JOURNAL OF BACTERIOLOGY,0021-9193/97/$04.0010

June 1997, p. 3443–3450 Vol. 179, No. 11

Copyright © 1997, American Society for Microbiology

Topology of the Outer Membrane Phospholipase A ofSalmonella typhimurium

KARIN B. MERCK,1* HANS DE COCK,1 HUBERTUS M. VERHEIJ,2 AND JAN TOMMASSEN1

Department of Molecular Cell Biology, Institute of Biomembranes,1 and Department of Enzymology and ProteinEngineering, Centre for Biomembranes and Lipid Enzymology,2 Utrecht University, 3584 CH Utrecht, The Netherlands

Received 23 December 1996/Accepted 13 March 1997

The outer membrane phospholipase A (OMPLA) of Enterobacteriaceae has been proposed to span themembrane 14 times as antiparallel amphipathic b-strands, thereby exposing seven loops to the cell surface. Wehave employed the epitope insertion method to probe the topology of OMPLA of Salmonella typhimurium. First,missense mutations were introduced at various positions in the pldA gene, encoding OMPLA, to create uniqueBamHI sites. These BamHI sites were subsequently used to insert linkers, encoding a 16-amino-acid B-cellepitope. Proper assembly of all mutant proteins was revealed by their heat modifiability in sodium dodecylsulfate-polyacrylamide gel electrophoresis. The accessibility of the inserted epitopes was assessed. Immuno-fluorescence analysis of intact cells with antibodies against the inserted epitope showed that three of sevenpredicted loops are indeed cell surface exposed. Trypsin accessibility experiments verified the cell surfaceexposure of two additional loops and provided support for the proposed periplasmic localization of threepredicted turns. For two other predicted exposed loops, the results were not conclusive. These results supportto a large extent the proposed topology model of OMPLA. Furthermore, the observation that the substitutionsGlu66Pro and Glu247Gly virtually abolished enzymatic activity indicates that these residues might play amajor role in catalysis.

Most bacterial outer membrane proteins are involved intransport processes, either as pore-forming proteins or as re-ceptors. The outer membrane phospholipase A (OMPLA)(also designated PldA protein, after its structural gene pldA) isone of the few enzymes present in this membrane. This Ca21-dependent enzyme has hydrolytic activity towards phospholip-ids and lipids (23, 31, 37). The pldA gene is widely distributedamong members of the family Enterobacteriaceae, and the pri-mary structure of this protein is highly conserved (8). The pldAgenes of Escherichia coli and Salmonella typhimurium encode30-kDa proteins of 269 amino acid residues, preceded by signalsequences of 20 residues (8, 22). The function of OMPLA isunknown. It has been shown that E. coli OMPLA is requiredfor efficient secretion of bacteriocins (28, 32), but it is unlikelythat this is the primary function of the protein. Since OMPLAis an outer membrane protein, it is embedded in its substrate.However, enzymatic activity is detected only when the integrityof the outer membrane is affected, e.g., by heat shock (18) orby phage-induced lysis (16). The mechanism of enzymatic ac-tivation of OMPLA, however, has not yet been elucidated. Thestructures and functions of many water-soluble phospholipasesare known in detail, but OMPLA has no sequence homologywith any of these well-characterized phospholipases.

To understand the catalytic and activation mechanisms ofthis intriguing enzyme, knowledge of its structure is needed.An E. coli OMPLA derivative has been overproduced, puri-fied, and refolded with good yields (19). Although the refoldedprotein did crystallize (5), its structure has not yet been re-solved by X-ray diffraction analysis. There are strong indica-tions that OMPLA belongs to the same class of b-sheet struc-ture integral outer membrane proteins as the porins. Thisassumption is based on the facts that OMPLA lacks hydropho-

bic sequences long enough to span the lipid bilayer and thatcircular dichroism measurements showed that its secondarystructure consists mainly of b-structure (19). A model for thetopology of OMPLA that was based on the prediction of turnsand loops and calculations of hydrophobicity and hydrophobicmoment has been proposed (8) (Fig. 1). In this model, OM-PLA traverses the outer membrane 14 times as antiparallelamphipathic b-strands, thereby exposing seven hydrophilicloops to the cell surface. On the periplasmic side, the b-strandsare connected by short turns.

In the absence of a crystal structure, valuable structuralinformation can be obtained by genetic approaches. A partic-ularly useful method is the insertion of a foreign epitope atvarious sites in the protein. The subsequent determination ofthe accessibility of the inserted epitope in intact and in per-meabilized cells will yield information on the topology of theprotein in the membrane. This approach, which was originallydeveloped in the study of outer membrane protein LamB (13)and successfully applied for other outer membrane proteins,such as PhoE (2), was used in the present study to determinethe topology of OMPLA.

MATERIALS AND METHODS

Bacterial strains and growth conditions. All strains used are E. coli K-12derivatives. Strain DH5a (20) was used as the recipient in cloning experiments.Strain CE1265 (26) has the phoE gene deleted, and it does not produce OmpFand OmpC proteins due to an ompR mutation. Additionally, the strain has a recAmutation and contains a phoR mutation, resulting in constitutive expression ofthe pho regulon. Bacteria were grown at 37°C in L broth (39) supplemented,when required for plasmid maintenance, with chloramphenicol (37 mg/ml) orwith ampicillin (100 mg/ml).

Plasmids and DNA techniques. Plasmid pST103 is a pUC19 derivative andcontains the pldA gene of S. typhimurium (8). Plasmid ppL302 is an expressionconstruct in which the E. coli pldA gene is cloned behind the phoE promoter (19).Plasmid pRM37 (a gift from R. Janssen) contains a phoE allele, encoding amutant PhoE protein with the B-cell epitope encompassing amino acid residues240 to 255 of membrane protein E2 of Semliki Forest virus (SFV) (35) insertedin loop 8. Other plasmids used were the cloning vector pACYC184 (12) andpUC4K (40), which carries a kanamycin resistance gene cassette. Plasmid DNA

* Corresponding author. Mailing address: Agrotechnological Re-search Institute ATO-DLO, P.O. Box 17, 6700 AA Wageningen, TheNetherlands. Phone: 31 317 475332. Fax: 31 317 475347. E-mail:k.b.merck@ato.dlo.nl.

3443

on April 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

was isolated by the alkaline lysis procedure (4), followed by anion-exchangechromatography on Qiagen columns (Diagen, Dusseldorf, Germany).

To clone the pldA gene of S. typhimurium under control of the phoE promoter,the part of the gene that encodes the mature protein was amplified by PCR withpST103 as the template and the following two oligonucleotides as primers: K005(59-CGGCgcATGCACAAGAAGC-39), with nonhybridizing nucleotides (inlowercase) to introduce an SphI site (underlined), and K006 (59-AAAgatCtcccgggAAAACACAATCCAA-39), with nonhybridizing nucleotides (in lowercase) tointroduce a BglII site (underlined). Primer K005 hybridizes with nucleotidescorresponding to amino acids 24 to 13 of OMPLA, whereas primer K006hybridizes with the nucleotides directly behind the stop codon. PCR was carriedout essentially as described previously (8). After digestion with SphI and BglII,the PCR product was used to replace E. coli pldA on pPL302. In the resultingplasmid, pPLST-wt, the S. typhimurium pldA gene is under control of the phoEpromoter.

To facilitate the insertion of a linker encoding a foreign epitope, uniqueBamHI sites were introduced in pPLST-wt at various positions in the pldA geneby a four-primer PCR method (29). The four primers used were a primerhybridizing to the phoE promoter region (59-TAATCTGTAAGATATCTTTAAC-39), primers K005 and K006 (described above), and a mismatch primer tointroduce the desired BamHI site. The PCR products were digested with SphIand BglII and exchanged for the SphI-BglII wild-type fragment of pPLST-wt.Plasmids containing the desired BamHI site were named after the position of themutation in the topology model (Fig. 1 and Table 1); e.g., pPLST-L1 refers to aplasmid with a BamHI site in the DNA corresponding to loop L1 in the protein.

Linkers encoding the B-cell epitope corresponding to amino acid residues 240to 255 of the SFV E2 protein were cloned into the created BamHI sites. Tomaintain the correct reading frame of the pldA gene, either linker A or B (Fig.2) was chosen. The SFV epitope linkers were unphosphorylated and added in a1,000-fold molar excess to BamHI-digested plasmid DNA. The presence andorientation of the linkers in recombinant plasmids obtained were checked bymeans of PCR, using the relevant mutagenic primer and primers hybridizing tothe 59 and 39 ends of the OMPLA-encoding DNA. In the nomenclature used,

plasmids containing the epitope linker received the extension E (Table 1). Cor-responding proteins are named after their plasmids; e.g., PLST-T1E is the pro-tein encoded by plasmid pPLST-T1E.

Mutant plasmid pPLST-T6E was obtained in an alternative way. Briefly, akanamycin resistance cassette was isolated by digesting pUC4K with EcoRI.After the EcoRI overhangs were blunted with the Klenow fragment of DNApolymerase, the kanamycin resistance cassette was inserted in the mung beannuclease-treated EcoNI site of the S. typhimurium pldA gene on pPLST-wt. Thekanamycin resistance cassette was then removed by BamHI digestion, and SFVepitope linker A (Fig. 2) was inserted. The sequence surrounding the insertion isgiven in Table 1.

Constructs were sequenced on double-stranded template DNA by the dideoxychain termination method (33) with the T7 DNA polymerase deaza sequencingkit (Pharmacia LKB Biotechnology). DNA-modifying enzymes were obtainedfrom either Pharmacia LKB Biotechnology or New England Biolabs.

Isolation of crude cell envelope fractions. Cells from 50-ml overnight cultureswere lysed by ultrasonication at 0°C in 5 ml of 50 mM Tris-HCl–2 mM EDTA(pH 8.5). After removal of unlysed cells by centrifugation (1,000 3 g, 20 min), thecrude membrane fraction was pelleted at 170,000 3 g for 15 min. The pellet wasdissolved in 300 ml of 0.225% Triton X-100–2 mM Tris-HCl.

Protein analysis. Proteins were separated by sodium dodecyl sulfate-poly-acrylamide gel electrophoresis (SDS-PAGE) (27). Gels were stained with Coo-massie brilliant blue R250. To evaluate heat modifiability of mutant and wild-type OMPLAs, crude cell envelope fractions were either boiled or incubated atroom temperature in sample buffer prior to SDS-PAGE.

The polyclonal mouse anti-OMPLA serum used for Western blot analysis wasraised against denatured OMPLA (8). The specificity of the antiserum wasincreased by affinity purification (11). Native OMPLA was heat denatured withinthe gels prior to electroblotting onto nitrocellulose filters as described previously(8). Blots were developed by using alkaline phosphatase-conjugated goat anti-mouse immunoglobulins.

Protein concentrations were determined as described by Bradford (6).

FIG. 1. Topology model for S. typhimurium OMPLA, based on the previously proposed model for E. coli OMPLA (8). The amino acids that were replaced whilecreating BamHI sites for subsequent insertion mutagenesis are shown in boldface. The numbers of the surface-exposed loops and periplasmic turns are indicated atthe top and the bottom, respectively.

3444 MERCK ET AL. J. BACTERIOL.

on April 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

Trypsin accessibility experiments. Cells of an overnight culture were pelletedand resuspended in ice-cold 100 mM Tris-HCl (pH 8.0)–0.25 M sucrose supple-mented with either 10 mM MgCl2 (buffer A) or 5 mM EDTA (buffer B). A100-fold-concentrated stock solution of bovine pancreas trypsin (Sigma) in 100mM Tris-HCl (pH 8.0)–10 mM MgCl2 was added to a final concentration of 100mg/ml, and the suspension was incubated at 37°C for 1 h. Subsequently, trypsinwas inhibited by the addition of 1 mM phenylmethylsulfonyl fluoride and 500 mgof hen egg trypsin inhibitor per ml. Cells were washed twice with either buffer Aor buffer B supplemented with trypsin inhibitor (500 mg/ml), pelleted, dissolvedin SDS-PAGE sample buffer, and incubated for 10 min at 100°C. Proteins wereseparated by SDS-PAGE and analyzed by Western blotting.

Phospholipase activity assay. OMPLA activity in crude cell envelope fractionswas determined spectrophotometrically (17) with the substrate 3-hexadecanoyl-thiopropyl-1-phosphocholine (1). The measurements were carried out at least intriplicate.

Immunofluorescence microscopy. Cells were harvested and fixed overnight atroom temperature in 2% paraformaldehyde–0.5% glutaraldehyde in phosphate-buffered saline (PBS). Occasionally, cells were incubated for 15 min at roomtemperature with 100 mM EDTA in PBS prior to fixation, in order to removepart of the lipopolysaccharide (LPS) molecules, which might shield exposedepitopes of outer membrane proteins (41). Fixed cells were preincubated in PBG(0.5% bovine serum albumin and 0.09% fish gelatin in PBS). Cells were treatedfor 1 h at room temperature with mouse monoclonal antibodies (MAbs) directedeither against the SFV B-cell epitope (1:2,000) (provided by I. Fernandez) oragainst the periplasmic carboxy-terminal tail of OmpA (1:100) (provided by M.Kleerebezem). After being washed, the cells were incubated for 1 h with thesecond antibody, goat anti-mouse immunoglobulin G/M, labeled with fluoresceinisothiocyanate (1:2,000 in PBG). After three washes with PBG, the labeled cells

were immobilized on glass cover plates. Labeled cells were visualized in a fluo-rescence microscope (Leitz Orthoplan) by either fluorescence or phase-contrastmicroscopy.

RESULTS

Construction of mutant pldA genes. In pPLST-wt, which wasconstructed as described in Materials and Methods, the DNAfragment encoding the mature domain of S. typhimuriumOMPLA is present behind the phoE promoter and signal se-quence-encoding domain. This construct allows overproduc-tion of OMPLA in normal strains under phosphate-limitingconditions. To examine the proposed topology model ofOMPLA (Fig. 1), a linker encoding an epitope of SFV (Fig. 2)was inserted at various sites in the pldA gene. First, the uniqueEcoNI site in S. typhimurium pldA was used to constructpPLST-T6E, which encodes OMPLA with the epitope insertedin periplasmic turn T6 (see Materials and Methods). To obtainadditional insertions, unique BamHI sites were first introducedby site-specific mutagenesis, and linkers encoding the SFVepitope were subsequently inserted into these BamHI sites.The constructs obtained are summarized in Table 1.

Expression of mutant pldA genes. In E. coli CE1265, the phoregulon is constitutively expressed as a result of a phoR muta-tion. This strain was used for expression of the mutantOMPLA proteins. The mutant proteins containing only one ortwo amino acid substitutions as a consequence of the introduc-tion of the BamHI sites in the gene had electrophoretic mo-bilities comparable to that of wild-type OMPLA (Fig. 3). Ex-cept for PLST-L1, all of these mutant proteins were expressedalmost as well as wild-type OMPLA was. The drastically re-duced detection of PLST-L1 on the Western blots was notcaused by the loss of a major antigenic determinant due to themutation, since this protein was, in contrast to wild-type OMPLA,also hardly detectable on Coomassie blue-stained gels. As ex-pected, the insertion mutant proteins had slightly lower elec-trophoretic mobilities than the wild-type OMPLA. Their ex-pression levels were in most cases lower than that of the wild-type protein (Fig. 3). Only PLST-T1E, PLST-L2E, PLST-L4E,and PLST-L5E were expressed approximately equally as wellas wild-type OMPLA. The amino acid substitutions inPLST-L1 and the insertions in many mutant proteins mightinterfere with the efficiency of assembly of these proteins intothe outer membrane, resulting in degradation of unassembledproteins by periplasmic proteases.

Like several other outer membrane proteins (see, e.g., ref-erence 21), OMPLA is a heat-modifiable protein (8, 31); i.e.,the compactly folded, native protein migrates faster in SDS-PAGE than the heat-denatured protein does. To assesswhether the mutant OMPLAs were correctly folded, their heat

FIG. 2. Nucleotide sequences of the complementary oligonucleotides encoding the SFV B-cell epitope. Additional nucleotides required for insertion of the linkersinto BamHI sites and for adapting the linker to the correct reading frame are in boldface.

TABLE 1. Positions of mutations and linker insertions in OMPLAand identities of the substitutions and insertions

Plasmid Insertion site(amino acid)a

Amino acid(s) mutated orinsertedb

pPLST-T1 E7 VH38DPpPLST-T1E E7 8DP-SFV-DPpPLST-L1 Q24 25EH325DPpPLST-L1E Q24 25DP-SFV-DPpPLST-L2 D65 66E366PpPLST-L2E D65 66P-SFV-DPpPLST-L3 E105 106SS3106DPpPLST-L3E E105 106DP-SFV-DPpPLST-L4 P150 Silent mutationpPLST-L4E P150 149DP-SFV-DPpPLST-T5 G167 168N3168SpPLST-T5E G167 168S-SFV-RSpPLST-L5 D183 184DN3184EDpPLST-L5E D183 184EDP-SFV-DPpPLST-T6E L204 205EFPGS-SFV-RSGVNpPLST-L6 G219 220Y3220SpPLST-L6E G219 220S-SFV-RSpPLST-L7 G246 247E3247GpPLST-L7E G246 247GS-SFV-RS

a Last unchanged amino acid, after which mutations are introduced.b SFV indicates the insertion of amino acids 240 to 255 of the SFV E2 protein

(Fig. 2).

VOL. 179, 1997 TOPOLOGY OF OMPLA 3445

on April 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

modifiabilities were tested (Fig. 4). In all mutants, at least partof the total amount of the mutant protein expressed migratedfaster in the unboiled samples than the denatured forms oftheir proteins did. Therefore, a major proportion of each of themutant proteins seems to be folded correctly (Fig. 4).

Enzymatic activity of mutant OMPLAs. Phospholipase Aactivities of crude outer membrane fractions of derivatives ofstrain CE1265 carrying the various plasmids were determinedquantitatively in a chromogenic assay (Table 2). To be able tocompare the observed activities and the specific activities ofthe mutant proteins with those of the wild-type OMPLA, thevarying expression levels should be taken into consideration.Since the levels of overproduction of the mutant proteins werehighly variable, it is not possible to accurately calculate specificactivities. Therefore, enzyme activities are presented only qual-itatively.

The low enzymatic activities of the mutant proteins PLST-

L1, -L1E, -L3E, -T5E, -T6E, and -L7E (Table 2) can be fullyaccounted for by the low expression levels of these mutantproteins (Fig. 3; Table 2). Mutant proteins PLST-L2 and -L7were well expressed (about 50% of the expression of wild-typeOMPLA [Fig. 3]), whereas the enzymatic activities were neg-ligible (Table 2). Obviously, the specific activities of thesemutants are much lower than that of wild-type OMPLA. Ap-parently, the mutations Glu66Pro and Glu247Gly each drasti-cally reduce or even completely abolish enzymatic activity.

FIG. 3. Expression of mutant OMPLAs. Western immunoblot analysis of cell envelope samples of E. coli CE1265 expressing the various mutant OMPLAs is shown.pACYC184 indicates that the cells carried the vector plasmid, which does not encode OMPLA. An affinity-purified OMPLA-specific antiserum was used to detectOMPLAs with point mutations on the blot, whereas a MAb directed against the SFV E2 epitope was used to identify the insertion derivatives of OMPLA. Either 1or 10 mg of cell envelope protein was loaded on the gel as indicated below the lanes.

FIG. 4. Heat modifiability of mutant OMPLAs. Western immunoblotting ofcell envelope samples of E. coli CE1265 expressing mutant OMPLAs carryingmissense mutations (A) or insertions (B) is shown. Samples were either boiledfor 10 min or incubated at room temperature prior to SDS-PAGE, as indicatedbelow the lanes. An affinity-purified OMPLA-specific polyclonal antiserum wasused to detect the mutant OMPLAs on the Western immunoblots. Each lanecontained 1 to 10 mg of cell envelope protein. wt, wild type.

TABLE 2. OMPLA activities in crude outer membrane fractionsfrom E. coli CE1265 carrying various plasmids

Plasmid Expression levela Phospholipase activityb

Wild type and point mutantspACYC184 2 2pPLST-wt 11 11pPLST-T1 11 11pPLST-L1 6 2pPLST-L2 11 2pPLST-L3 11 11pPLST-L4 11 11pPLST-T5 11 11pPLST-L5 11 11pPLST-L6 11 1pPLST-L7 11 2

Insertion mutantspPLST-T1E 11 11pPLST-L1E 6 2pPLST-L2E 11 2pPLST-L3E 6 2pPLST-L4E 11 2pPLST-T5E 6 2pPLST-L5E 11 11pPLST-T6E 6 2pPLST-L6E 6 1pPLST-L7E 6 2

a Expression levels, in E. coli CE1265, of mutant OMPLAs compared with thatof wild-type OMPLA. Symbols: 11, expression comparable with that of wild-type OMPLA (20 to 100%); 6, very low expression (,5%); 2, no expression.

b Phospholipase activities of mutant OMPLAs compared with that of wild-typeOMPLA and background activity. One unit is defined as 1 nmol of 3-hexade-canoyl-thiopropyl-1-phosphocholine converted per min. Symbols: 11, phospho-lipase activity comparable to that of wild-type OMPLA (500 to 2,000 U mg ofmembrane protein21); 1, phospholipase activity considerably lower than that ofwild-type OMPLA (50 to 500 U mg of membrane protein21); 2, phospholipaseactivity comparable to background levels (,50 U mg of membrane protein21).

3446 MERCK ET AL. J. BACTERIOL.

on April 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

Consistently, negligible OMPLA activities were detected in thecells expressing PLST-L2E and -L7E, which have epitope in-sertions at the same sites, although in the case of PLST-L7Ethis reduction in activity may also be a consequence of the lowexpression level. Furthermore, PLST-L4E has no or extremelylow specific activity. This is expected, since the insert is presentclose to residues His-142, Ser-144, and Ser-152, which havebeen shown to be important for enzymatic activity (8, 9, 24).

Accessibility of the inserted epitopes. To probe the locationof the inserted epitope in the various OMPLA derivativesrelative to the outer membrane plane, two approaches werechosen, i.e., indirect immunofluorescence labeling and trypsinaccessibility.

(i) Indirect immunofluorescence labeling. The accessibilityof the SFV epitope in the insertion mutants in intact cells wasdetermined by immunocytochemical techniques. Cells express-ing the mutant proteins were incubated with MAbs directedagainst the SFV epitope and subsequently with fluoresceinisothiocyanate-labeled secondary antibodies as described inMaterials and Methods. The cells were examined by fluores-cence microscopy. As a positive control, CE1265 cells carrying

plasmid pRM37, which express a derivative of porin PhoE withthe SFV epitope inserted in the exposed loop L8, were exam-ined. As expected, most of these cells were found to be labeled(Fig. 5). Cells expressing the wild-type OMPLA from plasmidpPLST-wt were not labeled when the anti-SFV epitope MAbwas used (Fig. 5). In an additional negative control experiment,CE1265 cells harboring plasmid pPLST-L2E were incubatedwith a MAb directed against the periplasmic carboxy-terminaltail of OmpA. Cells treated with this antibody were not labeled(Fig. 5). Immunostaining with the anti-SFV epitope MAb re-vealed surface labeling of cells expressing the OMPLA inser-tion derivatives PLST-L2E, PLST-L3E, PLST-L4E, and PLST-L5E (Fig. 5). From these results it can be concluded that loopsL2, L3, L4, and L5 are present on the cell surface, as predictedin the topology model (Fig. 1). Typically, only a small propor-tion of the cells was labeled in the case of mutant PLST-L4E.This observation is reminiscent of the results of labeling ex-periments with antibodies directed against cell surface-exposedepitopes of outer membrane protein PhoE (41). In that casealso, only a small proportion of the cells could be labeled,which was demonstrated to be due to shielding of the epitopes

FIG. 5. Immunofluorescence assay of cells expressing OMPLA insertion derivatives. Fixed CE1265 cells containing the indicated plasmids were stained as describedin Materials and Methods. Either a MAb against the SFV E2 epitope (a-SFV) or an anti-OmpA-specific MAb (a-OmpA) was used for labeling.

VOL. 179, 1997 TOPOLOGY OF OMPLA 3447

on April 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

by the outer core region of the LPS (41). To determinewhether the limited labeling of the cells expressing PLST-L4Ewas due to steric hindrance by LPS, cells were incubated inEDTA (see Materials and Methods) before fixation and im-munolabeling in order to remove part of the LPS molecules.The proportion of the cells that could be labeled with theanti-SFV epitope MAb was increased drastically by the EDTAtreatment, although still not all cells were labeled (results notshown). The epitope in the periplasmic extension of OmpAremained inaccessible to the anti-OmpA MAb after EDTAextraction (results not shown). These results are consistentwith the hypothesis that the inserted epitope in PLST-L4E ispartially shielded in intact cells by LPS molecules.

The predicted surface exposure of loops L1, L6, and L7could not directly be confirmed, since surface labeling was notdetectable on cells expressing the PLST-L1E, -L6E, and -L7Emutant proteins (results not shown). The low expression levelsof these proteins, possibly in combination with steric hindranceby LPS, could be responsible for the absence of surface label-ing. In the case of cells expressing PLST-L6E, steric hindrancecould be confirmed, since some labeling was observed afterEDTA treatment of the cells (results not shown).

Mutant protein PLST-T1E was well expressed (Fig. 3), butcells expressing this protein were not fluorescently stained af-ter being labeled with the anti-SFV MAb (Fig. 5). This indi-cates that the epitope in PLST-T1E is localized at the periplas-mic side of the outer membrane. The cells expressing PLST-T5E and -T6E also were not fluorescently stained, even afterEDTA extraction (results not shown). However, since the ex-pression levels of these proteins were very low, it cannot def-initely be concluded that the epitopes are not exposed at thecell surface.

(ii) Trypsin accessibility experiments. In general, outermembrane proteins tend to be protease resistant (30) by virtueof their compact b-structures, with the linking surface loopsbeing tightly packed or folded within the b-barrel (15). SinceOMPLA is thought to have a b-barrel-like conformation likethe porins (8, 19, 42), we expected it to be protease resistant.Indeed, wild-type OMPLA was fully resistant to trypsin both inintact cells and in EDTA-permeabilized cells (Fig. 6). Since theSFV epitope contains four potential trypsin cleavage sites, it isa sensitive probe in trypsin accessibility experiments. Cellsexpressing the various insertion mutant proteins were incu-bated with trypsin in the presence of Mg21 to stabilize theouter membrane or in the presence of EDTA to permeabilizethe outer membrane (38). To verify the reliability of the ap-proach, cells carrying pRM37, which express a mutant PhoEwith the SFV epitope inserted in exposed loop L8, were alsotreated with trypsin. Whereas wild-type PhoE is not degradedby trypsin (38), the mutant protein was degraded by trypsin inintact cells, showing that the insert is surface exposed (Fig. 6).The degradation product detected had an Mr in accordancewith the position of the trypsin cleavage site. As observedpreviously (38), the OmpA protein remained intact after tryp-sin treatment of the cells in the presence of Mg21, whereas theperiplasmic tail of the protein was digested in EDTA-perme-abilized cells (data not shown). Trypsin treatment of intactcells producing mutant OMPLAs resulted in partial or com-plete degradation of the mutant proteins PLST-L1E, PLST-L2E, PLST-L3E, PLST-L5E, and PLST-L7E (Fig. 6). Thisindicates that the loops L1, L2, L3, L5, and L7 are indeedsurface-exposed loops, in agreement with the proposed topol-ogy model (Fig. 1). Only a portion of the mutant proteinsPLST-L1E, PLST-L3E, and PLST-L5E was degraded (Fig. 6),indicating that the trypsin cleavage sites were not fully acces-sible to trypsin. In the cases of PLST-L3E and PLST-L5E,

proteolytic cleavage was more complete in the presence ofEDTA, probably because the shielding of the protease-sensi-tive sites by LPS molecules was alleviated. Mutant proteinsPLST-L4E and PLST-L6E were accessible to trypsin only afterpermeabilization of the cells with EDTA. This could mean thatthe trypsin cleavage sites in these proteins are accessible onlyfrom the periplasmic side of the membrane. However, anotherpossibility is that these sites are at the external side of the outermembrane but completely shielded by the LPS molecules.

In the presence of Mg21, mutant proteins PLST-T1E, -T5E,and -T6E were almost completely resistant to trypsin. Afterpermeabilization of the cells with EDTA, mutant proteinPLST-T5E was completely degraded by trypsin and mutantproteins PLST-T1E and PLST-T6E were partially degraded bytrypsin. These results are consistent with the periplasmic loca-tion of the corresponding insertion sites in the proposed to-pology model (Fig. 1).

DISCUSSION

The goal of this study was to determine the topology ofOMPLA of S. typhimurium by the epitope insertion method.This approach has proven to have utility in the study of severalother outer membrane proteins, including PhoE, LamB, andFhuA (2, 14, 25). Elaborating on a previously proposed topol-ogy model for OMPLA (8), a B-cell epitope of SFV was in-troduced at various positions in the protein, and its accessibil-ity in the cells to MAbs and trypsin was determined. Theinterpretation of the results was not always straightforward,first because of the low expression levels of some of the mutantproteins and second because the accessibility of the insert insome cases appeared to be sterically hindered by other mem-brane compounds. Nevertheless, to a large extent, the resultsobtained confirm the proposed topology model. In the mutantproteins PLST-L2E, -L3E, and -L5E the inserted epitope wasaccessible in intact cells to the SFV-specific MAb, as revealedby immunofluorescence microscopy, and to trypsin. Therefore,it can definitely be concluded that the predicted loops L2, L3,and L5 are indeed surface exposed. Also, PLST-L1E (partially)and PLST-L7E (completely) were cleaved by trypsin in intactcells, showing that the SFV epitope was cell surface exposed.The failure to detect the epitope on the cell surface in theseconstructs by immunofluorescence is most probably due totheir very low expression level. Hence, we conclude that loopsL1 and L7 also are cell surface exposed.

Apparently conflicting results were obtained with mutantprotein PLST-L4E. The cell surface of a small proportion ofthe cells expressing PLST-L4E was immunolabeled, suggestingthat the inserted epitope is located at the surface. However,the fact that only few cells were immunolabeled cannot beexplained by low expression levels, since PLST-L4E was wellexpressed. Possibly, the putative loop L4 of OMPLA is not verywell exposed but folds back into the b-barrel like the constric-tion loop L3 in the porin (15). Alternatively, the core region ofthe LPS molecules sterically hindered the binding of the anti-bodies to the surface-located epitope, as reported before forsurface-exposed epitopes of other outer membrane proteins(3, 41). Indeed, extraction of the cells with EDTA, whichremoves a portion of the LPS molecules, drastically increasedthe labeling efficiency, although still far fewer than 50% of thecells were labeled. Nevertheless, the observation that aperiplasmic epitope of OmpA was not unmasked by the EDTAtreatment suggests that the detection of the SFV epitope in theEDTA-treated cells expressing PLST-L4E is indicative of thecell surface location of the epitope. The concerns with respectto the poor immunostaining of the cells expressing PLST-L4E

3448 MERCK ET AL. J. BACTERIOL.

on April 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

were reinforced by the finding that PLST-L4E was not acces-sible to trypsin in intact cells. In contrast, after permeabiliza-tion of the cells with EDTA, the epitope became fully acces-sible to the protease. Previously, such results would have beeninterpreted as evidence for the periplasmic location of theinserted epitope in PLST-L4E (see, e.g., reference 38). How-ever, the positive reaction, even though of only a limited num-ber of cells, in the immunofluorescence microscopy studiescannot be neglected. Possibly, the inserted epitope in PLST-L4E is located near the cell surface but its accessibility issterically hindered by LPS molecules, which can be removed inpart by EDTA extraction. If this interpretation is correct, theresults described here impose severe limitations on a proce-dure commonly used to map protease-sensitive sites in (recom-binant) outer membrane proteins relative to the membraneplane; i.e., whereas protease sensitivity of a site in intact cellsin the presence of Mg21 can still be considered evidence forthe cell surface exposure of this site, the unmasking of a siteafter EDTA treatment cannot be considered definite proof ofits periplasmic location. An alternative explanation for ourresults is that loop L4 has a dual location with respect to themembrane plane, being partly exposed and partly on theperiplasmic side. It is important in this respect to note thatloop L4 contains the active-site serine, Ser-144, and two otherresidues, His-142 and Ser-152, which are required for enzy-matic activity (8–10). Thus, one could envisage a model for theactivation of the enzyme that involves a movement of L4 rel-ative to the membrane through the b-barrel. The enzymewould be inactive when L4 is exposed to the surface betweenthe LPS molecules, but it would be active when L4 moves tothe periplasmic side of the membrane where the phospholipidsare.

For PLST-L6E, considerations similar to those for PLST-L4E are applicable. Cells expressing PLST-L6E were not im-munostained, and PLST-L6E turned out to be trypsin inacces-sible in intact cells. However, after EDTA treatment of thecells, the epitope became accessible both to the MAb in im-munofluorescence microscopy and to trypsin. The putativeloop L6 is very short (Fig. 1), which could explain the pooraccessibility of the inserted epitope even if the loop is indeedlocated at the cell surface and the necessity to remove part ofthe LPS molecules to make it accessible. However, like for L4,no definite conclusions can be drawn from the data obtainedfor the location of L6 with respect to the membrane plane.

Cells expressing PLST-T1E, -T5E, and -T6E were not la-beled by immunofluorescence microscopy, even after extrac-tion of part of the LPS molecules with EDTA. These results

suggest that the epitope in these constructs is not exposed tothe cell surface. However, it cannot be excluded that the lackof labeling in the cases of PLST-T5E and -T6E is due to thelow expression levels of these proteins. Furthermore, PLST-T1E, -T5E, and -T6E were accessible to trypsin only in EDTA-permeabilized cells. These results are at least consistent withthe presence of T1, T5, and T6 at the periplasmic side of theouter membrane, as proposed in the topology model (Fig. 1),but, considering the discussion above with respect to PLST-L4E, they cannot be considered definite proof of periplasmicexposure of T1, T5, and T6.

To identify residues that are involved in the catalytic actionof OMPLA, the enzymatic activities of the various point andinsertion mutants were measured. Strikingly, the mutant pro-teins PLST-L2 and PLST-L7 displayed no or hardly any enzy-matic activity (Table 2). Apparently, the mutations Glu66Proand Glu247Gly influence enzymatic activity dramatically. Thisis the more interesting since it has been suggested that thecatalytic center of the enzyme contains a classical Asp-His-Sertriad (see reference 24 and references therein). The involve-ment of Ser-144 as the active-site serine residue, as originallydetermined by chemical modification (24), was recently con-firmed by Brok et al. (10) in a site-directed mutagenesis study.Furthermore, the involvement of His-142 in the catalytic mech-anism has been demonstrated (9). So far, the acidic componentin the catalytic triad has not been identified. The recentlydetermined structures of acetylcholine esterase (36) and lipase(34) show that glutamates instead of aspartates also can beinvolved in the triad. For this reason, our finding that theGlu66Pro and Glu247Gly substitutions render the enzyme in-active is of special interest. Although Glu-66 is present in loopL2, Glu-247 is present in loop L7, and the identified active-siteresidues are present in loop L4, they might be in close prox-imity in the three-dimensional structure of the protein. More-over, Glu-66 and Glu-247 are perfectly conserved in all en-terobacterial species for which the pldA gene has beensequenced (7, 8). It is therefore very well possible that eitherGlu-66 or Glu-247 is the acidic component of the catalytictriad. An alternative possibility is that either one of theseresidues is involved in Ca21 binding. Additional site-directedmutagenesis experiments, in combination with enzyme purifi-cation, accurate determination of specific activities, and Ca21-binding experiments, are to be carried out to unambiguouslydetermine the exact role of each of these glutamates in cata-lytic activity.

FIG. 6. Trypsin accessibility of OMPLA insertion mutants in intact or permeabilized cells. Derivatives of E. coli CE1265 expressing the various insertion mutantOMPLAs were incubated with trypsin treatment in the presence or absence of EDTA as indicated. Total cell proteins were separated by SDS-PAGE and analyzed onWestern blots by using an affinity-purified OMPLA-specific polyclonal antiserum, except in the case of cells containing pRM37, where a MAb against PhoE was used.

VOL. 179, 1997 TOPOLOGY OF OMPLA 3449

on April 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

ACKNOWLEDGMENTS

We thank Ruud Cox for the preparation of 3-hexadecanoyl-thiopro-pyl-1-phosphocholine, Michiel Kleerebezem for giving us the OmpA-specific MAb, Riny Janssen for providing plasmid pRM37 and the SFVlinkers, Isabelle Fernandez for the SFV epitope-specific MAb, JobVercauteren for constructing plasmid pPLST-wt, Ronald Brok fordonating the plasmid pST103, and Bruno Humbel for introducing us tothe immunofluorescence microscopy techniques.

REFERENCES1. Aarsman, A. J., L. L. M. van Deenen, and H. van den Bosch. 1976. Studies

on lysophospholipases. VII. Synthesis of acylthioester analogs of lysolecithinand their use in a continuous spectrophotometric assay for lysophospho-lipases, a method with potential applicability to other lipolytic enzymes.Bioorg. Chem. 5:241–253.

2. Agterberg, M., H. Adriaanse, and J. Tommassen. 1987. Use of outer mem-brane protein PhoE as a carrier for the transport of a foreign antigenicdeterminant to the cell surface. Gene 59:145–150.

3. Bentley, A. T., and P. E. Klebba. 1988. Effect of lipopolysaccharide structureon reactivity of antiporin monoclonal antibodies with the bacterial cell sur-face. J. Bacteriol. 170:1063–1068.

4. Birnboim, H. C., and J. Doly. 1979. A rapid alkaline extraction procedure forscreening recombinant plasmid DNA. Nucleic Acids Res. 7:1513–1523.

5. Blaauw, M., N. Dekker, H. M. Verheij, K. H. Kalk, and B. W. Dijkstra. 1995.Crystallization and preliminary X-ray analysis of outer membrane phospho-lipase A from Escherichia coli. FEBS Lett. 373:10–12.

6. Bradford, M. M. 1976. A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dye bind-ing. Anal. Biochem. 72:248–254.

7. Brok, R. G. P. M. 1996. Personal communication.8. Brok, R. G. P. M., E. Brinkman, R. van Boxtel, A. C. A. P. A. Bekkers, H. M.

Verheij, and J. Tommassen. 1994. Molecular characterization of enterobac-terial pldA genes encoding outer membrane phospholipase A. J. Bacteriol.176:861–870.

9. Brok, R. G. P. M., N. Dekker, N. Gerrits, H. M. Verheij, and J. Tommassen.1996. A conserved histidine residue of Escherichia coli outer membranephospholipase A is important for activity. Eur. J. Biochem. 234:934–938.

10. Brok, R. G. P. M., I. Ubarretxena Belendia, N. Dekker, J. Tommassen, andH. M. Verheij. 1996. Escherichia coli outer membrane phospholipase A: roleof two serines in enzymatic activity. Biochemistry 35:7787–7793.

11. Burke, B., G. Griffiths, H. Reggio, D. Louvard, and G. Warren. 1982. Amonoclonal antibody against a 135-K Golgi membrane protein. EMBO J.1:1621–1628.

12. Chang, A. C. Y., and S. N. Cohen. 1978. Construction and characterization ofamplifiable multicopy DNA cloning vehicles derived from the P15A crypticminiplasmid. J. Bacteriol. 134:1141–1156.

13. Charbit, A., J. C. Boulain, A. Ryter, and M. Hofnung. 1986. Probing thetopology of a bacterial membrane protein by genetic insertion of a foreignepitope: expression at the cell surface. EMBO J. 5:3029–3037.

14. Charbit, A., J. Ronco, V. Michel, C. Werts, and M. Hofnung. 1991. Permis-sive sites and topology of an outer membrane protein with a reporterepitope. J. Bacteriol. 173:262–275.

15. Cowan, S. W., T. Schirmer, G. Rummel, M. Steiert, R. Ghosh, R. A. Pauptit,J. N. Jansonius, and J. P. Rosenbusch. 1992. Crystal structures explainfunctional properties of two E. coli porins. Nature 358:727–733.

16. Cronan, J. E., and D. L. Wulff. 1969. A role for phospholipid hydrolysis in thelysis of Escherichia coli infected with bacteriophage T4. Virology 38:241–246.

17. de Geus, P., N. H. Riegman, A. J. G. Horrevoets, H. M. Verheij, and G. H.de Haas. 1986. Overproduction and high-yield purification of phospholipaseA from the outer membrane of Escherichia coli K-12. Eur. J. Biochem.161:163–169.

18. de Geus, P., I. van Die, H. Bergmans, J. Tommassen, and G. de Haas. 1983.Molecular cloning of pldA, the structural gene for outer membrane phos-pholipase of E. coli K-12. Mol. Gen. Genet. 190:150–155.

19. Dekker, N., K. B. Merck, J. Tommassen, and H. M. Verheij. 1995. In vitrofolding of Escherichia coli outer-membrane phospholipase A. Eur. J. Bio-chem. 232:214–219.

20. Hanahan, D. 1983. Studies on transformation of Escherichia coli with plas-mids. J. Mol. Biol. 166:557–580.

21. Heller, K. B. 1978. Apparent molecular weights of a heat-modifiable proteinfrom the outer membrane of Escherichia coli in gels with different acrylamideconcentrations. J. Bacteriol. 134:1181–1183.

22. Homma, H., T. Kobayashi, N. Chiba, K. Karasawa, H. Mizushima, I. Kudo,K. Inoue, H. Ikeda, M. Sekiguchi, and S. Nojima. 1984. The DNA sequenceencoding pldA gene, the structural gene for detergent-resistant phospho-lipase A of E. coli. J. Biochem. 96:1655–1664.

23. Horrevoets, A. J. G., T. M. Hackeng, H. M. Verheij, R. Dijkman, and G. H.de Haas. 1989. Kinetic characterization of Escherichia coli outer membranephospholipase A using mixed detergent-lipid micelles. Biochemistry28:1139–1147.

24. Horrevoets, A. J. G., H. M. Verheij, and G. H. de Haas. 1991. Inactivation ofEscherichia coli outer-membrane phospholipase A by the affinity label hexa-decanesulfonyl fluoride. Evidence for an active-site serine. Eur. J. Biochem.198:247–253.

25. Koebnik, R., and V. Braun. 1993. Insertion derivatives containing segmentsof up to 16 amino acids identify surface- and periplasm-encoded regions ofthe FhuA outer membrane receptor of Escherichia coli K-12. J. Bacteriol.175:826–839.

26. Korteland, J., N. Overbeeke, P. de Graaff, P. Overduin, and B. Lugtenberg.1985. Role of the Arg158 residue of the outer membrane PhoE pore proteinof Escherichia coli K-12 in bacteriophage TC45 recognition and in channelcharacteristics. Eur. J. Biochem. 152:691–697.

27. Lugtenberg, B., J. Meijers, R. Peters, P. van der Hoek, and L. van Alphen.1975. Electrophoretic resolution of the ’major outer membrane protein‘ ofEscherichia coli K12 into four bands. FEBS Lett. 58:254–258.

28. Luirink, J., C. van der Sande, J. Tommassen, E. Veltkamp, F. K. de Graaf,and B. Oudega. 1986. Effects of divalent cations and of phospholipase Aactivity on excretion of cloacin DF13 and lysis of host cells. J. Gen. Micro-biol. 132:825–834.

29. Mikaelian, I., and A. Sergeant. 1992. A general and fast method to generatemultiple site-directed mutations. Nucleic Acids Res. 20:376.

30. Morona, R., J. Tommassen, and U. Henning. 1985. Demonstration of abacteriophage receptor site on the Escherichia coli K12 outer-membraneprotein OmpC by the use of a protease. Eur. J. Biochem. 150:161–169.

31. Nishijima, M., S. Nakaike, Y. Tamori, and S. Nojima. 1977. Detergent-resistant phospholipase A of Escherichia coli K-12. Purification and proper-ties. Eur. J. Biochem. 73:115–124.

32. Pugsley, A. P., and M. Schwartz. 1984. Colicin E2 release: lysis, leakage orsecretion? Possible role of phospholipase. EMBO J. 3:2393–2397.

33. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequencing withchain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74:5463–5467.

34. Schrag, J. D., Y. Li, S. Wu, and M. Cygler. 1991. Ser-His-Glu triad forms thecatalytic site of the lipase from Geotrichum candidum. Nature 351:761–764.

35. Snijders, A., B. J. Benaissa-Trouw, T. A. M. Oosterlaken, W. C. Puijk,W. P. A. Posthumus, R. H. Meloen, W. A. M. Boere, J. D. Oosting, C. A.Kraaijeveld, and H. Snippe. 1991. Identification of linear epitopes on Sem-liki Forest virus E2 membrane protein and their effectiveness as a syntheticpeptide vaccine. J. Gen. Virol. 72:557–565.

36. Sussman, J. L., M. Harel, F. Frolow, C. Oefner, A. Goldman, L. Toker, andI. Silman. 1991. Atomic structure of acetylcholinesterases from Torpedocalifornica: a prototypic acetylcholine-binding protein. Science 253:872–879.

37. Tamori, Y., M. Nishijima, and S. Nojima. 1979. Properties of purified de-tergent-resistant phospholipase A of Escherichia coli K-12. Inactivation, andprotection with detergents and phospholipids. J. Biochem. 86:1129–1138.

38. Tommassen, J., and B. Lugtenberg. 1984. Amino terminus of outer mem-brane PhoE protein: localization by use of a bla-phoE hybrid gene. J. Bac-teriol. 157:327–329.

39. Tommassen, J., H. van Tol, and B. Lugtenberg. 1983. The ultimate localiza-tion of an outer membrane protein of Escherichia coli K-12 is not determinedby the signal sequence. EMBO J. 2:1275–1279.

40. Vieira, J., and J. Messing. 1982. The pUC plasmids, an M13mp7-derivedsystem for insertion mutagenesis and sequencing with synthetic universalprimers. Gene 19:259–268.

41. Voorhout, W., J. Leunissen-Bijvelt, J. Leunissen, and A. Verkleij. 1986.Steric hindrance in immunolabelling. J. Microsc. 141:303–310.

42. Weiss, M. S., U. Abele, J. Weckesser, W. Welte, E. Schiltz, and G. E. Schulz.1991. Molecular architecture and electrostatic properties of a bacterial porin.Science 254:1627–1630.

3450 MERCK ET AL. J. BACTERIOL.

on April 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from