Insight into DNA and Protein Transport in Double-stranded DNA Viruses: The Structure of Bacteriophage N4

SUMMARY

Bacteriophage N4 encapsidates a 3,500 amino acid-long DNA-dependent RNA polymerase (vRNAP), which is injected into the host along with the N4 genome upon infection. The three-dimensional structures of wild-type and mutant N4 viruses lacking gp17, gp50, or gp65 were determined by cryo-electron microscopy. The virion has an icosahedral capsid with T = 9 quasi-symmetry that encapsidates well-organized dsDNA and vRNAP. The tail, attached at a unique pentameric vertex of the head, consists of a neck, twelve appendages, and six ribbons that constitute a non-contractile sheath around a central tail tube. Comparison of wild-type and mutant virus structures in conjunction with bioinformatics established the identity and virion locations of the major capsid protein (gp56), a decorating protein (gp17), the vRNAP (gp50), the tail sheath (gp65), the appendages (gp66), and the portal protein (gp59). The N4 virion organization provides insights into its assembly, and suggests a mechanism for genome and vRNAP transport strategies utilized by this unique system.

Introduction

Transport of a virus’s genetic information across cell membranes into the host cytoplasm is a fundamental event in a viral life cycle. Animal viruses typically enter host cells via endocytosis, followed by disassembly of the capsid, and subsequent release of the viral genome. Tailed bacteriophages, on the other hand, leave their capsids outside the cell and have evolved elaborate DNA delivery strategies. Phages with contractile tails (Myoviridae) have a tail tube that penetrates the cell membrane followed by injection of the genome into the cytoplasm. T4, for example, attaches to the cell surface via fibers that are attached to a baseplate at the tip of the tail. Upon successful adsorption, conformational changes in the baseplate trigger contraction of a tail sheath, which then drives the tail tube through the cell envelope.¹ T4 transports a 169 kb genome into the cell cytoplasm in 30 seconds. This is one of the most efficient DNA transport processes known.² It has been proposed that the energy required for this process is provided by pressure inside the head due to the tightly-packaged, negatively-charged DNA. ^{3; 4} Phages with long, non-contractile tails (Siphoviridae), such as T5 with a 1,900 Å-long tail, insert a long, hollow tube into cell membranes through which DNA is delivered into the cell.^{5; 6} T5 initially transfers 8% of its 122 kbp genome. Then, pre-early phage-encoded proteins participate in the transport of the rest of the genome.⁷ For phages with short, non-contractile tails (Podoviridae), DNA injection is more problematic since the tail is too short to reach the cell’s cytoplasm.⁸ To overcome this problem, T7 ejects phage-encoded proteins along with the phage DNA during infection. These proteins have been proposed to form a channel that spans the entire cell envelope, providing a conduit through which DNA is transported into the cytoplasm.^{8; 9} However, the exact mechanism of DNA delivery into the cytoplasm utilized by tailed phages is not known.

N4, a member of Podoviridae, is a lytic dsDNA bacteriophage that infects E. coli K12 strains.¹⁰ The virus’s linear 72 kbp genome encodes three tRNAs and 72 open reading frames (ORFs) (GenBank accession number EF056008). N4 is unique among dsDNA bacteriophages in that it packages a phage-encoded RNA polymerase (vRNAP) into the capsid,¹¹ and ejects the protein into the host cell, presumably preceding the first ~500 bp of genomic DNA. Then, vRNAP catalyzes RNA synthesis initiated at a promoter present in the first 500 bp; this process pulls the next 10 to 40 kbp of the genome out of the virus and into the host (A. A. Demidenko and L. B. Rothman-Denes, unpublished data). N4 is the only known bacteriophage that does not require the activity of the host RNA polymerase for transcription of its early genes.¹² Transport of the remaining DNA requires RNA synthesis catalyzed by the N4-encoded RNA polymerase II, a product of early transcription^{13; 14} (A. A. Demidenko and L. B. Rothman-Denes, unpublished data).

The isometric head of N4 has a diameter of approximately 700 Å. A short, non-contractile tail is attached to the head.¹⁵ The mature virion contains at least 10 gene products (gp) (Table 1).¹⁶ Gp50, the vRNAP, consists of 3,500 amino acids and is present in the virion.¹⁶ The vRNAP N-terminal domain (~998 aa) is required for injection of the first 500 bp of the genome (A. A. Demidenko and L. B. Rothman-Denes, unpublished data), the central domain (~1100 aa, mini-vRNAP) possesses the RNA synthesizing properties of vRNAP,¹⁷ and the C-terminal domain (~1,400 aa) is required for vRNAP encapsidation.¹⁷ Gp56 is thought to be the major capsid protein based on an estimate of 520 ± 30 copies per virion.¹⁶ This number suggests a T = 9 symmetry for an icosahedral head, although no virus structure with T = 9 symmetry has been observed previously. Gp65 is the receptor-binding protein required for N4 adsorption to the host (J. McPartland and L. B. Rothman-Denes, unpublished data). Little is known about the other seven structural proteins.

Table 1.

Structural Proteins in Bacteriophage N4

Gene Product	Amino Acids (Met)	# Copies/Virion			Proposed Role
		35S-Met	SDSa	CryoEMb
17	279 (5 c)	147 ± 49	160 ± 20	175	Decorating protein
50	3,500 (114)	4 ± 1	1 or 2		vRNA Polymerase
51	644 (17)	20 ± 2	18 ± 1
52	150 (1 or 2d)	49 ±7	33 ± 5
54	299 (7)	17 ± 4	43 ± 9
56	401 (16c)	548 ± 45	520 ± 30	535	Major capsid protein
59	764 (23c)	17 ± 4	12 ± 2	12	Portal proteind
65	1,382 (38)	6 ± 1	5 ± 1	6	Non-contractile tail sheath
66	556 (16)	25 ± 6	27 ± 5	36	Appendagee
67	236 (3)	7 ± 1	13 ± 5

^aData are from Falco et al. ¹⁶.

^bEstimated from the cryoEM structures in this study.

^ccontains N-terminal methionine.

^dnot know if the mature protein contains the N-terminal methionine

^ePredicted from the sequences using bioinformatics.

Here, we report the three-dimensional, cryo-electron microscopic (cryoEM) structures of wild-type N4, as well as mutant virions lacking gp17, gp50, or gp65. Comparison of wild-type and mutant virus structures established the identity and locations of the major capsid protein (gp56), a decorating protein (gp17), the vRNAP (gp50), the tail sheath protein (gp65), the appendages (gp66), and the portal protein (gp59) in the virion. A T = 9 icosahedral head and a non-contractile tail sheath, as observed in bacteriophage N4, have not been reported previously. Based on the structure of the tail and the location of the vRNAP, we propose a mechanism for transport of the DNA and vRNAP from the virion into the host cell.

Results and Discussion

The N4 capsid has T = 9 quasi-symmetry

The three-dimensional, cryoEM structure of the N4 head was determined to 14 Å resolution by imposing icosahedral symmetry on mature virus particles (Table 2). The icosahedral head has a diameter of 695 Å between 2-fold vertices and 790 Å between 5-fold vertices with an average shell thickness of 27 Å. The surface of the capsid displays a complex pattern of protrusions representing the major capsid protein, gp56, and a decorating protein, gp17. The major capsid protein, gp56, forms both pentamers and hexamers with a T = 9 lattice (Fig. 1a, c). This is more apparent in the mutant that is missing the gp17 decorating protein because the latter has only three (not nine) copies per icosahedral asymmetric unit (Fig. 1a and next section). Six of the nine monomers in each icosahedral asymmetric unit reside in one hexamer located in a general position, two additional monomers are part of two neighboring hexamers coincident with the icosahedral 3-fold axes, and the ninth monomer is a part of the pentamer at each 5-fold vertex. Both the generally-positioned hexamers and the icosahedral 3-fold hexamers show good 6-fold symmetry. The centers of capsomers are approximately 140 Å apart. All the major capsid proteins have a small surface protrusion, separated by about 37 Å within pentamers and within hexamers. Based on the T = 9 triangulation, the total number of major capsid protein (gp56) copies is 535 per capsid (540 minus 5 copies since one pentamer is replaced by the portal or connector assembly). This number is in excellent agreement with the biochemically-determined values of 520 ± 30¹⁶ and 548±45 (Table 1).

Table 2.

Statistics of Image Reconstructions

	Number of Boxed Particles (Used)	Defocus, μm	Symmetry	Resolution, Å
Wild-type	2848 (2489)	1.13 – 3.75	icosahedral	14
			C5	19
			C1	29
gp17-minus	2555 (2490)	1.64 – 3.61	icosahedral	15
vRNAP-minus	2831 (2506)	1.02 – 3.64	icosahedral	14
			C5	20
gp65-minus	2587 (2096)	1.42 – 3.20	icosahedral	19
			C1	30

Figure 1.

Structures of the mature and gp17-minus N4 head. (a) Surface rendering of the N4 head viewed down an icosahedral 2-fold axis (top). The density is contoured at 3σ. The asymmetric unit is outlined by a black triangle connecting 5-, 3-, and 2-fold symmetry axes. One pentamer at the icosahedral 5-fold axis, one hexamer in a general position, and two hexamers at the icosahedral 3-fold axes are colored in orange, green, and purple, respectively. The scale bar represents 300 Å. Two independent hexamers and the pentamer in the gp17-minus virus are enlarged in order to show their similarities (bottom). (b) Central cross-section of the N4 head reconstruction. Density connecting the capsid to the most external DNA layer is indicated by a red arrow. The orthogonal icosahedral 2-fold axes are indicated with white arrows. (c) A schematic diagram of the T = 9 icosahedral lattice. The T = 9 hexagonal lattice and the icosahedral asymmetric unit are shown. The positions of nine major capsid protein subunits in an icosahedral asymmetric unit are labeled as 1-9. Subunit 1 forms one fifth of a pentamer (orange), subunit 2-7 form a general hexamer (green), each of subunit 8 and 9 form one sixth of a hexamer at the icosahedral 3-fold axis (purple). (d) Difference density between the wild-type and gp17-minus viruses. For clarity, only the front half of the map is shown. A T = 9 icosahedral lattice and asymmetric unit are shown. The three molecules of gp17 per icosahedral asymmetric unit are labeled A, B, and C. (e) The difference density between the wild-type and gp17-minus (green) superimposed with gp17-minus virus (blue to white with increasing radius). (f) Stereo view showing the fit of IgG domains into the difference density at two different contour levels. The grey caged density represents 3 σ, and a green solid density represent 5 σ above mean The three IgG domains (Protein Data Bank accession number 1FNH) were fitted as two separate rigid bodies (DI-II in magenta and DIII in blue). Three gp17 molecules are labeled as in (d). A symmetric unit (top) and a side view of the molecule B and C (bottom) are shown.

There are two major classes of icosahedral viruses both containing hexameric or pseudo-hexameric capsomers. In the first class, capsomers contain three copies of a monomer that has two independent jelly-roll folds. The centers between capsomers are separated by ~72 Å. In the second class, which includes the tailed phages T4, ϕ29, ε15, P22, and herpes simplex virus, the capsomers contain six monomers each consisting of HK97-like folds and whose centers are separated by ~135 Å.^18-22 The distance between the N4 capsomer centers is approximately ~140 Å suggesting that N4 may belong to the HK97-like group of viruses. Although the N4 major capsid protein, gp56, does not have detectable sequence similarity to other tailed phage capsid proteins, the predicted secondary structure of gp56 has a similar pattern as that of the HK97 and T4 major capsid protein crystal structures.^{22; 23} The similarity of the fold remains uncertain given that the cryoEM reconstruction is only at 14 Å resolution, yet fitting of the HK97 capsid protein structure into the N4 capsid cryoEM density shows that the overall shape and dimensions of gp56 monomers and capsomers are consistent with the HK97 capsid protein fold. N4 gp56 is 44 kDa, larger than the HK97 capsid protein (31 kDa), but similar in size to the capsid proteins in T4 (49 kDa), ϕ29 (48 kDa), and P22 (47 kDa). The extra residues in these proteins form an insertion domain on the viral surface,^{20; 21; 23} making it likely that the surface protrusion on gp56 corresponds to such an insertion domain.

Structure of the gp17-minus mutant virus

The complex pattern of protruding densities on the N4 capsid surface can be attributed to two different conformations of the decorating protein, gp17. The cryoEM structure of the gp17-minus mutant virus (Fig. 2) was determined to 15 Å resolution (Table 2) showing clearly the T = 9 symmetry (Figs. 1a-c). The gp17-minus capsid, with only small protrusions in the capsomer, has a smoother surface than the wild-type virus. The difference map calculated between wild-type and gp17-minus virions shows three significant density peaks representing gp17 per asymmetric unit (or 180-5 copies per virion) (Fig. 1d), consistent with the previously-determined value of 160 ± 20 copies per virion¹⁶ or 147 ± 49 (Table 1). The gp17 density was 3σ above the maximum noise level, indicating that the capsid protein structure is not significantly affected by gp17 binding.

Figure 2.

Protein composition of wild-type, gp17-minus, gp50-minus and gp65-minus virions. Arrows denote proteins absent in virions lacking gp50, gp65, or gp17. For properties of gene products, refer to Table 1.

Gp17 is an elongated molecule, located near quasi-3-fold axes relating any three capsomers, i.e, two hexamers and one pentamer, or three hexamers (Fig. 1d and e). One end of the molecule originates at the quasi-3-fold axis (thus creating a symmetry mismatch) and the other end extends towards the capsomeric boundaries. Despite the quasi T = 9 symmetry of the gp56 capsid lattice, there are only three copies of gp17 per icosahedral asymmetric unit: one at the juncture relating one pentamer and two hexamers (A in Fig. 1d), and two at the junctures relating three hexamers (B and C in Fig. 1d). As a result, the generally-positioned hexamers and the hexamers on the icosahedral 3-folds are differently decorated.

Two different conformations of the gp17 molecules were dependent on their locations. The gp17 molecule near position A has a “hook” shape, whereas the other two gp17 molecules near positions B and C have an elongated “rod” shape (Figs. 1d-f). Both densities are approximately 125 Å in length and 30 Å in width. At contour levels above 6σ, both hook- and rod-shaped densities separate into three smaller densities indicating that gp17 consists of three domains of about equal length, arbitrarily designated as DI, DII, and DIII, with DI being close to the quasi-3-fold axis. When the hook- and rod-shaped densities are compared, DI and DII superimpose well, whereas DIII from the hook-shaped gp17 is rotated about 140° with respect to domain III of the rod-shaped gp17. Thus, the difference of the two shapes is due to a change in the hinge angle between DII and DIII. Gp17 interacts with the capsid protein mainly through DI and DII, with DIII having little contact with the capsid surface in either the hook or rod conformation. DIII in the rod-shaped density has an almost radial orientation (Fig. 1f).

The fold and function of gp17

Analysis of the 279 amino acid sequence of gp17 using the program PSI-blast,^{24; 25} InterPro²⁶ showed that residues 93-200 were homologous to 5 proteins that had an immunoglobulin (IgG)-like fold with greater than 30% sequence identity. The IgG-like fold of gp17 is also readily recognized in domain search programs, such as InterPro and Pfam. Domain and secondary structure predictions using the programs DomPred²⁷ and PsiPred²⁸ identified three domains consisting of mostly β-strands. The predicted secondary structure pattern of gp17 was further searched against known secondary structure patterns in a three-dimensional structure database using the program DomSSEA²⁷ to identify proteins with similar folds. Two fibronectin-like proteins, each consisting of three IgG-like domains scored the highest (Protein Data Bank accession numbers 1TDQ and 1FNH), and differ from each other only in the angle between domains. Rigid-body fitting of these proteins into the cryoEM difference density produced a good fit except for the angle between DII and DIII. Thus, DI-DII and DIII were fitted as two separate rigid-bodies (Fig. 1f), resulting in an excellent fit, indicating that gp17 probably consists of three consecutive IgG domains.

Gp17 is not an essential component of the virus, since virions lacking gp17 are as infectious as wild-type virions. The location of the gp17 molecules around the periphery of capsomers might help to stabilize the capsid. A decrease in infectivity and the DNAse I sensitivity of the DNA in EDTA-treated virions (not shown) indicated that gp17-minus virions are unstable. The T4 soc, lambda gpD, and bacteriophage L Dec capsid proteins play a similar role in capsid stabilization.^{18; 29-31} Ig-like domain-containing proteins are found as components of many dsDNA bacteriophage virions.³² Based on the presence of Ig-like domains in bacterial surface glycohydrolases, Fraser et al. have suggested that phage Ig-like domains may aid in initial interactions with the bacterial cell surface. A similar role has been proposed for the bacteriophage ϕ29 head fibers and bacteriophage L decorating proteins;^{21; 33} whether gp17 plays such a role is not known. It may be of significance that the apparently unessential decorating proteins such as N4 gp17, T4 soc, the ϕ29 head fibers and the L decorating protein all bind at quasi-3-fold axes, although they form monomers in N4 and trimers in the other phages.

Interaction of capsid and DNA

The genomic DNA is organized as a series of concentric rings within the capsid shell, becoming weaker towards the center of the capsid. The central section of the icosahedrally-averaged reconstruction shows at least 8 layers of DNA with an average spacing of 24 Å (Fig. 1b). The spacing is similar to the dsDNA density observed in other bacteriophages, and is attributed to hexagonally-arranged DNA with an inter-layer spacing of 28 Å.³⁴ There are densities connecting the outermost DNA layer with the center of the hexameric capsomers, indicating DNA-protein interaction (Fig. 1b). No such interaction is visible at the center of the pentameric capsomers. It is not clear whether the interaction between the DNA and capsid is a direct interaction with the capsid protein gp56, or whether there is an additional protein that mediates the DNA-capsomer interactions. This would require 80 copies of such a mediating protein. However, apart from gp56 and gp17, no other protein is present in a sufficient number of copies in the virus. Thus, it is likely that the interaction is mediated by the capsid protein itself.

Tail structure

A 29 Å-resolution asymmetric cryoEM reconstruction showed that the 350 Å-long tail is assembled at the unique 5-fold vertex (Fig. 3). The tail components were named based on analogy with well-described tailed virus systems such as T4 and phi29, and the boundaries of each component were determined as a point where density disconnects and apparent symmetry breaks down. The neck protein between the head capsid and the tail also serves to attach twelve appendages, the latter being somewhat similar to the appendages of bacteriophage ϕ29.^{35; 36} The tail consists of a tail tube surrounded by a sheath. Each of the appendages consists of a thin, 200 Å-long stem and a distal domain. The density for the distal domain is weak and can only be seen at low contour levels, indicating flexibility.

Figure 3.

Structure of the mature N4 phage. (a) Surface representation of the asymmetric reconstruction of the mature N4 phage showing a side view (left) and an end-on view looking down the long axis of the phage (right). The head is colored from blue to white with increasing radius. The appendages (gp66), neck, non-contractile sheath (gp65), and tail tube are colored in orange, brown, green, and yellow, respectively. One icosahedral asymmetric unit is outlined in black. The scale bar represents 300 Å. (b) Central cross-section of the asymmetric reconstruction of the N4 phage. The distal domain of the appendages is marked with a star. (c) Enlarged tilted (left) view, end-on view (middle), and central section (right) of the 6-fold-averaged tail.

The 59 kDa gp66 is a good candidate for the appendage protein, although a BLAST search does not indicate homology to tail proteins in other phages. Secondary structure prediction of gp66 shows mostly α-helical structures at the N terminus, and β-structures at the C terminus. N-terminal residues between 50 and 210 contain leucine-rich regions that might form an interrupted trimeric coiled-coil. Assuming the pitch of a coiled-coil is 150 Å/100 residues, 160 residues would form a 240 Å-long coiled-coil, which would fit into the 200 Å-long and thin density of the appendage stem region. The BetaWrapPro program³⁷ predicts the C-terminal residues 253 to 392 to form right-handed β-helices. The estimated number of copies of gp66 is 27 ± 5¹⁶ or 25 ± 6 (Table 1), which is close to the 36 copies that would be present if each of the twelve appendages were trimers. Trimeric structures of coiled-coils or β-helices are common in phage tail fibers. Bacteriophage T4 fibritin is a 530 Å-long, interrupted, trimeric coiled-coil³⁸ and the cell-puncturing device is a 3-start β-helix.³⁹ P22, K1F, and the lactococcal phage p2 tailspikes are either three single-stranded β-helices or three-stranded β-helices.^40-42 A recent cryoEM reconstruction of bacteriophage ϕ29 also suggests that the ϕ29 appendages may consist of a trimeric coiled-coil attached to P22-like β-helical structures at their distal ends.³⁶

The central tail tube is a featureless, 310 Å-long cylinder with a 70 Å external diameter (Fig. 3c and 4). The inner diameter of the tail tube is about 25 Å, just large enough for dsDNA to pass through. In contrast to mature P22, T4, ε15, and ϕ29 phage structures,^{19; 36; 43; 44} no density is observed inside the tail tube. The very end of the tail tube is closed by a ‘plug’, which is presumably displaced upon infection. Surrounding the tail tube are six ribbons that form the sheath. The sheath is wider (~230 Å) near the neck and narrower (~75 Å) near the plug at the distal end of the tail. Each ribbon of the tail sheath makes only few contacts with the tail tube (Fig. 3c).

Figure 4.

Structure of gp65-minus mutant virus. (a) Surface representation of the asymmetric reconstruction of gp65-minus mutant virus showing a side view (left) and an end-on view looking down the long axis of the phage (right). The color scheme is the same as in Figure 3. One icosahedral asymmetric unit is outlined in black. The scale bar represents 300 Å. (b) Central cross-section of the asymmetric reconstruction of gp65-minus virus. (c) Tilted view of the tail for the 6-fold-averaged gp65-minus virus (left) and an end-on view (right). (d) Difference map calculated between asymmetric reconstructions of wild-type and gp65-minus viruses showing the non-contractile tail sheath (gp65).

Virions that lack gp65 do not adsorb to E. coli; moreover, purified gp65 blocks adsorption of wild-type phage to E. coli cells (J. McPartland and L. B. Rothman-Denes, unpublished data). Thus, gp65 is probably the major receptor-binding protein. An asymmetric reconstruction of a gp65-minus mutant virus (Fig. 2) lacked the tail sheath (Figs. 4a-c). A difference map calculated between the wild-type and gp65-minus mutant virions clearly showed the tail sheath consisting of six ribbons (Fig. 4d). The number of copies of gp65 in mature virions is consistent with the previously determined number.¹⁶ Each gp65 molecule is 265 Å long, and on average 35 Å wide. There are three larger lumps along the length of each ribbon that might represent separate protein domains.

It seems that the N4 tail assembles sequentially onto the DNA-filled capsids, similar to T7 or ϕ29 tail assembly,^{9; 45} since gp65-minus virions can successfully assemble a phage lacking only the tail sheath. When gp65 is added to the gp65-minus mutant, then the particles gain infectivity (J. McPartland and L. B. Rothman-Denes, unpublished data) suggesting that gp65 is the last protein to be added in the assembly process.

N4 is the first short-tailed phage to have a tail tube inside a non-contractile sheath. Whereas N4 uses the tail sheath to recognize the host (a function accomplished by the tail fibers of T4), the contractile sheath of T4 is part of a mechanism for inserting the tail tube into the periplasmic space of E. coli.

The portal protein and vRNAP location

A portal protein, similar in structure to the ϕ29 portal connector assembly⁴⁶ and also seen in other tailed phages,^{18; 19; 44; 47}, is apparent at the unique 5-fold vertex associated with the tail in the cryoEM reconstruction of N4. The N4 portal is about 170 Å wide and 90 Å high and has a central channel large enough to allow the passage of dsDNA. The portal consists of three domains, similar to the crown, wing, and stalk domains observed for bacteriophages ϕ29, T7, and SPP1.^{46; 48-50} The internal channel diameter of the portal ranges from a minimum of 20 Å at the top, to a maximum of 80 Å in the middle, and 45 Å at the bottom (distal) end. The dimensions of the N4 portal’s length and external diameter are larger than that of the ϕ29 connector.⁴⁶ Docking of the dodecameric ϕ29 connector crystal structure into the density shows a good fit in the wing and the stalk domains (Fig. 5a).

Figure 5.

The interior of bacteriophage N4. (a) The N4 portal assembly showing the crown, wing and the stalk domains. The atomic structure of the ϕ29 connector protein (Protein Data Bank accession number 1H5W) was fitted into the 5-fold-averaged cryoEM density of the N4 portal assembly. The α-helices and β-strands of the ϕ29 connector are shown in red and purple, respectively. (b) Comparison of wild-type and gp50-minus viruses. Cross section of the 5-fold-averaged wild-type (left) and gp50-minus (right) viruses viewed perpendicular to the long axis of the phage. The densities are displayed at 2.5σ above their respective means to show the molecular boundaries of portal and inner core proteins. The dsDNA, inner core proteins, and portal assembly are colored in dark green, orange, and red, respectively. All the other components of the phage are colored as in Figure 3.

Amongst phages, the connectors’ sequence identity is low and of variable size (40-100 kDa). However, secondary structure analyses of portal proteins, when compared with the ϕ29 portal structure, suggest a conserved folding motif.⁴⁷ Furthermore, all known bacteriophage portal proteins are dodecameric when assembled in capsids. Based on these observations and because the predicted secondary structure for residues 295-410 matches the portal folding-motif, gp59 was identified as the N4 portal protein (Table 1). The larger molecular weight (86 kDa) of gp59, compared to 40 kDa of the ϕ29 connector, accounts for the crown and wings seen in SPP1 crystal structure.

Previously, we determined that vRNAP was present in 1-2 copies per virion¹⁶. A re-evaluation, based on the presence of 114 methionines in the vRNAP polypeptide in mature virions, indicates that vRNAP is present in 4±1 copies per virion (Table 1). Biochemical analysis of vRNAP-less particles indicates that they contain a full complement of the N4 genome (data not shown), and all virion proteins except for vRNAP (Fig. 2). The 5-fold-averaged reconstruction of a gp50-minus virus, determined to 20 Å resolution, has the same surface features as wild-type N4 (Fig. 5b), indicating that the vRNAP resides inside the capsid although this is not apparent in the asymmetric reconstruction at only 30 Å resolution. In wild-type virions, cylinder-like densities on top of the portal protein disrupt the continuity of the circular DNA density (Fig. 5b). This internal core density is ~110 Å in height and ~80 Å in width, and comprises at least two components, an upper disk-like structure that sits on top of a lower, longer cylindrical density. The internal core density in gp50-minus virus is ~80 Å in height and ~100 Å in width, smaller than that of the wild-type. One major difference is that the density corresponding to the lower end of the cylinder (closest to the portal protein) in wild-type is missing in the gp50-minus virus, suggesting that at least a part of vRNAP is located close to the internal entrance to the portal channel. The proximity of the vRNAP to the portal channel and its requirement for genome injection,¹⁷ suggest that vRNAP exits the capsid prior to the ejection of the DNA.

The process of infection

The N4 tail assembly is different from other Podoviridae viruses, which might indicate the special requirement of the N4 tail to deliver a large protein. The N4 vRNAP and DNA transport requires host proteins from several different cellular compartments, reflecting the complexity of the transport process. The N4 phage first binds to the E. coli cell surface through an interaction between the tail sheath and the outer membrane N4 receptor NfrA.⁵¹ Adsorption must trigger conformational changes in the tail, which leads to the ejection of the 3500 amino acid vRNAP and the first 500 bp of the N4 genome into the host. The narrowest part of the tail tube is 25 Å in diameter and, thus, the large size of vRNAP precludes the passage in its native form. Ejection of proteins from the bacteriophage capsid into the host cell has been reported for T7 and T4, and denatured or partially-denatured state mechanisms have been proposed.^{8; 9; 52} Similarly, the vRNAP would have to exit the phage head in an unfolded or semi-unfolded form, which would need to be refolded in the cytoplasm.

Materials and Methods

Bacterial and phage strains

E. coli K12 W3350 was used for growth of wild-type N4. E. coli W3350 supF was used for growth of N4 am229 (ORF 65am), N4 am29 (ORF 50am), and N4 ORF 17am. The resultant virions were used to infect E. coli W3350, yielding virions lacking gp17, gp50, or gp65 (Fig. 2).

Bacteria were grown aerobically at 37 °C in Luria-Bertani (LB) medium to 3 × 10⁸ cells/ml. Cells were infected with phage at a multiplicity of infection of 10. The infected cells were lysed with chloroform 3 hrs. after infection. Cellular debris was pelleted and the supernatant was filtered through a 0.45 μm Millipore filter before loading onto 5%-40% glycerol step gradient. Following centrifugation at 35,000 RPM for 2 hrs. at 4°C in an SW41 Beckman rotor, the pelleted phage was resuspended in TM buffer (15 mM Tris-HCl pH 8.0, 10 mM MgCl₂) and purified by CsCl buoyant density centrifugation. CsCl purified virions were resuspended in TM buffer. Samples were heated for 2 min at 100 °C in 1.6% sodium dodecyl sulphate, 60 mM Tris-HCl pH 6.8, 100 mM DTT, 5% glycerol (loading buffer), and loaded onto a precast 4-20% polyacrylamide gradient gel (Bio-Rad Laboratories, Hercules, CA). Proteins were stained with Phastgel™ Blue R (GE Healthcare Biosciences, Piscataway, NJ).

Labeling of virion proteins and determination of protein stoichiometry in virions

E. coli W3350 cells were grown arerobically at 37°C in LB medium to 3 × 10⁸ cells/ml and infected with wild-type N4 phage at a multiplicity of infection of 10. After 6 minutes, cells were spun down, resuspended in LB medium containing 50 uCi/mL S35 L-Methionine (Perkin Elmer Life & Analytical Science, Shelton, CT), and lysed with chloroform at 3 hours post-infection. Virions were purified and processed as described above, and loaded onto a 10% SDS-polyacrylamide gel. The gels were dried, and the labeled proteins were detected by phosphorimaging and quantitated using ImageQuant Software. Several dilutions of virions were employed to assure a linear response. Virion copy number was based on the number of methionines in the mature virion protein and virion copy numbers of 6 and 535 for gp65 and coat protein, respectively.

Generation and characterization of N4 ORF 17am

An amber mutation in ORF 17 was introduced into the N4 genome. To that end, a plasmid derivative of pUC18 carrying a 2,189 bp N4 genomic fragment, spanning positions 5,430 to 7,619 (ORFs 16 and 17) and containing a TGTATA to TCTAGA change starting at position 6,922, inserted between the PstI and EcoRI sites was constructed. The sequence change, UAU (Tyr) to UAG (am) in ORF 17 (aa 56), introduced a new XbaI site overlapping with the UAG codon. E. coli W3350 pUC18 ORF 16/ORF 17am was grown in 5 ml of LB with 5 μl of 100 μg/ml ampicillin at 37 °C with aeration to 3 × 10⁸ cell/ml and infected with N4 am15 (ORF 16am)⁵³ at a multiplicity of 0.01. Recombination between the plasmid and the infected phage yielded phages that had incorporated the plasmid into the phage genome by homologous recombination.⁵⁴ These phages grow in E. coli W3350 because they carry wild-type as well as mutated copies of ORFs 16 and 17. After 15 min. incubation, aliquots were removed, and serial dilutions were plated onto E. coli W3350 and incubated overnight at 37 °C. Large plaques (phages that had integrated the plasmid into the N4 genome and, therefore, had wild-type copies of ORF 16 and ORF 17) were spotted onto plates seeded with E. coli W3350 and E. coli W3350 supF. Plates were incubated overnight at 37 °C and individual plaques growing in E. coli W3350 supF were inoculated into tubes containing 5 ml of LB media and E. coli W3350 supF. After 3 hrs. incubation with aeration at 37 °C, cells were lysed with the addition of chloroform and cell debris was removed by centrifugation. During this second period of growth the plasmid excises to yield two types of progeny: parental N4 am15 and the newly generated N4 ORF 17am. To determine whether the phages isolated carried the amber mutation in ORF 17, ORF 17 was PCR-amplified using primers: ORF 16_F-Pst: 5′ GCA-CTGCAGGCCAATAGGCTAAGACGCGA and ORF 17_R_Eco: 5′ GCAGAATTCCATTCATAAAAACAATCACATTACTCTGT. The resulting PCR-amplified fragments were digested with XbaI to ascertain the presence of the XbaI site in ORF 17, and sequenced to confirm the presence of the TAT to TAG change in ORF 17. All phages tested plated with the same efficiency in E. coli W3350 and in E. coli W3350 supF, indicating that gp17 is not essential for phage growth. SDS-PAGE analysis of the protein composition of N4 ORF 17am virions indicated that they contain all virion proteins except gp17 (Fig. 2).

CryoEM and three-dimensional image reconstruction

All micrographs were taken at a magnification of ~39,500 using a FEI-CM200 electron microscope. Micrographs were digitized using a Zeiss scanner with a step size of 14 μm, which resulted in a 3.54 Å/pixel size. Individual particles were selected and boxed by eye using RobEM (http://cryoem.ucsd.edu/programs.shtm). The boxed particles were corrected for linear gradient and normalized means and variances. The contrast transfer function parameters were determined with the program ‘ctfit’ in the EMAN package.⁵⁵

The initial model for the icosahedral head was generated from the class averages of 5-fold, 3-fold, and 2-fold views as determined by EMAN. The model for the tail was produced by putting a cylinder at one of the 5-fold vertices along the long axis of the phage. The icosahedral symmetry was released to produce an asymmetric reconstruction using a python script¹⁹ to search for 60 different orientations.³⁵

The resolutions of each reconstruction were determined by the Fourier shell correlation method. A correlation of 0.5 was used as a cut-off between two reconstructions from independent half-data sets. The number of particles, symmetry imposed, and the final resolution for each reconstruction are summarized in Table 2.

Difference maps between the reconstructions were calculated after adjusting radial and density scale factors using RobEM (http://cryoem.ucsd.edu/programs.shtm). The program SITUS was used to fit the structure of fibronectin into the cryoEM difference density between wild-type and gp17-minus virions.