Novel Listerial Glycerol Dehydrogenase- and Phosphoenolpyruvate-Dependent Dihydroxyacetone Kinase System Connected to the Pentose Phosphate Pathway

Abstract

Several bacteria use glycerol dehydrogenase to transform glycerol into dihydroxyacetone (Dha). Dha is subsequently converted into Dha phosphate (Dha-P) by an ATP- or phosphoenolpyruvate (PEP)-dependent Dha kinase. Listeria innocua possesses two potential PEP-dependent Dha kinases. One is encoded by 3 of the 11 genes forming the glycerol (gol) operon. This operon also contains golD (lin0362), which codes for a new type of Dha-forming NAD⁺-dependent glycerol dehydrogenase. The subsequent metabolism of Dha requires its phosphorylation via the PEP:sugar phosphotransferase system components enzyme I, HPr, and EIIA^Dha-2 (Lin0369). P∼EIIA^Dha-2 transfers its phosphoryl group to DhaL-2, which phosphorylates Dha bound to DhaK-2. The resulting Dha-P is probably metabolized mainly via the pentose phosphate pathway, because two genes of the gol operon encode proteins resembling transketolases and transaldolases. In addition, purified Lin0363 and Lin0364 exhibit ribose-5-P isomerase (RipB) and triosephosphate isomerase activities, respectively. The latter enzyme converts part of the Dha-P into glyceraldehyde-3-P, which, together with Dha-P, is metabolized via gluconeogenesis to form fructose-6-P. Together with another glyceraldehyde-3-P molecule, the transketolase transforms fructose-6-P into intermediates of the pentose phosphate pathway. The gol operon is preceded by golR, transcribed in the opposite orientation and encoding a DeoR-type repressor. Its inactivation causes the constitutive but glucose-repressible expression of the entire gol operon, including the last gene, encoding a pediocin immunity-like (PedB-like) protein. Its elevated level of synthesis in the golR mutant causes slightly increased immunity against pediocin PA-1 compared to the wild-type strain or a pedB-like deletion mutant.

INTRODUCTION

Bacteria usually take up glycerol via facilitated diffusion catalyzed by the glycerol facilitator GlpF. Depending on whether they are exposed to aerobic or anaerobic conditions, some bacteria, such as Escherichia coli, either first phosphorylate intracellular glycerol to glycerol-3-P by the enzyme glycerol kinase GlpK (10, 36) or first oxidize it to dihydroxyacetone (Dha) by means of a glycerol dehydrogenase (7). In the latter case, Dha is subsequently converted into dihydroxyacetone phosphate (Dha-P) by a Dha kinase (DhaK). Many proteobacteria, such as Citrobacter freundii (42) and Klebsiella pneumoniae (25), contain an approximately 60-kDa DhaK composed of two domains. The N-terminal domain covalently binds Dha by forming a hemiaminal bond with the Nϵ2 of a histidyl residue (19), whereas the C-terminal domain binds the phosphoryl donor ATP. ATP-requiring Dha kinases are also found in plants, fishes, and mammals. Most firmicutes, many actinobacteria, and several enterobacteria possess a similar Dha kinase, which, however, is split into two subunits, DhaK and DhaL. DhaK covalently binds Dha by forming a hemiaminal bond similar to the N-terminal domain of fused DhaKs (19, 43). DhaL corresponds to the C-terminal domain of fused Dha kinases and carries ADP or ATP in the active site. The most significant difference between the two types of Dha kinases is that in the case of the split enzyme, ADP remains tightly bound to DhaL after Dha phosphorylation and is therefore only very slowly replaced with an ATP molecule (3). ADP therefore rather functions as a cofactor, and for efficient Dha phosphorylation, it needs to be rephosphorylated to ATP, which is achieved in a phosphoenolpyruvate (PEP)-dependent reaction catalyzed by proteins of the PEP:sugar phosphotransferase system (PTS).

PTS proteins catalyze the uptake and concomitant phosphorylation of carbohydrates and also carry out numerous regulatory functions (12). The PTS is usually composed of four soluble proteins or protein domains, which form a phosphorylation cascade. Enzyme I (EI) autophosphorylates with PEP at a histidyl residue and transfers the phosphoryl group to HPr, which in turn phosphorylates one of usually several sugar-specific EIIAs present in a bacterium. Finally, an EIIB of the same sugar specificity accepts the phosphoryl group from P∼EIIA and transfers it to a sugar molecule bound to the corresponding membrane-spanning EIIC (in some cases, EIIC and EIID). The phosphorylation of the PTS proteins occurs at conserved histidyl residues, except for most EIIBs, which are phosphorylated at a cysteyl residue (34). While the phosphorylation of PTS substrates is coupled to their transport, Dha is not transported by the PTS and requires PTS components only for its phosphorylation. The number and domain organization of the PTS proteins participating in Dha phosphorylation vary depending on the organism (18). Operons containing the dhaK and dhaL genes for a Dha kinase composed of two distinct proteins also contain dhaM, a gene encoding an EIIA of the mannose class PTS called EIIA^Dha. Two domains resembling HPr (HPr^Dha) and the N-terminal part of EI (the first approximately 230 amino acids), containing the autophosphorylation but not the PEP binding site (EI^Dha), are sometimes fused to EIIA^Dha. This is the case, for example, for the PEP-dependent Dha phosphorylation systems of E. coli (21, 35) and K. pneumoniae. In these Dha kinase systems, the phosphoryl transfer occurs mainly from PEP via the common PTS components EI and HPr to the truncated EI^Dha and the HPr^Dha domains present in DhaM before it is finally transferred to the EIIA^Dha domain (21). The direct phosphorylation of HPr^Dha by the general PTS protein EI probably occurs in some actinobacteria (Corynebacterium diphtheriae, Arthrobacter chlorophenolicus, and Leifsonia xyli), which contain only HPr^Dha fused to the EIIA^Dha domain. DhaM of another actinobacterium, Mycobacterium smegmatis (MSMEG_2121), contains an entire EI fused to the EIIA^Dha and HPr^Dha domains and therefore might itself bind PEP and autophosphorylate. Finally, most firmicutes and several actinobacteria (streptomycetes and propionibacteria) possess a DhaM protein composed of only EIIA^Dha, which therefore was proposed previously to become directly phosphorylated by the general PTS components EI and HPr (49). Only very few archaea contain PTS components functional in carbohydrate transport and phosphorylation. Nevertheless, a split Dha kinase and the general PTS components EI, HPr, and EIIA^Dha, required for its phosphorylation, are also present in some archaea, such as Haloferax volcanii, Haloquadratum walsbyi, and Halorubrum lacusprofundi (49). After its phosphorylation, EIIA^Dha transfers the phosphoryl group to the ADP molecule in DhaL (21), and the resulting ATP:DhaL holoenzyme phosphorylates Dha bound to DhaK. The dependence of Dha phosphorylation on a functional DhaM was demonstrated previously for E. coli, where the replacement of the phosphorylatable His in any one of the three DhaM domains led to a loss of Dha phosphorylation (21).

Some proteobacteria, such as K. pneumoniae, contain an ATP-dependent (25) and a split PEP-requiring Dha kinase. Interestingly, the K. pneumoniae genes encoding the PEP- and ATP-dependent Dha kinases are located in the same operon. In addition, glycerol kinase often efficiently phosphorylates Dha, as was shown previously for the Enterococcus faecalis enzyme (13), whereas Dha kinases discriminate between glycerol and Dha, owing probably to the hemiaminal bond formed with Dha. Glycerol lacks a keto group and therefore cannot form a hemiaminal bond. Dha utilization by an E. coli glpK mutant depends strictly on PEP and a functional EI. Dha was therefore erroneously thought not only to be phosphorylated but also to be transported by the PTS (24). However, the uptake of extracellular Dha probably occurs via facilitated diffusion catalyzed by the glycerol facilitator GlpF or homologues of GlpF. In several bacteria, such as Lactococcus lactis and a few streptococci, a gene encoding a protein with significant similarity to GlpF is associated with the dha operon. Enterococci and clostridia also contain in the dha operon a gene encoding a glycerol dehydrogenase resembling the E. coli glycerol dehydrogenase GldA (13). Glycerol is converted by GldA into dihydroxyacetone, which is subsequently phosphorylated to Dha-P (6, 40). Enterococci possesses a glycerol dehydrogenase/DhaKLM system as well as GlpK (48) and the glycerol-3-P oxidase GlpO (8). In strain JH2-2, both pathways contribute to glycerol utilization under aerobic conditions. However, under anaerobic conditions, only the glycerol dehydrogenase/DhaKLM system was reported to be operative (6).

In numerous bacteria, the GlpK-mediated metabolism of glycerol also depends on the general PTS components EI and HPr. In enterobacteria, unphosphorylated EIIA^Glc was shown previously to interact with GlpK and to inhibit its activity, whereas the phosphorylation of EIIA^Glc by EI and HPr prevents its inhibitory effect on GlpK activity (23). A different mechanism of regulation is operative in most Gram-positive bacteria. In these organisms, the phosphorylation of GlpK by P∼His-HPr at a conserved histidyl residue is required for GlpK activation, and the absence of phosphorylation leaves the enzyme about 10-fold less active (11). As a consequence, in both types of organisms, mutants defective in EI or HPr phosphorylate glycerol only very slowly. This mode of regulation was called inducer exclusion, because glycerol-3-P, the inducer of the glp regulon, is barely formed when the use of glucose or other rapidly metabolizable carbon sources leads to a dephosphorylation of the PTS components (12).

For the utilization of glycerol and/or Dha, Listeria innocua possesses two PEP-dependent Dha systems and a glpFK operon. The genes glpFK and the dha-1 operon encoding the common DhaK, DhaL, and EIIA^Dha (DhaM) components (19) are found in all listerial species, including the pathogen Listeria monocytogenes. The second set of dha-like genes (dha-2) is part of a large transcription unit containing, in addition to dhaKLM-2, seven or eight other genes, one of which (golD) encodes a novel glycerol dehydrogenase. This operon is present in 5 of the 7 listerial species for which genome sequences have been determined (it is absent from Listeria grayi and Listeria marthii). Here we demonstrate that EI and HPr together with L. innocua EIIA^Dha-2, DhaK-2, and DhaL-2 can efficiently phosphorylate Dha formed from glycerol by the enzyme GolD. The established or deduced activities of the other enzymes encoded by this operon suggest that they catalyze the metabolism of glycerol mainly via the pentose phosphate pathway in the sense of the Calvin cycle, i.e., the production of ribulose-5-P from the two triosephosphates Dha-P and glyceraldehyde-3-P.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

L. innocua strain CLIP 11262 and mutants derived from it as well as L. monocytogenes strain EGD-e were grown in brain heart infusion (BHI) broth or in Luria-Bertani (LB) medium supplemented with either 50 mM glucose, glycerol, or Dha. L. monocytogenes strain EGD-e was also grown in minimal medium complemented with glucose or glycerol (38). E. coli strain NM522 (Stratagene) was used for cloning experiments and for protein overexpression. NM522 transformed with pQE30 (Qiagen) harboring various genes of the gol operon was grown at 37°C under agitation in LB medium supplemented with 100 μg/ml ampicillin.

RNA isolation, reverse transcriptase, and quantitative reverse transcriptase PCR experiments.

The extraction of total RNA from L. monocytogenes EGD-e and various L. innocua strains was performed on cells grown to the exponential phase in 30 ml LB or minimal medium in the presence or absence of 50 mM glucose, glycerol, or Dha, and cells were collected by centrifugation. RNA from the different strains was extracted by using the RNeasy minikit (Qiagen). Residual genomic DNA was digested with the Turbo DNA-free kit (Ambion). Reverse transcription (RT) experiments were performed by using Ready-To-Go RT-PCR beads (GE Healthcare) by incubating 1 μg of total RNA with 2 μg of random hexamers for 30 min at 42°C to allow cDNA synthesis. For the amplification of intergenic regions during RT experiments, primers were designed based on the coding sequence of the selected genes (Table 1). RT-PCR amplification of the following fragments covering one or several intergenic regions was carried out: 805 bp for lin0359-lin0360, 908 bp for lin0360-lin0362, 329 bp for lin0362-lin0363, 901 bp for lin0363-lin0364, 562 bp for lin0364-lin0365, 994 bp for lin0366-lin0369, and 325 bp for the coding region of lin0359. The RT-PCR products were analyzed by electrophoresis on 1% agarose gels. RNA extracts were tested for DNA contamination by preincubating the reaction mixture at 95°C for 10 min to inactivate the reverse transcriptase before the addition of hexamers and RNA. The expression of the gol operons of L. innocua and L. monocytogenes was determined by quantitative RT-PCR experiments on a LightCycler instrument (Roche) by using primers specific for either the dhaK-2 or the pedB-like gene (Table 1) and SuperScript III reverse transcriptase (Invitrogen), according to the protocol provided for the LightCycler Faststart DNA Master SYBR green I kit (Roche). Data were processed with Roche Molecular Biochemicals LightCycler software.

Table 1.

List of primers used in this study

Primer	Sequencea	Use
rpoB_FRT	GCGGATGAAGAGGATAATTACG	RT-PCR
rpoB_RRT	GGAATCCATAGATGGACCGTT	RT-PCR
DhaM-MK	CAATCTCTCGTTAATCCATCC	RT-PCR
DhaK-MK	CTTCTACTGCTTCATACCC	RT-PCR
DhaK-KL	GGAATCAAACGTACTTCCATCG	RT-PCR
DhaL-KL	GTAAGGTTGCAATATCGGG	RT-PCR
DhaK-RTF	CGAGCAATCGGTCATCTTGATCC	RT-PCR
lin0363-RTR	GTTCGTAACCCATTTCGTCACAAC	RT-PCR
lin0363-RTF	CGCCGTTTCGCTACTTGATAC	RT-PCR
lin0362-RTR	CGACTTGAAGTGCGAGCG	RT-PCR
lin0362-RTF	CTCTTTTTAGTAAGTGACG	RT-PCR
lin0360-RTR	CCAAAGCGAGAAAGCCATTG	RT-PCR
lin0360-RTF	GCGCAGCAAAAGCGAAAACCGG	RT-PCR
PedB1	GAAAAAGAATACTCCAGAGGC	RT-PCR
PedB2	GCTTGGAAATTGATTGGTC	RT-PCR
GolRint1 (BamHI)	TACGGATCCAAAATAACAAAATAACTGTACC	golR insertion
GolRint2 (EcoRI)	TAAGAATTCAAATCGGTTAAGCCTTCTTCC	golR insertion
GolR_ext3′	GTATGATCTAGAAGCGCAACC	Confirmation of golR mutant
GolR_ext5′	GAAATCATTGATAGGTAAAAGG	Confirmation of golR mutant
pedB1_BamHI	TCTGGATCCGAAGTACACC	pedB deletion
pedB1_SalI	TGGAAATTGATTGTCGACGAATAAATCTATAAGTGCC	pedB deletion
pedB2_SalI	ATAGATTTATTCGTCGACAATCAATTTCCAAGC	pedB deletion
pedB2_NcoI	CAACCATGGAAAGGAGAAAATCACATG	pedB deletion
pedBext5′	GCAAAAGCGAAAACCGG	pedB mutant verification
pedBext3′	TGTTAGTTAAAGTTTTTG	pedB mutant verification
DhaM_BamHI	AGCGGATCCATGATAAGTATTGTTTTAG	DhaM-2 purification
DhaM_SalI	AATGTCGACTTATTACTTTTTTAGTGC	DhaM-2 purification
DhaK_BamHI	GGAGGATCCATGAGACGTTTAGTGAATG	DhaK-2 purification
DhaK_SalI	TTAGTCGACCAGTAGTAATTTAGTCCAC	DhaK-2 purification
DhaL_BamHI	GGAGGATCCATGAGCGAATTAGTTATGGATAGCG	DhaL-2 purification
DhaL_SalI	AAGGTCGACCGCATAAATCAGACCGC	DhaL-2 purification
tpi_like_BamHI	GTCGGATCCATGCGTAAACCCTTAGTTGG	Tpi-2 purification
tpi_like_SalI	ATTGTCGACCTAACTCATCCTTTCGTTTTG	Tpi-2 purification
lin0362_BamHI	TAAGGATCCATGACTTTTAAAGGTTTT	Lin0362 purification
lin0362_PstI	AAGCTGCAGACTTTATTTAATCGTATAACCGC	Lin0362 purification
lin0363_SalI	TGAGTCGACATGAAAATCGCAATTGG	Lin0363 purification
lin0363_PstI	CCACTGCAGTTTTTATCTATTTTTTTGG	Lin0363 purification
rpoB-F2	CAGAAACTGGTGAAATTATCGC	Quantitative PCR
rpoB-R2	TAGATTGAACGAGTACGCTATCTT	Quantitative PCR
QDhaKF	AATGCAGATCCAATCGAAACT	Quantitative PCR dhaK-2
QDhaKR	CCAAGACGATTCAGAGCC	Quantitative PCR dhaK-2
QPedBF	GTCGAATTAGAAGCTCGTTCTG	Quantitative PCR pedB-like
QPedBR	TGCTTGGAAATTGATTGGTCAC	Quantitative PCR pedB-like
DpQE	GTTCTGAGGTCATTACTGG	Sequencing pQE-30 inserts
FpQE	CGGATAACAATTTCACACAG	Sequencing pQE-30 inserts
pMAD_5′	GCGAGAAGAATCATAATGGG	Sequencing pMAD inserts
pMAD_3′	GTTACACATTAACTAGACAG	Sequencing pMAD inserts

^aRestriction sites are indicated in boldface type.

Construction of lin0370 (golR) and lin0359 (pedB-like) mutants.

In order to construct a pedB-like deletion mutant, the regions upstream and downstream of the pedB-like gene (446 and 511 bp, respectively) were amplified by PCR by using L. innocua DNA as the template and the two primer pairs pedB1_BamHI/pedB1_SalI and pedB2_SalI/pedB2_NcoI (Table 1). The two amplified DNA fragments were cut with the indicated restriction enzymes and cloned into plasmid pMAD (2) digested with the corresponding enzymes. The correct sequence of the two amplified DNA fragments was verified by DNA sequencing. pMAD-derived plasmid pMAD-pedB-like was subsequently used to transform L. innocua strain CLIP 11262 by electroporation. A pedB-like deletion mutant was obtained by double recombination according to a previously described protocol established for L. monocytogenes (1). The correct deletion of the pedB-like gene was verified by carrying out PCR amplification using primers pedBext3′ and pedBext5′ and DNA sequencing of the amplified DNA fragment.

In order to construct a golR (lin0370) disruption mutant, an internal 526-bp fragment of golR was amplified by PCR by using L. innocua DNA as the template and primers GolRint1 and GolRint2 (Table 1). The amplified DNA fragment was digested with the restriction enzymes BamHI and EcoRI and cloned into plasmid pMAD cut with the same enzymes. The correct sequence of the insert in plasmid pMAD-golR was verified by DNA sequencing. pMAD-golR was subsequently used to transform L. innocua strain CLIP 11262 by electroporation. A golR insertion mutant was obtained by single recombination according to a previously described protocol (1). The correct insertion of pMAD-golR into this mutant was verified by carrying out PCR amplification with primers GolR_ext3′ and GolR_ext5′, followed by DNA sequencing of the amplified DNA fragment.

Overproduction and purification of proteins.

His-tagged HPr and EI from Bacillus subtilis were overproduced and purified by using plasmids pAG2 and pAG3, respectively (18). The genes encoding the L. innocua proteins DhaK-2, DhaL-2, EIIA^Dha-2, Lin0362 (GolD), Lin0363 (RpiB), and Lin0364 (triosephosphate isomerase 2 [Tpi-2]) were first amplified by PCR by using appropriate primers to introduce the restriction sites BamHI and SalI (BamHI and PstI for Lin0364) at the ends of the amplified DNA fragments (Table 1), which allowed their insertion into the His tag expression vector pQE30 (Qiagen) cut with the two corresponding restriction enzymes. The resulting plasmids were used to transform E. coli strain NM522. The expression of the various genes by induction with isopropyl-β-d-thiogalactopyranoside (IPTG) and the purification of the His-tagged proteins by ion chelate affinity chromatography were carried out as previously described (18).

Separation of phosphorylated and unphosphorylated proteins by electrophoresis on polyacrylamide-urea gels.

The phosphoryl group transfer from PEP to EI (P∼EI), to HPr (P∼His₁₅-HPr), and, finally, to DhaM-2 (P∼EIIA^Dha-2) was monitored by using reaction mixtures containing 50 mM Tris-HCl (pH 7.4), 1 mM PEP, 5 mM MgCl₂, 3 μg EI, and 3 μg HPr. These assay mixtures were preincubated for 20 min at 37°C to allow the efficient phosphorylation of HPr before DhaM-2 was added, and incubation was continued for another 20 min at 37°C. Reaction mixtures lacking one or two of the PTS proteins were also prepared. Phosphorylated and unphosphorylated proteins were separated by electrophoresis on 10% polyacrylamide–8 M urea gels (30). After electrophoresis, gels were stained with Coomassie brilliant blue.

Spectrophotometric assay of Dha phosphorylation.

In order to measure the PEP-dependent phosphorylation of Dha catalyzed by EI, HPr, DhaM-2, DhaK-2, and DhaL-2, we set up a coupled spectrophotometric assay. Dha-P formed by the PEP-requiring phosphorylation cascade is converted into glycerol-3-P by the enzyme glycerol-3-P dehydrogenase (Sigma). This reaction requires NADH, and its conversion to NAD⁺ was monitored by measuring the decrease of the absorbance at 340 nm (A₃₄₀) with a Kontron Bio-Tek spectrophotometer over a time period of 60 min. The 700-μl reaction mixtures contained 75 mM Tris-HCl (pH 7.4), 6 mM PEP, 1 mM NADH, 6 mM Dha, 10 mM MgCl₂, 1 U of glycerol-3-P dehydrogenase, 30 μg EI, 35 μg HPr, and 30 μg of the DhaK-2, DhaL-2, and DhaM-2 proteins. Reaction mixtures lacking any one of the three PTS proteins or DhaK or DhaL were prepared. After preincubation for 10 min at room temperature, the lacking protein was added, and the decrease of the A₃₄₀ was monitored for 60 min and compared to that of a reaction mixture which had not been complemented. A control experiment with all proteins but without Dha was also carried out.

Spectrophotometric assays of glycerol dehydrogenase, triosephosphate isomerase, ribose-5-P isomerase, and ribulose-5-P 3-epimerase activities.

The glycerol dehydrogenase activity of Lin0362 (GolD) was measured by using a 700-μl assay mixture containing 75 mM Tris-HCl (pH 7.4), 1 mM NAD⁺ or NADP⁺, 1.2 mM MgCl₂ or MnCl₂, and 1.2 mM glycerol. After preincubation for 5 to 10 min at room temperature, 40 μg of purified GolD was added, and the increase of the A₃₄₀ was monitored with a Kontron Bio-Tek spectrophotometer. We also measured the reverse dihydroxyacetone reductase activity of GolD by replacing NAD⁺ with NADH and by replacing glycerol with Dha. The reverse reaction was monitored by measuring the decrease of the A₃₄₀. All necessary control experiments were also carried out.

In order to measure the activity of Tpi-2, we set up another coupled spectrophotometric assay. We started from fructose-1,6-bisphosphate (FBP), which is split by the enzyme FBP aldolase into glyceraldehyde-3-P and Dha-P. As mentioned above, Dha-P is converted into glycerol-3-P by the enzyme glycerol-3-P dehydrogenase in an NADH-consuming reaction. Once all the FBP-derived Dha-P had been converted into glycerol-3-P (no further decrease of the A₃₄₀), we added either commercially available triosephosphate isomerase (Sigma) or purified L. innocua Tpi-2. Triosephosphate isomerase converts glyceraldehyde-3-P into Dha-P, and as a consequence, we expected a second decrease of the A₃₄₀. The reaction mixture contained 90 mM Tris-HCl (pH 7.4), 1 mM NADH, 0.2 mM FBP, 1 U of glycerol-3-P dehydrogenase, and 0.3 U of FBP aldolase. Under the reaction conditions employed, we observed a decrease of the A₃₄₀ of about 0.7 in the absence of Tpi. When either 1.5 U of commercially available Tpi or 20 μg of L. innocua Tpi-2 was added to the reaction mixture, a second decrease of the A₃₄₀ of about 0.7 was observed, which did not occur when Tris-HCl (pH 7.4) was added in place of the Tpi enzymes.

Finally, the ribose-5-P isomerase activity was measured by using a 700-μl assay mixture containing 50 mM Tris-HCl (pH 7.4), 0.5 mM NADH, 1.2 mM MgCl₂, and 0.5 mM ribose-5-P. After preincubation for 5 to 10 min at room temperature, 6 μg of purified Lin0363 (RpiB) was added, which was expected to transform ribose-5-P into ribulose-5-P. In order to monitor the formation of ribulose-5-P, the samples also contained 18 μg of Lactobacillus casei ribitol-5-P dehydrogenase, which transforms ribulose-5-P into ribitol-5-P in an NADH-consuming reaction; the decrease of the absorbance at 340 nm was therefore monitored. Because ribitol-5-P dehydrogenase is not commercially available, the gene LCABL_39250 from the presumed L. casei BL23 ribitol operon (32) was cloned into plasmid pQE30, and the His-tagged protein was purified. The presumed ribitol-5-P dehydrogenase activity of the L. casei enzyme was confirmed with ribulose-5-P and NADH as substrates before it was used in the ribose-5-P isomerase assay (A. Bourand and J. Deutscher, unpublished results). To determine whether Lin0363 might possess ribulose-5-P 3-epimerase activity (transformation of xylulose-5-P into ribulose-5-P), we carried out a similar assay by replacing ribose-5-P with xylulose-5-P. All other conditions were identical to those used for the ribose-5-P isomerase assay. All necessary control experiments were also carried out.

Bacteriocin assay.

Bacteriocin activity was measured by using a microtiter plate assay system as previously described (33). Five different purified pediocin-like (class IIa) bacteriocins, kindly provided by Jon Nissen-Meyer, were tested: the pediocin PA-1 K20R variant from Pediococcus acidilactici (22), curvacin A from Lactobacillus curvatus, leucocin A from Leuconostoc mesenteroides, sakacin P from Lactobacillus sakei, and enterocin A from Enterococcus faecium. Immunity against these bacteriocins was tested with wild-type L. innocua, the golR-disrupted mutant, and the pedB-like deletion mutant. Each well of the microtiter plate contained 200 μl of BHI broth, either the indicator strain (wild type) or one of the two mutants at an A₆₀₀ of about 0.01, as well as a specific concentration of one of the bacteriocins tested. The bacteriocin concentrations ranged from 25 nM to 0.05 nM, with a 2-fold difference for each dilution step. The microtiter plates containing the bacterial cultures were incubated for 12 h at 30°C. The growth of the wild-type strain and the two mutants was monitored spectrophotometrically by measuring the absorbance at 600 nm every 10 min with a microtiter plate reader. The MIC, which is defined as the concentration of a bacteriocin that inhibits growth by 50%, was determined for the wild type and the two mutant strains. The experiment was repeated twice.

RESULTS AND DISCUSSION

Listeria innocua contains two potential PEP-dependent Dha kinases.

A BLAST search on the ListiList Web server with the three proteins of the L. lactis Dha phosphorylation system (49) revealed that L. innocua CLIP 11262 and L. monocytogenes EGD-e possess two operons encoding homologues of the lactococcal proteins. One operon (dha-1) has the same gene order as that of L. lactis (dhaK-1, dhaL-1, and dhaM-1; lin2843-lin2845; and lmo2695-lmo2697). The encoded proteins exhibit 61, 53, and 48% sequence identities, respectively, to their L. lactis homologues. The second set of dha-like genes seems to be part of a large operon (dha-2) and exhibits a different gene order (dhaM-2, dhaK-2, and dhaL-2), with two genes encoding presumed membrane proteins of unknown function inserted between lin0369 (dhaM-2) and lin0366 (dhaK-2) for L. innocua (between lmo0351 and lmo0348 for L. monocytogenes) (Fig. 1). The Dha-2 proteins show between 32 and 40% sequence identity compared to their L. lactis homologues. The expression of the dha-2 genes of L. monocytogenes was reported to be upregulated during growth on glycerol, especially during late log phase (between 30- and 180-fold upregulation compared to glucose-grown cells), whereas the dha-1 operon was only slightly induced (26). A similar observation was made during adaptation to a soil environment, where an 8-fold upregulation was detected for the dha-2 genes (37), whereas the expression level of the dha-1 operon was only 2-fold elevated.

Fig 1.

Schematic representation of the lin0359-0369 (gol) operon and the gene encoding its repressor Lin0370 (GolR). Gray arrows indicate the dha-2 genes, and arrows filled with a motif indicate all other genes for which the function has been determined by in vitro or in vivo assays.

Six other genes also seemed to be cotranscribed with the dha-2 genes. They encode a protein with significant similarity to triosephosphate isomerase (Tpi-2 [lin0364]) and proteins resembling enzymes of the pentose phosphate pathway: lin0360 encodes a presumed transketolase, lin0361 encodes a transaldolase, lin0363 encodes a RpiB-type ribose-5-P isomerase (44), and lin0362 encodes enzymes resembling the short-chain dehydrogenase/reductase family. The last gene in the operon (lin0359) encodes a protein with 48% similarity to the pediocin immunity protein from P. acidilactici.

By carrying out RT-PCR with 6 different primer pairs (Table 1), we showed that in the corresponding L. innocua mRNA transcript, each of the 11 genes of the presumed dha operon is linked to its two neighboring genes, except for the distal pedB-like (lin0359) and dhaM-2 (lin0369) cistrons (data not shown). These results strongly suggest that the 11 L. innocua genes are organized into an operon (Fig. 1), which is in good agreement with previously reported L. monocytogenes transcriptome data acquired during soil adaptation (37), where all genes, except for the pedB-like gene, were reported to be transiently upregulated after 15 and 30 min of adaptation to soil extracts or soil microcosms. During growth on glycerol to the late log phase, 7 of the 11 L. monocytogenes genes were also reported to be upregulated (26).

Glycerol-inducible expression of the lmo0341-0351 operon in L. monocytogenes and constitutive expression of the lin0359-0369 operon in the lin0370 mutant.

In order to test whether the expression of the lin0359-lin0369 operon can also be induced in defined growth media, we carried out quantitative RT-PCR experiments. We tested the expression of the lin0359-0369 operon after growth in LB medium and LB medium containing 50 mM Dha, glycerol, or glucose. We carried out quantitative RT-PCR by using two primer pairs allowing the amplification of a 153-bp dhaK-2 DNA fragment and a 175-bp pedB-like DNA fragment. A small amount of dha mRNA was formed when the cells were grown in LB medium without an additional carbon source (Table 2). Even less dha mRNA was detected when the cells were grown in LB medium containing glucose, glycerol, or Dha. Apparently neither glycerol nor Dha induced lin0359-0369 expression during growth in LB medium. Unfortunately, L. innocua strain CLIP 11262 grows very poorly in minimal medium, even when complemented with glucose. Quantitative RT-PCR experiments were therefore carried out with L. monocytogenes strain EGD-e, which grows well in minimal medium, by employing the same dhaK- and pedB-specific primer pairs as those used for L. innocua (100% identity to the corresponding L. monocytogenes sequences). When L. monocytogenes EGD-e was grown in minimal medium with glucose, no significant expression of the dha-2 operon (lmo0341-0351) was observed. However, after reaching the stationary phase, glycerol-grown cells exhibited elevated expression levels of the dhaK-2 (500-fold) and the pedB-like (8-fold) genes (Table 3) compared to those exhibited by cells grown on glucose. No growth was observed in minimal medium containing Dha as the sole carbon source. These results are in agreement with data from transcriptome studies with glycerol-grown L. monocytogenes EGD-e cells, where a strong induction (up to 180-fold) of the genes of the lmo0341-0351 operon by glycerol was also found when the cells entered stationary phase (26). The dha-1 operon is not induced by glycerol, and its inducer is unknown.

Table 2.

Relative expression levels of the dhaK-2 and pedB-like genes determined by quantitative RT-PCR for L. innocua wild-type strain CLIP 11262 and the golR mutant derived from it after growth in LB medium containing 50 mM glucose, glycerol, or Dha

Strain	Mean relative expression level (SD)a
	LB		LB + Glc		LB + Gly		LB + Dha
	dhaK-2	pedB-like	dhaK-2	pedB-like	dhaK-2	pedB-like	dhaK-2	pedB-like
CLIP 11262	0.044 (0.036)	0.027 (0.007)	0.001 (0.001)	0.016 (0.012)	0.005 (0.002)	0.013 (0.009)	0.001	0.011 (0.009)
lin0370 (golR)	1b	1b	0.065 (0.024)	0.111 (0.074)	1.133 (0.423)	0.817 (0.318)	0.680 (0.337)	0.774 (0.117)

^aMean values and standard deviations (in parentheses) calculated from results obtained with at least four independent RNA preparations are shown.

^bTranscript levels were calculated relative to the LB medium-grown golR strain, which was set to 1.

Table 3.

Relative expression levels of the dhaK- and pedB-like genes determined by quantitative RT-PCR of L. monocytogenes wild-type strain EGD-e grown in minimal medium containing 50 mM glucose or glycerol

Strain	Mean relative expression level (SD)a
	MM glucose		MM glycerol
	dhaK-2	pedB-like	dhaK-2	pedB-like
EGD-e	0.002 (0.001)	0.132 (0.094)	1	1

^aMean values and standard deviations (in parentheses) calculated from the results obtained with at least four independent RNA preparations are shown. Transcript levels were calculated relative to the value obtained for glycerol-grown cells, which was set to 1. MM, minimal medium.

The lin0359-0369 operon is preceded by the lin0370 gene (lmo0352 in L. monocytogenes), oriented in the opposite direction and encoding a presumed transcription repressor of the DeoR family, which we suspected to control the expression of the operon. We therefore inactivated lin0370 and tested the effect of this mutation on the expression of the lin0359-0369 operon by quantitative RT-PCR. The inactivation of lin0370 caused 20- and 40-fold-elevated expression levels of the dhaK-2 and pedB-like genes, respectively, after growth in LB medium compared to the wild-type strain (Table 2). A similar increased expression level of the lin0359-0369 operon was observed when the lin0370 repressor mutant was grown in LB medium with glycerol or Dha. The presence of glucose strongly inhibited the constitutive expression of the dhaK-2 and pedB-like genes in the lin0370 mutant (Table 2). This repression is probably due to catabolite control protein A/P-Ser-HPr-mediated carbon catabolite repression (12, 17), because the first gene of the lin0359-0369 operon contains an almost perfect target site (catabolite response element, TGGAAGCGCTATCT) for the catabolite repressor complex in its coding sequence.

lin0367 and lin0368 encode proteins of unknown function, and lin0362 encodes a new type of glycerol dehydrogenase.

For 3 of the 11 proteins encoded by the dha-2 operon, a sequence comparison did not allow the prediction of a function. Two of them are small presumed membrane proteins encoded by lin0367 and lin0368, and homologues are present in most other listeriae. Homologues are also found in members of the orders Thermoanaerobacterales, Lactobacillales, and Clostridiales, where they are usually associated either with a gene encoding a predicted 6-P-β-glucosidase or with a glycerol (glp) or dha operon. For example, in Halanaerobium praevalens, two homologous genes are located in front of the glpK and glpO genes. In Tepidanaerobacter sp. strain Re1, the genes for these two proteins are located in an operon exhibiting the same gene order as that of the B. subtilis glycerol operon (glpPFKD—antiterminator, glycerol facilitator, glycerol kinase, and glycerol-3-P dehydrogenase) (5), with the exception that glpF is replaced with the two genes encoding the small membrane proteins. This organization suggests that in Tepidanaerobacter sp. Re1, the two small membrane proteins might catalyze glycerol uptake either via facilitated diffusion or active transport.

The gene lin0362 encodes a predicted short-chain dehydrogenase/reductase of unknown specificity. In order to determine the substrate of Lin0362, we purified the His-tagged protein as described in Materials and Methods. Enzymes of this family usually catalyze the reversible transformation of HCOH groups in the second position of polyols or related compounds into keto groups. Examples are the Rhodobacter sphaeroides sorbitol dehydrogenase/sorbose reductase (41) or the Gluconobacter oxydans xylitol dehydrogenase/xylulose reductase (45). We therefore tested the possibility that Lin0362 might catalyze the NADH- or NADPH-dependent reduction of xylulose-5-P or ribulose-5-P, two intermediates of the pentose phosphate pathway, into xylitol-5-P or ribitol-5-P, respectively. However, no such activity could be measured with the purified protein (data not shown). Because glycerol also contains an oxidizable HCOH group in its second position and because this triol induces the expression of the L. monocytogenes dha-2 operon (lmo0341-0351) (Table 3) (26), we tested the possibility that the protein Lin0362 might be a novel type of glycerol dehydrogenase/dihydroxyacetone reductase. When we tested this hypothesis, we indeed found that Lin0362 catalyzes the NAD⁺-requiring transformation of glycerol into Dha as well as the reverse reaction, the NADH-dependent reduction of Dha to glycerol. As observed previously for other alcohol dehydrogenase/keto reductase-catalyzed reactions, the equilibrium was shifted toward the formation of glycerol and NAD⁺, because the standard free energy strongly favors the formation of the alcohol (4). As a consequence, the formation of NADH during glycerol dehydrogenation rapidly slowed, owing to the preferred backward reaction. In addition, the initial rate of the NADH-dependent reduction of Dha was found to be nearly 10 times higher than the rate of the NAD⁺-dependent oxidation of glycerol (data not shown). For the efficient oxidation of glycerol inside the cells, the Dha formed will therefore have to be rapidly dissimilated, most likely by phosphorylation to Dha-P (Fig. 2, top right). The enzyme Lin0362 specifically uses NAD⁺ and NADH, because no activity was observed when its two functions were tested with NADP⁺ and NADPH, respectively. Divalent cations were necessary for both activities of Lin0362, and the Mg²⁺ and Mn²⁺ ions had similar stimulating effects, whereas the enzyme was inactive in the presence of Zn²⁺.

Fig 2.

Proposed model of glycerol metabolism by the enzymes encoded by the L. innocua gol operon. Glycerol taken up via facilitated diffusion possibly catalyzed by the small membrane proteins Lin0367 and Lin0368 or by GlpF encoded by the glpFK operon is intracellularly oxidized to Dha by the enzyme glycerol dehydrogenase (GolD). The PEP-requiring PTS phosphorylation cascade composed of EI, HPr, and EIIA^Dha-2 is connected to DhaL-2::ADP, which is converted to DhaL-2::ATP and subsequently catalyzes the phosphorylation of Dha bound to DhaK-2. Tpi-2 catalyzes the interconversion of Dha-P and glyceraldehyde-3-P (Gap). The two triosephosphates are condensed to FBP, which is subsequently converted into fructose-6-P (F-6-P). The enzyme encoded by the lin0360 gene resembles transketolases and probably allows entry into the pentose phosphate pathway (PPP) by catalyzing the formation of xylulose-5-P and erythrose-4-P from F-6-P and glyceraldehyde-3-P.

Similar to the dha operon in most enterococci and clostridia (6, 40), the L. innocua lin0359-lin0369 and L. monocytogenes lmo0341-lmo0351 operons contain a gene encoding a protein with NAD⁺-dependent glycerol dehydrogenase activity. However, while the enterococcal and clostridial enzymes exhibit between 40 to 60% sequence identity compared to glycerol dehydrogenase (GldA) from firmicutes (31) or enterobacteria (46), the listerial protein is, to the best of our knowledge, the first member of the short-chain dehydrogenase/reductase family exhibiting glycerol dehydrogenase/dihydroxyacetone reductase activity. The L. innocua lin0359-lin0369 and L. monocytogenes lmo0341-lmo0351 operons are therefore likely to encode a glycerol utilization system, which in the first step transforms glycerol into dihydroxyacetone, which is subsequently probably converted into Dha-P by the DhaKLM-2 proteins (Fig. 2). We therefore called Lin0362 GolD, for glycerol dehydrogenase, and the genes lin0359 to lin0369 the gol operon.

In vitro reconstitution of PEP-dependent Dha phosphorylation.

Based on the above-described results, we wanted to test whether the DhaKLM-2 proteins would indeed be able to phosphorylate Dha. For that purpose, we purified EI and HPr from B. subtilis and the three Dha-2 proteins from L. innocua. We first tested whether EI and HPr would be able to phosphorylate DhaM-2 (EIIA^Dha-2), encoded by lin0369. After the incubation of the three PTS proteins in the presence of PEP as the phosphoryl donor, the samples were separated on a 10% polyacrylamide gel containing 8 M urea. This gel system often allows the separation of phosphorylated and unphosphorylated proteins (30). Indeed, HPr incubated with EI and PEP migrates much faster than unphosphorylated HPr (Fig. 3, compare lanes 1 and 2). When the EIIA^Dha-2 protein was incubated with PEP, EI, and HPr (Fig. 3, lane 4), it also exhibited a slightly higher electrophoretic mobility than the EIIA^Dha-2 protein incubated with PEP and HPr (Fig. 3, lane 3) or only PEP (Fig. 3, lane 5). These results established that the EIIA^Dha-2 protein becomes phosphorylated by PEP, EI, and HPr.

Fig 3.

Phosphorylation of DhaM-2 (EIIA^Dha) detected by electrophoresis on an 8 M urea–10% polyacrylamide gel. All samples were loaded onto the gel in a volume of 25 μl containing 50 mM Tris-HCl (pH 7.4), 1 mM PEP, and 5 mM MgCl₂ and the proteins indicated for each lane: 1 μg of EI, 3 μg of HPr, and 3 μg of DhaM-2. Samples were incubated at 37°C as described in Materials and Methods. The drastic change of the migration behavior caused by the phosphorylation of HPr was described previously (30) (compare lanes 1 and 2). The difference in electrophoretic mobility between DhaM-2 and P∼DhaM-2 is less pronounced but still clearly visible (compare lanes 3, 4, and 5).

We next wanted to test whether Dha would become phosphorylated when the DhaK-2 and DhaL-2 proteins were included in the phosphorylation assay. We set up a coupled spectrophotometric test by also adding glycerol-3-P dehydrogenase to the assay mixture, which we expected to convert the P-Dha formed by the DhaKLM-2 system into glycerol-3-P in an NADH-consuming reaction. We indeed observed a decrease of the A₃₄₀ in the sample when all proteins were present (Fig. 4), whereas no change of the A₃₄₀ occurred when any one of the five proteins was lacking (see “−DhaK” in Fig. 4; identical straight lines were obtained when either one of the other proteins was lacking). These results confirm that Dha is indeed the substrate for this second Dha kinase system. Our results also confirm that the phosphoryl group transfer from PEP to Dha occurs in five steps, as depicted in Fig. 2 (left). While the oxidation of glycerol by GolD is reversible, and the equilibrium is shifted toward glycerol and NAD⁺, the phosphorylation of Dha by the PEP-dependent Dha kinase is almost irreversible owing to the high free energy of PEP. The efficient Dha kinase reaction therefore also provides the driving force for the glycerol dehydrogenase reaction.

Fig 4.

Spectrophotometric assays of Dha phosphorylation. Spectrophotometric assays were carried out at room temperature by monitoring the decrease in the amount of NADH at an A₃₄₀ over a time period of 60 min by using a Kontron Bio-Tek spectrophotometer. All samples consisted of 700-μl reaction mixtures containing 75 mM Tris-HCl (pH 7.4), 6 mM PEP, 1 mM NADH, 6 mM Dha, 10 mM MgCl₂, and 1 U of glycerol-3-P dehydrogenase. The following five proteins were assumed to be required for Dha phosphorylation: EI, HPr, DhaK-2, DhaL-2, and DhaM-2 (EIIA^Dha). Samples lacking any one of the above-mentioned five proteins were prepared in duplicates and preincubated for 10 min before the lacking protein was added to one of the two samples. The decrease of the A₃₄₀ was monitored for both samples. All complemented samples exhibited a more or less rapid decrease of the A₃₄₀. For the control assays, we show only the sample lacking DhaK-2 (−DhaK-2), because a similar straight line was obtained for all other “noncomplemented” samples. These results confirm that all five bacterial proteins are necessary for Dha phosphorylation.

When we carried out the spectrophotometric assays, we usually preincubated the assay mixture for 5 min before the reaction was started by the addition of the lacking enzyme. Interestingly, when HPr or DhaL was used to start the reaction, NADH consumption occurred at a significantly reduced rate (about one-third) compared to the reaction velocity obtained with the other three “starter” proteins (Fig. 4). A possible explanation for this unexpected behavior might be that during preincubation, different protein complexes are preformed, depending on which enzyme is lacking. The formation of a strong complex between EIIA^Dha and DhaL has indeed been reported for the L. lactis proteins, and the crystal structure of the transient complex has been determined (49). In addition, HPr is known to form transient complexes with EI and EIIA^Man (EIIA^Man belongs to the same EIIA family as EIIA^Dha), and the solution structures of the corresponding E. coli protein complexes have been determined (20, 47). Because slow Dha phosphorylation was observed when either HPr or DhaL was absent during preincubation, it is likely that the perturbation of the formation of the HPr and DhaL complexes with EIIA^Dha might be responsible for the slower Dha phosphorylation.

The protein encoded by lin0364 exhibits triosephosphate isomerase activity.

Like most firmicutes, L. innocua possesses an operon with genes encoding triosephosphate isomerase and three or four other enzymes of the lower part of the Embden-Meyerhof-Parnas pathway. However, as mentioned above, the sequence of the protein encoded by lin0364 also exhibits significant similarity to triosephosphate isomerases (29% identity to L. innocua triosephosphate isomerase of the gap-pgk-tpi-pgm-eno operon). We therefore overproduced and purified the lin0364-encoded protein and tested whether it indeed possesses a triosephosphate isomerase function. For that purpose, we used a spectrophotometric assay which starts from FBP, which is split by the enzyme aldolase into glyceraldehyde-3-P and Dha-P. The latter product is immediately reduced to glycerol-3-P in an NADH-dependent reaction catalyzed by the enzyme glycerol-3-P dehydrogenase, which was also included in the assay mixture. Once NADH consumption had stopped (no further decrease of the A₃₄₀), commercially available triosephosphate isomerase or the purified lin0364-encoded protein was added to the reaction mixture. In both cases, this led to a second phase of NADH consumption and, therefore, to a further decrease of the A₃₄₀, which was of an amplitude (ΔA₃₄₀ = 0.7) identical to that of the first observed decrease (data not shown). This result confirmed that the glyceraldehyde-3-P formed during the aldolase reaction in amounts identical to that of Dha-P is converted by the lin0364-encoded protein into Dha-P before it is reduced to glycerol-3-P. The lin0364-encoded protein therefore possesses triosephosphate isomerase activity, and we called the enzyme triosephosphate isomerase 2 and the corresponding gene tpi-2.

Lin0363 exhibits ribose-5-P isomerase (RpiB) activity.

The Dha-P formed by the Dha kinase system DhaK-2/DhaL-2/EIIA^Dha-2 is probably partially converted by triosephosphate isomerase 2 into glyceraldehyde-3-P. These two glycolytic intermediates might subsequently be fused by the enzyme aldolase, thereby forming FBP, which might be further converted to fructose-6-P. Together with glyceraldehyde-3-P, this glycolytic intermediate is probably transformed into xylulose-5-P and erythrose-4-P by the enzyme transketolase (Lin0360), which provides the entry point into the pentose phosphate pathway (Fig. 2, bottom right). It should be noted that in addition to Lin0360 and Lin0361, listeriae usually possess two other transketolase and transaldolase homologues. In three additional steps catalyzed by three enzymes encoded within the gol operon, the end product of the pentose phosphate pathway, ribulose-5-P, will be formed: transaldolase (Lin0361) produces sedoheptulose-7-P and glyceraldehyde-3-P from erythrose-4-P and fructose-6-P, transketolase (Lin0360) forms ribose-5-P and xylulose-5-P from sedoheptulose-7-P and glyceraldehyde-3-P, and finally, ribose-5-P isomerase B (RpiB [Lin0363]) converts ribose-5-P into ribulose-5-P.

In order to support this concept of glycerol metabolism, we purified the presumed ribose-5-P isomerase B encoded by lin0363 and tested its function as described in Materials and Methods. It indeed exhibited the expected ribose-5-P isomerase activity by catalyzing the interconversion of ribulose-5-P and ribose-5-P (data not shown). The only enzyme lacking in the gol operon-encoded pentose phosphate pathway is ribulose-5-P 3-epimerase, which converts the xylulose-5-P formed during the two transketolase reactions into ribulose-5-P. We therefore tested whether Lin0363 might also possess ribulose-5-P 3-epimerase activity. However, no such activity was exhibited by Lin0363. It is therefore likely that one or several of the five ribulose-5-P 3-epimerase homologues encoded by the listerial genome will complete the pentose phosphate pathway during glycerol utilization via the Dha-2 system.

What is the function of the PedB-like protein encoded by lin0359?

The gol operon seems to be specific for listeriae, because we detected operons encoding triosephosphate isomerase 2, the novel type of glycerol dehydrogenase GolD, and enzymes of the Dha kinase system and the pentose phosphate pathway in only four other listerial species, including L. monocytogenes, Listeria seeligeri, Listeria ivanovii, and Listeria welshimeri. The gol operon is lacking in the L. grayi and L. marthii strains, for which the genome sequences have been determined, as well as in L. ivanovii strain PAM55, while it is present in L. ivanovii strain FSL F6-596. When present, the Dha-2 proteins of these organisms exhibit between 83% and 96% sequence identities compared to the corresponding L. innocua proteins. While the gol operons of L. seeligeri, L. ivanovii, and L. welshimeri lack the distal pedB-like gene, it is present in the gol operon of all three sequenced L. innocua strains. L. monocytogenes takes an intermediary position because out of the 27 strains for which the genomes have been sequenced, only 12 contain a pedB-like gene in their gol operons.

The mode of action of the immunity proteins has not yet been entirely elucidated, but each immunity protein seems to act more or less specifically against its corresponding bacteriocin (14). Interestingly, an extracellular loop of the glucose/mannose-specific EIIC PTS component seems to be involved in the specific targeting by class IIa bacteriocins (28, 39), and the corresponding mutants therefore exhibit resistance to these bacteriocins (9). Organisms producing bacteriocins frequently contain the bacteriocin gene and the corresponding immunity gene in the same operon. Similar to lin0359 (pedB-like), immunity genes have been acquired by bacteria that are the targets of the bacteriocin. The protein encoded by lin0359 exhibits 36% sequence identity to Lactobacillus gasseri PedB and 33% sequence identity to PedB from P. acidilactici, with the latter providing immunity against pediocin PA-1. Two obvious questions were therefore whether the listerial PedB-like protein would protect against pediocin or any other known member of the class IIa bacteriocins and why the pedB-like gene is attached to the gol operon in L. innocua and about half of the L. monocytogenes strains.

In order to get an answer to the first question, we tried to determine whether the PedB-like protein would provide immunity against some of the most common pediocin-like bacteriocins (class IIa bacteriocins), including pediocin PA-1 from P. acidilactici (22), curvacin A, leucocin A, sakacin P, and enterocin A. Their effects on the growth of wild-type L. innocua and of mutants carrying either a golR disruption (expresses pedB constitutively) or a pedB deletion were tested. The concentration of each bacteriocin leading to 50% growth inhibition (MIC) was determined. We observed that only the pediocin PA-1 K20R variant had a pedB-related effect. Compared to wild-type L. innocua and the ΔpedB-like mutant, the strong constitutive expression of the pedB-like gene in the golR mutant rendered it reproducibly about 2 times less sensitive to the presence of pediocin PA-1 (Fig. 5). For all other bacteriocins tested, the observed MIC was the same for the wild type and the two mutant strains (data not shown). The weak protective effect against pediocin PA-1 suggests that pediocin PA-1 is not the physiologically relevant bacteriocin against which the listerial PedB-like protein provides immunity. It probably protects against a bacteriocin more or less closely related to pediocin PA-1. There are several bacteriocins that exhibit only 1 or 2 amino acid exchanges compared to the 44 residues of pediocin PA-1, such as pediocin Ach from P. acidilactici, the pediocin produced by Pediococcus sp. strain CRA51, or the coagulin CoaA from Bacillus coagulans (29). The corresponding PedB proteins protecting against these pediocins are nearly identical to PedB from P. acidilactici, whereas the PedB-like protein encoded by the listerial gol operon exhibits only 48% sequence similarity. It is therefore likely that the PedB-like protein protects against a bacteriocin more distantly related to pediocin PA-1. Among these bacteriocins, we have tested curvacin A, leucocin A, sakacin P, and enterocin A (16). However, there exist many more which were not at our disposal, such as mundticin from Enterococcus mundtii, avicin from Enterococcus avium, or piscicolin from Carnobacterium maltaromaticum. Again, none of the corresponding immunity proteins produced by the above-mentioned organisms exhibits more than 40% sequence identity to the listerial PedB-like protein. We therefore favor the idea that the bacteriocin to which the listerial PedB-like protein provides immunity has not yet been discovered. In order to protect itself, the organism synthesizing this bacteriocin is likely to produce an immunity protein nearly identical to the listerial PedB-like protein encoded by the gol operon.

Fig 5.

Pediocin immunity assays. The immunity of the L. innocua wild-type strain (wt), the golR mutant, and the pedB-like mutant against the pediocin PA-1 K20R variant from Pediococcus acidilactici (22) was determined. For that purpose, the growth of the three strains in microtiter plates containing BHI broth and the indicated pediocin PA-1 concentrations was monitored over a time period of 12 h by measuring the A₆₀₀ with a microtiter plate reader. The mean values from three independent experiments and the calculated standard deviations are presented. The growth of the pedB-like mutant was strongly reduced at 0.78 nM pediocin PA-1. At 1.54 nM pediocin PA-1, the growth of the wild-type strain was completely inhibited, whereas at this concentration, the golR mutant was still able to reach about half the maximal A₆₀₀.

Regarding the second question, we hypothesize that some listerial species probably fused the pedB-like gene to the gol operon because it is induced by glycerol during growth in soil (37). The organism producing the corresponding bacteriocin might therefore also be a soil bacterium. Interestingly, all serotype 4b L. monocytogenes strains, which are the most virulent strains, lack the PedB-like protein, thus reflecting that serotype 4b strains are probably more adapted to proliferation in host cells than to saprophytic life. Although glycerol is one of the preferred carbon sources during the growth of L. monocytogenes within host cells (15), the gol operon is not induced during intracellular growth (27). In contrast, the expressions of the glpFK genes lmo1538 and lmo1539 are upregulated during intracellular growth, and their inactivation leads to a significant attenuation of listerial virulence. Glycerol metabolism during the saprophytic life of L. monocytogenes seems to be mediated via the enzymes encoded by both the gol and the glpFK operons, whereas during intracellular growth, it is probably achieved only by the glpFK gene products. How L. monocytogenes distinguishes between glycerol present in the soil and glycerol encountered in host cells is not understood. Glycerol phosphorylated by GlpK is probably metabolized mainly via the Embden-Meyerhof-Parnas pathway, whereas glycerol first oxidized to Dha by GolD is probably metabolized mainly via the pentose phosphate pathway. The two modes of glycerol metabolism therefore respond to different metabolic requirements.