SEARCH

SEARCH BY CITATION

Keywords:

  • rodZ;
  • motility;
  • cell morphology;
  • peptidoglycan

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

RodZ (YfgA) is a membrane protein well conserved among bacterial species and important in the determination of cell shape and motility, although the molecular mechanism involved is not well established. We have characterized a ΔrodZ mutant and show that defective peptidoglycan synthesis might be the primary effect of the deletion. A motile pseudorevertant of ΔrodZ isolated possessed a near rod-shaped cell morphology, indicating that RodZ is not absolutely required for the elongation of the lateral cell wall and the synthesis of functional flagella.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Most membrane proteins of bacteria are involved in the complex metabolic and signal transduction network (Sargent, 2007), and consequently, elucidation of their functions and the detailed molecular mechanisms is awaited. Recently, we performed a genome-wide screening for genes that resulted in a reduced biofilm phenotype when disrupted and identified yfgA, a predicted Escherichia coli gene for a membrane protein, as one such gene. Mutants of yfgA were nonmotile and showed phenotypes characteristic of membrane deficiency (Niba et al., 2007).

Flagella of E. coli are synthesized under the tight regulation of coordinated transcription of over 50 genes categorized into three classes (Chilcott & Hughes, 2000). The class one genes flhD and flhC form the master operon, which is the sole determinant of the fate of flagella biogenesis and motility. FlhD and FlhC proteins form a heterotetrameric complex that binds and regulates promoters of class two genes necessary for hook and basal body formation as well as the flagella-specific sigma factor, fliA, which in turn is required for the expression of class three genes such as fliC that encodes flagellin.

Motility and flagellar assembly are dependent on environmental factors represented by stresses that are sensed by flhDC. In E. coli, several global regulators such as H-NS (Bertin et al., 1994), OmpR (Shin & Park, 1995), CRP-cAMP (Soutourina et al., 1999), LrhA (Lehnen et al., 2002) and RcsAB (Francez-Charlot et al., 2003) are directly involved in the complex genetic regulatory hierarchy that assures ordered assembly of flagellar components.

In rod-shaped cells, a connection between flagellar biosynthesis and cell morphogenesis has been reported. Without flhD, the cell morphology switched from rods to spheres (Prüss & Matsumura, 1996). Furthermore, microarray analysis of the flhD/flhC-regulated promoters identified mreBCD genes that are responsible for rod-shape determination (Prüss et al., 2001). Cell shape is mainly maintained by peptidoglycans that form a protective layer to ensure that cells are not lysed by high internal osmotic pressure (review by den Blaauwen et al., 2008; Vollmer & Bertsche, 2008). Reports have shown that elongation and septation of the peptidoglycan layer are the basis for cell division and growth. MreB, MreC, MreD and RodA as well as the penicillin-binding protein PBP2 are essential for peptidoglycan elongation. A defect in any of these proteins causes cells to become spherical. A number of proteins, including PBP3, are involved in septation, and their loss leads to a filamentous cell morphology.

More recently, yfgA has been shown to participate in rod-shape determination and hence it was named rodZ (Shiomi et al., 2008; Alyahya et al., 2009; Bendezúet al., 2009). The rodZ gene codes for an inner membrane protein that possesses a putative DNA-binding domain of a helix-turn-helix (HTH) motif. Furthermore, RodZ was shown to interact with bacterial actin MreB (Bendezúet al., 2009; van den Ent et al., 2010), which may determine the place of cell wall synthesis (Alyahya et al., 2009; White et al., 2010). However, the exact role of RodZ in cell-shape determination remains to be elucidated. Mutants of rodZ were found to be growth deficient and exhibited a spherical form instead of the normal rod shape.

In this work, we have further characterized the ΔrodZ mutant as well as its pseudorevertant and present evidence that strongly indicates the involvement of rodZ in the biosynthesis of peptidoglycans.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Bacterial strains and plasmids

All E. coli strains used were derivatives of KR0401 (Niba et al., 2007). Deletion mutations ΔrodZkan and ΔsurAkan were introduced from mutants of each gene in the Keio collection (Baba et al., 2006) by P1kc-mediated transduction. lacZ fusions for promoter analysis were constructed as follows: from approximately 500 bp upstream of the first gene of each operon together with approximately 12 bp from the translational start site inside the ORF was PCR-amplified and cloned into pJL28 or pJL30 protein-fusion vectors (Lucht et al., 1994). In the case of the flhB gene, the region from nucleotide −7 to +5 was replaced with that of fliE, because, for unknown reasons, no β-galactosidase activity was obtained with its own initiator codon. pJL28-fliA (pMW198) and pJL29-flgB (pMW211) have been described (Lehnen et al., 2002). Single-copy (chromosomal) lacZ fusion strains were subsequently obtained from plasmid-bearing strains via phage λInCh as described (Boyd et al., 2000). β-Galactosidase activity was measured according to Miller (1972). The plasmid vector for the expression of the cloned gene with the C-terminal S-tag, pBADs, was constructed based on pBAD322 (Cronan, 2006) by replacing the SphI–HindIII region with a 520-bp fragment amplified from pSHLeu (Gan et al., 2002). ΔHTH and Δ(30-133) deletions were introduced using a mutagenesis method of overlap extension reported by Warrens et al. (1997).

Electron microscopy

Cells were grown in Luria–Bertani (LB) medium at 37 °C to the exponential growth phase. Formvar-carbon-coated copper grids were floated for 3 min onto a drop of cell culture, washed on drops of 0.9% NaCl and distilled water and then stained with 1% uranyl acetate. Images were obtained by a transmission electron microscope (H-7100, Hitachi, Tokyo, Japan) at an acceleration voltage of 75 kV.

Isolation of sacculi and peptidoglycan measurement

Peptidoglycan was isolated essentially as described (Hervéet al., 2007). Cells from 200-mL LB cultures grown at 37 °C to an OD600 nm of ca. 0.75 were used. After lysis of cells by the addition of hot 4% sodium dodecyl sulfate solution, followed by overnight incubation at room temperature, a cell wall fraction was obtained by centrifugation at 100 000 g for 1 h, washed with water and suspended in 0.5 mL of water. The amount of peptidoglycan in the isolated sacculi was measured using the silkworm larvae plasma (SLP) reagent set (Wako Pure Chemical Industries Ltd, Osaka) as described previously (Tsuchiya et al., 1996; van Langevelde et al., 1998). The amount of peptidoglycan in the samples was calculated from the standard curve obtained with peptidoglycan of Micrococcus luteus (Wako Pure Chemical Industries Ltd).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The expression of flagella genes is reduced in ΔrodZ mutants

As reported previously, deletion mutants of rodZ (yfgA) are nonmotile (Inoue et al., 2007; Niba et al., 2007). In order to investigate whether this was due to the altered expression of flagella genes in them, their promoter activities were examined using three classes of lacZ fusion constructs of flagella genes (Table 1). The expression of most of the class 2 and class 3 genes examined was apparently reduced. In contrast, the transcription of the class 1 genes flhDC was not reduced, indicating that rodZ does not directly affect the master regulator of flagella synthesis. The tar operon of class 3 that contains genes required for chemotaxis was an exception, suggesting a regulatory mechanism that might not be quite the same as other flagella genes.

Table 1.   Expression of flagella genes in the ΔrodZ mutant and mot+ pseudorevertant cells
ΔrodZ mutant
Classification*β-Galactosidase activity (Miller units)
GeneWTΔrodZRatio (fold)
Class 1flhD6 ± 47 ± 01.17
Class 2fliA83 ± 5737 ± 30.45
flgA9 ± 33 ± 10.38
flgB312 ± 38176 ± 170.56
flhB15 ± 712 ± 20.80
fliE9 ± 44 ± 00.39
fliL14 ± 45 ± 10.33
Class 3fliC4448 ± 599754 ± 2470.17
fliD308 ± 17654 ± 80.17
motA§269 ± 4714 ± 10.05
tar§45 ± 24109 ± 472.41
Nonflagella generpsF§257 ± 21447 ± 411.74
mot+ pseudo revertant
Classification*β-Galactosidase activity (Miller units)
GeneWTΔrodZ-mot+Ratio (fold)
  • Strains carrying a lacZ fusion on the chromosome or plasmids (§) were grown in LB medium at 37°C to an OD600 nm of around 0.5 and their β-galactosidase activities were measured. Ampicillin (100 μg mL−1) was added for the strain harboring plasmid. The table shows averages with SD of three or more independent experiments.

  • *

    The class of flagellar transcriptional hierarchy.

  • Gene promoter that was fused to lacZ.

  • The ratio of β-galactosidase activity in ΔrodZ mutant (upper table) or mot+ pseudo revertant (lower table) to wild type.

  • §

    lacZ fusion is on the plasmid.

  • WT, wild type.

Class 1flhD5 ± 07 ± 31.39
Class 2fliA119 ± 13136 ± 61.14
Class 3fliC4292 ± 28694653 ± 12901.08

Because the growth rate of the ΔrodZ mutant was significantly reduced and the expression of flagella genes might depend on the growth phase, we also monitored β-galactosidase activities of the fusion genes at various growth stages. The fliA and fliC promoter activities were clearly reduced in the ΔrodZ mutant throughout the growth stages examined, while the flhD promoter exhibited similar activities between wild type and mutant cells (data not shown). In addition, during the course of the assay, we observed a significant lysis of ΔrodZ cells after the middle logarithmic growth stage. This seemed to reflect the cell wall defect as we reported previously (Niba et al., 2007).

ΔrodZ mutants are defective in peptidoglycan synthesis

As the expression of most flagella genes was reduced, but still present at a significant level in the ΔrodZ mutant, we examined their flagella by electron microscopy (Fig. 1). As reported (Shiomi et al., 2008; Bendezúet al., 2009), mutant cells were mostly round. Surprisingly, however, they possessed many flagella especially at the late logarithmic phase. At this growth stage, many of the mutant cells appeared not only of a spherical shape, but swollen with absorbed staining solution and their contours were not clear (Fig. 1c). Some resembled broken sacculi without contents (Fig. 1d). These aberrant phenotypes were suppressed by the introduction of a low-copy plasmid pBADs-rodZ that expressed a tagged RodZ. However, this was not the case with its derivative pBADs-rodZΔHTH that lacked the HTH motif of RodZ (amino acid residues 30–49). Therefore, we interpreted the results to indicate that the HTH motif is essential for the function of RodZ. The ΔrodZ cells carrying plasmid pBADs-rodZΔHTH also remained nonmotile (data not shown).

image

Figure 1.  Cell shape and flagella of strains carrying the rodZ mutation observed by electron microscopy. (a) KR0401/pBADs (wild type); (b–d) KR0401ΔrodZ; (e) KR0401ΔrodZ/pBADs; (f) KR0401ΔrodZ/pBADs-rodZ; (g) KR0401ΔrodZ/pBADs-rodZΔHTH; (h) KR0401ΔrodZ/prodZ-1-ΔHTH; (i) KR0401ΔrodZ-mot+; (j) KR0401ΔrodZ-mot+/pBADs-rodZ. Scale bar indicates 2 μm. (k) The structure of tagged RodZ and its ΔHTH derivative cloned in plasmid pBADs. CD, cytoplasmic domain; TM, transmembrane domain; PD, periplasmic domain.

Download figure to PowerPoint

The above observation and previous reports suggest that the periplasmic peptidoglycan structure may not be intact in the ΔrodZ mutant and consequently it cannot protect the cell from lysis by the internal osmotic pressure. To investigate this possibility, we first examined the sensitivity of the mutant cells to hyper- and hypo-osmotic conditions. As shown in Fig. 2a, the growth rate of mutants was about twofold reduced in hypoosmotic medium (LB without NaCl), whereas the effect of hyperosmotic medium (LB with 0.4 M NaCl) on mutant cells was smaller. In contrast, the growth rate of the deletion mutants of surA that encodes a periplasmic chaperon was drastically reduced in hyperosmotic medium, but only mildly under hypo-osmotic pressure. The surA gene product is important for the synthesis of OMPs (Lazar & Kolter, 1996; Rouvière & Gross, 1996) and its mutant showed a synthetic lethal phenotype with ΔrodZ (Niba et al., 2007). Furthermore, the culture of ΔrodZ mutant cells showed a sharp decline in OD600 nm when diluted with water instead of LB medium (Fig. 2b), whereas this was not observed with ΔsurA and wild-type cells. This strongly indicates that ΔrodZ cells are spheroplast like. However, this phenotype was less evident in stationary-phase cultures, which may be due to the physiological change of mureins associated with the growth stage or nutritional starvation (Goodell & Tomasz, 1980; Glauner et al., 1988).

image

Figure 2.  Growth of ΔrodZ mutant cells under hyper- and hypo-osmotic conditions. (a) Cells were grown at 37°C in LB (closed circle), LB with 0 M (open circle), 0.3 M (cross) and 0.4 M (triangle) NaCl, and OD600 nm were measured at indicated time points. The experiment was repeated three times and similar results were obtained. Representative data are shown for each strain. (b) Culture of cells at the mid- (black), late-logarithmic (stripe) or stationary phase (white) was diluted (10-fold) either with LB medium or water. The ratio of measured OD600 nm of water to LB dilution was presented. Averages of three independent experiments are shown with SD.

Download figure to PowerPoint

To further clarify whether peptidoglycan of the mutant cells was defective, we quantified peptidoglycan of the mutant and wild-type cells using SLP reagent. The amount of peptidoglycan in the ΔrodZ mutant was calculated to be about 20% of the wild type (Table 2), a value well below 50% at which no detectable morphological change or slow growth was observed (Prats & de Pedro, 1989). This strongly indicates that the defective synthesis of peptidoglycan was the reason why the ΔrodZ mutant was very sensitive to hypo-osmotic pressure and exhibited significant cell lysis in liquid culture. The severe reduction of peptidoglycan observed with the ΔrodZ mutant was, however, less apparent in a later growth stage as in the case of the spheroplast-like phenotype described above, which seems to suggest that the ΔrodZ mutant is basically able to synthesize peptidoglycan, but is unable to coordinate it with cell growth.

Table 2.   Amount of PG in ΔrodZ mutant cells
 AVSD%
  1. The amount of PG in isolated sacculi (μg) was normalized by the OD600 nm of culture. The averages (AV) and SD are from four independent analyses.

  2. WT, wild type.

KR0401 (WT)1891695
KR0401ΔrodZ33619718
KR0401ΔrodZ-mot+163767587

The rodZ and ispG genes comprise one operon

On the chromosome of E. coli and most of proteobacteria, rodZ is followed by ispG, an essential gene for isoprene synthesis. Because isoprene is required for the biosynthesis of peptidoglycan (Bouhss et al., 2008), the above results might support an idea that rodZ is functionally related to ispG. Therefore, we first investigated whether rodZ and ispG are transcribed together or not using lacZ fusion constructs, prodZ-1 and prodZ-2 (Fig. 3). The results showed that this was indeed the case and ispG is mostly expressed from the promoter located upstream of rodZ, although a minor transcription activity was still observed when this promoter was eliminated in prodZ-2 (Table 3).

image

Figure 3.  Structure of the rodZ-ispG operon and the lacZ gene-fusion. The rodZ-ispG operon and the region fused to lacZ (thick bar) in each plasmid are depicted. //, deleted part; TM, transmembrane domain.

Download figure to PowerPoint

Table 3.   Expression of ispG-lacZ fusion in the wild type and ΔrodZ mutant
Plasmid*Host strain
Wild typeΔrodZΔrodZ-mot+
ActivityRatio 1ActivityRatio 1Ratio 2ActivityRatio 1Ratio 2
  • Strains carrying indicated plasmids were grown in LB medium containing ampicillin (100 μg mL−1) at 37°C to an OD600 nm of around 0.5, their β-galactosidase activities (Miller units) were measured and average values with SD of three or more independent experiments are shown. The activity of the pJL30 vector only yielded a ribosome-binding site and initiation codon ATG was <0.1.

  • *

    The structure of lacZ-fusion plasmids is depicted in Fig. 3.

  • The ratio of β-galactosidase activity of the constructs containing either a ΔHTH or a Δ(30-133) deletion relative to that of the construct without deletion.

  • The ratio of the β-galactosidase activity of the mutant relative to that of the wild type.

  • §

    pJL30 of the lolB operon containing an ispElacZ fusion.

  • ND, not done.

prodZ-1934 ± 301026 ± 5821.1965 ± 631.0
prodZ-1-ΔHTH527 ± 660.56805 ± 1420.781.5517 ± 500.540.98
prodZ-1-Δ(30-133)469 ± 840.501610 ± 3361.63.4441 ± 860.460.94
prodZ-299 ± 3114 ± 201.1123 ± 231.2
prodZ-2-ΔHTH61 ± 80.61174 ± 471.52.982 ± 130.671.4
prodZ-2-Δ(30-133)52 ± 70.53130 ± 51.22.554 ± 80.441.0
prodZ-31591 ± 5752549 ± 2531.61238 ± 1560.78
prodZ-440 ± 273 ± 81.9ND  
prodZ-547 ± 0.4140 ± 13.0ND  
prodZ-679 ± 2146 ± 41.9ND  
prodZ-764 ± 4146 ± 592.3ND  
plolB-ispE§527 ± 56436 ± 640.8353 ± 510.67

Next, we examined the effect of the ΔrodZ mutation on the expression of rodZ and ispG by constructing prodZ-3 and derivatives of prodZ-1 and prodZ-2 with a deletion in the N-terminal part of RodZ that prevents the production of intact RodZ (Fig. 3) and introducing them into the ΔrodZ mutant and wild type. The β-galactosidase activity of prodZ-3, prodZ-1-ΔHTH and prodZ-1-Δ(30-133) was 1.6, 1.5 and 3.4-fold higher, respectively, in the ΔrodZ mutant. In wild-type cells, however, the expression of ispG was decreased about 50% when rodZ on the plasmid was partially deleted, which might indicate that the RodZ protein is required for the coordinated synthesis of rodZ and ispG. Interestingly, the expression of ispG from prodZ-1-ΔHTH was reduced in the rodZ mutant, although to a lesser extent compared with the wild type, indicating that the RodZ lacking the HTH domain might partially retain its function. Also, the ΔrodZ mutant carrying this plasmid grew slightly faster.

Finally, in order to locate the minor promoter(s) observed with prodZ-2, we constructed additional lacZ fusions (Fig. 3) and examined their β-galactosidase activity (Table 3). The results showed that, indeed, a promoter(s) existed within the rodZ-orf as well as in the intergenic region, both of which showed higher activity in the ΔrodZ mutant compared with the wild type, while the expression of ispE, another gene involved in isoprene synthesis and located in a different operon, was not increased, suggesting that the effect of RodZ is specific to the expression of the rodZ-ispG operon. It seems that a balanced expression of some sorts between rodZ and ispG might be important, although we were unable to explain these results in an unequivocal manner.

RodZ is not absolutely required for motility and rod shape

During the analysis described here, we often encountered inconsistent results with the ΔrodZ mutant and noticed that derivatives that were motile and grew faster emerged spontaneously within the population. By PCR analysis, we confirmed the presence of the ΔrodZ (rodZkan) mutation in those faster-growing derivatives (data not shown). Subsequently, we isolated one such pseudorevertant, termed KR0401ΔrodZ-mot+, and characterized the phenotype. The cells grew and expressed fliA and fliC at a level similar to that of the wild type (Table 1). The cell shape was almost rod type, although more irregular and asymmetrical compared with the wild type (Fig. 1i). The cells tended to be more elongated than the wild type in contrast to the original ΔrodZ mutant. When extra copies of rodZ were introduced, some cells showed a filamentous morphology (Fig. 1j). The amount of peptidoglycan was also significantly higher than the original ΔrodZ mutant (Table 2). Furthermore, the expression of the plasmid-borne ispG measured by the fused lacZ activity was decreased as in the wild type when either the ΔHTH or the Δ(6-30-133) deletion was introduced (Fig. 3, Table 3). Because this mutant was obtained spontaneously, it is likely that a single event that affected the property of the membrane or the cell wall led to all these phenotypic changes.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Our analyses of the ΔrodZ mutant showed that the absence of RodZ leads to the reduced expression of most of the flagella genes, but not of their master regulator, which seems to suggest that RodZ does not function as a regulatory factor for this large network. It seems that the reduced expression could be due to stress signals from a defective cell wall. The rodZ mutant was nonmotile, but possessed flagella that were indistinguishable from those of the wild type. The expression of the motA operon required for membrane-bound components of flagellar motor and chemotactic response was most severely reduced in the mutant (Table 1). This might explain the above-mentioned phenotype of the mutant because mutations in motA and motB impaired the motility, but not the assembly of flagellum (Blair & Berg, 1990). Cells of the rodZ mutant, however, were able to move actively in liquid medium, although the number of active cells was much less than that in the wild type. Therefore, the nonmotile phenotype might be due to both defective motor synthesis and loss of proper chemotactic function. It is also conceivable that the weakened membrane structure and/or the altered cell shape hindered the movement of cells through soft agar, where more pressure is expected than in liquid medium.

We have confirmed that rodZ and ispG comprise an operon and the absence of RodZ apparently affected the expression of ispG. Because its overproduction is toxic to cells (GenoBase: http://ecoli.aist-nara.ac.jp/index.html), IspG might play another role in the peptidoglycan metabolism in addition to isoprene synthesis (Campos et al., 2001), although it has not yet been revealed in E. coli. In Providencia stuartii, a homologue of ispG termed aarC regulates the expression of 2′-N-acetyltransferase that contributes to the O-acetylation of peptidoglycan, and a missense mutant of aarC showed a phenotype similar to the rodZ mutant (Rather et al., 1997). O-acetylation influences the activity of lytic transglycosylases involved in the biosynthesis and turnover of peptidoglycan (reviewed in Scheurwater et al., 2008). Therefore, RodZ might function in the fine-tuning of peptidoglycan biosynthesis.

Plasmid pBADs-rodZΔHTH could rescue neither the sphere cell shape nor the nonmotile phenotype contrary to the recent reports by Shiomi et al. (2008) and Bendezúet al. (2009). We assume that this discrepancy is due to the amount of ΔHTH molecules expressed, because we occasionally observed elongated cells among the ΔrodZ mutant carrying plasmid prodZ-1-ΔHTH that should produce more proteins than pBADs-rodZΔHTH (Fig. 1g). The growth rate of the ΔrodZ mutant with this plasmid was also higher than that with pBADs-rodZΔHTH. These observations, along with the inner membrane localization of RodZ, might suggest that the function of RodZ is to provide a gateway for regulating the transport through the cell membrane and assembly of molecules such as precursors of peptidoglycan, lytic glycosylases or flagella components. The HTH domain may contribute to this process by interacting with various protein molecules and localizing RodZ itself into the membrane. For these reasons, a higher expression of RodZΔHTH than the intact RodZ might have been required to complement defects caused by the ΔrodZ mutation.

Nonetheless, RodZ was not absolutely required for the rod shape. We isolated pseudorevertants of the ΔrodZ mutant (KR0401ΔrodZ-mot+). They possessed a rod shape, although cells were irregular and not well balanced as the wild type. It was reported that RodZ interacts with and anchor MreB to the inner membrane, promoting the helical assembly of actin cytoskeleton (Bendezúet al., 2009; van den Ent et al., 2010). We speculate that the function of RodZ in the lateral synthesis of the cell wall was somehow compensated in the pseudorevertants, although the proper assembly of MreB was still lost due to the absence of RodZ and consequently the rigid rod shape was not maintained. Because rodZ is an essential gene in Caulobacter (Alyahya et al., 2009), E. coli might have another gene or mechanism that can complement the loss of rodZ. Genome-wide differential gene expression analysis of the ΔrodZ-mot+ derivative will be interesting and important to elucidate the function of rodZ in relation to cell morphogenesis.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Drs Gottfried Unden (Johannes Gutenberg Universität Mainz, Germany) and John Cronan (University of Illinois, USA) for providing us with plasmids and Dr Francis Bivelle (Institut Pasteur, France) for λInCh. We are grateful to Dr Toshinobu Suzaki (Kobe University, Japan) and members of his laboratory for kindly providing TEM facilities and helping us in electron microscopic analysis. We also thank Dr Katsumi Isono of the Kazusa DNA Research Institute for his critical reading of the manuscript.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References