Bacteriophage P22 Tail Accessory Factor GP26 Is a Long Triple-stranded Coiled-coil*

P22 is a well characterized tailed bacteriophage that infects Salmonella enterica serovar Typhimurium . It is characterized by a “short” tail, which is formed by five proteins: the dodecameric portal protein (gp1), three tail accessory factors (gp4, gp10, gp26), and six trimeric copies of the tail-spike protein (gp9). We have isolated the gene encoding tail accessory factor gp26, which is responsible for stabilization of viral DNA within the mature phage, and using a variety of biochemical and biophysical techniques we show that gp26 is very likely a triple stranded coiled-coil protein. Electron microscopic examination of purified gp26 indicates that the protein adopts a rod-like structure (cid:1) 210 Å in length. This trimeric rod displays an exceedingly high intrinsic thermostability ( T m (cid:1) 85 °C), which suggests a poten- tially important structural role within the phage tail apparatus. We propose that gp26 forms the thin needle-like fiber emanating from the base of the P22 neck that has been observed by electron microscopy of negatively stained P22 virions. By analogy with viral trimeric coiled-coil class I membrane fusion proteins,

P22 is a well characterized double-stranded DNA bacteriophage that infects Salmonella enterica serovar Typhimurium. The mature virion is an icosahedral T ϭ 7 capsid about 650 Å in diameter, which has a short tail apparatus ϳ180 Å in length emanating from a single vertex. This tail or "portal vertex structure," which is responsible for binding to and injecting the DNA into the bacterial host, contains only five proteins: a dodecameric portal protein gp1 (12 ϫ 83.5 kDa) (1), six copies of the trimeric tail-spike protein gp9 (6 ϫ 215.4 kDa) (2), and small numbers of the three tail accessory proteins (also called "head completion proteins") gp4 (18.0 kDa), gp10 (52.3 kDa), and gp26 (24.7 kDa) (3,4). Electron microscopic examination of negatively stained P22 virions revealed a thin fiber protruding from the base of the phage tail (5,6). This fiber appears to emanate ϳ250 Å from the center of the P22 tail, but it has remained essentially uncharacterized. Despite its relative complexity, the P22 tail is remarkably simpler than those observed in many tailed bacteriophages. For instance, in the long-tailed bacteriophage T4, the tail machinery is formed by at least 22 different proteins (7,8), and even in phage T7, another shorttailed phage, the portal vertex structure may contain as many as nine proteins (9).
P22 virus cycle is characterized by the formation of a spherical procapsid (average diameter ϳ600 Å), which matures into infectious virus (average diameter ϳ650 Å) via a series of dramatic conformational changes (10,11), DNA insertion, and protein additions (12). The P22 chromosome (ϳ43.5 kbps) is packaged into the procapsid through the portal protein ring in an ATP-dependent process (13) with the aid of a virally encoded terminase complex (gp2 and gp3) (5). After DNA cleavage and encapsidation of the genome (12), the hole in the portal complex, through which DNA enters, is closed by the three tail accessory proteins, which add to the growing tail structure in the following order: gp4, gp10, and finally gp26 (14). This is followed by the attachment of the six tail-spikes (gp9) (15), which yields the mature infectious phage. gp4, gp10, and gp26 are required for the stabilization of newly packaged DNA within P22 capsids. P22 phages bearing mutations in any of these three genes assemble as procapsids and package DNA but are unstable and lose their DNA within minutes, even inside the infected cell (14,16). Of the tail accessory proteins, only gp10 and gp4 are required for gp9 addition (5,14). Tailspikes attach to capsids that lack gp26, while gp10 Ϫ capsids lack the tail-spike attachment site (13). This suggests that gp10, and not gp26, may physically connect the tail-spike protein to the virion. The "DNA-leakage" phenotype shows that gp4, gp10, and gp26 are not directly involved in DNA packaging but are critical to retaining the encapsidated dsDNA inside the capsid. Addition of extracts containing gp26 to P22 gp26 Ϫ (4 ϩ , 10 ϩ ) particles results in a dramatic stabilization of the capsid, which changes its half-life from the order of minutes to months or years (14), suggesting that it is gp26 that stabilizes the DNA within the capsid, perhaps by plugging the hole through which DNA entered.
In the work presented here we have isolated the gene encoding tail accessory factor gp26 and produced recombinant gp26 in Escherichia coli. Interestingly, we show that purified gp26 is a long triple-stranded coiled-coil that resembles class I viral membrane fusion proteins. We propose that gp26 represents the uncharacterized fiber emanating from the center of the phage tail.

MATERIALS AND METHODS
Cloning, Protein Expression, and Purification-The gene encoding the tail accessory factor gp26 was PCR amplified from P22 DNA with primers 5Ј-CACGATGCATATGGCAGACCCGTCACTTAATAATCC and 5Ј-GTAGAGAGGATCCCGATGTTGCGTGTTGGAGTG, cleaved with NdeI and BamHI, and ligated into similarly cleaved plasmid expression vector pET-15b (Novagen, Madison, WI). The resulting plasmid, whose insert was completely sequenced to confirm its structure, was named p26AA1-233.A. It contains P22 DNA bp 8210 -8941 DNA (17,18), and the gp26 protein it expresses has six histidines within the 20 vector-encoded amino acids that are fused to its NH 2 terminus. The P22 gene 26 terminates at nucleotide 8911, so 30 bp of gene 14 DNA are present at its 3Ј-end and its native termination codon is utilized. The recombinant protein was expressed in E. coli strain BL21 cells, in LB broth supplemented with 2.5 g/liter glucose. After growth at 37°C to an optical density of 0.6, gp26 expression was induced with 0.5 mM isopropyl ␤-D-thiogalactopyranoside and the culture was shaken at 22°C for 14 h. Recombinant His-tagged gp26 protein was purified by metalchelating chromatography using Qiagen nickel-agarose beads. gp26 bound to beads was washed with increasing concentrations of imidazole (5-20 mM) and finally eluted with ϳ200 mM imidazole. Purified gp26 was then concentrated to 5 mg ml Ϫ1 using a Millipore concentrator (cut-off 10 kDa) and analyzed by gel filtration chromatography on a Superdex 200 column (Amersham Biosciences) in 250 mM sodium chloride, 20 mM Tris, pH 8, 0.1 mM phenylmethylsulfonyl fluoride. Calibration of the gel filtration column was carried out using high molecular mass globular protein standards (Bio-Rad) consisting of thyroglobulin (670 kDa), bovine ␥-globulin (158 kDa), chicken ovalbumin (44 kDa), equine myoglobin (17 kDa), and vitamin B 12 (1.3 kDa).
Sedimentation Equilibrium Analysis-Sedimentation equilibrium experiments were conducted in a Beckman Optima XL-A analytical ultracentrifuge equipped with UV optics, at 4°C, using a six-channel ANTi60 rotor with 12-mm thick, charcoal-epon centerpieces. The sam-ple channels in each cell contained different loading concentrations of gp26 protein in 200 mM NaCl, 20 mM Tris-Cl buffer, pH 8.0, while reference channels contained the corresponding buffer only. Samples were centrifuged at 16,000 rpm until sedimentation and chemical equilibrium were attained. Cells were scanned radially in continuous mode, with data resulting from ten absorbance readings taken at 0.001-cm intervals. Equilibrium was confirmed by no change in scans taken at four hourly intervals. The partial specific volume (0.723 cc/g) and the extinction coefficient for gp26 were calculated from the amino acid sequence (17, 18) using described methods (19). Curve fitting and calculation of molecular mass were done with non-linear least square techniques using program NONLIN (20).
Circular Dichroism-Circular dichroism spectra were collected on an AVIV circular dichroism spectrometer model 62DS at 20°C. The concentration of purified oligomeric gp26 was ϳ6 M in 200 mM NaCl, 20 mM Na 2 HPO 4 , pH 8. The molar ellipticity was monitored using a 1-mm path length cell and 0.50 nm wavelength increments from 195 to 250 nm. The resulting spectra were corrected according to the base line by using above buffer. For thermal unfolding curves the ellipticity at 222 nm was monitored as a function of temperature between 15 and 100°C. The temperature was increased in 1°C increments followed by 2-min equilibration.
Negative Stain Electron Microscopy-For electron microscopic examination, a ϳ10-l drop of purified gp26 at 0.1 mg ml Ϫ1 was placed on a parafilm sheet and allowed to stand for 1.0 min. A carbon-coated electron microscope grid previously glow discharged in a Polaron E5100 vacuum evaporator was placed on top of the drop, and gp26 was allowed to adsorb for 30 s. The grid was then stained for 1 min with 1% uranyl formate, blotted, and allowed to air dry. Specimens were observed and photographed in a Philips CM120 electron microscope operated at 120 keV. Electron microscope images were collected on a Gatan CCD camera at ϫ52,000 magnification.

FIG. 1. Prediction of a coiled-coil domain in gp26.
A, probability of coiledcoil formation in gp26 protein scored with the algorithm by Lupas. Green, blue, and red plots display the results obtained from the analysis of the gp26 sequence using a 14-, 21-, and 28-residue window width (21). B, primary sequence of gp26. The regions of gp26 with high propensity to adopt a helical conformation are underlined. Predicted heptads are highlighted in red. The hydrophobic region 1-37 is highlighted in yellow. C, gel filtration analysis of recombinant oligomeric gp26. The Superdex 200 gel filtration chromatography was calibrated using molecular mass markers, whose elution volume is indicated by the arrows. The molecular mass for the gp26 oligomer corresponds to ϳ150 kDa. D, elution fractions of gp26 peak analyzed on SDS-PAGE 12.5%. Lanes 1-7 correspond to main peak fractions eluted in the range of 62-74 ml.

RESULTS
Bioinformatic Analysis of the Gene Encoding P22 gp26 -The P22 gp26 amino acid sequence (17,18) was analyzed using a computer-based prediction algorithm developed by Lupas (21), and we found four putative coiled-coil heptads in region 50 -152 as well as one conserved heptad in the COOH-terminal region 200 -230. Each of these heptads contains the motif abcdefg, where positions a, d, and g are occupied by hydrophobic residues. In gp26 a is always an Ile, d and g show a strong preference for Leu and Val, respectively (Fig. 1, A and B). This arrangement is characteristic of heptads found in trimeric coiled-coil proteins (22) and mediates formation of elongated triple-stranded helical structures.
Prediction of the secondary structure content with the program PHD (23) indicates that both regions 50 -152 (helix A) and 200 -210 (helix B) have high propensity to fold as ␣-helices. In contrast the NH 2 -terminal moiety (1-50) and the linker (153-200) are predicted to be random coil. This predicted topology is consistent with a hairpin-like structure formed by a helix A-loop-helix B motif. By analogy with known helical trimeric coiled-coil proteins, three gp26 hairpins are expected to oligomerize via heptad-mediated interactions to generate a trimeric coiled-coil similar to those seen in class I fusion proteins (22).
Purification of Recombinant P22 Tail Accessory Factor gp26 -The gene encoding gp26 was identified as the only open reading frame between P22 genes 10 and 14 (17)(18). The gene was PCR amplified from P22 DNA and cloned into a prokaryotic expression vector. Recombinant gp26 fused to an NH 2 -terminal His 6 tag was overexpressed in E. coli BL21 (see "Material and Methods"). After purification by metal chelating chromatography, gp26 was concentrated to ϳ5 mg ml Ϫ1 and analyzed by gel filtration on a Superdex 200 column, which resolves macromolecules in a molecular mass range of 10 -600 kDa. At NaCl concentrations below 150 mM, the tail accessory factor gp26 displayed a strong tendency to aggregate, yielding large species migrating close to void volume. In the presence of at least 200 mM NaCl, gp26 appeared as a more monodisperse major peak eluting after ϳ65 ml (Fig. 1C). Molecular mass calibration standards showed that gp26 elutes at an apparent molecular mass of ϳ150 kDa. This value is consistent with a hexamer of a globular 26.9-kDa gp26 protein or, alternatively, a smaller oligomeric species with an elongated shape (e.g. trimer, see below).
gp26 Is a Trimer in Solution-To determine the exact oligomeric state of gp26 protein in solution, gel filtration-purified gp26 major peak material (Fig. 1C) was subjected to sedimentation equilibrium analysis. This technique provides a shapeunbiased measurement of the protein mass (24). Fig. 2A shows the concentration of gp26 protein as a function of position in the rotor cell after equilibrium was attained. Models describing the concentration distribution were fit to equilibrium absorbance versus radius data using the non-linear least squares analysis program NONLIN (20). The data indicate that gp26 exists as a 80.7 Ϯ 6-kDa species, in striking agreement with the sequencederived molecular mass of 80.69 kDa for a trimer of the recombinant histidine-tagged gp26 protein (26,895 ϫ 3 ϭ 80,685 Da). The final fit shown in Fig. 2A resulted from the simultaneous fitting of two different concentration distributions. At higher concentrations there was evidence of further aggregation of these trimers. We conclude from this analysis that gp26 protein exists as a trimer in solution under these conditions. Thus, the apparent gp26 molecular mass of 150 kDa measured by gel filtration probably indicates that it is an elongated trimer.
Visualization of gp26 by Negative Stain Electron Microscopy-We used negative stain transmission electron microscopy to visualize the gp26 trimer. The elongated shape suggested by gel filtration chromatography and the conservation of four heptads found in the predicted helix (residues 50 -152) suggest this putative helix may represent the actual trimerization coiledcoil core of the protein. Micrographs of gel filtration-purified gp26 stained with 1% uranyl formate clearly show a rod-like structure having an average length of about 210 Å (Fig. 2B). These structures are remarkably similar to trimeric coiled-coil class I fusion glycoproteins from Ebola pIIGp2 (25), HIV-1 gp41 (26,27), and the low pH-induced conformation of influenza virus TBHA2 (28,29). For instance, rods of influenza virus HA2, which include a coiled-coil domain of ϳ65 amino acids, measure about 100 Å in length (30). gp26, which has a predicted coiled-coil domain of ϳ100 amino acids (residues 50 -151, Fig. 1B), forms an ␣-helical rod-structure of ϳ210 Å. This suggests that three helices forming the coiled-coil core are fully extended and not folded in bundles, since 100 amino acids are expected to form ϳ155 Å of trimeric coiled coil. In addition, a globular domain is also visible at one end of the rod (Fig. 2C), which may be the non-helical regions predicted between residues 1 and 50 or 155 and 210.
Structural Stability of Trimeric gp26 -Trimeric coiled-coil structures have been reported to have extraordinary thermodynamic stability (22). To investigate the folding and stability of recombinant trimeric gp26, we first analyzed gel filtrationpurified gp26 by circular dichroism (Fig. 3A). Spectra recorded in 20 mM Na 2 PO 4 and 200 mM NaCl displayed a double minimum at 208 and 222 nm and a maximum at 196 nm, characteristic of ␣-helices. Quantification of CD data indicates a helical content of ϳ55%, which agrees well with the expected secondary structure prediction. As seen in section 3.1, regions 50 -152 and 200 -230 are predicted to be helical, which correspond to ϳ55.7% of the gp26 sequence (130/233 residues).
To characterize the structural stability of gp26, we studied the thermal unfolding of the protein by circular dichroism. Denaturing curves were recorded following variation of ellipticity at 222 nm as a function of the temperature. Trimeric gp26 unfolded irreversibly as a single entity (Fig. 3B) with a thermal melting temperature T m ϳ85°C. The slope of the transition between native and denatured states is sigmoidal in nature, suggesting the denaturation of gp26 is highly cooperative (31). We also examined the thermal denaturation of the protein in the presence of 1% SDS. Strikingly, despite the high concentration of detergent, gp26 trimer unfolded irreversibly with a T m ϳ60°C. This remarkably high temperature of melting is higher than that observed in most globular proteins (31). The SDS resistance was also confirmed by gel electrophoresis. Both unboiled gp26 and protein boiled for 1 min migrated on SDS-PAGE as a trimeric species of 85 kDa (Fig. 3C, lanes 1 and  1Ј and 2 and 2Ј). In contrast, no SDS resistance was observed for gp26 boiled for more than 60 s (Fig. 3C, lanes 3 and 3Ј and  4 and 4Ј).

gp26 Forms the Thin Fiber Emanating from the P22 Capsid-
We have isolated the gene encoding tail accessory factor gp26 and characterized a recombinant gp26 protein. Our data indicate that gp26 exists in solution as a very stable elongated trimeric coiled-coil ϳ210 Å in length. The gp26 trimer is likely held together by four highly conserved heptads bearing the consensus IxxLxxV/Y. In other well characterized trimeric coiled-coil proteins, the side chains of these hydrophobic residues pack against each other in a "knobs-into-holes" manner, forming a continuous extended hydrophobic core (32). Analysis of the thermal stability reveals that gp26 trimer is extraordinarily resistant to heat denaturation and to sodium dodecyl sulfate. The oligomer melts irreversibly with a T m ϳ85°C, a value comparable with that observed for thermo-stable proteins purified from thermophilic bacteria.
It was observed over 30 years ago that isolated P22 capsids imaged by electron microscopy sometimes show an extended needle-like fiber emanating from the base of the tail (6,14). Although this fiber has remained essentially uncharacterized, several direct and indirect observations led to the hypothesis that gp26 may indeed form the phage fiber. As mentioned previously, the whole P22 tail (portal vertex structure) is formed by only five gene products. Of these, both portal protein (gp1) and tail-spike protein (gp9) have been extensively characterized, and they are not involved in fiber formation. This implies that the tail-fiber is formed by one or more of the tail accessory factors gp4, gp10, and gp26. The structural characterization presented here indicates that the trimeric putative coiled-coil quaternary structure of gp26 is completely consistent with an elongated fiber adding to the tail of the mature phage. The length and morphology of recombinant purified gp26 (ϳ210 Å) is also consistent with the P22 fiber observed in electron micrographs of mature P22 phage (5,6). Furthermore, Strauss and King (14) observed that gp26, which is present in two-to-three molecules per P22 virion, sediments with a sedimentation coefficient of ϳ5S, consistent with an elongated oligomer of multiple subunits (14). In contrast, neither gp10 nor gp4 are likely candidates to form a tail-fiber. gp10 is believed to form the attachment site for the tail-spike as revealed by 10 Ϫ phages, which lack both tail-spike and tail-fiber (14), and gp4 appears to be a monomer in solution that binds with high affinity to portal protein 1 and also has endolysin activity (33).
Resemblance of gp26 to Class I Membrane Fusion Proteins-Trimeric coiled-coil is a highly versatile protein folding and oligomerization motif. Coiled-coil structures have been observed in a variety of cellular and viral proteins (34). For instance, in class I viral fusion proteins, triple-stranded coiledcoil domains mediate fusion of the viral membrane with the host-cell membrane (22). Examples of class I fusion proteins are found in various viruses such as orthomyxovirus, paramyxovirus, retroviruses, and filoviruses. These fusion proteins share a similar quaternary structure that is usually described as a trimer-of-hairpins. Each subunit of the trimer presents an amino-terminal (or amino-proximal) highly hydrophobic peptide known as fusion-peptide, followed by a hairpin presenting a helix-loop-helix motif. The long helix following the fusion peptide contains conserved heptads, which hold together the trimeric coiled-coil core. The second shorter helix folds from the outside onto the trimeric coiled-coil core forming a six-helix bundle. A similar coiled-coil helical bundle motif is also found in non-viral membrane fusion devices like the SNARE complex (35), and it has been hypothesized to be a global motif for promoting membrane fusion events (22,36).
gp26 has a resemblance to class I fusion proteins. Both the trimer-of-hairpins topology and trimerization heptads may be present. gp26 exhibits a remarkable thermodynamic stability that is also observed for many class I fusion proteins (22). Analysis of gp26 primary sequence reveals that the NH 2 -terminal residues 1-37 are highly enriched in hydrophobic residues, especially at positions 9 PVVI 12 and 31 YLLYVIA 37 (Fig.   1B). These hydrophobic regions may represent a prototype of phage fusion peptide (37), which inserts into the host lipid bilayer. Although it is well known that P22 tail-spike protein bears the endoglycosidase activity required to cleave the Oantigenic repeating units present on the S. enterica cell surface (15), it remains unknown how the phage inserts a DNA passage structure into the host outer membrane bilayer. In this regard, gp26 tail-fiber could engage in two functions. At the portalproximal side, gp26 may stabilize the phage head by "plugging" the portal protein DNA-pumping channel. At the distal end, gp26 may harbor a class I-like fusion peptide that inserts into the host lipid bilayer. This putative lipid-penetrating activity remains speculative, and the exact nature of the binding interactions taking place within the P22 tail remains to be investigated.