MppA, a Periplasmic Binding Protein Essential for Import of the Bacterial Cell Wall Peptide l-Alanyl-γ-d-Glutamyl-meso-Diaminopimelate

Abstract

Mutants of a diaminopimelic acid (Dap)-requiring strain of Escherichia coli were isolated which failed to grow on media in which Dap was replaced by the cell wall murein tripeptide, l-alanyl-γ-d-glutamyl-meso-diaminopimelate. In one such mutant, which is oligopeptide permease (Opp) positive, we have identified a new gene product, designated MppA (murein peptide permease A), that is about 46% identical to OppA, the periplasmic binding protein for Opp. A plasmid carrying the wild-type mppA gene allows the mutant to grow on tripeptide. Two other mutants that failed to grow on tripeptide were resistant to triornithine toxicity, indicating a defect in the opp operon. An E. coli strain whose entire opp operon was deleted but which carried the mppA locus was unable to grow on murein tripeptide unless it was provided with oppBCDF genes in trans. Our data suggest a model whereby the periplasmic MppA binds the murein tripeptide, which is then transported into the cytoplasm via membrane-bound and cytoplasmic OppBCDF. In assessing the affinity of MppA for non-cell wall peptides, we have found that proline auxotrophy can be satisfied with the peptide Pro-Phe-Lys, which utilizes either MppA or OppA in conjunction with OppBCDF for its uptake. Thus, MppA, OppA, and perhaps the third OppA paralog revealed by the E. coli genome sequence may each bind a particular family of peptides but interact with common membrane-associated components for transport of their bound ligands into the cell. As to the physiological function of MppA, the possibility that it may be involved in signal transduction pathway(s) is discussed.

During growth, Escherichia coli breaks down over one-third of its cell wall each generation and efficiently reutilizes the tripeptide therefrom for synthesis of new murein in a sequence of events termed the recycling pathway (9, 11, 32; see reference 33 for a review). In this pathway, murein is degraded to N-acetylglucosaminyl-1,6-anhydro-N-acetylmuramyl-l-alanyl- γ-d-glutamyl-meso-diaminopimelate (GlcNAc-anhMurNAc-tripeptide) by the combined action of lytic transglycosylases, endopeptidases, and d,d- and l,d-carboxypeptidases which are present in the periplasm (39). The muropeptide, GlcNAc- anhMurNAc-tripeptide, presumably is transported into the cytoplasm via the membrane-bound AmpG permease (20, 24). The tripeptide is then released from the muropeptide by AmpD anhydro-N-acetylmuramyl-l-alanine amidase (19, 21). Surprisingly, almost all murein tripeptide for recycling is transported into the cell as GlcNAc-anhMurNAc-tripeptide via the AmpG permease and is then released by the cytoplasmic AmpD amidase (20, 32), rather than being transported as the free tripeptide via the oligopeptide permease (Opp) as was originally proposed (10). Direct utilization of the tripeptide for cell wall synthesis was assumed to depend on a hypothetical ligase which would attach tripeptide to UDP-MurNAc, thereby reintroducing it into the biosynthetic pathway for wall synthesis (9, 20, 33). In fact, the enzyme responsible for this activity has recently been identified, and the gene, mpl, was shown to be the open reading frame (ORF) yifG at 96 min on the E. coli map (29). An mpl null mutant was completely devoid of ligase activity, and cells of this mutant were viable and accumulated tripeptide in their cytoplasm (29).

During a search for mutants lacking this murein peptide ligase activity, four mutants were isolated from a pool of mutagenized diaminopimelic acid (Dap)-negative (dap) parental cells in a screen that assayed the growth of cells on free tripeptide as a source of Dap. In this report, we describe the isolation and initial characterization of one such mutant. A new genetic locus, mppA, has been identified which codes for a periplasmic binding protein required for uptake of murein peptides. Two other mutants, one with a mutation in oppB and the other with a mutation in groESL (unpublished), were found to be defective in Opp function because of their resistance to triornithine toxicity. The oppB mutation indicates that murein tripeptide is transported from MppA into the cytoplasm via membrane components of Opp, and the groE mutation suggests that the chaperonin is involved in the proper folding and assembly of the components of the peptide transport system.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The E. coli strains used in this study are listed in Table 1. Cells were grown at 37°C in L broth (38) or 2× YT (38) supplemented with 50 μg of Dap per ml where required. Antibiotic-resistant strains were selected in the presence of 15 μg of chloramphenicol (Cm) per ml, 30 μg of kanamycin (Kan) per ml, or 100 μg of ampicillin (Amp) per ml as needed. Tests for triornithine resistance were done on M9-glucose agar supplemented with 0.1% Casamino Acids, 1 μg of thiamine per ml, and 500 μM triornithine (Bachem). Tests for growth of proline auxotrophs were done as described in the footnote to Table 2.

TABLE 1.

Bacterial strains

Strain	Genotype	Source or reference
AT980	dapD2 relA1 spoT1 thi-1 λ− Hfr (defective)	E. coli Genetic Stock Center (Yale University, New Haven, Conn.), no. 4545
TP72	F−ampG::kan lysA opp araD139 rpsL150 relA1 deoC1ptsF25 flbB5301 rbsR Δ(argF-lac)	20
TP981	AT980 ampG::Kan	P1(TP72) × AT980
TP984	TP981 mppA::miniTn10Cm	This work
TP985	AT980 mppA::miniTn10Cm	P1(TP984) × AT980
AW1043	MC1000 Δ phoA (E-15) proC::Tn5	A. Wright
TP987	AT980 proC::Tn5	P1(AW1043) × AT980
TP988	TP985 proC::Tn5	P1(AW1043) × TP985
CH483	Δ(trp-tonB-oppABCDF)467 thi lac pro galE	18
VC6121	thi-1 relA1 spoT1 dapB17::Mu zaa1::Tn5 Hfr	E. Ishiguro
TP986	CH483 dapB17::Mu zaa1::Tn5	P1(VC6121) × CH483
SS320	F−lacI22 lacZ pro-48 met-90 trpA trpR his-85 rpsL azi-9 gyrA λ− P1S	3
SS5013	SS320 Δ(tdk-oppABCDF-tonB-trp)	S. Short
TP989	SS5013 mppA::miniTn10Cm	P1(TP985) × SS5013
TP73	F− ΔampDE lysA opp araD139 rpsL150 relA1 deoC1 ptsF25 flbB5301 rbsR Δ(argF-lac)	20
CH212	hsdS20 (rB− mB−) recA13 ara-14 proA2 lacY1 galK2 (SmR) xyl-5 mtl-1 supE44 oppA462	16
JM83	ara Δ(lac-proAB) rpsL thi [φ80 dlacΔ(lacZ)M15]	44
TOP10F′	F′ [lacIq Tn10(Tetr)] mcrA Δ(mrr-hsdRMS-mcrBC) φ80lacZΔM15 ΔlacX74 deoR recA1 araD139 Δ(ara-leu)7697galU galK rpsL (StrR) endA1 nupG	Invitrogen

TABLE 2.

Growth responses of strains to murein tripeptide and Pro-Phe-Lys

Experiment no.a	Strain	Relevant genotype	Plasmid genotype	Growthb
1	TP986	ΔoppABCDF dapB17		−
	TP986/pB2	ΔoppABCDF dapB17	oppBCDF	+
2	TP985	dapD2 mppA		−
	TP985/pHSl1	dapD2 mppA	mppA	+
3	SS5013/pB2	pro-48 ΔoppA		+
	TP989/pB2	pro-48 ΔoppA mppA		−
	TP989/pB2, pHSL1	pro-48 ΔoppA mppA	mppA	+
	SS5013/pB2, pBφ30	pro-48 ΔoppA	oppA	+++
4	TP987	proC		+
	TP988	proC mppA		+

^aExperiments 1 and 2 were performed in L broth with murein tripeptide as the sole source of Dap. Experiments 3 and 4 were carried out in M9-glucose-B1 medium with required amino acids except for Pro, and Pro-Phe-Lys was used at 30 μg/ml. Pro was shown to be required for growth. Growth was observed after 1-ml standing cultures were incubated at 37°C for 18 to 20 h. 

Mutagenesis.

E. coli TP981 was mutagenized by transposon mutagenesis with λNK1324, a lambda phage which carries the transposon miniTn10Cm that contains the Cm acetyltransferase (cat) gene in place of the tetracycline resistance determinant of Tn10 (16). λNK1324, with a titer of 10¹⁰ as determined on a suppressor strain (23), produced no plaques (<10⁵) on TP981, which is suppressor negative. Eighty microliters of λNK1324 were mixed with 0.1 ml of a 10-fold concentrated overnight culture of TP981. After adsorption for 15 min at room temperature and for 15 min at 37°C, 5 ml of L broth containing 50 mM sodium citrate was added; this was followed by centrifugation. The cells were resuspended in 5 ml of L broth with citrate and shaken at 37°C for 45 min. Aliquots of this suspension were plated on L agar-Cm-Dap agar plates (see Results).

Mapping, DNA, and transformation techniques.

Mapping was done by P1 transduction (30, 38), by Southern hybridization (31), and by DNA sequencing as indicated. Small- and large-scale plasmid isolations were carried out by the alkaline lysis method (38), and plasmids were further purified with cesium chloride-ethidium bromide gradients or with a plasmid kit (Qiagen Inc., Chatsworth, Calif.). Standard procedures for restriction endonuclease digestions, ligation, and agarose gel electrophoresis were followed (38). E. coli cells were made competent and transformed with plasmid DNA either by the method of Dagert and Ehrlich (6) or by electroporation (38).

Recovery of the mutant gene on plasmids and strategy for sequencing.

To recover the transposon-disrupted gene on a plasmid, chromosomal DNA from the mutant was digested with PstI or HindIII, restriction enzymes which do not cleave the miniTn10Cm transposon, and the resultant fragments were ligated to pBluescript SK⁺ (Stratagene) or pBR322 DNA digested with the corresponding restriction enzyme and with calf intestinal alkaline phosphatase. The ligation mixtures were used to transform competent TOP10F′ cells, and the cells were plated on L agar containing Dap and Cm. From Cm^R colonies, plasmids which carried the Cm marker and the flanking chromosomal DNA containing part or all of the gene of interest were recovered. The plasmid DNA served as a template for sequencing the chromosomal DNA flanking the transposon. The primers used for this purpose were complementary to the two ends of the cat gene present in the transposon (see Results). Primer 5′-CCTCCCAGAGCCTGATAA-3′ is complementary to the 5′ end of the noncoding strand of cat and was used to determine the sequence downstream of the cat gene. The primer 5′-AAGCACCGCCGGACATC-3′, complementary to the promoter region of the coding strand of cat, was used to sequence the DNA upstream of the transposon.

Construction of plasmids.

The cloning vectors used were pTrc99A (Pharmacia), pBluescript SK⁺, pACYC177, pACYC184, and pBR322. pB2 (Amp^R) is a pBR322 derivative that harbors the promoter distal region of the opp operon beginning at the EcoRV site in oppA and extends through the end of the operon. Thus, it encodes oppB, oppC, oppD, and oppF. Transcription is driven by the Tet promoter of pBR322 (39a). pBφ30 (Cm^R) is a pACYC184 derivative that contains the opp regulatory region, the oppA gene, and part of oppB (39a). Expression plasmids for overproduction of MppA were constructed by the following procedure. PCR primers were designed to incorporate a BspLU11I site (given here in bold) that included the initiation codon (underlined) of mppA (5′-AATTTACATGTCGGTTAGAGGGAAAC-3′) on the forward primer and a unique BglII site (in bold) on the reverse primer (5′-CGCCAGATCTCATCACATCAATGCTTCAC-3′). Since another potential initiation codon was found 21 bases downstream of the first ATG in the mppA sequence, PCR amplification of this shorter version was also done, using, in this case, 5′-AAACTCATGAAGCACTCTGTTTCAG-3′ as the forward primer that introduced a BspHI site (in bold) that included the initiation codon (underlined). These primers were used to amplify the two versions of the mppA gene from the chromosome of E. coli JM83. The amplified DNA products were treated with BspLU11I or BspHI and BglII, and the resulting fragments were ligated into the compatible NcoI and BamHI sites of the expression vector pTrc99A. The ligation mixtures were used to transform strain JM83, and the transformants were selected for Amp resistance at 37°C. Plasmid DNA from about 1,000 Amp^R colonies was restricted by BamHI to linearize the cloning vector (there is no BamHI site in the desired plasmids), and the restriction mixture was used to transform JM83. Amp^R clones selected this way all contained the expected plasmids. pMLD1285 (long form) and pMLD1493 (short form) allowing expression of the mppA gene under the control of the isopropyl-β-d-1-thiogalactopyranoside (IPTG)-inducible Trc promoter were selected for further study. pHSL1 (Kan^R) is a pACYC177 derivative in which the 2.3-kb fragment from pMLD1285 cut with HpaI and HindIII and carrying mppA was blunted with the Klenow fragment of DNA polymerase I and inserted into pACYC177 which had been cut with ScaI and blunted. Following ligation, a Kan^R Amp^S transformant was shown to express mppA (Table 2, experiment 2).

Preparation of crude protein extracts.

Cells (0.5-liter cultures) were grown from a 1% inoculum at 37°C in 2× YT medium with vigorous aeration. The inoculum was grown overnight from a single colony. All cultures contained 100 μg of Amp per ml. When required, 1 or 5 mM IPTG was added in early log phase and growth was continued for approximately 4 h. In all cases, cells were harvested in the cold when the optical density at 600 nm reached the range of 0.7 to 1 and were washed with 40 ml of cold 20 mM potassium phosphate (pH 7.4) containing 0.3 mM MgCl₂ and 0.1% β-mercaptoethanol. The cell pellet was suspended in 5 ml of the same buffer and disrupted by sonication (Sonicator 150; T. S. Ulltrasons, Annemasse, France) for 10 min with cooling. The resulting suspension was centrifuged at 4°C for 30 min at 200,000 × g in a Beckman TL-100 ultracentrifuge. The supernatant was dialyzed overnight at 4°C against 100 volumes of the phosphate buffer, and the resulting solution (5 ml; 10 to 12 mg of protein/ml), designated the crude extract, was stored at −20°C. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of proteins was performed as previously described with 13% polyacrylamide gels (25). Protein concentrations were determined by the method of Lowry, with bovine serum albumin as a standard (26).

Isolation of periplasmic proteins.

Cells (100-ml cultures) growing exponentially at 37°C in 2× YT medium were harvested and washed with 40 ml of 30 mM Tris-HCl (pH 8.0). After centrifugation, the cell pellet was resuspended in 8 ml of the buffer containing 30% (wt/wt) sucrose, and 20 to 30 min later EDTA was added to a final concentration of 10 mM. The suspension was kept at room temperature for a further 15 min and then centrifuged. The supernatant containing the released periplasmic proteins was dialyzed against 100 volumes of 30 mM Tris-HCl (pH 8.0) and concentrated on PM10 Amicon membranes. The cell pellet, following release of the periplasmic proteins, was resuspended in 2 ml of Tris buffer and sonicated, and the cytoplasmic proteins were recovered following high-speed centrifugation.

Murein tripeptide isolation.

E. coli TP73 (ΔampDE), which accumulates large quantities of anhMurNAc-tripeptide in its cytoplasm (20), was grown to early stationary phase in 20 liters of 2× YT broth. The cells (150 g of cell paste) were harvested, washed once in 10 mM potassium phosphate buffer (pH 7.0), and resuspended in 130 ml of cold 10% (wt/vol) trichloroacetic acid (TCA). After 10 min, the extract was recovered by centrifugation and the pellet was extracted twice with 100 ml of cold 2% TCA. The combined extracts were extracted three times with cold ethyl ether to remove the TCA. After removal of ether, the extract was neutralized with 1 ml of 10 N NaOH, lyophilized, and dissolved in 20 ml of water. After removal of high-molecular-weight materials by precipitation with 50% ethanol, anhMurNAc-tripeptide was isolated by molecular sieve chromatography on a Toyopearl HW-40S column (1.6 by 78 cm; TosoHaas, Montgomeryville, Pa.) that was equilibrated and eluted with 20 mM potassium phosphate buffer (pH 7.0). The fractions containing anhMurNAc-tripeptide were pooled, desalted on a Sephadex G-10 column, and then fractionated by high-pressure liquid chromatography on a C-18 reverse-phase column (LiChrospher RP-18; 250 by 4 mm; 3-μm particle size; E. Merck), employing a linear gradient of 0 to 20% acetonitrile in 0.05% trifluoroacetic acid (20). Because of the limited capacities of the chromatography columns, repeated runs were necessary, during each of which about 50,000 cpm of ³H-Dap-labeled TP73 extract was added to facilitate monitoring of the column fractions. All the batches of high-pressure liquid chromatography-purified anhMurNAc-tripeptide were pooled, lyophilized, dissolved in 1 ml of 5 mM Tris-HCl (pH 7.5), and digested with 2 μg of purified AmpD amidase (21) for 16 h at 37°C. Tripeptide was isolated from this digest by chromatography on a QMA MemSep 1010HP anion exchange membrane (Millipore Corp., Bedford, Mass.), employing a gradient of 0 to 500 mM NaCl in 20 mM Tris-HCl (pH 8.0) over 20 min. Tripeptide eluted after about 11 min.

Complementation of the mutant requiring murein tripeptide for growth.

Growth of the mutant, containing various plasmids, on L agar-Dap-Amp plates was compared with growth on L agar-tripeptide-Amp plates in which the concentration of tripeptide was twice the amount required for growth (about 10 μg/ml). IPTG (0.1 mM) was added when induction of the Trc promoter was required.

Nucleotide sequence accession number.

Our nucleotide sequence of the mppA gene has been deposited in the GenBank database under accession number U88242.

RESULTS

Isolation of mutants that require murein tripeptide for growth.

Prior evidence indicated that E. coli could utilize murein tripeptide in place of Dap for synthesis of cell wall murein (9). In order to search for mutants which could not utilize tripeptide, E. coli TP981 (opp⁺ ampG::kan lysA⁺ dapD2) was constructed. Opp was assumed to be required for uptake of the tripeptide, based on the report of Goodell and Higgins that an opp deletion strain and an oppA nonpolar mutant did not utilize tripeptide (10). Since GlcNAc-anhMurNAc-tripeptide, transported into the cell via AmpG, is the principal source of cytoplasmic tripeptide (20, 32), ampG::kan was introduced to prevent reutilization of tripeptide originating from this source. Dap decarboxylase (LysA) was present to destroy free Dap that might be released from tripeptide so that it could not be used for growth. It was shown that TP981 required Dap for growth and that murein tripeptide could replace the required Dap.

Aliquots (10 μl) of the mutagenized TP981 culture (see Materials and Methods) plated on L agar-Cm-Dap agar plates yielded about 300 Cm^R colonies. These master plates were replicated onto media containing murein tripeptide in place of Dap. From about 90 such master plates, four mutants were obtained which did not grow on tripeptide-containing medium.

Test of the murein tripeptide-requiring mutants for Opp function.

Since oligopeptides are normally transported into the cytoplasm by Opp, one class of mutants isolatable by the procedure employed would be opp mutants. Therefore, the mutants were tested for resistance to triornithine, a toxic tripeptide dependent on Opp for transport into the cell (4). Two of the mutants proved to be resistant to triornithine, a fact indicating a possible defect in the Opp pathway. The other two mutants remained sensitive to triornithine, a fact indicating mutations in loci other than opp, and sequencing of the DNA flanking the transposon insertions revealed that the two triornithine-sensitive mutants were identical.

Identification of the mppA locus in the triornithine-sensitive mutant TP984.

P1vir grown on selected strains from the collection described by Singer et al. (40) were used for mapping by transduction. The recipient was TP984 Cm^R opp⁺. When transduced with P1vir grown on zci-3117::Tn10kan, which is very close to opp at 28.0 min, all Kan^R transductants remained Cm resistant, indicating that the mutation was unlinked to opp. Transduction with P1vir grown on zda-3061::Tn10 (30.4 min) converted 24 out of 40 Tet^R transductants to Cm sensitivity, indicating that the mutation is about 2 min clockwise from the opp operon.

Sequencing with primers complementary to each end of the cat gene present in the transposon revealed an ORF with the transposon and the characteristic 9-base pair repeat sequence (GTTAAAGCG) located 171 nucleotides upstream from the termination codon. The cloned HindIII fragment (>10 kb) contained the complete gene, while the cloned PstI fragment (4.5 kb) lacked a short N-terminal region because of the presence of a PstI site.

Comparison of the amino acid sequence of the ORF (accession no. U88242) with sequences in the database by means of the BLAST algorithm (2) revealed that the mutation was in a previously unidentified gene whose product is about 46% identical to the amino acid sequence of OppA, the periplasmic binding protein of the opp operon (Fig. 1). The gene was named murein peptide permease A (mppA) because of its structural and functional similarity to oppA and because, as shown below, it is a periplasmic protein required for the uptake of murein tripeptide. Hybridization of a ³²P-end-labeled, single-stranded 18-mer mppA probe to a membrane containing an ordered set of Kohara phage clones with overlapping segments of the complete E. coli genome (29) showed that mppA was present in both λ260 and λ261. Comparison with the current physical map (5) places the gene at 29.95 min. Our mppA sequence agrees perfectly with the sequence of ORF_ID o260#13 (accession no. D90771) from the E. coli genome sequence (1).

FIG. 1.

Comparison of the amino acid sequences of MppA and OppA from E. coli. Identical amino acids are identified by their single-letter codes, conservative amino acids are indicated by +. Amino acid numbering for both precursor proteins is shown on the right. Aligned MppA and OppA protein sequences are derived from the nucleotide sequences with accession numbers U88242 and P23843, respectively. The shaded area represents the signal sequences of MppA and OppA, and the arrow shows the site of signal peptide cleavage.

Complementation of mppA::miniTn10Cm in TP984.

Expression vectors carrying the wild-type mppA gene or an alternate form starting at an ATG 21 bases from the presumed start codon (see Materials and Methods) were tested for their abilities to complement TP984. As illustrated in Fig. 2, both forms complemented the mutant for growth on tripeptide in the presence of IPTG and had no effect on sensitivity to triornithine. Actually, the basal level of expression from the Trc promoter was sufficient for complementation by pMLD1285, whereas pMLD1493 did not complement unless IPTG was added (data not shown).

FIG. 2.

Complementation of mppA by pMLD1285 or pMLD1493. The plate on the left shows growth of the wild-type and mppA cultures on L medium containing Dap, Amp, and 0.1 mM IPTG. The plate in the center shows growths of the same cultures on L medium containing murein tripeptide, Amp, and 0.1 mM IPTG. The plate on the right demonstrates that none of the strains can grow on M9 medium containing 0.1% Casamino Acids, 1 μg of thiamine per ml, Dap, and 500 μM triornithine. All grew on the M9 medium in the absence of triornithine (data not shown). Sector 1, AT980/pTrc99A (the parent); sector 2, TP985/pTrc99A (mppA); sector 3, TP985/pMLD1285 (pTrcmppA⁺); sector 4, TP985/pMLD1493 (pTrcmppA⁺).

Overproduction of MppA and demonstration of its presence in the periplasm.

Since MppA is expected to be a periplasmic binding protein, it is presumably first made in a precursor form and processed to the mature periplasmic form in a manner similar to that of OppA. Because OppA is a major protein in the periplasm (15), to test for overproduction of MppA, which is predicted to be of similar size, the expression plasmids were introduced into E. coli CH483, a strain in which the opp operon is deleted (18). As shown in Fig. 3, induction with IPTG caused overproduction of a protein of approximately 58 kDa that is present predominantly in the periplasm. For unknown reasons, the plasmid with the longer signal peptide (pMLD1285) overexpresses MppA significantly better than pMLD1493 (Fig. 3). MppA produced from pMLD1285 has an arginine in place of lysine at residue 49 of the precursor protein, presumably a mutation introduced during PCR, but this is unlikely to be the reason for the superior production of MppA from pMLD1285.

FIG. 3.

Overproduction and cellular localization of MppA in E. coli cells. Crude protein extracts (total soluble proteins) and periplasmic extracts prepared from strain JM83 or CH483 carrying either the pTrc99A plasmid vector, the pMLD1285 plasmid, or the pMLD1493 plasmid were analyzed by SDS-PAGE. Shown are crude extract from JM83 (lane A); crude extracts from CH483 cells carrying either pTrc99A (lane B), pMLD1285 (lane C), or pMLD1493 (lane D) after growth for 4 h in the presence of 5 mM IPTG; and periplasmic extracts from the strains listed for lanes A to D (lanes E to H). Molecular mass (mw) standards indicated on the left are as follows: pyrophosphorylase b (94 kDa), bovine serum albumin (67 kDa), ovalbumin (43 kDa), and carbonic anhydrase (30 kDa).

Periplasmic MppA is processed by signal peptide cleavage.

To determine if periplasmic MppA lacks the putative signal sequence, the N-terminal amino acid sequence of the highly overproduced 58-kDa protein expressed from pMLD1285 was determined from the corresponding band transferred to an Immobilon P membrane following SDS-PAGE. The N-terminal sequence found, AEVPSGTVLA, is identical to amino acid residues 30 to 39 of MppA (Fig. 1). This confirms that the overproduced protein is the processed form of MppA. Processing of the MppA precursor occurs between the two alanines at positions 29 and 30; the corresponding signal peptide cleavage site in OppA is between two alanines located at positions 26 and 27 (Fig. 1). However, the signal sequences of the two proteins differ remarkably. MppA has nine hydroxyl- or thiol-containing amino acids in the hydrophobic region, whereas OppA has only a serine (Fig. 1). The mature protein has a predicted molecular weight of 57,618 that agrees with the size deduced from its mobility on SDS-polyacrylamide gels. It has a pI of 8.32.

Uptake of murein tripeptide requires components of Opp.

In E. coli CH483, the entire opp operon is deleted (18). However, the CH483 genome carries the mppA locus because a DNA fragment of the expected size was obtained by PCR (with Taq polymerase) on CH483 chromosomal DNA with 5′-TATGTGCTTTACCGCATTTTG-3′, which begins 153 nucleotides upstream of the start codon on the mppA coding strand, and 5′-ATTTGAAGATTATCGATA-3′, which begins 154 nucleotides downstream of the stop codon on the antisense strand, as primers. The PCR product was cloned into pTrc99A (cut with EcoRV and a T-overhang added with Taq polymerase in the presence of dTTP as described in reference 13), and the resulting plasmid complemented the mppA mutant, TP984, for growth on tripeptide (data not shown). To test if Opp components might be involved in supporting growth on tripeptide as the source of Dap, dap was introduced into CH483 by cotransduction with zaa1::Tn5 from E. coli VC6121 to yield strain TP986 (ΔoppABCDF, dap). TP986 failed to grow on L agar plus murein tripeptide, whereas TP986 containing pB2 (oppBCDF⁺) grew well on L agar plus murein tripeptide (Table 2, experiment 1). This indicates that some or all of the membrane and cytoplasmic components of Opp are necessary and sufficient for transfer of murein tripeptide from MppA into the cytoplasm. OppA is clearly not required.

Is MppA specific for murein tripeptide?

To investigate whether MppA could facilitate transport of non-cell wall peptides via OppB, OppC, OppD, and OppF, we determined the ability of the tripeptide Pro-Phe-Lys to provide proline for growth of proline auxotrophs. As shown in Table 2, experiment 3, E. coli SS5013 (pro his trp met ΔoppABCDF) did not grow in minimal medium when Pro-Phe-Lys was used as the source of proline. However, E. coli SS5013 carrying plasmid pB2 (oppBCDF) did grow in the same medium (although growth was poor), indicating that MppA (or the third OppA paralog, ORF-f535, shown in Fig. 4) could facilitate transport of the proline-containing tripeptide when OppB, OppC, OppD, and OppF were present and that OppA was not essential for its uptake. E. coli TP989, an SS5013 derivative carrying the mppA::Cm mutation, could no longer grow on Pro-Phe-Lys in the presence of pB2, strongly suggesting that MppA was involved in peptide uptake. Introduction into TP989/pB2 of a second plasmid, pHSL1, compatible with pB2 and carrying the mppA locus, restored growth, confirming that MppA directly participated in peptide transport.

FIG. 4.

Alignment of E. coli MppA with its paralogs and selected orthologs. Positions with identical residues for at least five of the eight sequences are solid black, and gaps are indicated by dashes. The sequences were aligned by the Clustal method (MegAlign; DNASTAR). The sequences shown are as follows: Sty OppA, S. typhimurium OppA (17); Eco OppA, E. coli OppA (22); Eco ORF-f535, E. coli ORF-f535 (GenBank accession no. U28377); Eco MppA, E. coli MppA (GenBank accession no. U88242); Hin HI1124, H. influenzae OppA ortholog (7); Hin HI0213, H. influenzae OppA ortholog (7); Bsu SpoOKA, B. subtilis OppA (36, 37); and Bsu DPPE, B. subtilis DPPE (28). The putative hexapeptide motif in MppA and HI0213 and the two cysteines in Sty OppA, Eco OppA, and Hin HI0213 are highlighted in light grey.

To determine if Pro-Phe-Lys could also be transported via OppA, proC derivatives of TP981 (opp⁺) and TP984 (opp⁺ mppA::Cm) were tested. Pro-Phe-Lys supported growth of both strains (Table 2, experiment 4). Thus, OppA, in the absence of MppA, can also allow growth of a proline auxotroph on Pro-Phe-Lys. Though TP989 (mppA::Cm ΔoppABCDF)/pB2 (oppBCDF)/pHSL1 (mppA) grew poorly during 18 h on Pro-Phe-Lys, when the compatible plasmid, pBφ30, expressing oppA, replaced pHSL1, significantly more rapid growth occurred (Table 2, experiment 3). Since both pHSL1 and pBφ30 are pACYC derivatives, higher oppA gene dosage cannot explain this result. A more likely explanation is that Pro-Phe-Lys binds preferentially to OppA rather than MppA. We have recently observed that a 15-fold excess of Pro-Phe-Lys does not compete with murein tripeptide for binding to MppA (∼95% purity) in vitro (unpublished data). This is consistent with our observation that MppA transports Pro-Phe-Lys poorly and that MppA transports l-Ala-γ-d-Glu-meso-Dap efficiently.

Genomics of mppA.

Comparison of the amino acid sequence of MppA with the sequences in the GenBank protein database revealed that, in addition to a 46% identity with OppA from E. coli and Salmonella typhimurium, MppA is about 39% identical to E. coli ORF-f535 (accession no. U28377), 28% identical to Bacillus subtilis Spo0KA/OppA (36, 37), 25% identical to B. subtilis DPPE (28), and 39 and 24% identical to two ORFs in Haemophilus influenzae (7). A multiple sequence alignment of these seven paralogs and orthologs of MppA is shown in Fig. 4.

A comparison with the two ORFs in H. influenzae is interesting. ORF HI1124 (accession no. U32792), although most closely related to MppA, is clearly the ortholog of OppA (50% identity) because it is part of the opp operon of H. influenzae (6). Examination of the alignments in Fig. 4 reveals that the presence of two cysteines is not critical for OppA, even though the crystal structure of S. typhimurium OppA (43) shows the two cysteines (Cys296 and Cys442) forming a disulfide bridge that straddles the peptide ligand.

ORF HI0213 contains a hexapeptide sequence (179 RIELDK 184) which is tantalizingly similar to a motif (174 KIQLDK 179) in the aligned sequence of MppA (Fig. 4). It will be interesting to see if this hexapeptide motif (K/RIQ/ELDK) can serve as a signature for functional MppA orthologs in other gram-negative bacteria. At present, there are insufficient data to determine whether MppA is found in different bacterial species or is restricted to E. coli.

Examination of the E. coli genome sequence surrounding mppA at 29.95 min shows that there are no genes with partial identity to opp genes, as might be expected if an entirely independent Mpp transport operon were present. Immediately upstream of mppA, reading clockwise as with mppA, we find xapR (pndR) (accession no. P23841), a putative xanthosine operon regulatory protein, and immediately downstream is yggB (accession no. P11666) that is present in the opposite orientation (1). Furthermore, mppA has a putative transcription termination signal consisting of a 30-nucleotide stem and loop beginning 11 nucleotides after the stop codon, which also suggests that mppA is not at the beginning of a special transport operon.

DISCUSSION

Recycling of cell wall peptides required the presence of an enzyme that would link the tripeptide l-Ala-γ-d-Glu-meso-Dap to UDP-N-acetylmuramate, thereby directly reintroducing it into the biosynthetic pathway for wall synthesis (9, 20). In seeking this hypothetical murein tripeptide-adding enzyme, we undertook a search for mutations in the putative gene. We reasoned that a mutant of a Dap-requiring strain which lacked the tripeptide-adding enzyme would not be able to grow on media in which murein tripeptide replaced Dap. A mutant isolation procedure was designed which relied on the uptake of the exogenously added murein tripeptide for growth. The mutant sought should grow on Dap but not on tripeptide. Concurrently with our mutant search, we identified the tripeptide-adding enzyme, or murein peptide ligase gene, mpl, by its homology to murC and demonstrated that an mpl-null mutant could not grow when Dap was replaced by murein tripeptide (29). Thus, our genetic screen for mutant isolation was sound.

However, to our surprise, instead of finding a mutant of mpl, as was suggested earlier (34), we discovered that one of the mutations was in a gene required for the uptake of murein tripeptide, namely, mppA. The other two mutants we isolated were in genes required for a functional Opp system because these mutants were resistant to triornithine, which requires Opp for uptake.

The new gene, mppA, mapping at 29.95 min, codes for the precursor of a periplasmic binding protein that is required for uptake of murein tripeptide. We have cloned and overproduced MppA and shown that the mature protein is present in the periplasm (Fig. 3). Plasmids carrying the mppA gene under control of the IPTG-inducible Trc promoter complement the mppA mutant when induced with IPTG (Fig. 2).

MppA and OppA are similar in size, and their amino acid sequences are about 46% identical (Fig. 1). OppA is an abundant periplasmic protein, whereas E. coli CH212 oppA462, which lacks OppA, contains only a trace protein of similar size, suggesting that MppA is a minor component in the periplasm (16). CH212 does not transport murein tripeptide (10), and this led us to suspect that, contrary to the published report (10), oppA462 is a polar mutation because, as we have shown, the absence of OppA does not prevent the utilization of tripeptide via MppA and the membrane and cytoplasmic Opp components. In fact, we have now demonstrated that oppA462 is a polar mutation because CH212 (proA2) will grow on Pro-Phe-Lys as the source of proline only when carrying plasmid pB2 (oppBCDF). Thus, a nonpolar oppA mutation is not presently available in E. coli.

Two of the tripeptide-requiring mutants are resistant to triornithine, which is usually diagnostic for a defect in Opp function (4). We have recently found that one of these mutants has its transposon located in oppB and is complemented by pB2, a plasmid expressing OppBCDF (unpublished data). This is consistent with our result demonstrating that TP986 (ΔoppABCDF, dap)/pB2 will grow on murein tripeptide and strongly suggests that MppA uses some or all of the membrane and cytoplasmic Opp components for transfer of murein tripeptide into the cytoplasm.

The other triornithine-resistant mutant was found to have its transposon in the promoter region of groESL, and this results in greatly reduced levels of the encoded proteins in the mutant (25a). This is, to our knowledge, the first example indicating that GroESL chaperonin function is essential for formation of a functional ABC transporter.

Since the periplasmic binding proteins, MppA and OppA, are about 46% identical and perform similar functions, they, as well as their orthologs, presumably evolved from a common ancestor following gene duplications. For MppA to use the Opp pathway, one would expect that MppA and OppA have a common surface that interacts with one or both of the membrane-bound components of the Opp pathway. This would be comparable to the case with the hisJ and argT gene products of S. typhimurium, another example of apparent gene duplication that gave rise to the periplasmic binding proteins required for transport of histidine and arginine (15). Both proteins dock on shared membrane-bound components in order to discharge their cargo into the cytoplasm (15).

OppA is a remarkably nonspecific binding protein believed capable of binding most tri-, tetra-, and pentapeptides linked together by normal α-peptide bonds. Tame et al. (43) have recently explained the structural basis for this sequence-independent peptide binding. The peptide is completely enclosed in a voluminous hydrated cavity that can accommodate the various amino acid side chains, and all bonding is directed to the peptide bonds between the α-amino acids. In contrast, the second bond in murein tripeptide is an amide bond between the γ-carboxyl of d-Glu and the l-amino group of meso-Dap. It is thus conceivable that the binding sites in OppA are not positioned correctly to accommodate peptides containing this novel γ-d-glutamyl amide bond.

While MppA is essential for the transport of murein tripeptide, MppA can also transport ordinary α-linked tripeptides such as Pro-Phe-Lys, although, judging by the growth response, MppA is much poorer than OppA for the transport of Pro-Phe-Lys. The binding constants of different tripeptides for OppA vary over a wide range; e.g., triornithine is at least 600-fold less able to bind than Ala-Phe-Gly (12). Presumably, a similar variation in the abilities of various tripeptides to bind to MppA (and the other OppA paralog in E. coli) will be found, but, judging by the example of Pro-Phe-Lys, it seems likely that α-linked tripeptides will prove to be poor ligands for MppA compared to peptides that contain the γ-d-glutamyl bond.

Further complicating the transport of peptides in E. coli, a protein called OppE is required for the uptake of certain tripeptides and is independent of Opp (3). This is similar to the situation with S. typhimurium, in which TppB is a protein essential for the uptake of some tripeptides. However, these transporters apparently do not depend on periplasmic binding proteins for their function, since only a single mutation site is known for each.

MppA binds murein tripeptide precisely because it may have evolved for this purpose. Following gene duplications, one paralog, OppA, evolved to relative nonspecificity; another, MppA, evolved to high affinity for the unique murein tripeptide; and the properties of the third OppA paralog in E. coli (ORF-f535 [Fig. 4]) are completely unknown at present. E. coli periplasm contains lytic transglycosylases (39) which degrade at least 30% of the murein sacculus to GlcNAc-anhMurNAc-tri- and -tetrapeptide each generation (8, 32) as well as a periplasmic MurNAc-l-alanine amidase which can release the peptide from the anhydromuropeptide, although this activity is very poor (35). The principal pathway for uptake and reutilization of tripeptide from the cell wall is indirect, as it requires the transport of GlcNAc-anhMurNAc-tripeptide into the cell via the AmpG permease, followed by cleavage by AmpD amidase to release tripeptide (19, 20). Therefore, the question of why E. coli has this capacity to bind and import free murein tripeptide remains unanswered. It would seem that MppA serves some purpose other than recycling.

Goodell (9) estimated a K_m of 2 μM tripeptide for the incorporation of exogenously added tripeptide into murein sacculi by intact cells, suggesting that the true K_m may be even lower. Goodell and Schwarz (11) have demonstrated that E. coli accumulates murein tri- and tetrapeptides as well as the dipeptide Dap-d-Ala in the medium during growth, and it has recently been shown that medium conditioned by E. coli growth augments transcription from the two E. coli promoters upstream of the ftsQAZ gene cluster required for cell division (41). One of the promoters (P₁) is RpoS-stimulated, and the other (P₂) is regulated by SdiA, which is a member of the LuxR subfamily of transcriptional activators involved in quorum sensing (autoinduction) (8, 41). The nature of the factors in E. coli-conditioned medium that stimulate transcription from the P₁ and P₂ promoters is unknown, but it is unlikely that these factors include an N-acyl-l-homoserine lactone (41), especially since E. coli lacks an ortholog of the Vibrio fischeri LuxI required for production of N-acyl-l-homoserine lactone (8).

Since the recycling pathway presumably supplies a constant level of tripeptide in the cytoplasm, it is difficult to imagine how entry of a smaller amount through the MppA pathway would have any significant effect thereon. It is tempting to speculate that upon E. coli growth to a quorum (8), the accumulated murein peptides (11) may bind to MppA and, upon contact with the required Opp components or another specific membrane receptor, may stimulate expression from selected promoters through a signal transduction pathway. Experiments to test these possibilities and that may illuminate the physiological function of MppA are under way.

An example of periplasmic binding proteins mediating signal transduction is the chemotactic response (42). The periplasmic binding proteins for ribose, galactose, and glucose, when loaded with their specific sugars, trigger the chemotactic response by binding to the Trg receptor (42). Likewise, the periplasmic binding protein of E. coli dipeptide permease activates the chemotactic response through the Tap receptor when loaded with dipeptide (27, 42). In a similar vein, a signaling pathway dependent on liganded MppA may exist for chemotaxis or may affect entry of E. coli into the stationary phase by regulating expression of rpoS and/or the RpoS-dependent promoters (14).