Regulation of a Phage Endolysin by Disulfide Caging

Abstract

In contrast to canonical phage endolysins, which require holin-mediated disruption of the membrane to gain access to attack the cell wall, signal anchor release (SAR) endolysins are secreted by the host sec system, where they accumulate in an inactive form tethered to the membrane by their N-terminal SAR domains. SAR endolysins become activated by various mechanisms upon release from the membrane. In its inactive form, the prototype SAR endolysin, Lyz_P1, of coliphage P1, has an active-site Cys covalently blocked by a disulfide bond; activation involves a disulfide bond isomerization driven by a thiol in the newly released SAR domain, unblocking the active-site Cys. Here, we report that Lyz₁₀₃, the endolysin of Erwinia phage ERA103, is also a SAR endolysin. Although Lyz₁₀₃ does not have a catalytic Cys, genetic evidence suggests that it also is activated by a thiol-disulfide isomerization triggered by a thiol in the SAR domain. In this case, the inhibitory disulfide in nascent Lyz₁₀₃ is formed between cysteine residues flanking a catalytic glutamate, caging the active site. Thus, Lyz_P1 and Lyz₁₀₃ define subclasses of SAR endolysins that differ in the nature of their inhibitory disulfide, and Lyz₁₀₃ is the first enzyme found to be regulated by disulfide bond caging of its active site.

In infections by double-stranded DNA phages, host lysis requires degradation of the peptidoglycan by a phage-encoded endolysin (17). By far the most intensively studied endolysin is the T4 lysozyme E (EC 3.2.1.17), which attacks the glycosidic bonds between GlcNAc and MurNAc in the murein (1). During the latent period, canonical endolysins are produced as fully active enzymes sequestered in the cytoplasm, thereby preventing premature lysis. Another phage protein, the holin, terminates the infection cycle by suddenly forming extremely large, nonspecific holes in the membrane that allow the endolysin to escape and attack the murein layer. Recently, studies of the lysis system of bacteriophage P1 have revealed that phage-encoded endolysins are not always dependent upon holins for export (19, 20). Although it is an ortholog of T4 E, the P1 lysozyme, Lyz_P1, is translocated across the cytoplasmic membrane by the host sec system by virtue of an N-terminal transmembrane domain (TMD) that is absent in E (Fig. 1 A). Since this transmembrane domain is not removed by signal peptidase, nascent Lyz_P1 remains tethered to the membrane with its catalytic residues already present in the periplasm. The Lyz_P1 TMD exits the membrane and becomes part of the soluble, periplasmic form of the protein, either at a low spontaneous rate or, more efficiently, when the holin triggers to depolarize the membrane (11). Because of the unique ability to direct sec-mediated export and membrane insertion and to support release into the periplasm from the bilayer, the TMD of Lyz_P1 was designated a signal anchor release (SAR) domain. More recently, other SAR endolysins have been identified and characterized (15, 16). In fact, bioinformatic analysis suggests that most members of the T4 lysozyme family, recognizable by the Glu-8aa-(Asp/Cys)-5aa-Thr catalytic triad (Fig. 1A), are SAR endolysins (43 of 58 entries in the GenBank protein database) (16).

FIG. 1.

Sequence alignments. (A) Alignment of T4E, Lyz_P1, and Lyz₁₀₃. Catalytic residues are highlighted in blue, Cys residues in pink, and SAR domains in yellow. Locations of Leu substitutions are indicated below the corresponding Gly residue (green). Inactivating/caging disulfides for Lyz_P1 and Lyz₁₀₃, respectively, are shown as arrows connected with a black line above the participating Cys residues. Disulfides resulting from isomerization are shown below the participating Cys residues as arrows connected with a brown line. (B) Alignments of Lyz_P1→103 and Lyz_103→P1 conversion mutants. Catalytic residues are highlighted in blue, Cys residues in pink, and SAR domains in yellow. Locations of Cys substitutions are indicated below the corresponding residue.

Since the lyz_P1 gene is expressed well before progeny P1 phage have been assembled, there must be a mechanism to ensure that the membrane-tethered form of the protein is kept enzymatically inactive so that premature lysis is avoided. A key to the regulation of Lyz_P1 is the fact that the P1 enzyme has a catalytic cysteine residue, Cys₅₁ (Fig. 1A), in the central position of the catalytic triad, in contrast to E and most of its orthologs, which have an Asp residue in this position (20). Genetic, biochemical, and structural analysis of Lyz_P1 demonstrated that the membrane-tethered form is inactive for two reasons: first, the entire catalytic domain is misfolded, so the active-site cleft is completely missing, and second, the catalytic Cys₅₁ is covalently occupied in a disulfide bond with another Cys at position 44. This led to a model for activation in which a thiol (Cys₁₃) present in the SAR domain becomes unmasked upon membrane release and triggers a disulfide bond isomerization, liberating the thiol of the catalytic Cys₅₁. This model was confirmed by crystal structures showing the alternative disulfide linkages in the inactive and active forms of Lyz_P1 (19).

Lyz_P1 became the prototype of a class of SAR endolysins recognizable by the Asp→Cys substitution in the catalytic triad and the presence of activating Cys in the N-terminal SAR domain. However, most SAR endolysins belong to a second major class, represented by R₂₁, the endolysin of the lambdoid phage 21 (20). These enzymes have the canonical Glu-8aa-Asp-5aa-Thr catalytic triad and no Cys residue in the SAR domain. Instead, genetic and structural analysis revealed that in the inactive form, the catalytic domain has nearly the correct fold, except for a displacement of the active-site Glu, but is subject to steric hindrance by the proximity of the bilayer in which the SAR domain is embedded. In the soluble, active form, the SAR domain of R₂₁ has refolded into the main body of the enzyme, providing a floor to the active site and repositioning the catalytic glutamate to its proper place (16). Thus, the R₂₁ regulatory scheme is markedly different from that of Lyz_P1, where the released SAR domain provides only the free thiol for the disulfide bond rearrangement and makes few contacts with the enzyme itself.

Here, we examine the regulation of the endolysin Lyz₁₀₃ of the Erwinia amylovora phage ERA103 (GenBank accession no. EF160123), which seems to have characteristics of both of these major classes: it has a Cys residue in an N-terminal hydrophobic sequence but retains the canonical Asp residue in the catalytic triad. The results are discussed in terms of a model for SAR-dependent disulfide bond isomerization distinct from that of Lyz_P1 and its homologs but which nevertheless confers a covalent constraint on premature activation of the muralytic activity.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The Escherichia coli strains XL1-Blue and MC4100 and the MG1655 ΔfhuA lacI^q lacY mutant strain have been described previously (13, 16). Standard conditions for the growth of cultures and the monitoring of lysis kinetics have also been described previously (3, 13). All bacterial cultures were grown in standard LB medium, supplemented with 100 μg/ml ampicillin when appropriate. When indicated, isopropyl-β-d-thiogalactopyranoside (IPTG), dinitrophenol (DNP), and dithiothreitol (DTT) were added to achieve final concentrations of 1 mM, 10 mM, and 1 mM, respectively.

DNA procedures and plasmid construction.

Procedures for the isolation of plasmid DNA, DNA amplification by PCR, PCR product purification, DNA transformation, site-directed mutagenesis, and DNA sequencing have been previously described (5, 12, 14). The construction of the plasmid pLyz_P1, a derivative of pJF118EH, has been described previously (4, 20). The plasmid pLyz₁₀₃ was constructed by amplifying and inserting lyz₁₀₃ between the EcoRI and HindIII restriction sites of pJF118EH. Similarly, the construct pETLyz₁₀₃ was constructed by amplifying lyz₁₀₃ and inserting it between XbaI and BamHI of pET11a (Novagen). For overexpression purposes, the inactive allele lyz₁₀₃(D52N) was used. Derivative alleles of lyz₁₀₃ and lyz_P1 were made using site-directed mutagenesis. For detection and purification purposes, the lyz₁₀₃ allele was modified to encode an oligo-histidine tag (Gly₂His₆Gly₂) appended to Met₁₇₈ by site-directed mutagenesis. All purified proteins cited in this work refer to the oligo-histidine-tagged versions.

SDS-PAGE and Western blotting.

SDS-PAGE, Western blotting, and immunodetection experiments were performed as previously described (5). Antiserum against Lyz_P1 was prepared in chickens by Aves Labs (Tigard, OR) and was used at a dilution of 1:1,000. A mouse monoclonal antibody against the oligo-histidine epitope tag was purchased from Amersham and was used at a dilution of 1:3,000. Horseradish peroxidase-conjugated secondary antibodies against chicken IgY were purchased from Aves Labs and were used at a 1:2,000 dilution for colorimetric detection and a 1:300,000 dilution for chemiluminescent detection. The anti-mouse IgG horseradish peroxidase-conjugated secondary antibody was supplied with the SuperSignal chemiluminescence kit (Pierce) and was used at a 1:5,000 dilution. Blots were developed using the chromogenic substrate 4-chloro-1-naphthol (Sigma) or with the West Femto SuperSignal chemiluminescence kit (Pierce). Chemiluminescent signal was detected using a Bio-Rad ChemiDoc XRS.

Subcellular fractionation.

Soluble or membrane localization was determined as described previously (16, 20). Briefly, 25 ml of an induced culture was collected by centrifugation at 5,000 × g in a Sorvall Superspeed RC2-B centrifuge and resuspended in 2 ml of French press buffer (0.1 M sodium phosphate, 0.1 M KCl, 5 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, pH 7.0). Cells were disrupted by passage through a French pressure cell (Spectronic Instruments, Rochester, NY) at 16,000 lb/in². The membrane and soluble fractions were separated by centrifugation at 100,000 × g in a Beckman TL-100 ultracentrifuge for 60 min. Equivalent amounts of each fraction were examined by SDS-PAGE and Western blotting as described above.

Sulfhydryl modification using PEG-OPSS.

To detect the presence of free cysteines in SAR endolysins, a 5-ml sample of a culture induced in logarithmic phase for 25 min was precipitated by trichloroacetic acid (TCA). The pellet was resuspended in 1 ml of PEGylation buffer (500 mM Tris [pH 7.0], 1% SDS, 1 mM EDTA), and mPEG-OPSS (Nektar Transforming Therapeutics, Huntsville, AL) (8) was added to produce a final concentration of 3 μM. The mixture was incubated for 30 min at room temperature and then precipitated by the addition of 1 ml of ice-cold acetone. The samples were held at −20°C for 10 min, after which the precipitate was collected by centrifugation at 18,000 × g at 4°C for 15 min. The pellets were air dried, resuspended in nonreducing SDS sample buffer, and examined by SDS-PAGE and Western blotting. All samples were run with controls that had not been exposed to PEG-OPSS.

Lyz₁₀₃ expression and purification.

Since wild-type Lyz₁₀₃ lyses cells rapidly when overexpressed, the enzyme activity was abolished by replacing the catalytic Asp with Asn. Hence, all purified Lyz₁₀₃ cited in this work refers to Lyz₁₀₃(D52N). pET lyz₁₀₃(D52N) cHis was transformed into BL21(DE3) cells (Invitrogen) harboring pLysS, and fresh transformants were cultured and induced for 1 h at 30°C. Cells were collected at 4,000 rpm for 30 min at 4°C in a Sorvall RC-3B centrifuge and resuspended in Lyz₁₀₃ buffer (20 mM Tris-HCl [pH 8], 100 mM NaCl). Protease inhibitor cocktail for His-tagged protein (Sigma) was added as per the manufacturer's instructions, and cells were lysed by passage through a French pressure cell (Spectronic Instruments, Rochester, NY) at 20,000 lb/in². After unlysed cells and debris were removed, the lysate was filtered through a 0.2-μm syringe filter. The cleared lysate was then applied to Talon metal affinity resin (Clontech). Protein was eluted in elution buffer (20 mM Tris-HCl, 100 mM NaCl, 500 mM imidazole, pH 8) and was used with no further purification.

CDAP cleavage.

Purified Lyz₁₀₃(D52N) was precipitated and resuspended in 100 μl CDAP (1-cyano-4-dimethylaminopyridinium tetrafluoroborate) buffer (4 M guanidine-HCl, 0.1 M citrate, pH 3). CDAP (Sigma), prepared fresh in CDAP buffer, was added in 1,000-fold molar excess and was incubated at room temperature for 15 min. NH₄OH (EMD Chemicals) was added to achieve a 1 M final concentration, and the mixture was incubated at room temperature for 3 h (10, 18). Total protein was precipitated by the addition of 1 ml of ice-cold ethanol and incubated overnight at −20°C. Samples were pelleted, dried, and resuspended in sample loading buffer with or without 5% β-mercaptoethanol, as indicated. Equivalent amounts were analyzed by SDS-PAGE and Western blotting.

RESULTS AND DISCUSSION

The lysozyme from bacteriophage ERA103 has an N-terminal SAR domain.

We first wished to confirm that the N-terminal hydrophobic domain of Lyz₁₀₃ is a SAR domain. Induction of lyz₁₀₃ resulted in lysis of Escherichia coli, even without a holin gene present, indicating a spontaneous release from the membrane and consequent lysis (Fig. 2 A). Moreover, like Lyz_P1, Lyz₁₀₃ was found in both the soluble and membrane fractions of cells (Fig. 3 A). For both proteins, the membrane-associated and soluble forms migrate identically in SDS-PAGE, indicating that the latter is not derived by the proteolytic cleavage of the former. Additionally, energy poisons such as DNP accelerated the lysis of cultures expressing lyz₁₀₃ (Fig. 2A), indicating that the collapse of the proton motive force facilitates the membrane release and activation of Lyz₁₀₃, as it does for Lyz_P1 (20). The SAR domains characterized in Lyz_P1 and R₂₁ differ from conventional TMDs in that they have a high content of weakly hydrophobic and uncharged polar residues, such as Ala, Gly, and Ser (20). In both cases, substitution of Leu residues for Gly residues (3 in Lyz_P1 [Fig. 2, 3C] and 2 in R₂₁ [16]) in the SAR domain blocked the release from the membrane and host lysis. The same molecular and cellular phenotype was observed for Lyz₁₀₃ when Gly residues at positions 9 and 10 in the SAR domain were converted to Leu (Fig. 2C, 3D). These results are consistent with the requirement for the SAR domain to exit the membrane, liberating the thiol in the SAR domain to attack the inhibitory disulfide. Thus, Lyz₁₀₃ is the third SAR endolysin to be characterized in terms of its physiological and topological characteristics. The predicted amino acid sequence of Lyz₁₀₃ is identical to that of the lysozyme from bacteriophage φEa1h. Thus, the puzzling lethality observed when lyz^φEa1h was expressed in the absence of its holin (6, 7) is due to the fact that, as a SAR endolysin, it reaches the periplasm by a holin-independent mechanism.

FIG. 2.

In each experiment, cultures of XL1-Blue containing Lyz_P1 or Lyz₁₀₃ in pJF118EH were induced at time zero. The optical density of the culture was followed as a function of time in minutes. (A) Expression of Lyz₁₀₃ lyses the cell independently of a holin. ○, Lyz_P1; ▪, Lyz₁₀₃; □, Lyz₁₀₃ plus 10 mM DNP 20 min (arrow) after induction. (B) Lysis by Lyz₁₀₃ is dependent on the SAR domain Cys. ▪, Lyz₁₀₃(C12S); □, Lyz₁₀₃(C12S) plus 1 mM DTT at 36 min (arrow) after induction; ⧫, Lyz₁₀₃(C12,42,45S). (C) Lyz_P1 and Lyz₁₀₃ can be locked into the membrane by Leu titration into the SAR domain. ○, Lyz₁₀₃(G9,10L); ▪, Lyz_P1(G8,9,10L). (D) Lyz_P1 and Lyz₁₀₃ can be interconverted. Lyz_P1→103 was achieved by two amino acid mutations: A41C and C51D (•). Lyz_103→P1 was achieved by two amino acid mutations: C42S and D52C (□). ○, Lyz_P1→103(C13S); ▪, Lyz_P1→103(C13S) plus 1 mM DTT at 45 min (arrow); ⧫, Lyz_103→P1(C12S). (E) There is optimal positioning of the SAR domain Cys. Cys substitutions are labeled adjacent to their respective curves. •, Lyz_P1→103(G9C); ▪, Lyz_P1→103(G10C); ○, Lyz_P1→103(A11C); ⧫, Lyz_P1→103(I12C); ▵, Lyz_P1→103; □, Lyz_P1→103(A14C); ⋄, Lyz_P1→103(I15C); ▴, Lyz_P1→103(A16C). (F) The placement of the caging disulfide is stringent. •, Lyz_P1(I38C, C51D); ○, Lyz_P1(G39C, C51D); ▪, Lyz_P1(N40C, C51D); □, Lyz_P1(C13S, I38C, C51D); ⧫, Lyz_P1(C13S, G39C, C51D); ⋄, Lyz_P1(C13S, N40C, C51D).

FIG. 3.

The endolysins of bacteriophages P1 and ERA103 are SAR endolysins. For all panels, the total protein is represented by lane 1, the soluble protein fraction is represented by lane 2, and the membrane fraction is represented by lane 3. (A) Lyz₁₀₃ exists in the soluble and membrane fractions of the cell. (B) Lyz₁₀₃(C12S) displays a nonlytic phenotype but is released from the cytoplasmic membrane as efficiently as the wild-type protein. (C) Lyz_P1 can be confined to the inner membrane by replacing three glycines (residues 8 to 10) with leucine residues. (D) Lyz₁₀₃ can be confined to the inner membrane by replacing two glycines (residues 9 and 10) with leucine residues.

The activity of Lyz₁₀₃ is regulated by a disulfide “cage.”

Lyz₁₀₃, like T4 E but unlike Lyz_P1, has an Asp residue as the middle component of its catalytic triad and thus is not regulated by covalent blocking of an active-site Cys (Fig. 1A). Nonetheless, the N-terminal catalytic domain of Lyz₁₀₃ contains three Cys residues (Cys₁₂, Cys₄₂, Cys₄₅), including one in the SAR domain, suggesting that it might exist in two isomeric forms differing in the arrangement of intramolecular disulfide bonds, as documented for Lyz_P1. To test this possibility, the C12S mutant of Lyz₁₀₃ was tested for function. This mutant was found to be lytically inactive (Fig. 2B), although it was released from the inner membrane as efficiently as the wild-type protein (Fig. 3B). Moreover, the addition of the reducing agent, dithiothreitol, to cells expressing the C12S mutant resulted in lysis (Fig. 2B). Finally, the triple mutant, Lyz₁₀₃(C12,42,45S), was found to be lytically active (Fig. 2B). The behavior of these two mutants is consistent with the model that nascent Lyz₁₀₃ is inactive because of an inhibitory Cys₄₂-Cys₄₅ disulfide that is disrupted by Cys₁₂ after the release of the SAR domain from the membrane. Since the inhibitory disulfide predicted to exist in nascent Lyz₁₀₃ involves cysteines that flank an essential catalytic residue (Fig. 1A), we refer to it as a “caging” disulfide to distinguish it from the inactivating disulfide present in Lyz_P1. It is of interest to note that a similar inactive, disulfide-caged form of T4 gpe was constructed 21 years ago (9). In this form of T4 gpe, the active-site residues are unaltered but an engineered disulfide occludes the active site. As shown here with Lyz₁₀₃ and earlier with Lyz_P1, the disulfide-caged form of T4 gpe could be activated by reducing agents.

Disulfide bond isomerization.

The results presented above indicate that the Cys₄₂-Cys₄₅ disulfide cages the catalytic Glu₄₃ in inactive Lyz₁₀₃ and that the Cys₁₂ thiol from the SAR domain is required for activation, but the results do not reveal whether one of the two Cys residues flanking Glu₄₃ participates in a disulfide exchange with Cys₁₂ and, if so, which one (Fig. 1A). To address this question, we used the sulfhydryl cyanylation reagent 1-cyano-4-dimethylaminopyridinium tetrafluoroborate (CDAP). Treatment of a CDAP-cyanylated protein with a strong base cleaves the protein on the N-terminal side of the cyanylated Cys, leaving an N-terminal 2-iminothiazolidine-4-carboxylyl (itz) modification on the C-terminal product (10, 18). If activated Lyz₁₀₃ has a disulfide between Cys₁₂ and Cys₄₂, Cys₄₅ will be open for CDAP modification and subsequent cleavage would create two polypeptide fragments which could be resolved from uncleaved protein without the presence of a reducing agent (Fig. 4 A). In contrast, a linkage between Cys₁₂ and Cys₄₅ will leave Cys₄₂ open for modification. In this case, cleavage with a strong base would result in polypeptide fragments that were still covalently linked and which would require treatment with a reducing agent to be resolved from uncleaved material (Fig. 4B). When purified Lyz₁₀₃ was subjected to CDAP treatment and alkaline cleavage and analyzed by Western blotting with anti-oligo-histidine tag antibodies, a polypeptide product of an appropriate size (16 kDa) could be resolved from uncleaved protein (20 kDa) by SDS-PAGE without a reducing agent (Fig. 4C). The yield of cleavage product was low, approximately 10 to 15% of total protein. However, treatment of the CDAP-modified product with PEG-OPSS revealed that >60% of the protein was cyanylated by CDAP and therefore unPEGylated (data not shown). This indicates that the cleavage reaction, rather than sulfhydryl accessibility, was limiting, presumably due to the high incidence of the competing β-elimination reaction, as noted elsewhere (18). Taken together, these results indicate that the Lyz₁₀₃-activating linkage resulting from the disulfide bond isomerization after SAR extraction is between Cys₁₂ and Cys₄₂ rather than between Cys₁₂ and Cys₄₅, as predicted from the Lyz_P1 and Lyz₁₀₃ sequence alignments (Fig. 1A). Thus, the position of the disulfide bond exchange is not conserved.

FIG. 4.

Analysis of the cage-relieving disulfide. Purified Lyz₁₀₃(D52N) cHis was subjected to CDAP cyanylation and alkaline cleavage. Proteins were resolved by SDS-PAGE with and without a reducing agent, as indicated. (A) A cage-relieving disulfide between Cys₁₂ and Cys₄₂ will allow Cys₄₅ to be modified by CDAP. Cleavage of this protein results in a C-terminal cleavage fragment that is resolvable without the addition of a reducing agent (β-Me). (B) A cage-relieving disulfide between Cys₁₂ and Cys₄₅ results in a cleavage product that is resolved only upon the addition of β-Me. (C) Treatment of Lyz₁₀₃(D52N) with 1,000-fold molar excess of CDAP reagent resulted in a resolvable C-terminal cleavage product (16 kDa), indicating the formation of a Cys₁₂-Cys₄₂ disulfide.

The Lyz_P1 and Lyz₁₀₃ regulatory schemes are interconvertible.

Comparing the crystal structures of the active and an inactive form of Lyz_P1 demonstrates that much of the N terminus of the protein is capable of adopting markedly different conformations depending upon the placement of a single intramolecular disulfide (19). To explore the structural malleability of the N-terminal catalytic domain of Lyz_P1 (Glu₄₂, Cys₅₁, and Thr₅₇), we attempted to convert Lyz_P1, in which the inactivating disulfide covalently blocks a catalytic Cys, into a Lyz₁₀₃-type endolysin, with a caging disulfide sequestering a catalytic Glu, by introducing the A41C and C51D mutations (Fig. 1B, Table 1). This mutant, Lyz_P1→103, was found to be lytically active, although lysis was delayed and more gradual compared with that of the wild-type enzyme (Fig. 2D). As seen with Lyz_P1, removal of the SAR domain Cys in Lyz_P1→103(C13S) rendered the enzyme dependent upon the addition of an exogenous reductant (Fig. 2D). Moreover, introducing the C42S and D52C mutations into Lyz₁₀₃ (Fig. 1B) converted the enzyme into a Lyz_P1-type endolysin that still required the presence of the Cys residue in its SAR domain for lytic function (Fig. 2D). Thus, with regard to the nature of the inhibitory disulfide and the use of a catalytic Cys or Asp, Lyz_P1 and Lyz₁₀₃ are fully interconvertible. Since we had polyclonal antibodies at hand for Lyz_P1 which were much more efficient than the antibodies against the oligo-histidine tag for Lyz₁₀₃, the Lyz_P1→103 construct was selected for further analysis.

TABLE 1.

Catalytic triads, cysteines participating in enzyme regulation, mutations required for the Lyz_P1→103 and Lyz_103→P1 conversions, and approximate lysis times of Lyz_P1, Lyz₁₀₃, Lyz_P1→103, and Lyz_103→P1^a

Lysozyme	Catalytic triad	Regulating cysteines	Conversion mutations	Lysis time (min)
LyzP1	E42, C51, T57	C13 (SAR), C44, C51 (catalytic)	NA	15
Lyz103	E43, D52, T58	C12 (SAR), C42, C44	NA	35
LyzP1→103	E42, D51, T57	C13 (SAR), C41, C44	A41C, C51D	50
Lyz103→P1	E43, C52, T58	C12 (SAR), C44, C52 (catalytic)	C42S, D52C	45

^aCysteines located in the SAR domain and catalytic cysteines are indicated. NA, not applicable.

Optimal positioning of the SAR Cys residue.

Since we had previously demonstrated that the in vivo activity of Lyz_P1 depended upon the placement of the Cys₁₃ residue in its SAR domain (19), we reasoned that the poor lytic profile obtained with the Lyz_P1→103 protein might be due to the suboptimal positioning of this critical residue. To test this notion, constructs were generated in which the Cys₁₃ residue in the SAR domain was moved to eight new positions, four in each direction, from the parental site in the context of Lyz_P1→103. The results validated the hypothesis, in that earlier and sharper lysis profiles were obtained with a Cys occupying positions 12, 10, 14, and 16, in order of apparent lytic activity (Fig. 2E). This pattern is different from that observed using the equivalent substitutions in Lyz_P1, where the Cys₁₃ form is the most active and moving the thiol to position 12 eliminated the ability to activate (19). These differences could be due to a combination of factors, since the activation of Lyz_P1 involves major structural changes within the N-terminal catalytic domain and all of the mutations necessary to convert Lyz_P1 into Lyz_P1→103 and its derivatives occurred within this domain.

Strict positional requirement for the inhibitory disulfide cage.

We next determined the effect of moving the position of the first Cys (Cys₄₁) of the inhibitory disulfide cage in Lyz_P1→103 toward the N terminus of the protein (Fig. 1B). All three variants (N40C, G39C, and I38C) tested were lytic (Fig. 2F). Surprisingly, the lethality of these proteins was not completely dependent upon the presence of a Cys in the SAR domain, as lysis still occurs, although much later and more gradually (Fig. 2F). The simplest explanation for this finding is that disulfides between Cys₃₈, Cys₃₉, or Cys₄₀ and Cys₄₄ do not effectively cage the active-site glutamate. We directly assessed the status of the cysteines in Lyz_P1→103, Lyz_P1→103(C13S), and Lyz_P1(C13S, I38C, C51D), the most lytically active cage mutant, by testing their susceptibility to modification by PEG-OPSS. PEG-OPSS has a PEG 5000 moiety affixed to an orthopyridyl disulfide. When reacted with a free sulfhydryl, PEG-OPSS will covalently bind and will shift the apparent molecular weight of the protein (8). As expected, the active endolysin, Lyz_P1→103, was found to have a single free cysteine (Cys₁₃), while the inactive enzyme, Lyz_P1→103(C13S), had none (Fig. 5). The lytically active Lyz_P1(C13S, I38C, C51D) protein was not modified with PEG-OPSS, indicating that all six of its cysteine residues were present in disulfide linkages. Thus, while the Cys₃₈-Cys₄₄ disulfide forms, it is not inhibitory.

FIG. 5.

Assessment of the status of cysteine residues. Lyz_P1→103 is an active endolysin and has a single free sulfhydryl. It is modified by PEG-OPSS (lane 1). Lyz_P1→103(C13S) has no free cysteine residues and is not modified by PEG-OPSS (lane 2). Lyz_P1(C13S, I38C, C51D) is an active endolysin despite the presence of the caging disulfide (lane 3). As confirmation of the presence of this disulfide, the protein is not modified by PEG-OPSS, indicating that all of its cysteines are in disulfide bonds. Lanes 1 to 6 contain unPEGylated Lyz_P1→103, Lyz_P1(C13S), and Lyz_P1(C13S, I38C, C51D), respectively.

Conclusion.

Two modes of SAR endolysin negative regulation have previously been experimentally established. The first is that of an inactivating disulfide involving a catalytic Cys, as is present in Lyz_P1 (19); the presence of the disulfide not only precludes participation of the catalytic Cys but also stabilizes a drastically misfolded N-terminal domain. The second mode was found in R₂₁, in which the inactive protein has more subtle folding defects, but the active site is compromised by the proximity of the membrane to which it is tethered by the embedded SAR domain. The results presented here demonstrate that Lyz₁₀₃ represents a third distinct mode of SAR endolysin regulation: disulfide bond caging of the active site in the inactive form of the enzyme (Fig. 6). Despite these differences in the regulatory mode, however, the two proteins appear to be completely interconvertible by simply repositioning the regulatory Cys residues and adjusting the position of the activating Cys residue in the SAR domain. These findings serve to emphasize both the plasticity of the SAR regulatory system and the importance of regulating the timing of lysis in the infection cycle.

FIG. 6.

Model for Lyz₁₀₃ activation. In its inactive form, Lyz₁₀₃ is tethered to the charged cytoplasmic membrane. A caging disulfide exists between Cys₄₂ (blue) and Cys₄₅ (green) that sequesters the catalytic Glu from the remaining members of the catalytic triad. Upon SAR domain (hatched bar) release from the membrane, the SAR thiol, Cys₁₂ (red), attacks the caging disulfide at Cys₄₂, forming a cage-relieving disulfide. The catalytic Glu is released from the cage and can form the catalytic cleft.

Finally, it should be noted that another instance of disulfide bonding affecting the steric blockage of an enzyme active site has been reported in NADP-malate dehydrogenase from chloroplasts (2). In this case, intramolecular steric regulation is achieved when a C-terminal helix, positioned and stabilized by a disulfide bond, occupies the active site and forms hydrogen bonds with catalytic residues. The negative regulation is released when the conserved disulfide is reduced by thioredoxin in response to light stimulation. A key difference is that the Lyz₁₀₃ activation, driven by the release of the SAR domain from the bilayer, is almost certainly irreversible, and thus the stabilities of the inactive and active states are unlikely to be comparable.