Introduction

Xanthomonas vesicatoria (previously named X. campestris) is a gram-negative bacteria of widespread occurrence in nature. The phylogenetic relationships of all validly described species of Xanthomonas were analyzed by sequencing and comparing 16S ribosomal DNAs (Hauben et al. 1997). Members of the genus Xanthomonas are known to be plant pathogens. The plant pathogenicity involves the overcoming and breakdown of the cell wall (Williamson et al. 1998; Jones et al. 2000) and the action of virulence factors. Fenselau et al. (1992) characterized the genes from X. vesicatoria necessary for the induction of the hypersensitive response in resistant plants (hrp genes). They hypothesized that these genes are involved in the secretion of molecules essential for the interaction of X.c. pv vesicatoria with the plant. Virulence factors can be grouped into families, based either on their function or on their mechanism of export to the bacterial surface (Finlay and Falkow 1997).

There are two main reasons to study the esterolytic enzymes of X. vesicatoria: (1) their important role in phytopathogenic bacterial species causing plant diseases (Starr 1975) and (2) their biocatalytic potential for industrial applications (Van den Mooter and Swings 1990). Esterases in general represent a group of hydrolytic enzymes of broad natural variety concerning substrate and reaction type. However, the physiological functions of many esterases are not clear. Some of these enzymes are known to be involved in metabolic pathways connected to virulence (McQueen and Schottel 1987). Substrate-specificity and interface activation are used as a basis for distinguishing esterases from true lipases (Turner 1995). Lipases are known to be activated by oil/water interfaces before hydrolyzing water-insoluble substrates with long-chain fatty acid esters; and the knowledge of bacterial esterolytic enzymes is rapidly increasing. Three different classes of esterolytic enzymes were described for bacteria, including carboxylesterases (EC 3.1.1.1), true lipases (EC 3.1.1.3) and various types of phospholipase (Songer 1997; Titball 1998). An updated and extensive classification of bacterial esterases and lipases was proposed by Arpigny and Jaeger (1999), who defined eight different families.

Esterases are employed for the enzymatic hydrolysis of carboxylic esters into their corresponding carboxylic acids and alcohols and for transesterifications and ester synthesis. The ability of esterases to act both in aqueous and nonaqueous systems makes them good tools for organic synthesis (Sheldon 1993). Enantioselective esterases are especially of high interest for the synthesis of enantiomerically pure products (Turner 1994). To provide a good basis for biocatalytic applications, it is very desirable to obtain more information about the biocatalytic characteristics of esterases acting on organic substrates (Ozaki and Sakashita 1997).

In this paper, we report on a novel recombinant esterase, Xv_EstE, from X. vesicatoria. Our data indicate that: (1) this esterase is a member of the GDSL family of lipolytic enzymes, (2) it exhibits activity to polar water-soluble carboxylic esters of long-chain fatty acids and (3) it is integrated into the outer membrane, being the first example of an outer membrane β-barrel protein with a N-terminal periplasmic extension.

Materials and methods

Bacterial strains, plasmids and culture conditions

X. vesicatoria was obtained from the Deutsche Stammsammlung (DSM 50861).

Escherichia coli strains HB101 (ATCC 33649; F− hsdS20 (rB − mB −) supE44 araA2 rpsL20 (strR) xyl-5 mtlI recA13) and SURE {Stratagene; hsdR mcrA mcrB mrr endA1 supE44 thi-1, λ−, gyrA96, relA1, lac, recB, recJ, sbcC, umuC, uvrC, [F', proAB, laqIqZDM15, Tn10, (tetr)]} were used as hosts for gene libraries and basic cloning work. E. coli BL21(DE3) [ATCC 10248; F− ompT (lon) hsdSB rB − mB −] containing DE3, a λ prophage carrying the T7 RNA polymerase was used as host for overexpression experiments. The vectors pGEM-3Zf(+) and pGEM-7Zf(+) were used for standard cloning experiments. The vector pMS470Δ8 (Balzer et al. 1992; AmpR, lacIq, P tac , SD-T7) was used for expression constructs.

Bacterial cells were routinely grown in 250 ml of LB medium (1% bactotryptone, 0.5% yeast extract, 0.5% NaCl) or on LB agar plates containing 15 g agar l−1. For plasmid selection, ampicillin (100 mg l−1) was added (LBamp).

Recombinant DNA work and construction of genomic library

Recombinant DNA work and the preparation of plasmid DNA were performed according to standard methods described by Sambrook et al. (1989) or Ausubel et al. (2001). Genomic DNA of X. vesicatoria DSM 50861 was prepared by the method of Saito and Miura (1963). Fragments of approximately 10 kb were obtained by partial Sau3AI digestion and size-selection by sucrose gradient centrifugation. After dialysis against 1×TE (10 mM Tris, 1 mM EDTA) for 2 days, the fragments were ligated into pGEM-3Zf(+). Ligation products were used to transform E. coli HB101.

Screening the genomic library

About 14,500 clones, of the X. campestris library, with an average insertion size of about 7.5 kb (99.9% efficiency; Clarke and Carbon 1976), were plated at a density of about 100 colonies per LB/tributyrin plate, grown for 1–6 days at 37 °C and screened for esterolytic and/or lipolytic activity by visual detection of halo formation, according to Oterholm and Ordal (1966). Four active clones were initially isolated and clone pGSX9 (insert size 5.6 kb), which showed the highest activity on LB/tributyrin plates, was selected. After further deletion and subcloning experiments, the esterase-coding region was finally assigned to a 2.8-kb fragment (StuI/Sau3AI) which was cloned as a StuI-HindIII fragment into HindIII-SmaI cut pGEM-7Zf(+), resulting in plasmid pGSX9Δ0.

DNA sequencing, sequence analysis and computer-aided structure prediction

Sequencing was performed by the dideoxy chain-termination method (Sanger et al. 1977), using an automated DNA-sequencing system (ABI 373A sequencer) and the BigDyeDeoxy terminator-sequencing kit (Applied Biosystems, Foster City, Calif.). Sequence analysis was performed using the Wisconsin genetics computer group (GCG) sequence analysis software package (Devereux et al. 1984).

For the prediction of amphipathic β-sheets, the FORTRAN program AMPHI was used. This program was successfully used in predicting the structure of the β-barrel protein, OmpF (Vogel and Jähnig 1986), before crystal data became available. Moreover, it was also helpful for the prediction of the β-barrel transporter unit of the autotransporter proteins (Jose et al. 1995), which were subsequently confirmed by experiment (Maurer et al. 1999). The underlying equation is Hβ(i)=[h(i+/−4)+h(i+/−2)+h(i)]/5 and an amphipathic β-strand is predicted when the hydrophobicity Hβ(i) varies within a period of two residues by Hβ(i)≥0.6 and Hβ(i+1)≥0.4 (Jähnig 1990; Maurer et al. 1997). The Kyte and Doolittle hydrophobicity values (Kyte and Doolittle 1982) were utilized as h for each amino acid. The results obtained with AMPHI were confirmed by calculating regions of high surface probability, as described by Emini et al. (1985), using the GCG program from the Wisconsin sequence analysis package (Devereux et al. 1984).

Construction of Xv_EstE overexpression clone

The NcoI restriction site (nucleotide position 865 in the deposited sequence) at the predicted translation start site was converted into a NdeI site by PCR. A plasmid based on pGEM-Z5f(+) was constructed which included a small PCR-generated NdeI-NcoI N-terminal fragment and the remainder of the esterase-coding region taken as a partial NcoI (nucleotide position 934)–SphI (vector site) fragment from pGSX9Δ0. This generated plasmid p5ZXEX9. From there, the entire esterase gene was taken as a NdeI-SphI fragment and cloned into pMS470Δ8, to generate the expression plasmid pMSEX9. This construct was finally transformed into E. coli BL21(DE3) cells.

Overexpression and preparation of Xv_EstE

A preculture of 100 ml was usually grown overnight at 37 °C to the stationary phase. All subsequent cultures for enzyme preparations were grown at 30 °C. Then, 10 ml of the preculture were used for inoculation of a second preculture, which was grown to the late exponential phase. Again, 10 ml of the second preculture were used for the inoculation of the main cultures (250 ml). Induction of esterase expression was performed at an optical density at 595 nm (OD595) of 0.8, by adding isopropylthiogalactoside (IPTG) to a final concentration of 0.5 mM. After 6 h of incubation, the cells were harvested by centrifugation (10,000 g for 10 min). The pellet was resuspended in 3 ml of 100 mM Tris/HCl (pH 7.0). After freezing and a minimum storage of 30 min at −20 °C, the cells were thawed, vortexed and disrupted by sonication with a Branson Sonifier 250 three times (30 s each time) on ice, with a duty cycle set to 70%. The obtained total cell lysate was ultracentrifuged (45,000 g for 1 h) and the pellet fraction, which contained nearly all esterase activity, was resuspended in 3 ml of 100 mM Tris/HCl (pH 7.0).

Differential cell fractionation

Cells were grown overnight and 1.5 ml of the culture were used to inoculate 40 ml of LBamp medium. The fresh culture was incubated at 37 °C with vigorous shaking (200 rpm) until an OD578 of 0.6 was reached. IPTG was added to a final concentration of 1 mM and cells were incubated for a further 1 h at the same conditions. After harvesting and washing with phosphate-buffered saline (PBS), subcellular compartments were prepared by a protocol based on the method of Hantke (1981). Briefly, cells were resuspended in 1.5 ml of 0.2 M Tris/HCl (pH 8.0) and then 0.1 ml of 1 M saccharose, 0.1 ml of 10 mM EDTA, 0.1 ml of lysozyme (2 mg ml−1) and 3.2 ml of aqua bidest were added sequentially. The cell suspension was kept at room temperature until spheroblasts could be detected by microscopy. Then, phenylmethanesulfonylfluoride was added to a final concentration of 0.5 nM and aprotinin was added to a final concentration of 1 µg ml−1. After the addition of 5 ml of extraction buffer (50 mM Tris/HCl, pH 8.0, 2% Triton X-100, 10 mM MgCl2) and 100 µl of DNase (100 mg ml−1), the suspension was transferred onto ice and kept for at least 20 min, until a remarkable reduction in viscosity was detectable. After centrifugation at 1,500 g at 4 °C for 5 min to remove cell debris, the supernatant was taken as whole cell lysate. To obtain the membrane fractions, the supernatant was centrifuged at 25,000 g for 60 min. The resulting supernatant contained the cytoplasm proteins, whereas the membranes sedimented. The pelleted membranes were separated into a cytoplasm membrane fraction and an outer membrane fraction by washing with 1% N-laurylsarcosine in PBS. After another centrifugation at 25,000 g at 4 °C for 60 min, the cytoplasm membrane proteins were in the supernatant, whereas the outer membrane proteins precipitated as a white sediment. The outer membrane pellet was washed twice before being resuspended thoroughly in an adequate volume of aqua bidest. For trypsin digestion of the outer membrane preparations, 1 ml of outer membrane suspension was supplied with 10 µl of trypsin (2.5%) and incubated for 30 min at 37 °C. The reaction was stopped by adding an equal volume of sample buffer.

For whole-cell protease treatment, cells were harvested from liquid culture, washed and resuspended in 5 ml of PBS. Trypsin was added to a final concentration of 50 mg l−1 and cells were incubated for 30 min at 37 °C. Digestion was stopped by washing three times with PBS containing 10% fetal calf serum and cell fractions were obtained as described above.

Protein methods, SDS-PAGE and Western blot analysis

Protein concentration was determined by the method of Bradford (1976) with bovine serum albumin (BSA) as a standard, using a commercial reagent kit from Biorad. SDS-PAGE was used for monitoring proteins in cell extracts (Sambrook et al. 1989). For native PAGE, cell extracts were diluted 1:1 in 2× dissociation buffer (10 mM Tris/HCl, pH 7.4, 50% glycerol, 1% Triton X-100, 0.1% bromophenol blue) and incubated for 5 min at room temperature prior to loading onto the PA gel (12% separating gel in 0.33 M Tris/HCl, pH 7.8; 4% stacking gel in 0.062 M Tris/HCl, pH 6.8; running buffer 0.049 M Tris base, 0.038 M Glycine, pH 8.3). After electrophoresis, the gels were stained with Coomassie brilliant blue.

For Western blot analysis, after gel separation the protein pattern was transferred onto polyvinylidenedifluoride membranes and blotted membranes were blocked overnight in PBS containing 3% milk powder. For immunodetection, membranes were incubated with an anti-Xv_EstE rabbit antiserum diluted 1:1,000 in Tris-buffered saline (TBS) with 3% BSA for 3 h. Immunoblots were rinsed three times with PBS and then the secondary antibody, goat-anti rabbit IgG coupled with horseradish peroxidase (Sigma, Deisenhofen, Germany) diluted 1:400 in TBS with 3% BSA, was added for 1 h. After three washes with PBS, a color reaction was obtained by incubating the filter with a solution containing 2 ml of 4-chlor-1-naphthol (3 mg ethanol ml−1), 25 ml of PBS and 10 μl of H2O2 (30%).

Esterase activity staining

Native PA gels were incubated in a solution of 0.1 M sodium phosphate (pH 7.0), 2 ml of substrate stock solution [α-naphthylacetate 1% (w/v) in acetone] and 500 µl of fast blue B stock solution [2% (w/v) in H2O] and revealed esterase activity by the formation of dark red bands. After separation on SDS-PAGE, the gels were renaturated by incubation in buffer containing 50 mM NaH2PO4, 12.5 mM citric acid (pH 6.3) and 25% (w/v) isopropanol prior to activity staining.

When using the fluorogenic substrate, 1-hydroxy-pyrene-3,6,8-trisulfonic acid (HPTS), the gel was incubated in a solution of 0.1 M sodium phosphate buffer (pH 7.0) and 50 μl of the substrate stock solution (2 mg ml−1 in dimethylsulfoxide; DMSO) for about 5 min. Esterase activity was detected by green fluorescent bands visualized by UV excitation, using a Chroma 43B transilluminator (302 nm).

Rapid determination of esterase activity in bacterial clones was determined by the colony filter assay described by Schlacher et al. (1998), using routinely α-naphthylacetate as substrate. Esterase activity was detected by the development of a purple stain in the bacteria within 5 min.

Spectrophotometric assay

Esterase activity was determined spectrophotometrically at 405 nm using 4 mM o- and p-nitrophenyl fatty acid esters (Sigma; 100-fold concentrated stock solution dissolved in DMSO) in 100 mM Tris/HCl buffer, pH 7.0. Specific absorption coefficients (at 405 nm) of 2.4166 ml mol−1 cm−1 and 9.5946 ml mol−1 cm−1 for ortho-nitrophenol and para-nitrophenol, respectively, were used to calculate the activity. One unit of enzyme activity was defined as the amount of enzyme catalyzing the formation of 1 μmol o-, p-nitrophenol min−1 at room temperature.

Thin layer chromatographic analysis

Crude lysate of overexpressed cultures (0.1 ml) was suspended in 0.5 ml of reaction buffer (0.1 M NaH2PO4, pH 7.3). Then, substrate was added (20 µl or 20 mg for liquid or solid substrates, respectively) and the mixture was agitated on a rotary shaker (150 rpm) at ambient temperature. Samples were withdrawn after 2, 4 and 20 h and analyzed by thin layer chromatography (TLC), using Merck silica gel 60 F254. Eluent and visualization methods (spraying with vanillin/H2SO4 or molybdate and heat treatment and/or UV detection) depended on the nature of the substrate. Conversion was semi-quantitatively estimated from spot sizes.

Chiral gas chromatographic analysis

For chiral GC analysis, enzyme reactions were performed as for TLC analysis, but at a 10-fold larger scale. The mixtures were agitated on a rotary shaker (150 rpm) at room temperature. Samples (0.5 ml) were taken after 2, 5 and about 20 h and the reaction was stopped by the addition of 0.5 ml of CH2Cl2. After extraction, centrifugation and separation of the aqueous layer, the organic phase was dried over sodium sulfate and a further 0.5 ml of CH2Cl2 were added. Following filtration with HPLC membrane filters (Nylon, 3 mm diam., 0.2-µm pores; Roth), an aliquot of 1 µl was injected into a Shimadzu GC-14A GC/FID system using FS-Lipodex E and Chrompack Chirasil Dex columns. The reactions were monitored by the liberation of the resulting alcohol or acid.

In some cases, samples needed a special work-up procedure and derivatization was required before injection. In the case of substrate 1-benzylpyrrolidin-3-yl acetate, 300 µl of the aqueous buffer solution were applied to a ChemElut extraction column cartridge (Varian) and eluted with 3 ml of CH2Cl2. Hydrolysis reactions with tetrahydrofuran-3-acetate were stopped with 0.5 ml of ethyl acetate, extracted and centrifuged and the organic layer was separated and dried over anhydrous MgSO4 (Sigma). As derivatization reagent, 5 µl of butyric acid anhydride (Fluka) and some crystals of 4-(dimethylamino)-pyridine (Merck–Schuchardt) were used. The samples were then extracted with saturated NaHCO3 solution to separate the residues from derivatization. Before injection, drying over MgSO4 was required.

Isoelectric focusing

Analytical isoelectric focusing was performed with the pellet fraction of overexpression cultures, using a LKB Multiphor II unit with ready-to-use PA gradient gels (Ampholine 2.2% PAGE plate, pH 3.5–9.5, broad pI calibration kit standards) as recommended by Pharmacia (Upsalla, Sweden). Protein bands were stained with Coomassie brilliant blue R-250, according to the instructions of the ready-to-use gels. The pI of the esterase was calculated from a plot of pI vs mobility, using standard proteins in the pI range 4.6–9.6 and the Exel gel SDS buffer strips from Pharmacia.

Influence of temperature

The temperature optimum was determined by measuring the specific activity at various temperatures over 25–60 °C by the standard spectrophotometrical assay, using the pellet fraction of Xv_EstE overexpression cultures and p-nitrophenyl propionate as substrate.

Temperature stability was determined by long-term incubation of pellet fractions in 100 mM Tris/HCl buffer, pH 7.0, at 25, 30, 37, 45, 50 and 55 °C. Samples were taken at timed intervals and the residual activity was spectrophotometrically determined, as described above.

Results

Sequence analysis

Sequencing of the 2.8 kb insert in pGSX9Δ0 (GenBank accession number AF536208) revealed an ORF of 1,818 bp, encoding a polypeptide of 606 amino acid residues. The average high G+C content (69%) of the esterase-encoding sequence (designated as Xv_estE) was typical for Xanthomonas genes. Sequence analysis revealed a potential ATG start codon at nucleotide position 867, proceeded by a possible Shine Dalgarno sequence. A potential signal peptide of 25 amino acids indicated a secretory protein.

A BLAST (Altschul et al. 1990; Gish and States 1993) analysis of the encoded polypeptide, which was designated Xv_EstE, revealed homologies to various esterases, lipases and phospholipases, all belonging to the GDSL family of hydrolases (Upton and Buckley 1995; Arpigny and Jaeger 1999). The highest sequence similarity was found to a putative lipase/esterase from the plant pathogen Xylella fastidiosa (Silvestri 2000; a bacterium belonging to the Xanthomonas group), a phospholipase B from Moraxella bovis [an outer membrane esterase from Pseudomonas aeruginosa (Wilhelm et al. 1999) which is identical with the putative autotransporter protein PapA of this organism (GenBank accession number AJ277638)], an autotransporting lipolytic enzyme from Pseudomonas sp. HSM0414 (GenBank accession number AY055203), esterase EstA from P. aeruginosa [an outer membrane esterase of Salmonella typhimurium (apeE gene) which does not contain a close analogue in E. coli (Carinato et al. 1998)] and a lipase from Photorhabdus luminescens (Wang and Dowds 1993). The esterase EstA from Burgholderia gladioli (Schwab et al. 1992; Klingsbichel 1996), which is significantly smaller, shows about 38% identity with the N-terminal 370 amino acids of Xv_estE. The Bg_EstA esterase was demonstrated to be located at the outer membrane by a lipid anchor positioned at the N-terminal cysteine of the mature protein (Klingsbichel 1996).

Conserved sequence blocks, as typically found with GDSL hydrolases (Upton and Buckley 1995), were found in Xv_EstE and from there the elements of a potential catalytic triad, Ser-38, Asp-142 and His-285, could be assigned (Fig. 1).

Fig. 1.
figure 1

Multiple alignment of Xv_EstE with bacterial esterases and lipases classified in the autotransporter family. The putative signal sequence of Xv_EstE is underlined. The conserved blocks around the catalytic triad residues (in bold) are boxed, shaded boxes indicate putative amphipathic β-strands and the two fully conserved amino acids from the ATF are marked with an asterisk. St_ApeE Esterase from Salmonella typhimurium (Carinato et al. 1998), Pl_Lip1 lipase from Photorhabdus luminescens (Wang and Dowds 1993), Xv_EstE esterase from Xanthomonas vesicatoria (this work), Pa_EstA esterase from Pseudomonas aeruginosa (Wilhelm et al. 1999)

Full size image

A RGD tripeptide motif (amino acids 346–348), identified by Prosite scanning, is present in the region between the N-terminal part, representing the catalytic domain, and the C-terminal β-barrel domain. This motif belongs to the 1-X-3 type of motifs which are involved in inter-protein interactions (Verevka and Miroshnychenko 2001).

The C-terminal part of Xv_EstE shows putative amphipathic β-strands (Fig. 1) similar to those described for the esterase EstA from P. aeruginosa (Wilhelm et al. 1999). To elucidate whether the C-terminus of Xv_EstE is capable of forming a β-barrel-like structure, structure prediction was performed using the FORTRAN program AMPHI, which was also used as basic tool in the first-time description of the autotransporter family of secreted proteins (Jose et al. 1995). As shown in Fig. 2, AMPHI predicts ten antiparallel, amphipathic β-sheets spanning amino acids 416–605 of the Xv_EstE precursor protein. This result could be confirmed by calculating regions of high surface probability, according to Emini et al. (1985). Such regions are assumed to be incompatible with transmembrane strands. By this strategy, two additional amphipathic β-strands, which were only weakly predicted by the AMPHI program, could be excluded (Fig. 2).

Fig. 2.
figure 2

Amphipathic β-sheet prediction by the computer program AMPHI (Jähnig 1990). A hydrophobicity plot of the C-terminal 220 amino acids of EstE from X. vesicatoria is given. Within each predicted amphipathic transmembrane β-sheet, at least one amino acid has a higher hydrophobicity value than 1.2. Arrows to the right indicate transmembrane regions running from the surface to the periplasmic side of the outer membrane; and arrows to the left indicate an opposite orientation of transmembrane β-sheets. Regions of high surface probability (SP) were calculated according to Emini et al. (1985), using the GCG sequence analysis package (Devereux et al. 1984). Amino acid numbering follows Fig. 1

Full size image

Biochemical and biocatalytic properties—basic features

Xv_EstE could be highly overexpressed in the E. coli BL21 host bearing the expression plasmid pMSEX9. The protein was found exclusively in the insoluble (pellet) fraction of cell lysates. Enzymatic activity could be restored after SDS denaturation prior to PAGE analysis, allowing detection by SDS-PAGE activity staining (Fig. 3). Thereby, a sharp, smaller esterase-active polypeptide band specific for the pMSEX9-overexpression plasmid could be detected in addition to the major esterase band (Fig. 3A, lane 1). As the RGD motif at position 346 may function as a signal for proteolytic processing, the size of this band in the range of about 40 kDa may provide an indication for a specific proteolytic event releasing the esterase catalytic moiety from the C-terminal β-barrel domain.

Fig. 3A, B.
figure 3

Analysis of expression of the esterase Xv_EstE in Escherichia coli BL21(DE3) by SDS-PAGE and activity staining. A Separation in a 12% polyacrylamide gel of proteins, using the crude lysate of a culture induced with 1 mM isopropylthiogalactoside (IPTG; (induction 3 h after inoculation, total culture time 10 h). Lane 1 Gel activity-stained with α-naphthyl acetate and with Coomassie blue, lane 2 gel stained with Coomassie blue only. B Separation in a 7.5% polyacrylamide gel of proteins from the pellet fraction of uninduced and induced cultures. Lanes 1, 2 E. coli BL21(DE3) containing expression vector pMS470Δ8 without Xv_estE insert uninduced and induced with 1 mM IPTG at 2 h after inoculation, respectively. Lanes 3, 4, 5, 6 E. coli BL21(DE3) containing Xv_EstE expression plasmid pMSEX9 induced with 1 mM IPTG at 7, 4 and 2 h after inoculation and non-induced, respectively. Total cultivation time from inoculation until harvesting was 9 h in all cases. The arrows indicate the positions of Xv_EstE protein and processed forms. Positions of molecular mass markers are given in kiloDaltons at the right

Full size image

When analyzing the protein on lower-concentration PA gels, it became evident that the esterase protein was nearly totally present in the unprocessed form upon induction of expression. However, when an uninduced culture was analyzed, processed esterase proteins could be found in small quantities (Fig. 3B, lane 6). Induction conditions were crucial for expression and best results could be obtained when induction was in the late exponential growth phase (Fig. 3B, lanes 3, 4). Under these conditions, specific esterase activity determined in lysates were in the range of 3.8 units mg−1 protein (measured with p-nitrophenyl acetate). Induction in the early exponential phase resulted in essentially no esterase production (Fig. 3B, lane 5).

Xv_EstE has a temperature optimum around 44 °C and shows quite good activity up to 60 °C (about 50% of maximum activity). It is characterized by high stability; and no significant loss of activity could be observed for 6 days at room temperature or for 48 h at 55 °C, providing a good basis for industrial applications.

The capability of the Xv_EstE enzyme to hydrolyze different triglycerides, nitrophenol-, naphthyl- and polar hydroxypyrene trisulfonic fatty acid esters (HPTS esters) was semi-quantitatively analyzed from halo formation or from the intensity of bands on activity-stained native PA gels. The results presented in Table 1 show a rather broad acceptance of short-to-medium chain-length substrates, except with the polar HPTS esters. Most interestingly, with respect to this group of substrates, this enzyme shows an inverse behavior; and a clear preference for medium-to-long chains was found. This fact seems to be characteristic of this class of enzyme, as a similar behavior was found with a further esterase of the GDSL family, Bg_EstA (Schlacher et al. 1997).

Table 1. Activity of the Xanthomonas vesicatoria esterase, Xv_EstE, on various chromatogenic and fluorogenic substrates tested by colorimetric assay and by activity staining in native polyacrylamide gels. Activity against triglycerides was determined by halo formation on agar plates. Activity staining with β-naphthyl esters gave results similar to those with α-naphthyl esters. HPTS 1-Hydroxy-pyrene-3,6,8-trisulfonic acid, n.d. not determined, +++ strong activity, ++ active, + weak activity, – no activity
Full size table

Selectivity on chiral substrates

The Xv_EstE, catalyzing the hydrolysis of a set of substrates belonging to different structural classes and having the chiral center either in the fatty acid or the alcohol moiety, was in a first step analyzed by TLC. As shown in Table 2 Xv_EstE was most active (conversions up to 100%) in the case of oxiran-2-yl butyrate; and the other substrates showed moderate conversion rates.

Table 2. Thin layer chromatography screening results with Xv_EstE
Full size table

For the determination of enantioselectivity by chiral GC, only those substrates which showed conversions up to 50% (from TLC analysis) were chosen. In addition, 1-methylprop-2-ynyl acetate and 1-benzylpyrrolidin-3-yl acetate, for which no TLC analysis was possible, were employed. The results clearly showed no significant enantioselectivity with most substrates. Only for those substrates summarized in Table 3 was a higher level of enantioselectivity recognized. Most interestingly, the rather small molecule 1-methylprop-2-ynyl acetate showed good selectivity (93% ee after >20 h) for the remaining acetate. This is of high importance, as this chiral compound is of substantial commercial interest.

Table 3. Chiral GC screening results with Xv_EstE
Full size table

Subcellular localization of Xv_EstE expressed in E. coli

Previous results indicated this esterase, Xv_EstE, might be associated with the cell envelope. To determine the subcellular localization of the enzyme, differential fractionation experiments were performed with E. coli cells containing the plasmid pMSEX9. Upon induction with IPTG, a protein band with an apparent molecular mass of 61 kDa appeared in whole cell lysates (Fig. 3A, lanes 1, 2) and in the outer membrane fraction (Fig. 4A, lane 4). It was not detectable in the cytoplasmic and periplasmic fractions or in E. coli BL21 whole cell lysates without plasmid (data not shown). Uninduced E. coli cells containing the plasmid only exhibited a faint band in this molecular mass range (Fig.4A, lanes 2, 3). These results strongly indicated that the protein co-localizing with the outer membrane fraction was esterase Xv_EstE. This was verified by Western blot experiments, using Xv_EstE-specific rabbit antiserum (Fig. 4B) for detection. As the high over-expression obtained after 1 h induction with IPTG did not result in cross-contamination of the cytoplasmic or the periplasmic fraction, esterase Xv_EstE appeared to be an integral outer-membrane protein when expressed in E. coli. Extending the time of induction with IPTG up to 6 h resulted in contamination of all subcellular fractions with Xv_EstE protein (data not shown). This can be explained either by the incapacity of the outer membrane to pick up such large amounts of recombinant protein or by a general disturbance of the cell envelope due to the high expression of foreign protein.

Fig. 4A, B.
figure 4

SDS-PAGE and Western blot analysis of outer membrane preparations from E. coli BL21(DE3) pMSEX9, expressing Xv_EstE. Cultures were left uninduced (ind. –) or protein expression was induced with 1 mM IPTG for 1 h (ind. +) before outer membranes were prepared. Preparations were (+) or were not (–) digested with trypsin after outer membrane isolation. The molecular mass of marker proteins (M 2 , prestained marker) is indicated in kiloDaltons. Xv_EstE and the natural outer membrane proteins OmpF/C and OmpA are marked. For Western blotting, the samples are diluted 1:30

Full size image

Orientation of recombinant Xv_EstE in the E. coli cell envelope

External protease accessibility is a common tool to verify the surface translocation of autotransporters or other secreted proteins (Jose et al. 2001). Xv_EstE, as most proteins, contains a large number of predictable trypsin cleavage sites, 17 of which are located in the esterase domain. Given that they are accessible, the addition of trypsin should result in a rapid degradation of Xv_EstE. However, external addition of trypsin to physically intact E. coli cells harboring plasmid pMSEX9 did not result in a reduction, neither in size nor in the amount of Xv_EstE localized in the outer membrane (data not shown). This was the first indication that Xv_EstE is not directed to the surface but to the periplasm of the bacterial cell and that it is protected from external protease access by the outer membrane. To exclude that the lack of trypsin degradation is due to structural constraints of Xv_EstE itself, outer membranes were first isolated from E. coli pMSEX9 cells expressing Xv_EstE and then subjected to trypsin digestion. This also enabled trypsin access from the periplasmic side of the outer membrane. As seen in Fig. 4 (lanes 4, 5), trypsin addition to isolated outer membranes resulted in a rapid and almost complete degradation of Xv_EstE. This indicates that the trypsin cleavage sites are generally accessible in Xv_EstE but are protected by the outer membrane in intact cells.

Discussion

The novel esterase cloned from X. vesicatoria, Xv_EstE, was identified by sequence homology to belong to the GDSL family of hydrolases. Overexpression of this enzyme could be achieved at high levels in E. coli and it was found that the enzyme was present only in insoluble cell fractions. It could be clearly demonstrated that Xv_EstE was localized in the outer membrane, indicating that the present signal sequence efficiently directed the Xv_EstE protein into the secretory pathway. This esterase, in analogy to EstA from P. aeruginosa (Wilhelm et al. 1999), which was recently described to belong to the autotransporter family of secreted proteins, consists of two domains. The catalytic domain can be assigned to the N-terminal part and the C-terminal part of Xv_EstE shows putative amphipathic β-strands similar to those described for the esterase EstA from P. aeruginosa, indicating the potential to form a β-barrel structure. Most, if not all, integral outer membrane proteins in Gram-negative bacteria appear to be so-called β-barrels (Koebnik et al. 2000). Such structures may function as membrane anchors (Pautsch and Schulz 1998) or as autotransporter systems (Henderson et al. 1998).

The autotransporter proteins are synthesized as polyprotein precursors that consist of a signal peptide, one or several passenger domains and a transporter domain at the very C-terminus. This transporter domain folds as a β-barrel within the outer membrane and enables surface translocation of the N-terminal-attached passenger (Jose et al. 1995). In several examples, the passenger was shown to be a protease and in some cases the passenger remained anchored within the outer membrane by the transporter domain, the β-barrel. Summarizing the structure prediction data (Fig. 2), we conclude that the C-terminus of Xv_EstE might be capable of forming a membrane-embedded β-barrel consisting of ten amphipathic, antiparallel β-sheets. In analogy to all outer membrane β-barrel proteins known so far, the last transmembrane β-sheet of Xv_EstE is supposed to run from the cell surface to the C-terminal end, directed to the periplasm. The barrel would be closed by its interaction with the antiparallel first β-sheet, consequently running from the periplasm to the cell surface. This suggests a periplasmic location of the N-terminally attached esterase domain (Fig. 5), at least at an early stage, unless it was translocated by a specific mechanism. Both the absence of accessibility to externally added trypsin and the differential cell fractionation experiments supported this hypothesis. Moreover, the Xv_EstE protein contains an aromatic amino acid (phenylalanine) at its very C-terminus, which is a typical feature of the porins, the general example for outer membrane β-barrels (Schulz 2000). These proteins form diffusion pores for small hydrophilic molecules with no particular specificity.

Fig. 5.
figure 5

Schematic model of the β-barrel-mediated outer membrane anchoring of Xv_EstE from X. vesicatoria. The C-terminal part of Xv_EstE folds as a β-barrel consisting of ten amphipathic β-sheets in the outer membrane. The esterase domain is not translocated to the surface, but remains directed to periplasm, anchored within the outer membrane by the β-barrel. IM Inner membrane, OM outer membrane, PP periplasm

Full size image

Taking these results together, we conclude that Xv_EstE from X. vesicatoria possesses a C-terminal domain that is able to form a β-barrel within the outer membrane. This β-barrel consists most likely of ten amphipathic, antiparallel β-sheets; and it does not serve for the surface translocation but for the outer membrane anchoring of the N-terminal-attached catalytic esterase domain. Therefore, Xv_EstE is unlikely to be an autotransporter protein, but rather belongs to a new family of outer membrane esterases (or enzymes) that are fixed in the outer membrane by a β-barrel structure and possess a catalytic domain directed towards the periplasm (Fig. 5). There are several examples of β-barrel proteins in the outer membrane of Gram-negative bacteria that have a C-terminal periplasmic extension, e.g. OmpA from E. coli or Hsr from Helicobacter mustelae. Xv_EstE from X. vesicatoria, as described here, is the first example of an outer membrane β-barrel protein with a N-terminal periplasmic extension. The physiological function of this periplasmic esterase activity, however, still needs further elucidation. One might speculate that the membrane-spanning barrel structure may likely function as a transporter pore that directs specific substrates to the esterase.

Concerning enzymatic properties, esterase Xv_EstE was demonstrated to behave in vitro as a typical carboxyl esterase, showing a clear preference for short-chain fatty acid esters. Triglycerides of long-chain fatty acid esters which represent typical lipase substrates were not hydrolyzed. A very interesting behavior could be detected with HPTS esters, which are highly polar at the alcohol moiety due to the three sulfonic acid groups present. With this class of substrates, no reactivity was seen with short-chain fatty acid moieties whereas, with long-chain fatty acid moieties, good activity was detected. It seems that this feature is typical for esterases belonging to the GDSL family, as a very similar behavior could be found with esterase EstA from Burkholderia gladioli (Schlacher et al. 1997), which also belongs to this family. In contrast, esterase EstC from B. gladioli (Reiter et al. 2000), which belongs to the GXSXG active site motif esterases, can hydrolyze only short-chain HPTS esters. A possible explanation could be that the long-chain fatty acid HPTS esters are able to form micellar structures in aqueous solutions and Xv_EstE (being a typical membrane protein) can well attack such structures. This makes this enzyme an interesting candidate for biocatalytic applications with polar ester substrates.

Analysis of enantioselectivity for a series of typical esterase substrates revealed that Xv_EstE reflected only limited enantioselectivity. However, these studies were restricted to classical non-polar ester substrates and more detailed studies employing also polar esters have to be performed in future work.

In conclusion, Xv_EstE was demonstrated to possess interesting features with respect to both biological and biocatalytic functions. Its localization in the outer membrane with the catalytic domain oriented into the periplasm and its ability to hydrolyze polar esters possibly organized in a micellar organization suggest an interesting role for this esterase in vivo and in vitro.