The T4 RI Antiholin Has an N-Terminal Signal Anchor Release Domain That Targets It for Degradation by DegP

The bacterial strains, bacteriophages, and plasmids used in this work are described in Tables 1 and 2. Growth and lysis of cultures were monitored at A ₅₅₀. T4 phage stocks were prepared as described previously (23). Bacterial cultures were grown in standard LB medium supplemented with ampicillin (Amp) (100 μg/ml), kanamycin (Kan) (40 μg/ml), or chloramphenicol (Cam) (10 μg/ml) when appropriate. When indicated, isopropyl β-d-thiogalactoside (IPTG), NaN₃, or CHCl₃ was added to give final concentrations of 1 mM, 1 mM, or 1%, respectively.

Isolation of plasmid DNA, DNA amplification by PCR, DNA transformation, and DNA sequencing were performed as previously described (27). Oligonucleotides for PCR were obtained from Integrated DNA Technologies, Coralville, IA, and were used without further purification; sequences of the oligonucleotides used are available on request. Enzymes were purchased from New England Biolabs, except for Pfu polymerase, which was from Stratagene. Automated fluorescent sequencing was performed at the Laboratory for Plant Genome Technology at the Texas Agricultural Experiment Station. Site-directed mutagenesis of plasmids was performed as described previously (27).

Subcellular fractionation, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and Western blotting were performed as described previously (27). Briefly, to establish whether a protein was a membrane-associated or soluble species, cultures were chilled and concentrated ∼150:1 by centrifugation and disrupted by use of a French press. The lysates were cleared of debris by minicentrifuge centrifugation and separated into membrane and soluble fractions by ultracentrifugation at 100,000 × g. To establish periplasmic localization, cultures were concentrated 120:1 by centrifugation and incubated in 12.5% sucrose, 30 mM Tris-HCl, pH 8.0 supplemented with 200 μg of lysozyme and 1 mM EDTA. When ∼95% of cells were converted to spheroplasts, the samples were centrifuged at 9,000 × g for 10 min to separate the periplasm from the spheroplasts (membrane and cytosol).

A rabbit polyclonal antiserum against polypeptide AQSYARIMNIKLETE of RI was obtained from Sigma-Genosys. The blocking solution for Western blotting was a 1:16 mixture of 1% gelatin (Difco) with 1% bovine serum albumin (ICN) in Tris-buffered saline (150 mM NaCl and 10 mM Tris [pH 7.7]).

All DNA manipulations were done by standard techniques described previously (27). Empty vector pZE12-Δluc was constructed by deleting the luc insert between the KpnI and XbaI sites and ligating the backbone. Plasmid pZE12-500RI was constructed by inserting T4 DNA from nucleotide (nt) 58681 to nt 60000 between the KpnI and XbaI sites of pZE12, and plasmid pZE12-500ΔRI was constructed by deleting the rI sequence from pZE12-500RI.

The codons for Ala7, Thr6, and Ala5 of RI were changed sequentially to leucine codons (Fig. 1C) by use of a QuikChange kit from Stratagene with pZA-RI (27) as the template, generating plasmids pZA-RI^L, pZA-RI^2L, and pZA-RI^3L, respectively. The KpnI-AvrII fragments encoding these rI derivatives were transferred to the pZE-12 vector by ligation into similarly digested pZE12-luc, yielding pZE12-RI^L, pZE12-RI^2L, and pZE12-RI^3L. The same approach was used to generate pZE12-RI_TMDΦPhoA from the starting plasmids, pZA-RI_TMDΦPhoA and pZE12-luc.

T4ΔrI phage was made by homologous recombination between pZE12-500ΔRI and T4D. Plasmid pZE12-500ΔRI was transformed into MDS12 tonA::Tn10 lacI ^q, and the transformants were grown to an A ₅₅₀ of ∼0.4. The culture was infected with T4D at a multiplicity of infection (MOI) of ∼10 and was allowed to grow for an additional 3 h. The infected cells were lysed using CHCl₃, and the T4ΔrI in the lysate was enriched three times as follows. A culture of MDS12 tonA::Tn10 lacI ^q at an A ₅₅₀ of ∼0.3 was infected with the lysate containing T4ΔrI at an MOI of ∼0.1. Five minutes later, the infected culture was superinfected with T423_am at an MOI of ∼10 and incubated for an additional 30 min. The infected culture was rapidly filtered through a 0.22-μm syringe filter, and the filtrate was used for the next round of enrichment. At the end of the third enrichment, the filtrate was screened for phage that produced large plaques indicative of the rI deletion. The presence of the deletion in the phages recovered from the large plaques was confirmed by PCR.

Fresh transformants of strains harboring either pZE12-RI, pZE12-^ssPhoAΦRI_CTD, or pZE12-Δluc were grown in 25 ml of LB-Amp to an A ₅₅₀ of ∼0.2 and split into three 7-ml aliquots. Each aliquot was diluted with LB-Amp to make a 25-ml culture. The cultures were grown until reaching an A ₅₅₀ of ∼0.3 and induced with IPTG for 10 min. One of these cultures was left uninfected, while the other two were infected with either T4D or T4ΔrI at an MOI of ∼10, and the culture growth was monitored at A ₅₅₀.

The host strain harboring the plasmid with the indicated allele of rI was grown in 30 ml of LB-Amp to an A ₅₅₀ of ∼0.4. The culture was then split into two 15-ml aliquots, and each aliquot was diluted to 60 ml with LB-Amp. These cultures were grown until reaching an A ₅₅₀ of ∼0.4 and induced with IPTG for 1 h. Protein synthesis in one of the cultures was inhibited using Cam at a 300-μg/ml final concentration. At ∼1.5-min intervals for 10 min after the addition of Cam, 5 ml from each culture was taken and added to 5 ml of 10% ice-cold trichloroacetic acid (TCA). The TCA precipitates were collected by centrifugation and subjected to SDS-PAGE and Western blotting using anti-RI antibodies. The intensities of the bands corresponding to RI were analyzed using the Image J program from the NIH website (http://rsb.info.nih.gov/ij/ ).

MDS12 tonA::Tn10 lacI ^q was grown to an A ₅₅₀ of ∼0.4 and infected with either T4D or T4ΔrI at an MOI of ∼10. Fifteen minutes after infection with T4ΔrI, Cam (in 95% ethanol) was added to a final concentration of 300 μg/ml. As a control, an equivalent volume of 95% ethanol was added to a parallel culture. To cultures infected with T4D, the same additions were made 30 min after infection. For the T4D-infected cultures, total and free phage were determined at 1 h after infection. Total phage was assessed by determining the titer of a CHCl₃-treated aliquot of the infected culture. A second aliquot was rapidly filtered through a 0.22-μm filter to obtain free phage. The ratio of PFU in the filtrate to that in the CHCl₃-treated sample was used as the fraction of total phage that had been released.

Fresh transformants of MG1655 tonA::Tn10 lacI ^q harboring pQ and pT4T, with or without pZA-^ssPhoAΦRI_CTD, were grown at 37°C until reaching an A ₅₅₀ of ∼0.2. The cultures were split into three 7-ml aliquots and diluted with LB containing appropriate antibiotics to a final volume of 25 ml. The cultures were grown to an A ₅₅₀ of ∼0.2, and NaN₃ was added to one aliquot. Ten minutes after the addition of NaN₃, IPTG was added, and the A ₅₅₀ was followed until lysis occurred.

The fact that sec-mediated secretion of the C-terminal periplasmic domain of RI (RI_CTD) was both necessary and sufficient to block lysis mediated by the T4 holin (27) suggested that the NTD, RI_NTD, might be a cleavable signal sequence, as suggested by Paddison et al. (19) and as predicted, at a 99.8% probability, by SignalP analysis (Fig. 1A). However, in the course of these studies, we noted that the chimeric protein RI_NTDΦPhoA (Fig. 1B) existed in both soluble and membrane-associated forms indistinguishable by SDS-PAGE (27). Upon further examination, we found that a considerable fraction of the soluble portion is located in the periplasm (Fig. 2A, lane 7). This species had a mass similar to that of the unprocessed form of PhoA, which accumulates during the inhibition of the sec-mediated translocation pathway by sodium azide (Fig. 2A, lane 2). Thus, by several criteria, RI_NTDΦPhoA behaves like the endolysin of bacteriophage P1 Lyz, suggesting that the RI_NTD can function as a SAR domain. The N-terminal SAR domain of Lyz directs its sec-mediated integration into the cytoplasmic membrane and allows its subsequent release into the periplasm independent of proteolytic processing. Using antisera prepared against a C-terminal peptide of RI revealed that the RI protein itself was present in both the soluble and membrane fractions of cells induced for a plasmid-borne rI gene (Fig. 2B, lanes 2 and 3). Similarly, the soluble RI protein appeared to be periplasmic (Fig. 2C, lane 3). Since the masses of the soluble and membrane bound forms of RI were indistinguishable (Fig. 2B and C), the former was not generated from the latter by proteolysis. As a control, a Western blot of the samples from cells expressing FtsI^cmyc was probed with the antibodies for the c-myc tag; this inner membrane protein was detected only in the membrane and spheroplast fractions (not shown). The ability of both RI_NTDΦPhoA and RI to distribute between the membrane and periplasmic fractions strongly argues that the NTD of RI functions as a SAR domain, instead of the cleavable signal sequence predicted by SignalP.

The SAR domains of P1 Lyz and related endolysins have a high content of weakly hydrophobic and uncharged polar residues compared to conventional TMDs. This suggested that this compositional feature is one factor allowing SAR domains to exit the membrane. For this reason, we tested the effect of increasing the hydrophobicity of the SAR domain of RI on its ability to distribute in the soluble fraction and on its function as an antiholin. The progressive substitution of leucines for the Ala/Thr residues at positions 5 to 7 of RI (RI^L, RI^2L, and RI^3L; Fig. 1C) had the effect of increasing the fraction of RI that was membrane associated upon the separation of cellular contents into soluble and membrane fractions (Fig. 2B). Moreover, increasing the hydrophobicity of RI_NTD resulted in decreasing the ability of RI to inhibit T-mediated lysis (Fig. 3A).

Previously, we had observed that when the periplasmic C-terminal domain of RI (RI_CTD) was fused to the cleavable signal sequence of alkaline phosphatase (PhoA), the resulting chimera, ^ssPhoAΦRI_CTD (Fig. 1B), was more effective at eliciting LIN than RI itself (see the work of Tran et al. [27]). Induction of the ^ss phoAΦrI_CTD gene resulted in the accumulation of two anti-RI-reactive species (Fig. 2D, lane 5). The larger product had a mass approximately equal to that of RI (Fig. 2D, lane 2), while the mass of the smaller product was approximately the same as that of a polypeptide fragment comprising residues 25 to 97 of RI (Fig. 2D, lane 1). During subcellular fractionation, the larger product was spheroplast associated (Fig. 2D, lane 6), while the smaller product was periplasmic (Fig. 2D, lane 7) and, moreover, was absent from cultures treated with azide (Fig. 2D, lane 12). Since azide at 1 mM is known to inhibit the sec translocation pathway (6, 18), it is clear that the smaller product detected after induction of the ^ss phoAΦrI_CTD gene represents the soluble, periplasmic RI_CTD released by cleavage of the ^ssPhoAΦRI_CTD precursor by signal peptidase. The full-length ^ssPhoAΦRI_CTD product seen in the presence and absence of azide probably represents protein trapped in the cytoplasm. Since azide treatment also eliminated LIN (Fig. 3C), it was the periplasmic RI_CTD and not the precursor that inhibited the T holin.

At a low multiplicity of superinfection, LIN is transient and requires continued reinfection for its maintenance. This could be explained if RI were an unstable protein that is rapidly degraded in the absence of a signal generated by repeated superinfections. As one test of this hypothesis, we determined the half-life of RI protein. In this experiment, the rI gene under the control of the lac promoter was induced by the addition of IPTG, and 1 hour later Cam was added to block further protein synthesis. The level of RI present as a function of time after the addition of Cam was assessed by SDS-PAGE and Western blotting. In this case, the RI protein rapidly disappeared (Fig. 4A) after the addition of the protein synthesis inhibitor, showing a half-life of ∼2 min (Fig. 5). Similar results were obtained when RI protein synthesis was induced from the arabinose-inducible plasmid pBAD-RI and then repressed with fucose (not shown).

It was not possible to measure RI stability in T4-infected cells, since the level of RI produced was below the limits of detection with our antibody, even when 5 ml of T4-infected culture at an A ₅₅₀ of ∼0.5 was used for SDS-PAGE and Western blotting. Instead, to provide evidence that the continued synthesis of RI is necessary to maintain LIN during a T4 infection, we added Cam to cultures infected with wild-type T4. This treatment resulted in a gradual decrease in the culture A ₅₅₀ beginning immediately after the addition of Cam (Fig. 6A). After 1 h of infection, nearly 80% of the total phage present in the treated culture had been released to the medium, compared to 6% released in a parallel culture that was not treated with Cam (Fig. 6B). The ability of Cam to partially abrogate LIN is consistent with the rapid turnover of RI during phage infection.

When cultures expressing the ^ss phoAΦrI_CTD gene were treated with Cam, the periplasmic RI_CTD polypeptide was found to be relatively stable (Fig. 4B and 5). Moreover, the membrane-restricted RI^3L protein also had a half-life considerably longer than that of RI (Fig. 4C and 5). These data indicate that the degradation of RI occurred after its release from the membrane and was promoted by the exposure of its SAR domain to the periplasm. Additional evidence for the destabilization of RI by its SAR domain was obtained by complementing the LIN-defective phage, T4ΔrI, with plasmids carrying either the rI or the ^ss phoAΦrI_CTD gene. In these experiments, the plasmids were induced 10 minutes before infection with T4ΔrI. Since T4 infection results in the rapid destruction of host DNA (16), synthesis of RI or ^ssPhoAΦRI_CTD should cease shortly thereafter. Attempts to complement the LIN defect of T4ΔrI with rI resulted in a 5-min delay in lysis, compared to the 30- to 40-min delay seen with ^ss phoAΦrI_CTD (Fig. 3B). This result is consistent with the increased stability of the RI_CTD compared to that of RI with its N-terminal SAR domain.

We reasoned that the rapid degradation of periplasmic RI might due to the recognition of its exposed SAR domain by proteolytic machinery responsible for removing misfolded proteins from this compartment. Since DegP is known to be involved in the degradation of misfolded periplasmic proteins (13, 24, 25), we determined the effect of mutations that eliminate DegP activity on the half-life and function of RI. The RI protein was found to have a prolonged half-life in a degP mutant (Fig. 4D and 5). Unexpectedly, no significant lysis delay was observed when T4ΔrI was complemented with rI expressed from a plasmid in this mutant (Fig. 3D). Moreover, T4 plaque morphology is unaffected on degP lawns (not shown).

Recently, Isaac et al. (11) showed that degradation of a subset of misfolded periplasmic proteins by DegP required the adaptor protein CpxP. We hypothesized that both DegP and CpxP were necessary for RI degradation. According to this idea, even when DegP is absent and RI is stable, LIN might be compromised because CpxP binds to RI and prevents the interaction of RI with T. To test this hypothesis, we performed the complementation experiments in the absence of CpxP, DegP, or both. The lysis delay in the double mutant was not different from that seen for the degP mutant alone (Fig. 3D), indicating that CpxP did not interfere with RI antiholin function.