Approaching the physiological functions of penicillin-bindingproteins in Escherichia coli

Biochimie 83 (2001) 99−102 © 2001 Société française de biochimie et biologie moléculaire / Éditions scientifiques et médicales Elsevier SAS. All rights reserved. S0300908400012050/FLA

Approaching the physiological functions of penicillin-binding proteins in Escherichia coli Kevin D. Young* Department of Microbiology and Immunology, School of Medicine and Health Sciences, University of North Dakota, Grand Forks, 58202-9037 North Dakota, USA (Received 10 October 2000; accepted 16 November 2000) Abstract — A rigid shell of peptidoglycan encases and shapes bacteria and is constructed and maintained by a diverse set of enzymes, among which are the penicillin-binding proteins (PBPs). Although a great deal has been learned about how these proteins synthesize and modify peptidoglycan, the physiological functions of the multitude of bacterial PBPs remain enigmatic. We approached this problem by combining PBP mutations in a comprehensive manner and screening for effects on biochemical processes involving the passage of proteins or nucleic acids across the cell wall. The results indicate that the PBPs or their peptidoglycan product do have significant biological functions, including roles in determination of cell shape, in phage resistance, in induction of capsule synthesis, and in regulation of autolysis. © 2001 Société française de biochimie et biologie moléculaire / Éditions scientifiques et médicales Elsevier SAS peptidoglycan / penicillin-binding proteins / phage resistance / bacterial shape / capsule

1. Introduction The bacterial cell is contained, defined and protected by a peptidoglycan exoskeleton, a rigid wall so vital for bacterial survival that it is the target of our oldest and most important class of antibiotics, the β-lactams. These compounds inactivate the enzymes that synthesize and remodel peptidoglycan, the penicillin-binding proteins (PBPs). The PBPs were identified over 25 years ago by covalently labeling them with radioactive penicillin so they could be visualized by polyacrylamide gel electrophoresis [20, 21]. This technological achievement quickly led to the assignment of general physiological functions for some of the PBPs in Escherichia coli: PBPs 1a and 1b elongated the sacculus, PBP 2 maintained the cell’s rod shape, PBP 3 was required for septum formation during cell division, and three small PBPs were deemed to be non-essential [18–20]. Unfortunately, although our biochemical knowledge about these enzymes has increased dramatically, our understanding of the biological roles of the PBPs has not advanced much further during the passage of 25 years. The most that can be said is that recent work implies that peptidoglycan and the PBPs may play important roles in several physiological events, including protein secretion and DNA transfer [4]. One of the major problems in identifying or assessing the importance of the PBPs to * Correspondence and reprints. E-mail address: [email protected] (K.D. Young).

these processes is that many cells contain multiple PBPs with similar enzymatic activities. Thus, mutation of one PBP may have no effect on a bacterium because another PBP can perform the required biochemical steps. This idea, which we refer to as the ‘theory of equivalent substitution’, has often been raised to explain the many failures to find functions for the PBPs. In this article I will briefly summarize how we have begun to address the question of the biological functions of the PBPs in E. coli, and will introduce results that establish that peptidoglycan and the PBPs do play significant roles in basic bacterial physiology. 2. Genetic strategy and results The idea of equivalent substitution predicts that it may be impossible to select for PBP mutants that exhibit a phenotype because the effects of losing one enzyme would be masked by the continued activity of another. If true, then the only way to test the idea is to purposely and simultaneously mutate similar PBP genes so that such duplication cannot occur. An additional prediction would be that if a specific enzymatic activity were essential for bacterial survival then it should be impossible to create certain combinations of PBP mutations. These non-viable combinations would define different ‘substitution families’ of PBPs that could perform one another’s functions. A total of 12 PBPs have been identified in E. coli: five high molecular mass (HMW) PBPs (1a, 1b, 1c, 2 and 3) and seven low molecular mass (LMW) PBPs (4, 5, 6, 7,

100 DacD, AmpC and AmpH). PBPs 1a and 1b have both transglycosylase and transpeptidase activities, PBPs 2 and 3 are believed to have only transpeptidase activity, and PBP 1c is not completely characterized [5, 6, 14, 17]. The LMW PBPs are divided into those that are endopeptidases (PBPs 4 and 7), those that are carboxypeptidases (PBPs 4, 5, 6, and DacD), and those that belong to the class C β-lactamase family (AmpC and AmpH) [1, 2, 8–10, 12, 16]. In particular, the LMW groupings constitute what might be substitution families. We attempted to determine the existence of PBP substitution families by constructing 192 mutants of E. coli from which eight PBPs were deleted in every possible combination (PBPs 1a, 1b, 4, 5, 6, 7, and AmpC and AmpH) [3]. We began by deleting an internal portion of each cloned PBP gene and inserting a kanamycin resistance cassette bounded on either end by the res sequence of plasmid RP4. Each marked gene was recombined onto the chromosome to replace the wild type gene, after which the kanamycin resistance cassette was removed by the introduction of RP4 resolvase [13]. This procedure yielded antibiotic sensitive PBP mutants that were used in subsequent steps as the parents from which additional PBP genes were deleted until all combinations of mutations were constructed. As expected, mutants lacking both PBPs 1a and 1b were non-viable, confirming that one or the other of these two enzymes must be active for bacterial growth [11, 22]. Thus, these PBPs comprise a substitution family of two members in E. coli. However, quite unexpectedly, all other mutant combinations lived and grew with near normal kinetics. This was surprising because the mutants we constructed eliminated the LMW PBPs in all combinations, including strains from which all seven LMW PBPs were absent (PBPs 4, 5, 6, 7, DacD, AmpC and AmpH) [3]. Other combinations included mutants lacking all the carboxypeptidases (PBPs 4, 5, 6 and DacD), all the class C β-lactamases (AmpC and AmpH), and two of the three known endopeptidases (PBPs 4 and 7). Since then, the Höltje laboratory has constructed a viable mutant lacking all three endopeptidases (PBPs 4 and 7, and MepA) (J.-V. Höltje, personal communication). These results force us to draw one of two conclusions. Either unknown proteins are duplicating the activities of the missing PBPs or else none of these enzymatic activities are essential for primary peptidoglycan synthesis or bacterial survival. The first possibility is just an extension of the equivalent substitution argument in which we assume that the enzymatic activities (carboxypeptidase, endopeptidase, β-lactamase) must be essential, but for which we invoke ignorance about the enzymatic participants. However, because mutants lacking all seven LMW PBPs grow well and because no reasonable candidates encoding enzymatic homologues have been unearthed in the E. coli genome, we are now inclined to accept the second possibility that these proteins and their activities

Young are truly non-essential as far as (laboratory) bacterial viability is concerned. 3. Physiological strategy and results It seems unreasonable to think that E. coli retains these PBPs for no purpose. Therefore, we have begun to screen the family of PBP mutants to see if any are impaired in biochemical processes which must cross or interact with peptidoglycan. Such processes fall into three broad categories: those which enter the cell, those that exit the cell, and those housekeeping functions that maintain the state and integrity of the wall itself. For example, DNA enters bacteria by way of viral infection, conjugation or transformation. Exit processes would include protein secretion and the assembly or transport of large structures (flagella, pili, secretins, etc.) that are embedded in or anchored to the peptidoglycan. Internal domestic functions are exemplified by the proper assembly of peptidoglycan, its routine recycling, or its explosive disassembly during autolysis. By systematically screening the comprehensive set of PBP mutants, we found that the PBPs or their peptidoglycan product do contribute to several different physiological activities. 3.1. Effects on entry events The major discovery in this class of physiological processes is that subsets of PBP mutants became resistant to infection by different bacteriophages (K.D. Young, S.A. Denome, S. Maier, in preparation). For example, 31 of 64 mutants lacking PBP 1a were resistant to phage λ (defined as supporting the production of fewer than 10–3 plaques when compared to the wild type parent). Of this group, decreasing numbers of mutants were also resistant to phages T7, T6 and T4. Although all resistant mutants were missing PBP 1a, not all mutants lacking 1a were resistant. Thus, the loss of PBP 1a is essential for the phenotype but not sufficient – other factors or the deletion of additional PBPs are necessary. Beyond the reliance on PBP 1a there is no clear pattern of mutations in resistant mutants, so that it is not possible to describe the exact PBP composition that contributes to phage resistance. It may be that a complicated form of the theory of equivalent substitution is at work in which phage resistance depends on the opposing activities of several different proteins. One unmistakable conclusion is that the biochemical functions of PBPs 1a and 1b differ with respect to this phenotype. A similar observation was made for resistance to phage φK, a relative of φX174 that infects E. coli K12. In this case, several mutants were resistant to φK infection (K.D. Young, S.A. Denome, S. Maier, in preparation). Many of the resistant mutants were missing PBP 5, not all mutants lacking PBP 5 were resistant, again suggesting that there is a complex reliance on the presence or absence of other proteins.

Penicillin-binding proteins in E. coli Because DNA can enter cells via conjugation or transformation, we also screened the PBP mutants for effects on these processes. Every mutant could act as a recipient and as a donor of the F plasmid, indicating that conjugation was unimpaired for the transfer of this particular plasmid (K.D. Young, unpublished). Although some of the mutants appeared to be more difficult to transform than others, we were unable to measure the effect reproducibly (F. Sailer and K.D. Young, unpublished), leaving open the question of whether this property might be affected by the PBPs in E. coli as has been suggested for other organisms [4]. 3.2. Effects on exit events In this category, the clearest result was that every mutant from which PBP 1a was deleted became mucoid (F. Sailer, S.A. Denome, B.M. Meberg and K.D. Young, in preparation). This result was completely unexpected because PBP 1a mutants have been constructed and in use for many years but never reported to have this characteristic. Nonetheless, every ∆1a strain in the original mutant set exuded a thick capsule when incubated on a solid surface. The capsular material was identified as colanic acid because mutations in this pathway prevented capsule production by ∆1a mutants. In fact, on continued subculture, several ∆1a mutants lost their mucoid phenotype because of secondary mutations in rcsB, one of the regulatory genes in the colanic acid pathway. This may explain why the connection was not identified earlier – if previous 1a mutants were mucoid, they probably lost the capsule via secondary mutations so that the phenotype escaped notice. Once again, there is a clear distinction between the functions of PBPs 1a and 1b in that no ∆1b mutant produced this capsular material. Why the loss of PBP 1a should induce synthesis of colanic acid is unknown. In a separate (preliminary) screen, we observed that several PBP mutants were no longer motile (F. Sailer, unpublished), suggesting that the assembly or function of the flagella might be impaired. Although consistent with what we might expect of an altered peptidoglycan, this result is not yet confirmed. 3.3. Effects on housekeeping functions Among domestic biochemical processes, we observed alterations in at least three different areas: control of bacterial shape, response to antibiotics, and susceptibility to autolysis. While viewing the microscopic appearance of the mutants, we discovered that the presence of active PBP 5 was the most important determinant of the diameter, contour and topology of E. coli [15]. In the absence of PBP 5, individual cells increased in diameter and took on aberrant shapes. A PBP 5 mutation eliminated the unifor-

101 mity of width along a single cell (its contour) and produced cells of highly irregular form (topology), including cells with knobs, protrusions, and branching filaments. These mutants afforded us the opportunity to test the theory of equivalent substitution directly. On the one hand, the presence and extent of the morphological aberrations were greater in cells lacking multiple PBPs. In fact, the cells changed very little if only PBP 5 was absent. Thus, it would appear that other PBPs might substitute for the function of PBP 5. On the other hand, in those mutants where the phenotypes existed, the effects were reversed completely when complemented by expression of cloned PBP 5 but not by any other carboxypeptidase (D.E. Nelson and K.D. Young, submitted for publication). This, then, is a specific example where the theory of equivalent substitution may be at work in its mild form – although PBP 5 is the major determinant of sacculus shape and uniformity, the combined activity of other PBPs may perform similar, if not identical, duties. Another interesting observation was that some of the PBP mutants became refractory to lysis by a mixture of aztreonam and mecillinam (A. Paulson and K.D. Young, in preparation). These two antibiotics bind specifically to PBPs 3 and 2, respectively, and the simultaneous inactivation of these PBPs is associated with rapid lysis [3, 7]. In addition, the activity of PBPs 2 or 3 has been regarded as essential to peptidoglycan synthesis [10]. In contrast to these previous results and expectations, several of the PBP mutants resisted lysis when exposed to these two β-lactams. The refractory strains stopped dividing but individual cells continued to enlarge until they formed huge spherical cells (up to 5–10 µm in diameter!) [3]. Because the walls of these cells continued to enlarge in classical LB broth without osmotic stabilizers, the conclusion appears to be that peptidoglycan synthesis can continue without PBPs 2 and 3 and that lysis is not an inevitable conclusion of their loss. Because one of the mutants continued to grow while having only PBPs 1b and 1c active, it may be that peptidoglycan synthesis is somewhat simpler than usually envisioned. The set of resistant strains were all ∆1a mutants, though that one alteration did not of itself predict which mutants would exhibit the phenotype. Finally, several of the PBP mutants were temperature sensitive. These strains grew normally at 30 °C but lysed when transferred to 42 °C (J.L. Haug and K.D. Young, in preparation). Once again, this group was a subset of ∆1a mutants. Interestingly, the complete set of 64 ∆1a mutants could be divided into three non-overlapping phenotypic subsets: those that were temperature sensitive, those that were refractory to lysis by aztreonam plus mecillinam, and those that retained the normal parental responses (temperature resistant and sensitive to lysis by these antibiotics). Although the data are preliminary, the results suggest that various PBP combinations determine the different fates of their final peptidoglycan product.

102 4. Conclusion In general, peptidoglycan has been considered to be transparent to molecular processes that cross the bacterial cell wall. It is becoming clearer that this transparency is a function of active processes, several of which appear to be mediated by the penicillin-binding proteins. Because of potential overlap in the enzymatic activities of the PBPs and because many of these proteins are non-essential for bacterial viability, identifying their biological functions has required the construction of multiple mutants followed by mass screenings. It is likely that continued application of this approach will identify other interesting physiological roles not only for the PBPs but for other families of proteins, as well. Acknowledgments I thank past and present members of my laboratory for their contributions to the construction of the PBP mutant set and for examining the variety of physiological consequences exhibited by these strains. In particular, I would like to acknowledge the work of the following in making these discoveries possible: Sylvia Denome, Tom Henderson, Pam Elf, Shelly Maier, Jessica Haug, David Nelson, Fran Sailer, Bernadette Meberg and Avery Paulson.

References [1] Amanuma H., Strominger J.L., Purification and properties of penicillin-binding proteins 5 and 6 from the dacA mutant strain of Escherichia coli (JE11191), J. Biol. Chem. 259 (1984) 1294–1298. [2] Baquero M.-R., Bouzon M., Quintela J.C., Ayala J.A., Moreno F., dacD, an Escherichia coli gene encoding a novel penicillinbinding protein (PBP6b) with DD-carboxypeptidase activity, J. Bacteriol. 178 (1996) 7106–7111. [3] Denome S.A., Elf P.K., Henderson T.A., Nelson D.E., Young K.D., Escherichia coli mutants lacking all possible combinations of eight penicillin binding proteins: viability, characteristics, and implications for peptidoglycan synthesis, J. Bacteriol. 181 (1999) 3981–3993. [4] Dijkstra A.J., Keck W., Peptidoglycan as a barrier to transenvelope transport, J. Bacteriol. 178 (1996) 5555–5562. [5] Ghuysen J.-M., Dive G., Biochemistry of the penicilloyl serine transferases, in: Ghuysen J.-M., Hakenbeck R. (Eds.), Bacterial cell wall, Elsevier Science B.V, Amsterdam, 1994, pp. 103–129. [6] Goffin C., Ghuysen J.-M., Multimodular penicillin-binding proteins: an enigmatic family of orthologs and paralogs, Microbiol. Mol. Biol. Rev. 62 (1998) 1079–1093. [7] Gutmann L., Vincent S., Billot-Klein D., Acar J.F., Mrèna E., Williamson R., Involvement of penicillin-binding protein 2 with other penicillin-binding proteins in lysis of Escherichia coli by some β-lactam antibiotics alone and in synergistic lytic effect of amdinocillin (mecillinam), Antimicrob. Agents Chemother. 30 (1986) 906–912.

Young [8] Henderson T.A., Templin M., Young K.D., Identification and cloning of the gene encoding penicillin-binding protein 7 of Escherichia coli, J. Bacteriol. 177 (1995) 2074–2079. [9] Henderson T.A., Young K.D., Denome S.A., Elf P.K., AmpC and AmpH, proteins related to the class C β-lactamases, bind penicillin and contribute to the normal morphology of Escherichia coli, J. Bacteriol. 179 (1997) 6112–6121. [10] Höltje J.-V., Growth of the stress-bearing and shape-maintaining murein sacculus of Escherichia coli, Microbiol. Mol. Biol. Rev. 62 (1998) 181–203. [11] Kato J.-I., Suzuki H., Hirota Y., Dispensability of either penicillinbinding protein-1a or -1b involved in the essential process for cell elongation in Escherichia coli, Mol. Gen. Genet. 200 (1985) 272–277. [12] Korat B., Mottl H., Keck W., Penicillin-binding protein 4 of Escherichia coli: molecular cloning of the dacB gene, controlled overexpression, and alterations in murein composition, Mol. Microbiol. 5 (1991) 675–684. [13] Kristensen C.S., Eberl L., Sanchez-Romero J.M., Givskov M., Molin S., de Lorenzo V., Site-specific deletions of chromosomally located DNA segments with the multimer resolution system of broad-host-range plasmid RP4, J. Bacteriol. 177 (1995) 52–58. [14] Matsuhashi M., Utilization of lipid-precursors and the formation of peptidoglycan in the process of cell growth and division: membrane enzymes involved in the final steps of peptidoglycan synthesis and the mechanism of their regulation, in: Ghuysen J.M., Hakenbeck R. (Eds.), Bacterial cell wall, Elsevier Science B.V, Amsterdam, 1994, pp. 55–71. [15] Nelson D.E., Young K.D., Penicillin binding protein 5 affects cell diameter, contour, and morphology of Escherichia coli, J. Bacteriol. 182 (2000) 1714–1721. [16] Romeis T., Höltje J.-V., Penicillin-binding protein 7/8 of Escherichia coli is a DD-endopeptidase, Eur. J. Biochem. 224 (1994) 597–604. [17] Schiffer G., Höltje J.-V., Cloning and characterization of PBP 1C, a third member of the multimodular class A penicillin-binding proteins of Escherichia coli, J. Biol. Chem. 274 (1999) 32031–32039. [18] Spratt B.G., Distinct penicillin binding proteins involved in the division, elongation, and shape of Escherichia coli K12, Proc. Natl. Acad. Sci. USA 72 (1975) 2999–3003. [19] Spratt B.G., Properties of the penicillin-binding proteins of Escherichia coli K12, Eur. J. Biochem. 72 (1977) 341–352. [20] Spratt B.G., Pardee A.B., Penicillin-binding proteins and cell shape in E. coli, Nature 254 (1975) 516–517. [21] Suginaka H., Blumberg P.M., Strominger J.L., Multiple penicillinbinding components in Bacillus subtilis, Bacillus cereus, Staphylococcus aureus, and Escherichia coli, J. Biol. Chem. 247 (1972) 5279–5288. [22] Yousif S.Y., Broome-Smith J.K., Spratt B.G., Lysis of Escherichia coli by β-lactam antibiotics: deletion analysis of the role of penicillin-binding proteins 1A and 1B, J. Gen. Microbiol. 131 (1985) 2839–2845.