Overexpression of mltA in Edwardsiella tarda reduces resistance to antibiotics and enhances lethality in zebra fish


Xiao-Hua Zhang, College of Marine Life Sciences, Ocean University of China, 5 Yushan Road, Qingdao 266003, China. E-mail: xhzhang@ouc.edu.cn


Aims:  The aim of this study was to investigate the role of membrane-bound lytic murein transglycosylase A (MltA) in a bacterial fish pathogen Edwardsiella tarda.

Methods and Results:  An mltA in-frame deletion mutant (ΔmltA) and an mltA overexpression strain (mltA+) of Edw. tarda were constructed through double-crossover allelic exchange and by transformation of a low-copy plasmid carrying the intact mltA into the ΔmltA mutant, respectively. Either inactivation or overexpression of MltA in Edw. tarda resulted in elevated sensitivity to β-lactam antibiotics and lower viability in oligotrophic or high osmotic environment than wild-type strain. Autolysis induced by EDTA was reduced in ΔmltA strain, while mltA+ strain was virtually flimsy, indicating that MltA is responsible for the lysis effect. Moreover, mltA+ strain exhibited significant increases in lipopolysaccharide (LPS) biosynthesis and virulence to zebra fish compared with wild-type strain.

Conclusions:  The results indicated that MltA plays essential roles in β-lactam antibiotics and environmental stresses resistance, autolysis, LPS biosynthesis and pathogenicity of Edw. tarda. This is the first report that MltA has a virulence-related function in Edw. tarda.

Significance and Impact of the Study:  This study provided useful information for further studies on pathogenesis of Edw. tarda.


Edwardsiella tarda is a Gram-negative bacterium and a causative agent of septicaemia in freshwater and marine fish which is responsible for significant economic loss in the aquaculture industry (Thune et al. 1993; Abbott and Janda 2006; Mohanty and Sahoo 2007; Ye et al. 2009).

The pathogenicity of Edw. tarda is polyfactorial and some of the virulence properties have been implicated, including dermatotoxins (Ullah and Arni 1983), the haeme-iron uptake system (Janda and Abbott 1993), haemolysins (Hirono et al. 1997), phagocytic killing (Srinivasa Rao et al. 2001), a type III and a type VI secretion system (T3SS and T6SS) (Tan et al. 2005; Zheng and Leung 2007), the ability to produce biofilm (Zhang et al. 2008) and invasion of epithelial cells (Xu et al. 2010). However, the major virulence factors of Edw. tarda are still poorly understood. To prevent and treat edwardsiellosis, it is necessary to fully understand the pathogenicity mechanism of Edw. tarda.

Lytic transglycosylases (LTs) are a class of autolysins, which are capable of cleaving the β-1,4 glycosidic bond between N-acetylmuramic acid and N-acetylglucosamine residues of peptidoglycan, with concomitant formation of 1,6-anhydromuramic acid residues (Höltje et al. 1975). Previous studies have arranged known and hypothetical LTs into four families based on amino acid sequence similarities and identified consensus motifs (Blackburn and Clarke 2001). Of these, family 1 represents a superfamily containing five subfamilies (named 1A-1E) involving sequences with identity to Slt, MltC, EmtA, MltD and YfhD from Escherichia coli; family 2 and 3 represent enzymes involving sequences with identity to MltA and MltB from E. coli, respectively; family 4 represents enzymes deriving from bacteriophage lambda LT. However, the specific role of each family is largely unknown. By far, LTs have been suggested to be involved in numerous cellular processes including cell growth and separation (Höltje and Herdrich 2001; Heidrich et al. 2002), cell wall turnover (Dillard and Seifert 1997), motility (Rashid et al. 2002), drug resistance (Xu et al. 2011), protein secretion (Zahrl et al. 2005), differentiation (Foster 1994) and pathogenicity (Cloud-Hansen et al. 2006; Xu et al. 2011).

The annotation of the virulent Edw. tarda EIB202 genome indicates the presence of a putative membrane-bound LT gene with the locus tag as ETAE_0715 (Wang et al. 2009). This gene is predicted to encode a 383-amino acid protein which has a 3D domain at the C-terminal end with 75% identity to the membrane-bound lytic murein transglycosylase A (MltA) in E. coli. Therefore, the gene was designated as mltA and the putative protein was predicted to function as LT. In this study, as an attempt to interpret the role of MltA in Edw. tarda, we (i) constructed an mltA in-frame deletion mutant (ΔmltA) and an mltA overexpression strain (mltA+) of Edw. tarda EIB202; (ii) compared ΔmltA and mltA+ strains with the wild-type strain in terms of biological properties, antibiotic susceptibilities and virulence to fish to provide useful information for further studies on the pathogenesis of Edw. tarda.

Materials and methods

Ethics statement

The zebra fish (Dario rerio) used for virulence tests in this study are cultured animals, and all the experiments are programed in strict accordance with the regulations of local government.

Bacterial strains, plasmids and cultivation conditions

Bacterial strains and plasmids used in this study were represented in Table 1. Edwardsiella tarda EIB202 was isolated and identified by phenotypic tests and 16S rDNA sequencing in previous study (Xiao et al. 2008). Plasmids were introduced into E. coli strains by transformation and into Edw. tarda strains through mating with E. coli SY17-1 (λpir). Edwardsiella tarda strains were routinely grown at 28°C in tryptic soy broth (TSB; Difco, Detroit, MI, USA), tryptic soy agar (TSA) or Luria–Bertani (LB, Difco) broth, while E. coli strains were cultured at 37°C in LB broth. When required, antibiotics (Sigma) were added to the media at the following concentrations: ampicillin (Amp, 50 μg ml−1), kanamycin (Kan, 50 μg ml−1) and colistin B (ColB, 12·5 μg ml−1). The identity of Edw. tarda EIB202 was verified by 16S rDNA sequencing and gyrB-based PCR test (Lan et al. 2008).

Table 1.   Strains and plasmids used in this study
Strains or plasmidsCharacteristicsReferences or source
Edwardsiella tarda
 EIB202Pathogenic isolated from a mariculture farm in Yantai, China. CmrXiao et al. (2009)
 ΔmltAEIB202, in-frame deletion of mltAThis study
 mltA+EIB202, overexpression of mltA in ΔmltA complemented with intact mltAThis study
Escherichia coli
 SY327 (λpir)Δ (lac pro) argE(Am) rif malA recA56
rpoB λ pir, host for π-requiring plasmids
Miller and Mekalanos (1984)
 SY17-1 (λpir)Tpr Smr recA thi pro rK- mK-RP4:2-Tc:MuKm Tn7 λ pir (thi pro hsdR hsdM+ recA RP4-2-Tc:Mu-Km-Tn7)Simon et al. (1983)
 JM109endA1 hsdR17 gyrA96Δ(lac proA) recA1 relA supE44 thi F (lacIqlacZΔM15 proAB+traD36)Yanisch-Perron et al. (1985)
 pUCm-TCloning vector, AmprSangon, Shanghai
 pRE118Suicide vector for allelic exchange, KanrEdwards et al. (1998)
 pRE118ΔmltApRE118 derivative containing mltA bp 1–188 fused in-frame to bp 906–1152, KanrThis study
 pACYC184Cmr, TcrRose (1988)
 pACYC184KpACYC184 derivative with Kan fragment inserted in BamHI and XbaI sites, Cmr KanrThis study
 pACYC184K-mltApACYC184K derivative containing 1·4 kb fragment of mltA putative promoter and ORF, Cmr KanrThis study

DNA manipulations and analysis

Plasmid and genome DNA extraction, recombinant DNA techniques, bacterial transformation and agarose gel electrophoresis were performed as described by Ausubel et al. (1987), Joseph and David (2001) or according to the manufacturer’s protocols. Restriction enzymes and DNA modification enzymes were purchased from TaKaRa (Dalian, China).

Construction of mltA in-frame deletion mutant

All primers described in this study were shown in Table 2. The sequence (1152 bp) of mltA (ETAE_0715) (Genbank ID: YP_003294771.1) was acquired from NCBI (National Center for Biotechnology Information). An mltA in-frame deletion mutant (ΔmltA) of Edw. tarda EIB202 was constructed by overlap PCR (Mo et al. 2007) and double-crossover allelic exchange using suicide vector pRE118 (Edwards et al. 1998). Briefly, the upstream fragment (251 bp) and downstream part (277 bp) of mltA were amplified by PCR with primer pair mltA-MF/mltA-MR1 and mltA-MF1/mltA-MR, respectively. The DNA fragment containing the ΔmltA was amplified using primers mltA-MF and mltA-MR by overlap PCR. The fused segment (ΔmltA) was sequenced and then ligated into the XbaI/KpnI sites of pRE118. The resulting plasmid, pRE118-ΔmltA, was transformed into E. coli SY327 (λpir) and then introduced into E. coli SY17-1 (λpir) (Liang et al. 2003) for mating into Edw. tarda EIB202 by conjugation. The transconjugants carrying the plasmid pRE118-ΔmltA integrated into the genome through a homologous recombination were screened on TSA containing Kan and ColB. Allelic exchange between the chromosomal gene and the in-frame deleted plasmidic copy was achieved in a second crossover event which was counter-selected on TSA containing 10% sucrose to demonstrate the excision of pRE118 from the chromosome. The mutant, without the targeted fragment in the chromosome of Edw. tarda EIB202, was screened by Kan susceptibility and ColB resistance and determined by PCR using primers mltA-MF/mltA-MR.

Table 2.   Primers used for cloning and qRT-PCR
PrimersSequences (5′–3′)
  1. *Nucleotides in bold represent restriction enzyme sites added to the 5′ region of the primers.

  2. †Nucleotides underlined represent the 15-bp overlap sequence.


Construction of a low-copy plasmid (pACYC184K) carrying antibiotic (Kan) resistance gene for complementation

Considering that Edw. tarda EIB202 is capable of resisting both chloramphenicol (Cm) and tetracycline (Tc) (Wang et al. 2009), the low-copy plasmid pACYC184 carrying a Tc resistance gene was replaced by a Kan-resistance gene to be used as a complement plasmid. Briefly, the Kan-resistance gene carrying BamHI and XbaI restriction sites was amplified from pRE118 with specific primer pair KanF/KanR and sequenced, and it was then subcloned into BamHI- and XbaI-digested pACYC184. The resultant plasmid, designated as pACYC184K, was transformed into E. coli JM109 competent cells to confirm whether the antibiotic (Kan)-resistance gene was expressed.

Construction of the complemented strain (mltA+)

An intact mltA gene harbouring the putative promoter region, coding for MltA protein, was amplified using specific primers mltA-U/mltA-D, which was cloned into the linear plasmid pUCm-T for sequencing (Invitrogen Biotech, Shanghai, China). The resultant plasmid was designated as pUCM. The intact mltA gene fragment was obtained by digesting pUCM with XbaI and then subcloned into XbaI-digested pACYC184K to create pACYC184K-mltA. The constructed plasmid was then electrotransformed into the ΔmltA mutant to gain the complemented strain (mltA+). ColB- and Kan-resistant transformants were screened, and the existence of the plasmid was confirmed by PCR analysis and sequencing.

Quantitative reverse transcription PCR (qRT-PCR) analysis of RNA

Temporal expression of the mltA operon was monitored by qRT-PCR analysis. Edwardsiella tarda strains were grown in TSB to OD540 of 0·5, and RNAs were isolated from corresponding samples as described by Simms et al. (1993). Contamination DNA was removed from each sample by DNase I treatment with the RNase-free kit (MBI Fermentas, Heidelberg, Germany) according to the manufacturer’s protocols. Reverse transcription was performed as described previously (Rice et al. 2003). The random primer (6 mer) was used to produce cDNA templates.

The qRT-PCR was performed by conducting three independent experiments, each in triplicate, with an ABI 7500 detector (Applied Biosystems, Foster City, CA, USA), and transcript levels were normalized to 16S rRNA in each sample according to the method described by Tian et al. (2008). The primer pairs (16S-qF/16S-qR, mltA-qF/mltA-qR) for qRT-PCR listed in Table 2 were designed using Primer Express software (Applied Biosystems) with predicted products in the 100- to 200-bp size range.

Antibiotic susceptibility assay

Antimicrobial susceptibility test was performed as recommended by the Clinical and Laboratory Standards Institute (CLSI). Growing Edw. tarda strains were tested on Muller–Hinton Agar (MHA) supplemented with 1% NaCl in triplicates using disc diffusion method. Diameters of the inhibition zones were measured after incubation at 28°C for 24 h.

The bacteria were grown in LB broth at 28°C to an OD540 of 0·5 and then ampicillin was added at the concentration of 10 μg ml−1. The RNAs of corresponding samples were isolated after being placed at 28°C for 30 min without shaking. The qRT-PCR was performed as described above.

Determination of growth kinetics

Overnight cultures of the three strains were adjusted to equal density based on OD540 readings as bacterial pre-inoculum. They were inoculated into 300 ml fresh LB with the scale of 1 : 100 in triplicates and were then cultivated at 28°C with constant agitation. Growth in LB was monitored every 2 h by optical density readings in a Spectrophotometer before the bacterial densities began to decrease.

Stress survival assays

The methods for testing the changes in survivability in M9 medium supplemented with 0·02% glucose and LB medium added with 2 mol l−1 NaCl were conducted as described previously (Xiao et al. 2009). Aliquots of the cell suspension were taken after defined time periods for plate counts analysis after appropriate dilution.

Autolysis induced by EDTA was carried out by the method described previously with minor modifications (Leduc and van Heijenoort 1980). Briefly, early exponential phase cells (OD540 = 0·4) cultivated in TSB were harvested by centrifugation for 10 min at 3000 g. The supernatant was removed with cotton tips as much as possible. The unwashed pellet was immediately suspended by vigorous agitation in 25 ml of 0·01 mol l−1 EDTA (pH 8·5). After absorbance measurements at 0 h, each sample was transferred to a 100-ml flask and placed in a water bath (28°C) without shaking. An aliquot was taken to determine the survival rate every 2 h by plate counts after appropriate dilution.

Swimming motility assay

Swimming motility was evaluated by point inoculating Edw. tarda strains on TSA plates containing 0·3% agar. The plates were analysed after incubation at 28°C for at least 16 h. The experiment was performed in triplicate.

Quantitative biofilm formation assay

Quantification of biofilm formation was performed as described previously (Stepanovićet al. 2007) with minor modifications. Briefly, overnight cultures of Edw. tarda and its derivatives were adjusted to OD540 = 0·5, and they were then diluted 1 : 20 with fresh TSB in triplicates. 96-well microtiter plates were inoculated with bacterial suspensions (200 μl per well) and incubated statically at 28°C for 2 days. The suspensions were removed and each well was washed for three times with sterile distilled water, and then, the samples were fixed with 150 μl of methanol for 20 min before inversion for at least 30 min. The absorbance of each well stained with 150 μl of 2% crystal violet was measured at 570 nm using an ELISA reader following resolubilization of the samples for 30 min with 150 μl of 95% ethanol.

Extraction and analysis of LPS

Lipopolysaccharide (LPS) was extracted from plate cultures using the proteinase K method with minor modifications (Edwards et al. 2000). Briefly, the bacterial strains were harvested from TSA media and adjusted to OD540 = 1·0 with fresh PBS. One millilitre of the bacterial suspension was then mixed with 500 μl of concentrated LPS digestion buffer 1 [0·1875 mol l−1 Tris–HCl, pH 6·8; sodium dodecyl sulfate (SDS), 6% (w/v); glycerol, 30% (v/v)], and the mixture was boiled for 5 min to ensure complete cell lysis. One hundred microlitres of the lysate was then added to 350 μl of LPS digestion buffer 2 [0·0625 mol l−1 Tris–HCl, pH 6·8; SDS, 0·1% (w/v)] and incubated with proteinase K (100 μg ml−1) for 4 h at 55°C. The samples supplemented with bromophenol blue 1% (w/v) were separated by SDS-polyacrylamide gel electrophoresis. The LPS profile was visualized by silver staining (Tsai and Frasch 1982).

Haemolytic activity and chondroitinase assay

The strains were tested for haemolytic activity on agar base (Oxoid) supplemented with 5% sheep erythrocytes. The β-haemolytic activity was indicated by the presence of a colourless zone surrounding the colonies (Gerhardt et al. 1981). As previously described (Shotts and Cooper 1992), chondroitinase activity was identified as a clear area around the colonies against a cloudy background.

Virulence tests in zebra fish

Zebra fish (D. rerio) from quarantined stocks recognized as disease-free (Austin and Austin 1989), as model organisms, were used to assess the virulence of Edw. tarda strains (EIB202, ΔmltA and mltA+). The animals (average weight = 0·3 g) were randomly divided into 22 groups, each comprising 16 fish. Then, with 1 ml of disposable syringes, three paralleled groups were injected intraperitoneally with 20 μl bacterial cells adjusted to required concentrations (102, 103, 104, 105, 106, 107 and 108 CFU ml−1). Bacterial concentrations were determined by microscopy and confirmed with total viable (colony) counts after plating appropriate dilutions onto TSA media with incubation at 28°C for 2 days. Each fish from the control group was injected with 20 μl sterile saline (negative control). To make the experiments reproducible, bacteria were grown to late logarithmic phase in TSB broth, washed and diluted in sterile saline. The infected fish were maintained in static fresh water for 2 weeks, moribund fish were examined microbiologically, and mortalities were recorded. The lethal dose 50 was calculated by the statistics method (Wardlaw 1985).

Zebra fish were also used to assess the virulence of crude LPSs extracted from the three Edw. tarda strains. The animals (average weight = 0·3 g) were randomly divided into 12 groups, each comprising ten fish. Then, three paralleled groups were injected intraperitoneally with 20 μl LPSs diluted in sterile PBS (100, 10−1, 10−2). Each fish from the three control groups, also diluted in sterile PBS, was injected with 20 μl LPS digestion buffer 2 supplemented with proteinase K (100 μg ml−1). All the extracted crude LPSs and negative control were heated to 95°C for 10 min to inactivate proteinase K. The injected fish were maintained in static fresh water for 5 days, and mortalities were recorded.

Statistical analysis

Error bars shown in the graphs are standard deviations, and the data were represented as mean ± SEM. The statistical analysis (variance analysis) was performed using spss statistics 17.0 software, with the Tukey–Kramer post-test for multiple comparisons. Consequently, < 0·05 and < 0·01 were taken to show statistical significance and distinct significance, respectively.


Construction of mltA in-frame deletion mutant (ΔmltA) and complemented strain (mltA+)

Using the double selection strategy of allelic exchange mutagenesis with suicide vector pRE118, we deleted a 717-bp (residues 189–905) central portion of the mltA from Edw. tarda EIB202, thus obtaining the ΔmltA mutant with loss of the MltA from amino acid 63–302 function as binding peptidoglycan. The mutant was identified by the ability to survive on TSA supplemented with ColB and inability to grow with Kan. The correct deletion was confirmed by DNA sequencing of the resulting PCR product (data not shown).

Complemented strain was constructed by electrotransforming the plasmid pACYC184K harbouring the intact mltA gene into ΔmltA. ColB- and Kan-resistant transconjugants were selected, and the existence of the plasmid was confirmed by PCR analysis and sequencing. The ΔmltA and mltA+ mutants were grown on TSA for 30 generations, indicating that they were steady maintained.

Analysis of expression level of mltA using qRT-PCR

The qRT-PCR analysis showed that mltA was not transcribed in ΔmltA, however, the expression level of mltA in mltA+ increased by c. 118-fold compared with that in the wild-type strain, which was unexpected. Therefore, we redefined mltA+ as an mltA overexpression Edw. tarda strain.

Effect of mltA on antibiotics resistance

Antibiotics susceptibility test with disc diffusion method showed that the ΔmltA and mltA+ mutants had elevated sensitivity to β-lactam antibiotics, such as ampicillin (P < 0·01), carbenicillin (P < 0·01), piperacillin (P < 0·01) and rocephin (P < 0·01) (Table 3), compared with wild-type strain, indicating that ΔmltA and mltA+ mutants significantly reduced the resistance to β-lactam antibiotics.

Table 3.   Sensitivity of Edwardsiella tarda EIB202, ΔmltA and mltA+ strains to antibiotics
Antibiotics*Dosage (μg per disc)Diameter of inhibition zone (mm)
  1. *Only significant differences were shown. Data are the means for three independent experiments and presented as mean ± SEM.

  2. **< 0·01 (highly significant) vs corresponding values of Edw. tarda EIB202 (Tukey’s test).

Ampicillin1031·0 ± 1·035·0 ± 0·6**40·0 ± 1·0**
Carbenicillin10028·3 ± 0·636·7 ± 1·0**40·0 ± 1·5**
Piperacillin10026·0 ± 1·040·7 ± 1·2**44·3 ± 0·6**
Rocephin3031·0 ± 0·637·3 ± 0·6**45·0 ± 1·2**

The qRT-PCR analysis showed that the expression level of mltA in Edw. tarda EIB202 treated with ampicillin (10 μg ml−1) increased by c. 7-fold compared with that treated without ampicillin, confirming that MltA contributes to resist β-lactam antibiotics in Edw. tarda. However, the expression level of mltA in mltA+ in the presence of ampicillin (10 μg ml−1) decreased by c. 1·7-fold compared with that in the absence of ampicillin, further indicating that overexpression of mltA resulted in elevated sensitivity to these antibiotics.

Effect of mltA on growth

When cultured in LB at 28°C, the strain EIB202 and ΔmltA represented similar growth pattern. Both of them reached the maximum cell density (OD540 = 3·0) at 18 h, while the maximum cell density of the mltA+ (OD540 = 2·7) was observed at 24 h, which decreased by c. 10% (Fig. 1) compared with wild-type strain. The results that mltA+ had a minor growth defect indicated that MltA serves for cell growth and separation.

Figure 1.

 Growth curve of Edwardsiella tarda strains. Data are the means for three independent experiments and presented as mean± SEM. Error bars indicate the standard deviation for three triplicate samples. (inline image) ElB202; (inline image) ΔmltA and (inline image) mltA+.

Effect of mltA on carbon starvation, high osmotic pressure resistance, and autolysis induced by EDTA

The survival rates of Edw. tarda strains were determined by plate counts method after the strains were exposed to diverse environmental stresses. When the test strains were incubated in M9 medium supplemented with 0·02% glucose, all the strains displayed a reproducible initial increase in viable counts during the first 24 h of incubation. During the following 5 days of cultivation, the viable counts of ΔmltA exhibited faster decrease than that of EIB202, indicating that MltA had a critical contribution to carbon starvation endurance in Edw. tarda. Nevertheless, overexpression of mltA led to the sharpest decrease in viable counts after the sole carbon source (glucose) was depleted (Fig. 2a).

Figure 2.

 Stress survival assays of Edwardsiella tarda strains. Wild-type strain EIB202 (closed diamond), ΔmltA (closed square) and mltA+ (closed triangle) were exposed to M9 medium (a), 2 mol l−1 NaCl (b), 0·01 mol l−1 EDTA (c). One hundred per cent survival corresponds to the viable counts determined just prior to exposure to the indicated stress. Error bars indicate the standard deviation for three triplicate samples.

For the osmotic challenge, all the test strains showed a rapid decrease during the first 12 h (Fig. 2b). ΔmltA exhibited enhanced sensitivity to the high osmotic culture compared with that of EIB202, indicating that MltA played an essential role in high osmotic pressure resistance. mltA+ displayed much more elevated sensitivity to the high osmotic culture than the wild type and ΔmltA did.

When treated with 0·01 mol l−1 EDTA for 8 h, the survivability of the mltA+ declined by c. 24- and 45-fold compared with that of EIB202 and ΔmltA (Fig. 2c), respectively. These results indicated that MltA contributed to the stress adaptation and autolysis of Edw. tarda.

Effect of mltA on swimming motility and biofilm formation

Swimming motility assay showed that mltA+ exhibited an increased motility. By contrast, the ΔmltA mutant was defective for swimming motility (Fig. 3). These results demonstrated that MltA contributes to the motility of Edw. tarda.

Figure 3.

 Swimming motility in Edwardsiella tarda strains. ΔmltA and mltA+ displayed a decreased and increased swimming motility, respectively.

As shown in Table 4, the mltA+ exhibited increased biofilm formation generating 18·9% more than that of the wild-type strain. On the other hand, the ΔmltA was deficient in biofilm production, producing 10·6% less than that of EIB202, indicating that the biofilm formation might be positively affected by MltA in Edw. tarda.

Table 4.   Characterization of Edwardsiella tarda strain EIB202, ΔmltA and mltA+
  1. *Crystal violet staining quantification (OD570). Data are the means for three independent experiments and presented as mean ± SEM.

  2. **< 0·01 (highly significant) vs corresponding values of Edw. tarda EIB202 (Tukey’s test).

In vitro biofilm formation*0·303 ± 0·0040·274 ± 0·007**0·345 ± 0·003**
LD50 to zebra fish (CFU g−1)(3·39 ± 0·03) × 104(3·62 ± 0·03) × 105(2·63 ± 0·04) × 102**

Effect of mltA on LPS biosynthesis

Analysis of LPS (Fig. 4) extracted from Edw. tarda strains showed that ΔmltA was defective in the production of bands of low-molecular weight, while overexpression of mltA led to an increased generation of other two bands of low-molecular weight, indicating that MltA affected the LPS biosynthesis in Edw. tarda.

Figure 4.

 Silver staining of lipopolysaccharide (LPS) extracted from wild-type strain EIB202 (a), ΔmltA (b) and mltA+ (c). ΔmltA was defective in the production of LPS bands of low-molecular weight, while overexpression of mltA led to an increased generation of other two bands of low-molecular weight.

Virulence of MltA and LPS to zebra fish

MltA has little effect on haemolytic and chondroitinase activities (data not shown); however, it was involved in the virulence of Edw. tarda to zebra fish. The LD50 of EIB202, ΔmltA and mltA+ were 3·39 × 104, 3·62 × 105 and 2·63 × 102 CFU g−1, respectively. The ΔmltA mutant exhibited a tenfold decrease (no significant difference) in virulence compared with the wild-type strain, while the virulence of mltA+ strain increased (P < 0·01) by c. 129-fold compared with that of EIB202 (Table 4).

The fish injected with crude LPSs extracted from the three Edw. tarda strains without dilution were all dead after 2 days. However, when the LPSs were diluted for 100 times, the numbers of dead fish were significantly higher in the mltA+ group (5·0 ± 0·9) than the EIB202 and ΔmltA groups (2·3 ± 0·5 and 2·0 ± 0·9, respectively). The fish in the three negative control groups were all alive. This result indicated that the increased generation of low-molecular weight LPS bands contributes to the pathogenicity of Edw. tarda.


MltA was identified in a variety of bacteria and was found to be involved in numerous cellular processes. To study the role of MltA in Edw. tarda EIB202, an mltA in-frame deletion mutant and a complemented strain were constructed. However, as unexpected, the complemented strain, which was constructed by electrotransformating a low-copy plasmid (derived from pACYC184) carrying the intact mltA gene into ΔmltA mutant could overexpress the mltA in Edw. tarda. Unlike previous studies (Jennings et al. 2002; Stapleton et al. 2007), overexpression of recombinant LTs in E. coli BL21 (DE3) cells using expression vectors, the overexpression Edw. tarda strain we constructed in this study is capable of expressing MltA in excess without the presence of any inducer, such as isopropyl-β-D-thiogalactopyranoside (IPTG). Similar to Staphylococcus aureus (Stapleton et al. 2007), this surprising result might indirectly reflect that the expression level of mltA was controlled tightly by complex and multifactorial regulatory mechanisms from the genome. Further research is needed to find out the detailed regulation mechanisms of mltA in Edw. tarda.

In this study, we demonstrated that inactivation of mltA led to elevated sensitivity to β-lactam antibiotics, which was consistent with the results of previous studies (Heidrich et al. 2002; Korsak et al. 2005). Moreover, the peptidoglycan breakdown product called 1,6-anhydromuropeptides was considered to be a signal for inducing expression of the chromosomally encoded β-lactamase (Jacobs et al. 1994; Park 1995). Therefore, it may be a normal inference that growing bacteria in the presence of appropriate concentrations of β-lactam antibiotics would enhance the expression level of murein hydrolases. As expected, the wild-type strain of Edw. tarda indeed displayed moderately increased amounts of MltA in the presence of ampicillin (10 μg ml−1). However, mltA+ exhibited sharply decreased amounts of MltA when treated with ampicillin compared with that treated without ampicillin, which might be considered as a defence mechanism of bacteria to prevent any damage to the cell. Continuously increased expression level of mltA in mltA+ would certainly lead to extensive autolysis of bacteria induced by ampicillin, which had been indirectly verified by the evidence represented in this study that overexpression of mltA resulted in significantly elevated sensitivity to ampicillin. Therefore, it is necessary for mltA+ itself to reduce the expression level of mltA to help its own survival in the presence of ampicillin.

Peptidoglycan hydrolases were confirmed to be involved in resistance to a variety of environmental stress in E. coli and Staph. aureus (Templin et al. 1999; Stapleton et al. 2007) and contribute to autolysis induced by EDTA and high concentration of salts in Neisseria gonorrhoeae and Staph. aureus (Gilpin et al. 1972; Hebeler and Young 1975). In this study, the ΔmltA mutant exhibited more sensitivity than its wild-type parent to carbon starvation and high osmotic culture, demonstrating its essential roles in stress adaptation. Regarding to the significant decrease in viability caused by overexpression of mltA, there are two possible explanations. (i) Peptidoglycan hydrolases overexpression may play a role in the remodelling of peptidoglycan architecture leading to a low viability under some environmental stresses as showed in previous study (Vijaranakul et al. 1995). (ii) The excessive accumulation of MltA contributes to the autolysis induced by carbon starvation and high osmotic stress.

Biofilm is a complex aggregation of micro-organisms growing on solid surfaces, which was thought to contribute to the virulence of pathogenic bacteria considering that it could be used as a kind of protection mechanism to resist environmental challenges, avoid immune, cellular and chemical systems of host defence (Wakimoto et al. 2004). MltA has been implicated in biofilm formation in Staphylococcus epidermidis and Lactococcus lactis (Heilmann et al. 1997; Mercier et al. 2002). The ΔmltA mutant of Edw. tarda exhibited slightly reduced ability to form biofilms in vitro (Table 1). However, further immunofluorescence histopathology analysis with rabbit anti-Edw. tarda serum as primary antibody showed that the wild-type strain and the mutants of Edw. tarda could not form observable biofilm on the surface of tissues in moribund fish infected by Edw. tarda strains (data not shown). These results indicated that although the in vitro biofilm formation between the wild-type strain and the mutants of Edw. tarda had significant difference, they might not be biologically relevant in vivo. Therefore, the in vitro biofilm formation system may not exactly reflect the in vivo biofilm formation.

Enteropathogenic E. coli was found to display a type III secretion system (T3SS)-dependent, contact-mediated, haemolytic activity requiring a series of secreted proteins (Warawa et al. 1999). Meanwhile, analysis of bacterial genome sequences revealed that many, if not all, T3SS gene clusters encode murein-degrading enzymes, which belong to the lytic tranglycosylase family of proteins (Zahrl et al. 2005). The activity of haemolysin and chondroitinase, two putative virulence factors of Edw. tarda, was also detected. We found that MltA has little effect on haemolysin and chondroitinase activities. It was interesting to find that the mltA+ strain had a minor defection in growth (Fig. 1), while previous study showed that an E. coli mutant with deletion in seven murein hydrolases of different specificities was still able to grow at a normal rate (Heidrich et al. 2002). The results might be explained by assuming that activities of murein synthetases are unable to make up for that of murein hydrolases, which needs further research to confirm.

Although mltA+ only had a minor growth defection, it had a c. 129-fold lower LD50 for zebra fish than wild-type strain. LPS has been considered to be one of the virulence factors in Edw. tarda (Wang et al. 2010), and our results confirmed that the increased generation of two LPS bands with low-molecular weight contributes to the pathogenicity of Edw. tarda to zebra fish. There is another possible explanation as following. MltA in Edw. tarda might be involved in peptidoglycan-derived cytotoxin production as in N. gonorrhoeae (Cloud and Dillard 2002; Cloud-Hansen et al. 2008). The main contributor needs to be confirmed by further research. According to the results of this study, we speculate that although overexpression of mltA help to enhance the pathogenicity of Edw. tarda to zebra fish, the amount of MltA should be controlled by stringent regulatory mechanism, or it will make bacteria more sensitive to various environmental stresses and lose competitiveness for survival.

In conclusion, our results reported here indicate that in addition to its contribution to stress adaption, MltA plays important roles in Edw. tarda in the regulation of LPS biosynthesis. Further studies are necessary to determine the regulators of mltA expression as well as regulatory mechanisms of MltA involved in swimming motility, stress adaption, LPS biosynthesis and virulence in Edw. tarda.


This work was supported by grants from the National High Technology Research Development Program of China (863 Programs, no. 2008AA092501) and the International Science and Technology Cooperation Programme of China (no. 2012DFG31990).