Diversity of Oligopeptide Transport Specificity in Lactococcus lactis Species A TOOL TO UNRAVEL THE ROLE OF OppA IN UPTAKE SPECIFICITY*

The specific oligopeptide transport system Opp is essential for growth of Lactococcus lactis in milk. We ex-amined the biodiversity of oligopeptide transport specificity in the L. lactis species. Six strains were tested for (i) consumption of peptides during growth in a chemically defined medium and (ii) their ability to transport these peptides. Each strain demonstrated some specific preferences for peptide utilization, which matched the specificity of peptide transport. Sequencing of the binding protein OppA in some strains revealed minor differences at the amino acid level. The differences in specificity were used as a tool to unravel the role of the binding protein in transport specificity. The genes encoding OppA in four strains were cloned and expressed in L. lactis MG1363 deleted for its oppA gene. The substrate specificity of these engineered strains was found to be similar to that of the L. lactis MG1363 parental strain, whichever oppA gene was expressed. In situ binding experiments demonstrated the ability of OppA to interact with non-transported peptides. Taken together, these results provide evidence for

The oligopeptide transport system Opp has been described in many bacteria. This transport system may be involved in (i) nutrient acquisition in Lactococcus lactis (1,2) or Streptococcus thermophilus (3); (ii) recycling of cell wall peptides for peptidoglycan synthesis in Salmonella typhimurium and Escherichia coli (4); (iii) sensing of extracellular signaling molecules (pheromones) required for the initiation of competence and sporulation in Bacillus subtilis (5)(6)(7)(8), for the induction of conjugation in Enterococcus faecalis (9,10), and for the induction of virulence in several pathogenic bacteria (11)(12)(13)(14); and (iv) growth at low temperatures and intracellular survival in macrophages of Listeria monocytogenes (15).
Opp is a member of a superfamily of highly conserved ATPbinding cassette transporters. In Gram-negative bacteria, the transporter comprises a periplasmic solute-binding protein (OppA) and a translocon consisting of two integral membrane proteins (OppB and OppC) and two membrane-bound cytoplasmic ATP-binding proteins (OppD and OppF). In Gram-positive bacteria, OppA proteins are lipoproteins anchored to the cell membrane by their N-terminal lipid moiety. Although several copies of the gene encoding the binding protein might be present in Gram-positive bacteria (3,16), only one seems to be functional in L. lactis (17,18). OppA serves as an initial receptor. It binds the substrate and delivers it to the transmembrane complex. It is generally considered as the specificity determinant of the system, whereas the rate of peptide transport is imposed by the rate of peptide donation from OppA to the OppBC complex (19). The ATP-binding proteins couple ATP hydrolysis to the transport process.
The substrate specificity of Opp from S. typhimurium has been well established. S. typhimurium Opp transports peptides from two to five amino acids with a broad range of sequences (20). S. typhimurium OppA has a higher affinity for tripeptides than for dipeptides (21). The Gram-positive bacterium L. lactis shows significant differences in peptide uptake and affinity compared with S. typhimurium. First of all, L. lactis MG1363 is able to transport oligopeptides containing up to 18 amino acids (22). Its binding protein preferentially interacts with nonameric peptides, but is able to bind peptides containing up to 35 amino acid residues (23). Nevertheless, the strain preferentially uses hydrophobic basic peptides with molecular masses ranging between 600 and 1100 Da (24), whereas di-and tripeptides are not transported by the L. lactis Opp system (16,25).
In this work, we compared the ability of different strains of L. lactis to transport oligopeptides. In the L. lactis species, we demonstrate the existence of variability in both the specificity of peptide transport and the amino acid sequence of L. lactis OppA (OppA Ll ). 1 In an attempt to correlate these diversities, we expressed the different OppA Ll proteins in an oppA mutant with its native oppDFBC operon still functional. The ability of these engineered strains to transport peptides was compared with that of the corresponding wild-type strains. We reveal that, although OppA Ll was essential for peptide transport function, there was no correlation between the sequence of the binding protein and the specificity of peptide transport. These results suggest a role for the OppBCDF component in imposing the specificity of peptide transport in L. lactis.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Growth Conditions-The bacterial strains used in this study are listed in Table I. L. lactis strains were * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY189900, AY189901, and AY189902.
E. coli strains were grown with aeration at 37°C in LB broth supplemented with the appropriate antibiotics when necessary. L. lactis wild-type strains were grown at 30°C in M17 broth or in chemically defined medium (CDM) (33). Engineered strains (SL5145, SL5146, SL5147, SL5152, SL5174, and SL5175) were grown in similar media containing both 0.5% (w/v) glucose and 0.25% (w/v) lactose and supplemented with erythromycin (2.5 mg/liter) and chloramphenicol (2.5 mg/ liter). In some growth experiments, one of the essential amino acids (methionine, valine, histidine, glutamine, leucine, or isoleucine) was omitted from CDM and replaced with a peptide containing the omitted amino acid (final concentrations of 0.08, 0.28, 0.03, 0.27, 0.36, and 0.16 mmol/liter, respectively).
Growth experiments were performed with an ultramicroplate reader (Bio-Tek Instruments, Inc., Winooski, VT) using 96-well sterile microplates. Each well contained 200 l of culture medium and was inoculated with ϳ10 7 colony-forming units/ml of an overnight culture. Prior to inoculation, the strain was washed twice with 50 mmol/liter KH 2 PO 4 / K 2 HPO 4 (pH 6.9). The A 600 was measured over 10 h every 30 min after gentle shaking. To prevent evaporation of the culture medium, each well was overlaid with sterile paraffin oil. The apparent growth rate is defined as the maximal slope of the semilogarithmic plot against time of A 600 measurements.
DNA Manipulations and Sequencing-Total DNA of L. lactis strains was isolated from a 2-ml culture grown overnight in M17 broth. Cells were harvested by centrifugation at 8000 ϫ g for 10 min and resuspended for 2 h at 37°C in 0.1 mol/liter Tris, 0.1 mol/liter EDTA, 25% (v/v) glucose, and 0.1 g/liter mutanolysin (pH 7.0). Cells were lysed by incubation for 30 min at 37°C in 0.1 mol/liter Tris, 0.01 mol/liter EDTA, 0.5% (v/v) sarcosyl, 1 g/liter proteinase K, and 1.25 g/liter RNase. Proteins were removed by two successive 25:24:1 (v/v/v) phenol/chloroform/isoamyl alcohol extractions and one 24:1 (v/v) chloroform/isoamyl alcohol extraction. The last supernatant volume was adjusted to 400 l with Tris/EDTA/sarcosyl solution, and 50 l of 3 mol/liter potassium acetate (pH 4.8) was added. DNA was precipitated with 1.2 ml of ice-cold ethanol and finally resuspended in 200 l of 10 mmol/liter Tris buffer.
Expression of the OppA-His 6 protein was carried out in 500 ml of culture essentially as described by QIAGEN S. A. After removal of the culture medium, cells were resuspended in 10 mmol/liter Tris, 100 mmol/liter NaH 2 PO 4 , 8 mol/liter urea, 0.1% (v/v) Triton X-100, 20 mmol/liter ␤-mercaptoethanol, and 1 mg/liter lysozyme (pH 8.0). The cell suspension was incubated for 1 h at 25°C with gentle agitation (200 rpm), and a clear cell-free extract was obtained by centrifugation at 12,000 ϫ g for 15 min. Purification of the OppA-His 6 protein was carried out under denaturing conditions by applying the cell-free extract to nickel-nitrilotriacetic acid resin (QIAGEN S. A.) and using an elution system of 10 mmol/liter Tris, 100 mmol/liter NaH 2 PO 4 , and 8 mol/liter urea according to the manufacturer's instructions. The OppA-His 6containing fractions, as determined by SDS-PAGE analysis, were dialyzed against sterile water and lyophilized.
Anti-OppA antibodies were purified from the serum of an immunized rabbit (Valbex-Université Claude Bernard Lyon I, Villeurbanne, France) as described above, except that the resin was additionally washed with 40 ml of 50 mmol/liter Tris and 150 mmol/liter NaCl (pH 7.4) and then with 40 ml of 50 mmol/liter Tris and 1 mol/liter NaCl (pH 7.4). Antibodies were eluted during a 30-min incubation with 4 mol/liter MgCl 2 . The elution fractions were collected, dialyzed against sterile water, and lyophilized.
Milk Peptide Purification-Milk proteins were precipitated with 1% (v/v) trifluoroacetic acid. After removal of the proteins by centrifugation at 10,000 ϫ g for 10 min at 4°C, the supernatant was ultrafiltered through a 3000-Da cutoff membrane (YM3, Amicon, Inc., Beverly, MA). Peptides were isolated by solid-phase extraction using reverse-phase cartridges (Sep-Pak C 18 , Waters Associates, Milford, MA). The peptides were separated at 40°C by HPLC on a reverse-phase C 18 column (Nucleosil (250 ϫ 4.6 mm), Colochrom, Gagny, France) at a flow rate of 1 ml/min. Solvents A and B were 0.115% (v/v) trifluoroacetic acid and 0.1% (v/v) trifluoroacetic acid and 60% (v/v) acetonitrile in MilliQ water, respectively. A 5-min isocratic phase in solvent A was followed by a linear gradient of solvent B (0 -60% within 40 min). The collected fractions were submitted to a second separation using 5 mmol/liter KH 2 PO 4 /K 2 HPO 4 (pH 6.9) and 60% (v/v) acetonitrile in 5 mmol/liter KH 2 PO 4 /K 2 HPO 4 (pH 6.9) as the solvents. The eluted peptides were collected, dried in a SpeedVac concentrator (Savant Instruments, Inc., Farmingdale, NY), resuspended in MilliQ water, and desalted using the first HPLC separation system (trifluoroacetic acid/acetonitrile). Purified peptides were identified by mass spectrometric analysis and Nterminal microsequencing.
Peptide Transport-The transport assays were adapted from previously described procedures (2,37). Cells were grown to A 650 ϳ 0.8 in CDM containing free amino acids as the nitrogen source. Prior to transport assays, cells were washed twice with 50 mmol/liter KH 2 PO 4 / K 2 HPO 4 (pH 6.9) and then de-energized for 30 min at 30°C with 10 mmol/liter 2-deoxy-D-glucose (38). For each transport assay, cells (A 650 ϭ 1; corresponding to 0.2 g of cell protein/liter) were incubated for 5 min at 22°C in the presence of 25 mmol/liter glucose and 2 mmol/liter MgSO 4 . When required, the serine proteinase inhibitor 4-(2-aminoethyl)benzenesulfonyl fluoride (1 mmol/liter; Interchim, Montluçon, France) was added to the incubation mixture. We first verified that this inhibitor had no effect on transport process. Uptake was initiated by adding the peptide at a final concentration of 50 mol/liter, unless otherwise stated. One-ml samples were taken, and cells were separated from the incubation medium by filtration using cellulose acetate filters (0.45-m pore size; Schleicher & Schü ll). Cells were subsequently washed twice with 2 ml of ice-cold KH 2 PO 4 /K 2 HPO 4 (50 mmol/liter) at pH 6.9. Peptide uptake was monitored by determining the intracellular concentration of free amino acids constituting the peptide under study, as previously described (2). It is worth mentioning that intact peptide could not be detected inside the cells due to the high rate of peptide hydrolysis by internal peptidases (37). The amino acids were first derivatized with o-phthalaldehyde and then separated at 37°C on a reverse-phase HPLC C 18 column (UptiSelect (250 ϫ 4.6 mm), Interchim) at a flow rate of 1 ml/min. Solvent A was 50 mmol/liter sodium acetate (pH 5.7) and 3% (v/v) tetrahydrofuran, and solvent B was 95% (v/v) methanol and 5% (v/v) tetrahydrofuran. A 5-min isocratic phase in 18% (v/v) solvent B was followed by a linear gradient of solvent B (18 -100% within 35 min). For detection of fluorescence, the excitation and emission wavelengths were 340 and 455 nm, respectively.
Binding Assays-The ability of lactococcal strains to bind peptide VGDE was estimated at 30°C as follows. Concentrated de-energized cells (A 650 ϳ 15) were incubated for 2 min (23) in 50 mmol/liter KH 2 PO 4 / K 2 HPO 4 (pH 6.5) containing 500 mol/liter VGDE. Cells were collected on a 0.22-m pore size filter (Schleicher & Schü ll) and washed three times with ice-cold potassium phosphate buffer. The filter was then coated for 2 min with a solution of peptide YGGFL (500 mol/liter) in potassium phosphate buffer. After removing cells by filtration, the peptides contained in the buffer were concentrated by solid-phase extraction using an anion cartridge exchanger (Accell Plus QMA, Waters Associates) and analyzed by HPLC as described above.

RESULTS
Peptide Utilization by Different Strains of L. lactis-Six strains of L. lactis were grown in CDM lacking an essential amino acid in the free form and supplied by a pure peptide. None of the strains was able to grow when the omitted amino acid was not replaced by a peptide. Twenty-five different peptides were selected on the basis of their various biochemical characteristics and their origin (Table II). Six of them were purified from milk. They were initially chosen because of their disappearance from milk after growth of some of the strains (data not shown).
Most peptides (18 of 25) were able to sustain growth of the six strains at a maximal rate in CDM deprived of one essential amino acid in the free form and provided in peptide form (Fig.  1). Growth systematically corresponded to consumption of the peptide, as revealed by HPLC analysis of the culture medium (data not shown). Nevertheless, except for CNRZ261, none of the strains was able to use all the tested peptides as a source of amino acids. Five peptides (VGDE, DRVYIHPFHL, RPK-PQQFFGLM, ISQRYQK, and LPQY) were differently consumed by the six L. lactis strains, suggesting that the strains under study do not share the same preferences for peptide utilization. Moreover, L. lactis Wg2 grew very poorly in the presence of the basic heptapeptide ISQRYQK as the source of Gln or Ile, whereas this strain grew at a maximal rate in the presence of the basic heptapeptide YPFPGPI (source of Ile) or TVYQHQK (source of Gln). This indicates that previous observations made with L. lactis MG1363, which indicated a preference for peptide utilization related to both the mass and the charge of the peptide (24), cannot be extended to the L. lactis species.
Variability in Peptide Utilization Corresponds to Variability in Peptide Transport-The peptides used for growth experiments were incubated with cell-free extracts. They were all cleaved at the amino acid level. The lack of growth of some strains in CDM was therefore not due to an inability of the cells to cleave the peptide intracellularly. The peptides were also incubated either in the presence of the P I -type proteinase PrtP released from L. lactis E8, CNRZ261, and Wg2 by incubation in a Ca 2ϩ -free buffer (39) or in the presence of P III -type PrtP anchored to resting L. lactis CNRZ437 cells (note that autoproteolysis of P III -type PrtP affects its specificity) (40). Only one of them (RPKPQQFFGLM) was cleaved by PrtP. Other peptides were not hydrolyzed. This suggests that most (if not all) of the differences observed during growth experiments were not due to a difference in extracellular cleavage of the peptide by PrtP. Consequently, the most convenient hypothesis to explain the differences in growth is that the strains under study do not have identical oligopeptide transport capabilities.
To ascertain this hypothesis, we tested the ability of four of   Table II. Peptide concentration was adjusted to the corresponding free amino acid concentration in CDM (33) as described under "Experimental Procedures." the strains to transport three peptides that revealed a difference in preferences for peptide utilization between strains (VGDE, DRVYIHPFHL, and RPKPQQFFGLM). To prevent ex-tracellular cleavage of RPKPQQFFGLM by PrtP, uptake was performed in the presence of 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride. As expected, the ability of the strains to grow correlated with their ability to transport the peptides (Table  III). By analyzing the presence of free Val or free Met in the external medium, we have experimentally excluded the possibility that differences in peptide uptake rates could be due to differences in amino acid efflux rates between strains. Despite the fact that VGDE ensured a maximal growth rate of L. lactis Wg2 or CNRZ437, a large difference in the initial rate of uptake was observed between the two strains. On the other hand, IL1403 did not grow at a maximal rate in the presence of VGDE, although its initial rate of VGDE uptake was higher than that of Wg2. This indicates that the rate of peptide transport did not determine the growth rate, at least under our experimental conditions.
The Binding Protein OppA Is Not Responsible for Variability in Transport Specificity-The binding protein is generally considered to be responsible for the specificity of peptide transport in ATP-binding cassette transporters (19). The oppA genes from L. lactis strains MG1363, Wg2, and CNRZ437 were therefore cloned and sequenced. Comparison of the deduced amino acid sequences, including those available in the GenBank TM / EBI Data Bank (SSL135, IL1403, and SK11), revealed some differences at the amino acid level (Fig. 2). The OppA sequence of MG1363 was identical to that of SSL135, but differed from those of Wg2, CNRZ437, IL1403, and SK11 by 9, 10, 72, and 2 residues, respectively.
In an attempt to assign the differences in peptide transport pattern to the variations in the OppA amino acid sequence, the oppA genes from four different strains were introduced into the OppA-defective strain AMP15 (28). The oppA genes were cloned under the control of the T7 promoter into plasmid pLET5 (29). To ensure the expression of oppA, the two plasmids pILpOL and pMG820 were also introduced in the recipient strain, yielding L. lactis strains SL5147, SL5152, SL5174, and SL5175, carrying the oppA genes from MG1363, Wg2, IL1403, and CNRZ437, respectively (Table I). As controls, plasmid pLET5 free of cloned oppA plus pILpOL and pMG820 were introduced in MG1363 and AMP15, yielding MG3 ϩ and SL5146, respectively. The growth rates and maximal bacterial populations in milk of strains MG1363, MG3 ϩ , and SL5147 were similar (1.6 Ϯ 0.1 h Ϫ1 and (1.0 Ϯ 0.1) ϫ 10 8 colonyforming units/ml, respectively). The amounts of OppA expressed by SL5147, SL5152, SL5174, and SL5175 grown in the appropriate medium (CDM containing 5 g/liter glucose as the carbon source, 2.5 g/liter lactose as the inducer (29), and 2.5 mg/liter each erythromycin and chloramphenicol) were in the same range and slightly lower in each case than that produced by L. lactis MG1363 as revealed by Western blot analyses (Fig.  3). As a last control, no significant differences in the uptake of the control peptide YGGFL by MG1363 and MG3 ϩ or SL5147 could be detected, whereas SL5146 was unable to transport peptides.
As expected, the properties of peptide transport by SL5147 exactly matched those of MG1363. Surprisingly, L. lactis SL5152, SL5174, and SL5175, carrying the oppA genes from Wg2, IL1403, and CNRZ437, respectively, had the same preferences for peptide utilization as MG1363 (Fig. 4). For example, strains SL5152, SL5174, and SL5175 were unable to grow in CDM containing VGDE as the source of Val, whereas the corresponding L. lactis wild-type strains Wg2, IL1403, and CNRZ437, respectively, grew. Moreover, the growth rates of SL5152, SL5174, SL5175, and MG1363 in CDM containing RPKPQQFFGLM as the source of Met were comparable, although Wg2, IL1403, and CNRZ437 were unable to use this peptide as the source of Met (Fig. 1). It is worth mentioning that L. lactis strains SL5152, SL5147, SL5174, and SL5175 are Prt Ϫ strains and therefore unable to cleave peptides in the external medium prior to transport.
Uptake experiments confirmed that L. lactis strains SL5152, SL5174, and SL5175 transported the same peptides despite their expression of different OppA proteins (Table IV). The ability of the engineered strains to transport specific peptides was identical to that of MG1363. Initial rates of uptake were slightly lower than those obtained with the wild-type strain. This was presumably due to a lower expression level of the binding protein in the engineered strains compared with MG1363 (Fig. 3). The complementation of MG163⌬oppA with the binding protein isolated from L. lactis strain Wg2, IL1403, or CNRZ437 restored the substrate specificity of L. lactis MG1363 rather than that of the OppA donor strain.
OppA Ll Is Able to Bind a Non-transported Peptide-The ability of the binding protein OppA from L. lactis strain MG1363 to bind in situ non-transported peptides was estimated. De-energized cells were first incubated in the presence of the nontransported peptide VGDE (loading step). After extensive washing, cells were then incubated in the presence of YGGFL (chase step). The removal of VGDE from OppA by YGGFL was estimated by submitting the chase buffer to HPLC analysis after concentrating its peptide content by solid-phase extraction. No VGDE could be detected when YGGFL was omitted from the chase buffer. Similarly, only traces of VGDE were detected when using L. lactis AMP15, indicating that this strain did not bind a significant amount of VGDE during the loading step. In contrast, L. lactis MG1363 released ϳ0.35 nmol of VGDE/mg of protein in the presence of YGGFL (mean of three determinations, S.D. ϭ 0.02) (Fig. 5). These results demonstrate the ability of L. lactis MG1363 OppA to interact in situ with a non-transported peptide.

DISCUSSION
Previous studies of substrate specificity in peptide transport and/or utilization by L. lactis were performed using only one strain, MG1363 (22,24). The results demonstrated that the substrate specificity of L. lactis strain MG1363 is rather atypical compared with other model Gram-negative bacteria such as E. coli and S. typhimurium (20,21). The present study provides evidence for variability in the ability of L. lactis strains to consume peptides as a source of amino acids during growth. For instance, MG1363 was unable to grow using the tetrapeptide VGDE as the source of Val. This result is in agreement with the established preferences for peptide utilization by this strain, i.e. rejection of acidic peptides with low molecular mass (24). However, this peptide was utilized by all the other strains under study. Consequently, the MG1363 preferences are not representative of the genus Lactococcus. We were able to distinguish four strains on the basis of their capabilities to use several peptides. This variability in peptide consumption resulted from a difference in the ability of the strains to transport peptides. The specificity of peptide transport by L. lactis strain MG1363 did not mirror that of all the lactococcal strains. Our first original conclusion is that the L. lactis genus shows a biodiversity in oligopeptide transport specificity.
Oligopeptide utilization by L. lactis requires the presence of the functional oligopeptide transport system Opp. Previous  Table II. Peptide concentration was adjusted to the corresponding free amino acid concentration in CDM (33) as described under "Experimental Procedures." studies indicated that the binding protein OppA dictates the specificity of the Opp system (28,41). As proof, a ⌬oppA mutant was shown to be unable to transport peptides, and mutagenesis of OppA affected the overall specificity of the protein for peptides (28).
The biodiversity in peptide transport specificity in lactococci was used as a tool to unravel the role of the binding protein OppA Ll in determining the substrate specificity of L. lactis. Comparison of the OppA Ll sequences, including those of the four strains studied, showed some amino acid substitutions. To determine to what extent these substitutions were responsible for the specificity, we cloned and expressed the oppA genes of the four strains in the recipient strain L. lactis MG1363⌬oppA (strain AMP15) (28). Complementation by different oppA genes restored all four functional Opp hybrid transporters. This suggests that none of the substitutions in the different OppA Ll proteins impaired donation of the substrate from the binding protein to the transmembrane channel of MG1363. The substrate specificity of the four engineered strains was found to be identical to that of L. lactis strain MG1363. The substrate specificity of the complemented L. lactis strains SL5152, SL5174, and SL5175 did not match that of the corresponding wild-type strains (Wg2, IL1403, and CNRZ437, respectively).
These intriguing results question the role of OppA Ll in the specificity of peptide transport. If a peptide is not bound by OppA Ll , it will not be transported by the Opp system. In this respect, OppA Ll might be considered as one determinant of peptide transport specificity. Nevertheless, the specificity of peptide binding by OppA Ll is very broad (23). The use of recombinant strains allowed us to establish that the inability of lactococci to transport specific peptides was not due to peptide exclusion by OppA Ll . In other words, the specificity of peptide transport (i.e. the ability to ultimately transport specific peptides or not) is not solely dependent on OppA Ll . The broad binding specificity of OppA Ll excludes it from being the major specificity determinant of oligopeptide transport. Our second original conclusion is that, although OppA Ll is absolutely necessary for peptide transport, it does not (exclusively) determine its specificity.
Complementary results corroborate this interesting observation. Binding experiments indicated that the tetrapeptide VGDE was able to interact in situ with the oligopeptide-binding protein of L. lactis MG1363. This observation was in agreement with the reported competitive inhibition exerted by VGDE on the transport of a reporter peptide by L. lactis MG1363. 2 This indicates that a peptide could bind to OppA Ll even if it is not transported by the Opp system. These results further support our conclusion, i.e. the specificity of peptide uptake by L. lactis is not exclusively dictated by OppA Ll .
The proposed model for oligopeptide transport by L. lactis is a four-step process: (i) reversible binding of the substrate to the open form of the binding protein; (ii) conformational change of the binding protein, resulting in the partial entrapment of the substrate; (iii) transfer of the partially entrapped substrate from the binding protein to the transmembrane complex; and (iv) translocation of the substrate across the membrane (23,43). The peptide VGDE was translocated by the three L. lactis strains Wg2, IL1403, and CNRZ437 at different rates (Table  III), although its K m value was in the same range for the three strains (250 mol/liter). This result indicates that the first step of transport, peptide binding by OppA, is not a limiting step for the transport process. It is in agreement with previous kinetic analyses that demonstrated that the rate of transport is deter-mined by the kinetics of peptide donation from the binding protein to the translocator complex (19). In this kinetic model, the binding protein would act as a plug that blocks the ligand from returning to the external medium (44). A consequence of the peptide binding would be to transmit a signal via the transmembrane complex to the ATP subunits that results in an increase in the transporter affinity for ATP and subsequently leads to the opening of the translocation pore and the concomitant release of the substrate from the binding protein.
Transmembrane proteins OppB and OppC have been described as important actors in several bacterial phenomena. For example, bacterial adherence can be affected by mutations in the binding protein and other domains of the permease complex (42,45). Indeed, the adherence of Streptococcus gordonii is affected by mutating either the binding protein (SarA) or OppC (42). One possibility is that the transmembrane complex OppBC acts as a filter. It is worth mentioning that the three available lactococcal opp sequences display variability in the OppB and OppC sequences (16 -18). This sequence variability might explain the transport variability among L. lactis strains. If this hypothesis is correct, it remains to be shown how the transmembrane complex participates in the specificity of the oligopeptide transport process (e.g. exclusion of specific peptides from the channel or impairment of the interaction of the transmembrane complex with the binding protein liganded with specific peptides).
Our results show that the specificity of oligopeptide transport in L. lactis is determined by at least two successive filters. The first filter, OppA, captures oligopeptides and initiates the transport process, but is rather aspecific. Our further work will focus on the second filter, presumably the transmembrane channel, and will aim at identifying the ensemble of determinants involved in oligopeptide transport specificity in L. lactis.