INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Bacteriophage holins are small membrane proteins required for host cell lysis. For most phages, a bacteriolytic activity, or endolysin, is elaborated in the cytoplasm during the vegetative phase. Endolysins with a variety of enzyme activities have been reported, including true lysozymes (17, 20), transglycosylases (3), endopeptidases (24), and amidases (15, 22). In general, endolysins have no secretory signal sequence and thus accumulate as fully folded enzymes in the cytoplasm (46). The holin functions to effect release of the endolysin across the cytoplasmic membrane at a precisely scheduled time. There appears to be no specific interaction between the holin and endolysin, because unrelated endolysins with different enzymatic activities can act with the same holin to effect lysis (7, 13, 25, 33, 42). More than 60 different genes have been identified as holins by function or postulated to be holins by sequence analysis (38, 46, 47).

On the basis of primary sequence, holins appear to fall into two size classes: class I holins, the prototype for which is the protein S from phage , are 90 residues or more in length; class II holins, the prototype for which is the S protein from the lambdoid phage 21, are usually less than 75 residues (Fig. 1A) (7, 46, 47). Membrane topology seems obvious for the class II holins, which have two putative transmembrane domains (TMDs) and almost certainly are oriented with the N and C termini in the cytoplasm (Fig. 1B) (46, 47). The topology of S as the prototype class I is not settled, because there is a third uncharged although relatively hydrophilic domain which could be a transmembrane alpha-helix (Fig. 1B). Analysis of the distribution of positively charged residues according to the von Heijne "positive-inside" rule favors the three-domain model, with the N terminus outside (Fig. 1B) (31, 44).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   (A) Topological models for class I and II holins. Class I holins may have three TMDs, but class II holins are limited by their size to two TMDs. In both cases, the longer form ( S107 and 21 S71) is depicted. The C and N termini and charges are indicated. (B) Amino acid sequences of the class I holin of  and the class II holin of 21. Charged residues are denoted above the sequence, while putative TMDs (===), turn regions (ttt), and highly charged C-terminal regions (***) are indicated below the sequence. The ambiguous uncharged domain in  S (::::) is indicated. The # signs indicate the dual starts demonstrated for  S (5, 31) and suggested for 21 S (7). The vertical arrow indicates the site of the inserted G2H6G2 sequence in the S105H94 protein (see text).

Genetic analysis of S has been fruitful. Because S is small and lethal, it has been possible to select for a large number of missense mutations. During a screen for dominant mutations, it was found that some alleles with changes in the 5' flanking region of S had dominant character (31). This led to the identification of a secondary structure, called sdi (for structure-directed initiation), which directs translational initiation events not only to codon Met₁ but also to the Met₃ codon (Fig. 1A and 2). S thus produces two distinct proteins, called S107 and S105, reflecting the number of amino acid residues in the products resulting from initiation at codons 1 and 3, respectively (5). Remarkably, the two proteins have opposing functions in vivo, with the shorter product S105 acting as the lethal lysis effector, whereas the S107 protein functions as an inhibitor of S105 function (30, 31). The inhibitory function of S107 was shown to be due to the positively charged residue at codon 2 (4). Thus, it was apparent that one level of the temporal regulation of S-mediated lysis was the partition of S expression between S105 and S107 products. This "dual-start" mode of regulation appears to be conserved in lambdoid and other holins, including both class I and class II holins (6, 47). The conservation of this motif implies that the functional regulation of the two classes of holins is the same. This supports the notion that the N termini of both proteins are cytoplasmically disposed, which in turn requires that  S have two TMDs like the phage 21 class II holin (Fig. 1B).

Genetic analysis further revealed that the structure of S contains timing information independent of the N terminus. A missense mutant was isolated with the Ala-to-Gly change at position 52 (A52G) in the second TMD. This mutant had an unconditional defect in plaque formation resulting from host lysis which occurs so early that, on average, the first virion has not been assembled (23). This result demonstrates that holins have two essential functions: not only must holes be formed to give the endolysin access to the peptidoglycan but also these lesions must form only after a period adequate to support unabated intracellular production of the progeny virions. Quantitative analysis suggests that the  holin accumulates in the membrane to approximately 10³ molecules per cell before hole formation is triggered to terminate the infective cycle (10, 39, 48).

In this light, it was doubtful that purification of the S holin was going to be feasible, given the fact that this molecule evolved to kill the host cell at very low concentrations and is a small membrane protein without enzyme activity. However, we have found that, although hyperexpression of the S gene results in loss of more than 5 orders of magnitude of viability in the induced culture within 5 min, the induced cells are metabolically active for 10 to 15 min, allowing accumulation of S protein in the cytoplasmic membrane to levels 100-fold in excess of the normal triggering levels (39, 40). At this level of expression, milligram quantities of this small membrane protein could, in principle, be obtained from liter cultures. Here we report successful efforts to achieve purification of biochemically useful quantities of the  holin by using an inserted oligohistidine tag. The results are discussed in terms of the sensitivity of S and membrane proteins to multiple-residue insertions and the potential for developing an in vitro system for investigating the molecular basis of holin function and regulation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Strains, phages, plasmids, oligonucleotides and standard DNA manipulations and sequencing. Relevant Escherichia coli strains, bacteriophages, plasmids, and oligonucleotides used in this study are listed in Tables 1 and 2. Oligonucleotides were purchased from the Gene Technologies Laboratory in the Department of Biology at Texas A&M University. Luria-Bertani (LB), LB-AMP (LB medium supplemented with 100 µg of ampicillin per ml), TB, and TB-maltose (TBM) media were prepared by the methods of Miller (26). The pET11a plasmid vector was acquired from Novagen (Madison, Wis.). Octylglucoside (n-octyl--D-galactopyranoside) was purchased from ICN Biomedicals Inc. (Costa Mesa, Calif.). Egg yolk lecithin (L--phosphatidylcholine [PC]) was obtained from Avanti Polar Lipids Inc. (Alabaster, Ala.). Unless otherwise indicated, chemical compounds were obtained from Sigma (St. Louis, Mo.).

Site-directed mutagenesis of S by PCR. PCR was used to create a unique SmaI site in codons 38 and 39 of S in the plasmid pKB110 (Table 1) (Fig. 2). Two mutagenic oligonucleotides, ForSG38P and RevSG38P, were annealed to covalently closed circular pKB110 plasmid DNA under standard PCR conditions and subjected to 30 cycles of the following thermal cycler program: 95°C for 1 min, 50°C for 1 min, and 72°C for 12 min. The resultant PCR product was gel purified, digested with SmaI, ligated, and transformed as previously described. Plasmid DNA was obtained from transformant colonies and screened by restriction with SmaI. One SmaI-sensitive plasmid, pSg38p, was sequenced to confirm the mutagenized sequence.

View larger version (25K): [in this window] [in a new window]   FIG. 2.   Map of the  lysis gene region and plasmid derivatives. The sdi mRNA and N-terminal amino acid sequences of the S gene products are depicted. The boxed sequences in the mRNA indicate the Shine-Dalgarno sequences for the dual translational starts of the S gene (5). sdi is the RNA stem-loop structure in which single-base changes are known to cause dominant-negative lysis defects (31). The pR' and tR' promoter and terminator for the  late transcriptional unit are shown upstream of the EcoRI site. Restriction sites are indicated (H, HindIII; R, EcoRI; V, EcoRV; C, ClaI; N, NdeI; B, BamHI). pKB110 contains a HindIII/ClaI fragment corresponding to the wild-type "lysis cassette" of  (46). N- and C-terminal oligohistidine tags were inserted in frame into the S105 gene by Seamless Cloning. An isogenic trans activation plasmid, pS105, bearing the M1L mutation was also made. A derivative of this plasmid, pS105H94, encoding an oligohistidine tag between codons 94 and 95, was made by modified Seamless Cloning. The plasmid pS105H94 was subjected to QuikChange mutagenesis to introduce the A52V lysis-deficient allele, creating the plasmid pS105H94a52v. The S105 alleles of the internally tagged oligohistidine trans activation plasmids were subcloned into pET11a by standard PCR. In the diagram, oligohistidine tags in S at the N terminus or between codons 94 and 95 are mutually exclusive. The N-terminal coding regions of pETNHS105 and pETS105 and their translations are illustrated. The boxes indicate the Shine-Dalgarno sequences. rbs, ribosome binding site.

Mutagenesis of S using Seamless Cloning. All manipulations of S were done using the reading frame for S105, beginning with codon 3 and terminating at codon 107. For simplicity, this reading frame will be designated S105 throughout this work. Seamless Cloning (Stratagene, La Jolla, Calif.) was used to place the NHS105 allele under  transcriptional and translational control by substituting it for the S⁺ allele of pKB110. This plasmid and its derivatives are used as functional test vectors for all S alleles by introduction into a host carrying  Cm^r (SR). Induction of the lysis-defective prophage results in trans activation of the p_R' promoter and thus transcription of the lysis genes on the plasmid (Fig. 2). S105 alleles with the oligohistidine tags are designated as follows: NHS105 has the sequence MH₆GSH at the N terminus, CHS105 has the sequence G₂H₆ at the C terminus, and S105H94 has the sequence G₂H₆G₂ inserted between residues 94 and 95. The manufacturer's protocol was followed exactly. A vector fragment was synthesized by using the primers ForpKBEam and RevpKBEam with the SmaI-linearized pSg38p template. The S105 insert was synthesized with ForS105Eam and RevSEam. The NHS105 insert was amplified with ForNHS105Eam and RevSEam, while the CHS105 insert was amplified with ForS105Eam and RevCHSEam.

Modified Seamless Cloning. An oligohistidine tag was inserted between codons 94 and 95 without creating a new restriction site in the resulting plasmid (Fig. 2). The Seamless Cloning (Stratagene) protocol was modified in two respects. First, only two oligonucleotides or primers, ForPhe94His and RevPhe94His, were used, encoding the Eam1104I restriction site, half of the oligohistidine tag, and 19 to 21 nucleotides (nt) of homology with the S gene (Table 2). Second, the two primers were annealed to a covalently closed circular plasmid (pS105) and extended completely around the plasmid under the following conditions: 1 cycle of 94°C for 3 min, 50°C for 1 min, and 72°C for 12 min and then 10 cycles, with 1 cycle consisting of 94°C for 1 min, 50°C for 1 min, and 72°C for 12 min. 5-Methyl-dCTP was then added to the sample exactly as specified in the Seamless Cloning protocol, and five cycles were performed, with one cycle consisting of 94°C for 1 min, 50°C for 1 min, and 72°C for 12 min. The reaction mixtures were digested, ligated, and transformed, and transformants were screened exactly as described above for the Seamless Cloning technique.

QuikChange mutagenesis. Specific alleles were introduced into the plasmids pS105 and pS105H94 by site-directed mutagenesis by using the QuikChange kit from Stratagene (Fig. 2). Manufacturer's instructions were specifically followed. The plasmids pS105 and pS105H94 were used as template DNA (Table 1). Constructs were confirmed by diagnostic PCR and sequencing as previously described (39).

Induction of pET11 clones of the S gene. Cells carrying the T7 vector plasmid, pET11a (Novagen), or any of its derivatives (Table 1) were grown and induced as previously described (39, 40).

Induction of functional test vectors harboring the S gene. In each case, MC4100 [ Cm^r (SR)] harboring the trans activation plasmid derived from pKB110 was induced as previously described (10), except that the period of aeration at 42°C was decreased from 20 to 15 min.

Membrane protein sample preparation and analysis. Detergent-solubilized preparations of inner membrane proteins were obtained, resolved, immunoblotted, and analyzed as described previously (10, 39, 40).

Immobilized metal affinity chromatography (IMAC). Ni-NTA (nickel-nitrilotriacetic acid) (Qiagen, Chatsworth, Calif.) and cobalt-based Talon affinity resin (CLONTECH Laboratories, Inc., Palo Alto, Calif.) were purchased as 50:50 slurries. Columns (0.5- to 1.0-cm diameter) with bed volumes of 0.5 to 1.0 ml were poured and preserved in 30% ethanol. Prior to use, columns were equilibrated with 20 column volumes of a solution consisting of 1% Triton X-100, 20 mM Tris-HCl, and 10% glycerol (TTG) plus 0.5 M NaCl (TTGN), pH 8.0. Columns were developed at a flow rate of no more than 4 column volumes per h. Detergent-solubilized membrane extracts were loaded based on estimates of the extracts' S protein content and on an optimized 10-mg/ml binding capacity of the resin for oligohistidine-tagged protein. Columns were then washed with 10 column volumes each of the following: (i) TTGN at pH 8.0, (ii) TPGN (1% Triton X-100, 10% glycerol, 0.5 M NaCl, 100 mM sodium phosphate) at pH 6.5, and (iii) TPGN at pH 4.5. Columns were finally eluted with a solution consisting of 1% octylglucoside, 10% glycerol, and 100 mM sodium phosphate using a linear gradient from pH 4.5 to 2.5. Column fractions were analyzed for protein purity by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis.

Determination of protein concentration. Estimates of protein concentration of various samples were determined by DC (detergent-compatible) Bio-Rad Protein Assay (Bio-Rad Laboratories, Hercules, Calif.) according to the manufacturer's recommendations. Bovine serum albumin was used as a standard. Concentrations of purified samples of S protein were determined by the method of Pace et al. (28) after complete dialysis against 6 M guanidine hydrochloride (GdnHCl) in 20 mM Tris-HCl-150 mM NaCl (pH 7.6) (Tris-buffered saline [TBS]), yielding an ₂₈₀ of 11,460 M¹ cm¹.

CD. Purified samples of S protein were analyzed on an Aviv model 62DS circular dichroism (CD) spectrometer (Aviv Associates Inc., Lakewood, N.J.) at wavelengths from 195 to 320 nm in a 0.1-cm pathlength cuvette at a final concentration of 15 to 25 µM at room temperature (23°C). Scans were measured in millidegrees and converted to mean residue ellipticity. Alpha-helical content was estimated from the molar ellipticity signal at 222 nm.

Preparation of calcein-loaded liposomes. First, 350 µl of a solution of PC dissolved in hexane-ethanol (20 mg of PC per ml) was mixed with 143 µl of cholesterol dissolved in chloroform (7 mg/ml) and dried under air. After removal of residual organic solvents under a vacuum for 15 min, the lipids were resuspended in 1 ml of 0.2 M calcein in TBS with vigorous vortexing. The Liposofast apparatus (Avestin, Inc., Ottawa, Ontario, Canada) was fitted with polycarbonate membranes (100-nm pore size), and the multilamellar lipid suspension was extruded through the membranes 19 times. The nominally unilamellar liposomes loaded with calcein were separated from unincorporated calcein by gel filtration chromatography on a 150-ml Sephadex G50 resin column equilibrated in TBS. We estimate that 10¹² liposomes were added to each assay based on the lipid content of E. coli (27).

Dye release assay. Dye release assays were performed in an SLM-Aminco 8000 or 8100 fluorescence spectrophotometer (Spectronic Instruments Inc., Rochester, N.Y.) in 2.5-ml glass cuvettes with all four sides polished. Two milliliters of TBS was added to a cuvette with 40 µl of liposomes. The samples were excited at a wavelength of 490 nm, and emission was monitored at 520 nm. Baseline, or 0% relative fluorescence, was defined as the signal acquired upon addition of a volume of buffer (6 M GdnHCl in TBS) equal to the volume of protein added. Total release, and therefore 100% relative fluorescence, was obtained by adding 10 µl of 10% Triton X-100. Protein (2 to 15 µg in a volume of 5 to 40 µl) was added to initiate the assay. Alpha-hemolysin (9 to 18 µg in a volume of 5 to 10 µl) was used as a control to establish the quality and fraction of unilamellar vesicles (2, 14). Relative fluorescence was plotted versus time.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Preliminary purification of unmodified S105 and S107 proteins. No holin has been purified in biochemically useful quantities nor has an in vitro assay been devised for crude preparations. Gene fusion approaches in which easily purifiable domains (i.e., -galactosidase or ubiquitin) were fused with proteolytically cleavable linkers to the N- or C-terminal end of the S protein were unsuccessful, because in every case cleavage of the hybrid purified in detergent was at best inefficient (11). Taking advantage of the recent achievement of hyperexpression of S (39), we undertook to obtain enriched fractions of S holin by ion-exchange chromatography and preparative-scale isoelectric focusing of inner membrane extracts in detergent. The S holin could be purified to a state where only one contaminant was detectable, but the available concentration of S holin in such preparations was limited by its tendency to precipitate near its isoelectric point (38). Consequently, purification of S protein by standard chromatographic methods was abandoned.

Oligohistidine tags at the N and C termini of S. The potential for single-step purification of S from detergent extracts by IMAC led us to create oligohistidine-tagged versions of S (21). The reading frame encoding S105 was fused to an N-terminal oligohistidine tag sequence under the translational control of the consensus T710 ribosome binding site and the T7 promoter of pET11a (Novagen) (Fig. 2). Induction of this construct, designated as pETNHS105, resulted in a sudden halt in culture growth after about 10 min, followed by a gradual loss of turbidity (Fig. 3), which correlated with the accumulation of translucent, vacuolized cells as determined by phase-contrast microscopy (data not shown). These growth characteristics were very similar to those observed when the normal S105 reading frame, lacking an oligohistidine tag, is fused to the same gene expression signals (39), suggesting that the addition of the N-terminal oligohistidine tag did not grossly affect the toxicity of the S105 sequence, at least at these hyperexpression levels. Coomassie blue-detectable protein species of the expected sizes accumulated in membrane extracts (Fig. 4A, lane 1), and immunoblot analysis using the C-terminal antibody confirmed that the accumulated protein was S (Fig. 4B, lane 1).

View larger version (16K): [in this window] [in a new window]   FIG. 3.   Profiles of BL21(DE3) pETNH induction. Cells harboring plasmid pETNH () or pETNHS105 () were induced at time zero and followed by monitoring A550.

View larger version (33K): [in this window] [in a new window]   FIG. 4.   NHS105 accumulates in inner membrane and can be purified by metal affinity chromatography. (A) Coomassie blue-stained SDS-polyacrylamide gel revealing hyperexpression of NHS105 and its extraction from the inner membrane (lane 1), column fractions from purification (flowthrough [lane 2], pH 4.5 wash [lane 3], and pH 4.5 to 2.5 gradient fractions [lanes 4 to 11]). S protein dimers are visible in the peak fractions from the gradient. (B) Western immunoblot of identical fractions from the purification revealing oligomers up to tetramers. Note that some NHS105 is detected in the pH 4.5 wash. (C) Purification of S105H94. The fractions are inner membrane extract (lane 1), metal affinity column flowthrough (lane 2), pH 4.5 wash (lane 3), and pooled fractions from the pH 4.5 to 2.5 gradient (lane 4). The positions of molecular size markers (in kilodaltons) are shown to the left of the gels.

View larger version (13K): [in this window] [in a new window]   FIG. 5.   CD spectra of NHS105 () and S105H94 () proteins. See Materials and Methods for conditions of measurement.

In vivo functional assay for oligohistidine-tagged S alleles. Our inability to detect pore formation in vitro by a variety of methods prompted a more rigorous examination of the biological function of the NHS105 protein. Although the physiological effects of inducing the pETNHS105 plasmid closely resembled those obtained from inducing the isogenic constructs without the oligohistidine tag, the abnormal physiology derived in these T7 polymerase expression systems precludes useful phenotypic analysis. The S gene is normally subject to sophisticated and complex transcriptional and translational controls, upon which the timing of lysis, and thus the ultimate productivity of the vegetative cycle, is dependent (6, 46). To assess biological function, it was necessary to reinsert the NHS105 allele into the normal transcriptional and translational context of the phage.

View larger version (14K): [in this window] [in a new window]   FIG. 6.   Profiles of MC4100 [ Cmr (SR)] induction. (A) Cells harboring pKB1 (), pS105 (), pNHS105 (), or pCHS105 () were induced at time zero and monitored for 1.5 h. (B) Cells harboring pKB1 (), pS105 (), pS105H94 (), pS105H94a52v (), or pS107H94m3l () allele.

Construction of a functional internally tagged S105 allele. Despite the fact that the S105 constructs with oligohistidine tags at the N and C termini had proven to have aberrant lysis behavior in terms of timing or efficiency of endolysin release, the ease with which the tagged S protein could be purified from membrane extracts prompted us to look for internal sites in S where the tag would not seriously affect S-holin function. Constructs containing the sequence G₂H₆G₂ inserted in different locations within S105 were made by using a modified version of Seamless Cloning, and each was tested for S function at physiologically relevant levels of S expression (41). One allele, in which the tag was inserted between codons 94 and 95, exhibited a lysis phenotype sufficiently similar to wild-type S to warrant its use as a source of S protein (Fig. 6B). This allele, designated S105H94, was inserted into the pET11a hyperexpression vector, generating pETS105H94 (Fig. 2). Induction of this plasmid resulted in accumulation of S protein at levels indistinguishable from that obtained with pETS105 (Fig. 4C). As a control for in vitro hole formation assays, S105H94a52v, an isogenic allele bearing the A52V lysis-defective missense change (30), was constructed (Fig. 2). This allele was tested in the trans activation plasmid context and found to have the expected lysis-defective phenotype (Fig. 6B). Induction of an isogenic pET11a construct of this defective allele also yielded Coomassie blue-detectable quantities of S protein, at approximately twofold-higher levels than for the S105H94 lysis-proficient allele, presumably because mass accumulation continues a few minutes longer after induction (data not shown). Purification of the S105H94 (and S105H94a52v [data not shown]) protein was accomplished in the same manner as described above (Fig. 4C). All the purified S protein preparations had CD spectra indistinguishable from that observed for the original purified N-tagged S protein, with approximately 40% alpha-helical conformation in 1% octylglucoside detergent solution (Fig. 5).

In vitro assay for S function. The purified S105H94 protein, solubilized in octylglucoside, exhibited a marked tendency to aggregate in insoluble masses, which were readily resolubilized by resuspension in 6 M GdnHCl in TBS (data not shown). After replacing the chaotrope with detergent by dialysis, the S protein refolded to a conformation indistinguishable from that of S eluted from the IMAC column in octylglucoside, as determined by CD spectroscopy (data not shown). We reasoned that this renaturability might allow denatured S to insert into a target membrane in vitro. To test this idea, calcein-loaded PC-cholesterol liposomes were created by extrusion of lipid suspensions through polycarbonate membranes. At the concentrations used, the dye is self-quenched, and only when it is released from the liposomes and diluted into the assay solution does it become highly fluorescent (1). For a positive control, the alpha-hemolysin from Staphylococcus aureus, a soluble hole-forming toxin (2, 19, 45), was used to demonstrate that most of the liposomes could be made permeable efficiently by attack from the external aqueous environment and were thus not multilamellar (Fig. 7A). When S105H94 was added to the suspension of calcein-loaded liposomes, a fluorescent signal was obtained, indicating that a fraction of the dye was released, whereas the S105H94a52v protein was essentially inactive under these conditions (Fig. 7A). We reasoned that the S protein could have several fates at the instant of dilution into the aqueous suspension of liposomes (Fig. 8). The desired pathway, in which S inserted into the membrane bilayer, would compete with various unproductive pathways, including the formation of insoluble aggregates and the accretion of S protein onto the surfaces of the liposomes. The latter pathways might be favored by higher concentrations of S, and consequently the assay was repeated by adding the same total amount of S in a series of smaller sequential aliquots. Using this protocol, more-efficient calcein release was indeed observed, although the final yield of dye released never exceeded approximately 50% of that obtained using the alpha-hemolysin (Fig. 7B). Interestingly, each addition of S protein resulted in dye release with slow kinetics on the order of minutes. Because the partition between membrane insertion and aggregation is likely to be extremely rapid after dilution out of detergent or chaotrope, we presume that the slow kinetics reflects the time required for the productive formation of holes within the membrane. In any case, because this dye release was obtained with purified S protein and protein-free artificial membranes, we conclude that S protein is necessary and sufficient to make a bilayer permeable, at least sufficiently to allow passage of a fluorescein dye. The fact that the isogenic A52V protein is incapable of the same reaction strongly suggests that this permeabilization capacity is at least related, if not identical, to the "hole formation" which allows passage of endolysin through the cytoplasmic membrane in phage lysis. Moreover, we observed that prior treatment of the liposomes with S105H94a52v followed by treatment with S105H94 resulted in a more gradual dye release (Fig. 7C), suggesting that the preinserted defective holins interfered with the hole-forming activity of the active holin. This result is consistent with the partial dominant character of the A52V allele when it is present on a multicopy plasmid in trans to the parental S105 allele on an induced prophage (41).

View larger version (13K): [in this window] [in a new window]   FIG. 7.   Permeabilization of calcein-loaded liposomes by purified S protein. (A) Kinetics of dye release by holin and alpha-hemolysin. Total release is effected by adding 0.05% Triton X-100 (). Relative fluorescence was monitored continuously for the first 5 min. Single point measurements were taken between 10 and 20 min postaddition of protein. Amounts of protein added to initiate the assay were as follows: 8.75 µg of alpha-hemolysin (), 15.2 µg of S105H94 (), 3.8 µg of S105H94 (), and 17.6 µg of S105H94a52v (). (B) Sequential amounts (1.9 µg) of S105H94 were added to calcein-loaded liposomes, as indicated by the vertical arrows. Fluorescence was monitored continuously. (C) Pretreatment of calcein-loaded liposomes with a S105H94a52v. Maximum release of dye was established by addition of 0.05% Triton X-100 (). A baseline signal was established by adding an equal volume of 6 M GdnHCl in TBS (). At point a, 7.6 µg of S105H94 protein (+) or 8.8 µg of S105H94a52v protein () was mixed separately with calcein-loaded liposomes, and fluorescence was monitored for 1 h. At point b, 7.6 µg of S105H94 was then added to both S105 (+) and S105H94 () samples, and fluorescence was monitored.

View larger version (37K): [in this window] [in a new window]   FIG. 8.   Model for permeabilization of calcein-loaded liposomes by purified S protein. Purified S protein is diluted out of 6 M GdnHCl into the assay. Two immediate fates are portrayed: (i) insoluble aggregation and (ii) association with liposome membrane bilayers. S protein association with the bilayer is depicted as having three basic outcomes: (i) association with the bilayer in a nonintegral fashion, (ii) spontaneous integration into the bilayer, and/or (iii) oligomerization and hole formation. A successful hole formation event would be detected as an increase in fluorescence, as the dye goes from a self-quenched state (filled circles) to a nonquenched state (open circles) due to release and dilution into the assay buffer.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Holins constitute one of the largest functional groups of proteins known, with more than 60 reported holin sequences from more than 30 nonorthologous groups (40, 46, 47). This work reports the first purification, the first biophysical characterization, and the first in vitro assay of a holin, the S protein of bacteriophage lambda.

Positioning of the oligohistidine tag in the S protein. Oligohistidine tags are frequently added to proteins to facilitate purification by IMAC (21), and in most cases, the tag is placed at the amino or carboxy terminus of the polypeptide chain. The initial choice of the N-terminal histidine tag position for S was based on the fact that our best antibody was raised against a C-terminal epitope (10). N-tagged S protein was stable and straightforward to purify in milligram quantities from detergent-solubilized membranes. Under overexpression conditions, NHS105 affected cellular morphology and culture growth very similarly to the untagged protein. Moreover, the CD spectrum of detergent-solubilized NHS105 was exactly as expected from genetic and modeling considerations. However, in the exact transcriptional and translational context of the S gene in lambda, NHS105 was found to function much more like the inhibitor form of S, S107, than the effector form, S105. This serves as a cautionary note for the future analysis of other holins. The holin protein is designed to cause hole formation and thus terminate the vegetative cycle, at a precisely defined time, which corresponds to a certain concentration in the membrane (10, 39). It is thus unproductive to assess the function of a holin variant, such as an oligohistidine-tagged form, under conditions where the rate or amount of holin production is significantly different from the normal context of the lytic cycle. The inhibitor character of the NHS105 allele has been verified by demonstrating its ability to inhibit S function in trans (41). Assuming that this inhibitory property is based on the same requirement for N-terminal positive charge found for the S107 and P22 gp13108 inhibitor forms, this phenotype suggests that significant positive charge exists on the side chains of the hexahistidine sequence in vivo.

Purification and characterization of the holin. The use of IMAC for the purification of the holin was successful, achieving apparent homogeneity in a single step from solubilized inner membrane material. This is important because holins are small membrane proteins which have evolved to kill E. coli and thus cannot be synthesized in vivo at levels exceeding approximately 1 mg per liter of culture (39). Multistep purifications would thus be likely to reduce useable yields to unacceptable levels. However, it should be noted that the site used (between residues 94 and 95 in the middle of the highly charged C terminus) was one of only two positions of 13 tested for the oligohistidine tag in the S105 reading frame which left the timing and lysis functions of the holin largely unaffected (41). Assuming that a similar site for the oligohistidine tag would be effective in other holins, these findings should allow rapid purification of other holins of interest, including the prototypical class II holin S²¹ and the nonorthologous class I holins of phages P1 and P2 (7, 35, 49).

An in vitro assay for holin function. This work demonstrates that purified S protein, when diluted out of a chaotropic solution, is capable of forming holes in artificial liposomes. This is a major step in the effort to understand how holins achieve the hole-forming function. It is not yet clear that the holes responsible for the release of the dye are the same as those formed in vivo. However, it is unlikely that the A52V missense change would adversely affect the ability of the S protein to enter the membrane, since it actually increases hydrophobic character. Thus, the inability of S protein carrying this missense change, which abolishes hole formation in vivo, to form these holes is support for the notion that the channel formation being measured by the release of the encapsulated carboxyfluorescein is related, if not identical, to the lethal biological lesion. Moreover, each addition of S protein was followed by a definable kinetics of fluorescein release over a period of minutes (Fig. 7B). This is a tantalizing result which may reflect the other required biological function of S: not just to form a hole but to do so slowly to allow a macroscopic vegetative phase on the order of many minutes (23). Nevertheless, in its present form, this in vitro assay for holin function is still far from an optimal tool for detailed structure-function analysis of this unusual class of membrane proteins. Most S protein added to the assays ends up as insoluble aggregates, and very little S protein actually ends up associated with the liposomes, as determined by Western blot analysis of the liposome and aqueous fractions after addition of the holin (data not shown). Moreover, the assay appears limited to the release of only a fraction of the total dye entrapped in the liposome preparation, approximately 50% by the sequential method of S addition (Fig. 7B). One interpretation of this limitation is that significant amounts of S become trapped on the surface of the liposome and thus block entry of further S molecules (Fig. 8). Some of the material which is bound to the membranes is apparently bound nonproductively. In any case, substantial effort will now be required to optimize this assay. The use of altered ionic and osmotic conditions, different phospholipid mixtures, and soluble chaperones to help prevent excessive aggregation of S may be indicated. It should be noted that the lysis protein L of the RNA phage MS2 has previously been studied using similar technology (18). In that case, a fragment of the lysis protein was synthesized chemically and tested for its ability to liberate calcein from liposomes. An important difference is that L is not a holin and does not function by allowing endolysin to penetrate the cytoplasmic membrane and thus gain access to the peptidoglycan. Moreover, there is no evidence that the fragment of L used functions in the same way as the L protein does in vivo. That is, there are no mutant forms of L which have been demonstrated to fail in the same biochemical assay. In any case, experiments to optimize the dye release assay are now under way in this laboratory.