Three redundant murein endopeptidases catalyse an essential cleavage step in peptidoglycan synthesis of Escherichia coli K12


For correspondence. E-mail; Tel. (+91) 40 27192523; Fax (+91) 40 27160591.


Bacterial peptidoglycan (PG or murein) is a single, large, covalently cross-linked macromolecule and forms a mesh-like sacculus that completely encases the cytoplasmic membrane. Hence, growth of a bacterial cell is intimately coupled to expansion of murein sacculus and requires cleavage of pre-existing cross-links for incorporation of new murein material. Although, conceptualized nearly five decades ago, the mechanism of such essential murein cleavage activity has not been studied so far. Here, we identify three new murein hydrolytic enzymes in Escherichia coli, two (Spr and YdhO) belonging to the NlpC/P60 peptidase superfamily and the third (YebA) to the lysostaphin family of proteins that cleave peptide cross-bridges between glycan chains. We show that these hydrolases are redundantly essential for bacterial growth and viability as a conditional mutant lacking all the three enzymes is unable to incorporate new murein and undergoes rapid lysis upon shift to restrictive conditions. Our results indicate the step of cross-link cleavage as essential for enlargement of the murein sacculus, rendering it a novel target for development of antibacterial therapeutic agents.


Peptidoglycan (PG or murein) is a unique and essential constituent of the eubacterial cell wall. It is a large, covalently linked, mesh-like macromolecule that completely surrounds the cytoplasmic membrane and protects the cell from intracellular turgor pressure in addition to contributing to cell shape (Weidel and Pelzer, 1964; Park, 1996; Holtje, 1998; Vollmer et al., 2008a). Structurally, PG sacculus consists of multiple linear glycan strands made up of alternating, β-1,4-linked N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM) disaccharide units. The lactoyl moiety of NAM is covalently attached to a peptide chain of two to five amino acids with the pentapeptide consisting of l-alanine (ala)−d-glutamic acid (glu)−meso-diaminopimelic acid (mDAP)−d-ala−d-ala. Normally, the ε-amino group of mDAP of a stem peptide is bonded to d-ala of another stem peptide of a neighbouring glycan strand resulting in a cross-linked, mono- or multi-layered structure (Fig. 1). In Escherichia coli, approximately one-third of the peptide chains are cross-linked to each other and of these, 93% are between mDAP and d-ala (D-D conformation), with a minor fraction being that between mDAP and mDAP (L-D) (Glauner et al., 1988). The degree of cross-linkage depends on growth conditions and contributes to the flexibility of PG sacculus. Generally, the length of glycan chains varies from 8 to 100 disaccharide units with one end terminating in a 1,6-anhydromuramic acid residue (Park, 1996; Holtje, 1998; Vollmer et al., 2008a).

Figure 1.

Schematic diagram depicting peptidoglycan sacculus and its structure. Murein or PG sacculus is situated between outer membrane (dark blue) and inner membrane (light blue) of Gram-negative bacteria such as E. coli and is made up of short overlapping glycan chains (dark pink) that are arranged perpendicular to the long axis of the cell. Glycan chains consist of disaccharide [with alternating NAG (blue) and NAM (pink) residues] muropeptides connected to each other by cross-links (black arrows). Cross-links between mDAP of one muropeptide to d-ala of another muropeptide of an adjacent glycan chain results in formation of a mesh-like sacculus that totally surrounds the bacterial cytoplasmic membrane.

As the murein sacculus completely encircles the cytoplasmic membrane, the growth of a cell is required to be intimately coupled to growth and expansion of PG. To enlarge such a covalently closed mesh-like molecule, the bonds connecting the glycan chains would need to be cut for incorporation of new glycan strands. It was proposed nearly five decades ago that growth of PG is initiated by local breakdown of existing glycan chains followed by insertion of new material without disrupting the continuity of the sacculus (Weidel and Pelzer, 1964). Subsequently, several in vivo labelling experiments confirmed that PG synthesis occurs diffusely at multiple sites with incorporation of new glycan strands adjacent to the pre-existing glycan chains (Burman et al., 1983; Goodell and Schwarz, 1983; Burman and Park, 1984; Cooper et al., 1988; de Jonge et al., 1989; de Pedro et al., 1997). The PG synthesis is known to be mediated by the penicillin-binding proteins, Pbp1A, Pbp1B, Pbp1C (the bifunctional transglycosylases/transpeptidases), Pbp2 and Pbp3 (the monofunctional transpeptidases) (Spratt, 1975; Sauvage et al., 2008). However, the murein hydrolytic enzymes that cleave the covalent cross-links for insertion of newly synthesized glycan strands have not been identified, although their ‘space-maker’ function has been predicted to be essential for murein enlargement and hence for cell growth (Burman and Park, 1984; Tomasz, 1984; Koch, 1990; Holtje, 1998).

Escherichia coli has several murein hydrolases that cleave total murein sacculus or its soluble subunits with a wide variety of specificities. They perform multiple functions such as murein maturation, turnover, recycling, autolysis, and cleavage of septum during cell division (Holtje, 1998; Vollmer et al., 2008b; van Heijenoort, 2011). However, mutants of E. coli lacking most of these hydrolases singly or in various combinations are viable and show no significant alterations in their growth properties (Denome et al., 1999; Heidrich et al., 2002).

In this study, using combined genetic and biochemical approaches, we discovered three novel murein hydrolases that specifically cleave the d-ala−mDAP cross-links in murein. An E. coli mutant that is deficient in any one of the hydrolases is viable, but a mutant that is deleted for all the three is inviable and undergoes rapid lysis, indicating that these three are redundantly essential for normal physiological growth. 3H-mDAP labelling experiments showed the requirement of these endopeptidases for incorporation of new murein. In summary, our results provide compelling evidence for involvement of these three redundant endopeptidases in murein synthesis and identify them to be the long-postulated space-maker hydrolases required for enlargement of the murein sacculus.


Spr is a murein DD-endopeptidase that cleaves d-ala−mDAP cross-links

Escherichia coli encodes a class of paralogous proteins that belong to the NlpC/P60 peptidase superfamily (Pfam entry: PF00877) which include Spr, NlpC, YdhO and YafL (Anantharaman and Aravind, 2003). Although members of this class of proteins are in general predicted to be cell-wall hydrolysing enzymes in bacteria, their functions in E. coli have not been studied so far.

To decipher the function of these proteins, initially, we investigated the role of Spr, a predicted outer membrane lipoprotein in murein metabolism. An overexpressed and purified C-terminal histidine-tagged Spr derivative showed murein hydrolase activity on intact peptidoglycan sacculi of E. coli in a zymogram assay (data not shown). To examine the precise enzymatic specificity, we used murein sacculi of E. coli predigested with a muramidase (mutanolysin) as a substrate for Spr. The resulting muropeptides were fractionated by reverse phase-HPLC (high-performance liquid chromatography) and analysed using mass spectrometric methods as described in Experimental procedures. Incubation of the total soluble muropeptides with Spr resulted in complete conversion of a cross-linked dimer of disaccharide-tetrapeptide (Tetra-Tetra; D44) into a monomeric disaccharide-tetrapeptide (Tetra; M4) (Fig. 2A and B; inset to Fig. 2). Other cross-linked molecules such as trimers of disaccharide-tetrapeptide (Tetra-Tetra-Tetra; T444) or anhydro form of Tetra-Tetra (Tetra-Tetra−anhydro; D44N) were also hydrolysed as the peaks corresponding to these muropeptides disappeared after incubation with Spr (Fig. 2A and B). Cleavage of Tetra-Tetra into Tetra was also verified using purified Tetra-Tetra as the substrate, confirming that Spr is a DD-endopeptidase with specificity for d-ala−mDAP cross-links (Fig. 2C and D). Spr appeared to also have a weak LD-carboxypeptidase activity [cleavage of (l)-mDAP−d-ala peptide bond] as a small amount of disaccharide tripeptide (Tri; M3) was always observed in the above reactions (Fig. 2D and F; inset to Fig. 2). However, Spr exhibited no activity on purified Tri (M3) or Penta (M5) muropeptides (data not shown). In addition, Spr did not show any activity against intact peptidoglycan sacculi under similar reaction conditions (Fig. S1).

Figure 2.

Determination of enzymatic specificity of Spr by HPLC analysis. Total soluble muropeptides (obtained by mutanolysin digestion of E. coli PG sacculi) were incubated with buffer (A) or 5 μM Spr (B) and resulting reaction products were analysed by reverse phase-HPLC. Muropeptide peaks were identified using MALDI-TOF and MS/MS. Peak 1 – disaccharide tripeptide (Tri; M3); 2 – disaccharide tetrapeptide (Tetra; M4); 3 – cross-linked disaccharide tetrapeptide (Tetra-Tetra; D44); 4 – trimer of disaccharide tetrapeptide (Tetra-Tetra-Tetra; T444); 5 – anhydro form of Tetra-Tetra (Tetra-Tetra−anhydro). Muropeptide fractions of Tetra-Tetra or Tetra were collected, dried, reconstituted and used as substrates for incubation with 5 μM Spr (D and F) or buffer (C and E). Spr cleaved Tetra-Tetra into Tetra (B and D); Tetra into Tri (B and F); and Tetra-Tetra-Tetra and Tetra-Tetra−anhydro into Tetra monomers (B). The reaction scheme shown in the box is a representation of DD-endopeptidase (Δ) and LD-carboxypeptidase activities (−) of Spr.

Multiple copies of ydhO or yebA substitute for the absence of Spr

Escherichia coli is known to possess three other DD-endopeptidases with specificity for d-ala−mDAP peptide cross-links, namely, Pbp4, Pbp7 and MepA encoded by dacB, pbpG and mepA respectively (van Heijenoort, 2011). Pbp4 has a weak DD-carboxypeptidase activity in addition to its DD-endopeptidase activity (Korat et al., 1991). On the other hand, Pbp7 is a DD-endopeptidase that is active only on intact murein (Romeis and Holtje, 1994a) and in multiple copies is known to suppress the phenotypic defects of a spr deletion mutant (Hara et al., 1996). Unlike these two Pbps, MepA is a penicillin-insensitive DD-endopeptidase (Keck and Schwarz, 1979). However, the absence of these three endopeptidases individually or in combination does not confer any discernible phenotype (Heidrich et al., 2002).

Based on the assumption that an E. coli mutant totally lacking DD-endopeptidase activity would be inviable and our observation that a mutant lacking the four endopeptidases (Δspr ΔdacB ΔpbpG ΔmepA) was not significantly defective in growth than a single spr deletion mutant (Table S1), we predicted that E. coli may possess additional hydrolases with similar specificity as that of Spr. To search for such hypothetical DD-endopeptidases, a multicopy suppressor screen was performed as described in Experimental procedures. An spr deletion mutant is unable to grow on nutrient agar (NA) at high temperature (Hara et al., 1996) and hence clones from a multicopy plasmid library of E. coli that conferred significant growth advantage to a Δspr derivative were obtained. The analysis of the clones indicated that multiple copies of ydhO or yebA suppress the growth phenotypes of Δspr on NA at high temperature (Fig. 3).

Figure 3.

Multiple copies of yebA and ydhO suppress Spr phenotype. Cultures of MG1655Δspr deletion mutant carrying the vector alone (pTRC99a) or vector encoding either spr (Ptrc::spr), yebA (Ptrc::yebA) or ydhO (Ptrc::ydhO) were grown in LB. Cultures were serially diluted and 5 μl from the indicated dilutions were spotted onto NA plates containing various amounts of IPTG and grown at 42°C. spr and yebA suppressed the Spr phenotype at an inducer concentration of 50 μM whereas basal expression of ydhO (no IPTG) was adequate for the growth of Δspr deletion mutant.

YdhO and YebA are DD-endopeptidases

As mentioned above, YdhO is paralogous to Spr within the NlpC/P60 family of peptidases (Anantharaman and Aravind, 2003). On the other hand, YebA is a paralogue of EnvC, NlpD and YgeR, which belong to the lysostaphin (M23) family of peptidases (Uehara et al., 2009). As the biochemical functions of these two putative peptidases are not known, we examined their enzymatic activity after overexpression and purification of the C-terminal histidine-tagged derivatives.

Both proteins showed murein hydrolase activity on intact peptidoglycan sacculi in a zymogram assay (data not shown). Both of them, like Spr, also cleaved various soluble cross-linked muropeptides such as Tetra-Tetra, Tetra-Tetra-Tetra and Tetra-Tetra−anhydro into Tetra monomers, demonstrating that they are DD-endopeptidases with specificity for d-ala−mDAP cross-links (Fig. 4). However, these two proteins did not exhibit any LD-carboxypeptidase activity. Moreover, both of them, unlike Spr, exhibited partial cleavage activity on d-ala−mDAP cross-links in intact PG sacculi (Fig. S1). In addition, the activity of YebA was dependent on the presence of a divalent metal ion with Zn2+ being the most optimal (Fig. S2), confirming that it is a metallo-endopeptidase.

Figure 4.

YebA and YdhO are DD-endopeptidases. Total soluble muropeptides obtained by treatment of PG sacculi of MG1655 with mutanolysin were incubated with buffer (A), 3.5 μM YdhO (B) or 2 μM YebA (C) and processed as described in the legend to Fig. 2. Purified Tetra-Tetra was also used as the substrate in the similar way (D, E and F). Both YebA and YdhO cleaved Tetra-Tetra, Tetra-Tetra-Tetra and Tetra-Tetra−anhydro cross-links.

spr, yebA and ydhO are redundantly essential for growth of E. coli

In order to examine the physiological role of these newly-identified DD-endopeptidases, deletions of yebA or ydhO were introduced into a spr deletion mutant by phage P1-mediated transduction (Miller, 1992). Deletion of ydhO, but not of yebA, could be introduced into the spr mutant. A spr yebA double deletion mutant could be constructed only when the strain harboured a plasmid encoding either spr or yebA, indicating that these two mutations are synthetic lethal with each other.

This phenotype was further examined by constructing a pair of chromosomal Δspr ΔyebA derivatives in which spr or yebA was conditionally provided from plasmids encoding Para::spr or Ptrc::yebA such that their expression was induced by addition of arabinose or IPTG. Figure 5A shows that depletion of Spr in a Δspr ΔyebA/Para::spr strain results in complete cessation of growth on Luria–Bertani (LB)-based media. Similar result was also obtained when Δspr ΔyebA/Ptrc::yebA strain was depleted of YebA by growth without IPTG (as seen in Fig. 7B).

Figure 5.

spr, yebA and ydhO are essential for growth of E. coli.

A. Growth of mutants lacking spr, yebA and ydhO. Cultures of MR501 (BW27783 Δspr/Para::spr); MR504 (BW27783 Δspr ΔydhO/Para::spr); MR506 (BW27783 Δspr ΔyebA/Para::spr) or MR508 (BW27783 Δspr ΔydhO ΔyebA/Para::spr) were grown overnight in LB-arabinose (0.2%). Serial dilutions were prepared and 5 μl from the following dilutions (10−2, 10−4, 10−5, 10−6) were spotted onto indicated plates and grown overnight at 37°C. MM-plates were supplemented with 0.2% maltose and additionally either with 0.2% arabinose or glucose as indicated.

B. Growth of mutants lacking spr, yebA and ydhO. Cells of MR506 (BW27783 Δspr ΔyebA/Para::spr) or MR508 (BW27783 Δspr ΔydhO ΔyebA/Para::spr) were grown overnight in LB-arabinose (0.2%). Next morning, cells were pelleted, washed twice with LB to remove traces of arabinose and diluted 1:100 into either LB or MM (supplemented with 0.5% casamino acids as carbon source) containing 0.2% arabinose (MR506-open blue squares; MR508-solid blue squares) or glucose (MR506-open red triangles; MR508-solid red triangles) and grown at 37°C. Growth was monitored by measuring absorbance at 600 nm at regular intervals. Colony-forming units were also measured at each hour interval and the growth curves depict cfu ml−1.

C. Cells lacking Spr, YebA and YdhO undergo lysis. Cells of MR508 (BW27783 Δspr ΔydhO ΔyebA/Para::spr) were grown overnight in LB-arabinose. Cells were pelleted, washed with LB to remove traces of arabinose and diluted 1:100 into fresh LB broth supplemented with 0.2% of arabinose or glucose and grown at 37°C. At each hour, 1 ml aliquots were withdrawn and cells were stained with a LIVE/DEAD BacLight bacterial viability kit (Molecular Probes) according to manufacturer's instructions. Cells were observed by DIC (differential interference contrast) microscopy (panel I) and fluorescence microscopy using GFP and DsRed filters (panel II) on a Zeiss apotome microscope. In this assay, live cells appear green and dead cells stain red. Scale bar represents 5 μm.

D. Scanning electron microscopy (SEM) of triple deletion mutant. Cultures for SEM were grown as described above and processed as explained in Experimental procedures. Scale bar represents 3 μm.

The Spr-depleted double deletion mutant (Δspr ΔyebA/Para::spr) did not grow on LB (Fig. 5A) or on minimal A medium (MM) supplemented with 1% tryptone (data not shown); yet, it grew normally on MM supplemented with either 0.2% maltose or 0.5% casamino acids suggesting a growth-rate regulation of this process (Fig. 5A and B). However, a Spr-depleted triple deletion mutant (Δspr ΔyebA ΔydhO/Para::spr), did not grow on any media (Fig. 5A and B). The data above indicate that the activity of either Spr or YebA is essential for viability of E. coli in conditions of fast growth rate (i.e. during growth on rich media such as LB or MM-tryptone) possibly due to higher requirement of macromolecular synthesis whereas the activity of YdhO alone is sufficient for viability during conditions of low growth rate (i.e. growth on nutrient-poor media such as MM-maltose). The triple deletion mutant lacking all the three endopeptidases grew well when any one of these (Spr, YebA or YdhO) was provided in multiple copies confirming their essential and redundant functions in vivo (Figs 5A and S3).

Role of other peptidases in growth of E. coli

We further examined the phenotypes of an E. coli mutant deleted for all the other members of NlpC/P60 family. A Δspr ΔydhO double deletion mutant or a quadruple mutant lacking all the members of the family (Δspr ΔydhO ΔnlpC ΔyafL) did not grow on NA even at 30°C; however, none of these mutants showed any discernible phenotype on LB, LBON (LB lacking NaCl) or MM (Table S1). In addition, deletion of any of the genes encoding the other NlpC/P60 family members, nlpC, yafL, or other LytM family members, envC, nlpD, ygeR, or other DD-endopeptidases, dacB, pbpG or mepA did not confer additional sickness to either a single spr deletion mutant or spr yebA double deletion mutant (Table S1) demonstrating that spr, yebA and ydhO are alone redundantly necessary for growth of E. coli under normal physiological conditions.

Absence of Spr, YebA and YdhO leads to cell lysis

The terminal phenotype in both double and triple deletion mutants (Δspr ΔyebA/Para::spr and Δspr ΔyebA ΔydhO/Para::spr) grown under restrictive conditions was extensive cell lysis as evidenced by rapid decrease in optical density and colony-forming units (cfu) (Fig. 5B). Live-dead cell staining and viability measurements showed that depletion of Spr leads to immediate growth inhibition followed by lysis (Fig. 5B and C). DIC microscopy (Fig. 5C) and scanning electron microscopy (Fig. 5D) of Spr-depleted cells showed formation of large ovoid cells that ultimately lyse. Prior to lysis, most of the cells become huge and deformed with blebs, kinks and protrusions all over the surface. The pattern of lysis did not appear to be like that of a typical β-lactam-induced lysis in that there were no blebs or protrusions at the midcell (Chung et al., 2009; Uehara et al., 2009). Time-lapse microscopy images also showed rapid lysis of cells growing on LB-agarose pads (Movie S1). In addition, lysis in these mutant strains was observed only in growing cells; depletion of Spr in late exponential or early stationary-phase significantly reduced cell lysis (data not shown). Lysis (that is, the presence of cytoplasmic contents in the spent medium) was also evident by 100-fold higher levels of β-galactosidase activity in the supernatants of glucose-grown cultures compared with those from arabinose-grown cultures (Fig. S4).

Requirement of Spr, YebA and YdhO for new murein incorporation

Based on the data above that at least one of the three murein DD-endopeptidases is essential for cellular integrity of E. coli, and the fact that inhibition of murein synthesis is known to trigger cell lysis (Tomasz, 1979), we speculated that the cross-link cleaving activity of these hydrolases is required for new murein incorporation. To investigate this possibility, we measured incorporation of 3H-mDAP into the PG sacculi of the endopeptidase-deficient mutants. Because these cells lyse rapidly, incorporation studies were done before the onset of lysis. Figure 6 shows the incorporation of 3H-mDAP into insoluble murein material in wild-type, ampicillin-treated (wild-type), and Spr-depleted cultures at various time intervals. The data indicate that in both ampicillin-treated and endopeptidase-deficient cultures, 3H-mDAP incorporation into PG sacculi is decreased by about 75–80% compared with that in the wild-type strain confirming the requirement of these three endopeptidases in new murein incorporation (Fig. 6).

Figure 6.

Incorporation of new murein in the absence of Spr, YebA and YdhO. Incorporation of new murein using radioactive mDAP (3H-mDAP) was measured in strains MR500 (BW27783 ΔlysA::Kan) and MR509 (BW27783 ΔlysA::Kan Δspr ΔydhO ΔyebA/Para::spr). lysA deletion was used to prevent conversion of added mDAP into lysine (Wientjes et al., 1985). Cells were grown overnight in LB-arabinose (or LB for MR500), pelleted, washed and diluted into prewarmed MM supplemented with 0.5% casamino acids and 0.2% glucose and allowed to grow at 37°C. Five hundred microlitre aliquots were taken out at 90, 120, 150 min (prior to initiation of cell lysis in MR509) and pulsed with 3H-mDAP and were further processed as described in Experimental procedures. Strain MR500 treated with ampicillin (to inhibit murein synthesis) was used as a control. Cells were grown as described above and after 90 min of growth, ampicillin (Calbiochem, USA) was added to a final concentration of 6 μg ml−1 and grown further. Aliquots were collected at 90, 120, 150 min (equivalent to 0, 30, 60 min time points after addition of ampicillin) and processed as described above. The graph shows the incorporation of 3H-mDAP (the actual counts) in the above strains at various time intervals. Values are representative of three independent experiments.

DD-endopeptidase activity is fundamental for viability

To examine the importance of murein cross-link cleavage in growth of E. coli, we created site-directed mutations in the predicted active site of Spr. Spr belongs to C40 clan of peptidases with a conserved Cys (68)-His (119)-His (131) catalytic triad, a distinctive signature of the NlpC/P60 family proteins (Aramini et al., 2008). As cysteine at the catalytic site is predicted to be most crucial for peptidase activity, a mutant Spr (Spr-C68A) was generated and tested in vitro for DD-endopeptidase activity and in vivo for physiological activity. Consistent with the notion above, this mutation completely abolished the DD-endopeptidase activity of Spr (Fig. 7A) and also failed to support growth of a Δspr deletion mutant (on NA; data not shown) or a Δspr ΔyebA deletion mutant (Fig. 7B). Based on these results, we conclude that the cross-link cleavage activity is indeed essential for cell growth and viability.

Figure 7.

Effect of C68A substitution on the function of Spr.

A. DD-endopeptidase activity of Spr-C68A. HPLC profile of Tetra-Tetra treated with buffer (i) 5 μM Spr (ii) or 5 μM Spr-C68A (iii). The reaction products were processed as described in legend to Fig. 2. Spr-C68A (in which the active site cysteine is substituted with alanine) did not exhibit DD-endopeptidase activity (even with excess enzyme addition; data not shown) as can be seen by the presence of intact peak 3 (Tetra-Tetra) in panel iii. Peak 1 – disaccharide tripeptide (Tri); 2 – disaccharide tetrapeptide (Tetra); 3 – cross-linked disaccharide tetrapeptide (Tetra-Tetra).

B. Effect of Spr-C68A on growth of Δspr ΔyebA double mutant. Cells of MR511 (MG1655 Δspr ΔyebA/Ptrc::yebA) were transformed with plasmids encoding either Para::spr or Para::sprC68A. The transformed cells were grown overnight in LB-IPTG (50 μM) at 37°C. Cells were serially diluted and 5 μl of the dilutions were spotted on LB plates or LB plates supplemented with either IPTG (50 μM) or arabinose (0.2%) and incubated overnight at 37°C. Plasmid Para::spr was able to complement the strain Δspr ΔyebA/Ptrc::yebA in absence IPTG whereas Para::sprC68A was unable to complement the mutant strain. Growth defect of this double deletion mutant (Δspr ΔyebA/Ptrc::yebA/ Para::spr) on LB plates (with no added arabinose or IPTG) indicates a tight regulation of Para::spr and Ptrc::yebA.

Muropeptide composition in absence of Spr, YebA and YdhO

To examine the composition of murein in the absence of these endopeptidases, we prepared PG sacculi (Glauner, 1988) from single, double or triple deletion mutants lacking spr, yebA or ydhO in various combinations. Analysis of muropeptides indicated a significant increase in the amount of anhydro derivative of Tetra-Tetra (D44N) by about 100% (Table 1; Fig. S5) in the PG sacculi of Spr-depleted double (Δspr ΔyebA/Para::spr) or triple (Δspr ΔyebA ΔydhO/Para::spr) deletion mutants compared with that of wild-type strain. However, murein composition of other deletion mutants (Δspr, Δspr ΔydhO or ΔyebA ΔydhO) did not considerably alter from that of wild-type (data not shown).

Table 1. Muropeptide composition of DD-endopeptidase deficient mutant strains.a
MuropeptideWT (%)cΔsprΔyebA (%)ΔsprΔyebAΔydhO (%)
  1. aMurein was isolated from cells grown in restrictive conditions as described in Experimental procedures.
  2. bAmount of Tetra-Tetra−anhydro muropeptide was consistently higher in the endopeptidase-deficient mutants.
  3. cRelative molar percentage of each muropeptide and total cross-links were calculated from integration results of HPLC profiles as described by Glauner (1988). The values are from three independent experiments and are shown along with standard deviation.
Tri7.95 ± 0.456.04 ± 1.836.80 ± 0.91
Tetra30.73 ± 2.134.87 ± 3.0232.87 ± 2.08
Tri-Lys-Arg2.91 ± 1.612.54 ± 0.792.05 ± 1.11
Tetra-Tri (4-3)3.45 ± 0.982.88 ± 0.374.26 ± 1.71
Tetra-Tetra27.61 ± 3.6829.99 ± 1.0426.65 ± 2.21
Tetra-Anh1.22 ± 0.470.96 ± 0.281.01 ± 0.14
Tetra-Tri-Lys-Arg2.75 ± 0.332.21 ± 0.562.25 ± 0.34
Tetra-Tetra-Tetra2.44 ± 0.171.57 ± 0.121.95 ± 0.43
Tetra-Tetra-Anhb1.54 ± 0.253.61 ± 0.253.59 ± 0.40
Total cross-linkage22.02 ± 0.3023.88 ± 0.4226.65 ± 0.35


In the present study, we identified three new murein hydrolases, Spr, YebA and YdhO that are redundantly essential for growth of E. coli. In vitro, all of them show cleavage specificity for d-ala−mDAP peptide cross-bridges confirming their redundancy. Absence of these hydrolases inhibits new murein incorporation and leads to autolysis. The significance of their identification in this study stems from the fact that essential hydrolases that function in murein growth and enlargement are not known in E. coli till now (Tomasz, 1984; Holtje, 1998; Vollmer et al., 2008b; van Heijenoort, 2011; Typas et al., 2012). Our results indicate that the step of cross-link cleavage is a prerequisite for growth of murein making it an excellent target for development of novel inhibitors of bacterial cell wall synthesis.

Cross-link cleavage is essential for growth of PG sacculus

The molecular structure of murein sacculus intuitively suggests that cleavage of covalent bonds is essential for incorporation of new material for its enlargement (Weidel and Pelzer, 1964). Over the years, several models have been suggested to explain the mechanism of insertion of new glycan strands into the pre-existing murein sacculus, including an endopeptidase-mediated cleavage of existing cross-links to initiate the process of murein growth with simultaneous formation of cross-links between the new and old glycan strands (Burman and Park, 1984); and a ‘three-for-one’ growth mechanism in which initially three new glycan strands get attached to a pre-existing strand of the sacculus by the action of murein synthases followed by an endopeptidolytic cleavage of the old strand, finally resulting in replacement of one old strand with three new strands (Holtje, 1993). Several variations of these models have also been proposed (Holtje, 1996; Koch, 1998). However, there is no strong experimental support for any of these models and irrespective of the mechanism proposed, all of them envision cleavage of cross-links for new murein synthesis and have predicted the existence of essential cross-link cleaving enzymes (Weidel and Pelzer, 1964; Burman and Park, 1984; Tomasz, 1984; Koch, 1990; Holtje, 1998), also termed ‘space-maker hydrolases’ (Tomasz, 1984). The three endopeptidases identified in this study aptly fit the description of these long-postulated space-maker hydrolases.

Physiological functions of Spr, YebA and YdhO

As discussed below, several lines of evidence indicate that the primary role of these murein hydrolases is to cleave the d-ala−mDAP cross-links to facilitate growth of PG. Most importantly, these endopeptidases are required for new murein incorporation (Fig. 6) and consistent with this vital role, they are absolutely essential for growth and viability of E. coli (Fig. 5). Absence of these endopeptidases results in cell lysis and the morphology of lysing cells indicate induction of autolysins as a cause of lysis (Typas et al., 2012).

The enzymatic specificity of Spr, YebA and YdhO also denotes the significance of cross-link cleavage in murein synthesis. Hydrolases that participate in murein growth are predicted to be DD-endopeptidases as cleavage of d-ala−mDAP cross-links of the pre-existing strands is expected to generate additional acceptor tetrapeptide moieties for the formation of new cross-links (Tomasz, 1984). Though highly speculative, in this regard, accumulation of Tetra-Tetra−anhydro in the PG sacculi of mutants lacking these endopeptidases (Table 1; Fig. S5) raises an interesting scenario. Because Tetra-Tetra−anhydro residue is normally present at ends of cross-linked glycan strands (depicted in Fig. 8) as a natural by-product of glycan strand maturation, it is possible that cleavage of Tetra−anhydro residue from Tetra-Tetra−anhydro at the termini of pre-existing glycan strands is crucial for the entry of nascent glycan strands into the sacculus; these ends could be normally cleaved by DD-endopeptidases, and if not processed, the Tetra-Tetra−anhydro termini could prevent further attachment of glycan strands/peptides and may result in inhibition of nascent murein incorporation and autolysis (Fig. 8). Alternatively, Tetra-Tetra−anhydro accumulation could be merely a signature of autolysis mediated by the action of lytic transglycosylases in these endopeptidase-deficient mutant cells (Kohlrausch and Holtje, 1991). However, surprisingly, no other significant signatures of autolysis are observed in these mutants (Table 1; Fig. S5).

Figure 8.

Schematic diagram indicating the importance of cross-link cleavage in murein growth. Model depicts growth of murein sacculus. It is proposed that cleavage of cross-links in the pre-existing glycan strands requires murein hydrolytic enzymes to enable incorporation of new glycan strands for enlargement of the murein sacculus (Weidel and Pelzer, 1964). Our study identifies Spr, YebA and YdhO to be the murein hydrolytic enzymes that cleave the cross-links to facilitate murein synthesis. Absence of these enzymes leads to inhibition of murein synthesis and results in cell lysis.

Considering the proposed role of these hydrolases in the growth of murein sacculus, it is reasonable to expect that these will be active against intact PG sacculi. However, of the three hydrolases identified in this study, only YebA and YdhO exhibited DD-endopeptidase activity on intact PG sacculi in vitro under the experimental conditions tested, although, all of these were highly effective in cleaving the muropeptide cross-links (Fig. S1). Though, Spr did not exhibit any activity against the intact sacculi in these conditions, we strongly believe that it will be active on PG sacculus in vivo for following reasons. It has been earlier observed that multiple copies of Pbp7 known to act only against intact murein suppress the defects of an spr deletion mutant. In addition, the recently identified orthologues of Spr which are effectors of type VI secretion system, Tse1, Tae2 and Tae3 are shown to be active on intact PG sacculi of E. coli (Russel et al., 2012). It is also possible that Spr may require an additional cofactor or a protein (or specialized reaction conditions) for its optimal activity.

DD-endopeptidases of E. coli

The NlpC/P60 class of proteins defines cell-wall peptidases that predominantly hydrolyse d-glu−mDAP or N-acetylmuramate−l-ala linkages of murein (Anantharaman and Aravind, 2003). Incidentally, Spr and YdhO are among the very few reported members of NlpC/P60 family that show specificity for d-ala−mDAP peptide cross-links (Figs 2 and 4; Russel et al., 2012). Conversely, YebA belongs to LytM or lysostaphin family of proteins (M23 class of metallo-endopeptidases) that are known to cleave peptide cross-links in various bacteria (Firczuk et al., 2005). It has recently been shown that YebA may also have a minor role in cell septation (Uehara et al., 2009; Babu et al., 2011). However, the other members of this family, EnvC and NlpD do not have murein hydrolytic activity of their own (Uehara et al., 2010).

Escherichia coli has three other DD-endopeptidases, Pbp4, Pbp7 and MepA, which cleave the d-ala−mDAP cross-bridges of murein. Pbp4 and Pbp7 are implicated in maintenance of cell shape and cell-cell separation (Meberg et al., 2004; Priyadarshini et al., 2006) and Pbp7 is also believed to be a component of a hypothetical multienzyme complex that functions in murein synthesis (Romeis and Holtje, 1994b). MepA, a penicillin-insensitive, lysostaphin-like-metallopeptidase, is not implicated in any biological process in the cell though its enzymatic activity is very well-studied in vitro (Keck and Schwarz, 1979; van Heijenoort, 2011). AmpH, a low-molecular-weight penicillin-binding protein has recently been shown to possess both DD-carboxypeptidase and DD-endopeptidase activities (Gonzalez-Leiza et al., 2011).

Regulation of cross-link cleavage

It is expected that growth and enlargement of PG sacculus would be a highly regulated event requiring fine balance between murein synthesis (PG synthases) and cleavage (PG hydrolases). Unregulated activity of hydrolases will lead to destruction of murein and it is not clear how this potentially lethal activity is controlled in the cell. Regulation of division-specific cell wall amidases AmiA, -B, -C has been unravelled recently (Uehara et al., 2010; Yang et al., 2012). One attractive possibility for the modulation of the endopeptidases identified in this study, is that the cross-link status of PG itself may regulate the cleavage activity such that the rigidity (or flexibility) of PG in presence of high cross-linking induces the cleavage followed by insertion of new glycan chains and formation of new cross-links leading to the enlargement of murein sacculus. That is, these endopeptidases could be ‘smart autolysins’ as predicted (Koch, 1990) in that they recognize and cleave the cross-links that are in ‘stressed’ but not in ‘relaxed’ conformation. In this case, the mechanisms of signalling between cross-link status and cleavage, as also the cross-talk between the cleavage and cross-link forming machinery remain to be elucidated.

Alternatively, these endopeptidases could be governed by components of cell elongation machinery. It is believed that the cell-wall elongation apparatus comprising of helical cytoskeletal proteins, MreBCD and RodZ along with murein biosynthetic enzymes, Pbp2 and RodA may form a complex (or elongase) that catalyses elongation-specific murein synthesis coupling lateral wall growth to cell shape (den Blaauwen et al., 2008; Bendezu et al., 2009; Gerdes, 2009; Vats et al., 2009). At present, it is not known whether any of the endopeptidases identified here physically interact with such cell elongation machinery as in B. subtilis, in which a cell wall hydrolase LytE is specifically associated with MreB elongation sites (Carballido-Lopez et al., 2006).

Experimental procedures

Media, bacterial strains and plasmids

Strains were grown in LB (1% tryptone, 0.5% yeast extract, 1% NaCl) or MM (Miller, 1992) supplemented with either 0.5% casamino acids or 0.2% maltose as carbon source. Nutrient agar contained 0.5% bacto-peptone and 0.3% beef extract (Miller, 1992). Unless otherwise indicated, antibiotics were used at the following concentrations (μg ml−1): ampicillin-50; chloramphenicol-25; and kanamycin-25. l-arabinose or d-glucose (0.2%) was used to induce or repress the expression from the Para vector. Growth temperature was 37°C unless otherwise indicated.

The bacterial strains and plasmids used in this study are listed in Table 2. All strains are derivatives of MG1655 (Coli Genetic Stock Centre) or BW27783 (Khlebnikov et al., 2001) unless otherwise indicated. The deletion mutations from the Keio mutant collection (Baba et al., 2006) were used after confirming their authenticity by linkage analysis, PCR, and phenotype if known. Marker-less strains were made by flipping out the antibiotic resistance (KanR) marker using pCP20 plasmid that encodes a Flp recombinase (Datsenko and Wanner, 2000).

Table 2. Bacterial strains and plasmids used in this study
  1. aDeletion alleles used in this study are sourced from Keio collection (Baba et al., 2006). The deletion mutations were used after testing for their authenticity (by linkage analysis, PCR and sequence analysis) and introduced into different strain backgrounds by P1 phage-mediated transduction (Miller, 1992). Marker-less strains were made by flipping out the antibiotic resistance (KanR) marker using pCP20 plasmid that encodes a Flp recombinase (Datsenko and Wanner, 2000).
BL21 (λDE3)ompT rB mB (PlacUV5::T7gene1)Lab collection
C41 (λDE3)BL21 (λDE3) with an uncharacterized mutationWagner et al. (2008)
MG1655rph1 ilvG rfb-50Lab collection

lacIq rrnB3 ΔlacZ4787 Δ(araBAD)567

Δ(rhaBAD)568 hsdR514

Datsenko and Wanner (2000)
BW27783BW25113 Δ(araFGH) Φ(ΔaraEp PCP8araE)Khlebnikov et al. (2001)
MR500BW27783 ΔlysA::KanRThis study
MR501BW27783 Δspr/pMN83This study
MR504BW27783 Δspr ΔydhO/pMN83This study
MR506BW27783 Δspr ΔyebA/pMN83This study
MR508BW27783 Δspr ΔyebA ΔydhO/pMN83This study
MR509BW27783 Δspr ΔyebA ΔydhO ΔlysA::KanR/pMN83This study
MR510MG1655 Δspr ΔyebA/pMN83This study
MR511MG1655 Δspr ΔyebA/pMN81This study
MR512MG1655 Δspr ΔyebA ΔydhO/pMN81This study
MR513MG1655 Δspr ΔyebA ΔydhO/pMN82This study
MR514MG1655 Δspr::KanR /pTRC99aThis study
MR515MG1655 Δspr::KanR /pMN80This study
MR516MG1655 Δspr::KanR /pMN82This study
MR517MG1655 Δspr::KanR /pMN81This study
PlasmidsRelevant features 
pBAD33ColE1, CmR, Para BADGuzman et al. (1995)
pET21bColE1, AmpR, lacIq, T7lacNovagen
pTRC99aColE1, AmpR, lacIq, PtrcLab collection
pMN80pTRC99a−sprThis study
pMN81pTRC99a−yebAThis study
pMN82pTRC99a−ydhOThis study
pMN83pBAD33−sprThis study
pMN84pET21b−spr28-188This study
pMN85pET21b−ydhO28-271This study
pMN86pET21b−yebA40-440This study
pMN87pET21b-spr28-188 C68AThis study
pMN88pBAD33-spr C68AThis study

MG1655 genomic DNA was used as template for all amplifications by PCR. Pfx polymerase (Invitrogen) was used for amplifications and the cloned fragments were routinely confirmed by sequence analysis. PCR products encoding Spr28-188 or YdhO28-271 were cloned at NdeI-XhoI sites whereas YebA40-440 was cloned using NdeI-SacI sites to generate C-terminal His6 fusions in pET21b (Novagen). Full length spr, ydhO and yebA ORFs lacking their respective promoter regions were cloned into pTRC99a vector downstream to the trc promoter for regulated expression using IPTG as an inducer. spr was also cloned using XbaI and PstI sites of pBAD33 vector under the control of an arabinose-inducible promoter (Para::spr). (Construction of these plasmids is described in detail in Supporting Information.)

Standard plasmid transformations, P1 phage-mediated transductions and β-galactosidase measurements were done as previously described (Miller, 1992).

Identification of multicopy plasmid suppressors of Δspr

A multicopy plasmid library carrying overlapping E. coli genomic DNA fragments cloned at BamHI site in a p15A-based plasmid, pACYC184 (obtained from Miroslav Radman's laboratory) was transformed into MG1655 Δspr deletion mutant. Transformants were plated on NA-Cm plates at 42°C and plasmids were isolated from colonies that grew very well on these plates. The ability of their suppression was reconfirmed after an additional round of transformation. Subsequent restriction analysis and sequencing with the flanking vector primers 184TetA [5′-CGCCGAAACAAGCGCTCATGAGCC-3] and 184TetB [5′-CTATGCGCACCCGTTCTCGGAGCAC-3] yielded various classes of plasmids that restored viability to Δspr mutant on NA to different extents. The putative candidate genes were cloned into other plasmid vectors and their ability to suppress the NA-Ts phenotype of spr was reconfirmed.

Microscopy and viability measurements

To monitor growth, indicated cells were grown overnight in permissive conditions and next day they were washed and diluted 1:100 into fresh medium supplemented with or without the inducer. At indicated time points, absorbance at 600 nm and cfu were measured and additional 0.5 ml culture was drawn for microscopy. Cell viability was measured using LIVE/DEAD Baclight bacterial viability kit (Molecular Probes, Invitrogen). The immobilized cultures on a thin agarose pad were visualized on a Zeiss apotome microscope by DIC (Nomarski optics) and fluorescence microscopy using GFP and DsRed filters.

Scanning electron microscopy

Scanning electron microscopy was performed essentially as described earlier (Legaree et al., 2007). In brief, cells were pelleted, washed and resuspended in 0.07 M Sorensen's phosphate buffer (pH 6.8). Ten microlitre of cell suspension was spotted on a poly-l-lysine coated coverslip, incubated for 15 min and fixed by addition of Sorensen's buffer containing 2.5% glutaraldehyde for 1 h at room temperature. Samples were washed repeatedly with buffer and then dehydrated by a series of 10 min ethanol washes followed by critical point drying with CO2. Coverslips were sputter coated with gold for 90 sec in Polaron Range SC7620 sputter coater (Quorum Technologies). Samples were scanned by Hitachi 3400 scanning electron microscope (Hitachi Science Systems, Hitachinaka, Japan) at an accelerated voltage of 5 kV.

Time-lapse microscopy

Time-lapse microscopy was done using Zeiss confocal microscope equipped with an incubation chamber. Appropriate numbers of cells were placed on a thin LB-agarose pad on a shallow depression microscopic slide and growth was continuously monitored at 37°C by DIC imaging.

Protein overexpression and purification

Spr and YebA were purified from C41λDE3 (Wagner et al., 2008) whereas YdhO was purified from BL21λDE3 (Novagen). All proteins were fused to a C-terminal His tag and were hence purified by metal-affinity chromatography using Ni2+-NTA agarose column (Qiagen) as per the manufacturer's instructions. Protein samples were regularly stored at −30°C in 50 mM NaH2PO4, 100 mM NaCl, 1 mM DTT and 50% glycerol. Details of protein purification are given in Supporting Information.

Preparation of murein sacculi

Peptidoglycan or murein sacculi were prepared as described earlier (Glauner, 1988). Cells from 1 l of exponentially growing culture (OD600 ∼ 0.6) of indicated strains were harvested by centrifugation at 10 000 g for 20 min at 4°C. Cell pellet was resuspended in 6 ml of ice-cold deionized water and added drop-wise to 6 ml of boiling 8% SDS within a time period of 10 min with vigorous stirring. Boiling was extended for next 45 min to destroy high molecular weight DNA and for complete membrane solubilization. The mixture was incubated overnight at room temperature and sacculi were collected by ultracentrifugation (130 000 g, 60 min, room temperature). The pellet was washed repeatedly with deionized water to remove SDS from the pellet. Concentration of SDS was assayed as previously described (Hayashi, 1975). High molecular weight glycogen and covalently bound lipoprotein were destroyed by addition of α-amylase (100 μg ml−1 in 10 mM Tris-HCl, pH 7.0, 2 h at 37°C) and pre-digested pronase (200 μg ml−1, 90 min at 60°C) respectively. Samples were boiled with equal volume of 8% SDS for 15 min to inactivate the enzymes. Sacculi were collected by ultracentrifugation and washed repeatedly with deionized water to remove traces of SDS. Finally, the pellet was resuspended in 25 mM Tris-HCl (8.0) and stored at −30°C.


Zymography was performed using intact murein sacculi of MG1655 as described earlier (Bernhardt and de Boer, 2004). Purified proteins were run on SDS-PAGE gels impregnated with 0.7% sacculi (wet w/v). They were electrophoresed, renatured and cleared zones were visualized by staining the gel with methylene blue/KOH.

Determination of enzymatic activity and HPLC analysis of muropeptides

The HPLC analysis was performed as previously described (Glauner, 1988; Uehara et al., 2009) with slight modifications. Total muropeptides were prepared by digesting the sacculi with 20 U mutanolysin (Sigma-Aldrich) at 37°C in 25 mM Tris-HCl (pH 8.0) for 16 h and insoluble material was removed by centrifugation (10000 g, 15 min at room temperature). The soluble muropeptide fraction was incubated with either 5 μM Spr, 2 μM YebA or 3.5 μM YdhO at 26°C for 20 h. Samples were heat inactivated and reduced with 1 mg of sodium borohydride in 50 mM sodium borate buffer (pH 9.0) for 30 min. Excess borohydride was destroyed by addition of 20% phosphoric acid and pH was adjusted to 3–4 before loading onto a Zorbax 300 SB RP-C18 (250 × 4.6 mm, 5 μm) column. HPLC was performed using Agilent technologies RRLC 1200 system. Samples were injected onto a pre-heated column at 50°C and binding was allowed at a flow rate of 0.5 ml min−1 with 1% acetonitrile in water containing 0.1% trifluoroacetic acid (TFA) for 10 min. A gradient of 1–10% acetonitrile containing 0.1% TFA was used for elution at a flow rate of 0.5 ml min−1 for next 60 min (using RRLC online software). Absorbance was detected at 205 nm and the fractionated muropeptides were identified by MALDI-MS and MS/MS analysis.

DD-endopeptidase activity on intact PG sacculi

To test the activity of murein hydrolases on intact sacculi, purified proteins (5 μM Spr, 2 μM YebA or 3.5 μM YdhO) were incubated overnight with sacculi in 25 mM Tris-HCl (pH 8.0) at 37°C. Enzymes were inactivated by boiling for 5 min at 100°C. Samples were then incubated overnight at 37°C with 20 units of mutanolysin (Sigma). Insoluble material was removed by centrifugation at 10 000 g for 15 min and the supernatant was analysed by reverse phase HPLC (RP-HPLC) as described above.

MALDI-TOF analysis

Muropeptides collected from the fractions of HPLC were dried and reconstituted into 50% acetonitrile containing 0.1% TFA. An aliquot was mixed with equal volume of matrix (α-cyano-4-hydroxycinnamic acid; CHCA) and spotted onto a target plate. Data were acquired in positive ion reflector mode on 4800 plus MALDI TOF/TOF system (ABI systems). Peptide mass was analysed by GPS software version 3.6 (ABI). Relevant peaks were further analysed by MS/MS.

Measurement of 3H-mDAP incorporation into murein sacculi

Incorporation of 3H-mDAP into murein sacculi was done as described (Wientjes et al., 1985). Indicated strains were grown overnight in LB broth, washed and diluted 1:100 into prewarmed MM supplemented with 0.5% casamino acids and 0.2% glucose and allowed to grow. For pulse-labelling of murein, 0.5 ml aliquots were removed at regular intervals and incubated with 5 μCi ml−1 [3H] DAP (14.6 Ci mmol−1, Moravek Biochemicals, USA) for 10 min with gentle shaking at 37°C. Cells were immediately lysed by addition of 3 ml of 4% SDS and boiled for 1 h. The mixture was allowed to cool to room temperature and filtered through 0.22 μm filters (Millipore). The filters were washed with 30 ml of Milli-Q water and dried completely before counting the radioactivity in a liquid scintillation counter (Perkin-Elmer) using toluene-based scintillation fluid.


We thank NBRP (Japan): E. coli for Keio mutant collection; C Subbalakshmi, V Krishna Kumari for HPLC analysis; C Sundaram for mass spectrometry; and J Gowrishankar for advice and suggestions on the manuscript. This work is supported by funds from Council of Scientific and Industrial Research and Department of Biotechnology, Government of India. SKS acknowledges financial support from Indian Council of Medical Research.

Conflict of interest

The authors declare that they have no conflict of interest.