Peptidoglycan degradation by specialized lytic transglycosylases associated with type III and type IV secretion systems

Type III secretion systems (T3SS) and type IV secretion systems (T4SS) of Gram-negative bacteria are composed of supramolecular protein complexes which span the cell envelope and secrete proteins and/or DNA in a contact-dependent manner (for recent reviews on T3SS see Ghosh, 2004; Tampakaki et al., 2004; for a recent review on T4SS see Cascales & Christie, 2003). Due to structural constraints imposed by the peptidoglycan (PG) meshwork and its sieving capacity (Demchick & Koch, 1996; Pink et al., 2000; Vázquez-Laslop et al., 2001; Yao et al., 1999), it has been proposed that for the assembly of a macromolecular secretion system specialized enzymes are required to enlarge holes in the PG by controlled local degradation (Dijkstra & Keck, 1996; Koraimann, 2003). Analyses of bacterial genome sequences revealed that many, if not all, T3SS and T4SS gene clusters encode their own murein-degrading enzymes, which belong to the lytic transglycosylase (LT) family of proteins. These small proteins (usually between 150 and 250 aa residues) have been termed specialized LTs because of their defined role in macromolecular transport (Koraimann, 2003) and contain the characteristic SLT sequence motif (see the PFAM database, accession code PF01464; ). Originally LTs were characterized as enzymes involved in growth and division of the stress-bearing murein sacculus (Höltje & Heidrich, 2001). LTs are widely distributed among Gram-negative bacteria and cleave the β-1,4 glycosidic bond between N-acetylmuramic acid (MurNAc) and N-acetylglucosamine (GlcNAc). Despite their observed redundancy, LTs have been proposed to represent potential targets for novel antibacterial agents (Korsak et al., 2005). Known inhibitors of LTs are hexa-N-acetylchitohexaose (Leung et al., 2001; Song et al., 1994) and bulgecin A (Templin et al., 1992); very recently a novel LT inhibitor, N-acetylglucosamine thiazoline, has been identified (Reid et al., 2004a, b).

In contrast to structurally and functionally related lysozymes, the reaction catalysed by LTs is not a hydrolysis but rather a transglycosylase reaction which results in formation of an internal 1,6 anhydro bond in the MurNAc residue (Höltje et al., 1975). Experimental data strongly suggest that the cleavage reaction is triggered by a conserved glutamyl residue in the active site of LTs (van Asselt et al., 2000). A natural breakdown product produced by LTs, the muropeptide GlcNAc-(1,6 anhydro)MurNAc-tetrapeptide has been shown to induce inflammatory cytokines in human monocytes (Dokter et al., 1994). An intracellular receptor and component of the human innate immune system, Nod1, has been shown to specifically recognize Gram-negative PG and PG breakdown products (Girardin et al., 2003). Recognition of PG and PG breakdown products from Gram-negative and Gram-positive bacteria has been acknowledged as a key part of the innate immune system present in diverse organisms (Girardin & Philpott, 2004), yet the source and structural composition of PG breakdown products that elicit a defence reaction in the host of a pathogen are unknown. A potential link between the innate immune system recognizing PG from Gram-negative bacteria and the activity of specialized LTs could be the finding that the delivery of PG to epithelial cells by the human pathogen Helicobacter pylori is mediated by the T4SS, which is an essential virulence determinant of this organism (Viala et al., 2004).

Here we focus on biochemical characterization of specialized LTs from diverse T3SS and T4SS. We cloned and overexpressed specialized LTs from both types of secretion systems. From T3SS: IpgF from Shigella sonnei, which is 98 % identical to IpgF, encoded within the entry region of the virulence plasmids pWR100 and pCP301 of Shigella flexneri serotypes 5a and 2a, respectively (Allaoui et al., 1993; Buchrieser et al., 2000; Jin et al., 2002); and IagB, which is encoded by the Salmonella enterica pathogenicity island SPI-1 (Miras et al., 1995). From T4SS: P19 encoded by the conjugative resistance plasmid R1 (Koraimann et al., 1993), VirB1 encoded by the Ti plasmid of Agrobacterium tumefaciens (Thompson et al., 1988), VirB1 encoded by the small chromosome of Brucella suis (O'Callaghan et al., 1999), TrbN from the IncP plasmid RP4 (Pansegrau et al., 1994) and HP0523 encoded by the cag pathogenicity island (cag-PAI) of H. pylori (Tomb et al., 1997). Inactivation of the genes encoding these proteins in the respective system mostly resulted in virulence-attenuated phenotypes which could have arisen from the observed redundancy of LTs in bacteria (Koraimann, 2003). Whereas mutations in ipgF and iagB were reported to have no obvious effects on pathogenicity or needle complex formation (Allaoui et al., 1993; Sukhan et al., 2001), a deletion of the corresponding LT gene in the PAI of Citrobacter rodentium, rorf3, led to decreased T3SS assembly, effector translocation and virulence (Deng et al., 2004). Mutations inactivating P19 of the derepressed F-like plasmid R1-16 reduced the conjugation frequency five- to tenfold (Bayer et al., 1995, 2001). Mutations inactivating VirB1 encoded by the Ti plasmid of A. tumefaciens led to attenuated tumour formation in plants (Mushegian et al., 1996). Notably, HP0523, the cag-PAI-encoded LT from H. pylori was found to be essential for both CagA translocation into the human host cells and IL-8 induction (Fischer et al., 2001).

The main aim of the work presented here was to unambiguously demonstrate the muralytic activity of these enzymes, which up to now has not been shown for any of the specialized LTs encoded by T3SS or T4SS. This goal was reached by combined in vivo studies, zymogram analyses and in-solution muramidase assays, including a novel test system employing Cy3-labelled PG.

Bacterial strains, plasmids and growth conditions.
The bacterial strains and plasmids used in this work are listed in Table 1. Escherichia coli cells were grown in 2x TY (16 g tryptone l¹, 10 g yeast extract l¹ and 5 g NaCl l¹). The following antibiotics, when appropriate, were added at the indicated final concentrations: 100 µg ampicillin ml¹, 40 µg kanamycin ml¹.

Table 1. Bacterial strains and plasmids

Enzymes, chemicals and oligonucleotides.
Restriction endonucleases and other enzymes used for standard cloning and DNA modification procedures were purchased from TaKaRa Biomedicals, Roche Molecular Biochemicals or New England Biolabs. Oligonucleotides were purchased from MWG Biotech. Rabbit anti-MalE antibody was purchased from Merck; goat anti-rabbit and anti-GroEL antibodies were purchased from Sigma-Aldrich. For DNA manipulations, PG isolation and purification, zymogram analyses, Cy3 labelling of PG, and the Cy3 PG spot assay RNase-free, double-distilled water (Fresenius Kabi Austria) was used.

DNA manipulations and recombinant plasmids.
Recombinant DNA techniques were performed as described by Sambrook et al. (1989) or Ausubel et al. (1987) or according to the manufacturer's protocols. DNA fragments amplified by PCR using the primers given below were cloned into pMAL-p2X (New England Biolabs) for periplasmic and into pMAL-c2X (New England Biolabs) for cytoplasmic expression of fusion proteins; the inserted sequences were verified by DNA sequencing in each case. Signal sequences, when present, were removed during the cloning procedure. All constructs and their relevant features are listed in Table 1. For gene 19 amplification the forward primer BamP19_T20 (5'-ACGCGGATCCACTGATTGCTTTGATCTTGC-3') was used. Reverse primers were P19end_Hind (5'-ACCCAAGCTTTTAATTATTCTGCAGCGTG-3'), 19Hind150G (5'-CATTAAGCTTTTATTAACCTTTACTGCTTTTTATCCC-3'), 19Hind144G (5'-CCTAAAGCTTTATTACCCGGTATAAATCC-3') and 19Hind140R (5'-CATTAAGCTTTTATTACCGGTAAACCTCTG-3'). Primers for site-specific PCR mutagenesis for the creation of the P19_E44A mutant were P19E44A_1 (5'-CATGGAAAGCATCCCGTTAC -3') and P19E44A_2 (5'-GTAACGGGATGCTTTCCATG-3'), which were used in combination with the primers BamP19_T20 and P19end_Hind. Plasmids pCK217 (Koraimann et al., 1993) and pETBla19E44QHis+ (Bayer et al., 2001) were used as templates for the PCR amplifications. hp0523 was amplified from a DNA clone containing the cag-PAI of H. pylori using primers Bam_HP0523 (5'-CTTAGCCGGATTCTGCCAAAAGGTAACCATTAGTTTC-3') and HP0523_end_Hind (5'-GGGAAAGCTTACTACTCGTTATATCGCACTTGAG-3'). All gene 19 fragments and the hp0523 fragment were digested with BamHI and HindIII and ligated into the pMAL vectors. Oligonucleotides used to amplify ipgF from a clinical isolate of Sh. sonnei were IpgF-EcoRI (5'-GCTTGAATTCGATTGTTGGGATAAGGCTGG-3') and IpgF_HindIII (5'-GAGCAAGCTTTTATTATATCCTTCGATTATTCTGCTTGCTC-3'). Site-specific mutagenesis resulting in the IpgF_E42Q mutant was done with the mutagenic primers IpgF_EQ_fw (5'-GCGATTGCGGAAAAACAGTCCGGATTTA-3') and IpgF_EQ_rev (5'-ATTAAATCCGGACTGTTTTTCCGCAATCG-3') in combination with the two ipgF-specific primers above. The iagB gene was amplified from Sal. enterica using primers IagB_EcoRI (5'-GATTGAATTCGATTGCTGGCTTCAGGC-3') and IagB_HindIII (5'-CCGCAAGCTTTTATTATTTGTTTACCGCGATAGAAAGTC-3'). The B. suis virB1 gene was amplified using primers Bruc_EcoRI (5'-GTACGAATTCGTGCCATTCCTTGTCCTCGC-3') and Bruc_HindIII (5'-CAGTAAGCTTTTATTAGAAAACAACTACGCCGTC-3'). ipgF, iagB and virB1 fragments were digested with EcoRI and HindIII and ligated into the pMAL vectors. Primers for the amplification of virB1 from A. tumefaciens were VirB1_BamHI (5'-CTAAGGATCCGAGTTCGACCATGTTGCTCG-3') and VirB1_NotIa (5'-GATTGCGGCCGCTTATTAGTATAAGTCGAATAAGAC-3'); finally, trbN primers were TrbN_BamHI (5'-CTAAGGATCCCCCGATCTGTCGGAACAGATGG-3') and TrbN_NotI (5'-CTAAGCGGCCGCTCCTTATTATGGCGTGTTGTTG-3'). The amplified fragments were digested with BamHI and NotI, cloned into the vector pET-32a(+) (Novagen), and subsequently BamHIXhoI fragments were ligated into the BamHI and SalI restriction sites of the pMAL vectors.

DNA and protein sequence analyses.
The Wisconsin Package Version 10.3 (Accelrys) was used for the analysis and in silico manipulations of DNA and protein sequences. The calculation of similarity scores was performed using the BESTFIT program with the BLOSUM 62 matrix (gap weight 6; length weight 2, gaps longer than 5 not penalized).

Expression and affinity chromatography purification of fusion proteins.
E. coli SF100 cells (ompT) harbouring recombinant plasmids for cytoplasmic expression were cultivated at 30 °C to an OD₆₀₀ of 0·40·7. Expression of fusion proteins was induced by addition of 0·3 mM IPTG. After 1·5 h cells were harvested, resuspended in 2 ml chromatography buffer (20 mM Tris/HCl pH 7·4, 200 mM NaCl, 1 mM EDTA) and lysed using a French pressure cell. Cell debris was removed by low-speed centrifugation at 500 g; the supernatants were diluted to 10 mg total protein ml¹ and applied to an amylose resin (New England Biolabs) column of 2 ml bed volume. The column was washed first with 20 ml chromatography buffer and subsequently with 10 ml chromatography buffer supplemented with complete protease inhibitor (Roche Molecular Biochemicals). Fusion proteins were eluted with 4 ml elution buffer (20 mM Tris/HCl pH 7·4, 200 mM NaCl, 1 mM EDTA, 10 mM maltose). Protein concentrations in the samples were determined using the Bio-Rad Protein Assay.

Anion-exchange chromatography.
Affinity-purified fractions of MalE-IpgF and the active-site mutant MalE-IpgF_E42Q were dialysed against 200 vols 20 mM Tris/HCl pH 7·4 at 8 °C overnight. To allow dissociation of GroEL from the fusion protein, the dialysed fractions were incubated with 1 mM ATP, 1 mM MgCl₂ and 0·1 % Triton X-100 for 30 min at 25 °C (Keresztessy et al., 1996; Mattingly et al., 1995). The fractions were centrifuged at 10 000 g to remove any insoluble material. The supernatant corresponding to 8 mg fusion proteins was applied to a Mono Q HR 5/5 column (Amersham Pharmacia) using FPLC and eluted with a linear gradient from 0 to 2 M NaCl in 20 mM Tris/HCl pH 7·4 containing 0·1 % Triton X-100 at 8 °C. Fractions were collected and analysed by SDS-PAGE, Western blotting, zymogram analyses, in-solution muramidase assay, and the Cy3-PG spot assay.

MALDI-TOF analysis.
Samples (12 µg) of affinity-purified fractions of MalE-P19_153L and MalE-IpgF were separated electrophoretically on a 12·5 % SDS polyacrylamide gel. Proteins were Coomassie stained and the bands corresponding to the unknown protein were excised. The gel slices were destained with 500 µl 50 % methanol/50 % acetic acid in water at room temperature overnight. After a second destaining procedure, the gel slices were dehydrated in 200 µl acetonitrile. The samples were then reduced in 50 µl 10 mM DTT for 30 min at room temperature. After removal of DTT, proteins were alkylated with 50 µl 50 mM iodoacetamide for 30 min at room temperature and then washed with 100 µl 100 mM ammonium bicarbonate for 10 min. Subsequently, the gel slices were dehydrated as described above. The gel slices were then rehydrated by swelling in 100 µl 100 mM ammonium bicarbonate for 10 min. Ammonium bicarbonate was removed and the gel dehydrated as above. Acetonitrile was removed and 200 µl acetonitrile was added. Subsequently, the gel slices were dried in vacuo. Reswelling of the gel slices was achieved by incubation with 50 µl 20 ng µl¹ trypsin (Promega) in ice-cold 50 mM ammonium bicarbonate for 10 min on ice. Excess trypsin solution was removed and 10 µl ammonium bicarbonate was added. Tryptic digestion of the proteins in the gel slices was performed overnight at 37 °C. Peptides were first extracted with 30 µl 100 mM ammonium bicarbonate. After collection of the supernatant, peptides were extracted twice with 30 µl formic acid in 50 % acetonitrile. The three supernatant fractions were pooled and the sample volume was reduced to 15 µl in a vacuum centrifuge. MALDI-TOF analysis was performed by piChem, Graz. The GroEL chaperone of E. coli was identified using the mass fingerprint data obtained by MALDI-TOF and the PeptIdent software ().

Complementation assays.
Complementation of the gene 19 defect was performed with donor strains carrying either the resistance plasmid R1-16 or R1-16/mut1 together with a complementation plasmid. The latter plasmids, pMalpP19 wt, pMalpP19_150G, pMalpP19_144G, pMalpP19_E44Q, pMalpP19_E44A, pMalpIpgF, pMalpIpgF_E42Q, pMalpIagB, pMalpVirB1_AT, pMalpVirB1_BS, pMalpHP0523 and pMalpTrbN, are described in Table 1. Induction of transcription from the tac promoter with IPTG was not necessary for complementation. Overnight cultures from single colonies grown in 2 ml 2x TY, supplemented with kanamycin and ampicillin in case of the donors, were used for mating. Forty microlitres of the donor cells, i.e. E. coli MC1061 harbouring R1-16 or R1-16/mut1 and the different complementation plasmids, were pipetted into 0·9 ml prewarmed 2x TY medium and incubated for 60 min at 37 °C. Then 100 µl of an overnight culture of recipient cells, i.e. E. coli J5, was added, and the mixture was incubated for 60 min at 37 °C without shaking. Conjugation was interrupted by vigorously mixing for 1 min and placing the tubes on ice. Dilutions of 10³ to 10⁷ in 0·9 % NaCl were plated on lactose MacConkey agar plates containing kanamycin. The conjugation frequency was determined by counting white donor and red transconjugant colonies and is expressed as the number of transconjugants per 100 donor cells.

PG isolation and purification.
PG was isolated from 1 litre of E. coli J5 cells in stationary phase according to Rosenthal & Dziarski (1994). Cells were harvested and washed with 40 ml 10 mM Tris/HCl pH 6·8. After washing, cells were resuspended in 30 ml 10 mM Tris/HCl pH 6·8. This suspension was added dropwise to 300 ml boiling 4 % SDS. After an additional 45 min of boiling, PG sacculi were collected by ultracentrifugation at 200 000 g for 20 min at 20 °C. The pellet was resuspended in 150 ml 2 M NaCl and incubated overnight at room temperature. After ultracentrifugation as described above, sacculi were washed with 60 ml water, collected by ultracentrifugation and resuspended in 20 ml water containing 0·1 mM MgCl₂. The suspension was incubated first with 50 µg ml¹ DNase I, 50 µg ml¹ RNase A and 200 µg ml¹ α-amylase (Roche Molecular Biochemicals) for 90 min at 37 °C and second, after addition of 200 µg ml¹ Pronase (Roche Molecular Biochemicals) at 60 °C for 60 min. Enzymes were inactivated by addition of SDS to a final concentration of 8 % and 15 min boiling. PG was collected by ultracentrifugation and washed twice with 20 ml water. The final pellet was resuspended in 5 ml water and stored at 20 °C. The amount of PG was estimated by evaporation and weighing an aliquot of the dried material.

Zymogram analyses.
Zymogram analyses for the determination of PG-degrading activity was performed according to Bernadsky et al. (1994). Purified PG was incorporated into 12·5 % and 16 % polyacrylamide gels to a final concentration of 0·05 % (w/v). Protein samples were separated electrophoretically on polyacrylamide gels containing 0·02 % SDS. After electrophoresis, the gels were incubated with water for 60 min at 4 °C, then transferred to 25 mM potassium phosphate buffer pH 5·2 containing 0·1 % Triton X-100 (renaturation buffer) and incubated for 60 min at 4 °C. After this equilibration step, the gels were incubated with fresh renaturation buffer for 72 h. Peptidogylcan degradation was visualized by staining of gels with 0·1 % methylene blue/0·01 % KOH for 60 min at 4 °C. Gels were destained with water for 4560 min.

Muramidase assay.
Muramidase activity was determined by measuring solubilization of radiolabelled murein polymer (Engel et al., 1991). [³H]A₂pm-labelled murein sacculi (2·5 µg; 5000 c.p.m.) were incubated in the presence of enzyme samples (36 µM) in a total volume of 100 µl 20 mM Bistris buffer pH 5·3 for 30 min at 37 °C. To precipitate the insoluble substrate, 100 µl of 1 % cetyltrimethyl ammonium bromide solution was added. The samples were kept on ice for 30 min. After centrifugation for 2 min at 17 000 g in an Eppendorf centrifuge at 4 °C, 100 µl of the supernatant was added to 1·5 ml scintillation cocktail, and radioactivity was measured in a Beckman LS 6500 scintillation counter. Inhibition of muramidase activity was assayed by addition of 1 µM GroEL (20 µl) to the reaction mixture prior to incubation at 37 °C.

Cy3 labelling of PG.
A 500 µl sample (approx. 1 mg) of PG purified as described above was incubated with 50 µl 1 M borate buffer pH 9·6 and 5 µl Cy3 NHS ester (1 mg ml¹ in dimethylformamide, Amersham Biosciences) for 60 min at room temperature. Excess Cy3 was removed by dialysis against 500 vols water using Slide-A-Lyser dialysis cassettes (MWCO 10 000, Pierce). Cy3-labelled PG was stored at 20 °C.

Cy3-PG spot assay.
Eight-well glass slides (ICN) were coated with poly-L-lysine (Sigma-Aldrich) for 30 min at room temperature. After removal of unbound poly-L-lysine by washing with water, the slides were dried. Then 10 µl water and 1 µl Cy3-labelled PG were spotted onto the coated wells and incubated at room temperature for 45 min in the dark. To remove residual SDS and low-molecular-mass material the slides were rinsed with deionized water for 2 min and then dried at room temperature. Ten microlitres of buffer (10 mM or 25 mM sodium phosphate or ammonium acetate, pH 6) was pipetted on each of the Cy3-PG spots. Before and after addition of 1 µl enzyme, fluorescence microscopy images were taken using a Zeiss Axioskop and Zeiss filter set #15 (BP 546/12, FT 580, LP 590). Fluorescence microscopy images were taken using a cooled CCD camera and analysed with the Metamorph 5.0 software package (Visitron Systems). To determine the pH optimum of MalE-IpgF the Cy3-PG spot assay was performed as described above using different buffers for incubation (pH 4 and 5, 10 mM sodium citrate buffer; pH 6 and 7, 10 mM sodium phosphate buffer; pH 6·8 and 8, 10 mM Tris/HCl). After addition of 1 µl MalE-IpgF the slides were incubated at room temperature for 15 min. To remove the digested material, the slides were rinsed with deionized water for 2 min and dried at room temperature. To quantitate the fluorescence signals, dry slides were scanned before and after incubation with the enzyme using an array scanner (Axon GenePix 4000B, PMT setting 320, scan power 33 %) and images were obtained with the GenePix Pro 4.1 program; the 16 bit tiff files were subsequently analysed using ImageQuant 5.1 software. PG-degrading activity of IpgF was calculated as the ratio of fluorescent PG spot values before and after treatment minus one. To test the effects of bulgecin A a gift of A. J. Dijkstra (HoffmannLa Roche, Basel, Switzerland) and hexa-N-acetylchitohexaose (Seikagaku, Japan), various amounts of the compounds were added to the reaction mixture prior to the addition of the enzyme. The stocks of the various enzyme preparations had the following concentrations: MalE-IpgF, 6 µg µl¹; MalE-IpgF_E44Q, 4 µg µl¹; lysozyme from chicken egg white (Sigma), 20 µg µl¹; GroEL, 1·6 µg µl¹.

Overexpression and purification of MalE-LT fusion proteins
Earlier attempts to overexpress specialized LTs in the cytoplasm of E. coli resulted in formation of inclusion bodies which could only be solubilized in 8 M urea. Renaturation and recovery of active proteins was not possible. This was observed with native proteins as well as with different fusion constructs that were tested (His-tag, Trx-tag, S-tag; data not shown). The proportion of the soluble target protein was increased significantly when fused to the maltose-binding protein (MBP or MalE) from E. coli. Therefore, we constructed plasmids for the expression of N-terminal MalE fusions of P19 wild-type, C-terminally truncated versions of P19 (P19_153L, P19_150G, P19_144G, P19_140R the C-terminal amino acid residue in the deletion variant is indicated), and mutated versions of P19 (P19_E44Q and P19_E44A). The C-terminal deletions of P19 were included in the analysis to determine the minimal P19 core that is sufficient to confer in vivo and in vitro activity. We also constructed plasmids for the expression of MalE fusions with IpgF (Sh. sonnei, virulence plasmid T3SS) and the assumed active site mutant IpgF_E42Q, IagB (Sal. enterica, chromosomal T3SS), VirB1_AT (VirB1 from A. tumefaciens, Ti plasmid T4SS), VirB1_BS (VirB1 from B. suis, chromosomal T4SS), TrbN (Pseudomonas sp., conjugative resistance plasmid RP4 T4SS) and HP0523 (H. pylori, cag-PAI T4SS). The relevant protein data and accession numbers for the protein sequences in the databases are listed in Table 2. Both periplasmic and cytoplasmic expression constructs were made (see Table 1). Whereas the periplasmic constructs were used for the in vivo complementation assays, overexpression and affinity-purification steps were performed with fusion proteins expressed in the cytoplasm. In the latter case protein yield and stability were considerably higher. Signal sequences, when present, were removed during the cloning procedure. Despite the sequence similarity of the LTs chosen for expression and purification (see Table 3), the corresponding fusion proteins behaved quite differently during affinity chromatography. In the case of MalE-VirB1_AT and MalE-HP0523 at least one-third of the fusion proteins present in the crude extract did not bind to the amylose affinity resin. Best results in the first purification step were obtained with MalE-IpgF and MalE-P19; the average yield from 500 ml cultures was 815 mg of approximately 50 % pure fusion protein.

Table 2. Relevant protein data and sequence accession numbers

Table 3. Similarity scores *The BESTFIT algorithm with the BLOSUM 62 matrix (gap weight 6; length weight 2, gaps longer than 5 not penalized) of the Wisconsin Package, Version 10.3 was used. The numbers indicate similarity/identity values.

In vivo activity of MalE-LT fusion proteins
To determine the in vivo activity of the MalE-P19 fusion protein we used conjugation assays to assess the effects of mutations in protein P19 (Bayer et al., 2001). In addition to investigating the biological activity of P19 and P19 mutants, we were interested in cross-complementation phenotypes of specialized LTs encoded by heterologous T3SS and T4SS. Complementation of the gene 19 defect was taken as an indication that the MalE-LT fusion protein tested degraded E. coli PG in vivo, and that it was able to appropriately interact with the DNA transport apparatus (i.e. the T4SS) encoded by plasmid R1. Cells harbouring either plasmid R1-16 (a derepressed variant of plasmid R1) or R1-16 and pMAL-p2X (the vector plasmid) served as positive controls. Cells harbouring R1-16/mut1 (R1-16 defective in gene 19) and pMAL-p2X served as negative control. Table 4 shows that MalE-P19 wild-type fully complemented the conjugation deficiency of R1-16/mut1. We concluded that the MalE portion of the fusion protein did not interfere with the in vivo activity of P19 or other LTs. Complementation analysis revealed that amino acid residues between positions 144 and 150 of P19 wild-type are critical for its activity: MalE-P19_150G complemented the conjugation defect, whereas in the case of MalE-P19_144G and MalE-P19_140R no complementation was observed. As expected, the glutamyl residue at position 44 of P19 wild-type, which is invariantly conserved in the SLT family of proteins, was essential for its in vivo activity. Neither MalE-P19_E44Q nor MalE-P19_E44A could complement the conjugation defect. Finally, the heterologous complementation experiments showed that MalE-IpgF and MalE-TrbN restore conjugation to wild-type levels. As for MalE-P19_E44Q and MalE-P19_E44A, the exchange of the conserved glutamyl residue in MalE-IpgF_E42Q resulted in loss of the ability to complement the conjugation defect. MalE-IagB, MalE-VirB1_AT and MalE-VirB1_BS did not complement the DNA transfer defect; MalE-HP0523 showed an intermediate phenotype (Table 4).

Table 4. In vivo complementation experiments Mean values of at least three independently carried out experiments are given.

PG degradation by MalE-LT fusion proteins: zymogram analyses
A method which has been widely used in the past for the detection of PG degradation by muramidases is the analysis of proteins in zymograms (Bernadsky et al., 1994). In zymograms, polymeric PG is incorporated into a polyacrylamide gel matrix. After electrophoresis under denaturing conditions and removal of SDS, the proteins are allowed to refold (for details see Methods). After methylene blue staining and subsequent destaining, clear, non-stained zones indicate degradation of the PG substrate by the refolded muramidase. In our zymogram analyses we first tested MalE-P19 (Figs 1a and 2c). As expected, clear zones appeared in the case of MalE-P19, suggesting a PG-degrading activity which was virtually indistinguishable from the activity seen with the positive control protein MltB (Fig. 2a), the well-characterized membrane-bound LT of E. coli (van Asselt et al., 2000). MalE did not produce clear zones (Fig. 2b), unambiguously demonstrating that the activity seen with MalE-P19 was due to the LT part of the fusion protein. Investigation of C-terminally truncated versions of MalE-P19 revealed that both MalE-P19_153L and MalE-P19_150G were able to degrade PG, indicating that amino acid residues between positions 150 and 169 of P19 are not essential for protein folding, substrate recognition and degradation (Fig. 1b, c). In contrast, zymogram analysis of MalE-P19_144G (Fig. 1d) and MalE-P19_140R (data not shown) clearly demonstrated that these C-terminal deletions of the P19 fusion proteins are inactive. These results suggested that these truncated proteins either did not fold correctly or lost the ability to bind to and/or degrade the substrate.

(72K): Fig. 1. Zymogram analyses of MalE-P19 and MalE-IpgF proteins. Affinity-purified fusion proteins as indicated were separated by modified SDS-PAGE and renatured in the PG-containing gel as described in Methods. For all proteins, images showing the relevant details of the same gel, produced by staining with either methylene blue (Z) or Coomassie (C), are represented. Arrowheads point to clear, non-stained zones, indicating PG degradation by MalE-P19, MalE-P19_153L, MalE-P19_150G and MalE-IpgF. In contrast, MalE-P19_144G and MalE-IpgF_E42Q are inactive.

(56K): Fig. 2. Zymogram analyses of MalE-LT fusion proteins from different secretion systems. The indicated untreated () and Factor Xa treated (+) affinity-purified fusion proteins as indicated were separated by modified SDS-PAGE and renatured in the PG-containing gel as described in Methods. For all proteins, images showing the relevant details of the same gel, produced by staining with either methylene blue (Z) or Coomassie (C), are represented. Black and white arrowheads indicate active fusion proteins and active LTs after protease cleavage, respectively. MltB served as a positive control, MalE was used as a negative control.

MalE-IpgF also produced cleared zones in the zymogram gels and can therefore be regarded as a PG-degrading enzyme (Fig. 1e). The catalytic activity of this specialized LT was lost when the conserved glutamyl residue in the presumed active site of this enzyme was changed to glutamine (Fig. 1f). The analysis of the PG-degrading activities of MalE-LT fusion proteins was not only carried out with full-length fusion proteins but also with fusion proteins treated with Factor Xa. This serine protease causes the release of the respective LT from MalE via a specific cleavage site. The tested wild-type forms of the MalE-LT fusion proteins (MalE-P19, MalE-IpgF, MalE-IagB, MalE-VirB1_AT, MalE-VirB1_BS and MalE-HP0523), as shown in Fig. 2(ch), produced cleared zones in zymogram gels, indicating degradation of polymeric PG. After cleavage with Factor Xa only the released LTs produced cleared zones, whereas the MalE part, as expected, did not show a positive signal.

Interaction of MalE-LT fusion proteins with GroEL
SDS-PAGE of affinity-purified MalE-LT fusion proteins revealed that in all cases except MalE-P19_140R (data not shown), a protein with a molecular mass of approximately 60 kDa co-purified with the MalE-LT fusion protein (arrowhead in Fig. 3b). The protein corresponding to that band was identified by MALDI-TOF analysis of peptide fragments produced by an in-gel tryptic digest. Six prominent peptide fragments were detected which gave an 86·6 % match to the E. coli chaperonin GroEL (data not shown). The results obtained using peptide mass fingerprinting were confirmed by Western blotting and detection using a commercially available anti-GroEL antibody (Fig. 3c). GroEL did not co-purify with a preparation of MalE alone; thus, GroEL specifically interacted with the LT part of the fusion protein. Several attempts to remove GroEL from MalE-LTs originating from T4SS failed. MalE-P19 turned out to be unstable and precipitated as soon as GroEL was removed. Removal of GroEL was only accomplished in the case of affinity-purified MalE-IpgF (Fig. 3) and MalE-IpgF_E42Q (data not shown) after addition of ATP and Mg²⁺ and subsequent anion-exchange chromatography. In this way, MalE-IpgF and MalE-IpgF_E42Q proteins with an estimated purity of >95 % suitable for further enzymic characterizations were obtained.

(55K): Fig. 3. Purification of MalE-IpgF and removal of GroEL. (a) Elution profile of proteins separated by anion-exchange chromatography. Numbers indicate peaks that were analysed by SDS-PAGE (b) and Western blotting (c, d). (b) SDS-PAGE (12·5 % polyacrylamide) of selected fractions. AC, 10 µg affinity-chromatography-purified MalE-IpgF; fraction 2 contains >95 % pure MalE-IpgF; fractions 35 contain MalE-IpgF and degradation products. (c) Western blot analysis of the indicated fractions using anti-GroEL antibody. A GroEL signal was detected after affinity purification (AC) and in fractions 68. No GroEL signal was detected in fractions 15, indicating successful removal of GroEL from MalE-IpgF. (d) After detection of the GroEL signal, the membrane was stripped and analysed with anti-MalE antibody. The most intense MalE signals were detected in fractions 25. Double bands in fractions AC, 6, 7 and 8 result from incomplete removal of the anti-GroEL antibody during the stripping procedure.

MalE-IpgF degrades PG in solution
MalE-IpgF, purified as described above, was subjected to an in-solution muramidase assay to demonstrate PG-degrading activity. In this assay purified, radioactively labelled polymeric PG from E. coli was used as a substrate. An active muramidase generates soluble fragments that remain in the supernatant after a precipitation step (for details see Methods). As shown in Fig. 4, only purified, GroEL-free, MalE-IpgF degraded PG in a manner comparable to lysozyme, which was used in the same experiment as a positive control. The negative controls MalE and thyroglobin did not show any PG-degrading activity. When GroEL (1 µM) was added to the reaction mixture, the activity of MalE-IpgF (3 µM) decreased by 80 %, demonstrating that GroEL not only interacted with IpgF but also inhibited its muralytic activity. In contrast, the enzymic activity of lysozyme was not adversely affected by addition of GroEL (Fig. 4).

(12K): Fig. 4. Muramidase activity of MalE-IpgF. Degradation of murein was determined by measuring the amount of soluble radiolabelled fragments released after incubation without enzyme (blank), and with MalE, lysozyme, MalE-IpgF or thyroglobin (grey bars). Radioactivity increased at least fivefold in the supernatants derived from lysozyme- and MalE-IpgF-treated murein sacculi compared to the negative controls (blank, MalE- and thyroglobin-treated sacculi), demonstrating muramidase activity of MalE-IpgF. To test the effect of GroEL on the activity of MalE-IpgF, GroEL was added to the reaction mixture (white bars). Addition of GroEL inhibited the activity of MalE-IpgF whereas no inhibitory effect on lysozyme was detectable.

A new test system for PG-degrading enzymes: the Cy3-PG spot assay
The in-solution muramidase assay using radiolabelled PG sacculi has been a useful tool over the last 20 years to investigate the activity of murein-degrading enzymes. Nevertheless, it is a time-consuming method, and PG sacculi need to be radioactively labelled in vivo. To permit rapid testing for peptidoglycanase activity we developed a new assay which is easier to perform, reliable, and suitable for screening for inhibitors of murein-degrading enzymes. We used the fluorescent dye Cy3 to label purified PG isolated from the Gram-negative bacterium E. coli. Labelling of large amounts of sacculi can be performed within 12 h, and storage of the labelled (insoluble) substrate is possible for at least 6 months at 20 °C. The labelling protocol is very efficient, so that small amounts of the substrate are sufficient for the assay (for details see Methods). Fluorescence microscopy showed that Cy3-labelled PG sacculi appear as intensely fluorescing aggregates (Fig. 5a, no enzyme); at higher magnifications single labelled PG sacculi were visible (Fig. 5c). Addition of MalE-IpgF to the buffer resulted in rapid dissolution of the aggregates and diffusion of the breakdown products through the liquid (Fig. 5a); the fluorescing sacculi collapsed and net-like, less well defined irregular structures were observed (Fig. 5c). After 10 min the Cy3-labelled PG aggregates in the area where MalE-IpgF had been pipetted were completely dissolved (Fig. 5b). Incubation of Cy3-labelled PG with buffer alone or with the mutant MalE-IpgF_E42Q did not lead to any visible changes in the appearance of the aggregates (Fig. 5b). The newly developed test system was verified using lysozyme from chicken egg white. As can be seen in Fig. 5(c), lysozyme degraded Cy3-labelled PG in a manner comparable to IpgF. To verify that not only were aggregates dissolved by the muralytic enzymes but specific breakdown products were formed we performed a modified version of FACE (fluorophore-assisted carbohydrate electrophoresis), a method that has been introduced as an alternative to HPLC analysis of PG degradation products (Li et al., 2004). Specific, low-molecular-mass degradation products were seen only when MalE-IpgF was used and not in the case of MalE-IpgF_E42Q (our unpublished data).

(101K): Fig. 5. Muramidase activity of MalE-IpgF detected by Cy3-PG spot assay. Cy3-labelled PG was bound to poly-L-lysine-covered glass slides as described in Methods. Images were taken using a fluorescence microscope. The PG-degrading activity of MalE-IpgF results in dissolution of PG aggregates and diffusion of breakdown products over the time of incubation. (a) Cy3-PG spot before (no enzyme) and after 2, 4 and 6 min incubation with MalE-IpgF. (b) Degradation of Cy3-PG by wild-type IpgF in comparison to IpgF_E42Q. Images were taken before (no enzyme) and after 10 min incubation with the respective enzyme preparation. (c) Higher magnifications show oval-shaped fluorescing sacculi corresponding to the size of E. coli cells. After treatment with IpgF or lysozyme the sacculi collapse to net-like structures indicating degradation of PG. Hexa-N-acetylchitohexaose (4 mM) inhibits the muralytic activity of IpgF. Bars represent 1 mm in (a) and (b), and 10 µm in (c).

pH optimum and inhibition of MalE-IpgF by known LT inhibitors
In further experiments using a variation of the Cy3-PG spot assay in combination with an array scanner (see Methods) the optimal pH for IpgF and the effects of LT inhibitors were determined (Fig. 6). As shown in Fig. 6(a), optimal activity of IpgF was observed at pH 6, and less activity at pH 5 and pH 7; no activity could be measured at pH 4 and very little at pH 8. We conclude that the pH optimum of IpgF lies at pH 6. The same assay system was used to determine the activities of IpgF in comparison to IpgF_E42Q and lysozyme. As expected, IpgF and lysozyme were active, whereas IpgF_E42Q was not. As observed earlier, GroEL, when added to the substrate together with the enzyme, inhibited IpgF but not lysozyme (data not shown). A similar observation was made with hexa-N-acetylchitohexaose, which completely inhibited IpgF at a concentration of 4 mM (Fig. 6b). In contrast, lysozyme was not affected by the presence of N-acetylchitohexaose (data not shown), which is in accordance with the observation that this substance is a substrate for lysozyme and solely inhibits LTs (Leung et al., 2001; Song et al., 1994). We also tested bulgecin A, which has been described as an inhibitor of LTs (Templin et al., 1992) and found that a concentration of 20 mM was sufficient to inhibit IpgF (Fig. 6c).

(11K): Fig. 6. pH optimum and inhibition of MalE-IpgF. Quantitative analysis of the Cy3-PG spot assay as described in Methods demonstrated that the pH optimum of the MalE-IpgF fusion protein is at pH 6 (a). The same assay was used to determine the inhibition by known LT inhibitors hexa-N-acetylchitohexaose (b) and bulgecin A (c). The enzyme's muralytic activity was calculated (see Methods) and is expressed as arbitrary units. ND, Not detectable. The results of one representative experiment are shown in each panel.

In this paper we show for the first time that several selected members of the LT family of proteins, termed specialized LTs (Koraimann, 2003), are indeed capable of degrading PG in vitro as suggested by the results of our zymogram analyses. Furthermore, the IpgF protein from Sh. sonnei, purified to homogeneity, degraded polymeric radioactively labelled PG in solution, and Cy3-labelled PG in a newly developed Cy3-PG spot assay. We could show by complementation analyses and in vitro assays that the conserved glutamyl residue corresponding to the catalytic residue of LTs position 42 in IpgF is absolutely necessary for its activity. Since IpgF was inhibited by both bulgecin A and N-acetylchitohexaose we conclude that IpgF cleaves the β-1,4 glycosidic bond between the MurNAc and GlcNAc moieties in PG isolated from E. coli in a manner characteristic for other well-characterized LTs like Slt70, Slt35 or bacteriophage λ lysozyme (Leung et al., 2001; Thunnissen et al., 1995; van Asselt et al., 2000). In addition to the demonstrated in vitro PG-degrading activity of IpgF, this T3SS-encoded LT also cross-complemented a defect in P19, the LT in the conjugation system of the IncF plasmid R1 (a T4SS). We have proposed earlier that P19 of plasmid R1 interacts with the T4SS apparatus encoded by the conjugative plasmid and that the protein facilitates the efficient assembly of the transport complex in the cellular envelope by local degradation of the cell wall (Bayer et al., 2000, 2001). Our results are not only fully consistent with this view but also suggest that LTs from different secretion systems can fully or partially substitute each other. A similar observation has been made in the case of the VirB1 proteins of A. tumefaciens and B. suis (Höppner et al., 2004). This intriguingly low level of specificity and the observed genetic redundancy of LT genes could explain the sometimes weak effects of disruptions of a single LT gene in a given secretion system (Koraimann, 2003). IpgF is one of the closest sequence relatives of P19, although it is encoded by a T3SS of Sh. sonnei. The sequence identity (43 % in the SLT domain) was sufficient for the restoration of the wild-type conjugation frequency. In contrast to IpgF, IagB did not complement the gene 19 defect in vivo (39 % sequence identity in the SLT domain, Table 3). The differences between the IpgF and IagB proteins could be sufficient to disable complementation by IagB and may hint at important residues necessary for a specific interaction of P19 with its postulated but as-yet-unidentified interaction partner. The fact that VirB1_AT and VirB1_BS did not complement the gene 19 deficiency in vivo is probably due to major differences between the VirB1 proteins and P19. VirB1_AT and VirB1_BS both have a C-terminal extension which is only present in the VirB1 class of specialized LTs. In T-DNA transfer of A. tumefaciens VirB1_AT is processed and a peptide, VirB1*, representing this C-terminal extension, is secreted into medium or loosely interacts with the bacterial cell surface (Baron et al., 1997). VirB1_AT was shown to interact with several VirB proteins, among them VirB8, which has been proposed to recruit VirB1 to the assembly site for the VirB transporter complex (Ward et al., 2002). Similarly, VirB1_BS has been shown to interact with core components VirB8, VirB9 and VirB11 (Höppner et al., 2005). In complementation assays it was shown that VirB1_BS can complement a VirB1_AT deletion, whereas P19 encoded by the F plasmid (termed Orf169) and HP0523 did not restore wild-type levels of tumour formation and IncQ plasmid transfer (Höppner et al., 2004). Thus, the cross-complementation observed by us and by the Baron laboratory (Höppner et al., 2004) are fully consistent and suggest that specialized LTs are probably late additions to secretion systems and function to replace existing inefficient cellular enzymes designed for PG turnover.

One intriguing observation during our efforts to purify MalE-LT fusion proteins was that they strongly interacted with the chaperone GroEL in the cytoplasm. Since LTs normally possess signal sequences and are transported into the periplasm like P19 (Bayer et al., 2000), further investigations are currently being performed in this laboratory to elucidate whether the binding of GroEL to the LT is solely an artifact created by overexpression of the fusion proteins with concomitant accumulation of misfolded protein in the cytoplasm or serves a physiological role. Interestingly, the interaction of GroEL with the MalE-IpgF fusion protein also inhibited the PG-degrading activity of this enzyme. Presumably, the LT proteins are instable or misfolded in the cytoplasm and are recognized by the GroEL/ES complex unless they are transported to the periplasm by the Sec machinery. At least in the case of the 169 aa P19 protein, residues between R140 and G144 (RIYTG) are mainly responsible for the strong interaction with GroEL. Substrate recognition by the GroEL/ES complex is not well defined; the molecular basis of substrate recognition and binding by GroEL is supposed to be the presentation of hydrophobic amino acids on the surface of misfolded proteins (Wang et al., 1999).

It has been recently recognized that PG fragments are specifically recognized by NOD proteins, which have been identified as key constituents of the innate immune system of animals and humans (for reviews see Dziarski, 2003; Girardin & Philpott, 2004; Royet & Reichhart, 2003). Human Nod1 is a cytosolic receptor in epithelial cells and detects a unique diaminopimelate-containing GlcNAc-MurNAc tripeptide found in Gram-negative bacterial PG, resulting in activation of the transcription factor NF-κB pathway (Girardin et al., 2003). Only very recently has it been found that non-invasive H. pylori elicits this innate immune response by delivering bacterial PG fragments into epithelial cells via the H. pylori cag-PAI (Viala et al., 2004). It is tempting to speculate that in this case HP0523, the specialized LT encoded by the H. pylori cag-PAI, could produce the PG fragments that are then translocated via the T4SS into epithelial host cells. HP0523 from H. pylori was shown here to be capable of degrading PG in vitro and it was found earlier to be essential for both CagA translocation and IL-8 induction (Fischer et al., 2001). Future studies will be required to address the question whether HP0523 or other LTs like IpgF can produce PG fragments that are recognized by Nod1 and thus can trigger an innate immune response.

We thank Arnoud J. Dijkstra for providing purified MltB and bulgecin A and Joachim V. Höltje for providing radiolabelled murein sacculi. We thank Ingeborg Lederer for providing heat-inactivated Sh. sonnei isolates, David O'Callaghan for DNA clones containing B. suis virB1, Christian Baron for pointing to the similarity of HP0523 with the LT family of proteins and for sending H. pylori hp0523 clones, Peter Christie for sending A. tumefaciens virB1-containing DNA and Erich Lanka for a plasmid construct containing RP4 trbN. Work in our laboratory on the function of specialized LTs is supported by the EU project QLK22-CT-2001-01200 within the fifth framework (FP5).