ABSTRACT
Tailed double-stranded DNA (dsDNA) bacteriophages, herpesviruses, and adenoviruses package their genetic material into a precursor capsid through a dodecameric ring complex called the portal protein, which is located at a unique 5-fold vertex. In several phages and viruses, including T4, Φ29, and herpes simplex virus 1 (HSV-1), the portal forms a nucleation complex with scaffolding proteins (SPs) to initiate procapsid (PC) assembly, thereby ensuring incorporation of only one portal ring per capsid. However, for bacteriophage P22, the role of its portal protein in initiation of procapsid assembly is unclear. We have developed an in vitro P22 assembly assay where portal protein is coassembled into procapsid-like particles (PLPs). Scaffolding protein also catalyzes oligomerization of monomeric portal protein into dodecameric rings, possibly forming a scaffolding protein-portal protein nucleation complex that results in one portal ring per P22 procapsid. Here, we present evidence substantiating that the P22 portal protein, similarly to those of other dsDNA viruses, can act as an assembly nucleator. The presence of the P22 portal protein is shown to increase the rate of particle assembly and contribute to proper morphology of the assembled particles. Our results highlight a key function of portal protein as an assembly initiator, a feature that is likely conserved among these classes of dsDNA viruses.
IMPORTANCE The existence of a single portal ring is essential to the formation of infectious virions in the tailed double-stranded DNA (dsDNA) phages, herpesviruses, and adenoviruses and, as such, is a viable antiviral therapeutic target. How only one portal is selectively incorporated at a unique vertex is unclear. In many dsDNA viruses and phages, the portal protein acts as an assembly nucleator. However, early work on phage P22 assembly in vivo indicated that the portal protein did not function as a nucleator for procapsid (PC) assembly, leading to the suggestion that P22 uses a unique mechanism for portal incorporation. Here, we show that portal protein nucleates assembly of P22 procapsid-like particles (PLPs). Addition of portal rings to an assembly reaction increases the rate of formation and yield of particles and corrects improper particle morphology. Our data suggest that procapsid assembly may universally initiate with a nucleation complex composed minimally of portal and scaffolding proteins (SPs).
INTRODUCTION
Double-stranded DNA (dsDNA) tailed bacteriophages, herpesviruses, some dsDNA archaeal viruses, and adenoviruses have a pseudoicosahedral capsid with a portal protein incorporated at a unique 5-fold symmetry axis (1–3). The portal protein is a ring of 12 identical subunits constituting a channel for the active packaging and subsequent ejection of the dsDNA viral genome (4–8). The existence of a single portal ring is crucial to the formation of infectious virions. Evidence from dsDNA bacteriophages, such as T4, SPP1, and Φ29, and from herpes simplex virus 1 (HSV-1), indicates that portal proteins play a key role in the nucleation and assembly of proper procapsids, thereby ensuring one portal ring per capsid (9–13).
The current hypothesized mechanism for portal ring incorporation suggests that the portal interacts with the assembly chaperone, scaffolding protein (SP). This scaffolding protein-portal protein complex acts as a nucleus around which the coat proteins (CP) are copolymerized with SP, thereby ensuring that all procapsids have only one portal. This theory is supported by ample data from in vivo and in vitro studies of several bacteriophages and viruses. For instance, in vivo studies have demonstrated interactions between scaffolding and portal proteins (12, 14–16). Additionally, in vitro studies using bacteriophage Φ29 have shown that the yield and rate of PC assembly increase in the presence of portal protein, consistent with its role as an assembly nucleator (10). Similarly, data from in vitro assembly of HSV-1 indicates that the portal gets integrated into the procapsids during initiation, highlighting the involvement of the portal during early steps of virus assembly (9).
The work of Bazinet and King (4) showed that the presence or absence of the bacteriophage P22 portal protein in vivo did not affect PC assembly kinetics in radioactive pulse-chase experiments, leading the authors to propose that the P22 portal protein might not nucleate PC assembly (4, 17). Thus, P22 was suggested to employ a different mechanism to incorporate portal protein at a 5-fold vertex. This observation is not unique to P22, as phage SPP1 also does not show a change in vivo in PC assembly kinetics in the absence of its portal, although for SPP1 the portal protein does increase the yield of properly sized procapsids (12).
Recently, we established an in vitro assembly assay for bacteriophage P22 in which portal protein can be coassembled into procapsid-like particles (PLPs) at a single vertex (18). Furthermore, SP was shown to interact with portal monomers and catalyze the oligomerization of dodecameric portal rings. This interaction possibly leads to formation of a scaffolding protein-portal protein nucleation complex that results in one portal per P22 procapsid. Here, we present evidence that P22 portal protein, in a fashion similar to that of the portal proteins of other dsDNA viruses, can act as an assembly nucleator. As observed for bacteriophage Φ29 and HSV-1, we found that the presence of P22 portal protein is able to increase the rate of PLP assembly in vitro. Furthermore, the role of P22 portal protein as an assembly initiator was confirmed by its ability to correct the morphology of particles assembled at a high SP:CP ratio. Our results highlight that the assembly initiator function of portal proteins is well preserved among dsDNA viruses.
(This article was submitted to an online preprint archive [19].)
RESULTS
For dsDNA bacteriophages, including T4, SPP1, Φ29, and for the eukaryotic virus HSV-1, portal proteins have a demonstrated role in the nucleation of PC assembly (9–13). On the other hand, in vivo studies with bacteriophage P22 showed that portal protein was not involved in the initiation of PC assembly and had no effect on the rate of procapsid assembly (4, 17). This led to the hypothesis that for P22 the portal protein is not involved in the nucleation of PC assembly. However, subsequent studies showed that excess portal protein in vivo led to aberrant structures, as well as petite T=4 capsids, clearly indicating that portal protein affects morphology and suggesting that it may indeed nucleate P22 PC assembly (17). Having recently established conditions where P22 portal rings can be incorporated into PLPs assembled in vitro (18), we decided to reinvestigate if portal protein can nucleate PLP assembly in vitro. In vitro assembly reactions allow precise control over the solution conditions, including the concentrations of portal, coat, and scaffolding proteins, and accurate determination of assembly rates by monitoring the light scattering the reactions. We also determined if the addition of portal rings can correct morphology defects.
Determination of optimal assembly conditions.High concentrations of SP lead to aberrant structures in vitro. In reactions with high scaffolding protein concentrations, coat protein limits the reaction, resulting in overnucleation. Consequently, mostly incomplete particles are formed because there is not enough CP to complete the shell (20, 21). In the experiments described here, we wanted to control the assembly conditions such that proper PLPs or aberrant particles would result, as desired. A HEPES acetate buffer [20 mM HEPES and 70 mM potassium acetate (KAc)] is critical for the successful incorporation of portal protein (18). This buffer is so favorable for assembly that the critical concentration of CP for assembly is 0.075 mg/ml, or 1.6 μM, 4-fold lower than previously determined (see reference 22 and data not shown). Since the concentration of SP is important for proper PC assembly, new optimal concentrations for proper or aberrant particle assembly were determined. Assembly at increasing scaffolding protein:coat protein ratios (SP:CP) with coat protein held at 6.4 μM was monitored by light scattering at 500 nm (Fig. 1A). The relative slope of the early elongation phase of PLP assembly was plotted.
Effect of high-concentration scaffolding protein on PLP assembly. (A) PLP assembly kinetics in the presence of varied concentration of scaffolding protein monitored by light scattering at 500 nm. The scaffolding protein (3.0 μM to 44.7 μM) was incubated in a cuvette for 5 min in 20 mM HEPES (pH 7.5) and 70 mM KAc buffer. The assembly reaction was started by adding 6.4 μM coat protein monomers. The experiment was performed three independent times with different preparation of scaffolding protein. Shown are representative traces. (B) Relative slope of the elongation phase directly after the initial lag plotted against the SP:CP molar ratio. (C) PLPs assembled in vitro in the presence of various concentration of scaffolding protein were run on a 1.0% agarose gel. The scaffolding protein concentration and the SP:CP ratio are indicated below the gel. (D) Negative-stain electron micrographs of in vitro assembled PLPs at varied SP:CP ratios. Bar, 100 nm. Red arrows indicate half-PLPs, and yellow arrows indicate two half-procapsids fused together.
Figure 1 shows that at low SP:CP ratios, scaffolding protein is limiting for assembly. Increased concentrations of SP yielded more particles, until at higher scaffolding protein input concentrations, where SP is in large excess (>26.8 μM or SP:CP ratio of 4.18), the kinetics of assembly plateaued (Fig. 1A and B). An aliquot of the assembly product obtained from each in vitro reaction was run on a native agarose gel, and the particles were visualized via transmission electron microscopy (TEM) (Fig. 1C and D). At low SP:CP ratios (≤1.39), a distinct PLP band is observed on the gel, and these particles had normal spherical procapsid-like morphology. However, at high SP:CP ratios (>1.39) the intensity of the PLP band decreased, and a band corresponding to assembly intermediates appeared. Examination of electron micrographs of these particles revealed the presence of bowl-shaped “half-procapsid” structures along with normal PLPs, as seen previously at a high SP:CP ratios (20, 21). These half-procapsids increased in number with increasing SP concentration. Thus, setting the SP:CP ratio to ≤1.39 leads to PLPs, and when the scaffolding protein is increased to an SP:CP ratio of >1.39, overnucleation occurs, resulting in aberrant particles.
Portal protein rings can nucleate procapsid assembly.We tested the effect of portal rings on the kinetics of PLP assembly, as was done previously for Φ29 portal protein (10), to assess if the rings can nucleate assembly. In all of the experiments, the procapsid conformation of portal rings (PC portal) was used (23). The SP (9 μM) was preincubated with varied concentrations of PC portal rings (0.0167 μM to 0.1 μM rings) in a cuvette. Because high portal expression in vivo affects particle morphology (17), the portal concentration was kept low in the reactions. The cuvette was incubated at room temperature (RT) for 5 min, and assembly was initiated by the addition of 6.4 μM CP monomers. The rate of PLP assembly in the presence of PC portal rings was monitored by the time-dependent increase in the scattered light at 500 nm, measured in counts per second (CPS).
Both the rate of assembly and yield of PLPs increased with increasing portal ring concentration to 0.05 μM, after which increased portal concentration decreased the rate of assembly (Fig. 2A). The initial rate of each assembly reaction was determined from the slope of the elongation phase directly after the initial lag phase (24) and plotted as the rate relative to the rate of the assembly reaction conducted in the absence of portal protein versus the concentration of portal rings (Fig. 2A, inset).
Portal protein rings can nucleate PLP assembly. PLP assembly kinetics in the presence of varied concentrations of PC portal rings was monitored by light scattering at 500 nm. (A) The scaffolding protein (9 μM) and portal rings (0.0167 μM to 0.1 μM) were incubated in a cuvette for 5 min in 20 mM HEPES (pH 7.5) and 70 mM KAc buffer. The assembly reaction was started by adding 6.4 μM coat protein monomers. The concentration of each protein used in the assembly reaction is indicated. The experiment was performed three independent times with different preparations of coat protein monomers. Shown are representative traces. (Inset) Relative slope of the elongation phase directly after the initial lag plotted against portal ring concentration. (B) The scaffolding protein (20.8 μM) and PC portal rings (0.0267 μM to 0.2 μM) were incubated in a cuvette for 5 min in 20 mM HEPES (pH 7.5) and 70 mM KAc buffer. The assembly reaction was started by adding 6.4 μM coat protein monomers. The concentration of each protein used in the assembly reaction is indicated. The experiment was performed three independent times with different preparations of coat protein monomers. Shown are representative traces. (Inset) Relative slope of the elongation phase directly after the initial lag plotted against portal ring concentration.
The positive effect of portal rings on the rate of capsid assembly suggests that bacteriophage P22 portal rings can act in assembly nucleation. However, high concentrations of PC portal rings (> 0.1 μM) decreased the rate and yield of assembly (Fig. 2A). A similar result was observed when portal ring concentration was further raised to 0.2 μM (data not shown). Two models could explain the negative effect of high portal ring concentration on assembly. First, the decrease in the rate of PLP assembly at high concentration of PC portal rings (>0.1 μM) could be due to interaction of portal protein with SP (18), decreasing the amount of free SP for interaction with CP. Second, the portal protein could be causing too many nuclei to form, decreasing the concentration of CP available to assemble into PLPs. In the first instance, addition of extra SP should relieve the inhibition of assembly, while in the second instance, we anticipated no effect on assembly by the addition of extra SP.
PC portal protein can act as an assembly nucleator even at a high scaffolding protein:coat protein ratio.To further understand the role of P22 portal protein as an assembly nucleator, we investigated the effect of portal protein in assembly reactions when SP is in excess and causing overnucleation (Fig. 1). Varied concentrations of PC portal rings (0.0267 μM to 0.2 μM) were incubated with SP held at 20.8 μM (SP:CP ratio = 3.25) for 5 min in a cuvette at RT. Coat protein monomers were added, and the PLP assembly was observed by light scattering. As shown in Fig. 2B, even at this high SP:CP ratio, the presence of PC portal rings steadily increased the assembly kinetics. Maximal effect on the assembly rate was observed with 0.053 μM or 0.075 μM PC portal rings in the reaction. Furthermore, at high concentrations of PC portal rings (>0.1 μM) the rate of PLP assembly gradually declined, similar to the effect noted when the scaffolding protein:portal protein ratio was low (Fig. 2A). In these experiments, the best portal ring concentration (0.075 μM) is unlikely to affect the SP concentration. Rather, the data are consistent with portal rings affecting the nucleation complex. Thus, our results show that there is a fine requisite balance between the critical protein components for assembly to occur properly.
Portal protein decreases aberrant structures.In other systems, the portal protein can correct morphology defects that occur during PC assembly (12, 16, 22, 25, 26). We tested if the presence of PC portal rings can complete the half-procapsids and aberrant structures observed at high SP:CP ratios (Fig. 1). The partial PLPs are kinetically trapped structures; in this case, the result of too many nuclei forming is that CP becomes limiting and the particles are unable to complete (21, 27).
We hypothesize that the portal protein will affect the initiation of assembly by building a different nucleation complex with SP and CP monomers, thereby allowing more PLPs to form closed particles with a portal integrated at a single vertex, even at high SP:CP ratios. To test this, 0.053 or 0.1 μM PC portal rings were incubated with SP at 20.8 μM for 5 min at RT (Fig. 3A). As a mock control, SP (20.8 μM) was incubated in buffer alone. The assembly reaction was initiated by adding 6.4 μM CP monomers. This assembly reaction was allowed to reach completion, and the PLPs were purified over a Superose 6 gel filtration column. When Alexa Fluor 488 dye (AF488)-labeled portal rings were used in the reactions, the labeled rings could be visualized in the PLP peak, as shown previously (Fig. 3B and C) (18). The purified PLPs were analyzed by TEM and prepared as described in Materials and Methods. Over 2,000 PLP particles from each reaction were analyzed to quantify the number of complete and half/partial PLP structures in the micrographs. Consistent with our hypothesis, the PLP particles assembled in the presence of PC portal rings at high SP:CP ratios were more uniform and spherical than those assembled without portal rings (Fig. 3E). Additionally, the percentage of aberrant or half-procapsids was significantly reduced from 43% to 26% in the presence of 0.053 μM PC portal rings and to 32% in the presence of 0.1 μM PC portal rings, concomitant with an increase in normal-sized and closed PLPs (Fig. 3D). Thus, our data indicate that, similarly to those of other dsDNA viruses (HSV-1, Φ29, T4 and SPP1), the bacteriophage P22 portal protein can act as an assembly initiator, a feature of portal protein that is likely conserved among all dsDNA viruses.
P22 portal protein completes half-procapsids assembled at a high SP:CPs ratio. (A) Alexa Fluor 488 dye (AF488)-labeled PC portal rings (0.053 μM or 0.1 μM) were incubated with scaffolding protein (20.8 μM) in 20 mM HEPES (pH 7.5) and 70 mM KAc buffer for 5 min. CP monomers (6.4 μM) were added to initiate capsid assembly process. The reaction was allowed to reach completion at room temperature (RT) for 4 h, and the in vitro assembled PLPs were separated by gel filtration chromatography. The purified PC peaks were analyzed by SDS-PAGE. (B) Superose 6 elution profiles of assembly reactions. Rx1, CP and SP with 0.053 μM AF488-labeled portal rings; Rx2, CP and SP with 0.1 μM AF488-labeled portal rings; and Rx3, CP and SP alone (mock reaction). (C) SDS-PAGE gel of the purified PLP peak fractions from Rx1 to Rx3. The presence of labeled PC portal rings in these fractions was visualized using a PharosFX Plus molecular imager. (D) Percentage of uniform and spherical procapsids (black bars) and aberrant or half-procapsids (gray bars) assembled in vitro in the absence (no PC portal) or presence of unlabeled PC portal rings (0.053 μM or 0.1 μM). More than 2,000 particles were analyzed for each reaction. (E) Representative electron micrographs of the in vitro assembled PLPs assembled in the absence or presence of unlabeled PC portal rings (0.053 μM or 0.1 μM protomer). Red arrows indicate normal spherical PLPs and yellow arrows indicate aberrant or half-procapsids. Scale bar, 100 nm.
DISCUSSION
Portal proteins are essential to formation of infectious virions in the tailed dsDNA phages, herpesviruses, and adenoviruses (1–3). The portal is the conduit for the DNA into the heads during packaging and out of heads during infection. As such, portal protein is a potential antiviral therapeutic target (3). Recently, we found conditions that allowed the portal protein to be incorporated into assembling PLPs (18). This advance allowed us to revisit the role of portal protein in initiation of P22 assembly. Here, we showed that the rate of assembly is increased by the addition of portal rings, as expected if portal rings are involved in initiation of PLPs (Fig. 2). We also demonstrated that the addition of portal rings could improve PLP morphology (Fig. 3).
The assembly of bacteriophage P22 is one of the most thoroughly studied assemblies of this class of viruses. In vivo and in vitro scaffolding protein is essential to incorporation of the portal protein (18, 28). Scaffolding protein is also required for the minor ejection proteins to be assembled into PCs (28) (Fig. 4A). The ejection proteins are located on the portal barrel in the mature virion (29). The barrel only forms after DNA packaging (23). However, the ejection proteins are assembled into PCs even in the absence of the portal protein (30). How correct numbers of ejection proteins are assembled into PCs, and how and when they localize to the portal barrel, are unknown. Perhaps the ejection proteins are also involved in the nucleation step, allowing them to be assembled near the portal. This is an experiment we are now poised to do.
Cartoon comparing the in vivo and in vitro assembly reactions. (A) In an infected cell, the assembly proteins are synthesized and then self-assemble. Scaffolding protein is essential to the incorporation of portal rings, the ejection proteins, and for the growing procapsid shell, where scaffolding protein coassembles with the coat protein. We propose that the nucleation complex involves a portal ring, along with an undetermined number of scaffolding and coat proteins. If the ejection proteins are members of the nucleation complex or added later in PC assembly is not known. The ejections proteins are located on the portal barrel in mature virions, but how they arrive there is not understood. The barrel is not formed in portals in solution or in PCs, and it is induced to fold by DNA packaging. (B) In vitro assembly reactions at optimal protein concentrations with or without portal protein. In either condition, completed particles with the correct morphology are formed. (C) In vitro assembly conditions where coat protein is limiting. In the absence of portal protein, the reaction overnucleates. This yields kinetically trapped incomplete particles. The addition of portal rings to the reaction corrects nucleation so that PLPs form.
As with incorporation of HSV-1 and phage Φ29 portal rings into PLPs in vitro, the protein concentrations need to be carefully balanced (9, 10). At low SP:CP ratios (≤1.39), PLP assembly with or without portal rings occurs efficiently (Fig. 4B). However, we observed that in the absence of portal rings at high SP:CP ratios (>1.39), PLP assembly is affected, leading to the production of aberrant particles. The excess SP topples the balance of the critical components, resulting in rapid nucleation and formation of kinetically trapped metastable half-particles (Fig. 4C). The presence of an optimal concentration of portal rings in assembly reactions carried out at high SP:CP ratios rebalances the reaction (Fig. 4C).
Over 30 years ago, Bazinet and King presented evidence that the absence of portal protein did not affect the kinetics of P22 PC assembly in vivo (4). We believe this result could be due to (i) a 3-min radioactive amino acid pulse that is nearly as long as the time PCs require to assemble in vivo (31, 32), which could obscure changed assembly rates, or (ii) the sampling after the addition of chase was sparse, so perhaps any changes in kinetics were simply not observed. We plan to revisit these in vivo experiments in the future.
MATERIALS AND METHODS
Purification of PC portal rings, SP, and coat monomers for assembly reactions.Portal rings in the procapsid conformation (PC portal) were purified and assembled as described previously (18, 23). Briefly, to purify and assemble PC portal rings, full-length His-tagged portal protein was purified by metal-chelating affinity chromatography using high-affinity nickel-nitrilotriacetic acid (Ni-NTA) resin (GenScript), and concentrated to ∼200 mg/ml using a 30-kDa Millipore-Amicon centrifugal filter. The protein was incubated at room temperature for 24 h to promote oligomerization and purified by size-exclusion chromatography (SEC) over a Superose 6 Increase GL 10/300 gel filtration column (GE Healthcare) (23). The portal protein concentration in all experiments is given as the 12-mer ring concentration. Scaffolding protein (SP) was purified as previously described (33). Coat protein monomers were prepared from empty procapsid shells. These shells were generated from an “assembler” plasmid containing only the genes for coat and scaffolding proteins, thereby eliminating the possibility of contaminating portal protein (34). The shells were denatured in 6.75 M urea and 20 mM HEPES (pH 7.5) at 2 mg/ml final concentration for 30 min at room temperature (RT). The denatured coat protein monomers were diluted (1:1) with 20 mM HEPES (pH 7.5) and extensively dialyzed against 20 mM HEPES (pH 7.5) at 4°C. Aggregates and uncontrolled shell assembly products were removed by ultracentrifugation at 221,121 × g at 4°C for 20 min in a Sorvall S120AT2 rotor.
Procapsid-like-particle assembly kinetics monitored by light scattering.For the assembly reactions, scaffolding protein was preincubated in a cuvette with PC portal rings for 5 min in 20 mM HEPES (pH 7.5) with potassium acetate (KAc) added to 70 mM final concentration. The assembly reaction was initiated by the addition of coat protein monomers in 20 mM HEPES (pH 7.5) to a final concentration of 6.4 μM. The final scaffolding and portal protein concentrations were varied, as described in the figure legends. The light scattering at 500 nm was monitored using a Horiba FluoroMax 4 spectrofluorometer with the cuvette held at 20°C. The excitation and emission monochromators were set at 500 nm, with slits set to 3.5 nm. A 2.0-optical density (OD) neutral density filter was placed in the excitation beam, and the scattered light at 500 nm was recorded in counts per second (CPS) for 30 to 60 min.
Agarose gel electrophoresis of in vitro assembled PLPs.The yield and oligomeric state of the PLPs and particles assembled with varied concentrations of scaffolding protein (3.0 to 44.7 μM) were analyzed by agarose gel electrophoresis. Twenty to thirty microliters of an in vitro assembly reaction were loaded onto 1.0% SeaKem agarose gel in 1× TAE (40 mM Tris base, 20 mM acetate, 1 mM EDTA). The gels were run at 100 V for ∼75 to 90 min. The agarose gels were stained with Coomassie blue and imaged with a Bio-Rad Gel Doc imager.
Negative-stain electron microscopy.Samples to be viewed by transmission electron microscopy (TEM) were prepared by applying 3 to 5 μl of in vitro assembled PLP (with or without PC portal rings) onto 300-mesh carbon-coated copper grids (Electron Microscopy Sciences). Samples were absorbed for 1 min and washed with 2 to 3 drops of water. Finally, the grids were stained with 1% uranyl acetate solution (wt/vol). The grids were visualized using an FEI Technai G2 Spirit BioTwin transmission electron microscope equipped with an AMT 2k XR40 charge-coupled device (CCD) camera at ×68,000 magnification.
ACKNOWLEDGMENT
This work was supported by NIH grant GM076661 to C.M.T.
FOOTNOTES
- Received 3 February 2019.
- Accepted 12 February 2019.
- Accepted manuscript posted online 20 February 2019.
- Copyright © 2019 American Society for Microbiology.
