Structure of the Staphylococcus aureus bacteriophage 80α neck shows details of the DNA, tail completion protein and tape measure protein

SUMMARY

The Staphylococcus aureus pathogenicity islands (SaPIs), including SaPI1, are a type of mobile genetic elements that are mobilized at high frequency by “helper” bacteriophages, such as 80α, leading to packaging of the SaPI genomes into virions made from helper-encoded structural proteins. 80α and SaPI1 virions consist of an icosahedral head connected via a portal vertex to a long, noncontractile tail. A connector or “neck” forms the interface between the tail and the head. Here, we have determined the high-resolution structure of the neck section of SaPI1 virions, including the dodecameric portal and head-tail-connector proteins, and the hexameric head-tail joining, tail terminator and major tail proteins. We also resolved the DNA, the tail completion protein and the tape measure protein inside the tail, features that have not previously been observed at high resolution. Our study provides insights into the assembly and infection process in this important group of mobile genetic elements.

Graphical Abstract

eTOC BLURB

S. aureus pathogenicity island 1 (SaPI1) is mobilized by helper bacteriophage 80α and packaged into virions made from bacteriophage structural proteins. Kizziah et al. present the cryo-EM structure of the SaPI1 virion neck between head and tail, providing insights into the assembly and infection process of staphylococcal phages and SaPIs.

INTRODUCTION

Tailed bacteriophages (phages) with double-stranded DNA genomes belonging to class Caudoviricetes (order Caudovirales) are abundant in all environments and play important roles in biomass turnover and bacterial evolution¹. Phages often exist as prophages, integrated as mobile genetic elements (MGEs) into their host genomes, and frequently carry genes encoding virulence factors and antibiotic resistance^{2, 3}.

The Caudovirales phages consist of an icosahedral or prolate head (capsid) attached to a tail via a unique vertex that incorporates a dodecameric portal. The portal is assumed to act as the nucleus for capsid assembly, and serves as an entry and exit point for the DNA^4–6. Bacteriophages are traditionally divided into groups based on tail morphology: siphoviruses have long, non-contractile tails, myoviruses have long contractile tails, while podoviruses have short tails. However, these groups are not monophyletic and are no longer considered viral families⁷. The tail generally makes the first contact with the host and plays important roles in host recognition, cell wall penetration, and injection of the genome^8–10. Long tails are assembled through a separate pathway and are attached to heads via a connector or neck region, consisting of several proteins that together act as a conduit for the DNA during infection¹¹.

Staphylococcus aureus is an opportunistic human pathogen that encodes a large array of virulence factors^{12, 13}. Most of these are encoded on MGEs, including prophages and chromosomal islands^{14, 15}. Transduction by phages represents the main mechanism by which MGEs are transferred between hosts in S. aureus^{16, 17}. S. aureus pathogenicity islands (SaPIs) are a type MGEs that encode superantigen toxins, adhesins and other virulence factors. SaPIs become mobilized at high frequency by so-called “helper” phages^18–20 and packaged into transducing particles made up of phage-encoded structural proteins. SaPIs suppress the replication of their helpers, often by redirecting the assembly pathway to form capsids of a smaller size than that normally made by the phage^{20, 21}.

Phage 80α is a typical temperate staphylococcal siphovirus that can act as helper for a number of SaPIs, including SaPI1¹⁹. Phages closely related to 80α are found in many staphylococcal strains, including MRSA strain USA300 LAC, commonly involved in community-acquired infections²². The 80α virion has a 63 nm icosahedral head with T=7 architecture, a 190 nm long flexuous tail, a complex baseplate^23–25, and a 43,864 base pair double-stranded (ds)DNA genome²⁶(NCBI RefSeq NC_009526.1)(Fig. 1A). When 80α acts as a helper for SaPI1, the assembly pathway is redirected to form small capsids with T=4 architecture^{27, 28} (Fig. 1B). DNA is packaged into the capsids by a terminase complex, consisting of small (TerS) and large (TerL) subunits. SaPI1 encodes its own TerS protein, which provides specificity for its own genome^{29, 30}. Both 80α and SaPI1 genomes are packaged via a headful mechanism yielding blunt-ended, circularly permuted copies with terminal sequences distributed throughout the genome³¹.

Figure 1.

Structure of the SaPI1 neck. (A) Schematic diagram of part of the late operon of 80α, showing the genes encoding neck proteins. ORF numbers and protein names are shown: TerS, small terminase subunit; TerL, large terminase subunit; PP, portal protein; EP, ejection protein, SP, scaffolding protein; CP, major capsid protein; HTCP, head-tail connector protein; HTJP, head-tail joining protein; TCP, tail completion protein; TrP, tail terminator protein; MTP, major tail protein; TMP, tape measure protein. The schematic diagram indicates the location of the neck proteins in the virion. The DNA is shown as a purple double helix. (B) Cryo-electron micrograph of SaPI1 virions. Scale bar = 50 nm. (C) Isosurface representation of a composite of the C6 neck reconstruction, colored by radius from the central axis, and the previously determined C12 reconstruction of the portal (gray)³². (D) Cutaway view of the asymmetric (C1) reconstruction of the neck, showing the DNA, TCP and TMP inside the neck, colored by radius according to the color bar. (E) Segmented reconstruction, colored by protein as in panel A: PP (gp42), gray; HTCP (gp49), blue; HTJP (gp50), green; TrP (gp52), pink; MTP (gp53), yellow. (F) Cutaway view of segmented reconstruction, showing density inside neck, colored as in panel C. In addition, the DNA is purple; TCP (gp51), orange; TMP (gp56), brown. The segmented maps were Gaussian filtered to 4Å resolution for clarity.

We previously determined structures of 80α and SaPI1 procapsids and mature capsids^{24, 28}, as well as the 80α portal protein expressed in E. coli and in situ in SaPI1 virions³². We also determined the structure of the 80α baseplate, a complex of 7 different proteins, including three presumed receptor binding proteins²⁵. However, the organization of other tail-related components, including the neck that connects the tail and the head, were not resolved in these structures.

Here, we have used cryo-electron microscopy (cryo-EM) to determine the structure of the neck region of SaPI1 virions that constitutes the interface between the head and the tail. The reconstruction resolves several ring-like proteins that connect the portal to the major tail protein. We were also able to resolve in molecular detail the DNA, the tape measure protein and the tail completion protein inside the tail. Our detailed structural description of these features provides insights into the architecture and assembly of phage tails, and a structural basis for understanding the infection process in this important group of MGEs and their helper bacteriophages.

RESULTS

The head-to-tail connector

We previously described the structure of the capsid and portal from SaPI1 virions³². In the focused reconstruction of the in situ portal, there was a dodecameric ring immediately below the portal that we identified as gp49, product of open reading frame (ORF) 49 in the RefSeq entry for 80α (NC_009526.1) (Fig. 1A). We refer to this as the “head-to-tail connector protein” (HTCP), consistent with usage in phage lambda (Table 1). In Bacillus subtilis phage SPP1 and Pseudomonas aeruginosa phage JBD30 it is referred to as an “adaptor”, as it connects the dodecameric portal with the hexameric proteins below it^{33, 34}, whereas in Salmonella phage Chi it is referred to as Neck1³⁵. In the portal reconstruction³², the C-terminus of the HTCP could be seen to add a β-strand to the clip domain of PP, forming a three-stranded β-sheet together with one strand from each of two adjacent PP subunits³². However, the rest of the HTCP was not well resolved, and the density had been truncated due to the applied mask.

Table 1.

List of 80α neck proteins and equivalent proteins from other phages.

80α (Staphylococcus aureus)				Lambda (Escherichia coli)47, 49 57 RefSeq: NC_001416		SPP1 (Bacillus subtilis)33, 42, 45 RefSeq: NC_004166		HK97 (Escherichia coli)37, 46, 58 RefSeq: NC_002167		JBD30 (Pseudomonas aeruginosa)34 RefSeq: NC_020198		GTA (Rhodobacter capsulatus)38.		Chi (Salmonella enterica)35 RefSeq: NC_025442
Protein	Abbr¶	Olig§	ORF	Name	ORF	Name	ORF	Name	ORF	Name	ORF	Name	ORF	Name	ORF
HEAD ASSOCIATED:
Portal	PP	12	42	Portal	B	Portal	6	Portal	3	Portal	32	Portal	3	Portal	14
Head-Tail Connector	HTCP	12	49	Connector	W	Adaptor	15	Adaptor, Ad1; Connector	6	Adaptor	41	Adaptor	6	Neck1	13
Head-Tail Joining	HTJP	6	50	Head-tail joining	FII	Stopper	16	Head closure, Hc1; Adaptor	7	–	–	Stopper	7	Neck2	19
TAIL ASSOCIATED:
Tail Completion	TCP	1	51	Tail completion	Z	TCP	16.1	–	10	–	–	–	–	–	–
Tail Terminator	TrP	6	52	Tail terminator	U	THJP	17	Tail terminator	11	Stopper*	42*	Terminator	8	Terminator	21
Major Tail	MTP	6	53	Tail tube	V	TTP	17.1	MTP	12	MTP	44	TTP	9	TTP	22
Tape measure	TMP	3	55	TMP	H	TMP	18	TMP	16			TMP	11	TMP	25

^¶Additional abbreviations not defined in this column: THJP, tail-head joining protein; TTP, tail tube protein.

^§Oligomeric state of protein

^*Identified as “stopper” in Valentova et al.³⁴, but is actually equivalent to TrP.

Here, we re-extracted the particles from the same data set (59,457 particles; Fig. 1B), centered on the neck region below the portal, and carried out a separate focused reconstruction that included part of the portal, the HTCP and additional structures associated with the neck and tail. While the HTCP was expected to have C12 symmetry, the tail is known to have C6 symmetry²⁵; therefore, the focused reconstruction was refined with the application of C6 symmetry, reaching a final resolution of 3.1 Å (Fig. 1C; Fig. S1A,C; Table S1).

A model for gp49 was generated using AlphaFold2³⁶ and fitted into the map density immediately below the portal, followed by refinement in ISOLDE (Fig. 1E, 2A–E; Fig. S2A). The HTCP forms a dodecamer in which each subunit consists of a short N-terminal α-helix followed by three 25 Å-long antiparallel α-helices interspersed with a β-hairpin (Fig. 2A, F). The HTCP model was merged with the previous portal model³² using the C-terminal extension as a guide (Fig. 2A–C, F). The three long α-helices from all 12 subunits form a 95 Å-wide outer ring, while the β-hairpins define a barrel with 31 Å inner diameter that lines the central tube of the neck, constituting the narrowest point of the neck below the portal (Fig. 2D, E). (The portal itself has an inner diameter of 26 Å at its narrowest³².) The β-barrel inserts into the tube formed by the other neck proteins below (Fig. 2D, E).

Figure 2.

Atomic modeling of the neck. (A) Ribbon representation of the atomic model from the C6 neck reconstruction. The two copies of HTCP (gp49) per asymmetric unit are colored light and dark blue; the HTJP (gp50), TrP (gp52) and MTP (gp53) hexamers are shown in green, pink and yellow, respectively. The C-terminus of HTCP, which was built from the previous portal structure³² is in cyan. (B) Atomic model of the neck reconstruction with one asymmetric unit colored as in panel A, the rest gray. The previously determined portal protein model is included, with the two copies of PP (gp42) per asymmetric unit colored light and dark brown. (C) An isolated asymmetric unit, colored as in A and B. (D) Slab through the neck model, colored as in panel A. (E) Sections through the neck model at the levels indicated by the arrows in panel D. From left to right: HTCP and HTJP; HTJP and TrP; TrP and MTP. The smallest inner diameters are indicated. (F) Monomers of neck proteins, from left to right: HTCP, HTJP, TrP and MTP. (G) Interaction between HTJP (green surface) and the HTCP (blue) and TrP (pink). (H) Interaction between HTJP (green), HTCP (blue) and TrP (pink) shown as molecular surfaces. (I) AlphaFold model of HTJP (orange) superimposed on the HTJP model from the neck reconstruction (green), showing the predicted inner ring of α-helices. (J) Cutaway view showing the electrostatic potential surface of the tail interior. (K) MTPs from the 80α baseplate reconstruction (orange and tan) superimposed on the MTP from the neck model (yellow), showing the difference in the C-arm.

The structure of gp49 is similar to the equivalent proteins from other phages, including gp6 from Escherichia coli phage HK97³⁷, gp15 from phage SPP1³³ and gp41 from JBD30, to which it aligns with root-mean-square deviations (RMSD) of 3.61Å, 4.84Å and 5.21Å, respectively (Fig. S3A; Table S2). The equivalent protein from a phage-like “gene transfer agent” (GTA) from Rhodobacter capsulatus, gp6,has a large insertion between the α-helices, reflecting a difference in its interaction with the portal³⁸ (Fig. S2A). By comparison, the HTCP from E. coli phage lambda, gpW, is shorter, lacking the N-terminal two α-helices (Fig. S3A). The remaining two α-helices of gpW are tilted ≈45° outward compared to 80α. However, the β-hairpin that lines the interior of the 80α and SPP1 HTCPs, but is missing in HK97 and JBD30, is retained in lambda, indicating that the proteins are indeed structurally and evolutionarily related³⁹. In the solution structure of gpW⁴⁰, the hairpin is folded over the helices, suggesting that the protein undergoes conformational change upon oligomerization and portal binding.

The head-to-tail joining protein

The focused reconstruction of the neck revealed two additional protein rings with C6 symmetry between the HTCP and the major tail protein (MTP, gp53) (Fig. 1C, E). Based on the principle of co-linearity of genome and structure observed in other phages, we presumed that these proteins corresponded to gp50, gp51, and/or gp52 (Fig. 1A), and made AlphaFold2 models for each protein. Each model was then tested for fit into the ring-like densities. While gp51 did not match either ring, gp50 and gp52 matched the top and bottom ring, respectively (Fig. 1E, 2A–E; Fig. S2A).

The gp50 protein, which makes up the topmost ring, is equivalent to the head-tail joining protein (HTJP) gpFII of lambda⁴¹, gp16 of SPP1³³, gp19 from phage Chi³⁵ and gp7 from GTA³⁸ (Fig. 2F; Fig. S3B; Table S2). In SPP1, gp16 was associated with heads extracted from tail-less DNA-filled capsids³³ and was considered part of the head-tail connector complex, assumed to be added to packaged heads prior to attachment of the tail. SPP1 gp16 was designated as “stopper”, but whether it serves a functional role as a stopper—to prevent the DNA from falling out of the capsids after packaging—in 80α is not known. We will refer to this protein by the term “head-tail-joining protein” (HTJP), consistent with the terminology used by Casjens et al.³⁹ (Table 1).

The 80α HTJP consists of a seven-stranded β-sandwich with two extended loops (Fig. 2F). The top of the sandwich inserts into a groove between the outer α-helix and the β-hairpins formed by two copies of HTCP (Fig. 2D,G,H). The two extended loops at the other side of the sandwich form a clamp that grabs the inside and outside of the ring below it, made of gp52 (Fig. 2D,G,H; see below). Alhough the loop comprised of residues 43–55 exhibits a β-hairpin structure in the reconstruction, AlphaFold predicted residues 43–50 to be an α-helix that would narrow the inner diameter of the hexamer to 16 Å (Fig. 2I), possibly representing a pre-tail joining state where the protein ring could act as a stopper, similar to SPP1³³.

The β-sandwich of HTJP resembles a classic “tail tube protein” (TTP) fold that is found in the major tail protein (MTP, gp53), as well as the distal tail protein (Dit) and the tail tip protein (Tal)²⁵. The β-hairpin (residues 43–55) corresponds to the previously described “stacking loop” that forms part of the interaction between successive rings of MTP²⁵(Fig. 2F). The 80α HTJP is quite similar to gp16 of SPP1, with an RMSD of 6.99Å for all equivalent residue pairs, but is more divergent from gpFII of lambda (RMSD=14.43Å), which lacks the extended external loop and has an additional α-helix at its N-terminus that forms a ring around the barrel made from the β-hairpins of gpW⁴⁰(Fig. S3B; Table S2). The solution structure of gpFII lacks this α-helix⁴¹, suggesting that this ring is formed only when the protein oligomerizes upon binding to the the gpW connector.

The tail terminator protein

The AlphaFold2 model for gp52 matched the hexameric protein ring below the HTJP (Fig. 1E, 2A–E; Fig. S2A). This is the “tail terminator protein” (TrP)^{10, 39}, equivalent to lambda gpU and GTA gp8 (Fig. S2C)^{10, 42}, that are added to the tail prior to its attachment to the capsid.⁴³

The TrP consists of a four-stranded β-sheet with two α-helices (Fig. 2F), resembling one-half of a TTP fold, indicating a structural and evolutionary relationship, as suggested by Cardarelli et al.³⁷. The β-sheet provides a predominantly negatively charged surface that is continuous with the β-sheets of MTP that form the inner lining of the tail tube (Fig 2J). Two loops at the top of the sheet mesh with the extended loops of two HTJP subunits above it, while the loops at the bottom connects with the MTP²⁵ (Fig. 2A–D, G–H). The 80α TrP protein is similar to those from lambda (gpU; RMSD=5.4 Å for 114 equivalent atoms) and GTA (gp8; RMSD=10.86 Å) (Fig. S3C; Table S2).

Below the TrP, a sixfold symmetric ring could be identified as the major tail protein (gp53), for which the structure was previously determined from the reconstruction of the 80α baseplate²⁵ (Fig. 1E, 2A–D). It consists of an 8-stranded β-sandwich where one side presents a negatively charged surface to the inside of the tail (Fig. 2F, J). The neck reconstruction showed part of the second MTP ring, but the density was too poor to be modeled. A total of 39 or 40 MTP rings form the complete 80α and SaPI1 tail. In the tail, the C-terminus (C-arm) of MTP interacts with an extended “stacking loop” and extends into the β-sheet of the MTP ring above it²⁵. The same C-arm interacts with the shorter stacking loop of TrP, while the N-terminus interacts with the second loop of the adjacent TrP subunit (Fig. 2K).

The tail interior

In the C6 symmetrical reconstruction, an elongated density could be clearly discerned in the inner tube of the neck stretching all the way from the portal through the neck and into the tail tube; however, the density was not resolvable to high resolution, presumably due to smearing caused by the sixfold averaging. To resolve this internal density, we carried out a symmetry expansion from C6 to C1, followed by masked classification and asymmetric reconstruction focused on the tail interior. The mask was then widened, resulting in a C1 reconstruction encompassing the entire neck that reached a resolution of 3.4 Å (FSC=0.143; Fig. 1D; Fig. S1B, D; Fig. S2B; Table S1).

Although the quality of the inner density was not as high as for the outer proteins (Fig. S1D), the reconstruction clearly showed density corresponding to a double-stranded DNA helix in the upper part of the neck, extending about 72Å from the clip domain of PP, through the HTCP, and ending near the interface between the HTCP and the HTJP (Fig. 1D, F; Fig. S4). Immediately below the DNA, tubular densities corresponding to α-helices could be discerned (Fig. 1D, F). We initially expected these densities to correspond to the tape measure protein (TMP), gp56, which is assumed to extend through the entirety of the tail⁴⁴. To our surprise, these α-helical densities instead matched the AlphaFold2 model of gp51, which could be fitted with high confidence into the density as a monomer with minimal changes to the model (Fig. 3A; Fig. S5). To model the internal density accurately, we generated an arbitrary DNA sequence using AlphaFold3, which was combined with the Alphafold2 model of gp51 to initiate the model refinement, resulting in a structure that included residues 7–115 of gp51 and 20 base pairs of B-DNA (Fig. 3A–E; Fig. S5).

Figure 3.

Modeling of the inside of the neck. (A) Detail of internal density with the TCP (gp50, orange), TMP (gp56, shades of green) and DNA (purple) models fitted in. External proteins (HTCP, HTJP, TrP and MTP) are colored as in Fig. 2. (B) Atomic model of the inside of the neck, including the DNA double helix (purple and magenta), TCP (orange) and the three copies of TMP in green, turquoise and aqua, inside the segmented density of the asymmetric reconstruction. (C) Same model as in (B) with the atomic model of the surrounding proteins shown in gray. (D) slabs through the tail model at the levels indicated in (C). (E) Closeup view of the internal proteins and the DNA. (F) Ribbon diagram of the TCP model. (G) Ribbon diagrams of the AlphaFold3 models of the TCPs from phage lambda (gpZ), HK97 (gp10) and SPP1 (gp16.1). (H) The interaction between TCP, TMP and the DNA. The TCP interacting loop is shown in stick representation, colored by element. (I) Detail of the interaction between TCP and the DNA. The 3’ and 5’ ends of the DNA and pertinent residues are indicated. (J) Electrostatic potential surfaces of the inside of the tail. (K) Electrostatic surface of TCP with surrounding proteins and DNA shown in ribbon representation. (L) Top view of the electrostatic surface of TCP with the HTCP shown in ribbon representation. (K) Same view as panel L with TCP and HTCP both shown as electrostatic surfaces.

The gp51 structure consists of a 55 Å long, bent N-terminal α-helix, an extended loop interspersed with a β-hairpin and a short α-helix, followed by a 29 Å long C-terminal α-helix (Fig. 3E, F). Based on the location of ORF51 in the genome (Fig. 1A), gp51 is equivalent to the “tail completion protein” (TCP) of phage lambda (gpZ)¹⁰ and its equivalents gp16.1 in SPP1⁴⁵ and gp10 in HK97⁴⁶. While no structures of these or any other TCPs have been determined, AlphaFold models of lambda gpZ, SPP1 gp16.1 and HK97 gp10 reveal the structural similarity (Fig. 3G). Based on these models, the 80α TCP is most similar to gp16.1 from SPP1 (RMSD=6.93 Å for 108 equivalent Cα atoms), while lambda gp6 is more distantly related, with an additional β domain inserted at its apex (Fig. 3G), perhaps reflecting the difference in packaging strategy. However, the locations of these proteins in their respective phages have not been determined: Lambda gpZ was not observed in recent high-resolution structures of phage lambda^47–49, and although gp16.1 was shown to be tail-associated in SPP1, its exact location and structure has remained unknown⁴⁵. A recent structure of the phage T5 tail indicated that the TCP was located on the outside of the baseplate⁵⁰, but it is not clear whether this feature is related to the 80α TCP.

Since there is no unique sequence at the ends of the genome in 80α or SaPI1, it was not possible to model the DNA accurately in our structure. The end of the modeled DNA appeared to be blunt, with a phosphate group at the 5’ end, as expected for DNA packaged by a headful mechanism³¹ (Fig. 3A,H,I). However, additional density at the 3’ end of the DNA that could not be modeled might represent a single-nucleotide overhang. Residues Y78 and F81 in the extended loop of the TCP cap the DNA double-helix by forming pi-stacking interactions with the final base pair. (Fig. 3I) The 5’ and first backbone phosphates are stabilized by a hydrophilic pocket formed from residue N55 and exposed backbone peptide bonds in the TCP loop (Fig. 3I). The last backbone phosphate at the 3’ end is stabilized by the guanidinium group of R87 (Fig. 3I). These non-specific interactions accommodate the end of the DNA terminus regardless of DNA sequence.

Below the TCP, several α-helical densities could be discerned, which were presumed to belong to the TMP (Fig. 1D, F, Fig. 3A; Fig. S4). Several AlphaFold3 predictions incorporating varying combinations of HTCP, HTJP, TrP, TCP, and DNA, together with three copies of TMP (consistent with our previous reconstruction of the 80α baseplate²⁵) were generated. While neither of these models accurately represented the entire neck structure, one model included an interaction between TCP and TMP that accurately matched the density below the TCP. In this model, the N-termini of three copies of TMP each contributes a β-strand to a five-stranded β-sheet that includes the β-hairpin of the TCP (Fig. 3B–E). The N-termini of the three copies of TMP are shifted by about 6–7 residues relative to one another, so that the following α-helices are out of register (Fig. 3E; Fig. S4). The N- and C-terminal α-helices from the TCP together with the three α-helices from TMP form a five membered superhelix (Fig. 3D–E). Only residues 1–37 of TMP could be modeled into the density. Below this, the density deteriorates and the organization of TMP becomes unclear (Fig. 1D). Once its C-terminus reaches the baseplate, however, the three copies must be back into register again, since the three C-termini are known to interact with the Tal protein with identical interactions, as shown in our previous reconstruction of the 80α baseplate²⁵. It is not clear how this is accomplished, but TMP is often seen by SDS-PAGE to be fragmented, suggesting that it might be cleaved in some way inside the tail.

The TMP is predominantly positively charged, complementary to the negatively charged inner lining of the neck and tail (Fig. 3J), presumably providing resistance against the pressure from the packaged DNA. The TCP forms a connector between the DNA and the TMP, presenting a positively charged surface to the DNA and a negatively charged surface to the TMP (Fig. 3K–M). There are minimal interactions between TCP and TMP and either of the exterior proteins (HTCP, HTJP, TrP and MTP). Interactions between TCP and the HTCP dodecamer make up 60% of the total buried surface area between the interior and exterior neck proteins, about half (48%) of which involves two adjacent HTCP monomers that interface with a large positively charged surface at the base of the TCP extended loop (Fig. 3K–M), suggesting that this area is primarily responsible for stabilizing the position of the TCP, and thereby the DNA and TMP, in the neck. The HTCPs from HK97 and JBD30 lack the β-hairpin that interacts with TCP in 80α (Fig. S1A). It is unclear if the TCPs of these phages are located inside the neck and how they might interact with the HTCP³⁴.

DISCUSSION

Here, we have determined the structure of the neck of SaPI1 virions, consisting of structural proteins provided by the 80α helper phage. We were able to resolve all the proteins that form the connecting region between the portal and the major tail protein, as well as the DNA and proteins on the inside of the tail. We have also resolved a long-standing question in the field: the structure, location and interaction of the tail completion and tape measure proteins.

The terminology used to describe the proteins involved in the head-to-tail connection in various systems is somewhat confusing. In some systems, such as ϕ29 and P2, the portal is traditionally called the connector, though the term “connector” (or “head-to-tail connector”) is more often associated with the protein ring below the portal and sometimes to the whole complex of proteins between head and tail. We have chosen to refer to the protein making up the ring immediately below the portal as the head-tail connector protein (HTCP), and the ring below it as the head-tail joining protein (HTJP), consistent with usage in phage lambda (Table 1). In SPP1 the two proteins are referred to as “adaptor” and “stopper”, but this implies a functional role that has not been fully characterized in 80α. In some systems, the two proteins are referred to as Neck1 and Neck2 (Table 1). This organization with two “connector” rings (HTCP and HTJP) is common, but not universal: in the T5-like E. coli phage DT57C, there is no HTJP equivalent; instead, the HTCP appears to be highly modified to interact directly with the TrP⁵¹. The same might be true in JBD30, where the protein identified as “stopper” is, in fact, a TrP-like protein³⁴. It has been proposed that the types of HTCP present in lambda (gpW) and HK97 (gp6) are unrelated³⁹, but our analysis suggests that the two folds are in fact structurally and presumably evolutionarily related (Fig. S3, Table S2). The HTCP and HTJP proteins are also referred to collectively as “head completion proteins” to reflect the fact that they are added to the heads after assembly, but before joining with the tail^{10, 37}.

Similarly, “tail completion proteins” are proteins added to the tail prior to head-tail joining. This includes the tail terminator (TrP), which caps off the tail after addition of a sufficient number of rings of MTP⁴⁴. However, “tail completion protein” is also used to more specifically denote lambda protein gpZ and its homologs^{10, 39}, which includes 80α gp51, identified in this study. Until the present study, the structure and location of these ubiquitous proteins was unknown. In the 80α/SaPI1 tail, its location inside the neck suggests that it might be involved in both head-tail joining and DNA ejection, consistent with the observations in SPP1⁴⁵. To avoid confusion, we have chosen to retain the “tail completion protein” (TCP) terminology for gp51.

DNA packaging requires the terminase complex, consisting of TerS and TerL, which dock with the portal protein in the procapsids to initiate packaging. TerL, at least, must remain attached to the portal during packaging, presumably via the clip domain of the PP³². Since TerL is presumed to be a pentamer⁵², there is a symmetry mismatch between the portal and the terminase. Upon completion of packaging, TerL is removed and is replaced by the HTCP, which, like the portal, is a dodecamer and forms a very tight interaction with the PP³². How this exchange between TerL and HTCP is coordinated in order to avoid DNA escape is not clear. Presumably, the DNA remains inside the portal, perhaps involving the channel loops of the PP³², at least until the HTCP ring is added. In phage lambda, the HTCP itself, comprised of gpW, provides a narrow channel (20 Å wide) that might be sufficient to prevent DNA leakage^{47, 49}. In the 80α HTCP, this channel is 31 Å at its narrowest (Fig. 2E), probably too wide to prevent DNA escape. Instead, the hexameric HTJP ring might serve in this role. In SPP1, gp16 residues 40–60 in neck complexes extracted from tail-less, DNA-filled capsids formed a ring of α-helices that narrowed the diameter of the internal tunnel of the HTJP ring to ≈11Å and was thus proposed to act as a “stopper”³³. Although the 80α HTJP lacks this α-helix, and the HTJP tunnel is ≈36 Å wide at its narrowest, the AlphaFold2 model for gp50 predicted an α-helix in residues 43–52 (Fig. 2I), suggesting that the open state we observed might represent a reorganized protein that forms once the HTJP ring interacts with the tail (the ring formed by TrP). Once the tail is attached, the DNA would be blocked by the TCP, which occupies the central tunnel.

Long-tailed phages (siphoviruses and myoviruses) assemble their tails through a pathway separate from that of the head, followed by attachment to the heads after completion of DNA packaging. The TMP is a large protein that defines the length of the tail⁴⁴. By comparison with other systems, tail assembly is assumed to be nucleated from a complex made from Tal, TMP and the distal tail protein (Dit) and proceed by addition of hexameric MTP rings (39 or 40 rings in the case of 80α and SaPI1) until it reaches the end of the fully extended TMP, upon which the terminator TrP is added to cap off the stack of MTP rings⁴⁴. In phage lambda, the TCP (gpZ) is the last protein to be added to the tail prior to head-tail attachment⁴⁴. It was proposed that gpZ binds specifically to one end of the DNA⁵³. This may differ from 80α and other headful packaging phages, where the packaged DNA does not have a specific sequence at the ends. The location of gpZ is not known and it was not observed in high-resolution reconstructions of lambda virions^{47, 49}. Nor was the TCP observed in high-resolution reconstructions of JBD30 or DT57C^{34, 51}.

Here, we have shown that a monomer of TCP (gp51) is located inside the SaPI1 tail where it interacts with the N-termini of the TMP trimer in an asymmetric manner (Fig. 3). The tight five stranded β-sheet formed between TMP and TCP might suggest that the complex is formed early in tail assembly, but there is no additional evidence to support this conjecture. Indeed, gp51 was not seen by mass spectrometry in a crude procapsid preparation, which includes numerous tails and tail proteins⁴³, suggesting that the TCP is either added late in tail assembly or only after DNA packaging, for example during head-tail joining. Its location inside the HTJP—assumed to be a part of the head—also points to the TCP being the last protein to be added to the tail, maybe during head-tail joining. Insertion of the TCP into the HTJP ring might cause a reorganization and opening of the channel α-helices of HTJP (Fig. 2I), upon which the TCP itself may serve to plug up the tail and prevent the DNA from spontaneously ejecting (functionally a “stopper” protein).

Consistent with this role, absence of gp16.1 in SPP1 led to a 100-fold reduction in titer due to a failure to join heads and tails, and also affected injection of DNA through the cell wall⁴⁵. Similarly, in phage lambda, complete virions could be assembled in the absence of gpZ, but had 500-fold lower infectivity⁵³. The location of neither gpZ nor SPP1 gp16.1 has been ascertained, but in light of the similarity of their predicted structures to the 80α TCP, it seems likely that they could occupy a similar location inside their respective tails. The proposed location of the phage T5 TCP on the outside of the baseplate suggests that this protein may serve a different purpose⁵⁰.

In phage lambda, the DNA was observed far inside the tail, suggesting that it drops down during head-tail joining, ready for ejection⁴⁷. Likewise, in SPP1, between 55 and 67 base pairs of genomic DNA were protected by the tail⁵⁴. In contrast, in the SaPI1/80α tail, the DNA extends only as far down as the barrel of β-hairpins in the HTCP, where it is blocked from sliding further down by the TCP (Fig. 3). The same is the case in JBD30 and DT57C^{34, 51}. Thus, either the HTCP itself is capable of holding back the DNA until head-tail joining occurs, or the TCP would have to push the DNA back into the head, perhaps using the TMP as a spring.

In addition to its role in tail assembly and length determination, the TMP is also involved in the DNA ejection process, probably being ejected along with the DNA as a kind of “pilot” protein²⁵. We previously showed how the C-terminal α-helices of the TMPs interact with the α-helices in the Tal protein at the tip of the baseplate, providing a model for how interaction of Tal with the host cell membrane could cause a conformational change that is coupled to TMP release to initiate DNA ejection²⁵. Given the location of TCP inside the tail, it too presumably escapes together with the TMP and the DNA, consistent with the observed effect on DNA routing in SPP1⁴⁵. It is worth noting that the 80α minor capsid protein gp44, which serves to protect the DNA from degradation post injection, is also ejected together with the DNA^{55, 56}. It is not known whether this protein has a specific location in the capsid or the tail or whether it affects the organization of the DNA. We could not have observed gp44 in this study, because the SaPI1 construct that we used was made from an 80α Δ44 deletion, which we previously showed to slightly increase SaPI1 yields, while greatly reducing phage titers^{27, 55}.

In this paper, we have presented the structure of the SaPI1 neck, comprised of proteins encoded by the staphylococcal siphovirus 80α. In particular, we were able to provide a detailed description of the organization of the DNA, TCP and TMP. It seems likely that many other long-tailed phages, both siphoviruses and myoviruses, may share the same structure and organization of the TCP in the tail. In some cases, the failure to observe the TCP in high resolution reconstruction of phage necks and tails may be due to the application of threefold averaging during the reconstruction procedure⁵¹. It could also be that the TCP does not always have a consistent orientation relative to the outer neck proteins in all phages; indeed, in the 80α tail, the TCP is stabilized by minimal interactions with the HTCP. Additionally, phages employing sequence-specific DNA packaging (such as lambda) may employ other strategies for positioning their DNA in the tail and readying it for ejection.

RESOURCE AVAILABILITY

Lead contact

Request for further information and resources should be directed to the lead contact, Terje Dokland (dokland@uab.edu).

Materials availability

No unique reagents were generated in this study.

Data and code availability

• The final cryo-EM density maps and atomic coordinates were submitted to EMdep with identifiers EMD-48617 and EMD-48618 and Protein Data Bank identifiers 9MU2 and 9MU3. They are publicly available as of the date of publication. Accession codes are also listed in the key resources table.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

KEY RESOURCES TABLE

REAGENT OR RESOURCE	SOURCE	IDENTIFIER
Bacterial strains
ST65	Dearborn et al., 201127	N/A
Deposited Data
C6 neck structure	This paper	EMD-48618; PDB ID 9MU3
C1 neck structure	This paper	EMD-48617; PDB ID 9MU2
Software and algorithms
cryoSPARC	Punjani et al., 201759	https://cryosparc.com
UCSF Chimera	Goddard et al., 200760	https://www.cgl.ucsf.edu/chimera/
UCSF ChimeraX	Pettersen et al., 202161	https://www.cgl.ucsf.edu/chimerax/
ISOLDE	Croll et al., 201862	https://tristanic.github.io/isolde/
AlphaFold2	Jumper et al., 202136	https://github.com/googledeepmind/alphafold
AlphaFold3	Roy et al., 202463	https://alphafoldserver.com
Other
Vitrobot Mark IV	Thermo Fisher Scientific	N/A
FEI Titan Krios	Thermo Fisher Scientific	N/A
BioQuantum K3 detector	Gatan Ametek	N/A

STAR METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

The S. aureus strain used in this study, ST65, was derived from strain RN10628 by deletion of 80α ORF44 by allelic exchange, as described in Dearborn et al 2011²⁷. RN10628 is a derivative of the transformable S. aureus lab strain RN4220, containing SaPI1 tst:tetM as well as an 80α prophage³⁰. The cells were grown in CY broth at 32 °C as previously described^{27, 32}.

METHOD DETAILS

Production of SaPI1 virions

The production and purification of SaPI1 virions were described previously³². Briefly, SaPI1 virions were produced by mitomycin C induction of S. aureus strain ST65, which contains SaPI1 tst::tetM as well as an 80α prophage with a deletion of ORF44 (80αΔ44)²⁷. The virions were purified by PEG 6,000 precipitation, followed by CsCl and sucrose gradient centrifugation. Fractions containing SaPI1 particles were concentrated by pelleting and resuspended in phage dialysis buffer (20 mM Tris-HCl pH 7.8, 50 mM NaCl, 4 mM CaCl₂, 1 mM MgSO₄).

Electron microscopy

Cryo-EM was carried out as previously described³². SaPI1 particles were treated with 1μl Benzonase^® nuclease at room temperature for 1 hour prior to cryo-EM grid preparation, followed by dialysis on a 0.025 μm MCE Membrane filter (MF-Millipore) floating on phage dialysis buffer. Cryo-EM samples were prepared using a Vitrobot Mark IV with glow discharged nickel Quantifoil R2/1 grids, and imaged using an FEI Titan Krios microscope operated at 300 kV and equipped with a BioQuantum K3 energy filter and detector at the Midwestern Center for Cryo-Electron Microscopy (MCCEM) at Purdue University. A total of 2,796 images were collected at a magnification of 64,000 × (pixel size 1.33 Å), and an electron dose of 35.26 e⁻/Ų (Table S1).

Three-dimensional reconstruction of the SaPI1 neck

The electron micrographs of the SaPI1 virion were motion corrected in RELION-4⁶⁴. The rest of the reconstruction procedure was done in cryoSPARC⁵⁹, as outlined in Fig. S6. A total of 82,618 particles of full virion capsids were subjected to ab initio reconstruction with C5 symmetry, followed by icosahedral refinement, symmetry expansion to C1 and 3D classification (13 classes). After aligning on the Z axis and re-centering of the portal, the images were re-extracted and symmetry expanded from C5 to C1, followed by 3D classification, resulting in a data set of 59,457 particles that were used for the previously described C1 portal reconstruction³². These images were re-centered on the neck region and reconstructed with the application of C6 symmetry, reaching a final resolution of 3.12Å (FSC=0.143) (Fig. S1; Table S1). To resolve the internal structures, the particles from the C6 reconstruction were symmetry expanded to C1 (yielding a total of 356,742 particles). A mask was applied that only covered the internal density, followed by 3D classification without alignment with 10 classes. The class showing the most clearly resolved internal density (53,086 particles) was refined with a static mask after removing duplicates (final 35,724 particles). Subsequently, the mask was expanded to encompass the whole neck region and refined without symmetry. The final C1 reconstruction reached a resolution of 3.54 Å (Fig. S1; Table S1).

Model building

Initial models were made in AlphaFold2³⁶ and AlphaFold3⁶³. Model building was done in ChimeraX⁶¹ with refinement using ISOLDE⁶². RMSDs were calculated from within UCSF Chimera⁶⁰ using the BLOSUM-62 matrix, a pruning distance of 3.5Å, and a secondary structure weight of 50%.

QUANTIFICATION AND STATISTICAL ANALYSIS

The quantification and statistical analysis are integral parts of the cryoSPARC software used for reconstruction and ChimeraX and ISOLDE software used for model building and refinement, listed in the key resources table. Cryo-EM data collection, reconstruction and refinement statistics are provided in Table S1.

HIGHLIGHTS.

• SaPI1 virions are made of structural proteins encoded by S. aureus bacteriophage 80α

• The cryo-EM structure of the SaPI1 virion neck was determined

• The outer neck tube consists of a HTCP dodecamer and HTJP and TrP hexamers

• The neck structure reveals the tail completion protein and the TMP inside the tail