Peptidoglycan lytic activity of the Pseudomonas aeruginosa phage φKZ gp144 lytic transglycosylase

Authors


  • Editor: Craig Winstanley

Correspondence: Roger C. Levesque, Département de Biologie Médicale, Pavillon Charles-Eugène Marchand, Université Laval, Sainte-Foy, Québec, Canada, G1K 7P4. Tel.: +1 418 656 3070; fax: +1 418 656 7176; e-mail: rclevesq@rsvs.ulaval.ca

Abstract

The gp144 endolysin gene from the Pseudomonas aeruginosa phage φKZ was cloned and studies of gp144 expression into Escherichia coli showed host cell lysis. The gp144 protein was purified directly from the culture supernatant and from the bacterial cell pellet and showed in vitro antibacterial lytic activity against P. aeruginosa bacteria and degraded purified peptidoglycan of Gram-negative bacteria. MS analysis identified the gp144 peptidoglycan cleavage site and confirmed a lytic transglycosylase enzyme. Studies of gp144 expression in the presence of sodium azide (NaN3), an inhibitor of the protein export machinery, and into an E. coli MM52 secAts mutant at permissive and restrictive temperatures showed that gp144 was secreted independently of the Sec system. The solution conformation of purified gp144 analyzed by circular dichroism spectroscopy was 61% in α-helical content, and showed a 72% decrease when interacting with dimyristoylphosphatidylglycerol (DMPG), one of the major components of bacterial membranes and less than 10% with dimyristoylphosphatidylcholine (DMPC) found in eukaryotic membranes. Membrane vesicles of DMPG anionic lipids containing calcein indicated that gp144 caused a rapid release of fluorescent calcein when interacting with synthetic membranes. These results indicated that gp144 from φKZ is a lytic transglycosylase capable of interacting with and disorganizing bacterial membranes and has potential as an antipseudomonal in phage therapy.

Introduction

To promote bacterial host cell lysis and release of phage progeny, double-stranded DNA phages typically use a holin–endolysin mechanism. Endolysin is restricted to the cytoplasm until the holin destabilizes the internal membrane, permitting hydrolysis of peptidoglycan and cell lysis (Young et al., 2000; Wang et al., 2000; Fischetti, 2005).

Phage φKZ infects Pseudomonas aeruginosa, a Gram-negative opportunistic pathogen causing nosocomial infections difficult to treat because of antibiotic resistance (Davies, 2002; Hancock & Brinkman, 2002). The Myoviridae φKZ contains the largest sequenced double-stranded DNA phage genome with 306 ORFs (Mesyanzhinov et al., 2002).

During the course of this work, Mesyanzhinov and colleagues described some of the properties of gp144 (Miroshnikov et al., 2006). Recombinant gp144 purified from Escherichia coli effectively degraded chloroform-treated P. aeruginosa cells. The gp144 protein was found in solution in stoichometric monomer: dimer and trimer equilibrium. In this study, we extend these observations and have developed a method for expression, purification and analysis of a biologically active recombinant gp144. We also demonstrate that gp144 is a lytic transglycosylase, and that the enzyme goes through conformational changes when interacting with dimyristoylphosphatidylglycerol (DMPG) calcein vesicles reflecting the net anionic nature of bacterial membranes, which only have 15–30% phosphatidylcholine.

Materials and methods

Bioinformatics

The φKZ 280.3 Kb sequence was analyzed for ORFs using software programs of the University of Wisconsin Package (GCG, version 10.2, Madison, WI). The sequence was split into 312 overlapping fragments and compared with sequences of the viral, bacterial and phage divisions of GenBank using tblastx (Benson et al., 2005). Similar search was done with blastx against the 1137 reported holins. φKZ ORFs were compared with the PFAM-A database using hmmpfam HMMER (Eddy, 1998). The results were analyzed using a developed python program (van Rossum & de Boer, 1991). Retrieval from databases and sequence splitting were performed with emboss (Rice et al., 2000). Gp144 was analyzed for Sec, signal-arrest-release sequence (SAR) and TAT signal sequences and characterized using blast, cdart and expasy algorithms (Berks et al., 2000; Geer et al., 2002; Xu et al., 2004).

Cloning of φKZ ORF144

Phage amplification was performed using phage PhiKZ DNA genomic DNA (Sambrook et al., 1989). DNA was purified with the Lambda Maxi Kit (Qiagen, Mississauga, Ontario, Canada). PCR was used to obtain ORF144 encoding gp144 fused with a C-terminal His-tag using genomic DNA (300 ng), the primers 5′-GTAGAGGTTATCATATGAAAGTATTA-3′ and 5′-TGCTACCTCGAGTTTTCT-3′, 5 mM MgCl2 and 2.5 U of Hot start TAQ polymerase (Qiagen) at touchdown temperatures of 40–30°C. The amplified product was purified, digested with NdeI and Xho1 and ligated into corresponding sites of pET24a (Novagen, Madison, WI). Plasmid pMON2266 was electroporated into E. coli recA ElectroMAX™ DH10B™ and sequenced using the T7 primers (Novagen).

Overexpression and purification of gp144

Plasmid pMON2266 was transformed into E. coli BL21 (λDE3) using CaCl2 (Sambrook et al., 1989). Isopropyl-β-d-thiogalactopyranoside (IPTG) at 1 mM was added to a tryptic soy broth culture (kanamycin 50 μg mL−1) at an OD600 nm of 0.8 and incubated for 6 h at 37°C under agitation. The supernatant from a 1-L culture was treated with DNAse and concentrated to 200 mL using a stirred ultrafiltration cell and a filter with a 10 kDa cut-off. Expressing cells were resuspended in sonication buffer [50 mM Tris-HCl pH 8.6, 1 mM ethylenediaminetetraaectic acid (EDTA)], sonicated 30 s mL−1 before addition of protease inhibitors. Fractions were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Gp144 was purified from the supernatant and the cell pellet by affinity chromatography using a HisTrap™ HP nickel column (Amersham Biosciences, Baie d'Urfé, Québec, Canada). Final purification was performed by gel filtration (Superdex 75 10/300 column, LMW gel filtration Calibration Kit, Amersham Biosciences). Fractions containing gp144 were concentrated, maintained at −80°C in 50 mM Tris-HCl pH 7.4, 25 mM MgCl2 and 10 mM NaCl and N-terminal sequencing was performed (Paradis-Bleau et al., 2005).

Peptidoglycan degrading assay

Peptidoglycan from P. aeruginosa PAO1, E. coli ATCC 25922, Staphylococcus aureus ATCC 25923 and Bacillus cereus ATCC 27877 was prepared from 50-mL cultures of Gram-positive bacteria and 500 mL for Gram-negatives (Moak & Molineux, 2004). Peptidoglycan was sonicated, diluted in water to an OD600 nm of 1, monitored for 250 min with and without purified gp144 at various concentrations and performed in triplicate.

Identification of the gp144 peptidoglycan cleavage site by LCMS

The molecular mass of the gp144 peptidoglycan degradation products was determined using an Agilent 1100 LCMS. A 200 μg aliquot of peptidoglycan was incubated with 250 μM of purified gp144 at 37°C in 100 μL. Peptidoglycan hydrolysis was stopped by adding 10% trifluoroacetic acid (TFA) and TFA-soluble muropeptides were collected (Navarre et al., 1999). Dried muropeptides were resuspended in 70 μL of mobile phase A (H2O, 0.1% formic acid, 0.025% TFA) and 10 μL was injected into a ZORBAX 300 SB-C18 reversed-phase column protected by a Zorbax 300 SB-C18 analytical guard-column (Agilent Technologies, Montreal, Québec, Canada). Separation of muropeptides was performed at a flow rate of 200 μL min−1. A gradient of mobile phase B (acetonitrile, 0.1% formic acid, 0.025% TFA) was used from 1% increasing 1.36% min−1. The eluted muropeptides were identified at 214 nm at 4°C and the column was at 40°C. The molecular mass of muropeptides was determined with an Agilent 1100 MSD ESI mass spectrometer using the positive electrospray ionization mode, in full scan from 50 to 2950 m/z with a step of 0.1 and a cycle time of 2.25 s.

Expression of gp144 in E. coli with sodium azide (NaN3)

Gp144 was expressed at an OD600 nm of 0.3, 0.6, 0.7, 0.8 and 0.9 with and without 1 mM NaN3. Escherichia coli BL21-induced and noninduced cultures for gp144 were monitored. CFUs were determined in triplicate at time points 0, 3, 6, 9, 12 and 25 h after induction at an OD600 nm of 0.8. Cell pellets and supernatants from 0, 1, 2, 3, 4, 6, 8, 10 and 12 h were analyzed by SDS-PAGE. Western blots of gp144 supernatants were performed using rabbit polyclonal anti-gp144 antibodies (Harlow & Lane, 1988). Immunodetection was performed with the ECL Advance Western Blotting Detection Kit using the rabbit anti-gp144 serum 1 : 50 000 and the ECL anti-rabbit peroxidase-linked antibody 1 : 15 000 (Amersham). Escherichia coli XL1-blue was lysogenized using the λDE3 Lysogenization Kit (Novagen) and transformed with pMON2266. Expression of gp144 was determined at OD600 nm values of 0.6, 0.7 and 0.8 and was monitored with and without NaN3.

Expression of gp144 in an E. coli secAts mutant

Escherichia coli MM52 (pMON2266) was grown at 30°C for SecA expression and at the nonpermissive (42°C) temperature with induction of gp144 at an OD600 nm of 0.05 to 0.7 (Oliver & Beckwith, 1981). Cell extracts and supernatants were analyzed by SDS-PAGE and gp144 in the supernatants was confirmed by Western blot. As control, cultures containing 0.4% maltose to induce the SecA-translocated maltose-binding protein (MBP) were used. At the time of induction (after 4 h at a permissive or nonpermissive temperature), 30 mL of each culture was treated with lysozyme-EDTA (Oliver & Beckwith, 1981). Spheroplasts were resuspended in 200 μL of sonication buffer and sonicated for 30 s. OD600 nm values were used to standardize periplasmic and cytoplasmic proteins analyzed by SDS-PAGE. The MBP content was evaluated by Western blot using the anti-MBP monoclonal antibody conjugated to HRP 1 : 500 (New England Biolabs, Mississauga, Ontario, Canada) and the Immobilon™ Western Chemiluminescent HRP Substrate detection assay (Millipore, Nepean, Ontario, Canada).

Circular dichroism

Fifteen scans were collected from 190 to 250 nm using a 1-nm wavelength increment on a Jasco J-710 spectropolarimeter with a 2-mm path length cell. Spectra were obtained by comparisons with the base line of a CD buffer (2.5 mM Tris-HCl pH 7.4, 1.25 mM MgCl2, 0.5 mM NaCl). Ellipticity was measured at 225 nm. Spectra were collected using 30 μM of gp144 and with concentrations of 1–25 mM of dithiothreithol (DTT). For interactions with synthetic membranes, 166 μM of gp144 was used with zwitterionic dimyristoylphosphatidylcholine (DMPC) or anionic DMPG lipid vesicles. The data were analyzed using the cdestima software to predict the percentage of α-helix content.

Gp144 permeabilization assays of calcein-containing vesicles

For calcein release assays, vesicles were prepared by dispersing 5 mg of DMPC or DMPG dried lipids into 250 μL of phosphate buffer (50 mM pH 7.4). After five cycles of freeze in liquid nitrogen, thaw at 37°C and mixing, the dispersions were extruded 11 times using a polycarbonate 0.1 μm filter. The stability of calcein-containing vesicles was confirmed by monitoring the calcein release for each preparation.

DMPC and DMPG calcein vesicles and gp144 permeabilization assays were performed as described (Biron et al., 2004) with the following modifications: 20 mg of DMPC or DMPG was directly dispersed in 1 mL of internal buffer and five cycles of freeze at −180°C, thaw at 37°C and mixing were performed before sonication. A final concentration of 4 μM of gp144 was added to 3 mL of external buffer at room temperature to yield a 100 : 1 lipid gp−1144 ratio.

Results

Bioinformatic analysis for an endolysin–holin system in the φKZ genome sequence

Bioinformatic analysis of the 280 334 bps sequence of φKZ and the 306 ORFs identified ORF144 as a putative peptidoglycan-degrading enzyme. We noted 52% amino acid identity between the catalytic domain of gp144 and the C-terminus of gp181, a φKZ tail fiber protein. The deduced amino acid sequence of the φKZ endolysin (gp144) indicated a putative cytoplasmic protein of 28.8 kDa with a pI value of 9.1 having no transmembrane domains, and secondary structures containing 51–54% of α-helix, 1–3% of β-sheet and 44–46% of loop. CDART identified a putative peptidoglycan-binding domain composed of three α-helices at the N-terminus and a lytic transglycosylase catalytic domain at the C-terminus. Exhaustive bioinformatic analysis did not identify a typical holin sequence.

Expression and purification of gp144 into E. coli

Analysis of the gp144 expression into E. coli was studied using bacterial growth curves. The gp144-expressing culture contained 1.8 × 107 CFUs mL−1 after 5 h in comparison with 4.4 × 109 CFUs mL−1 for the noninduced culture (data not shown). Significant differences between induced and non induced cultures for expression of gp144 were c. 3 logs in CFUs after 9 h. The E. coli colonies containing cells expressing gp144 appeared irregular, rough and smaller than nonexpressing colonies on Tryptic Soy Agar plates.

Analysis by SDS-PAGE indicated that the 30.1 kDa band (28.8 kDa plus a C-terminus fusion of LDLEHHHHHH) identified as gp144 was released directly into the culture supernatant extracellular media (300 mg L−1) and was also present in the cell pellet as soluble (100 mg L−1) and insoluble fractions (data not shown). Based upon SDS-PAGE analysis (1 μg lane−1) and SYPRO® Orange staining, gp144 was purified to >99% homogeneity and N-terminal sequencing of the 15 first amino acids confirmed gp144 identity. As previously reported, purification by gel filtration gave two major peaks corresponding to the monomer (30.1 kDa) and dimer (60.2 kDa) forms of gp144 (data not shown; Miroshnikov et al., 2006). The dimer form was predominant in the culture supernatant, whereas the monomer was predominant in the cytoplasm of the cell pellet fraction (data not shown). SDS-PAGE analysis showed that purified gp144 was present as monomers, dimers and trimers that were resistant to boiling and to 1% SDS; nondenaturing PAGE showed gp144 oligomers of high molecular mass (data not shown). The monomer form was found to dimerize rapidly after 24 h at 4°C, whereas DTT-treated gp144 gave only monomers.

Peptidoglycan-degrading activities

Addition of purified gp144 to purified peptidoglycan from four bacterial species caused a rapid OD600 nm decrease of peptidoglycan from Gram-negative bacteria in a dose-dependent fashion. As presented in Fig. 1a, 100 μM of gp144 hydrolyzed P. aeruginosa peptidoglycan in 2 min and in c. 25 min for E. coli. In contrast, an apparent lag of more than 60 min was observed for peptidoglycan of B. cereus and S. aureus and significant hydrolysis was detected only after 120 min. These results indicated that gp144 caused specific hydrolysis of peptidoglycan from Gram-negative bacteria.

Figure 1.

 Gp144 peptidoglycan-degrading activity. (a) Spectrophotometric curves showing hydrolysis of purified peptidoglycan from Escherichia coli, Pseudomonas aeruginosa, Staphylococcus aureus and Bacillus subtilis by gp144 visualized as a decrease of OD600 nm absorption. (b) Determination of the gp144 peptidoglycan cleavage site by LCMS. HPLC chromatogram of the gp144-digested peptidoglycan products from P. aeruginosa. (c) Schematic representation of the LCMS parent ion characteristics of lytic transglycosylase.

Identification of gp144 cleavage site in peptidoglycan

The HPLC chromatogram of peptidoglycan from P. aeruginosa hydrolyzed by gp144 gave major peptidoglycan degradation products, and the results are summarized in Table 1. The first peak identified contained gp144 amino acid autodegradation products with molecular masses of 119.11 and 370.23 Da because the same elution products were obtained for gp144 only (data not shown). By using the peak surface area, we selected seven major elution products as shown in Fig. 1b. The main peptidoglycan degradation product GlcNac-anhydroMurNac–Ala–iGln–mDAP–Ala–mDAP–iGln–Ala–Ala was defined as the parent ion as all muropeptides identified by MS converged to this compound (Table 1, Fig. 1b). The unique cell wall-degrading enzyme that could lead to this activity is a lytic transglycosylase. This enzyme cleaves the β(1,4)-glycosidic bond between N-acetylmuramic acid and N-acetylglucosamine and adds a new glycosidic bond within the C6 hydroxyl group of the same muramic acid residue. The lytic transglycosylase reaction gave distinctive (1,6)-anhydro N-acetylmuramic muropeptides frequently recovered in the gp144 digestion pattern, and the identified compounds are indicated in bold in Table 1. The molecular masses obtained from gp144 hydrolysis did not correspond to the size of degradation products for other types of peptidoglycan degrading enzymes such as N-acetylglucosaminidase, amidase or peptidase.

Table 1.   Molecular mass of the gp144-digested P. aeruginosa cell peptidoglycan obtained by LCMS and corresponding muropeptide structures
m z−1 observedMH+ predictedProposed structure
  1. The major cell wall degradation products considered for each peak is indicated by an asterisk, the parent ion is identified by two asterisks and the lytic transglycosylase characteristics (1,6)-anhydro N-acetylmuramic muropeptides are indicated in bold.

  2. N.D., not defined.

Peak 1
 119.11N.D.gp144 degradation product
 370.23N.D.gp144 degradation product
Peak 2
 629.33629.5*anhydroMurNac–Ala–iGln–mDAP (−OH)
 415.2415.46mDAP–Ala-mDAP (–H2O)
 186.1186.2Ala-iGln (−CH2–OH)
 424.15424.43Ala-iGln-mDAP (+OH)
Peak 3
 347.17347.19*anhydroMurNac–Ala (+H)
 187.09187.2Ala-iGln (−CH–OH)
Peak 4
 560.26560.6*iGln-mDAP–Ala-mDAP (−H)
 390.24390.43Ala-iGln-mDAP (+H)
Peak 5
 509.27508.96half mass
 1018.471017.92*anhydroMurNac–Ala–iGln–mDAP–Ala–mDAP–iGln
Peak 6
 557.3556.59Ala–iGln–mDAP–Ala–Ala (+2H+Na)
 786.36786.65anhydroMurNac–Ala–iGln–mDAP–Ala–Ala (-2H)
 1114.471114.84*GlcNac–O–anhydroMurNac–Ala–iGln–mDAP–Ala–mDAP (−H+Na)
Peak 7
 672.33671.74iGln–mDAP–Ala–mDAP–iGln (−H2O)
 694.31695.41GlcNac-anhydroMurNac-Ala–iGln (+H2O)
 887.5887.78anhydroMurNac–Ala–iGln–mDAP–Ala–mDAP (−2H)
 1344.551345.15**GlcNac-anhydroMurNac–Ala–iGln–mDAP–Ala–mDAP–iGln–Ala–Ala (−H2O)

Expression of gp144 in the presence of an inhibitor of the protein export machinery

As in silico bioinformatics analysis for Sec, SAR and TAT signal sequences using blast, cdart and expasy algorithms was negative for gp144 and as the cytoplasmic gp144 was secreted directly into the culture supernatant, we evaluated whether typical bacterial secretion systems could still potentially be involved. It is well known that bacterial cell cultures treated with NaN3 show inhibition of azide-sensitive components of one of the major bacterial protein export machineries such as Sec (Fortin et al., 1990). To evaluate the role of the Sec system in gp144 translocation, growth rates of E. coli were measured in liquid media induced for the expression of gp144 with or without 1 mM NaN3. As depicted in Fig. 2a, the results obtained showed that NaN3 did not affect gp144 hydrolytic activity. We noted a difference of 3 logs in CFUs from E. coli cultures expressing gp144 containing NaN3. SDS-PAGE analysis at selected time-points of the cell culture supernatants and cell pellets treated or not with NaN3 (Fig. 2b) showed that the concentration of gp144 with NaN3 was similar to the control without NaN3 after 1 h of expression (Fig. 2b, upper panels, left and right). Specific anti-gp144 immunodetection revealed gp144 in culture supernatants apparent after 1 h and up to 12 h after induction. Gp144 was also found in cultures grown with NaN3 and could be detected from 1 h up to 12 h (Fig 2b, lower panels, left and right). These data indicate that gp144 transport does not depend on the typical bacterial transport machinery.

Figure 2.

 Expression of gp144 in Escherichia coli with NaN3. (a) Bacterial growth curves of E. coli BL21 cultures noninduced (□) and induced for gp144-expression (▪) without NaN3; noninduced (▵) and induced for gp144-expression (▴) with NaN3 added at the time of induction with IPTG. (b) Time-course expression of gp144 with and without NaN3; SDS-PAGE showing gp144 in the cell pellet and from the culture supernatants by specific detection in Western Blots using an anti-gp144 polyclonal antibody. Total cell protein content was prepared by adding 50 μL of SDS-PAGE sample buffer and boiling for 5 min. CFUs were determined to standardize the cell numbers used and for estimation of the protein concentration. Fifteen microliters of culture supernatant was analyzed directly into gels.

Expression of gp144 in E. coli secAts defective mutant

To confirm the NaN3 data, we studied expression of the gp144 lytic transglycosylase into an E. coli secAts defective mutant. As a control, we first tested the SecA-dependent MBP translocation at both the permissive and restrictive temperatures. Immunodetection of the SecA-translocated MBP at the time of induction indicated that the 43.4 kDa protein accumulated in the periplasm at 30°C, while only a trace of MBP was detected in the periplasm at 42°C, confirming the SecAts phenotype (data not shown). To confirm gp144 activity at the restrictive temperature, we analyzed expression of gp144 in E. coli BL21 and E. coli XL1-blue grown at 30°C and at 42°C. Growth curve analysis demonstrated that gp144 was functional at both temperatures because OD600 nm values declined from 0.85 to 0.4, confirming gp144 activity and thermostability at 42°C (data not shown).

As shown in Fig. 3a, induction of gp144 at 30°C caused cell lysis and reduced CFUs at c. 0.5 log10 when compared with the noninduced control but not as significantly as at 37°C. This indicated less nonspecific cell lysis caused by the release of gp144 at lower temperature. At the nonpermissive temperature (42°C), E. coli cells showed a delay in growth for 6.5 h until the exponential phase was reached (Fig. 3a). Induction of gp144 caused cell lysis at 42°C and gave growth curves similar to gp144 at normal temperature (37°C). Gp144 was found in the cell pellet (Fig. 3b, upper left panel) after 1 h of expression at 30°C, accumulated until 4 h and decreased after 10 h. Gp144 was constantly present after induction at 42°C from 1 h up to 12 h, but at a much lower concentration (Fig. 3b, upper right panel). Specific anti-gp144 immunodetection confirmed that gp144 was released in the culture supernatant after 4 h of expression at 30°C, indicating a tight control of expression (Fig. 3b, lower left panel). In contrast, gp144 was found in the culture supernatant 1 h and up to 12 h after expression at 42°C (Fig. 3b, lower right panel), indicating that SecA was not necessary for secretion.

Figure 3.

 Expression of gp144 in an Escherichia coli secAts mutant. (a) E. coli MM52 growth curves showing noninduced (□) and induced (▪) cultures at a permissive temperature (30°C) and noninduced (▵) and induced (▴) cultures and at a nonpermissive temperature (42°C) using a SecAts phenotype. An overnight E. coli MM52 (λDE3) culture incubated at 30°C, supplemented with 50 μg mL−1 kanamycin, was used to inoculate two 500 mL Tryptic Soy Broth cultures. Cultures were adjusted to OD600 nm of 0.015 and 0.15 and incubated for 1 h at 30°C. The 0.15 OD600 nm starting culture was switched to 42°C for 4 h while the 0.015 OD600 nm starting culture remained at 30°C. Expression of gp144 was induced with 1 mM of IPTG 4 h after the temperature shift at an OD600 nm of 0.5. (b) Time-course expression of gp144 at permissive and nonpermissive temperatures; SDS-PAGE showing gp144 in the cell pellets and from culture supernatants. Specific detection of gp144 in culture supernatants was performed by Western Blots using an anti-gp144 polyclonal antibody.

Analysis of gp144 by circular dichroism and interaction with lipid vesicles

Analysis of purified gp144 by circular dichroism showed a content of 61±5% in the α-helix structure stable up to 50°C. To evaluate whether gp144 could interact directly with bacterial membranes and be liberated into the extracellular media, we used a biophysical approach with DMPG-containing vesicles. As shown in Fig. 4, gp144 was added to preparations of anionic DMPG vesicles and subsequent analysis of the CD spectra gave a decrease of 72±3% in α-helical content. In contrast, a small increase of gp144 α-helical content was observed with zwitterionic DMPC vesicles, a compound mimicking eukaryotic membranes. These studies indicated that gp144 was capable of interacting directly with anionic membranes. This indicated the potential of gp144 to interact with bacterial membranes that would facilitate its liberation.

Figure 4.

 Structural changes of gp144 interacting with anionic lipids. Circular dichroism spectra showing gp144 secondary structure alone and with DMPC or DMPG small unilamellar lipid vesicles. 50 μL of each vesicle preparation was combined to 166 μM of gp144 to obtain a 100 : 1 lipid gp 144−1 ratio before collection of the CD spectra at room temperature.

To extend these experiments, we decided to incubate gp144 with calcein-containing DMPG vesicles and measure indirectly the permeabilization of these artificial membranes by gp144 by measuring the fluorescence of calcein. As shown in Fig. 5, addition of gp144 to DMPG vesicles caused a rapid calcein release, reaching 55% after 50 s and reaching a plateau at 90%. Control experiments showed that calcein-containing vesicles were stable, that bovine serum albumin (BSA) did not enhance calcein release and that Triton released 100% of the fluorescent calcein. In contrast, gp144 did not release fluorescent calcein from DMPC vesicles (Fig. 5). These data would indicate that gp144 interacted with, disorganized and augmented the permeability of these artificial membranes.

Figure 5.

 Permeabilization and release of fluorescence from DMPG calcein-containing vesicles. Vesicles were prepared (as described in materials and methods) and were analyzed by spectrophotometry. Spectrophotometric curves indicate specific release of fluorescent calcein from anionic DMPG vesicles when interacting with the gp144 lytic transglycosylase.

Discussion

Phage lysis proteins are exciting prospects as antibacterial agents (Fischetti, 2005). Overexpression of the φKZ gp144 endolysin into E. coli retained peptidoglycan lytic activity against Gram-negative bacteria. Gp144 was released into culture supernatants after IPTG induction and lysed neighboring cells.

Gp144 peptidoglycan-degrading activity indicated a predominant activity against Gram-negative bacteria, and LCMS analysis of gp144 hydrolytic products identified a lytic transglycosylase. We noted a weak hydrolytic activity against the peptidoglycan of Gram-positive bacteria that O-acetylate their peptidoglycan, a modification that totally precludes the function of lytic transglycosylases because the O-acetylation occurs at the C-6 hydroxyl of muramyl residues, the same group involved in the formation of the 1,6-anhydromuramyl product of lytic transglycosylase activity (Ginsburg, 2002). The O-acetyl groups on the tested peptidoglycan may have been lost during its purification (a strong possibility, given the liability of the acetate ester linkage), which could explain residual activity against S. aureus and B. cereus. An apparent lag exists in hydrolysis of peptidoglycans from Gram-positive bacteria, which could reflect the prior need for spontaneous de-esterification of acetyl groups (if still present on the substrates) and the poor recognition of the peptidoglycan-binding domain of gp144 of Gram-positive bacteria.

The vast majority of secreted proteins use the Sec or Tat pathways requiring specific N-terminal signal sequences (Berks et al., 2000). Gp144 does not possess typical Sec or Tat secretion motifs. We assessed whether gp144 exploits the Sec pathway via a nontypical signal as for phage P1 endolysin (Sao-Jose et al., 2000). Inactivation of SecA blocked translocation and lytic activities of phages P1 and fOg44 endolysins (Sao-Jose et al., 2000; Xu et al., 2004). Gp144 activity and cell lysis was evident in a SecAts mutant at a non permissive temperature. We also studied the effect of NaN3 on gp144 expression and transport; NaN3 did not affect cell lysis by gp144, suggesting a bypass of typical secretion pathways (Fortin et al., 1990). However, expression of gp144 from a multicopy plasmid and subsequent cell lysis could be caused by levels of expression toxic to the host cell where gp144 is liberated and lytic transglycosylase activity may lyse neighboring cells.

LytA autolysin and Eji phage endolysin activation depends upon dimerization (Diaz et al., 1989; Saiz et al., 2002). Gp144 has a secondary structure similar to the Eji endolysin (Saiz et al., 2002). Recently, a SAR mediated export and control of phage P1 endolysin was identified (Xu et al., 2004). Lyz of P1 caused host lysis without a holin. Instead, export was mediated by an N-terminal transmembrane domain and required host sec function. Exported Lyz of identical SDS-PAGE mobility was found in both the membrane and periplasmic compartments, indicating that periplasmic Lyz was not generated by the proteolytic cleavage of the membrane-associated form. The N-terminal domain of Lyz is both necessary and sufficient not only for export to the membrane but also for release into the periplasm. The unusual N-terminal domain, rich in residues weakly hydrophobic, functions as an SAR sequence, acting as a normal signal-arrest domain to direct the endolysin to the periplasm in membrane-tethered form and then allows its release in the periplasm.

Disulfide isomerization after membrane release of its SAR domain also activates the P1 lysozyme (Xu et al., 2005). Crystal structures confirmed the alternative disulfide linkages in the two forms of Lyz and revealed dramatic conformational differences in the catalytic domain. Thus, the exported P1 endolysin is kept inactive by three levels of control-topological, conformational and covalent-until its release from the membrane is triggered by the P1 holin.

Examination of the protein sequences of related bacteriophage endolysins suggests that the presence of an N-terminal SAR sequence is not unique to Lyz. Analysis of the gp144 N-terminal region revealed a weakly positive charged region, followed by a hydrophobic α-helical region. This region may serve as a leader sequence and define a unique translocation process. Experiments are in progress to engineer a truncated derivative and confirm the function of this region as a secretion signal peptide for gp144.

Acknowledgements

We thank Dr Sylvain Moineau and Denise Tremblay from the Banque de phages Félix d'Hérelle, Université Laval, Québec, Canada, for providing φKZ. This work was funded by the CREFSIP, by an FQRNT Infrastructure Team grant to R.C. Levesque and by an FRSQ studentship to Catherine Paradis-Bleau.

Ancillary