Shared catalysis in virus entry and bacterial cell wall depolymerization

Summary

Bacterial virus entry and cell wall depolymerization require the breakdown of peptidoglycan (PG), the peptide cross-linked polysaccharide matrix that surrounds bacterial cells. Structural studies of lysostaphin, a PG lytic enzyme (autolysin), have suggested that residues in the active site facilitate hydrolysis, but a clear mechanism for this reaction has remained unsolved. The active site residues and a structural pattern of β-sheets are conserved among lysostaphin homologs (such as LytM of Staphylococcus aureus) and the C-terminal domain of gene product 13 (gp13), a protein at the tail tip of the Bacillus subtilis bacteriophage φ29. gp13 activity on PG and muropeptides was assayed using high performance liquid chromatography, and gp13 was found to be a D,D-endopeptidase that cleaved the peptide cross-link. Computational modeling of the B. subtilis cross-linked peptide into the gp13 active site suggested that Asp195 may facilitate scissile bond activation and His247 is oriented to mediate nucleophile generation. This is the first model of a Zn²⁺-metallopeptidase and its substrate to our knowledge. Residue Asp195 of gp13 was found to be critical for Zn²⁺-binding and catalysis by substitution mutagenesis with Ala or Cys. Circular dichroism and particle induced X-ray emission spectroscopy showed that the general protein folding and Zn²⁺-binding was maintained in the Cys mutant but reduced in the Ala mutant. These findings together support a model where the Asp195 and His247 in gp13 and homologous residues in the LytM and lysostaphin active sites facilitate hydrolysis of the peptide substrate that cross-links PG. Thus, these autolysins and phage entry enzymes have a shared chemical mechanism of action.

INTRODUCTION

Bacterial cell growth and infection by bacteriophages require remodeling of the peptidoglycan (PG) cell wall. Autolysis is the process of cell wall degradation by self-produced enzymes that normally permit bacterial cell expansion and division. Phages are extremely efficient at crossing the cell wall barrier during infection by both mechanical (phages T4, P22)^1–3 and enzymatic (phages GA-1, φ29, K1E, K1–5, P22, PRD1, T4, T5, T7/T3)4–11 means. The latter involves a battery of lysozymes, amidases, transglycosylases and endopeptidases that cleave either the polysaccharide or peptide bonds of the PG, providing access to the lipid bilayer. The PG of Bacillus subtilis, the host of bacteriophage φ29, is depicted in Fig. 1a.

Figure 1. B. subtiliscell wall peptidoglycan structures.

(a) B. subtilis cell wall peptidoglycan is composed of alternating N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM) residues with peptide side chains. Peptides can be cross-linked via the D-ala in position 4 and the free ε-amino group of diaminopimelic acid (Dpm). Over 95% of the Dpm free carboxylate groups are amidated³¹. The predominant cross-linked muropeptide fragment, p21, is two disaccharides cross-linked as shown via their peptide side chains (panel a, left). The activity of gp13 cleaves p21 to produce a disaccharide-tripeptide and a disaccharide-tetrapeptide. This could result from hydrolysis of either of two bonds (panel b, sites a and b) surrounding the cross-link (panel a, bold amino acids and bonds). However, the potential disaccharide-tetrapeptide products differ in the amino acids that have free amino groups (panel b, circles) and thus can be differentiated following modification of these amino groups followed by amino acid analysis (Table 1). The cross-link fragment used for modeling is shown and the chiral centers are indicated as L or D (panel b, dashed box).

Structural and biochemical studies have enlightened autolysin mediated PG remodeling for cell division and lysozyme-mediated phage entry into host bacteria. Atomic structures are available for both the inactive¹² and active¹³ forms of the Zn²⁺ metallopeptidase LytM of Staphylococcus aureus, a homolog of lysostaphin of Staphylococcus simulans. Lysostaphin is a well-studied metallopeptidase that cleaves the pentaglycine cross-link of the staphylococcal cell wall during autolysis^14–19. The atomic structures of the lysostaphin homolog Ale-1 of Staphylococcus capitis²⁰ and LAS protein MepA of Escherichia coli²¹ have also been determined. However, the precise molecular mechanism for catalysis by the conserved amino acids within LytM, lysostaphin and the lysostaphin-like enzymes remains unknown^{14, 18}. Models and cocrystallized structures of lysostaphin-like peptidases with bound substrates are also not available. Sequence alignments have suggested the presence of lysostaphin-like homologs in phages of both B. subtilis²² and Lactobacillus²³, and these are virion components. However, biochemical characterization of these phage proteins has not confirmed the proposed activity.

Bacteriophage φ29 has served as a model for virus assembly mechanisms24, 25. Early studies of the φ29 interaction with the host cell wall revealed that glucosylated teichoic acids rather than a protein serve as the receptor^{26, 27}. Furthermore, the ability of the phage to adsorb to teichoic acids has been shown to reside in the phage tail appendages²⁸. However, adsorption alone is insufficient for infection and progeny production; successful puncture of the cell wall is also required. Here, biochemical study of the cell wall depolymerizing metallopeptidase of bacteriophage φ29 of B. subtilis, gene product 13 (gp13), a homolog of the S. aureus autolysin LytM, is complemented by computational modeling. A possible mechanism of the unified action of phage entry enzymes and autolysins is proposed, based on biochemical and computational characterization of the phage φ29 morphogenetic factor/cell wall depolymerizing enzyme gp13^{22, 29}.

RESULTS

Purified φ29 particles and gp13 cleave the Bacillus subtilis peptidoglycan cross-links

Previous studies localized gp13 to the distal φ29 tail, the site of DNA injection following host cell binding, by immuno-electron microscopy²² and cryo-electron microscopy³⁰. A β-sheet rich C-terminal domain of gp13^{22, 30} is similar to LytM, the lysostaphin homolog in Staphylococcus aureus, while the N-terminal domain is α-helix rich.

To test the activity of φ29 on the purified peptidoglycan (PG) of B. subtilis (Fig. 1a), purified cell wall sacculi were pseudo-infected with wild-type (wt) φ29(see Methods). It is estimated that there are hundreds of glucosylated teichoic acid φ29 receptors per bacterium²⁶, and each cell has been shown to adsorb ~400 φ29²⁸. Intact PG was largely insoluble and could be pelleted for subsequent muramidase solubilization (see Methods) and high pressure liquid chromatography (HPLC) analysis of muropeptides from the pelleted PG. Following pseudo-infection of sacculi with φ29, there was release of muropeptides into the supernatant that was predominantly p7 (Supplementary Fig. 1d), a disaccharide-tetrapeptide component of the B. subtilis PG (Fig. 1b), while intact PG had little soluble material (Supplementary Fig. 1b). There was also a modest decrease in the cross-linked muropeptide p21 in the φ29 treated insoluble fraction (Supplementary Fig. 1c) compared to the untreated control PG (Supplementary Fig. 1a).

Incubation of sacculi (Fig. 2, blue trace) with purified gp13 (Fig. 2, red trace, offset by 1 min) also revealed a modest decrease in cross-linked species with a corresponding increase in uncross-linked species. gp13 did not cause cleavage of pentaglycine, the lysostaphin substrate in a thin-layer chromatography assay often used for lysostaphin study¹² (data not shown).

Figure 2.

gp13 activity on purified B. subtilis peptidoglycan. Mutanolysin digestion of B. subtilis sacculi produces a characteristic profile of peaks (blue line). The untreated sacculi (blue) show the characteristic pattern of cross-linked (>60 min) and uncross-linked (<40 min) muropeptides. Following incubation with purified gp13 protein a modest increase in monomer disaccharide (DS) tripeptide (TriP) and DS tetrapeptide (TP) peaks (under 40 min) is apparent with a slight decrease in cross-linked peaks beyond 60 min (red line). The 43.5 min peak present in the gp13 sample (* red line) is a contaminant from the purified protein solution.

To ensure that gp13 was a tail component with enzymatic activity, HPLC tracking of the dominant cross-linked muropeptide p21 was employed with φ29 virions and gp13 (Fig. 3a). Incubation of φ29 with purified p21 resulted in modest cleavage of p21 to products of smaller molecular mass (Fig. 3a, bottom red line), visualized most clearly on subtraction of the baseline absorption from the raw traces of particles incubated with p21 (Supplementary Fig. 2). Product peaks are at ~15 min and ~30 min (Fig. 3a). The precursor prohead particle, that lacks gp13, does not show the 15 or 30 min peaks (Fig. 3a, bottom black line). This subtle difference suggested that gp13 of the virion tail may have catalytic activity.

Figure 3.

Purified gp13 and φ29 cleaved p21 to the same products, and this activity functions on p15, p22 and p23. Purified p21 alone (a, top red) shows a lone peak at ~81 min, while purified gp13 alone (a, blue) lacks the 81 min peak but contains a 49 min contaminant (*). Incubation of gp13 with p21 (a, top black) shows reduction of p21, with the appearance of two product peaks at 15 min and 30 min. These peaks are present on p21 incubation with φ29 (a, bottom red), but not with gp13-lacking proheads (a, bottom black). These latter two lines are flat because they were produced by subtractions of background runs (Supplementary Fig. 2). (b) Deacetylated (deacetyl) and amidated (amid) forms of the cross-linked muropeptide of B. subtilis are all cleaved to lower molecular weight products. p3: DS-TriP^Amid, p4: DS^Deacetyl-TriP^Amid, p7: DS-TP^Amid, p15: DS-TriP^Amid-TP-DS, p21: DS-TriP^Amid-TP^Amid-DS, p22: DS-TriP^Amid-TP^Amid-DS^Deacetyl, p23: DS^Deacetyl-TriP^Amid-TP^Amid-DS, I: DS-TP, II: DS^Deacetyl-TP^Amid.

Purified cross-linked species were utilized to determine if gp13 contained the enzymatic activity on p21 observed with the virion. gp13 cleaved p21 to products that co-migrated with the φ29 produced products (Fig. 3a, top black line). p15, p21, p22 and p23 (Fig. 3b) are various forms of the cross-linked muropeptide group³¹. The predominant form, p21, is acetylated on both of the glucosamine residues (Fig. 1a), while p22 and p23 are deacetylated on one of the two glucosamine residues, and p15 is amidated on only one of the Dpm residues. gp13 incubated with each purified cross-linked muropeptide showed substantial product formation (Fig. 3b) to species that migrated at ~15 min and ~30 min. These products had elution times consistent with those of previously identified muropeptides^{31, 32} (Supplementary Fig. 1a).

φ29 and gp13 cleave the site of the original peptide cross-link formation

Cleavage of the cross-linked peptides may occur at a number of sites: two are illustrated (Fig. 1b). Mass spectrometry analysis of the peak contents that eluted at 15 and 30 min demonstrated masses consistent with p3 and p7 (Fig. 1b, site a) or p3 and p7′ (Fig. 1b, site b) (Table 1). Fluorodinitrobenzene (FDNB) modification of free amino groups in the cleavage products, followed by amino acid analysis, was used to differentiate p7 from p7′ and to determine the specific cleavage site of gp13. p7 and p3 (Fig. 1b, site a) are the products of p21 cleavage by gp13, because only Dpm reacted with FDNB (Table 1). Alanine would have reacted with FDNB if p7′ had been the other product (Fig. 1b, site b).

Table 1.

Muropeptide structure analyses.

				Amino acid analysis (molar ratios)
Muropeptide Name	Muropeptide Structure	m/z (M-H)−		−FDNB			+FDNB
		Predicted	Observed	Ala	Dpm	Glua	Ala	Dpm	Glua
p21	DS-TP-DS-TriP	1790.8	1790.6	1.6	1.1	1	1.5	0.5	1
p3	DS-TriP	868.9	868.3	1.1	0.9	1	1.0	0.0	1
p7	DS-TP	939.4	939.3	2.0	1.1	1	1.5	0.1	1

^aGlu was set to 1 for calculation of molar ratios.

Preliminary assessment of product formation kinetics (Supplementary data Fig. 3) suggested that the gp13 solution cleavage rate is slow (~1–2/min), however a similar rate, reported for the phage P22 tail spike endorhamnosidase gp9³³, is sufficient for infection.

The C-terminal domain of gp13 is active on the peptidoglycan cross-link

Predictions in previous work^{22, 30} suggested that the gp13 C-terminal domain (CTD) may contain the enzymatic activity on the PG cross-link. To determine if the CTD contains the cross-link-cleaving activity of the protein, the N-terminal domain (NTD) and CTD were expressed separately, purified and tested for their activity in the HPLC assay on the cross-linked muropeptide p21. The NTD is 159 amino acids (aa) and 18.4 kDa, while the CTD is 202 aa and 22.3 kDa (wild-type gp13 is 365 aa and 40.9 kDa) (calculated using ExPASy Protparam³⁴). The mole fraction of the substrate and products are shown for the CTD and NTD alone compared to full-length gp13 (Fig. 4). p3 and p7 were produced by the CTD but not the NTD.

Figure 4. Domain specificity of gp13-mediated cross-link cleavage.

Mass corrected peak area from HPLC analysis of purified N-terminal domain (NTD) and C-terminal domain (CTD) of gp13 incubated with p21. p21 remained uncleaved with the NTD, while product formation was nearly complete with the CTD. p21 (solid bars), p3 (hatched bars), p7 (open bars).

Computational modeling of the gp13-muropeptide complex shows the orientation of the scissile bond

To elucidate the catalytic mechanism underlying the cross-link cleavage mediated by the gp13 CTD and therefore the autolysins such as LytM, the central segment of the p21 muropeptide containing the Dpm-D-ala-Dpm region (Fig. 1b, dashed box) was modeled into the putative active site of gp13. The catalytic binding site of gp13 was identified as the region enclosing the cleft of the gp13 β-sheet domain³⁰ containing a Zn²⁺cofactor. This region also contains several residues that are highly conserved across LytM and gp13-like proteins of other φ29-related and lactococcal phages (Fig. 5). The original X-ray structure of apo gp13 contains a Zn²⁺ ion tetracoordinated to D195, H188, H280 and a conserved crystallographic water. The closest oxygen of Aspartate195 (D195) of gp13 is 1.94 Å from Zn²⁺, and the other oxygen is 2.87 Å from Zn²⁺. Successful docking of the modeled substrate requires the removal of the zinc bound crystallographic water. Of the 50,000 conformations searched, only seven poses were found, all of which shared a common mode of binding involving the carbonyl oxygen atom of the scissile amide bond positioned within 2.1 to 2.4 Å adjacent to the Zn²⁺ ion at the original crystallographic water site (Fig. 6a, arrow). The pose with the best docking score is reported here, and the overall model suggests the requirement of water displacement by the potential substrate from the Zn²⁺ ion as well as chelation of the carbonyl oxygen atom to Zn²⁺ ion during substrate binding. This mode of binding is supported by a previously solved X-ray structure of latent LytM of Staphylococcus aureus, which showed the Zn²⁺ ion chelated directly to the carbonyl oxygen atom of the N117 sidechain¹². In comparison to the hydrolytic mechanism of another mononuclear zinc metallopeptidase^{19, 35}, the placement of the carbonyl oxygen atom adjacent to the Zn²⁺ ion is necessary in the activation of the scissile peptide bond of during hydrolysis and the subsequent stabilization of the oxyanion atom of the tetrahedral intermediate. Based on this model, replacement of D195, which is directly chelated to the Zn²⁺ ion, is expected to alter the overall stabilization effect of the Zn²⁺ ion and, hence, indirectly affect the overall rate of enzymatic hydrolysis.

Figure 5.

Sequence alignment of gp13 in φ29, B103, GA-1, with LytM and lactococcal phage homologs from bIL285, TP90101, Tuc2009 and ul36. Sequence alignment of φ29 gp13 (top line) compared to other φ29 family members GA-1 and B-103 (lines 2, 3), LytM (line 4) and four lactococcal phages (lines 5–8) over the segment of gp13 containing the Zn²⁺-binding histidine motif and Asp195 The conserved Zn.²⁺ chelating residues (φ29 gp13: H188, D195, H280; LytM*: H210, D214, H293) and the two nearby histidines (φ29 gp13: H247, H278; LytM*: H260, H291) have been specifically labeled. The residue number for the first amino acid of each protein is listed in parentheses.

Figure 6.

Modeling of the p21 substrate cleavage target into the putative gp13-Zn²⁺ containing active site. (a) Modeling of the p21 substrate target (carbon: yellow, oxygen: red, nitrogen: blue, hydrogen: white), containing the peptide bonds of Dpm-Ala-Dpm resulted in energy minimizations that favor orientation of the carbonyl group adjacent to the scissile bond (arrow) as the fourth Zn²⁺ ligand in gp13. All residues within 5 Å of the model fragment are shown (sticks) with six essential residues as rods. (b) The entire gp13 crystal structure is shown as a ribbon diagram for reference.

The active-site aspartate residue of gp13 binds Zn for polarization

To demonstrate the importance of D195 and its role as part of the catalytic site of gp13, site-directed mutagenesis of the gp13 expression plasmid was performed to generate cysteine and alanine substitutions. The cysteine substitution (D195C) would be expected to maintain Zn²⁺ binding but may be more electron withdrawing than aspartate³⁶. The alanine substitution (D195A) is expected to block both Zn²⁺-binding and catalytic activity. Both the D195C and D195A proteins did not show cleavage of p21 to p3 or p7 (Fig. 7a). Thus, D195 was important for enzyme activity.

Figure 7.

gp13 aspartate 195 is involved in Zn²⁺ binding that is essential for proper protein folding and activity. (a) Mass corrected peak area from HPLC analysis of gp13 point mutants D195C and D195A. p21 (solid bars), p3 (hatched bars), p7 (open bars). (b) Particle Induced X-ray Emission spectra of wild-type gp13 (black) and mutants D195C (blue) and D195A (red). (c) Circular dichroism of wild-type gp13 (black) and mutants D195C (blue) and D195A (red).

To confirm that the D195C change did not prevent Zn²⁺-binding by gp13, particle induced X-ray emission (PIXE) spectra of wild-type gp13, gp13 (D195C), and gp13 (D195A) were obtained (Fig. 7b). The similarity of the PIXE profiles of the wild type and D195C proteins indicate that the D195C protein bound Zn²⁺, while the D195A protein did not. Alanine substitution of the homologous aspartate in MepA (D120A) rendered this protein insoluble³⁷. To confirm that gp13 D195A was soluble and to compare the secondary structure folding of wild-type gp13 to both the D195A and D195C proteins, circular dichroism (CD) was employed. CD spectra of gp13 and D195C show the same profile (Fig. 7c) (the vertical shift of D195C may be due to slight differences in protein concentration), while the D195A spectrum was characteristic of substantial disorder of the secondary structure that was absent in wild-type gp13 and gp13 D195C data. These results support the hypothesis that the CTD of gp13 is a PG cross-link-cleaving Zn²⁺metalloprotease and that D195 is active in binding Zn²⁺, which contributes to the protein stability, and substrate cleavage.

DISCUSSION

Essential bacterial cell wall PG cleavage that occurs both in growth and cell division and during phage infection is enabled by Zn²⁺ metallopeptidases. Here, Zn²⁺ metallopeptidase action of the C-terminal domain (CTD) of the phage φ29 gp13 on B. subtilis PG is documented, biochemical characteristics studied and a computational model of the enzyme-substrate complex presented.

The φ29 virion and gp13 demonstrate cross-link-cleaving capacity

The formation of p3 and p7 cleavage products from the cross-link p21 by gp13 and the gp13-containing virion suggests a mechanism for φ29 entry and confirms the multifunctional roles of gp13^{22, 29}. φ29 gp13-defective mutant particles (sus13(342, 330, 53)) are non-infectious, likely because they rapidly lose the packaged DNA^{29, 38} even though the gp13 fragments are in the phage²² (sus13(53) contains gp13 by Western blot analysis [Supplementary Fig. 4]). In addition to this DNA retention function, the enzymatic activity of gp13 on PG cross-links (this study) suggest that the few copies of gp13 per virion^{22, 30} facilitate entry of φ29 through the cross-linked PG layers.

The domain architecture of gp13 is similar to lysostaphin

The PG cross-link-cleaving activity of gp13 resides in the isolated CTD and is independent of the N-terminal domain (NTD) (Fig. 4). In another experiment, particles of the gp13-defective mutant sus13(53), where the CTD of gp13 is truncated and the NTD is intact²², resembled the isolated gp13-NTD; sus13(53) particles were unable to cleave p21 (data not shown). Activity of the CTD is also observed for the PG crosslink-cleaving enzyme LytM of Staphylococcus aureus, where the N-terminus is proposed to be cleaved from the precursor to produce the active enzyme¹³.

The acetylation of the cross-linked muropeptide does not affect cleavage by gp13 (Fig. 3b). The N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM) sugar residues of the PG may be expected to interact with the lysozyme-like NTD of gp13³⁰. A model for enzyme activity of gp13 dimers has been proposed³⁰ where the NTD of one monomer, by binding the NAG-NAM substrate of the PG, aids activity and/or binding of the CTD of the other monomer to a substrate. The absence of dependence on modifications of NAG-NAM for cross-link-cleaving activity of gp13 suggests that either 1) the NTD coordination is independent of acetylation of NAG-NAM, 2) oligomers of NAM-NAG, greater than the two present in the purified muropeptides studied here, are needed for NTD binding or 3) that the NTD does not influence the CTD domain in a coordinated fashion as proposed³⁰. An alternative model could involve the binding of the NTD domain of gp13 that effectively reduces the diffusion coefficient of the gp13 CTD, thereby enhancing the local cleavage of the cross-linked PG in the face of a modest catalysis rate (Supplementary Fig. 3). A similar division of target binding and catalysis between protein domains has been observed for zoocin A, a lysostaphin-like enzyme³⁹.

Dual domain enzymes are known in bacteriophages. As the dual domain endolysins of the phages B30⁴⁰ and φ11⁴¹ enable progeny exit, it is of interest to know if these and other phages also have low copy number virion-associated entry enzymes equivalent in function to gp13 of φ29.

Proposed mechanism of φ29 entry into Bacillus subtilis

Based on current observations and the literature, a model for φ29 entry into B. subtilis is proposed that includes adsorption, PG binding, lysozyme and metallopeptidase mechanisms of PG degradation, membrane puncture and genome injection. Adsorption of the virion to glucosylated teichoic acid surrounding the PG and the likely teichoic acid cleavage by the tail appendages have been discussed previously⁴². Considering PG cross-link degradation, the CTD of gp13 may be activated much as the CTD of LytM is cleaved away from the N-terminus¹³, or alternatively, the connected NTD may aid CTD cleavage by limiting diffusion. The in vitro experiments presented here cannot distinguish these in situ events.

After successful localized reduction in PG cross-linking, the φ29 tail knob gp9 may insert into the cell membrane, triggering conformational changes within the tail axis that deliver injection signals to the head-tail connector to initiate genome expulsion. A portion of genome would be propelled by pressure from the packaged DNA^{43, 44}, and subsequent traction of the remaining DNA would continue⁴⁵.

Zn²⁺ polarization is essential for catalysis in lysostaphin-like enzymes

D195 of gp13 is essential for catalysis and is superimposed on the LytM aspartate 214²². Support for the essentiality of D195 is found in Ale-1 of S. capitis⁴⁶ and MepA of E. coli⁴⁷, where mutation to alanine abrogates function (Ale-1) or renders this expressed protein insoluble and inactive (MepA). However, differentiation between Zn²⁺-binding for secondary structure determination and Zn²⁺ polarization for catalysis has not been discussed in other systems. Here, particle induced X-ray emission (PIXE) demonstrated that D195 of gp13 participates as a Zn²⁺ ligand as suggested by sequence homology²² and the gp13 crystal structure³⁰. Zn²⁺-binding was demonstrated for the Cys substitution mutant (D195C) by PIXE, but the Ala substitution mutant (D195A) showed appreciably reduced Zn²⁺signal (Fig. 7b). Although the D195C protein binds zinc, it is inactive for structural and/or electronic reasons. These data suggest that D195 is sufficiently electronegative to facilitate catalysis. Residues in LytM, lysostaphin, Ale-1 and MepA homologous to gp13 D195 may also provide the necessary action for catalysis.

In the current model of hydrolysis by lysostaphin-like enzymes¹⁹, gp13-D195 would properly bind Zn²⁺, polarizing the carbonyl-carbon of the substrate for susceptibility to nucleophilic attack by a hydroxyl group generated from water by a histidine on the opposite site of the active-site (Fig. 8). A hydroxyl nucleophile generated by His247-mediated deprotonation of water most likely facilitates hydrolysis in this geometry because His247 is 3.5 Å from the carbonyl group and is in the same plane while His278 is 0.5 Å further from the carbonyl group and ~45 degrees off axis compared to His247 (Fig. 6). Similar to the previously established model of VanX³⁵, hydrolytic cleavage of the peptide bond occurs after the formation of a tetrahedral intermediate, which forms a bidendate complex with Zn²⁺ (Fig. 8, step 3). gp13-His247 is homologous to LytM-His260, Ale-1-His200, lysostaphin-His 225^{22, 48} and MepA-His206 (structural superposition, not shown).

Figure 8.

Proposed hydrolysis reaction of the Bacillus subtilis peptidoglycan cross-link p21 catalyzed by the bacteriophage φ29 gene product 13. Step 1, substrate binding to gp13 active-site, Asp195 mediated polarization of the scissile bond carbonyl carbon and deprotonation of water. Step 2, hydroxyl nucleophile attack of the scissile bond carbon leading to the bidentate complex and tetrahedral intermediate. Step 3, proton removal from His247 by the lone pair on the nitrogen of the scissile bond. Step 4 hydrolysis. Step 5 product release.

gp13-like enzymes in other phages

Other phages encode gp13/lysostaphin-like enzymes that may reflect parallel evolution of PG cross-link cleaving, Zn²⁺ metallopeptidases. The lactococcal phages and φ29 family members share the His-x(3,6)-Asp and His-x-His motifs (Fig. 5). As infection by the 10^31 phages⁴⁹ is a frequent genetic transfer event, the likelihood of parallel evolution of entry mechanisms of different phages in conjunction with possible lateral gene transfer from the host warrants further investigation.

METHODS

Phage and protein preparation

Phage φ29 with wild-type (wt) gp13 was prepared using the delayed-lysis fiberless mutant sus8.5(900)14(1241) as described previously²². The gp13 N-terminal domain (NTD), C-terminal domain (CTD), full length and Aspartate 195 mutant gp13 proteins were expressed in either the CodonPlus (Strategene) or the Rosetta (Novagen) strains as previously described³⁰ with modifications. A cocktail of metals (ZnCl₂, 1 μM; MnCl₂, 0.5 μM; CoCl₂, 0.37 μM) was modified from prior work⁵⁰ and included in the induction medium. EDTA was omitted from all buffers and protease inhibitors. The gp13 C-terminal domain (CTD), and the D195A and D195C mutants were expressed in Rosetta. Cells (OD600nm= 0.5, Klett =44) were grown from overnight cultures with chloramphenicol (34 μg/mL) and ampicillin (100 μg/mL) to select for the Rosetta plasmid and the pTYB1 plasmid, respectively. Cultures were chilled to 0°C in a water/ice bath for 5 minutes prior to IPTG induction at 1 mM and incubated at 20°C overnight in a shaking water bath. Cells harvested by centrifugation were resuspended in 1/20 the original volume of HST (20 mM Hepes, 500 mM NaCl, 0.1% (v/v) Triton-X 100, 4°C) modified from the manufacturer’s guidelines (NEB, Ipswich, MA) to omit the Zn²⁺chelating agent EDTA, washed of residual media, and resuspended in 30 mL HST with an EDTA-free protease inhibitor mixture (Complete Mini EDTA-free, Roche, Indianapolis, IN) per the manufacturer’s recommendations (1 tablet per 10 mL concentrated cells). The cell concentrate was lysed by two or three passages through a chilled French press and checked for lysis by phase contrast microscopy. The crude lysate was clarified by centrifugation at 4°C and loaded onto a pre-HST buffer-equilibrated 5–8 mL chitin-resin column (NEB) at 4°C and run at a flow rate not exceeding 0.5 mL/min. Self-cleavage of the intein fusion was induced, after a 20 column volume HST wash to remove non-specifically bound material, with HST + 50 mM DTT by room temperature incubation for 36–48 hrs. HST elution with 10 column volumes permitted sufficient protein recovery for subsequent concentration by Centriprep 30 kDa mwco filters (Millipore). Dialysis against TS (20 mM Tris, pH 7.4, 150 mM NaCl) buffer was performed with a 10 kDa mwco Slide-a-Lyzer (Pierce) at 4°C with two 1L exchanges over 4–8 hrs each. Concentrated and buffer exchanged protein was cryo-protected with the addition of sterile glycerol to 10% final and stored at −80°C. The protein was quantified by SDS-PAGE densitometry with BSA standards.

Pfu polymerase (Stratagene) was used in circular PCR to generate the D195A and D195C mutants by modifying the full-length gp13 plasmid pTYB1³⁰ using standard conditions and primers as follows:

gp13 D195A forward primer 5′GGAACACTTTGTATAGCCTTTGTGGGCAAAACTGAAAAGTACCC3′ gp13

D195A reverse primer 5′GGGTACTTTTCAGTTTTGCCCACAAAGGCTATACAAAGTGTTCC3′

gp13 D195C forward primer 5′GGAACACTTTGTATATGCTTTGTGGGCAAAACTGAAAAGTACCC3′

gp13 D195C reverse primer 5′GGGTACTTTTCAGTTTTGCCCACAAAGCATATACAAAGTGTTCC3′

Peptidoglycan purification and HPLC analysis

PG from vegetative B. subtilis cells was purified and digested with muramidase (Mutanolysin, Sigma-Aldrich) as previously described³¹. Muropeptides were reduced with borohydride and separated by HPLC using a NaPO₄ buffer and a methanol gradient as described^{31, 51}. Muropeptides in enzyme assay samples were analyzed using the same HPLC system without borohydride reduction. Muropeptides to be purified for use as substrates in enzymatic assays or for structural analyses were further purified using HPLC and a trifluoracetic acid/acetonitrile buffer system³². The structures of purified muropeptides were verified using amino acid analysis⁵² and mass spectrometry⁵³ as previously described. Muropeptides (250 pMoles) were incubated with gp13 (100 pMoles) in TS (20 mM Tris, pH 7.4, 150 mM NaCl) buffer for 3 hours at 37C and the reaction was halted by acidification with phosphoric acid added to 5% (v/v). Chromatograms were analyzed using PowerChrom (eDaq, Colorado Springs CO). The raw peak areas were converted to molar equivalence to p3 using calculated extinction coefficients⁵⁴ for muropeptides (p21, p3, p7) and then normalized to mole fraction of total area per run.

Amino acid analysis

Purified putative p7 muropeptide was either modified by fluorodinitrobenzene (FDNB) or left untreated. Samples in 0.1 M NaHCO₃ were incubated with 16 mM FDNB for 16 hrs at room temperature in the dark. Both samples were then dried, acid hydrolyzed, and subjected to amino acid analysis as previously described⁵².

Computational modeling

All modeling and atomic distance measurement was done with the X-ray crystallographic structure of gp13³⁰ using Maestro (Schrödinger, LLC, New York, NY, 2007). The fragment model of the Bacillus subtilis cross-link was built and energy minimized using the OPLS 2005 force field. The docking of the modeled fragment was carried out using the well validated⁵⁵ and widely used Glide (Schrödinger, LLC, New York, NY, 2007) with default extra precision parameters. The Zn²⁺ ion, His247 and His278 side chains were defined as potential hydrogen bond donor/acceptor constraints for the docking.

Spectroscopy

Circular dichroism spectra of gp13 and the mutant proteins D195A and D195C (0.1 mg/mL) was conducted from 195 nm to 260 nm at a data pitch of 0.1 nm, acquiring 100 nm/min, with 4 sec response and 1 nm bandwidth using a 0.1 cm cuvette in a Jasco 815. Five measurements were accumulated per sample. Purified proteins were dialyzed against 10 mM potassium phosphate, pH 7.4, buffer using 10 kDa microdialysis cups (Pierce). Protein concentration was assayed by SDS-PAGE densitometry with BSA standards.

Particle induced X-ray emission (PIXE) was conducted with acetone precipitated protein (400 μg) and filter collected on 1 cm² nylon membranes (Boehringer Mannheim). Briefly, samples were mixed with −20°C acetone to 80% acetone (v/v) and incubated at −20°C for ten min. Visible precipitates were present. Nylon filters with 80% acetone washed samples and dried for at least 10 hrs at 25C were analyzed using a 1.7-MeV Pelletron accelerator (National Electrostatic Corporation) configured to generate a 4.0-MeV He⁺⁺ ion beam, at a beam current of ~30 nA in a circular beam 1.5–2 mm in diameter. Sample-emitted X-rays were counted and energy analyzed using a Kevex Si(Li) X-ray detector with 5-mm Be window and energy resolution of 145 eV; a capping Al foil also was used to reduce the photon flux to the detector to prevent pile-up. PIXE spectra were acquired using the Hypra program (Charles Evans and Associates) during doses of 40 μC or 20 μC, summed to a total of 80 μC per sample. The Zn Kα_1,2 (8.63 keV) line intensity was used to detect and gauge Zn content.

Sequence Alignment

Sequence alignment of LytM [GI:87160448] and gp13 homologs in φ29 family members (φ29 [GI:137932], GA-1 [GI:12248120] and B103[GI:2285516]) was conducted using Clustal W within Align X of the Vector NTI suite (v.10, Invitrogen, Carlsbad, CA). Lactococcal phage related proteins were identified by an ACLAME^{56, 57} search using gp13. The proteins include bIL285 endopeptidase [GI:13095734], TP901-1 “ORF47” [GI:13786578], Tuc2009 “Minor structural protein 3″ [GI:13487849], and ul36 “structural protein” [GI:21716123]. The alignments to lactococcal phages were added to the gp13-LytM alignment and displayed using AlignX.

Abbreviations

aa

Amino acid

NTD

amino-terminal domain

CTD

carboxy-terminal domain

Dpm

diaminopimelic acid

gp

gene product

HPLC

high performance liquid chromatography

NAM

N-acetyl muramic acid

NAG

N-acetyl glucosamine

PIXE

particle induced X-ray emission

PG

peptidoglycan

SDS-PAGE

sodium dodecylsulfate-polyacrylamide gel electrophoresis

sus

suppressor-sensitive