To compare the functions of the MurE genes of cyanobacteria, the moss P. patens and A. thaliana, a complementation assay was carried out using the P. patens PpMurE knock-out line. We previously showed that this line produces huge chloroplasts, and that transient expression of the functional PpMurE gene in this line can restore the phenotype to that of the wild type (Machida et al., 2006). We constructed plasmids containing the AtMurE or Anabaena MurE (AnaMurE) genes driven by the CaMV 35S promoter. As the AnaMurE amino-acid sequence is closely related to that of PpMurE among cyanobacterial MurE, we used AnaMurE for complementation tests. To construct a plasmid containing the AnaMurE genes, the coding region corresponding to the plastid-targeting sequence of PpMurE was fused to the AnaMurE gene to direct the plastid localization of AnaMurE. Polyethylene glycol (PEG)-mediated DNA transfer into protoplasts of the PpMurE knock-out transformant was carried out, and stable transformants were observed by microscopy (Figure 7a). When transformation was performed with the plasmid containing the AnaMurE gene with the PpMurE plastid-targeting sequence, P. patens cells showed normal chloroplast phenotypes. In contrast, no restoration of the wild-type phenotype occurred when transformation was performed with the plasmid containing the AtMurE gene.

Figure 7.

 Complementation of the Physcomitrella patens PpMurE knock-out line.
(a) Protonema cells of the wild-type plants (WT) contained approximately 50 chloroplasts, whereas the PpMurE knock-out lines (ΔPpMurE) contained macrochloroplasts. Plasmids directing the expression of the Anabaena MurE fused to the PpMurE transit peptide (Pptp-AnaMurE) or AtMurE were constructed, and stable transformants were generated. When transformation was performed with the Pptp-AnaMurE plasmid, P. patens cells showed normal chloroplast phenotypes (+AnaMurE). In contrast, the WT phenotype was not restored when transformation was performed with the AtMurE plasmid (+AtMurE).
(b) Transient expression assay with AtMurE fused to GFP in P. patens cells. Fluorescent images of GFP (left) and chlorophyll autofluorescence (middle), and a merged image (right), are shown.

Although A. thaliana and P. patens are both land plants, it was possible that mislocalization of the AtMurE protein in P. patens cells caused the lack of complementation of the PpMurE knock-out line. To assess the localization of the AtMurE protein in P. patens, we constructed a plasmid directing the expression of the AtMurE transit peptide fused to GFP. PEG-mediated transformation of P. patens with this plasmid resulted in GFP fluorescence in chloroplasts (Figure 7b). This result suggests that in P. patens cells, the AtMurE protein exhibits the same localization as the PpMurE protein (Machida et al., 2006) in the chloroplast stroma. These results suggest that the function of the AtMurE protein changed during the evolution from mosses to higher plants.

Discussion

The A. thaliana MurE gene was analyzed. We obtained four atmurE mutant alleles tagged with a T-DNA or a Ds transposon. These four alleles exhibited the same white phenotype, and MurE antisense plants had pale-green leaves (Figure 3d), indicating that disruption of the AtMurE gene causes these phenotypes. These results suggest that AtMurE is essential for normal chloroplast development and has an important role in normal plant growth. In the moss P. patens, disruption of the PpMurE gene results in the appearance of giant chloroplasts resulting from the inhibition of chloroplast division (Machida et al., 2006), but no other deficits were observed, including in thylakoid membrane formation. Several plastid division-deficient mutants of A. thaliana have been reported, including plastid ftsZ mutants (Miyagishima, 2005; Osteryoung et al., 1998). Although the cells of these mutants contain giant chloroplasts, the growth of the plants is similar to that of wild-type plants under laboratory conditions, and the plants produce green leaves. In contrast, atmurE mutants show a severe white-leaf phenotype that differs from the phenotype of the plastid division-deficient mutants. Electron microscopy revealed many irregularly shaped plastids in leaf cells of these mutants (Figure 5). However, the lack of giant plastids in the atmurE mutants suggests that the AtMurE gene is not related to plastid division in A. thaliana, in contrast to the case of the related gene in P. patens. Therefore, although the MurE genes are functional in both P. patens and A. thaliana, their functions may differ in the two plant species. The alignment analysis with MurE protein also suggests functional divergence.

Different functions for the P. patens and A. thaliana MurE genes were also suggested by the complementation assay. The cyanobacterial MurE with the chloroplast targeting signal was able to complement the knock-out phenotype in the moss, but AtMurE was not (Figure 7). The P. patens genome contains all of the genes that are needed for the creation of peptidoglycan from its precursor UDP-N-acetylglucosamine (UDP-GlcNAc; Machida et al., 2006; K. T. and H. T. unpublished data). This result suggests that a plastidic peptidoglycan exists in P. patens. Because no peptidoglycan-like layer was observed between the two chloroplast envelopes in P. patens, using standard electron-microscopic techniques, the plastidic peptidoglycan may be thin or this synthetic pathway may be specific to chloroplast division. In contrast to moss, no MurA, MurB, MurC or MurD genes are present in the A. thaliana genome. Therefore, it is likely that no substrate for MurE exists in A. thaliana, unlike the situation in bacterial cells and moss chloroplasts. So, even though plastid peptidoglycans may not be present in A. thaliana, AtMurE functions in chloroplast development. These results suggest that a change in the function of the protein occurred during the evolution of seed plants.

It was reported that AtMurE was present in an affinity-purified PEP complex from tobacco chloroplasts (Suzuki et al., 2004), and in the pTACs isolated from mustard and A. thaliana (Pfalz et al., 2006). In the paper by Pfalz et al. (2006), this group isolated 18 new pTACs and analyzed three T-DNA insertion mutants in the three corresponding genes: pTAC-2, pTAC-6 and pTAC-12. Interestingly, the phenotypes of these tagged mutants (ptac2, ptac6 and ptac12) are similar to those of the atmurE mutants. Homozygotic seedlings of each mutant develop white cotyledons. On sucrose medium, ptac2 develops yellow cotyledons and greenish primary leaves, whereas leaves of the other two mutants are yellowish. The expression patterns of plastid-encoded genes in all of the ptac mutants resemble those of the Δrpo mutants, which show defects in PEP-dependent transcription: the expression of class-I plastid genes with PEP promoters is downregulated, and the expression of class-III genes with NEP promoters is upregulated. These similar phenotypes of the atmurE and ptac mutant plants suggest that AtMurE is a functional component of the PEP complex and functions in plastid gene expression in plastid nucleoids.

What is the function of AtMurE in seed plants? MurE proteins are enzymes that add diaminopimelic acid or l-Lys to the UDP-MurNAc-dipeptide in bacteria (van Heijenoort, 2001). Studies have shown that MurE is an essential gene in Escherichia coli, because bacteria do not make cell walls without a functional MurE. In the peptidoglycan synthetic pathway, MurC, MurD and MurE are amino-acid transfer enzymes. The amino acid sequences of these enzymes are highly similar to those of folyl-γ-polyglutamate ligases, which function in the synthesis of folic acids (Eveland et al., 1997). Folyl-γ-polyglutamate ligases catalyze the ATP-dependent addition of successive glutamyl residues to either a folate monoglutamate in eukaryotes or to a dihydropteroate in prokaryotes. A common feature of these enzymes is amino-acid ligase activity. Therefore, a change in the substrate may have occurred during the evolution from bryophytes to seed plants, although it is unknown which amino-acid transfer function is related to plastid gene expression. The alignment analysis with AtMurE also suggested the change in the substrate (Figure 1b). It may also be possible that AtMurE has acquired new functions in plastid nucleoids. For instance, plastidic sulfite reductases (SiRs) have DNA-binding activity and DNA-compacting ability that regulate plastid gene expression, in addition to exhibiting SiR activity (Sekine et al., 2002, 2007). In addition, the possibility of functional change from bacterial MurG to plastidic monogalactosyldiacylglycerol synthase was pointed out in cucumber (Price et al., 1997; Shimojima et al., 1997). AtMurE may have an unknown function in plastid nucleoids.

Experimental procedures

Plant materials and growth conditions

The mutant line atmurE-2 (13-6090-1) was isolated by screening approximately 11 000 Ds-tagged Arabidopsis (Nossen ecotype) F3 lines that were generated at the RIKEN Institute (Ito et al., 2002, 2005; Kuromori et al., 2004, 2006). We identified atmurE-1 (SALK_126518) from the SALK T-DNA collection (Alonso et al., 2003), and atmurE-3 (WiscDsLox477-480E17) and atmurE-4 (WiscDsLox311G11) from the Wisconsin T-DNA collection (Sussman et al., 2000). We confirmed the insertions of atmurE-1, atmurE-3 and atmurE-4 with the left border primers of the tags. For atmurE-2, we sequenced both insertion sites. For normal conditions, the transgenic and mutant plants were grown for 3 weeks in growth chambers (CF-405; TOMY-Seiko, http://bio.tomys.co.jp) provided with approximately 75 μmol photon m−2 sec−1 at 22°C under a 16-h light/8-h dark cycle (long days), and were then transferred to soil and grown under the same conditions. For northern analysis, morphological observations, and measurements of chlorophyll content and fluorescence, plants were grown for 3 weeks on plates under approximately 75 μmol photon m−2 sec−1 at 24°C and in a 9-h light/15-h dark cycle (short days), and were then transferred to soil and grown under the same conditions.

Protonemata of the moss P. patens ssp. patens were grown on BCDAT medium solidified with 0.8% agar, in a regulated chamber at 25°C under continuous light (35 μmol photon m−2 sec−1; Nishiyama et al., 2000).

RNA gel-blot analysis

For northern hybridization, 10 μg of Dnase I-treated total RNA isolated from A. thaliana plants using an RNeasy Plant Mini kit (Qiagen GmbH, http://www.qiagen.com) were used. Northern blots were performed with Church’s phosphate buffer (Sambrook and Russell, 2001), and the AtMurE probe was synthesized using a PCR DIG Probe Synthesis kit (Roche Ltd, http://www.roche.com) with the primers forward, 5′-CTGATGACGATCCACCTGAAG-3′, and reverse, 5′-CTTAGCCATCAAACTCTGAACC-3′. After hybridization, the filters were washed four times at 68°C for 15 min, and signals were detected using a DIG DNA Labelling and Detection kit.

Subcellular localization of GFP fusion proteins

A DNA sequence encoding the N-terminal region of AtMurE (amino acids 1–50), which includes a putative chloroplast transit peptide region of 40 amino acids, was amplified by RT-PCR using the primers 5′-GGATCCTCTCTGTCTCCAATGGCGTTCACC-3′ and 5′-GGATCCGTACGAATTCCGGCGAGCAGGTC-3′ (the BamHI site is in italics). The fragment was ligated into the BamHI site of 35S:sGFP, which encodes a synthetic green fluorescent protein (sGFP) under the control of the constitutive CaMV 35S promoter. Using the same protocol, a fusion construct of the coding regions of the AtMurE cDNA and sGFP was generated using the primers 5′-GGATCCTCTCTGTCTCCAATGGCGTTCACC-3′ and 5′-GGATCCCCGCCATGGGAACTCGCTTGTGTC-3′ (the BamHI site is in italics). The resulting constructs and that of the 35S:sGFP control were introduced into Nicotiana tabacum SR1 leaves using a pneumatic particle gun (PDS-1000/He; Bio-Rad Laboratories, http://www.bio-rad.com). Particle bombardment and observation of GFP signals were performed as described by Motohashi et al. (2001).

RT-PCR analysis of the AtMurE gene

Total RNA was extracted from wild-type and atmurE plants using a NucleoSpin RNA Plant isolation kit (Macherey-Nagel, http://www.mn-net.com), and were then treated with Dnase I. The primers used to amplify the AtMurE gene were 5′-ATGGCGTTCACCTTTCTCTCTCC-3′ (forward primer; first ATG sequence corresponds to the predicted first methionine codon of AtMurE) and 5′-CTTAGCCATCAAACTCTGAACC-3′ (reverse primer; 1227–1206 from the first A in the first methionine codon). The primers used to amplify ACT2 (At3g18780) were 5′-CTAAGCTCTCAAGATCAAAGG-3′ (ACT2F forward primer) and 5′-ACATTGCAAAGAGTTTCAAGGT-3′ (ACT2R reverse primer). Then, 200 ng of RNA was reverse-transcribed into cDNA with an oligo(dT) (16-mer) primer using a PrimeScript RT-PCR kit (Takara Biomedicals, http://www.takara-bio.com). The first-strand cDNA was used as a template for PCR with Blend Taq polymerase (Toyobo Co., Ltd, http://www.toyobo.co.jp/e). PCR was carried out, beginning with an incubation at 94°C for 4 min, followed by 30 cycles of 94°C for 30 sec, 58°C for 1 min and 72°C for 1 min, concluding with an extension at 72°C for 7 min.

Generation of antisense transformants

The RAFL09-95-J16 clone containing the AtMurE cDNA was obtained from the RIKEN Arabidopsis full-length cDNA collection (Seki et al., 2002). Although sequencing of this plasmid showed that one intron remained in the coding region, we used this clone to construct antisense transformants. To produce transgenic Arabidopsis plants in which an antisense AtMurE gene was overexpressed under the control of the constitutive CaMV 35S promoter (Chiu et al., 1996), the SfiI fragment of RAFL09-95-J16 was inserted into the SfiI site of pBE2113SfiI, as described (Myouga et al., 2006). Arabidopsis plants (Columbia) were transformed with Agrobacterium tumefaciensGV3101 using the vacuum-infiltration method (Liu et al., 1998). Transgenic plants containing 35S:antisense-AtMurE were obtained after selection on GM agar plates containing 1% sucrose and 30 μg ml−1 kanamycin.

Total RNA was extracted from the antisense transgenic lines using a NucleoSpin RNA Plant isolation kit. Total RNA (2 μg) was reverse-transcribed using Superscript II RTase (Invitrogen, http://www.invitrogen.com) and 4 μl of random hexamers (50 μm; Amersham, http://www.amersham.com), and 1 μl of the product was used as a template for RT-PCR with Takara ExTaq polymerase (Takara). The primers used to amplify AtMurE were 5′-TCTCTGTCTCCAATGGCGTTCACC-3′ (forward primer) and 5′-GTACGAATTCCGGCGAGCAGGTC-3′ (reverse primer), and those used to amplify ACT2 were ACT2F and ACT2R. PCR was carried out on a Perkin-Elmer 9700 instrument (Applied Biosystems, http://www.appliedbiosystems.com) at 94°C for 1 min followed by 30 cycles of 94°C for 30 sec, 55°C for 30 sec and 72°C for 30 sec, and a final 72°C extension for 8 min.

Measurements of chlorophyll content and fluorescence

Mutant plants were grown for 2 weeks on medium containing 2% sucrose, transferred onto plates containing 5% sucrose and were grown for 1 month under short-day conditions (9-h light/15-h dark). Chlorophyll was extracted from leaves using dimethylformamide. Chlorophyll contents (Chlorophyll a + Chlorophyll b = 17.67A646.8 + 7.12A663.8) were measured with a UV-3000 spectrophotometer (Shimadzu Corp., http://www.shimadzu.com) and calculated according to Porra et al. (1989). The chlorophyll fluorescence of leaves was measured using an FMS1 Pulse Modulated Chlorophyll Fluorometer (Hansatech Instruments Ltd, http://www.hansatech-instruments.com).

Microscopic observations

Arabidopsis thaliana protoplasts were isolated as described by Qi et al. (2004), and were stained for 10 min with a 1/1000 (v/v) dilution of SYBR Green I solution. Fluorescent images of cells were recorded with a CCD camera (Nikon DXM1200, http://www.nikon.com, or Zeiss AxioCam, http://www.zeiss.com) using an Olympus BX60 or Zeiss Axioskop 2 Plus microscope.

For electron microscopy, samples of 14-day-old plants were fixed in 2% glutaraldehyde buffered with 20 mm sodium cacodylate for 4 h and then transferred into 2% osmium tetroxide overnight. The samples were dehydrated through an ethanol series and embedded in Spurr’s resin. Thin sections were cut and then stained with uranyl acetate and lead citrate, and the sections were observed with a JEM-1200EX electron microscope (JEOL, http://www.jeol.com).

Expression analysis of chloroplast-encoded genes

We used the plastid genes for a photosystem-I reaction center protein (psaA), D1 protein of the core complex of photosystem II (psbA), subunit of ATP synthase (atpB), catalytic subunit of the Clp protease (clpP), plastid homolog of the respiratory chain NADH dehydrogenase (ndhB) and a subunit of the acetyl-CoA carboxylase (accD). First-strand cDNA was synthesized as described for the RT-PCR analysis of AtMurE. The sequences of the primers used for the RT-PCR of plastid genes are as follows: 5′-CCATGCTTTAGCACCTGGTGT-3′ and 5′-CCACCCAGAAGGTAATGGGTT-3′ for psaA, 5′-TCTGGGAAGCTGCATCCGTT-3′ and 5′-AGCCTCAACAGCAGCTAGGT-3′ for psbA, 5′-GATACTCGCACAACATCTCC-3′ and 5′-CCGCTGGATAGATACCTTTG-3′ for atpB, 5′-TTCGAAGTCCTGGAGAAGGAG-3′ and 5′-CATGAGCTTGGGCTTCTGTTG-3′ for clpP, 5′-TCTCCCACTCCAGTCGTTGC-3′ and 5′-GGGTATCCTGAGCAATCGCA-3′ for ndhB, and 5′-AGGTGACAACGATCTGCACT-3′ and 5′-CTCCTCCGGAAGAACACACT-3′ for accD. PCR was carried out at 94°C for 4 min followed by an appropriate number of cycles of 94°C for 30 sec, 58°C for 1 min and 72°C for 1 min, with a final extension at 72°C for 7 min. The numbers of PCR cycles used were 25 for psbA and accD, 27 for psaA and ndhB, and 30 for atpB, clpP and Actin.

Moss transformation

A complementation assay was performed using a moss knock-out line in the PpMurE gene, which produces giant chloroplasts (Machida et al., 2006; Figure 7a). A transformant in which the PpMurE gene was replaced with the AnaMurE gene fused to the PpMurE transit peptide coding region was generated as follows. The published genome sequence of Anabaena sp. PCC 7120 was consulted (Kaneko et al., 2001). The AnaMurE gene was amplified by PCR from Anabaena genomic DNA using LA Taq polymerase (Takara), and the forward forward primer 5′-GGGGATATCATGAAATTGCGGGAATTACTA-3′ and the reverse primer 5′-GGCATTGTCCACCCTACACATTCC-3′, which are specific for AnaMurE. In the forward primer, an EcoRV site (set in italics) was added just before the start codon of the AnaMurE gene. The amplified band corresponding to the AnaMurE genomic region was fused into the pT7Blue vector (Merck/EMD Biosciences, Inc., http://www.merckbiosciences.com). The coding region for the PpMurE transit peptide, the functioning of which was experimentally confirmed using GFP fusion protein analysis (Machida et al., 2006), was amplified by PCR with the primers 5′-TTGCAAGCATACACCGGCTG-3′ and 5′-TCTGTCTGTTTGATCCAGTACA-3′. The amplified DNA was subjected to blunting and kination with a Takara BKL kit, and was inserted into the EcoRV site just before the start codon in the AnaMurE gene to create a plasmid encoding AnaMurE fused to the PpMurE transit peptide (PpT-AnaMurE). A fragment corresponding to the PpT-AnaMurE coding region was extracted from the plasmid by digestion at the SpeI and SmaI sites in the pT7Blue vector, and was inserted between these sites in the pBSIIKS-Hm-35S/NT vector to generate the plasmid pBSIIKS-Hm-CaMV35S-PpT-AnaMurE-NT. This vector contains a CaMV 35S promoter-driven hygromycin resistance gene, and a cloning site between another CaMV 35S promoter and the nopaline synthase terminator (Araki et al., 2003). The plasmid pBSIIKS-Hm-CaMV35S-PpT-AnaMurE-NT was linearized by digestion with PvuII. To replace the PpMurE region with PpT-AnaMurE by homologous recombination, the PpT-AnaMurE gene linked to the hygromycin-resistance gene was inserted into the middle of the coding sequence of the cloned PpMurE gene. The linearized fragment was then inserted into the ClaI site of a PpMurE genomic region clone (Machida et al., 2006) to generate a plasmid for moss transformation. Transformation of P. patens was carried out as described by Nishiyama et al. (2000).

To generate a plasmid for replacing the PpMurE gene with the AtMurE gene, AtMurE and the hygromycin-resistance gene were inserted into the cloned PpMurE genomic region. AtMurE was amplified by RT-PCR from 2 μg of A. thaliana total RNA using the primers 5′-TTGCTTTCTCAGTCTCTCTGTCTCCA-3′ and 5′-GTTGGTTGTTTTAGCTGTTACGGTTC-3′. The amplified band was fused into the pT7Blue vector. AtMurE was extracted from the plasmid by digestion with SpeI and SmaI, and was inserted into the pBSIIKS-Hm-35S/NT vector to generate the plasmid pBSIIKS-Hm-CaMV35S-AtMurE-NT. The plasmid pBSIIKS-Hm-CaMV35S-AtMurE-NT was linearized by digestion with PvuII and inserted into the ClaI site of the genomic clone of PpMurE to generate the plasmid PpMurE-Hm-AtMurE-PpMurE-0. Unfortunately, sequence analysis of the clone showed that one intron remained near the end of the AtMurE coding region. Therefore, this region was replaced with the correct one as follows. An accurate DNA sequence was amplified from the correct cDNA clone by PCR using the primers 5′-AGCAGTGGCGGTTGTAGCTAGCAA-3′ and 5′-TACCCACCGCTATGGGAACTCGCTTG-3′. The fragment was digested with BstXI and NheI and inserted between the BstXI and NheI sites of PpMurE-Hm-AtMurE-PpMurE-0 to create PpMurE-Hm-AtMurE-PpMurE-1. Linearized PpMurE-Hm-AtMurE-PpMurE-1 was used for moss transformation.

AtMurE-GFP in moss

To construct the AtMurE-sGFP fusion gene, the region encoding the 100 N-terminal amino acids of the AtMurE gene was amplified from the AtMurE cDNA clone using the primers 5′-TTGTCGACATGGCGTTCACCTTTCTCTCTC-3′ and 5′-AACCATGGCGGAGAGATAAGTGTTCCGATT-3′. The amplified DNA was digested with SalI and NcoI to cleave the restriction sites encoded by the primers, and was inserted into the SalI/NcoI-digested 35SΩ-sGFP(S65T) plasmid (Chiu et al., 1996). P. patens was transformed as described previously (Nishiyama et al., 2000).

Acknowledgements

The authors thank Drs N. Sato (University of Tokyo) and Y. Niwa (University of Shizuoka) for providing Anabaena sp. PCC7120 genomic DNA and a plasmid containing the sGFP(S65T) gene, respectively. The authors also thank the Salk Institute Genomic Analysis Laboratory and the Wisconsin pDS/Lox project for providing the sequence-indexed Arabidopsis T-DNA and DS/Lox insertion mutants, respectively. This study was supported by Grants-in-Aid for Scientific Research on Priority Areas (number 16085208 to HT) from the Ministry of Education, Science, Sports and Culture of Japan, and a grant for Genome Research from RIKEN to KS.

Ancillary