Crystal structure of Bacillus subtilis SPP1 phage gp22 shares fold similarity with a domain of lactococcal phage p2 RBP

Abstract

SPP1 is a siphophage infecting the gram-positive bacterium Bacillus subtilis. It is constituted by an icosahedric head and a long non-contractile tail formed by gene products (gp) 17–21. A group of 5 small genes (gp 22–24.1) follows in the genome those coding for the main tail components. However, the belonging of the corresponding gp to the tail or to other parts of the phage is not documented. Among these, gp22 lacks sequence identity to any known protein. We report here the gp22 structure solved by X-ray crystallography at 2.35 Å resolution. We found that gp22 is a monomer in solution and possesses a significant structural similarity with lactococcal phage p2 ORF 18 N-terminal “shoulder” domain.

Introduction

Phages represent a highly diverse group of viruses infecting bacteria and are the most populated biological entity on earth.¹ The vast majority of them belong to the Caudovirales order and is composed by a double-stranded DNA enclosed in an icosahedral capsid to which is attached a tail. More than 60% of known phages are member of the Siphoviridae family characterized by the presence of a long non-contractile tail, as is the case of SPP1. SPP1 is a virulent Bacillus subtilis phage encapsulating its genome in a 60-nm wide isometric capsid connected to a 160-nm long tail.^2,3 Host infection is initiated by the binding of its tail-tip to YueB, a membrane protein with a large ectodomain protruding out of the thick peptidoglycan layer.^4,5 This specific and irreversible interaction triggers a cascade of events resulting in DNA ejection into the B. subtilis cytoplasm.

The tail components of SPP1 have been thoroughly identified, and the tail overall structure was studied by electron microscopy, making it possible to assign most of its components (gp17–gp21) to electron densities.² However, a large volume of electron density remained unassigned at the tail tip distal end. This fact and the identification of five gene products coded downstream of gp21 (gp22, gp23, gp23.1, gp24, gp24.1), the most distally assigned tail component, led to hypothesize that these gps might form the tip of the tail.² To bring further structural insights to prove or disprove this hypothesis, we have undertaken the structural study of these orphan gps.

We report in this contribution the crystal structure of SPP1 gp22 at 2.35 Å resolution. We describe the overall structure and discuss its oligomerization state based on the crystal lattice organization and on static plus dynamic light scattering measurements performed in solution. The gp22 fold is similar to lactococcal phage p2 ORF18 (p2 RBP) shoulder domain, suggesting a possible common origin for these two protein domains.

Results and Discussion

Gp22 crystallization and structure determination

Gp22 crystallized in the orthorombic space group P2₁2₁2₁ with unit cell parameters a = 42.9 Å, b = 64.1 Å, and c = 101.05 Å (Table I). The diffraction limit was 2.35 Å and data quality allowed us to find the six Se sites present in the two molecules of the asymmetric unit and then to phase and build a model. The final model was refined to R_work and R_free values of 20.66 and 24.36%, respectively. The two monomers of one asymmetric unit are very similar with a r.m.s.d value of 0.5 Å over the first 120 Cα positions [Fig. 1(A)].

Tabel I.

Data Collection and Refinement Statistics

SPP1 gp22
Data collectiona
Beamline	BM-14 (ESRF)
Space group	P212121
Unit cell dimensions (Å)	a = 42.9, b = 64.1, c = 101.05
Wavelenght (Å)	0.97872
Resolution (Å)	54.138–2.35 (2.48–2.35)
Rsymb (%)	9.0 (46.3)
Mn(I)/σI	27.3 (6.3)
Completeness (%)	100 (100)
Redundancya	14.2 (14.6)
Refinementa
Resolution (Å)	39.68–2.35 (2.57–2.35)
Rwork/Rfree (%)	20.7/24.4 (21.9/27.6)
No. of reflections	12,092 (2846)
Mean isotropic B-factor (Å2)	33.0
r.m.s.d. bond lengths (Å)/angles (°)	0.015/1.53

^aValues in parentheses refer to the highest resolution shell.

^bR_sym =∑ (|I(h,i) − I(h)|))/∑ (I(h,i)

Figure 1.

Overall phage SPP1 gp22 structure. A: Ribbon representation of the gp22 dimer in the asymmetric unit. The C-terminal part of monomer C (a symmetry related gp22 monomer) is depicted in purple. B: Stereo ribbon representation of the gp22 monomer. C: Gp22 topology diagram. The disulfide bridge is indicated by a green bar. D: Stereo ribbon representation of the superimposition between SPP1 gp22 and phage p2 ORF18 shoulder domain. Gp22 is depicted in blue and ORF 18 shoulder domain in gold. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Overall structure

Gp22 is a 146 residue-long all-β protein formed by two small three-stranded antiparallel β sheets (βB and βC) and one large five-stranded mixed β sheet (βA). The three β sheets are arranged in two superimposed layers, one of these being formed by βA and the other by βB and βC [Fig. 1(A–C)]. The gp22 fold can thus be described as a pseudo-β-sandwich. The electron density map allowed us to build all residues from the first glycine of our construct to residue 121/120 for both molecules (A/B) of the asymmetric unit. Only a small part of each C-terminus was ordered and thus modeled in the final structure: residues 126–137 and 126–132 in molecule C (a gp22 monomer related to the molecule A by a crystallographic symmetry operation) and B, respectively. Indeed, only a few residues are visible in a region that is stabilized both by hydrogen bonds and van der Waals contacts established with a neighboring monomer in the crystal lattice and involving strand 7 in βC, strand 9 in βA and the loop connecting strands 9–10 in βA. It should be noted that the relative orientation between each monomer of the asymmetric unit and their C-terminal part is different, probably reflecting a high flexibility of the gp22 C-terminus. An intramolecular disulfide bridge is observed in each monomer between cysteine residues 76 and 86 located in strand 9 of β-sheet A and in the turn following strand 9, respectively.

Oligomeric state

In the crystal, each asymmetric unit is composed by a gp22 dimer. The interface buried surface area between the two corresponding monomers is of 1515.7 Ų with 703.8 Ų and 811.9 Ų contributed by monomer A and B, respectively. This interface is comparable in size to typical values found in protein–protein complexes⁶ and involves salt bridges, hydrogen bonds and van der Waals contacts. However, SEC/MALS/QELS/RI measurements revealed that gp22 is monomeric in solution with a measured mass of 15.3 kDa (theoretical monomer M_W= 16.7 kDa) and a hydrodynamic radius of 2.3 nm. This strongly suggests that gp22 is monomeric and that the observed dimer is only crystallographic. Other phage proteins were shown to oligomerize and adopt their correct stoichiometry only upon partner binding, as is the case of SPP1 phage adaptor (gp15) and stopper (gp16) proteins.⁷ We thus cannot rule out the possibility that gp22 is oligomeric in an in vivo context.

Biological role

Gp22 does not have any significant sequence similarity to any other protein hindering thus the possibility to postulate a function based on sequence conservation. However, a DALI search for structurally similar proteins provided a very interesting result: the gp22 structure superimposes onto the shoulder domain of the receptor-binding protein (RBP, ORF 18) of the lactococcal phage p2.^8,9 The structural superimposition between gp22 and the chain A of PDB 2WZP yields a Z score of 5.7 with a r.m.s.d value of 3.7 Å over 91 equivalent Cα positions. The two proteins exhibit very close pseudo β-sandwich folding pattern with only subtle differences [Fig. 1(D)]. Nevertheless, the role of the p2 ORF 18 N-terminus being only structural (anchoring the RBP to the baseplate),⁸ it is difficult to propose a function based on this observation. It should be noted that the RBP is trimeric in its native state^8,9 and that the shoulder domain oligomerization interface is formed by a three-helix bundle (each monomer contributing an α-helix). As this α-helix is absent in gp22, this oligomerization strategy should not be conserved for this protein in vivo. This illustrates a frequently observed phenomenon in phages that is the reuse of a single protein module to fulfill different functions. Finally, we did not find any unambiguous way to fit the gp22 structure into the distal part of the tail tip electron microscopy reconstruction, due to the low resolution (∼20 Å) of this contrast EM density.

Materials and Methods

Gp22 cloning, expression, and purification

The SPP1 gp22 nucleotidic sequence was PCR amplified and cloned by Gateway recombination into the pETG-20A vector.¹⁰ The resulting construct encoded a N-terminal fusion with a His₆-tagged thioredoxine followed by a TEV (tobacco etch virus) protease cleavage site. The plasmid was transformed in Escherichia coli T7 Express I^q pLysS strain (New England Biolabs) and expression was carried out at 25°C overnight with 0.5 mM IPTG in a minimal medium containing 50 mg/L seleno-methionine for production of seleno-methionine-labeled protein.¹¹ After harvesting, cell lysis was done by addition of 0.25 mg/mL lysozyme, a freezing/thawing cycle and sonication. Gp22 purification was performed via a first Ni²⁺-affinity step using 50 mM imidazole for elution. After desalting and TEV protease cleavage [10:1 (w/w) protein:TEV protease ratio, 4°C overnight], a second Ni²⁺-affinity step was performed followed by a gel-filtration on a preparative Superdex 200 26/60 column. MALDI-TOF mass spectrometry analysis was used to check protein integrity as well as Se-Met incorporation.

Crystallization and structure determination

We performed an initial nanocrystallization screening¹² in 96-well Greiner plates with a protein concentration of 7.5 mg/mL. We optimized a promising hit, obtained in condition 73 of the Structure screen (0.01M CoCl₂, 0.1M MES pH 6.5, 1.8M (NH₄⁺)₂SO₄, Molecular Dimensions Limited), by varying precipitant concentration and pH. The gp22 crystal used for data collection was obtained in 0.1M MES pH 7.2, 0.01M CoCl₂, 2.05M (NH₄⁺)₂SO₄ and flash frozen in mother liquor supplemented with 14% glycerol. Data were collected at the European Synchrotron Radiation Facility (ESRF, Grenoble, France) on BM14 (SeMet peak data set at the Se K edge, λ = 0.97872 Å). Data processing and scaling were done with XDS,¹³ POINTLESS¹⁴ and SCALA.¹⁴ A random selection of 7.5% of the data (test set) was assigned for calculation of the free R factor¹⁵ and was not included in the refinement. Phasing was performed by the single anomalous diffraction method using the phenix.autosol wizard.^16,17 Resulting phases were the starting point for automatic model building with ArpWarp.¹⁸ Model building and refinement were done with Coot,¹⁹ phenix.refine,²⁰ and BUSTER-TNT.²¹ TLS groups definition was assisted by the TLSMD server.²² Final refinement statistics and model quality are summarized in Table I. Structure analysis was helped by the protein interfaces, surfaces and assemblies server (PISA),²³ the protein–protein interface server,²⁴ the DALI server²⁵ as well as with promotif2.²⁶ Figures were generated with Chimera²⁷ and topdraw.²⁸ The coordinates have been deposited to the PDB with accession code 2XC8.

Light scattering measurements

The oligomerization state and size of gp22 in solution were studied by MALS/QELS/UV/RI coupled on-line with an analytical SEC column, as described.^29–31 MALS, QELS, UV, and RI measurements were achieved with a MiniDawn Treos (Wyatt technology), a Dynapro (Wyatt technology), a Photo Diode Array 2996 (Waters) and an Optilab rEX (Wyatt technology), respectively. The SEC column was a 15-mL KW-803 column (Shodex) run at 0.5 mL/min on an Alliance HPLC 2695 system (Waters) in a buffer containing 10 mM Hepes pH 7.5, 150 mM NaCl.

Glossary

Abbreviations:

r.m.s.d.

root mean square deviation

RBP

receptor binding protein

MALS/QELS/UV/RI

on-line multiangle laser light scattering, quasi-elastic light scattering, UV absorbance and refractive index detectors

EM

electron microscopy