Autolysis of Lactococcus lactis Is Increased upon D-Alanine Depletion of Peptidoglycan and Lipoteichoic Acids

Mutations in the genes encoding enzymes responsible for theincorporation of D-Ala into the cell wall of Lactococcus lactisaffect autolysis. An L. lactis alanine racemase (alr) mutantis strictly dependent on an external supply of D-Ala to be ableto synthesize peptidoglycan and to incorporate D-Ala in thelipoteichoic acids (LTA). The mutant lyses rapidly when D-Alais removed at mid-exponential growth. AcmA, the major lactococcalautolysin, is partially involved in the increased lysis sincean alr acmA double mutant still lyses, albeit to a lesser extent.To investigate the role of D-Ala on LTA in the increased celllysis, a dltD mutant of L. lactis was investigated, since thismutant is only affected in the D-alanylation of LTA and notthe synthesis of peptidoglycan. Mutation of dltD results inincreased lysis, showing that D-alanylation of LTA also influencesautolysis. Since a dltD acmA double mutant does not lyse, thelysis of the dltD mutant is totally AcmA dependent. Zymographicanalysis shows that no degradation of AcmA takes place in thedltD mutant, whereas AcmA is degraded by the extracellular proteaseHtrA in the wild-type strain. In L. lactis, LTA has been shownto be involved in controlled (directed) binding of AcmA. LTAlacking D-Ala has been reported in other bacterial species tohave an improved capacity for autolysin binding. Mutation ofdltD in L. lactis, however, does not affect peptidoglycan bindingof AcmA; neither the amount of AcmA binding to the cells northe binding to specific loci is altered. In conclusion, D-Aladepletion of the cell wall causes lysis by two distinct mechanisms.First, it results in an altered peptidoglycan that is more susceptibleto lysis by AcmA and also by other factors, e.g., one or moreof the other (putative) cell wall hydrolases expressed by L.lactis. Second, reduced amounts of D-Ala on LTA result in decreaseddegradation of AcmA by HtrA, which results in increased lyticactivity.

INTRODUCTION

Top

Introduction

AcmA, the major autolysin of Lactococcus lactis MG1363, is responsiblefor stationary phase cellular lysis and is involved in cellseparation of this organism (9). The enzyme consists of twodomains: the N-terminal region contains an N-acetyl-glucosaminidaseactive site domain (9; A. Steen, G. Buist, G. Horsburgh, S.J. Foster, O. P. Kuipers, and J. Kok, unpublished data) whilethe C-terminal region contains three so-called LysM domains,with which it specifically binds to peptidoglycan of L. lactisand of other gram-positive bacteria (49). Peptidoglycan, themajor cell wall component in bacteria and the substrate of AcmA,consists of glycan strands cross-linked by peptide side chains.The peptide chain contains alternating L- and D-amino acids.D-Alanine (D-Ala) is incorporated into the peptidoglycan peptidemoiety as a D-Ala-D-Ala dipeptide, where it is involved in cross-linkingof adjacent peptidoglycan strands. In many bacteria alanineracemase is responsible for the synthesis of D-Ala from L-Ala,the naturally occurring alanine isomer (53). Bacillus subtilisexpresses at least one alanine racemase: Dal (14). A dal mutantis dependent on D-Ala supplementation to be able to grow ina rich medium; cells start to lyse in the absence of D-Ala (4,14, 20). In minimal medium the mutant is D-Ala dependent whenL-Ala is supplemented, suggesting that a second, L-Ala-repressibleracemase is present (4, 14). Lactobacillus plantarum probablyexpresses only one alanine racemase, as an alr mutant is totallydependent on D-Ala for growth (22). D-Ala deprivation of analr mutant of L. plantarum resulted in growth arrest, a rapidloss of cell viability, and an aberrant cell morphology (43).Electron microscopy analyses showed that mainly the cell septumis affected in this mutant. Like L. plantarum alr, L. lactisalr is totally dependent on the addition of D-Ala to the growthmedium (23); when D-Ala was removed from the growth medium whenthe cells were in exponential growth phase, L. lactis alr growthwas impaired and the culture started to lyse (17). The alr genewas used as a food-grade plasmid selection marker in the alrmutants of L. plantarum and L. lactis, complementing the D-Alaauxotrophy (7). Moreover, the alr mutants of L. plantarum andL. lactis were used in a mucosal vaccination study, in whichthese two mutants were shown to enhance the mucosal deliveryof the tetanus toxin fragment C model antigen in mice (17).

Although peptidoglycan covers the whole surface of L. lactis,AcmA binds to lactococcal cells at specific loci, namely aroundthe poles and septum of the cell, exactly those places wherecell lysis has been shown to start (35, 49). Trichloroaceticacid treatment of cells causes binding of AcmA over the wholecell surface. Lipoteichoic acid (LTA) is a candidate-hinderingcomponent that is removed by this treatment, as it seems tobe present in L. lactis at those positions where AcmA is notable to bind (49). LTA is a secondary cell wall polymer suggestedto be involved in the control of autolysin activity (5, 15),in determining the electrochemical properties of the cell wall(42), in establishing a magnesium ion concentration (2, 21,26, 31), and in determining the physicochemical properties ofthe cytoplasmic membrane (18). LTA can be modified by variouscompounds, such as glycosyl residues (15) and D-Ala esters (1).

In gram-positive bacteria, the products of the dlt operon areinvolved in D-alanylation of LTA. The operon comprises fourgenes: dltA, encoding D-alanine-D-alanine carrier protein ligase(Dcl), the product of which is an activated D-Ala; dltC, specifyingthe D-alanyl-carrier protein (Dcp); dltB, which encodes a putativetransmembrane protein involved in the secretion of the activatedD-alanine; and dltD, encoding a protein that facilitates thebinding of Dcp and Dcl for ligation with D-Ala and has thioesteraseactivity for mischarged D-alanyl-acyl carrier proteins (11,28, 38). A link between D-alanylated LTA and autolysin activityhas been reported earlier: autolysis of B. subtilis dltA, dltB,dltC, or dltD mutants was enhanced, and the bacteria were moresusceptible to methicillin, which resulted in accelerated cellwall lysis, a faster loss of cell viability, and a slower recoveryof the cells in the post-antibiotic phase (52). Furthermore,the absence of D-Ala in the LTA of a dltB as well as a dltDmutant of B. subtilis causes an increase in the net negativecharge of the cell wall, resulting in an increase in the rateof posttranslational folding of some exported proteins (27).In the absence of D-alanylation, the yield of secreted recombinantanthrax protective antigen was increased 2.5-fold (50). B. subtiliscell growth, basic metabolism, cellular content of phosphorus-containingcompounds, cell separation, and surface charge were not altered(52). Insertional mutagenesis in the dlt operon of Staphylococcusaureus resulted in methicillin resistance and an increased autolysis(36). In L. lactis subsp. lactis IL1403, the dlt operon comprisesfour genes: dltA, dltB, dltC, and dltD (6). An L. lactis dltDmutant was obtained by random insertion mutagenesis and screeningfor UV-sensitive mutants (13). Apart from its UV sensitivity,the dltD mutant was characterized by having a lower plasmidtransfer rate during conjugation, and it was possible to makethe mutant electrocompetent without the addition of glycineto the growth medium (13). Insertion mutagenesis of dltA ofL. lactis resulted in secretion defects of the staphylococcalnuclease, which was used as a reporter for secretion. The secretiondefect is most probably caused by an entrapment of the reporterprotein in the cell wall, which could be the result of the interactionof the positively charged nuclease with the expected negativelycharged cell wall of the dltA mutant (41).

In this paper we show that mutations in the alr and dltD genesof L. lactis differentially affect autolysis, and we investigatethe role of the major autolysin AcmA therein.

MATERIALS AND METHODS

Top

Materials and Methods

Bacterial strains, plasmids, media, and growth conditions. The bacterial strains and plasmids used in this study are listedin Table 1. L. lactis was grown at 30°C in M17 broth (DifcoLaboratories, Detroit, Mich.) or on twofold diluted M17 (1/2M17)medium containing 0.5% glucose (GM17 or G1/2M17, respectively).When appropriate D-alanine was added to reach a 2 mM end concentration.For plasmid selection in L. lactis, chloramphenicol or erythromycin(Sigma-Aldrich, St. Louis, Mo.) was added (each to a concentrationof 5 µg/ml). For plasmid selection in Escherichia coli,erythromycin or ampicillin (Sigma-Aldrich) was used at a concentrationof 100 µg/ml.

View this table:
[in this window]
[in a new window]
TABLE 1. Strains and plasmids used in this study

Chemicals and enzymes. All chemicals used were of analytical grade and, unless indicatedotherwise, were obtained from Merck (Darmstadt, Germany). Enzymesfor molecular biology were purchased from Roche Molecular Biochemicals(Mannheim, Germany) and used according to the suppliers' instructions.

DNA manipulations and transformation. Molecular cloning techniques were performed essentially as describedby Sambrook et al. (47). Electrotransformation of E. coli andL. lactis was performed by using a gene pulser (Bio-Rad Laboratories,Richmond, Calif.) as described by Zabarovsky and Winberg (54)and Leenhouts and Venema (32), respectively. Mini-preparationsof plasmid DNA from E. coli and L. lactis were obtained by thealkaline lysis method as described by Sambrook et al. (47) andSeegers et al. (48), respectively. PCR products were purifiedby using a High Pure PCR purification kit (Roche Molecular Biochemicals).

Construction of an alanine racemase-deficient mutant and acmA knockouts in L. lactis. The suicide plasmid used to generate the alr disruption wasconstructed as follows. A 0.68-kb PCR product was amplifiedfrom MG1363 chromosomal DNA by using primers derived from theL. lactis IL1403 genome sequence (LLALR3, 5'-CCTGTCGAAAATATTTATAAAGCTG,and LLALR4, 5'-CGAGGATCCCCAAATTTCCGCATTAGGGTGAATATG [BamHI sitein italics]). This DNA fragment was restricted with PstI andBamHI and cloned into the corresponding sites of the pBLUESCRIPTSK⁺ (Stratagene GmbH, Heidelberg, Germany) plasmid. The DNAfragment was next recovered as a SalI-BamHI fragment and clonedinto the corresponding sites of the suicide plasmid pGIP011(described in reference 23) to generate plasmid pGIP016. Theresulting plasmid contains an in-frame deletion of the alr gene(30 nucleotides), resulting in the removal of the pyridoxal-Pbinding site of the Alr enzyme. The stable alr mutant was constructedby two successive crossover events. First-step integrant candidatesof strain NZ3900 (isogenic to MG1363) resulting from a singlecrossover recombination of the pGIP016 suicide vector at thealr locus were selected on GM17 plates containing both erythromycinand D-Ala. Three D-Ala auxotroph candidates were obtained andwere grown in GM17 (plus D-Ala) without erythromycin during120 generations in order to allow excision of the plasmid throughintrachromosomal recombination. A total of 1,500 colonies (GM17plus D-Ala) were examined by replica plating on GM17 containingboth D-Ala and erythromycin or on GM17 without D-Ala. Twenty-sevenerythromycin-sensitive candidates were obtained, one of whichwas a D-Ala auxotroph. This mutant strain (named PH3960) wasretained for detailed phenotypic analysis following furthervalidation by PCR amplification and Southern blotting (datanot shown).

The acmA deletion derivatives of L. lactis MG1363 (dltD) andL. lactis PH3960 ( ${Delta}$ alr) were obtained by replacement recombinationwith plasmid pINTAA, replacing acmA with the acmA ${Delta}$ 1 gene whichhas an internal deletion, as described by Buist et al. (9).

Enzyme assays and optical density measurements. AcmA activity was visualized on G1/2M17 agar plates containing0.2% autoclaved lyophilized Micrococcus lysodeikticus 4698 cells(Sigma-Aldrich) as halos around L. lactis colonies after overnightgrowth at 30°C.

X-prolyl dipeptidyl aminopeptidase (PepX) was measured as describedby Buist et al. (10). Similar patterns were obtained in threeindependent experiments. Optical densities at 600 nm (OD₆₀₀)of cell cultures were measured in a Novaspec II spectrophotometer(Pharmacia, Uppsala, Sweden).

To study lysis caused by AcmA, 1 ml of an overnight cultureof L. lactis MG1363acmA ${Delta}$ 1 or L. lactis MG1363 (dltD acmA ${Delta}$ 1) wasmixed with 1 ml of AcmA-containing supernatant of an overnightL. lactis MG1363 culture. After overnight incubation at 37°C,the amount of PepX released from the cells was measured as describedabove.

Lactate dehydrogenase activity in the supernatant was determinedby using pyruvate as a substrate as previously reported (23).Aliquots (1 ml) of the culture were withdrawn at various timepoints, cells were removed by centrifugation at 12,000 x g for5 min, and the supernatant was stored immediately at –20°Cuntil enzyme assays were performed. One unit of activity correspondsto the oxidation of 1 µmol of NADH per min.

Total protein concentration was measured by the Bio-Rad proteinassay method.

Triton X-100 induced lysis under nongrowing conditions. The procedure of Triton X-100-induced lysis was performed asdescribed previously (46) with the following modifications.Cells were grown in GM17 for 16 h at 30°C, harvested bycentrifugation (12,000 x g for 5 min at room temperature), washedthree times with an equal volume of a potassium-sodium phosphate(K-NaPO₄) buffer (10 mM; pH 6.5) and resuspended in K-NaPO₄buffer (200 mM; pH 6.5) containing 0.05% Triton X-100 (vol/vol).The cell suspension was incubated at 37°C under agitation,and autolysis was monitored by the decrease in OD₆₀₀ in time.

SDS-PAGE, AcmA zymograms, and Western hybridization. L. lactis cell and supernatant samples were prepared as describedpreviously (10). AcmA activity was detected by a zymogram stainingtechnique by using sodium dodecyl sulfate-polyacrylamide gelelectrophoresis (SDS-PAGE) with 12.5 or 17.5% polyacrylamidegels containing 0.15% autoclaved, lyophilized M. lysodeikticusATCC 4698 cells as described before (9). The standard low-rangeand prestained low- and high-range SDS-PAGE molecular weightmarkers of Bio-Rad were used as references. Proteins were transferredfrom SDS-10% polyacrylamide gels to polyvinylidene difluoridemembranes (Roche Molecular Biologicals) as described by Towbinet al. (51). MSA2 antigen was detected with 10,000-fold dilutedrabbit polyclonal anti-MSA2 antiserum (45), and horseradishperoxidase-conjugated goat anti-rabbit secondary antibodies(Pharmacia) by using an ECL chemiluminescence detection systemand protocol (Amersham, Piscataway, N.J.).

Binding of MSA2cA and AcmA to L. lactis cells. Binding of the MSA2cA fusion protein was studied by growingL. lactis NZ9000 (pNG3041) at 30°C until an OD₆₀₀ of 0.4was reached. L. lactis NZ9000 (pNG3041) was induced with nisinby adding a 1/1000 volume of the supernatant of the nisin producerL. lactis NZ9700 and incubated at 30°C for 2 h. The cellsof 1 ml of L. lactis MG1363acmA ${Delta}$ 1 or L. lactis MG1363 (dltD acmA ${Delta}$ 1)culture were resuspended in 1 ml of supernatant of the inducedL. lactis NZ9000 (pNG3041) culture. The suspensions were incubatedfor 5 min at room temperature and spun down, after which thecells were washed with 1 ml of fresh 1/2M17 medium and resuspendedin 100 µl of denaturation buffer (3). After boiling for5 min, the samples were subjected to SDS-10% PAGE, followedby Western hybridization by using anti-MSA2 antibodies. Thebinding of AcmA to lactococcal cells was studied by mixing thecells of 1 ml of an overnight culture of L. lactis MG1363acmA ${Delta}$ 1or L. lactis MG1363 (dltD acmA ${Delta}$ 1) with 1 ml of AcmA-containingsupernatant of an L. lactis MG1363 culture. The suspensionswere incubated for 5 min at room temperature and centrifuged.The supernatant (0.5 ml) was mixed with 0.5 ml of an M. lysodeikticussuspension (0.2% in 100 mM potassium phosphate buffer, pH 5.4).The OD₆₀₀ of the suspensions were measured at time zero andat 30 min. To determine the maximal amount of AcmA activitypresent in the MG1363 supernatant, 0.5 ml of L. lactis MG1363supernatant was mixed with the M. lysodeikticus cells (controlsuspension). The ${Delta}$ OD₆₀₀ of the M. lysodeikticus suspensions withMG1363 supernatant preincubated with wild-type or mutant cellswas calculated as a percentage of the ${Delta}$ OD₆₀₀ of the control suspension,resulting in the percentage of AcmA that did not bind to thecells. The unbound percentage was then subtracted from 100%to give the bound percentage.

Immunofluorescence microscopy, electron microscopy, and measurement of cell wall and septum thickness. Immunofluorescence studies to localize MSA2cA on the cell surfaceof L. lactis were performed as described before (49) by usinganti-MSA2 antibodies and Oregon green-labeled secondary antibody(Molecular Probes, Oreg.).

Samples for transmission electron microscopy were prepared asfollows. Cells (mid-exponential or stationary phase) were harvestedby low-speed centrifugation (3,000 x g for 5 min). The cellswere washed twice in sodium cacodylate buffer (200 mM; pH 7.3),prefixed in 2.5% (wt/vol) glutaraldehyde and fixed with 1% (wt/vol)OsO₄. The samples were embedded in Epon resin (Fluka Chemie,Buchs, Switzerland), and thin sections were prepared by usinga Reichert-Jung Ultracut microtome (Milton Keynes, United Kingdom).The sections were stained with 4% (wt/vol) uranyl acetate andthen with 0.4% (wt/vol) lead citrate and were examined witha JEOL JEE 1200 EXII electron microscope at 100 kV. Cell walland septum thickness was measured by using the software programXL Docu version 3.0 (Soft Imaging System GmbH, Munster, Germany).

RESULTS

Top

Results

Mutation of alr in L. lactis increases autolysis. The stable isogenic L. lactis NZ3900 alr deletion mutant L.lactis PH3960 ( ${Delta}$ alr) is only able to grow when D-Ala is presentin the growth medium (17, 23). When D-Ala is removed from thegrowth medium in exponential phase by washing the culture withfresh medium and resuspending the culture in medium withoutD-Ala, the growth of L. lactis PH3960 ( ${Delta}$ alr) is impaired andcell lysis starts (reference 17 and Fig. 1A). Depletion of D-Aladid not affect L. lactis NZ3900 with respect to growth and autolysis.L. lactis PH3960 ( ${Delta}$ alr) behaved like the control strain L. lactisNZ3900 in the presence of D-Ala (Fig. 1A).

View larger version (18K):
[in this window]
[in a new window]
FIG. 1. Effect of D-Ala depletion on growth and lysis of L. lactis PH3960. (A) Growth and lysis, as measured by following the OD₆₀₀, of L. lactis strains NZ3900 (control) and PH3960 ( ${Delta}$ alr) in the presence and absence of D-Ala. The strains were grown in GM17 supplemented with 2 mM D-Ala until an OD₆₀₀ of 0.6 was reached. Subsequently, the cultures were split: one half was grown in GM17 with 2 mM D-Ala while the other half was grown without D-Ala. (B) Growth and lysis of PH3960 ( ${Delta}$ alr) after D-Ala depletion and in the presence of 0.5 M sucrose or 0.5 M NaCl. Similar results were obtained in independent experiments.

When glycine is added to the growth medium, it will be incorporatedin the peptidoglycan in place of D-Ala, affecting the peptidoglycancross-linking and resulting in a destabilized cell wall (19).Indeed, when glycine was added to the growth medium, more lysiswas observed than when no glycine was added, both for the wild-typeL. lactis NZ3900 and for strain PH3960 ( ${Delta}$ alr) (Table 2). However,strain PH3960 ( ${Delta}$ alr) is more sensitive to glycine and lyses toa greater extent than strain NZ3900, even when grown in thepresence of D-Ala (Table 2). When D-Ala is depleted, the additionof glycine (1 as well as 2%) increases the lysis of strain PH3960( ${Delta}$ alr); lysis of strain PH3960 ( ${Delta}$ alr) under these conditions wasalmost complete after 22 h, as evidenced by a very low OD₆₀₀(Table 2). To examine whether osmotic stabilizers could reducelysis of L. lactis PH3960 ( ${Delta}$ alr), sodium chloride or sucrosewas added at a concentration of 0.5 M during D-Ala starvation.An effect on lysis was observed, as the OD₆₀₀ of the culturesgrown in the presence of the osmotic stabilizers were higherthan when no osmotic stabilizers were present (Fig. 1B). Protoplastformation in the presence of 0.5 M NaCl after D-Ala starvationwas observed by using a light microscope (data not shown). Themorphology of strain PH3960 ( ${Delta}$ alr) during D-Ala starvation wasstudied by using transmission electron microscopy; the septumand cell wall were thinner than those of the wild-type (Fig.2, compare A and B), and occasionally holes were observed inthe septal region (Fig. 2C).

View this table:
[in this window]
[in a new window]
TABLE 2. Effect of glycine on lysis of L. lactis NZ3900 (control) and L. lactis PH3960 ( ${Delta}$ alr) in the presence or absence of D-Ala^a

View larger version (39K):
[in this window]
[in a new window]
FIG. 2. Effect of D-Ala depletion on the cell integrity. (A) Electron microscopy image of L. lactis NZ3900 in stationary phase. (B) Electron microscopy image of L. lactis PH3960 ( ${Delta}$ alr) after 5 h of D-Ala depletion. Arrows show the thinner cell wall around the septum. (C) Electron microscopy image of L. lactis PH3960 ( ${Delta}$ alr) after 2 h of D-Ala depletion. The white rectangle shows where release of cytoplasmic material occurs.

The major autolysin AcmA of L. lactis MG1363 binds around theseptal region of the cell (49). To study the possible involvementof AcmA in the lysis of L. lactis PH3960 ( ${Delta}$ alr), an isogenicacmA alr double mutant was constructed. In the absence of D-Ala,L. lactis GB3960 ( ${Delta}$ alr acmA ${Delta}$ 1) still lyses and releases intracellularproteins into its medium, albeit to a much lesser extent thanstrain PH3960 ( ${Delta}$ alr) (Fig. 3). After 20 h of D-Ala starvation,the OD₆₀₀ of strain GB3960 ( ${Delta}$ alr acmA ${Delta}$ 1) was approximately twotimes higher (Fig. 3) than that of strain PH3960 ( ${Delta}$ alr). Totalprotein release from strain GB3960 ( ${Delta}$ alr acmA ${Delta}$ 1) is two timeslower than that from strain PH3960 ( ${Delta}$ alr) upon D-Ala starvation(Fig. 3). In an acmA ${Delta}$ 1 mutant of L. lactis MG1363, the isogenicparent of all strains used in this study, neither lysis nora decrease in the OD₆₀₀ value was observed (Fig. 4A) (9).

View larger version (22K):
[in this window]
[in a new window]
FIG. 3. Growth (OD₆₀₀, solid lines) and release of cytoplasmic proteins (dashed lines) by L. lactis PH3960 ( ${Delta}$ alr) and L. lactis GB3960 ( ${Delta}$ alr acmA ${Delta}$ 1). The two strains were grown in GM17 medium with 2 mM D-Ala until exponential phase, after which D-Ala was removed from the growth medium by centrifugation and cells were resuspended in fresh GM17 broth without D-Ala.

View larger version (24K):
[in this window]
[in a new window]
FIG. 4. (A) Effect of AcmA on growth and lysis of L. lactis NZ3900, MG1363acmA ${Delta}$ 1, MG1363 (dltD), and MG1363 (dltD acmA ${Delta}$ 1). The OD₆₀₀ of GM17 cultures of L. lactis was followed for 24 h (solid lines). Lactate dehydrogenase release in culture supernatant of the same strains was measured at time (t) = 7 and t = 22 h (dashed lines). The data are from a single experiment that was representative of two independent analyses in which similar results were obtained. (B) Effect of Triton X-100 on lysis of cell suspensions of L. lactis NZ3900 (control), MG1363acmA ${Delta}$ 1, MG1363 (dltD), and MG1363 (dltD acmA ${Delta}$ 1). Cells were harvested, washed, resuspended in phosphate buffer supplemented with 0.05% Triton X-100, and incubated at 37°C. The OD₆₀₀ of the cell suspensions was followed for 5 h (solid lines). Total protein release in the supernatant of the cell suspensions was measured at t = 0, t = 0.5, and t = 5 h (dashed lines). The data are from a single experiment that was representative of two independent analyses in which similar results were obtained.

The thickness of the septum and cell wall (excluding septum)of the mutant strains was measured. Global cell wall thicknessof strains PH3960 ( ${Delta}$ alr) and GB3960 ( ${Delta}$ alr acmA ${Delta}$ 1) are not affectedby the absence of D-Ala. The septum of strain PH3960 ( ${Delta}$ alr) isa bit thinner than that of strain NZ3900. Strain GB3960 ( ${Delta}$ alracmA ${Delta}$ 1) has a thinner septum than strain PH3960 ( ${Delta}$ alr), whichis an effect of the absence of AcmA (Table 3 and see below).

View this table:
[in this window]
[in a new window]
TABLE 3. Cell wall and septum thickness of the strains used in this study^a

A mutation in dltD affects growth and autolysis of L. lactis. A mutation in alr affects both peptidoglycan synthesis and LTAdecoration with D-Ala, while a mutation in dltD only affectsdecoration of LTA with D-Ala (37). In L. lactis MG1363 the organizationof dltA, dltB, dltC, and dltD and their surrounding genes isidentical to that in L. lactis IL1403. L. lactis MG1363 dltDwas obtained by insertion of ISS1 in the dltD gene (13), andthis mutant was used to investigate the effect of changes inLTA on AcmA activity. Growth and lysis of L. lactis MG1363 (dltD)were followed in time (Fig. 4A). L. lactis MG1363 (dltD) growsmore slowly and lyses to a greater extent than strain MG1363,releasing more intracellular protein in the culture supernatantwhen stationary phase is reached, as attested in the supernatantby an increase in activity of the cytoplasmic enzyme lactatedehydrogenase (Fig. 4A).

As the mutation in dltD clearly affects autolysis, the roleof AcmA in this phenomenon was investigated by constructingan isogenic dltD acmA double mutant. Growth and lysis of L.lactis strains MG1363acmA ${Delta}$ 1 and MG1363 (dltD acmA ${Delta}$ 1) are similar,and no proteins are released (Fig. 4A), which is a clear indicationthat AcmA is responsible for the increased lysis of MG1363 (dltD).L. lactis MG1363 (dltD) is more sensitive to incubation withTriton X-100 (a lysis inducer) than L. lactis MG1363 (dltD acmA ${Delta}$ 1)(Fig. 4B). The OD₆₀₀ of L. lactis MG1363 (dltD) drops dramaticallyin the first 30 min of incubation compared to the value in thecontrol strain L. lactis NZ3900, and more proteins are releasedinto the supernatant. Only limited lysis is observed for strainsMG1363 (dltD acmA ${Delta}$ 1) and MG1363acmA ${Delta}$ 1, with small amounts of proteinsbeing released under these circumstances.

The mutation of dltD does not affect septum and cell wall thickness(Table 3) and cell shape (data not shown). By contrast, thesepta of L. lactis MG1363acmA ${Delta}$ 1 and L. lactis MG1363 (dltD acmA ${Delta}$ 1)are thinner than that of L. lactis NZ3900. Apparently, AcmAis involved in the synthesis of the septum.

AcmA binding to dltD mutant cells is not affected. AcmA has to bind to the peptidoglycan of the lactococcal cellwall to be able to lyse the cell (49). It is an enzyme witha high pI of approximately 10 and is, therefore, positivelycharged at the pH value of the medium (pH 6.8). During growthof L. lactis the pH value drops due to the production of lactate.L. lactis MG1363 (dltD) contains a fivefold reduced amount ofD-Ala on its LTA compared to MG1363 (D-Ala to GroP ratio of5.8% in L. lactis MG1363 [dltD], compared to D-Ala to GroP ratioof 28.5% in L. lactis MG1363; N. Kramer, S. Morath, T. Hartung,E. J. Smid, E. Breukink, J. Kok, and O. P. Kuipers, unpublisheddata).

The LTA of D-Ala-depleted L. lactis MG1363 (dltD) is expectedto have an increased net negative charge, which could possiblyallow this mutant to bind, via electrostatic interactions, moreAcmA at the pH value of the medium than the wild-type strain.To investigate whether an increase in the binding of AcmA couldexplain the increase in lysis of L. lactis MG1363 (dltD), equalamounts of exponential phase cells of L. lactis MG1363acmA ${Delta}$ 1and L. lactis MG1363 (dltD acmA ${Delta}$ 1) were mixed with an AcmA-containingsupernatant of an L. lactis MG1363 culture. After mixing, thecell-AcmA suspensions were centrifuged and the supernatants,containing nonbound AcmA, were mixed with M. lysodeikticus autoclavedcells. The resulting decrease in the OD₆₀₀ value was taken asa measure of the amount of AcmA present in the supernatantswith the original AcmA-containing supernatant set at 100%. Equalamounts of AcmA bound to the cells of L. lactis MG1363 (dltDacmA ${Delta}$ 1) and L. lactis MG1363acmA ${Delta}$ 1 (Table 4), and lysis of bothstrains, as measured by the release of the intracellular markerenzyme PepX, was the same (data not shown).

View this table:
[in this window]
[in a new window]
TABLE 4. Quantification of AcmA-binding to L. lactis MG1363acmA ${Delta}$ 1 and L. lactis MG1363 (dltD acmA ${Delta}$ 1)^a

AcmA binds with its C-terminal domain (cA) to specific locion the lactococcal cell surface, namely, around the septum andat the poles of the cell (49). To examine whether a mutationin dltD affects the localization of AcmA and, consequently,influences its activity, immunofluorescence studies were performed.The MSA2cA protein, which is a fusion of the reporter proteinMSA2 of the malaria parasite Plasmodium falciparum and the cAdomain of AcmA (49), bound to the same loci on the cell surfaceof strains MG1363 (dltD acmA ${Delta}$ 1) and MG1363acmA ${Delta}$ 1 (Fig. 5), indicatingthat the distribution of AcmA over the cell surface is not alteredby the inactivation of dltD. Western hybridization analysiswith MSA2-specific antibodies showed that equal amounts of MSA2cAwere bound to cells of L. lactis MG1363acmA ${Delta}$ 1 and L. lactis MG1363(dltD acmA ${Delta}$ 1) (Fig. 5A).

View larger version (44K):
[in this window]
[in a new window]
FIG. 5. (A) MSA2cA binding to cells. Equal amounts of cells of overnight cultures of L. lactis strains MG1363acmA ${Delta}$ 1 and MG1363 (dltD acmA ${Delta}$ 1) were mixed with MSA2cA, a fusion protein consisting of the Plasmodium falciparum protein MSA2 and the C-terminal cell wall binding domain of AcmA. The suspensions were incubated for 5 min and centrifuged. The cells were washed once and cell-bound protein was visualized by Western hybridization with anti-MSA2 antibodies. Lane 1, MSA2cA bound to L. lactis MG1363 (dltD acmA ${Delta}$ 1); lane 2, MSA2cA bound to L. lactis MG1363acmA ${Delta}$ 1. (B) Localization of MSA2cA on the cell surface of L. lactis. MSA2cA was visualized on cells of L. lactis strains MG1363acmA ${Delta}$ 1 and MG1363 (dltD acmA ${Delta}$ 1) by immunofluorescence microscopy by using anti-MSA2 antibodies and a fluorescently labeled secondary antibody.

A dltD mutation reduces the breakdown of AcmA. From studies in B. subtilis, it is known that D-alanylationof LTA affects the stability and breakdown of several heterologousproteins (27). To examine whether the stability of AcmA of L.lactis is affected by a mutation in dltD, supernatants of stationaryphase cultures of L. lactis strains MG1363 and MG1363 (dltD)were analyzed on a zymogram containing M. lysodeikticus cellwall fragments (Fig. 6). Next to mature AcmA, several smallerbands of activity of breakdown products of AcmA were observedin the supernatant of L. lactis MG1363, as shown previously(9). In the supernatant of L. lactis MG1363 (dltD) only full-sizeAcmA was detectable.

View larger version (34K):
[in this window]
[in a new window]
FIG. 6. Zymographic analysis of AcmA activity in the supernatant of L. lactis MG1363 and L. lactis MG1363 (dltD) after culturing for 16 h or 48 h in G1/2M17 at 30°C. dAcmA, breakdown products of AcmA generated by the protease HtrA.

DISCUSSION

Top

Discussion

D-Ala is essential for L. lactis. An L. lactis alr mutant, whichdoes not synthesize D-Ala from L-Ala, starts to lyse upon D-Aladepletion. This mutant is expected to be defective in cross-linkingof the glycan strands, which is necessary for the rigid structureof peptidoglycan, since D-Ala in the disaccharide pentapeptideprecursor of peptidoglycan plays an important role in the cross-linkingprocess. In addition, D-alanylation of the LTA in this mutantwill be affected since this process also depends on the formationof D-Ala from L-Ala. In an analysis of the lysis of the alrmutant, we show that cell lysis increases when glycine is addedto the growth medium, even in the presence of D-Ala, showingthat glycine and D-Ala compete; it has been shown that duringthe biosynthesis of the peptidoglycan precursor (disaccharidepentapeptide), glycine can replace D-Ala in, e.g., L. plantarumand Staphylococcus aureus (19). While L. lactis Alr⁺ is ableto synthesize D-Ala from L-Ala, the L. lactis alr mutant dependson the D-Ala that is added to the culture growth medium. Whenthe D-Ala concentration becomes limiting for the alr mutant,the glycine incorporation into the peptidoglycan will increase.The alr mutant of L. lactis is therefore more sensitive to theaddition of glycine than L. lactis Alr⁺.

Analysis of L. lactis cell morphology by electron microscopyconfirms that mutation of alr affects cell wall synthesis; cellwall defects are observed, mainly in the septal region of thecell. The major autolysin, AcmA, is involved in cell separationand has been shown to bind and act in this region (49). We showhere that, indeed, AcmA is involved in the increased lysis ofL. lactis (alr). To our knowledge, this is the first time thatan autolysin is shown to be directly involved in the lysis ofmutants defective in peptidoglycan precursor biosynthesis.

The lysis observed in the alr mutant is not fully AcmA mediated.The L. lactis alr acmA double mutant still lyses, albeit toa lesser extent than L. lactis alr. Other factors, e.g., theactivity of one or more other cell wall hydrolases, could beinvolved. Indeed, genes for three homologues of AcmA (AcmB,AcmC, and AcmD) and a putative lytic endopeptidase (YjgB) arepresent in the genome of L. lactis (6), and recently activityof these peptidoglycan hydrolases has been shown in vitro (24).AcmB has autolytic activity, since an acmB mutant of MG1363lyses to a lesser extent than its parent (25). Lytic activityof AcmB, however, depends on the presence of AcmA as MG1363(acmA ${Delta}$ 1 acmB) lyses to the same extent as MG1363acmA ${Delta}$ 1 (25). AcmBis, therefore, not expected to be involved in the lysis of L.lactis GB3960 ( ${Delta}$ alr, acmA ${Delta}$ 1) observed in this study.

Since mutating alr will not only affect peptidoglycan synthesisbut also the D-alanylation of LTA, we also investigated whethera reduction of the D-alanyl substitution level of LTA couldstimulate autolysis. A dltD mutant of L. lactis was examinedwith respect to lysis behavior, since a mutation in the dltoperon is expected to affect only the D-alanylation of LTA andnot the peptidoglycan cross-linking (37).

D-Ala substitution in LTA is strongly reduced in the dltD mutantof L. lactis; a fivefold reduced amount of D-Ala in its LTAcompared to levels in the wild-type L. lactis MG1363 is observed(D-Ala to GroP ratio of 5.8% for the dltD mutant compared toD-Ala to GroP ratio of 28.5% for the wild-type strain; N. Krameret. al., unpublished data). We show here that this strong reductionin D-Ala substitution in LTA results in increased lysis of L.lactis. A slower growth rate of the mutant is also observed.The same phenomena have been observed in B. subtilis (52). AdltD acmA double mutant of L. lactis does not lyse more, andgrowth is completely restored; the increased lysis and the effecton growth of L. lactis MG1363 (dltD) are, therefore, dependenton the presence of AcmA. LTA lacking D-Ala substitutions isexpected to be more negatively charged. Fischer et al. (15)have reported in S. aureus that acidified LTA, due to a lackof D-Ala, has an improved capacity for binding of the S. aureusautolysin. The possibility that the increased lysis of L. lactis(dltD) was the result of increased binding of AcmA to the cellwall was investigated, since autolysins, including Atl of S.aureus and AcmA, are positively charged enzymes. Mutation ofdltD in L. lactis, however, did not influence the binding ofAcmA; similar amounts of AcmA bound to cells of L. lactis MG1363acmA ${Delta}$ 1and L. lactis MG1363 (dltD acmA ${Delta}$ 1), and the binding was at thesame loci in both strains. Also lysis of both strains upon theexternal addition of AcmA was identical, indicating that thelysis activity of AcmA remained unchanged and is not influencedby D-alanylation of LTA. In a previous study (49) it has beenshown that LTA is likely to be involved in hindering of AcmAbinding to cells. Since D-Ala in LTA has no effect on AcmA binding,the present study supports the conclusion that D-Ala in LTAis not involved in hindering AcmA binding.

Although the quantity and location of AcmA binding are not affectedin L. lactis (dltD), the strain lyses to a higher extent thanits parent. The thickness of the cell wall and septum is notaffected by the dltD mutation, precluding the possibility thatthe lysis increase of L. lactis (dltD) is a consequence of athinner cell wall. On the other hand, a direct correlation betweenthe mutation of AcmA and a thinner septum was observed, indicatingthat AcmA is important for the development of a normal septum.

The cell wall of L. lactis MG1363 (dltD) does seem to be weakerthan that of the wild-type strain. When lysis is induced withTriton X-100, L. lactis MG1363 (dltD) lyses to a greater extentthan its parent. Triton X-100-induced lysis of L. lactis MG1363(dltD acmA ${Delta}$ 1) and L. lactis MG1363acmA ${Delta}$ 1 is of the same low level.Incubation of stationary phase cells of L. lactis MG1363acmA ${Delta}$ 1or L. lactis MG1363 (dltD acmA ${Delta}$ 1) with AcmA does not result inincreased lysis of the double mutant, which indicates that AcmA,only when expressed in L. lactis MG1363 (dltD) during growth,results in a weaker cell wall.

D-Alanylation of LTA has been shown previously to be an importantfactor in the stability of secreted proteins. A dlt mutant ofB. subtilis showed an increase in the rate of posttranslationalfolding of exported proteins, especially those proteins thatare susceptible to proteolytic degradation (27). In the absenceof D-alanylation, the yield of secreted recombinant anthraxprotective antigen was increased 2.5-fold (50). The proteasethat is involved in this phenomenon has not been identified.Here we show that in L. lactis a mutation in the dltD gene alsostabilizes a secreted protein, i.e., AcmA. Zymographic analysisof AcmA activity in supernatants of L. lactis MG1363 and itsdltD mutant revealed that the HtrA-mediated breakdown of AcmA(44), which can be seen in MG1363, is not detectable in themutant.

L. lactis HtrA cleaves AcmA in the C-terminal peptidoglycan-bindingdomain (8, 44). The AcmA breakdown products are still active(Fig. 6) but will bind less strongly to the lactococcal cellwall, and, as a consequence, they have a lower lytic ability(A. Steen et al., unpublished data, and reference 8). HtrA degradationof AcmA could be a way for the cell to modulate the activityof this potentially lethal enzyme. This control mechanism onAcmA activity is absent in L. lactis (dltD) and could resultin the weakening of the peptidoglycan sacculus by the hydrolyticaction of AcmA, resulting in increased lysis.

The mutation in dltD could influence the stability of AcmA intwo ways. First, HtrA activity could be decreased by the alteredphysicochemical properties of the cell wall. Second, since foldedproteins are expected to be less prone to degradation by HtrA,the folding of AcmA could be faster, due to the proposed increaseof cation concentration in the cell envelope (26), a factorknown to influence the folding rate of secreted proteins (27,50). The exact mechanism in the case of AcmA is unknown.

Interestingly, B. subtilis expresses autolysins which containpeptidoglycan binding domains homologous to that of AcmA (33,34), as well as homologues of HtrA (39, 40). It would be interestingto investigate the role of HtrA in the stability of autolysinsand other secreted proteins in dlt mutants of B. subtilis, asit is known that mutation in the dlt operon increases cell lysisin B. subtilis (52).

In conclusion, whereas in the dltD mutant the increased lysisis caused by an effect of reduced D-alanylation levels of LTAon HtrA activity, cell lysis of the alr mutant of L. lactisis likely a combined effect of a defective peptidoglycan synthesisand a reduced D-alanylation level of LTA (Fig. 7). Since a dltDacmA double mutant of L. lactis does not lyse, D-alanylationof LTA only affects cell lysis when AcmA is expressed; D-alanylationof LTA is therefore not involved in the observed lysis of thealr acmA double mutant. The lysis of this double mutant is mostlikely caused by the defective peptidoglycan synthesis, whichmakes the cell wall more sensitive to other factors, including,e.g., one or more of the (putative) cell wall hydrolases otherthan AcmA.

View larger version (38K):
[in this window]
[in a new window]
FIG. 7. Schematic representation of the main conclusions of this paper. (A) In L. lactis MG1363 (and, for that matter, L. lactis NZ3900) AcmA hydrolyses peptidoglycan, resulting in cell lysis and subsequent protein release. AcmA is degraded by HtrA in the C-terminal domain which contains the LysM motifs (small black ovals). AcmA degradation results in a less active enzyme. (B) Since no AcmA is present in the acmA mutant of L. lactis MG1363, no lysis and protein release are observed for this strain. (C) In the dltD mutant of L. lactis, the reduction of D-alanylation of LTA results in a strong reduction of degradation of AcmA by HtrA. As a result, increased cellular lysis and protein release are observed. When AcmA is deleted in this dltD mutant, no lysis is observed (not shown in this figure, but compare with panel B). (D) An alr acmA double mutant, however, lyses, although AcmA is not present in this strain. The lysis of this strain is caused by an unknown factor. The high cellular lysis of the alr single mutant is most likely a combination of the reduced degradation of AcmA by HtrA, as shown in panel C and the AcmA-independent lysis as shown in panel D. A, AcmA; H, HtrA.

ACKNOWLEDGMENTS

This research has been carried out with financial support fromthe Commission of the European Communities, specific RTD projectsLABVAC (BIO4-CT96-0542), LABDEL (QLK3-2000-00340) and FAIR (CT98-4396).The content of this article does not necessarily reflect theCommission's views and in no way anticipates the its futurepolicy in this area.

E.P. holds a doctoral fellowship from FRIA. P.H. holds a scientificcollaborator fellowship from FNRS.

We are grateful to K. Schanck for skillful help in the constructionand characterization of the stable alr mutant. We thank E. Maguinfor providing strain MG1362 (dltD) and P. Renault for sequenceinformation on the alr gene from L. lactis IL1403 precedingpublication. We thank Naomi Kramer, Siegfried Morath, ThomasHartung, Eddy J. Smid, Eefjan Breukink, Jan Kok, and Oscar P.Kuipers for sharing the results on the D-Ala determination priorto publication.

FOOTNOTES

* Corresponding author. Mailing address: Molecular Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Groningen, The Netherlands. Phone: 31 50 3632287. Fax: 31 50 3632348. E-mail: G.Buist{at}biol.rug.nl.

REFERENCES

Top