This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Goffin, C. Articles by Ghuysen, J.-M. Search for Related Content PubMed PubMed Citation Articles by Goffin, C. Articles by Ghuysen, J.-M.

 Previous Article  |  Next Article 

Microbiology and Molecular Biology Reviews, December 2002, p. 702-738, Vol. 66, No. 4
1092-2172/02/$04.00+0     DOI: 10.1128/MMBR.66.4.702-738.2002
Copyright © 2002, American Society for Microbiology. All Rights Reserved.

Biochemistry and Comparative Genomics of SxxK Superfamily Acyltransferases Offer a Clue to the Mycobacterial Paradox: Presence of Penicillin-Susceptible Target Proteins versus Lack of Efficiency of Penicillin as Therapeutic Agent

Center for Protein Engineering, Institut de Chimie, University of Liège, B-4000 Sart Tilman, Liège, Belgium

SUMMARY

INTRODUCTION

PREDICTIVE STUDIES

LIPID II PRECURSORS: PEPTIDOGLYCAN ASSEMBLY

SxxK ACYLTRANSFERASE SUPERFAMILY

STRUCTURE-ACTIVITY RELATIONSHIPS OF ß-LACTAM ANTIBIOTICS

GROUP I SxxK ACYLTRANSFERASES: IMPLICATED IN (4->3) PEPTIDOGLYCAN BIOCHEMISTRY

Class A PBP Fusions: Nascent (4->3) Peptidoglycan

Class B PBP Fusions: Morphogenetic Apparatus

Free-Standing PBPs: Auxiliary Cell Cycle Proteins

E. coli PBP Fusions as Targets for ß-Lactam Antibiotics

Mosaic PBP Fusions

GROUP II SxxK ACYLTRANSFERASES

GROUP III SxxK ACYLTRANSFERASES: PENICILLIN-RESISTANT PROTEIN FUSIONS

Class B Penr Protein Fusions of Gram-Positive Bacteria

Class A Penr Protein Fusions of Gram-Negative Bacteria

SxxK ACYLTRANSFERASES IN SPECIFIC ORGANISMS

Class A PBP Fusions of M. tuberculosis, M. leprae, M. smegmatis, and C. diphtheriae

Class A Penr Protein Fusions of M. tuberculosis, M. leprae, M. smegmatis, and C. diphtheriae

Subclass B3 and B-Like I PBP Fusions of M. tuberculosis, M. leprae, and C. diphtheriae

Class B-Like II Penr Protein Fusions of M. tuberculosis, M. leprae, and C. diphtheriae

Free-Standing PBPs, ß-Lactamases, and Related Polypeptides of M. tuberculosis and M. leprae

PBP Fusions and Penr Protein Fusions of S. coelicolor

Detected versus Predicted PBPs of M. tuberculosis H37Ra

WHERE NEXT?

ACKNOWLEDGMENTS

REFERENCES

Top Next References

INTRODUCTION

Top Previous Next References

 
The bacterial wall peptidoglycans are covalently closed, net-like polymers (77, 203). The glycan chains are made of alternating ß-1,4-linked N-acetylglucosamine and N-acetylmuramic acid residues. The D-lactyl groups of the muramic acid residues are amidated with L-alanyl--D-glutamyl-(L)-diaminoacyl-D-alanine stem tetrapeptides and L-alanyl--D-glutamyl-L-diaminoacid stem tripeptides. The stem peptides can be branched, in which case the amino group of the diaminoacid residue is substituted by an additional amino acid residue or a short peptide. Unbranched or branched stem peptides belonging to adjacent glycan strands are covalently linked, resulting in polymeric peptidoglycan.

All peptidoglycan-containing bacteria in the exponential phase of growth manufacture a (43) peptidoglycan (Fig. 1 ) in a penicillin-susceptible manner. Peptidoglycan crosslinking extends from the carboxy-terminal D-alanine residue at position 4 of a stem tetrapeptide to the lateral amino group at position 3 of another, unbranched or branched, stem peptide. The (43) interpeptide linkages or cross-bridges are made by specialized acyltransferases which are immobilized by penicillin in the form of stable, enzymatically inactive penicillin-binding proteins (PBPs) (214). Escherichia coli, the mycobacteria, and leprosy-derived corynebacteria (111) produce unbranched (43) peptidoglycans with meso-diaminopimelic acid at position 3 of the stem peptides. Peptidoglycan crosslinking is mediated by direct (D)-alanyl-(D)-meso-diaminopimelic acid interpeptide linkages (Fig. 1). In mycobacteria, however, muramic acid is either acylated or glycolylated (12). The -carboxylate of D-glutamic acid can be amidated. A glycine residue substitutes for the L-alanine residue at the amino end of the stem peptides in Mycobacterium leprae.

View larger version (25K): [in this window] [in a new window]   FIG. 1. (43) peptidoglycans. D-Alanyl-diaminoacyl interpeptide linkages and cross-bridges (boxed) between L-Ala--D-Glu-(L)-diaminoacyl-D-alanine stem peptides. The diamino acid residues are meso-diaminopimelic acid (Dpm), L, L-diaminopimelic acid, or L-lysine. The stem peptides are unsubstituted at position 3 in Mycobacterium, Corynebacterium, and Bacillus spp. and E. coli. They are substituted at position 3 by one or several additional amino acid residues () in the other organisms shown. G, N-acetylglucosamine; M, N-acetylmuramic acid (see Fig. 4); COX = COOH or CONH2. In S. pneumoniae, the cross-bridges are N-(L-alanyl-L-alanyl)- or (L-alanyl-L-seryl)-L-lysine. In S. aureus, the cross-bridges comprise five glycine residues or three glycine and two L-serine residues.

View larger version (24K): [in this window] [in a new window]   FIG. 2. (33) peptidoglycans. Diaminoacyl-diaminoacyl interpeptide linkages and cross-bridges (boxed) between L-Ala--D-Glu-(L)-diaminoacyl-D-alanine stem peptides. The stem peptides are unsubstituted at position 3 in Mycobacterium spp. and E. coli. They are substituted at position 3 by Gly in Streptomyces spp. and Clostridium perfringens and by ß-D-Asp in E. faecium. For details, see the legend to Fig. 1.

Subsequently, it was found that Escherichia coli manufactures a (33) peptidoglycan (Fig. 2) in increasing proportions of total peptidoglycan and becomes increasingly more resistant to ß-lactam antibiotics as the generation time increases (227). Penicillin-induced lysis also causes an increased proportion of (33) peptidoglycan (127). The contribution of (33) peptidoglycan in Aeromonas spp., Acinetobacter acetoaceticus, Agrobacterium tumefaciens, Enterobacter cloacae, Proteus morganii, Pseudomonas aeruginosa, Pseudomonas putida, Salmonella enterica serovar Typhimurium, Vibrio parahaemolyticus, Yersinia enterocolitica, and E. coli grown to late exponential phase varies from 1% to 45% of total peptidoglycan (183, 184). Finally, Enterococcus faecium is also a (33) peptidoglycan manufacturer (Fig. 2). A laboratory mutant has been isolated which resists penicillin in the exponential phase of growth, conditions under which it manufactures a wall peptidoglycan of the (33) type exclusively (145, 205).

From the foregoing, it follows that, in all likelihood, (33) peptidoglycan crosslinking is carried out by acyltransferases that escape penicillin action and that, in particular genetic backgrounds or under specific growth conditions, the penicillin-resistant (33) peptidoglycan assembly molecular machine can substitute for the penicillin-susceptible (43) peptidoglycan assembly molecular machine. This conclusion raises questions of fundamental and practical importance, related, in particular, to the mycobacterial pathogens.

E. coli has a 4.60-Mb genome (24). M. tuberculosis H37Rv has a 4.41-Mb genome (38). Mycobacterium leprae has a 3.27-Mb genome (39) that shows extensive decay and downsizing. M. tuberculosis and M. leprae share about 1,500 genes. As stated above, E. coli and Mycobacterium spp. manufacture similar, unbranched, meso-diaminopimelic acid-containing peptidoglycans of the (43) and (33) types. The nonpeptidoglycan wall polymers, however, are different. In E. coli, several lipoproteins are linked to the peptidoglycan. The 56-amino-acid residue Braun's lipoprotein (29) contains one fatty acid bound as an amide to the amino group of a cysteine residue and two fatty acids bound as esters to the hydroxyl groups of S-glyceryl cysteine (Fig. 3). The fatty acids are embedded in the outer membrane. Lipoprotein molecules occur both in a free form and in covalent linkage with the underlying peptidoglycan. The linkage is an amide bond between the -amino group of the L-lysine residue at the carboxy end of the lipoprotein and the carbonyl group at the L-center of the meso-diaminopimelic acid residue at position 3 of a stem tripeptide of the peptidoglycan.

View larger version (12K): [in this window] [in a new window]   FIG. 3. Braun's lipoprotein-peptidoglycan complex of E. coli. For details, see the legend to Fig. 1.

M. tuberculosis and M. leprae have different life styles. M. tuberculosis enters host macrophages at cholesterol-rich domains of the plasma membrane (75). It survives in the macrophage by preventing fusion of the mycobacterial phagosomes with lysosomes (50, 201, 218, 236). It may persist inside macrophages lodged in calcified structures of the lungs and be reactivated decades after initial infection (153). It grows in synthetic media with a generation time of 12 to 24 h. Tuberculosis, once almost vanquished, resurged in the late 1980s because of the emergence of multidrug-resistant strains (25, 250). Often acting in deadly combination with AIDS, it kills more than 2 million people each year. Mycobacterium leprae enters the body via the mucosal linings of the nose or through open wounds. It shows a marked tropism for myelin-producing Schwann cells, which insulate nerves (187, 188). It may be dormant for years until the body's immune system attacks the infected cells, destroying the nerves and degrading soft tissues and even bones. It is unculturable but can be grown in the nine-banded armadillo and the footpad of mice with an estimated generation time of about 2 weeks. Leprosy still flourishes in developing countries, where more than 750,000 people contract the disease each year.

Consistent with its ability to manufacture a (43) peptidoglycan, M. tuberculosis H37Ra (ATCC 25177, an attenuated laboratory strain) produces four major PBPs of 94, 82, 52, and 37 kDa (35). The inactivation of the 94-, 82-, and 52-kDa PBPs in cells grown to mid-exponential phase is associated with antibacterial activity. ß-Lactamase production (240), more than the relatively low permeability of the cell envelope, is the major determinant of resistance (35, 63, 112, 185). Consequently, the MICs of ampicillin associated with the ß-lactamase inactivator sulbactam are 0.1 µg ml^-1 for M. tuberculosis H37Ra and 2 µg ml^-1 for M. tuberculosis H37Rv (35). The drug combination is bactericidal to M. tuberculosis in exponential-phase cultures. It is bactericidal to M. tuberculosis multiplying in macrophages (35) and to M. leprae multiplying in mouse footpads (180, 189).

Paradoxically in view of the above data, the ß-lactam antibiotics are not effective therapeutic agents for the treatment of tuberculosis and leprosy (101). This lack of efficiency could be due to a shift of peptidoglycan synthesis from the (43) type to the (33) type. Information that can help apprehend the problem was sought from comparative genomics of bacterial species (E. coli, Enterobacter faecium, Mycobacterium spp., and others) which have the ability to assemble, presumably from the same pool of precursors, peptidoglycans of the (43) and (33) types. Searches were also made for the positions of genes on the chromosomes because operons are common in prokaryotes and genes that are neighbors tend to be functionally linked.

PREDICTIVE STUDIES

Top Previous Next References

 
Data concerning unfinished genome sequences were from the Institute for Genomic Research (TIGR) website at http://www.tigr.org. They were produced at TIGR with the support of the National Institute of Allergy and Infectious Diseases for Mycobacterium avium 104, at the Sanger Center/Institut Pasteur/VLA Weibridge with the support of MAFF/Beowulf Genomics for Mycobacterium bovis AF2122/97, at the Sanger Center/World Health Organization/Public Health Laboratory with the support of Beowulf Genomics for Corynebacterium diphtheriae NCTC 13129, and at the Sanger Center/John Innes Center with the support of BBSRC/Beowulf Genomics for Streptomyces coelicolor A3 (2).

Amino acid sequence similarities were searched with the Basic Local Alignment Search Tool Blast programs of the National Center for Biotechnology Information (NCBI) from their website at http://www.ncbi.nlm.nih.gov/Blast (4). Program BlastP compares a query amino acid sequence against a protein database. Program tBlastN compares a query amino acid sequence against a nucleotide sequence database dynamically translated in all six reading frames. E. coli, M. tuberculosis, M. leprae, C. diphtheriae, and S. coelicolor proteins were searched in the nonredundant protein database with BlastP and in the finished and unfinished genome database with tBlastN. The Expect value E of the NCBI's programs allows estimation of how far from the background noise is the similarity between aligned sequences (3). For values of 0.01, E becomes equivalent to the probability P that the sequences align by chance (118). P values smaller than 10^-3 to 10^-4 (depending on the sizes of the databanks) indicate statistically significant similarity.

The P values depend on the number and size of the gaps introduced to optimize the alignments. When the P value is small, equal or close to 0.0, the percentage of identical amino acid residues allows the sequences to be hierarchically classified. Combining the P values, the lengths of the overlapping regions, and the percentages of identities (I) present in the aligned regions gives an estimate of the extent of similarity between the sequences under comparison. Proteins encoded by genes having a common ancestor are homologues (193). Proteins encoded by genes related by duplication within a genome followed by diversification normally evolve new functions (103); they are paralogues. Homologous proteins encoded by genes in different organisms separated by speciation are orthologues (51, 103).

Protein orthologues that are essential and related by small P values and large I values normally retain similar biological functions. This conclusion is especially pertinent when low P values and large I values apply, over the entire sequences, to protein fusions for which the constitutive modules evolved from different protein ancestors. In contrast, little information is obtained when the similarities are restricted to peptide stretches or amino acid groupings. Conserved motifs defining the boundary of a catalytic center give valuable information on the catalytic mechanism, not on the specificity and fate of the reactions catalyzed.

LIPID II PRECURSORS: PEPTIDOGLYCAN ASSEMBLY

Top Previous Next References

 
Lipid II molecules are the immediate biosynthetic precursors of the wall peptidoglycans (102). A disaccharide peptide exposed on the outer face of the plasma membrane is linked to a C₅₅H₈₉ undecaprenyl carrier via a pyrophosphate bridge involving carbon C-1 of N-acetylmuramic acid. The stem peptide borne by N-acetylmuramic acid is a pentapeptide terminating in D-alanyl-D-alanine. It can be unsubstituted or branched at position 3.

In E. coli, the synthesis of lipid II (Fig. 4) involves an interchange of carriers that are compatible with the environments of the cell (231). UDP-N-acetylglucosamine is converted into UDP-N-acetylglucosamine-enolpyruvate by MurA and from this into UDP-N-acetylmuramic acid by MurB. The Ddl (ATP,ADP + P_i) ligase catalyzes the formation of a D-alanyl-D-alanine dipeptide, and the MurC, MurD, MurE, and MurF (ATP,ADP + P_i) ligases catalyze the formation of UDP-N-acetylmuramoyl pentapeptide by the sequential additions to UDP-N-acetylmuramic acid of L-alanine, D-glutamic acid, meso-diaminopimelic acid, and the preformed D-alanyl-D-alanine dipeptide. Then, MraY transfers the phospho-N-acetylmuramoyl pentapeptide from its uridylic carrier to a membrane-bound C₅₅-isoprenoid alcohol phosphate, giving rise to lipid I. MurG transfers the N-acetylglucosamine from its uridylic carrier to carbon atom C-4 of N-acetylmuramic acid, giving rise to lipid II. Somehow, lipid II flips over the membrane bilayer, and the disaccharide pentapeptide moiety is exposed on the outer face of the membrane (Fig. 4). The genes ddl, murC, murD, murE, murF, murG, and mraY reside in the dcw (or mra) cluster at the 2-min region of the chromosome (Fig. 5).

View larger version (30K): [in this window] [in a new window]   FIG. 4. Lipid II precursor (bottom) and polymeric (43) peptidoglycan (top) of E. coli. Reactions catalyzed by the glycosyl- and acyltransferases. For details, see the legend to Fig. 1.

View larger version (28K): [in this window] [in a new window]   FIG. 5. Organization of the division cell wall (dcw) gene clusters of M. tuberculosis (Mtu), M. leprae (Mle), Streptomyces coelicolor (Sco), and E. coli (Eco). All the coding sequences are on the same DNA strands. Open arrows, mtu3, mle3, sco3, and eco3 (synonymous to ftsI), coding for cell septation SxxK PBP fusions. Shaded arrows, with fts standing for the filamentous phenotype of thermosensitive E. coli mutants, ftsW, ftsQ, ftsA, and ftsZ code for cell septation-related, non-penicillin-binding proteins. Genes murE, murF, murX (synonymous to mraY), murD, murG, and murC code for enzymes of the lipid II-synthesizing pathway. In S. coelicolor, murC (shaded rectangle) is not in the dcw cluster. Stippled grey arrows, genes of a dcw cluster having no homologues in the other clusters. In M. tuberculosis, the genes inserted between mtu3 and murE probably code for a glycine-rich protein of repetitive structure (Rv2162c), a protein of the lincomycin-synthesizing pathway (Rv2161c), and proteins of unknown function (Rv2160c and Rv2159c). A gene homologous to E. coli ddl (coding for the ligase which catalyzes the formation of D-alanyl-D-alanine) is located elsewhere in the chromosomes of M. tuberculosis, M. leprae, and S. coelicolor. E. coli ftsA codes for a cell division, actin-like ATPase. Solid, dashed, and dotted lines connecting the dcw genes indicate the extents, in decreasing order, of similarity between homologous genes. The E. coli cell septation genes mraZ, mraW, and ftsL (not shown) are upstream from eco3.

In Staphylococcus aureus, there are four glycyl tRNAs. One is involved in protein synthesis. The three others seem to be used exclusively for peptidoglycan synthesis. Lipid II precursor molecules with one, three, and five glycine residues attached to the -amino group of the L-lysine residue are formed by the sequential actions of the glycine-adding enzymes FemX (also called FemhB), FemA, and FemB (128, 197). Loss of FemX is lethal. In some strains, the FemAB-like Lif and Epr incorporate an L-serine residue instead of a glycine residue within the branches at positions 3 and 5 (60). In Streptococcus pneumoniae, the N-(L-alanyl-L-alanyl)-L-lysine and N-(L-alanyl-L-seryl)-L-lysine branches are synthesized by MurM and MurN, respectively (64, 242), with MurM being responsible for the addition of the first amino acid residue to the -amino group of L-lysine.

Vancomycin-resistant enterococci are programmed to produce lipid II precursor molecules which terminate in D-alanyl-D-lactate (8, 136). Five enzymes are necessary and sufficient. The Zn²⁺-dependent VanX dipeptidase hydrolyzes the D-alanyl-D-alanine dipeptide produced by the Ddl ligase. VanH and VanA act sequentially to synthesize the depsipeptide D-alanyl-D-lactate, which is incorporated in lipid II rather than the dipeptide D-alanyl-D-alanine. The production of VanHAX is controlled by a two-component regulatory system that involves the transmembrane sensor kinase VanS and the transcription factor VanR, which becomes active when phosphorylated by VanS (8).

The conversion of the disaccharide peptide (depsipeptide) moiety of lipid II into polymeric (43) peptidoglycan requires the sequential actions of two transferases (Fig. 4). A glycosyltransferase catalyzes attack of carbon C-1 of N-acetylmuramic acid of a lipid II molecule by the acceptor nucleophile OH of carbon C-4 of N-acetylglucosamine of another lipid II molecule (or at the nonreducing end of a growing glycan chain; not shown). At each transfer, the delivery of a disaccharide peptide unit generates an undecaprenyl pyrophosphate which is dephosphorylated. The C₅₅-isoprenoid alcohol phosphate turns over the membrane bilayer so that the phosphate group faces the cytosol, allowing a new cycle to start.

In turn, the required D,D-acyltransferase (Fig. 4 and reaction I in Fig. 6) catalyzes attack of the carbonyl of the D-alanine residue at position 4 of a pentapeptide (depsipeptide) borne by a glycan chain, by the lateral amino group at position 3 of a peptide borne by another glycan chain. The reaction proceeds via the transitory formation of an N-acyl-D-alanyl-enzyme intermediate in which the D-alanine residue is linked as an ester to a serine residue at the catalytic center (69, 70, 78). The carbonyl donor involved in enzyme acylation is invariably a D-alanyl-D-alanine(-D-lactic acid) sequence. In contrast, the acceptor involved in enzyme deacylation can be the amino group of a glycine residue, the -amino group of an L-lysine residue, or an amino group borne by an L- or a D-configured carbon atom (Fig. 1).

View larger version (18K): [in this window] [in a new window]   FIG. 6. Acyl transfer reactions. Formation of (43) peptidoglycan interpeptide linkages by penicillin-susceptible D,D-N-acyl-D-alanyl-D-alanine transpeptidases (reaction I). Formation of (33) peptidoglycan interpeptide linkages by penicillin-resistant L,D-N-acyl-L-diaminoacyl-D-dipeptidyl transpeptidases (reaction II) and L,D-N-acyl-L-diaminoacyl-D-alanine transpeptidases (reaction III). Hydrolysis of stem pentapeptides into stem tetrapeptides by penicillin-resistant D,D-N-acyl-D-alanyl-D-alanine carboxypeptidases (reaction IIIa).

No one knows exactly how the (33) peptidoglycans are made in a penicillin-resistant manner. The (33) interpeptide linkages or cross-bridges could be made through the action of an L,D-acyltransferase that cleaves the bond between position 3 and position 4 of pentapeptide carbonyl donors, with release of the dipeptide D-alanyl-D-alanine (reaction II in Fig. 6). Alternatively, they could be made through the sequential actions of a D,D-carboxypeptidase catalyzing reaction IIIa (Fig. 6) and an L,D-acyltransferase that cleaves the bond between position 3 and position 4 of tetrapeptide carbonyl donors (reaction III in Fig. 6). Reactions IIIa and III each cause the release of a single D-alanine residue.

SxxK ACYLTRANSFERASE SUPERFAMILY

Top Previous Next References

 
The serine acyltransferases implicated in wall peptidoglycan assembly are members of a superfamily of SxxK serine enzymes. The term superfamily is used to include proteins with no statistically significant sequence similarities but with similar structures in the classical sequence-based families (169). With x denoting a variable amino acid residue, the acyltransferases of this superfamily have a specific bar code in the form of three motifs, SxxK, SxN (or analogue), and KTG (or analogue). The motifs occur at equivalent places and with roughly the same spacing along the polypeptide chains. In the three-dimensional structures, they are brought close to each other at the immediate boundary of the catalytic center between an all- domain and an /ß domain, the five-stranded ß-sheet of which is covered by -helices (79, 123, 124).

The serine residue of the invariant motif SxxK occupies a central position in the catalytic center and is central to the enzyme acylation and deacylation steps, which, mechanistically, are similar to the proteolytic reactions of the trypsin protein family. The mature SxxK acyltransferase of Streptomyces sp. strain K15 is 262 amino acid residues long. It is one of the smallest members of the superfamily (Fig. 7) (68). Diverging evolution gave rise to a constellation of SxxK acyltransferases with various amino acid sequences. Fusion to other polypeptide chains resulted in a combinatorial system of structural modules that contributed to a massive increase in functional diversity.

View larger version (53K): [in this window] [in a new window]   FIG. 7. Basic polypeptide fold of the acyltransferases of the SxxK superfamily and spatial disposition of the three catalytic center-defining amino acid groupings. The structure shown is that of the 262-amino-acid DD-transpeptidase-PBP of Streptomyces sp. strain K15. The SxxK acyltransferases comprise an all- domain (left side) and an /ß domain (right side). The invariant motif 1, S*xxK, where S* is the essential serine nucleophile, forms the amino-terminal turn of helix 2 of the all- domain. Motif 2, most often SxN or SxD (here SxC), is on a loop connecting two helices of the all- domain. Motif 3, most often KTG or KSG, is on strand ß3 of the /ß domain. The structure was built with Molscript (130) and Raster3D (155). Illustration courtesy of Paulette Charlier, Eveline Fonzé, Michaël Delmarcelle, and André Piette, Center for Protein Engineering.

Top Previous Next References

View larger version (23K): [in this window] [in a new window]   FIG. 8. ß-Lactam antibiotic family. Only 6-epoxypenicillin S is active. *, D-configured carbon atom. ("Mecillinam" is another name for amdinocillin.)

View larger version (27K): [in this window] [in a new window]   FIG. 9. Common backbone of extended N-acyl-D-alanyl-D-alanine peptides and penams (top) and ab initio electrostatic potentials of bisacetyl-L-lysyl-D-alanyl-D-alanine and benzylpenicillin (bottom) optimized at level AM1 (the terminal carboxylates are protonated). The electrostatic negative wells I, II, and II are shown at levels of -40 kcal (solid contours) and -30 kcal (dotted contours). They are coplanar. The negative wells generated by the acetyl substituents of the - and -amino groups of the L-lysine residue of bisacetyl-L-lysyl-D-alanyl-D-alanine are above and below the plane, respectively. The negative well of small amplitude seen between well I and well III of benzylpenicillin is due to the pair of free electrons of the nitrogen atom of the azetidinone ring. Illustration courtesy of Georges Dive, Center for Protein Engineering.

As a result of the coplanarity of the negative wells at positions I, II, and III, the -face of the azetidinone ring of the ß-lactam antibiotics is well exposed, facilitating the attack of the electrophilic center at position I by the serine nucleophile of the SxxK D,D-acyltransferases. The backbone CON-C₂-CON-C₃-COO^- is modified into CON-C₂-CON-SO₃^- in the monobactam aztreonam (Fig. 8). It is modified into C=N-C₂-CON-C₃-COO^- in the penam amdinocillin. The carbon atom C-2 of the carbapenem imipenem has an L-configuration. In spite of these structural variations, the disposition of the sulfamate in aztreonam is roughly comparable to that of the carboxylate at position III in the bicyclic penams and 3-cephems. The environment of the CH=N amidino bond of amdinocillin generates a negative well at position II, albeit of reduced amplitude. Because the rotation angle (Fig. 9, upper part) is less open in the carbapenems than in the penams and 3-cephems, the carboxylate at position III of imipenem is moved upward, ensuring coplanarity with the oxygen atom at position I and the alcohol oxygen atom at position II. Of the two 6-epoxypenicillin isomers (Fig. 8), the side chain of which, at position II, has a frozen conformation due to the epoxy cycle, the S isomer is active, but the R isomer is not.

Depending on the (noncyclic, monocyclic, or bicyclic) framework and conformation of the backbone and the presence of bulky, ionized, and electron-withdrawing side chains, the electrostatic negative wells at positions II and III around the electrophilic center at position I of the carbonyl donors can vary in shape and strength, be displaced along the reference plane, be better expressed in other sections of the molecules, be fused, or be masked. In turn, the SxxK acyltransferases are large bodies studded with positively and negatively charged magnets. Upon binding of a carbonyl donor to the catalytic center, a dense hydrogen bonding network is formed, and a multimembered ring that is both enzyme and ligand specific is created, in which the scissile CO-N amide bond at position I of the carbonyl donor is connected to the catalytic serine OH of motif SxxK of the acyltransferase by one or several water molecules and the side chains of several amino acid residues (53, 54, 80). This multimembered ring is utilized as a motorway, allowing the proton of the serine OH to be transferred via a transition state to the nitrogen atom of the scissile bond, resulting in enzyme acylation. For the reaction to reach completion, the serine ester-linked acyl enzyme must adopt a conformation that allows entry of an amino group or a water molecule and formation of a new multimembered ring that performs enzyme deacylation.

Potential energy hypersurfaces best portray the geometric rearrangements, electronic redistributions, and free energy barriers that occur along the reaction pathways (80). Depending on the freedom of the water molecules, the ease with which the ligands undergo deformation, and the ease with which the enzyme backbone undergoes relaxation, numerous possible routes for proton transfer exist. The energetically most favorable route dictates the specificity of the SxxK acyltransferases and the completeness (protein binding, protein acylation, protein acylation, and deacylation) and productiveness of the catalyzed reactions.

The D,D-acyltransferases-PBPs of group I perform multiple functions related to (43) peptidoglycan assembly (see below). Changes in protein structure, with conservation of the overall fold, could go as far as a change from D,D specificity to L,D specificity (see the chapter on SxxK acyltransferases of group III).

GROUP I SxxK ACYLTRANSFERASES: IMPLICATED IN (4->3) PEPTIDOGLYCAN BIOCHEMISTRY

Top Previous Next References

 
The SxxK D,D-acyltransferases-PBPs of group I occur as independent entities referred to as free-standing PBPs. They also occur as modules of PBP fusions. A D,D-acyltransferase module of class A or class B is fused, in a class-dependent manner, to polypeptides having their own bar codes and three-dimensional structures (78, 84). Traces of similarity between the free-standing PBPs and the acyltransferase modules of class A or B are generally limited to the three active-site defining motifs. Similarity between the acyltransferase modules of class A and class B is marginal (P 1 x 10^-5).

The PBPs are bound to the plasma membrane, with the bulk of the polypeptide chains exposed on the outer face. The PBP fusions are synthesized with an amino-terminal hydrophobic segment that functions as both a signal sequence for secretion and a stop transfer signal that serves as a membrane anchor. Despite a great divergence in the amino acid sequences, the bar codes SxxK, SxN, and KTG of the free-standing PBPs and the acyltransferase modules of the PBP fusions are conserved except that, occasionally, SxD substitutes for SxN and KSG substitutes for KTG.

The core of an SxxK acyltransferase is defined as the sequence starting about 70 amino acid residues upstream from the SxxK motif and terminating about 70 amino acid residues downstream from the KT(S)G motif. Adducts may occur as inserts and/or as carboxy-terminal extensions. The free-standing PBP5 of E. coli (45) and the class B PBP fusion Spn2x of Streptococcus pneumoniae (86, 173) are of known structure. They each have carboxy-terminal extensions. That of E. coli PBP5 is made of ß-structures that form a loose ß-barrel; that of Snp2x is made of two /ß/ß/ß domains.

View larger version (99K): [in this window] [in a new window]   FIG. 10. Hierarchical distribution of the SxxK PBP fusions and Penr protein fusions of classes A and B. The occurrence of class-specific motifs 1 to 9 (class A) and 1 to 7 (class B) along the polypeptide chains is shown. Adapted from reference 84. Scores (vertical axis of the dendrograms) are the standard deviation values above that expected from a run of 100 randomized pairs of sequences with the same amino acid composition as the two sequences under comparison. The protein identifiers (bottom of the dendrograms) are defined in Table 1. The clusters are labeled by two circles, one of which defines a particular subclass (A1 to A5 and B1 to B5) and the other the prototypic protein. Solid arrowheads help identify proteins that are discussed in the text. Solid stars help identify the mycobacterial proteins. The Penr acyltransferase modules are underlined.

E. coli produces two class A PBP fusions, Eco1a and Eco1b (Fig. 11). The Eco1a-encoding gene, at the 75-min region of the chromosome, and the Eco1b-encoding gene, at the 4-min region, are not linked to particular operons. An insert occurs downstream from the membrane anchor in Eco1b. Inserts occur downstream from the intermodule junction and between motifs 8 and 9 in Eco1a. Eco1b occurs in three forms, each of which is functional and encoded by the same gene (219). The form, shown in Fig. 11, has a cytosolic amino-terminal tail which is 70 amino acid residues long. The glycosyltransferase modules of Eco1a and Eco1b are related by a P value of 4 x 10^-33 (identity [I], 35%). The acyltransferase modules of Eco1a and Eco1b are distantly related by a P value of 2 x 10^-15 (identity, 27%) for overlaps 190 amino acid residues long. They belong to distinct subclasses. Eco1a (subclass A1) and Eco1b (subclass B2) are produced in large quantities, amounting together to several thousand molecules per cell. In in vitro assays in the absence of preformed peptidoglycan primer, they each catalyze the conversion of the disaccharide peptide moiety of lipid II into polymeric (43) peptidoglycan.

View larger version (26K): [in this window] [in a new window]   FIG. 11. Class A PBP fusions. Bar code, motifs 1 to 9 and inserts (underlined). For protein identifiers, see Table 1. mTgase, free-standing transglycosylase of E. coli.

Crosslinking between peptide-substituted glycan chains is an acylation-deacylation reaction that is strictly dependent on the serine residue of motif SxxK of the acyltransferase module. It is aided by amino acid residues of the other catalytic center-defining motifs and one or several catalytic water molecules. In in vitro assays with lipid II, Eco1b transfers the carbonyl of the D-alanine residue at position 4 of stem pentapeptides to water (reaction products, tetrapeptide monomers) and to the D-amino group of the meso-diaminopimelic acid residue at position 3 of pentapeptide and tetrapeptide monomers [reaction products: (43)-linked tetrapeptide-pentapeptide and tetrapeptide-tetrapeptide dimers]. Peptide oligomers larger than dimers are not produced in detectable amounts, indicating that, in vitro, the stem pentapeptide of the tetrapeptide-pentapeptide dimers is not used as a carbonyl donor for further crosslinking.

The acyltransferase of Eco1b is inert on exogenous N-acyl-D-alanyl-D-alanine-terminated peptides (222). However, it catalyzes hydrolysis and aminolysis of ester and thiolester analogues, indicating that the D-alanyl-D-alanine-cleaving activity is glycosyltransferase dependent. Conversely, Eco1b, the acyltransferase module of which is inactivated by penicillin, catalyzes glycan chain elongation, indicating that the glycosyltransferase module is acyltransferase independent.

Loss of Eco1a and Eco1b is fatal, but loss of either Eco1a or Eco1b is tolerated, indicating that Eco1a and Eco1b can substitute for each other (119, 220, 252). PBP fusions of subclass A2 are dispensable in some gram-negative bacteria. The coccus-shaped Neisseria meningitidis, the genome sequence of which is known (223), produces a single class A PBP fusion of subclass A1 (199). Consistently, the cluster formed by the acyltransferase modules of subclass A1 (to which Eco1a belongs) is more populated, i.e., comprises more sequences, than the cluster formed by the acyltransferase modules of subclass A2 (to which Eco1b belongs) (Fig. 10). It remains possible that Eco1a and Eco1b cause distinct, subtle traits in peptidoglycan crosslinking in E. coli that are difficult to detect (34).

The acyltransferase and linker modules diverged in concert (Fig. 10). In gram-negative bacteria, an acyltransferase module of subclass B2 or B3 is fused to a linker module of subclass B2 or B3, respectively. In gram-positive bacteria, an acyltransferase module of subclass B4 or B5 is fused to a linker module of subclass B4 or B5, respectively. E. coli produces two class B PBP fusions, Eco2 of subclass B2 and Eco3 of subclass B3 (Fig. 10 and 12). The 588-amino-acid Eco3 is processed into M1-V577 Eco3 (161).

View larger version (40K): [in this window] [in a new window]   FIG. 12. Class B PBP fusions. Bar code, motifs 1 to 7. For protein identifiers, see Table 1.

The cell septation apparatus of E. coli, known as the divisome, is a Lego-like interlocking of a set of at least 10 proteins that assemble into a septal ring at midcell. With Fts standing for temperature-sensitive filamentous phenotype, the PBP fusion Eco3 (or FtsI), FtsW, FtsQ, FtsA, FtsZ are encoded by genes of the 2-min dcw cluster, which also contains several lipid II synthetase-encoding genes (Fig. 5) (162, 237). FtsK, ZipA, FtsN, and YgbQ are encoded by genes occurring at different places on the chromosome. Like Eco3, FtsL, YgbQ, FtsQ, and FtsN are bitopic membrane proteins with a short cytoplasmic amino tail and a relatively large periplasmic domain (31, 36).

FtsL and YgbQ have a leucine-like zipper motif in the periplasmic domain. They belong to a family of proteins that exhibit a great propensity to form coiled-coil structures (31). FtsW is a homologue of RodA (28). By analogy with FtsW of S. pneumoniae (76), the E. coli FtsW probably comprises 10 transmembrane segments, with a large extracytoplasmic loop extending between segments 7 and 8. FtsA and the eukaryotic actin are members of the same ATPase domain protein superfamily (to which Hsp70 also belongs) (237). FtsZ is a GTPase homologue of tubulin, the building block of the eukaryotic cell division microtubules (97, 142, 143).

FtsZ arrives first at mid-cell; it serves as scaffold holding the other proteins of the divisome together, and it provides the driving force for cytokinesis (2). FtsA and ZipA localize to the septum independently but in an FtsZ-dependent manner. The other proteins then assemble in a sequential dependency order as follows: FtsK, FtsQ, FtsL, YgbQ, FtsW, Eco3, and FtsN. It is likely that class A PBP fusions are also components of the divisome (233). The class A PBP fusion Bsu1 (subclass A3) of Bacillus subtilis is localized at the division septum in vegetative cells (175).

The cell shape PBP fusion Eco2 of subclass B2 and the cell septation PBP fusion Eco3 of subclass B3 are produced in small amounts: a few tens and about 200 molecules per cell, respectively. Their acyltransferase modules are distantly related by a P value of 6 x 10^-15 (identity, 29%). Loss of Eco2 results in a block of cell division, transformation of the E. coli cells into spherical bodies, and cell death. Loss of Eco3 causes a block of cell septation, formation of multigenomic filaments, and cell death. Loss of either Eco2 or Eco3 is fatal (211). PBP fusions of subclass B2, however, are dispensable in some gram-negative bacteria. Neisseria meningitidis possesses a single PBP fusion of class B that belongs to subclass B3 (255). This PBP fusion may fulfil functions comparable to those of Eco2 and Eco3 in E. coli (14). Consistently, the clusters formed by the linker and acyltransferase modules of the PBP fusions of subclass B3 are more populated than the clusters formed by the corresponding modules of the PBP fusions of subclass B2 (Fig. 10).

Both Eco2 and Eco3 are implicated in the exponential phase of growth of E. coli cells (19), conditions under which the peptidoglycan precursors are incorporated all over the lateral wall in a diffuse way (49). Newly formed (43) peptidoglycan is mixed with existing (43) peptidoglycan except at the time of cell septation. At this stage, peptidoglycan synthesis is strictly localized to the septum in an Eco3-dependent manner (27). In the course of the hierarchical assembly of the divisome, FtsZ, FtsQ, FtsL, and YgbQ are required for the septal localization of FtsW, and FtsW is essential for the subsequent recruitment of Eco3 (154).

Consistently, the linker module of Eco3 appears to be designed in such a way that the acyltransferase module is positioned, in an active form, within the divisome where it needs to be (148) (Fig. 13 ). The membrane anchor (243) and the segment upstream from motif 1 of the linker module (148) have the information ensuring that Eco3 localizes at the cell septation site. Motifs 1, 2, and 3 and other peptide segments which form the core of the linker module have the information ensuring that Eco3 folds correctly and that the acyltransferase catalytic center adopts the active conformation. The Glu206-Val217 peptide segment at the surface of the linker module has the information ensuring that Eco3 fulfils its cell septation activity within the fully complemented divisome by interacting with other components of the morphogenetic apparatus.

View larger version (26K): [in this window] [in a new window]   FIG. 13. Schematic structure of the membrane-bound SxxK subclass B3 PBP fusion protein Eco3 of E. coli. Spatial disposition along the polypeptide chain of the essential S*307 of the SxxK motif of the acyltransferase module (see Fig. 7) and of motifs 1, 2, and 3 (identified by the residue at the amino side of the sequences) and segment E206 to V217 of the linker module. The acyltransferase module is in yellow, with S*307 of the SxxK motif in red. The linker module is in black with motif 1 (R71 to G79) in green, motif 2 (R167 to G172) in dark blue, motif 3 (G188 to D197) in orange, and segment E206 to V217 in light blue. Motifs 1, 2, and 3 and other peptide segments form the core of the linker module in interaction with a noncatalytic groove of the acyltransferase module. The peptide segment M1 to R71 is of unknown structure. Adapted from reference 148. Illustration courtesy of Robert Brasseur, Faculté Universitaire des Sciences Agronomiques, Gembloux.

E. coli produces, in the cytosol, penicillin-resistant peptidases that hydrolyze the bond between meso-diaminopimelic acid (L-center) and D-alanine at positions 3 and 4 of the stem peptides. An L,D-endopeptidase (85) and an L,D-carboxypeptidase I (156) act on the nucleotides UDP-N-acetylmuramoyl-(D-alanyl-D-alanine-terminated) pentapeptide and UDP-N-acetylmuramoyl-(D-alanine-terminated) tetrapeptide, respectively. The L,D carboxypeptidase LdcA (221, 228) is made without signal peptide. Loss of LdcA causes cell lysis, and the effect is restricted to the onset of the stationary phase. LdcA does not belong to the SxxK peptidase family. The required stem tripeptides can also be produced, in the periplasm, by the action of the L,D-carboxypeptidase II (17). LdcII acts on stem tetrapeptides of nascent peptidoglycan at the time of cell division. It may be present in nondividing cells, but then its activity is masked. LdcII has not been characterized biochemically.

Eco2 is essential to multiplying rod-shaped E. coli cells. It may be also important in the stationary phase (23). In rounded E. coli cells, in which Eco2 is impaired, the incorporation of the peptidoglycan precursors is a zonal process (49). There is no mixing of new peptidoglycan and old peptidoglycan, and the wall polymer cannot undergo remodeling. In vitro assays that would help identify the reaction that the acyltransferase module of Eco2 performs in vivo have not been developed. Eco2 binds benzylpenicillin and ampicillin with high affinity, suggesting that the acyltransferase module is targeted to N-acyl-D-alanyl-D-alanine sequences. However, Eco2 is also very susceptible to the nonclassical penam amdinocillin, the carbapenem thienamycin, and the oxapenem clavulanic acid (213) (Fig. 8).

The linker modules of Eco2 and Eco3 have the same conserved motifs 1, 2, and 3. They probably adopt the same fold. As a corollary, peptide segments other than the conserved motifs must serve as recognition sites specifying the morphogenetic apparatus to which Eco2 and Eco3 attach. Cell division and viability in the absence of Eco2 but in the presence of Eco3 are restored by increasing the pool of ppGpp or the level of FtsQAZ (115, 238). Hence Eco2, but not Eco3, is dispensable in particular genetic backgrounds. It is likely that the morphogenetic apparatus is not composed of fixed structures and proteins can belong to several morphogenetic apparatuses simultaneously or at different times.

View larger version (25K): [in this window] [in a new window]   FIG. 14. Occurrence of catalytic center-defining motifs along the sequences of free-standing SxxK polypeptides of E. coli, M. tuberculosis, and M. leprae. Polypeptides of group 1 bear the bar codes SxxK, SxN, and KTG. They are auxiliary cell cycle PBPs in E. coli. Inserts are underlined. The M. tuberculosis polypeptides of groups 2 and 3 have no equivalent in E. coli. Their bar codes are modified or incomplete. It is likely that they are not implicated in wall peptidoglycan metabolism. The M. tuberculosis ß-lactamase of class A (group 4) has a class-specific Ex2LN motif. E. coli can produce two ß-lactamases, one of which (plasmid coded) is of class A (Fig. 15).