|Home | About | Journals | Submit | Contact Us | Français|
Bacteriophage HK97 maturation involves discrete intermediate particle forms, comparable to transitional states in protein folding, before reaching its mature form. The process starts by formation of a metastable prohead, poised for exothermic expansion triggered by DNA packaging. During maturation, the capsid subunit transitions from a strained to a canonical tertiary conformation and this has been postulated to be the driving mechanism for initiating expansion via switching hexameric capsomer architecture from skewed to 6-fold symmetric. We report the subnanometer electron-cryomicroscopy reconstruction of the HK97 first expansion intermediate prior to any crosslink formation. This form displays 6-fold symmetric hexamers, but, unexpectedly, capsid subunit tertiary structures exhibit distortions comparable to the prohead forms. We propose that release of this coat subunit strain acts in synergy with the first crosslinks to drive forward maturation. Finally, we speculate that the energetic features of this transition may result from increased stability of intermediates during maturation via enhanced inter-subunit interactions.
Virus maturation corresponds to a transition from an initial non-infectious, often fragile assembly product to an infectious and robust virion (Veesler and Johnson, 2012). Initial subunit interactions occur under conditions where the assembling entities have an association energy that favors assembly over disassembly, but that is near equilibrium to allow “self-correction” of misassembled subunits through annealing (Katen and Zlotnick, 2009). As viruses require sturdy stability to survive in the extra-cellular environment, they undergo a staged assembly process due to a mechano-chemical reorganization program, encoded in the capsid structure, that governs events underlying maturation.
Assembly and maturation of dsDNA phage capsids are tightly regulated processes, at both the genetic and biochemical levels, exhibiting conserved features in all Caudovirales and in some eukaryotic viruses such as herpesviruses (Johnson, 2010; Steven et al., 2005; Veesler and Cambillau, 2011; Veesler and Johnson, 2012). Moreover, the striking conservation of the coat subunit fold observed in all tailed phages and herpesviruses as well as some archeal viruses, suggests that it is derived from a common ancestor preceding the divergence of eukaryotes, bacteria and archea (Baker et al., 2005; Heinemann et al., 2011; Veesler and Cambillau, 2011; Veesler and Johnson, 2012). The lambdoid dsDNA phage HK97 constitutes an accessible model system for studying maturation of such viruses due to its well-characterized genetics and ease of handling. Its capsid maturation pathway involves discrete intermediate particle forms, comparable to transitional states in protein folding, that can be isolated using a combination of molecular biology and biochemical techniques.
The HK97 capsid precursor protein is a fusion of the scaffolding protein (δ-domain, residues 2–103) and of the coat subunit (residues 104–385) that forms a mixture of hexameric and pentameric capsomers upon expression. In vivo, 415 coat subunits (60 hexamers and 11 pentamers) assemble with a dodecameric portal and ~60 copies of the viral protease to form the first icosahedral particle termed Prohead-1 (Fig. 1). Activation of the viral protease results in digestion of the scaffolding domains and auto-digestion to produce small peptide fragments that diffuse out of the particle to yield Prohead-2. The two prohead particle forms exhibit distorted tertiary subunit structures readily recognized by the bent spine helix and the twisted P-domain β-sheet. The quaternary structures of these particles also display distortions from canonical symmetry as the hexameric capsomers are skewed, displaying only 2-fold symmetry (Gertsman et al., 2009; Huang et al., 2011). These structural distortions are believed to be induced by the scaffolding domain interactions when capsomers are formed and are stabilized by quaternary interactions following δ-domain proteolysis in Prohead-2 (Gertsman et al., 2010a). Prohead-2 is thus a metastable intermediate, trapped in a local free energy minimum that is primed for transition to a lower-energy conformation in response to small perturbations. Initiation of dsDNA packaging triggers Prohead-2 expansion, resulting in the formation of successive maturation intermediates (termed expansion intermediates) characterized by an increase of capsid diameter, a reduction of the shell thickness and a “curing” of the hexon asymmetry (Gan et al., 2006; Wikoff et al., 2000). Moreover, H/D exchange experiments coupled to mass spectrometry (HDXMS) suggested that in intermediates later than the Prohead-2 form, the capsid subunit tertiary structure transitions to a relaxed state similar to those observed in the Balloon and Head-2 crystal structures (Gertsman et al., 2010a; Gertsman et al., 2010b). The large conformational changes occurring during formation of the first expansion intermediate (EI-1) makes it cross-link competent with isopeptide bonds forming immediately through an autocatalytic mechanism between residues Lys169, on the E-loop of one coat subunit, and Asn356, on the P-domain of an adjacent subunit in a neighboring capsomer (Duda et al., 1995; Popa et al., 1991; Wikoff et al., 2000). Crosslink formation is not concerted, making EI-1 particles transient and the population heterogeneous. Crosslinking promotes formation of the subsequent expansion intermediates and has been proposed to modulate the capsid structural reorganization by biasing thermal motions via a Brownian ratchet mechanism based on the capture of the subunit E-loops (Lee et al., 2008; Ross et al., 2005). The in vivo maturation endpoint, Head-2 bears 415 crosslinks with a chainmail topology that dramatically stabilizes the capsid enclosing the genome packaged at near liquid-crystalline density (Helgstrand et al., 2003; Wikoff et al., 2000).
Here, we used an HK97 subunit mutation that prevents formation of crosslink or comparable non-covalent interactions and an expression system that produces virus-like particles indistinguishable from authentic proheads but with the portal replaced by a twelfth coat subunit penton. The mutation stops maturation at the EI-1 intermediate generating a homogeneous population of these particles without E-loop “chainmail” interactions. Comparing these particles with mature Head-2 allows the mechanical role of the Brownian ratchet in maturation to be identified. We determined the subnanometer structure of the crosslink-free EI-1 particle with electron cryo-microscopy (CryoEM) employing single particle protocols. The reconstruction, unexpectedly, reveals that coat subunit monomers exhibit distortions comparable to those observed in the prohead forms although the hexamers are approximately 6-fold symmetric. The observed coat subunit conformations suggest that release of their structural strain adds an energetic assist to crosslinking, driving capsid maturation forward with multiple energetic components. In addition, the structure suggests that the exothermic nature of capsid maturation (Galisteo and King, 1993) is a consequence of enhanced quaternary interactions that stabilize the downstream intermediates.
A construct encoding an E-loop truncation of the coat subunit was used to produce homogenous virus-like particles stalled at the EI-1 stage of maturation and lacking crosslinks. We carried out an icosahedral reconstruction of this maturation intermediate using 21,964 particle images and single-particle techniques. The resulting structure has a resolution of 9.3 Å (Supplemental Fig. 1), exhibits pronounced icosahedral facets and its overall size and morphology are consistent with a previously reported reconstruction obtained at lower resolution (Lee et al., 2008; Ross et al., 2005). The capsid forms a 43 Å thick and 600 Å wide (along 5-fold axes) hollow shell made of 420 coat subunits arranged with a T=7 laevo symmetry and with protruding hexamers and pentamers (Fig. 2A–C). The resolution of the reconstruction is qualitatively demonstrated by observed secondary structure elements of the subunits and the straightforward segmentation of individual proteins, either visually or by automated procedures. We further improved the quality of the map by averaging the density of the seven subunits within the icosahedral asymmetric unit.
We initially generated an EI-1 pseudo-atomic model by fitting the mature Head-2 X-ray coordinates in the reconstruction, as tertiary structure distortion was not anticipated (Gertsman et al., 2009). Previous HDXMS experiments suggested that early HK97 expansion intermediates share the relaxed major capsid protein conformation with the late maturation intermediates as well as with the final Head-2 (Gertsman et al., 2010a; Gertsman et al., 2009; Gertsman et al., 2010b; Wikoff et al., 2000). Rigid-body docking of the seven individual subunits forming the icosahedral asymmetric unit revealed a striking discrepancy between the EM reconstruction and the Head-2 atomic coordinates in the spine helices and the adjacent P-domain β-sheets (Fig. 3A). As the Prohead-2 coat subunits are characterized by a twisting around the P-domain β-sheet along with bending of the spine helix, we used this model to fit into the EI-1 reconstruction. This model dramatically improved the agreement with the density in these regions without compromising agreement with the rest of the density (Fig. 3A–B). The CryoEM density also shows that the coat protein hexamers are approximately 6-fold symmetric (Fig. 1B), in agreement with a previous EI-1 reconstruction (Lee et al., 2008; Ross et al., 2005).
The discrepancy between this EI-1 reconstruction and previous mass spectrometry data can be explained by considering the constructs used to produce the virus-like particles in the two studies (Gertsman et al., 2010a; Gertsman et al., 2010b). While we used the E-loop deletion mutant to overexpress EI-1 in the current study, the previously characterized “EI-1” harbored a wild-type coat subunit E-loop allowing immediate initiation of crosslink formation. The latter particle form must therefore correspond to EI-2, which is EI-1 with crosslinks but that has not transitioned to the Balloon particle (Lee et al., 2008).
The EI-1 hexons are 140 Å wide and 43 Å thick in our pseudo-atomic model, corresponding to an intermediate configuration between Prohead-2 (123 Å wide and 55 Å thick) and Head-II (157 Å wide and 32 Å thick). The reorganization of the subunits between Prohead-2 and Head-2, from approximately radial to approximately tangential orientation relative to the capsid surface, is associated with a 2-fold increase in the buried surface area at the interfaces between coat subunits. Accordingly, we observe that the rotation undergone by the subunits to reach the EI-1 state accounts for a substantial portion of the increased buried surface area.
Subunit interactions established at 3-fold and quasi 3-fold axes are known to be unchanged during maturation and to serve as anchoring points allowing preservation of capsid integrity (Gertsman et al., 2009; Wikoff et al., 2000). Although side chain positioning can not be achieved at the resolution of our reconstruction, our model is fully compatible with retention of the salt bridges established at 3-fold contact points between residues Arg194 and Glu363 as well as between Arg347 and Glu344 (Gertsman et al., 2010a). This observation, along with the observed approximate 6-fold symmetry of the coat subunit hexamers, validates the quality of the pseudo-atomic model and further reinforces the conclusions drawn from it.
No attempt was made to model the conformation of the E-loop in the previously reported EI-1 studies due to the limited resolution of the reconstructions (Lee et al., 2008; Ross et al., 2005). While in our case the E-loop is absent from the expressed molecule, the density clearly shows the location of the truncated loop and by inference the trajectory of the E-loop, if it were there (Fig. 4). They extend toward the capsid exterior, parallel to the spine helix of the neighboring subunit that is within the same hexon or penton, and interact with it though probably less extensively than in Prohead-2. These interactions may be involved in maintaining the distorted conformation of the coat subunit in EI-1, prior to disengagement as seen in later maturation intermediates. The crosslink formation requires a major repositioning of the E-loop to place the Lys169 side chain near the Gln356 side chain of a subunit belonging to a neighboring capsomer to form the isopeptide bond. As a result, the E-loop conformational change disrupts the intra-capsomeric interactions established with the spine helix probably allowing refolding of the coat subunit to reach its relaxed conformation observed in EI-2 and subsequent particle forms.
It was previously suggested that EI-1 represents the maturation ground state and that crosslinking acts via a ratcheting mechanism to shift the global equilibrium toward the balloon and then the Head-2 forms (Lee et al., 2008; Ross et al., 2005). The results presented here show that EI-1 still embodies a significant degree of conformational strain, due to the bent spine helix and twist of the subunits around the P-domain β-sheet. The structure implies that the transition to the relaxed coat subunit conformation is tightly correlated with formation of the first crosslinks and/or non-covalent quaternary interactions established by the tip of the E-loops. Indeed, the ability to arrest maturation at the EI-1 stage constituted a unique opportunity to demonstrate that in the absence of crosslinks or such non-covalent E-loop interactions the capsid still resides in a stressed conformation harboring distorted coat subunits despite the formation of ~6-fold symmetric hexamers. The observation that expansion can be induced by various physico-chemical stimuli (such as pH change or iso-butanol) and the characteristic two-state transition between Prohead-2 and EI-1 evidenced by SAXS measurements indicate that the latter has a lower energy than its precursor (Gertsman et al., 2010b; Lee et al., 2005). However, our results indicate that EI-1 is still storing energy in its structure, probably to ensure that, in combination with formation of the first crosslinks, the maturation moves forward (Fig. 1) and reaches the EI-2 particle form (with numerous crosslinks and coat subunits with canonical tertiary structure). Disruption of the spine helix/E-loop interactions facilitates Brownian motion-mediated sampling of the conformational space by uncrosslinked E-loops and formation of additional crosslinks between capsomers, making the maturation process irreversible. It should be noted that expansion from Prohead-2 to EI-1 results in a ~10 % increase in particle dimension but further expansion does not occur in the overall particle dimensions until 60% of the crosslinks are formed. When the expansion occurs from EI-1 to the Balloon it is also a two-state transition with no detectable intermediates as demonstrated by time-resolved SAXS experiments (Lee et al., 2008).
Differential scanning calorimetry (DSC) studies of bacteriophage P22 maturation revealed that its expansion is strongly exothermic (Galisteo and King, 1993). The striking conservation of the coat subunit fold among tailed phages as well as of many aspects of their maturation suggests that exothermic expansion is a common feature of such viruses (Johnson, 2010; Veesler and Cambillau, 2011; Veesler and Johnson, 2012). During expansion, coat subunits establish an increasing number of interactions with each other to stabilize the capsid concomitantly with dsDNA packaging in order to withstand the remarkable pressure generated by the genome (Fuller et al., 2007; Veesler and Johnson, 2012). The gradual enhancement of subunit intertwining increases by a factor ~2 the total buried surface area involving the subunits from a given icosahedral asymmetric unit as well as the number of residues participating to these contacts during the transition from Prohead-2 to Head-2. During this transition both hydrophobic and polar interactions are increased, but the proportions of each of these are dramatically modified in favor of hydrophobic stabilization (Table 1). However, electrostatic complementarity of the coat subunit A domains seems to play a major role in the reorganization observed. In the skewed Prohead-2 hexamers, subunits B and E adopt a specific conformation involving only tenuous interactions of their A-domains with the anti-clockwise located neighboring subunits (view from the capsid exterior) in comparison to the other four subunits (Fig 5A–F). In contrast, all the Head-2 hexamer subunits are characterized by the formation of identical interactions locking the capsomer conformation by the high level of A-domain charge complementarity (Fig. 5G–J). Polar and covalent interactions create thus specific contacts and provide the directionality for rearranging subunit-subunit interactions. It is worth mentioning that release of the strain in coat subunit pentamers during expansion provokes a dramatic reorganization of the hexamers/pentamers interactions. Therefore, the energetically unfavorable coat subunit refolding event occurring during capsid expansion is likely compensated by the enthalpy gain from the increase in polar interactions and by enhancement of hydrophobic interactions in addition to the contribution of chainmail crosslinks. This drives the maturation process forward, ensuring its irreversibility (beyond the EI-1 stage) by a progressive stabilization of the different expansion intermediates likely explaining the exothermic nature of bacteriophage expansion. Work is still in progress to address the discrepancy between the EM and SAXS data discussed here and some previously reported DSC results (Ross et al., 2005; Ross et al., 2006).
Overall, the program encoded in the initial assembly products that directs particle maturation to the fully mature particle, is slowly being discerned as a remarkable integration of chemistry, thermodynamics and mechanobiology that evolution has tuned to a very high level.
We used a mutated version of the HK97 gp5 coat subunit in which residues 159–171 are replaced with residues APGD (Gertsman et al., 2009). This construct harbors an E-loop shortened at its distal part preventing the formation of crosslinks and of part of the quaternary interactions. The capsid protein (gp5) and the protease (gp4) were coexpressed using E. coli BL21pLysS cells induced with 0.4 mM IPTG at 28°C overnight. After harvesting, cells were lysed using the Bugbuster reageant (Merck) supplemented with 20 μg/mL of DNAse I and 10 mM MgSO4. Cell debris were removed by centrifugation and capsids were precipitated in presence of 0.5 M NaCl and 6% polyethylene glycol 8000. Remaining Prohead-1 particles were disassembled by incubation in 2M KCl, 100mM CHES, pH 9.5 for 5–6 h before purification on a 10–30% glycerol gradient. Prohead-2 expansion was triggered by incubation in a buffer Na-acetate pH4.0, 300 mM NaCl during 6h at RT. The pH was raised to 7.5 and an anion exchange chromatography (5mL FF DEAE) was carried out before exchanging the buffer of the particles by ultracentrifugation to 10 mM Tris pH7.5, 40 mM NaCl.
Purified EI-1 capsids were prepared for cryoEM analysis by placing 3 μl of sample on a C-flat carbon-coated grids (Protochips, Inc.) previously glow-discharged in a Solarus plasma cleaner (Gatan, Inc.). Grids were manually blotted before plunging into liquid ethane and subsequently transferred to liquid nitrogen in which they were stored. Data were acquired on a Tecnai F20 Twin transmission electron microscope operated at 200 keV, using a dose of 20 e-/Å2, a nominal magnification of 62,000 and a nominal underfocus ranging from 1.0 to 3.5 μm. One data set containing 1,714 images was automatically collected using the Leginon data collection software (Suloway et al., 2005) using a Tietz F415 4K x 4K pixel CCD camera (15 μm pixel).
We extensively relied on the Appion processing pipeline for initial processing of the images (Lander et al., 2009). The contrast transfer function for each micrograph was estimated using CTFind3 and applied to each micrograph before particle extraction (Mindell and Grigorieff, 2003). We manually masked all the micrographs to exclude the particles lying on the carbon regions before carrying out an automated particle picking using FindEM (Roseman, 2004). Capsids were extracted using a box size of 704 pixels and binned by a factor of 2 for processing yielding a stack of 24,394 particles. 3D reconstruction was performed using Frealign (Grigorieff, 2007) including 21,964 particle images and an initial model obtained by low-pass filtering at 50 Å the EI-2 pseudo-atomic model previously reported (Lee et al., 2008). The resolution of 9.3 Å for the EI-1 reconstruction was assessed by calculating the Fourier shell correlation at a cutoff of 0.143 (Grigorieff and Harrison, 2011). The amplitudes of the resulting refined structure were adjusted with the SPIDER software package to more closely resemble those of an experimental low-angle X-ray scattering data (Frank et al., 1996; Gabashvili et al., 2000). Averaging of the 7 subunits belonging to the icosahedral asymmetric unit has been carried out using the RAVE package (LSQMAN, MAMA, IMP and AVE) (Kleywegt, 2001).
We generated a pseudo-atomic model of the EI-1 capsid by rigid-body fitting the Prohead-2 atomic coordinates (PDB 3E8K) in the reconstruction using UCSF Chimera (Goddard et al., 2007) before carrying out an energy minimization imposing strict icosahedral symmetry with CNS 1.3 (Brunger, 2007; Brunger et al., 1998). Vizualization was carried out with Coot (Emsley et al., 2010). Interface and interaction analyses were done using ViperDB (Carrillo-Tripp et al., 2009). Electrostatic surface potential calculations were done using pdb2pqr (Dolinsky et al., 2004) and APBS (Baker et al., 2001).
The resolution is estimated to be 9.3 Å at FSC=0.143.
We thank Andrew Routh for critical reading of the manuscript. This project was supported by grants from the NIH (R01 AI040101), NCRR (2P41RR017573-11) and the NIGMS (9 P41 GM103310-11) as well as a FP7 Marie-Curie IOF fellowship (273427) attributed to D.V. Molecular graphics and analyses were performed with the UCSF Chimera package. Chimera is developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from the National Institutes of Health (National Center for Research Resources grant 2P41RR001081, National Institute of General Medical Sciences grant 9P41GM103311).
The EI-1 cryoEM map and averaged icosahedral asymmetric unit has been deposited to the EM data bank under accession numbers.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.