Crystal Structure of ORF12 from Lactococcus lactis Phage p2 Identifies a Tape Measure Protein Chaperone

Abstract

We report here the characterization of the nonstructural protein ORF12 of the virulent lactococcal phage p2, which belongs to the Siphoviridae family. ORF12 was produced as a soluble protein, which forms large oligomers (6- to 15-mers) in solution. Using anti-ORF12 antibodies, we have confirmed that ORF12 is not found in the virion structure but is detected in the second half of the lytic cycle, indicating that it is a late-expressed protein. The structure of ORF12, solved by single anomalous diffraction and refined at 2.9-Å resolution, revealed a previously unknown fold as well as the presence of a hydrophobic patch at its surface. Furthermore, crystal packing of ORF12 formed long spirals in which a hydrophobic, continuous crevice was identified. This crevice exhibited a repeated motif of aromatic residues, which coincided with the same repeated motif usually found in tape measure protein (TMP), predicted to form helices. A model of a complex between ORF12 and a repeated motif of the TMP of phage p2 (ORF14) was generated, in which the TMP helix fitted exquisitely in the crevice and the aromatic patches of ORF12. We suggest, therefore, that ORF12 might act as a chaperone for TMP hydrophobic repeats, maintaining TMP in solution during the tail assembly of the lactococcal siphophage p2.

During industrial milk fermentation, Lactococcus lactis cells are added to transform milk into an array of fermented products such as cheese. However, this manufacturing process may be impaired by lytic phages present in the factory environment as well as in the milk itself (30). Due to the destructive effects of phage infections on bacterial fermentation, much effort has been undertaken to isolate and study the biodiversity of these bacteriophages. Lactococcal bacteriophages belong to at least 10 different genetically distinct species of double-stranded DNA viruses (9). Of them, three lactococcal phage species, all belonging to the Siphoviridae family, are the major source of problems in milk fermentation, namely, the 936, P335, and c2 species (7, 28, 29). Furthermore, members of the 936 species are by far responsible for the majority of infections (50 to 80%) (1, 24, 41). Numerous phages of the 936 species have been isolated, and several have been characterized at the genome level (25). However, little is known concerning their molecular mechanisms of infection, although we recently solved the structure of the receptor-binding protein (RBP) of our model 936-like phage, namely, the virulent phage p2 (38, 43), and of phages belonging to the P335 species (27, 34, 37, 38).

As with all viruses, bacteriophage genomes are quite compact, leaving little room for noncoding sequences (4). In fact, phage genes are disposed in an operon-type organization (4), and the order of genes corresponds to the different phases of the infection cycle. Moreover, genes are often in clusters (referred to as modules), with gene products from adjacent genes generally found to interact with each other. Interestingly, phage genome organization, including individual gene order, is often conserved within a given species, particularly within the Siphoviridae family. In the case of L. lactis virulent phages belonging to the 936 or P335 species, this principle applies particularly to the morphogenesis gene module, which includes all the genes coding for the phage structural protein genes. For the tail assembly, a module comprises a set of genes between the portal protein, which is connecting the tail to the capsid, and the RBP, which is located at the tip of the tail and is involved in host recognition (39, 43).

The characterization of tail assembly genes of lactococcal phages has been more extensive for temperate siphophages belonging to the P335 species (27, 34, 37, 38). Because of the similarities in genome organization, the findings in this phage species can, in some cases, be used as clues toward understanding the morphology of 936-like phages. For the temperate phage Tuc2009 (P335 species), all structural proteins required for tail and baseplate assembly have been identified (27, 34, 37, 38). Genes located between those coding for the tape measure protein (TMP) and BppL (RBP) were identified as corresponding to components of the baseplate structure, located at the tail distal end. Furthermore, a gene coding for the major tail protein (MTP) was also identified at a position upstream from tmp. Between the genes coding for the MTP and those coding for the TMP in Tuc2009 are two gene products identified as gpG and gpGT, which are not present in the phage particle. These two proteins were named based on their likely role analogous to the tail assembly proteins present in coliphage lambda, a model virus belonging to the Siphoviridae family (21, 27, 47). gpGT has an essential role in lambda tail assembly, acting prior to tail shaft assembly, while the role of gpG in tail assembly is not known (21). Both gpG and gpGT are also absent from mature lambda virions (21). It has been argued that they may act as assembly chaperones (47).

A close examination of 936 genomes indicates the presence of two genes coding for gpG and gpGT-like proteins. Analysis of the phage p2 genome, closely related to that of lactococcal phage sk1 (6), revealed that the putative tail assembly proteins could correspond to gene products ORF12 and ORF13. These two genes are followed by the TMP gene corresponding to orf14, other genes coding for other structural proteins, and the RBP gene orf18. During our ongoing investigation of the structure of phage p2, we report here the cloning, expression, and crystal structure of ORF12 in order to decipher its role in the tail assembly process.

MATERIALS AND METHODS

Bacterial strains and phage.

Lactococcus lactis subsp. cremoris MG1363 (14) was grown at 30°C in M17 supplemented with 0.5% glucose (GM17). In phage p2 (31) infection experiments, 10 mM CaCl₂ was added to plates or medium. Propagation of phages and determination of the titers of the lysates were performed as described previously (12).

Intracellular detection of ORF12 during phage infection.

L. lactis MG1363 was grown in GM17 until the optical density at 600 nm reached 0.5, and then it was infected with virulent phage p2 at a multiplicity of infection of 5. Samples were taken at 5-min intervals and flash-frozen (−80°C). Cell pellets were resuspended in 10 mM Tris-Cl, pH 8.0, 1 mM EDTA, 0.3% sodium dodecyl sulfate (SDS) and lysed with a bead beater. The cytoplasmic extracts were then dosed by a standard Bradford assay, and 5 μg of each sample was migrated on a 15% SDS-polyacrylamide gel. The gel was electrotransferred (30 V) overnight at 4°C with a transblot apparatus (Bio-Rad) onto a polyvinylidene difluoride membrane (Hybond P; GE Healthcare) using Tris-glycine-methanol buffer (25 mM Tris, pH 8.3, 192 mM glycine, 10% methanol). The intracellular production of ORF12 during the infection was subsequently detected with a protein A purified anti-ORF12 antibody (Davids Biotechnologie GmbH, Germany). Briefly, the membrane was first blocked with 5% skim milk in PBST (phosphate buffer supplemented with 0.1% Tween 20) for at least 1 h on a rotational shaker. The membrane was then treated with a primary antibody, anti-ORF12, diluted 1:100,000 (in blocking buffer) for 1 h at room temperature. Following washes with PBST, the anti-ORF12 antibody was detected after a 1-h incubation with a secondary antibody, horseradish peroxidase-labeled anti-rabbit immunoglobulin G, diluted 1:100,000 in blocking buffer (Rockland Immunochemicals). After other washes in PBST, the membrane was rinsed with phosphate-buffered saline before the final detection with the ECL Plus detection kit (GE Healthcare) following the manufacturer's instructions.

ORF12 cloning, expression, and purification.

The orf12 gene of phage p2 was cloned into the Gateway destination vector pETG-20A (Arie Geerlof, EMBL, Hamburg, Germany) for protein production according to the standard Gateway protocols and using the following primers for the initial PCR: forward primer, 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTTAGAAAACCTGTACTTCCAGGGTGCAAAACAATTGAGTACAGCACG-3′; reverse primer, 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTTTATTAAATTTCTTTCTGCCACAATTCG-3′. The att sequences are in italic, the tobacco etch virus (TEV) recognition site coding sequence is in bold, and stop codons are underlined. The final construct encoded a thioredoxin fusion protein containing an N-terminal hexahistidine tag followed by a TEV protease recognition site. Protein expression was done in the Escherichia coli Rosetta(DE3)pLysS strain (Novagen). Production of the selenomethionine (SeMet)-labeled protein was performed by blocking the methionine biosynthesis pathway (11). Briefly, cells were grown at 37°C in M9 broth supplemented with 0.1 mM CaCl₂, 4 mM MgSO₄, 1.2% glycerol, 100 mg/liter lysine, 100 mg/liter phenylalanine, 100 mg/liter threonine, 50 mg/liter isoleucine, 50 mg/liter leucine, 50 mg/liter valine, 50 mg/liter SeMet, and 1 ml/liter oligonucleotide elements. When the optical density reached 0.5, protein expression was induced with 0.5 mM isopropyl-β-thiogalactoside (IPTG) and cells were left overnight at 25°C. Bacterial cells were then harvested by centrifugation at 3,300 × g for 10 min, resuspended in lysis buffer (50 mM Tris-HCl, pH 8.0, 300 mM NaCl, 0.25 mg/ml lysozyme, and EDTA-free antiprotease cocktail) (Roche), and frozen at −80°C.

Pellets were quickly thawed at 37°C followed by an incubation with shaking at 4°C in the presence of 20 mM MgSO₄ and 10 μg/ml of DNase. Cells were then sonicated and cleared by centrifugation at 21,400 × g. After filtration (0.45-μm-pore-size filter), the supernatant was loaded on a 5-ml HiTrap nickel affinity column (GE Healthcare) preequilibrated with buffer (10 mM Tris, pH 8.0, 300 mM NaCl, 10 mM imidazole). Proteins were eluted using the same buffer containing 50 mM and 250 mM imidazole. Prior to TEV protease cleavage, buffer was changed to 10 mM Tris, pH 8.0, 300 mM NaCl, 10 mM imidazole on a HiPrep 26/10 desalting (GE Healthcare) column. The cleavage was performed at 4°C overnight using a 1:10 (wt/wt) ratio of TEV protease to target protein. After cleavage, protein was recovered in the flowthrough fraction of a 5-ml HiTrap nickel affinity column. The eluted protein was further purified on a HiLoad 26/60 Superdex 200 (GE Healthcare) gel filtration column in a buffer containing 10 mM Tris, pH 8.0, and 300 mM NaCl. Purified material was visualized on a 15% SDS-polyacrylamide gel and concentrated to appropriate crystallization concentrations on an Amicon Ultra-15 centrifugal filter unit with a cutoff size of 5 kDa.

ORF12 biochemical and biophysical characterization.

Purified protein was first analyzed for size and incorporation of SeMet by matrix-assisted laser desorption ionization-time of flight mass spectrometry and trypsin peptide mass fingerprinting (Brüker Autoflex2; Daltonics Bremen, Germany). Then, analytical size-exclusion chromatography was carried out with online multiangle static light scattering on an Alliance 2695 high-pressure liquid chromatography system (Waters) on a silica gel KW804 column (Shodex) in 10 mM Tris, pH 8.0, 300 mM NaCl at a flow rate of 0.5 ml/min. The protein was loaded at a concentration of 6 mg/ml or 8 mg/ml. Detection was performed using a UV/VIS absorbance photodiode array detector (2996; Waters), a triple-angle static light-scattering detector (MiniDAWN Treos; Wyatt Technology), a quasielastic light-scattering instrument (Dynapro; Wyatt Technology), and a differential refractometer (Optilab rEX; Wyatt Technology) (37). Molecular weight and hydrodynamic radius determinations were performed with the ASTRA V software (Wyatt Technology) using a dn/dc value of 0.175 ml/g.

Crystallization, data collection, structure determination, analysis, and topological modeling.

Crystallization trials were performed using a sitting drop vapor diffusion technique implemented on a nanodrop-dispensing robot (PixSys or Honeybee-X8; Cartesian Inc.) in Greiner 288-well plates (10, 19, 40). The protein was initially screened using commercially available crystallization screens: crystal screens I and II (Molecular Dimensions Ltd.), Stura footprint screens (Molecular Dimensions Ltd.), and the NeXtal SM1 suite (Qiagen). The initial crystal hit was obtained in the Stura II footprinting screen condition number 2 (0.1 M HEPES, pH 7.5, 18% polyethylene glycol 600). Optimization screens were prepared by varying buffer pH values and precipitant concentrations based on this initial crystallization buffer hit (10, 19, 40). All crystallization plates were stored in a thermoregulated room at 291 K. After 48 h, diffracting crystals grew to a dimension of 50 × 50 × 150 μm in 0.1 M HEPES, pH 6.8, 15.9% polyethylene glycol 600 with the use of 0.1 μl of protein mixed with 0.2 μl of reservoir solution. Crystals were quick-frozen in a liquid nitrogen flux with 10% glycerol as a cryoprotectant. A complete data set was collected at the European synchrotron radiation facility (Grenoble, France) on beamline ID14-EH4. A total of 360 images were collected with an oscillation range of 1° at a wavelength of 0.9785 Å.

The data set was integrated and reduced using MOSFLM and SCALA from the CCP4 suite (8). Phases were calculated with SHELX (36). Phase extension to 2.9 Å was performed, and a partial model was built using RESOLVE (42). Cycles of manual model rebuilding were carried out using Coot (13). Refinement was performed with REFMAC (32) using TLS (Translation/Libration/Screw) segments defined by the TLS Motion Determination server (http://skuld.bmsc.washington.edu/∼tlsmd/). Threefold noncrystallographic symmetry restraints were applied throughout refinement to the homotrimer found in the asymmetric unit. The final structure was analyzed using PROCHECK (20) and MolProbity (22). A summary of the structure determination and refinement statistics is presented in Table 1. Interaction models were generated using Coot (13) and refined using Turbo-Frodo (35). An all-canonical α-helix was built with Turbo-Frodo. The helix was subsequently slightly bent at the display to adapt to the groove shape and geometrically refined with the Turbo-Frodo “refine” option. The helix was further docked in ORF12 groove manually at the display with the Turbo-Frodo FBRT option. Figures were generated with Pymol (http://pymol.sourceforge.net/) and Turbo-Frodo (35).

TABLE 1.

Data collection and refinement statistics for p2 ORF12

Parameter	Valuea
Data collection
PDB access code	3D8L
Space group	R 3 2
Unit cell (Å)	158.3 158.3 99.5 90.0 90.0 120.0
Beamline	ID14-EH4
Detector	ADSC Q315
Wavelength (Å)	0.9785
Rotation range (°)	360
Resolution range (Å)	35.0-2.90 (3.06-2.90)
No. of observations	230,325 (31,122)
No. of unique reflections	10,739 (1,556)
Completeness	99.9 (100)
Redundancy	21.4 (20.0)
I/sI	29.1 (6.4)
Rsym (%)	9.2 (44.1)
Refinement
Resolution range (Å)	30.0-2.9 (2.98-2.90)
No. of unique reflections	10,218
No. of atoms of protein per water/buffer	22,114
R/Rfree all (last shell)	25.3/20.3 (33.5/29.4)
RMSD
Bonds (Å)	0.015
Angles (°)	1.591
Mean B value (Å2)
Protein	64.43
Water	51.33
Ramachandran (%)
Favored region	92.8
Allowed regions	5.7
Generously allowed regions	1.5

^aAll values in parentheses belong to the last shell.

Protein structure accession number.

Final coordinates and structure factors were deposited in the Protein Data Bank (PDB) (http://www.rcsb.org/PDB) with code 3D8L.

RESULTS AND DISCUSSION

Production and biochemical characterization of ORF12 from the virulent lactococcal phage p2 of the 936 species.

Using well-established laboratory screening procedures (44-46), the cloning and subsequent overproduction of a SeMet-labeled ORF12 full-length protein were successfully carried out. Optimal conditions gave a yield of approximately 8 mg of soluble and purified protein per liter of E. coli culture in minimal medium. Furthermore, the protein was shown to be quite soluble, up to >10 mg/ml. The combination of good protein yield and the high level of protein solubility allowed us both to carry out biophysical characterization and to perform crystallization trials for ORF12.

Basic characterization on a 15% SDS-polyacrylamide gel identified a single 10-kDa band (expected size, 10,552 Da). The SeMet protein was also subjected to matrix-assisted laser desorption ionization-time of flight mass spectroscopy, which identified a 10.7-kDa protein. Analysis of trypsin digests by mass spectroscopy also confirmed the full incorporation of three SeMets in the protein. A first indication of the oligomerization state of ORF12 arose from protein elution off the HiLoad 26/60 Superdex 200 gel filtration column. Based on the calibration curve established for this particular column, ORF12 was identified by SDS-polyacrylamide gel electrophoresis in fractions eluting between 55 and 65 kDa (data not shown). This suggested that the protein is likely present as a hexamer in solution. ORF12 was then subjected to weight and size analysis using MALS/UV/RI (multiangle light scattering-UV absorbance-refractive index) spectroscopy (37), which confirmed a higher oligomerization state in solution (Fig. 1). The native protein (Fig. 1, blue line) was injected at a concentration of 6 mg/ml. The curve exhibited a slow decrease from the main peak, which corresponded to a mass of 58 ± 3 kDa, comparable to a theoretical mass of 64.1 kDa for a hexamer. The SeMet-labeled ORF12 was injected at 8 mg/ml and exhibited a comparable behavior but with the main peak at a mass of 158 ± 8 kDa (Fig. 1, red line), which corresponded to a 15-mer polymerization (theoretical size, 160.3 kDa). In both cases, however, all forms between the oligomer and the monomer were observed. The hydrophobic effect of the SeMet labeling explains the higher oligomerization state of ORF12 (2).

FIG. 1.

Online multiangle laser light-scattering, absorbance, and refractive index analysis of ORF12 in solution. The abscissa indicates the time of elution from the high-pressure liquid chromatography column; the left ordinate indicates the molar mass in g/mol (Da). The absorption peaks are in blue (native) and red (SeMet), and the dashed lines indicate the molar masses. The experimental masses are given in black (kDa). The native ORF12 has been injected at a concentration of 6 mg/ml, and the SeMet derivative has been injected at 8 mg/ml.

Structure of ORF12.

Crystal structure determination yielded a solution with a well-defined electron density map relative to the 2.9-Å resolution. The resulting refined structure had an R/R_free of 20.3/25.3 (Table 1 shows a complete list of data collection and refinement statistics). The asymmetric unit of these R32 (a = b = 158.3, c = 99.5; α = β = 90.0, γ = 120.0) crystals contained three monomers of ORF12 (Fig. 2A). The electron density allowed the reconstruction of all 91 amino acids (aa) in chains A and B (Fig. 2B), while the first 3 aa of chain C could not be modeled. These three molecules could be superimposed with main-chain root mean square deviation (RMSD) values of 0.32 to 0.45 Å, as determined by the SuperPose server (26), within the experimental error at this resolution. The coordinates of ORF12 were submitted to different structure analysis servers (DALI and ProFunc) (16) with a view to the identification of closely related structures and hence a putative function. No structural neighbors were identified, meaning that ORF12 has an original, previously unobserved fold.

FIG. 2.

Structure of p2 ORF12. (A) Crystallographic trimer ribbon representation with monomers identified by their letter in the PDB (3D8L). The ribbon is colored blue to red from the N to the C terminus. The surface interaction areas between each monomer are given. (B) Stereo view of ORF12 monomer, with the same coloring as in panel A. The helices are numbered 1 to 5.

ORF12 is an entirely α-helical protein, composed of five α-helices in total (H1 to H5) (Fig. 1). All helices, with the exception of helix H2, are perfectly amphiphilic helices oriented in such a way as to form a hydrophobic cleft on the inner surface of the protein. H2, on the other hand, has few amino acids involved in the cleft and a majority of outside surface residues. The fifth helix (H5) both serves to form the nonpolar cleft and serves in the interaction between two monomers. Two types of interfaces can be observed between the three monomers of the asymmetric unit in the crystals. The first two monomers (chains A and B) interact side by side through a single helix (helix H5) with the third monomer (chain C) sitting in a head-to-tail position with the second (chain B). The side-by-side helices are also oriented in a head-to-tail manner with Glu 82 of one chain H bonding with Lys 89 of the other and vice versa (Fig. 3C). The interface formed by the head-to-tail association of two monomers creates an extended hydrophobic cleft whose surface is formed in large part by the nonpolar face of helix H3. The interface is mainly formed by nonpolar interactions of residues present in the loops and strands between the defined helices. Notably, one monomer has a 10-aa loop between H4 and H5 and a 6-aa loop between H2 and H3 and the first two residues of H3 at the surface. The other presents both the N and the C termini, including the last residues of H5, plus the first two residues of H2. This later helix includes Phe 20, which forms with the C-terminal residue Trp 87 an aromatic residue pocket situated in the hydrophobic cleft (see Fig. S1 in the supplemental material).

FIG. 3.

View of the spirals formed by ORF12 crystal packing. The crystal is formed from such spirals packed side by side. (A) Side view of the two spirals side by side. Each monomer is colored brown, violet, and yellow repeatedly. (B) View of the spirals rotated by 90° around the vertical axis. The gross dimensions of the spiral are displayed. (C) Stick representation of a stretch of the two side-by-side-positioned spirals illustrating the dense network of charge interactions, but loose packing, between the two spirals. The color code is blue for positively charged residues, red for negatively charged residues, pale green for semipolar residues, green for methionines, yellow for aliphatics, and violet for aromatics.

The three monomers of the asymmetric unit have very few buried residues in their core (probe radius, 1.6 Å; no cutoff). Chains A, B, and C have 87, 86, and 84 surface-exposed amino acid residues, respectively (with a special note concerning chain C, which has three residues missing), indicating that virtually all residues are either accessible to solvent or available for interactions among themselves in the crystal packing.

As mentioned above, the monomers in the asymmetric unit reveal two types of interactions, which were analyzed for the individual monomer positions in the asymmetric unit trimer using the EMBL-EBI Protein Interfaces, Surfaces and Assemblies (PISA) server (http://www.ebi.ac.uk/msd-srv/prot_int/cgi-bin/piserver). The largest surface interaction area was found to involve monomers B and C in the trimer, with an interface of 642 Ų. A smaller surface was observed between the interactions described above for H5 of chains A and B (149 Ų), while A and C had virtually no interactions (Fig. 2A). The larger surface area is involved in the B-C-type interaction, which forms the aromatic pocket at the interface of the two monomers.

Generating symmetry mates of the ORF12 trimer leads to the formation of two large spirals of compactly packed monomers (Fig. 3). A cross section of the spirals revealed an inner solvent channel of 36 Å with a complete diameter of 80 Å. The two spirals display weak mutual contacts, namely, those described above between monomers A and B. The contacts within the spiral are like those described for monomers B and C, with interaction surface areas of 709 Ų for monomer B and its interacting partner, monomer A, in a symmetry mate. The B-C interface corresponds to 642 Ų, while C and its symmetry mate A have an interface of 754 Ų. Therefore, the average value for this type of interaction is 702 Ų, corresponding to 11% of the total surface area of the ORF12 monomer. This kind of surface area is comparable to that observed for Fab/protein interactions (23) and is large enough to be significant for true biological interactions (17). We therefore propose that a crystallographic spiral corresponds to the elongation of the oligomers (15-mers in the case of SeMet ORF12) observed in solution, triggered by the decrease in solubility resulting from the increase of the concentration of precipitant occurring during the crystallization process.

ORF12 is a nonstructural protein but is expressed during phage infection.

Phage p2 was purified through a CsCl gradient to obtain highly concentrated preparations (10¹¹ to 10¹² PFU/ml), which were used to infect L. lactis MG1363. Then, a time course infection was performed and samples were taken at intervals. Intracellular extracts were tested for the production of ORF12 (Fig. 4). The phage protein ORF12 was first detected at 15 min after the beginning of the infection and at the expected size of 10 kDa, confirming that it is a late-expressed phage protein (5). The production of ORF12 then peaked at 30 min, and its concentration started to decrease coinciding with lysis of the host culture. This is in agreement with previous data that estimated the latency period of phage p2 (the time from infection to release of new phage progeny) as being up to 30 min (15). As expected, we could not find ORF12 in the structure of phage p2; thus, its role in the phage infection process was still unknown.

FIG. 4.

Intracellular detection of ORF12 during phage p2 infection of L. lactis MG1363. Samples were taken at various time intervals after phage infection, and ORF12 was detected by Western blot analysis. For ORF12, 5 ng of purified protein and, for p2, 1 × 10¹⁰ PFU of CsCl-purified phages were loaded, which corresponded to 7.8 ± 1.5 μg.

Putative function of ORF12: model of the TMP repeat segment and its binding to ORF12.

Each spiral formed by the crystal packing of the nonstructural phage p2 protein ORF12 displays several noteworthy features. Firstly, nonpolar crevices are located regularly at the inner face of the spiral, within a continuous cleft. In contrast, the external face of the spiral exhibits polar residues facing the solvent. Secondly, within the nonpolar cleft, a highly repetitive motif of aromatic residues can be identified (Fig. 5A).

FIG. 5.

Model of complex between the TMP hydrophobic helix and ORF12 spiral. (A) Sphere representation of a segment of four ORF12 modules in the spiral. The hydrophobic patches are visible in the center of the twisted spiral, in a twisted crevice, formed of aromatic residues. The color code is blue for positively charged residues, red for negatively charged residues, pale green for semipolar residues, green for methionines, yellow for aliphatics, and violet for aromatics. (B) View of the TMP segment 777 to 818, modeled as a curved α-helix. All atoms are colored orange, and the aromatic side chains are colored pink. The periodicity of the aromatic residues of TMP coincides with that observed in the ORF12 spiral, as outlined by the red arrows. (C) Model of a complex between the TMP segment 777 to 818 and four ORF12 modules in the spiral. Color coding is the same as for panels A and B. (D) Sequence of p2 TMP (1 to 999). The repeat area (522 to 898) is underlined. The segment chosen in the above model is identified in a yellow box with aromatic residues shown in violet.

Based on its position in the genome, and by comparison with other Siphoviridae, we hypothesized that ORF12 may play a role in tail assembly. Due to the peculiar characteristics of the TMP (TMP/ORF14) of phage p2, with two hydrophilic domains at each sequence extremity, but with a long hydrophobic helix, likely insoluble by itself (Fig. 5D), we examined the possibility that ORF12 might be a chaperone complexing to the hydrophobic central part of TMP and maintaining it in solution before its assembly with the MTP to form the phage tail. The TMP acts as a ruler or template that measures length during tail assembly (18). Besides the N (1 to 522 aa) and C (898 to 999 aa) termini of the TMP, which are likely globular and interact with the portal protein on the N-terminal side and with the baseplate on the C-terminal side, the middle part exhibits a 40-aa repeat of evenly spaced aromatic residues (Fig. 5D). Furthermore, this type of repeat (although varying in length from phage to phage) appears to be a common feature of TMPs in Siphoviridae (3, 6, 27).

Since this repeat region of the TMP molecule is predicted to be an α-helix, we generated a small 42-mer helical model of this repeat to verify if it could bind within the twisted hydrophobic crevice identified in the large twisted spiral structure of ORF12 (Fig. 3). This structure-structure docking revealed that the amphiphilic helix has aromatic residues correctly spaced to fit into the aromatic residue binding pockets of the ORF12 spiral (Fig. 3B and C; see also Fig. S2 in the supplemental material). These residues are observed on the nonpolar surface of the hydrophobic crevice (Fig. 3A), while the other side of the spiral displays a solvent-exposed polar side bearing many charged residues.

A possible function of ORF12 might therefore be to cover the central segment of p2 TMP composed of repeated hydrophobic motifs, to keep them in solution. The sequences of the N terminus of TMP (1 to 521 aa) and its C terminus (898 to 999 aa) resemble more globular proteins and are probably soluble by themselves. We hypothesize that maintaining the TMP in solution would help the assembly of MTPs around the TMP hydrophobic helix, a scheme seen in the virulent Bacillus siphophage SPP1 tail structure (33).

Conclusion.

Using structural genomic approaches, it is expected that the knowledge of the three-dimensional structure of a protein of unknown function may reveal its function if its fold resembles that of a protein of known function. The structure of p2 ORF12 being novel, such an approach was not possible. However, because the number of open reading frames in phages is limited and gene organization may be conserved, genomic comparisons may be the source of functional hypotheses, which can then be checked in silico or experimentally. Here, we suggest, based on strong topological observations, that the nonstructural ORF12 of phage p2 (and possibly other siphophages) may serve as a chaperone of the TMP central domain to maintain it in solution and present it to MTPs to facilitate the tail assembly process. The perfect match of ORF12 spiral (observed as an oligomer in solution) hydrophobic patches with the aromatic residues of TMP repeats leads us to postulate that a complex between the TMP residues 522 to 898 and a spiral of the ∼70-mer might be stable long enough in solution to allow MTP to approach TMP and the tail to be formed.