Structural atlas of the intact jumbo phage phiKZ

Abstract

Jumbo phage phiKZ, a key model for studying phage nucleus formation and bacterial defense mechanisms, possesses a highly complex tail machine that is essential for infection. Here, we present the structural atlas of the intact jumbo phage phiKZ by cryo-EM, thereby identifying 40 constituent proteins and unveiling its modular architecture. The virion, with a length of approximately 360 nm, is comprised of an icosahedral capsid of 2520 polypeptide chains from 11 proteins, and a massive tail machine of over 900 polypeptide chains from 29 proteins. The tail features a unique, multi-layered neck and a highly elaborate baseplate. The neck is reinforced by whisker-like proteins and anchors the contractile tail, which terminates in the baseplate. The baseplate is constituted by a central hub, an inner periphery of interlocking wedge heterotrimers and hexagonal rings, and an outer periphery with a striking hexagonal star configuration. This intricate peripheral region of the baseplate serves as an extended platform for twelve peripheral fibers, which mediate host cell adsorption. Our findings provide a structural framework for understanding jumbo phage assembly and infection, thus contributing to the foundation for future functional studies and rational engineering of these phages for potential therapeutic applications.

Jumbo phages are bacterial viruses with large genomes, oversized icosahedral capsids and complex tail architectures. Here, the authors use cryo-EM to provide a detailed structural analysis of jumbo phage phiKZ.

Introduction

Jumbo phages, a remarkable subgroup of bacteriophages, are characterized by exceptionally large genomes (typically exceeding 200 kbp) and unique structural-functional adaptations^1–3. These phages are distinguished by their oversized icosahedral capsids and complex tail architectures, which facilitate the efficient packaging and delivery of their large genomes. Notably, diverse jumbo phages encode specialized proteinaceous compartments, such as phage nucleus-like structures that physically exclude host nucleolytic defense systems, including CRISPR-Cas complexes and restriction-modification enzymes^4–8. The ecological importance of these phages is underscored by their capacity to infect a broad spectrum of bacterial hosts, including multidrug-resistant pathogens. This suggests their potential as tools for therapeutic and biotechnological applications^2,9,10. Despite their advantages, the study of jumbo phages remains challenging due to their genomic complexity and intricate structural features.

In recent years, the rapid advancement of cryo-electron microscopy (cryo-EM) has enabled the structural elucidation of numerous jumbo phage capsids at low to intermediate resolutions (approximately 10 Å). Representative examples include the capsid structures of phages G¹¹, phiRSL1¹², phiRSL2¹³, phiXacN1¹³, and Bellamy¹⁴. These structures have provided critical insights into capsid assembly mechanisms and functional organization. However, only a limited number of jumbo phage capsids have been resolved at high resolution, including the 4.1 Å resolution structure of the phiKp24 capsid¹⁵, and the 3.5 Å resolution structure of the phiKZ capsid¹⁶, which revealed a complex internal proteinaceous network.

Despite the advances made in the field of capsid studies, high-resolution structural characterization of jumbo phage tails remains a formidable challenge. The tail is a pivotal structural element for the process of infection, mediating host recognition, cell wall penetration, and genome ejection^15,17. Consequently, significant aspects, including the molecular details and assembly of tail proteins, the spatial distribution of tail fiber proteins, and the molecular mechanisms of tail remodeling during infection, remain to be fully elucidated. It is therefore vital to achieve high-resolution visualization of jumbo phage tails if a comprehensive understanding of their infection mechanisms is to be obtained, and if their potential in biomedical applications is to be realised.

Phage phiKZ is regarded as a prototypical jumbo phage model system¹⁸. It possesses a genome of approximately 280 kbp and is capable of infecting multidrug-resistant Pseudomonas aeruginosa¹⁹, suggesting its potential as a therapeutic agent against antibiotic-resistant pathogens. Furthermore, phiKZ is a pivotal system for the study of the phage nucleus and bacterial defense mechanisms against jumbo phages^5–7,20. As demonstrated in previous studies, the phiKZ capsid has a diameter of approximately 146 nanometers and displays canonical icosahedral symmetry with a triangulation number (T) of 27²¹. The capsid is composed of at least 30 proteins, and the tail consists of at least 32 different proteins^22,23. Preliminary cryo-EM studies at ∼18 Å resolution have demonstrated that the capsid is predominantly composed of the major capsid protein (gp120), with a small number of minor proteins located in the proximity of the capsid vertices²¹. However, the identities and functions of these minor components remained elusive for an extended period. Subsequent studies utilized cryo-EM to achieve a high-resolution structure of the phiKZ capsid, thereby facilitating the construction of an atomic model comprising 2520 polypeptide chains¹⁶. This structure revealed the presence of up to ten distinct minor capsid proteins that interact with major capsid proteins (MCPs), gp120. Specifically, proteins gp35 and gp244 have been identified as being located on the exterior surface at the capsid vertices, while the remaining eight minor proteins have been shown to form a complex network on the interior capsid surface¹⁶. In contrast to the advances in the capsid structure of phiKZ, the structure of the intact tail machine of phiKZ remains poorly characterized, with only low-resolution structure currently available²⁴. Despite the fact that the phiKZ tail adopts a typical contractile myophage architecture, the structural details of the neck, tail tube, tail sheath, baseplate, and fiber complexes are yet to be elucidated. Addressing these structural gaps is critical for elucidating the molecular basis of phiKZ infection and for deepening our mechanistic understanding of jumbo phage architecture and function.

In this study, we resolved the asymmetric structure of the phiKZ particle using cryo-EM, which enables us to elucidate the molecular architecture of the intact phiKZ virion, identify almost all the constituent proteins, and provide the first detailed structural insights into the previously unresolved neck, massive baseplate, and fibers of a jumbo phage. These findings provide valuable insights into the infection mechanism of jumbo phages, and contribute a structural foundation for utilizing phiKZ as a model system in structural virology and may inform the future development of novel phage-based therapeutic strategies against antibiotic-resistant pathogens.

Results

Overall structure of the intact phiKZ virion

Using cryo-electron microscopy single particle analysis, we reconstructed the three-dimensional structure of the Pseudomonas aeruginosa jumbo phage phiKZ. Firstly, we resolved the phiKZ head to ~3 Å resolution by local refinement and reconstruction method^25,26, thereby unveiling a structure comprising 11 major and minor capsid proteins, which is consistent with a previous report¹⁶. Next, we reconstructed the phage structure incorporating the unique portal–neck complex to a resolution of approximately 10 Å using our symmetry-mismatch reconstruction method²⁷. This asymmetric structure shows the portal and the multi-layered neck proteins connecting to the tail (Fig. 1a). However, despite obtaining a complete reconstruction of the head with its portal-neck, no discernible density corresponding to the proteinaceous inner body was observed within the capsid interior in the asymmetric structure (Fig. 1a). This finding precluded any further attempts at local refinement on this specific feature. The absence of the inner body signal is likely a consequence of positional heterogeneity relative to the portal-neck-focused alignment.

Fig. 1. Overall structure of phiKZ.

a Surface (left) and cut-open (right) views of the composite cryo-EM structure of the intact phiKZ virion, with dimensions indicated. Zoomed-in views of the neck and baseplate complexes are shown at center. b Schematic of the structural region of the phiKZ genome (colors as in panel a). Numbers appended to protein names denote genomic locus tags. Genes shown in gray encode capsid structural proteins, whereas those in white encode fiber structural proteins.

Subsequently, we reconstructed a 7.2 Å resolution map of the intact tail by imposing C6 symmetry using cryoSPARC software²⁸. The overall composite map indicates that the phiKZ virion, with a length of approximately 360 nm, is comprised of an icosahedral head with a diameter of ~145 nm and a tail with a length of ~215 nm, and the contractile tail sheath and tube terminate in a highly elaborate baseplate with a diameter of ~107 nm (Fig. 1a). Finally, a series of high-resolution local reconstructions were performed on different regions (Figs. S1–S4). The structures resolved in this study included the 3.5 Å resolution structure of the neck with C6 symmetry (Figure. S3), the 3.3 Å resolution structure of the adaptor with C12 symmetry (Fig. S3), the 3.1 Å resolution structure of the tail sheath and tube with helical symmetry, and the 3.5 Å resolution structure of the baseplate complex with C6 symmetry as well as at 4.0 Å with C3 symmetry (Fig. S4). All reconstructions exhibited high-quality side-chain densities (Fig. S5). Utilizing ModelAngelo software²⁹, we de novo built the atomic models for 20 open reading frames (ORFs) (Tables S1 and S2). By integrating AlphaFold3-prediction³⁰ and EMBuild software³¹, we built the backbone models for the nine proteins comprising the outer peripheral region and tail fiber (Fig. S6). The structural atlas of phiKZ presented in this study encompasses a total of 40 proteins (Fig. 1 and Table S1), thus demonstrating the intricate structural complexity of the intact tail machine architecture of the jumbo phage phiKZ.

Structure of the portal-neck complex

The portal-neck complex of phiKZ, with a height of 470 Å and a diameter of 330 Å, contains five components (Fig. 2a, b): a dodecameric portal (gp129), a dodecameric adaptor (gp42), a dodecameric stopper (gp99), a hexameric terminator (gp98), and a hexameric whisker (gp34 and gp144). The first four components bind sequentially to form a conserved channel for genome release and a connection between tail and head (Fig. S7a). The whisker collar surrounding the adaptor and terminator (Fig. 2a) is connected to the tail sheath via six copies of protein gp144 (Fig. 2c), a lytic transglycosylase³². The analogous architectural feature has been observed in bacteriophage Sf14³³.

Fig. 2. Structure of the phiKZ portal and neck.

a Surface view of the portal and neck regions with annotated dimensions. Individual protein names are annotated at the top. b Cut-open view of the portal-neck complex. Interactions between adaptor gp42 and stopper gp99 are shown in a blue dashed box. The C-terminal extension of the terminator gp98 projects towards the most proximal tail sheath ring, making contact with it (green dashed box). c The hexameric whisker interacts with the tail sheath via gp144. A single subunit of the hexameric whisker is composed of eight gp34 monomers and one gp144 monomer. The eight gp34 monomers assemble into four dimers, which in turn form a tetramer. d Side view of whisker gp34 binding to sheath gp29 via gp144, highlighting the interactions among subunits within the whisker gp34 tetramer and the connections between two adjacent whisker gp34 tetramers.

The portal protein gp129 shares highly conserved regions with its T4 counterpart³⁴, with significant variations restricted to the wing and barrel domains (Fig. S7b). Twelve gp42 molecules assemble into a conserved adaptor, linking the portal and the stopper (Fig. 2b). The basin-like stopper, comprising 12 copies of protein gp99, encapsulates the adaptor complex by the tight interaction with the C-terminal domain of gp42 (Fig. 2b). Notably, the N-terminal domains of the twelve gp99 proteins exhibit two distinct conformations. The six gp99 whose N-terminal domain forms a “basin floor” are designated gp99A, while the remaining six gp99 are designated gp99B (Fig. S7c). Through the two distinct configurations of the N-terminal domains, stopper forms tight interactions with the terminator and the whisker, respectively. The terminator, comprising six copies of gp98, fulfills two dual critical functions: it prevents further elongation of the tail tube and sheath, and reinforces the neck-tail junction (Fig. 2b). Similar to other myophages, the terminator of phiKZ also extends to the tail sheath and interacts with it (Fig. 2b). However, the terminator of phiKZ is larger and establishes tighter interactions with the sheath, which may facilitate the attachment of the massive tail complex in phiKZ.

The whisker ring of phiKZ is composed of 48 copies of gp34, with four gp34 dimers forming a tetramer (Fig. 2c), and six tetramers forming the hexameric whisker ring (Fig. S7d). The N-terminal domain of gp34 is composed of an antiparallel three-β-stranded sheet and an α-helix connected by loops, while the C-terminal domain of gp34 contains a core region, formed by a β-barrel and two α-helices connected by several loops, and an extended α-helix (Fig. S7e). In contrast to bacteriophage Sf14³³, the whisker of phiKZ is connected to the tail sheath by six gp144 monomers (Fig. 2a, c). Each gp144 likes a “wedge” configuration, situated between the tail sheath, terminator, and whisker (Fig. 2d). This sophisticated and precisely organized neck machinery may serve as a critical stabilizing element that connects the tail and head modules in the jumbo phage phiKZ (Fig. S8).

Structure of the tail sheath and tube

The 2000 Å-long helix tail of phiKZ, with a helical rise of 37.73 Å and a twist of 22.11°, is comprised of 44 stacked hexameric sheath rings, wrapping 43 stacked hexameric tube rings, to form a central lumen of approximately 68 Å (Figs. 1a and S9a and S10a). The tail sheath protein gp29 and tail tube protein gp30 exhibit a high degree of structural similarity to their counterparts in jumbo phage phiKp24¹⁵, with an RMSD of 1.34 Å and 1.07 Å, respectively (Figs. S9b, S10b).

The tail sheath protein gp29 is comprised of an inner and an outer sheath region (Fig. S9b). The outer sheath region projects radially outward along the helical tail and lacks in-subunit contacts within the same helical layer. The inner sheath region is constituted by an N-terminal domain, a C-terminal domain and two extensions (Fig. S9b). The two extensions of each gp29 interact with the C-terminal domain of the gp29 monomer in the underlying layer (Fig. S9c). The C-terminal domain also establishes connections with two neighboring sheath proteins in the upper ring. The atomic structure reveals that these extensions create a “mesh-like” structure to stabilize the tail sheath during contraction (Fig. S9c).

The tail tube protein gp30 exhibits two key features: firstly, a long C-terminal extension that can reach another gp30 in the adjacent upper layer, and secondly, an N-terminal domain that can extend toward a neighboring gp30 within the same layer (Figs. S9d and S10b). Adjacent-layer tube subunits are stabilized by the long C-terminal extensions that mediate intermolecular contacts (Figs. S9d and S10b). Our bioinformatic analysis, combined with AlphaFold3 predictions, reveals that this specific tail tube organization is conserved among other jumbo phages that share sequence homology with the phiKZ tail tube protein (Figs. S11 and S12). Moreover, the sheath-tube interface is stabilized by electrostatic complementarity, whereby the negatively charged inner surface of the sheath interacts with the positively charged outer surface of the tube (Fig. S10c).

Structure of the central region in the baseplate complex

The phiKZ baseplate complex is organized into three distinct regions (Fig. S13a): the central region, the inner peripheral region, and the outer peripheral region. The central region is comprised of six components (Fig. 3a): a tail tube initiator (gp88), a spacer (gp182), a hub (gp101), a spike (gp164), a spike tip (gp163.1), and a pin (gp174). Analogously to the inner baseplate structure of phage T4³⁵, the central region of the phiKZ baseplate possesses two rings at its junction with the tail tube (Fig. 3a–c). The first ring, designated as tube initiator, consists of six copies of the tube initiator protein gp88, which connects to the tail tube and initiates tube assembly (Fig. 3c). Each gp88 exhibits a high degree of alignment with the core domain and E-loop of the tail tube protein gp30 (Fig. S13b). The second ring, designated as spacer, is formed by six copies of the spacer protein gp182 (Fig. 3b, c). Each gp182 is composed of a β-sheet domain, an N-terminal domain, and an N-terminal extension (Fig. S13c). The β-sheet domains from the six gp182 monomers interconnect to form a 24-stranded β-barrel hexamer, onto which the tube initiator protein ring docks (Fig. 3c, d).

Fig. 3. Structure of the central region in the baseplate complex.

a Cut-open view of the central region in the phiKZ baseplate complex. Individual protein names are annotated at the right. The outer peripheral region and fibers of the phiKZ are rendered in white. The inner peripheral region and sheath of the phiKZ are rendered semi-transparent. b Side view of the ribbon models of the central region. c The interface between the tube initiator (first ring, blue) and the spacer (second ring, red). d Close-up view showing six gp182 monomers interconnecting to form a 24-stranded β-barrel hexamer. e Interactions between pin protein gp174 (pink), spike protein gp164 (red), and hub protein gp101 (cyan). f The spike inserts into the center of the pin ring and connects to the spike tip via the C-terminus of its β-helical domain.

The hexameric spacer ring is connected to the trimeric hub. Hub protein gp101, a homologue of the hub protein gp27 of T4³⁵, contains a β-barrel domain, a conserved central globular domain, and an extended β-hairpin (Fig. S13d). The β-hairpins from the three gp101 monomers extend to the OB-fold domains of the trimeric spike protein gp164 (Fig. 3e). Notably, the combined gp101 and gp164 proteins are homologous to the gp8 of PVC (Photorhabdus Virulence Cassette)³⁶, and share analogous features with their counterparts in T4³⁵. Six gp174 monomers, which exhibit two distinct conformations, form a pin ring (Fig. S13e). The N-terminal extension of one gp174 monomer interacts directly with the β-hairpin of hub protein gp101 (Fig. 3e). The spike protein gp164 inserts into the center of the pin ring, with the C-terminus of its β-helical domain connecting to the monomeric spike tip protein gp163.1 (Fig. 3b, f). In addition, weak electrostatic interactions exist between the pin protein gp174 and the spike protein gp164 (Fig. S14).

Structure of the inner peripheral region in the baseplate complex

The inner peripheral region of phiKZ is constituted by six baseplate wedges, six sheath initiators, and two hexagonal rings (designated as HR1 and HR2) (Fig. 4a, b). Each baseplate wedge is a heterotrimer, comprising one gp128 monomer and two conformers of gp87 (designated as gp87A and gp87B) (Fig. 4b). Each gp87 contains an N-terminal α-helical extension, N- and C-terminal domains, and a C-loop (Fig. S13f), while protein gp128 consists of an N-terminal domain and a C-terminal domain (Fig. S13g). The N- and C-terminal domains of gp87 are β-sheets-rich, whereas those of gp128 are α-helix-rich. The N-terminal regions of the three proteins that constitute a heterotrimer are approximately parallel, forming a core three-helical bundle (Fig. 4b, c).

Fig. 4. Structure of the inner peripheral region in the baseplate complex.

a Cut-open view of the inner peripheral region in the phiKZ baseplate complex. Individual protein names are annotated at the right. The outer peripheral region and fibers of the phiKZ are rendered in white. The central region and tube of the phiKZ is rendered semi-transparent. b Side view of the ribbon models of the inner peripheral region (left panel). The inner peripheral region comprises the sheath initiator, the baseplate wedges (upper right panel), HR1 (middle right panel), and HR2 (lower right panel). The right panels show top, side, and bottom views of the electron density maps for these three regions. The baseplate-wedge region consists of six heterotrimeric baseplate wedges, each composed of two conformers of gp87 and one gp128. HR1 is assembled from twelve gp49 dimers linked via handshake interactions. HR2 comprises six gp161 dimers. c Zoomed-in view of the black box in panel B (left) shows that the sheath initiator docks onto the three-helix bundle of the baseplate wedge. d Atomic model of the “handshake” interaction (the red dashed box in panel b) between two adjacent gp49 dimers within HR1. e Top (left panel) and bottom (right panel) views showing the relative positions of HR1 and HR2 to the pin (gp174). In HR2, six gp161 dimers are located beneath each vertex of the gp49 hexagonal ring. The central domain of each gp161 links to the pin gp174 in the central region of the baseplate.

The HR1 consists of 12 gp49 dimers, with 6 dimers located at the vertices of the hexagon and the remaining 6 dimers positioned at the midpoints of the six edges (Fig. 4b). Each gp49 monomer is composed of a core domain and two extended α-helices (Fig. S13h). Adjacent vertex and edge-midpoint dimers are tightly linked through a “handshake” interaction by a four-stranded antiparallel β-sheet (Fig. 4b, d).

The HR2 is constituted of six horn protein gp161 dimers (Fig. 4b). Each horn dimer is organized in a “Y-shaped”, bilaterally symmetric dimer, with two globular N-terminal domains, a central domain with four antiparallel β-sheets and four α-helices, and a handle domain formed by two long C-terminal α-helices (Fig. S13i). The handle domain of each horn is located on the HR2 vertex, corresponding to the HR1 vertex (Fig. 4b, e).

The six sheath initiator proteins (gp62) are sandwiched between the spacer ring and sheath layer, simultaneously connecting the inner gp182 and outer sheath subunits (Fig. S15) while also linking to the baseplate wedge’s α-helical bundle to secure the wedge complex (Fig. 4b, c). The six-wedge heterotrimers form a barrel-like structure, which is integrated with a “horseshoe-shaped” structure comprised of HR1 and HR2 (Fig. 4b, e). The C-loops of wedge proteins gp87A and gp87B extend to the C-terminal α-helix of horn gp161 and interact with the N-terminal domain of HR1 gp49, respectively (Fig. 4b). The globular N-terminal domains of each horn gp161 in HR2 connect to the N-terminal domains of adjacent edge-midpoint gp49 in HR1, while the central domain of each gp161 links to the pin protein gp174 in the central region of the baseplate (Fig. 4b, e).

Structure of the outer peripheral region in the baseplate complex

The outer peripheral region surrounds the inner periphery (Figs. 5a, b, and S16) and functions as the fiber attachment platform. Two gp139 dimers and one gp27 monomer bridge these two regions, stabilizing the overall architecture (Figs. 5b and S17a, b). The outer peripheral region adopts a remarkably intricate hexagonal star-like configuration with an overall diameter of ~1070 Å, defined by six star lobe complexes positioned at the vertices of the star (Figs. 5c and S16). These star lobe complexes are further interconnected around the baseplate by a ring composed of 18 gp157 dimers (Figs. 5d and S16, 17c), thereby reinforcing interactions within the outer periphery. Each star lobe complex contains five gp139 dimers, five gp26, two gp27, one gp127, and two gp130 proteins (Figs. 5e and S17a–f). Structurally, each star lobe complex can be subdivided into two sub-units (designated as unit1 and unit2) (Fig. 5e–g).

Fig. 5. Structure of the outer peripheral region in the baseplate complex.

Protein names and color codes are shown in the upper left. Key proteins (gp157, gp139, gp130, and gp26) are highlighted. Side (a) and top (b) views: Cryo-EM density of the outer peripheral region of the phiKZ baseplate. The outer peripheral region comprises six star-lobe complexes (c) and a ring of gp157 dimers (d). In (c), gp130 connects neighboring star-lobe complexes (red dashed circle) and links unit1 and unit2 within the same star-lobe complex (black dashed circle). Gp139 (orange dashed ellipse) extends across the outer peripheral region, adopting a C-shaped architecture. e Top: the star lobe complex subdivided into unit1 and unit2. Middle: three gp157 dimers associated with a single star lobe complex. Bottom: density map of a star lobe complex within the outer peripheral region of the baseplate, superimposed on the corresponding protein models. Constituent protein models of unit1 (f) and unit2 (g).

Unit 1 is constituted of five gp139 dimers (designated as dimers A–E), one gp127, one gp130 (designated as gp130A), two gp27 (designated as gp27 A and B), and three gp26 (designated as gp26 A-C) proteins (Figs. 5f and S17a; 17f). The gp139 dimer A establishes contact with the baseplate wedge protein gp128. The remaining gp139 dimers (B–E) align sequentially with gp139 dimer A, thereby forming a continuous band that traverses the platform and terminates at gp127 (Figs. 5f and S17a; S17b). The two proteins gp27A and gp27B (each 898 residues) connect to gp139 dimer A and gp139 dimer E (Fig. S17a), respectively, and together constitute the structural framework of unit1. The gp26B protein inserts its C‑terminus into the gp157-based brace ring (Fig. 5a). The gp26C interacts with gp130A within the same star lobe, thereby stabilizing the interface between unit1 and unit2 (Figs. 5c and S17e–f).

Unit 2 comprises one gp130 (designated as gp130B) and two gp26 (designated as gp26D-E) proteins (Figs. 5g and S17e). The gp26D connects to the C-terminus of the gp26C in the unit1 (Fig. S17e). The gp130B monomer bridges across neighboring star lobes to promote lateral inter-star lobe connectivity (Fig. 5c), thereby contributing to the outer peripheral region closure and mechanical rigidity.

Notably, gp27A simultaneously connects to both the unit1 and unit2 (Fig. 5e, f), binding two highly similar regions. Each region is formed by the N-terminal extensions of two gp26 proteins: gp26A and gp26B in unit1 and gp26D and gp26E in unit2 (Figs. 5f, g and S17e). However, due to the poorly resolved density map, a complete model of the gp26 N-terminal region could not be built, likely because of its inherent flexibility. In addition, at the interface between gp130 and an adjacent gp139, we identified an additional density, which could be fitted with the C‑terminus of gp26 (Figs. 5g and S17d).

In summary, the structural integrity of the outer peripheral region is orchestrated by five proteins with highly specialized functions. Protein gp27 constitutes the core framework of the region, and gp26 is responsible for organizing its internal components by cross-linking them. The entire region is secured to baseplate wedges via gp139, which mediates this critical interface. The primary function of gp130 is to facilitate the connection of adjacent star lobe complexes, as well as the unit1 and unit2 within the same star lobe complex. The overall region is further stabilized by gp157, which fortifies the linkages between neighboring star lobe complexes.

Structure of the hetero-hexameric fiber complex

In the cryo-EM density map of the outer peripheral region in the baseplate, we identified independent fiber‑attachment sites located beneath protein gp130 in both unit1 and unit2 (Fig. 6a). Analysis of these densities indicated that the two fiber regions are almost indistinguishable in architecture. Each region comprises six gp131-like N-terminal domains and a helical bundle formed by six α-helices (Fig. 6b). Protein gp131 contains numerous paralogs in the phiKZ tail genome, including gp132, gp133, gp134, and gp135. Among them, gp133, gp144, and gp135 exhibit a high degree of similarity with the N-terminal region of gp131 (Fig. 6b). However, this similarity is not observed in its C-terminal domain³⁷.

Fig. 6. Heterohexameric organization of the phiKZ tail fibers.

a Cryo-EM density map of the central and inner peripheral region of the baseplate (transparency), outer peripheral region of baseplate (colored identical to Fig. 1), and tail fibers (gray, orange). The tail fibers are anchored to the baseplate outer peripheral region via the orange region, which comprises three gp131-like N-terminal segments. Insets below show that both the unit1 fiber and unit2 fiber (low-pass filtered to 6 Å) feature a characteristic six-helix bundle and are nearly identical in architecture. b Six N-terminal segments from gp131, gp133, gp134, and gp135 were fitted into the density maps of the unit1 fiber and unit2 fiber. The lower panel shows AlphaFold3-predicted models of the N-terminal regions of gp131, gp133, gp134, and gp135. Rigid-body fitting indicates high structural similarity among these N termini.

The fiber-attachment sites connect downward to an extended region (Figs. 6b and S17g). However, the density is inadequate to build an atomic model. Notably, the C-terminal domain of gp131 adopts a seven-bladed propeller fold, a motif frequently associated with enzymatic activity within this protein family³⁷. Based on these observations, both unit1 and unit2 fiber docking sites engage the candidate hetero-hexameric basal complex of the tail fiber, composed of gp131×3, gp133, gp134, and gp135, which connect downward to a gp131 trimer (Figs. 6b and S18a–c). We compared the structural characteristics of the C-terminus of gp131 with AlphaFold3-predicted homologs from other jumbo phages, including possible tail fiber or tailspike proteins³⁷: gp184 from phage psa21, gp216 from phage 201phi2-1, and gp150 from phage phiPA3. Structural modeling shows that the C-terminal domain of gp131 is located at the distal end of the tail fiber (Fig. S18b, c). Together with the observed structural similarity and the conserved sequence features reported in previous studies³⁷, these findings suggest that the C-terminus of the gp131 trimer may contribute to host recognition or facilitate the initial stages of bacterial infection (Fig. S18d). These docked assemblies mediate the stable anchoring of two fibers per star lobe, distributing mechanical load onto the integrative framework jointly formed by the band-like gp139 anchors, the cross-linking gp26 network, and the star-shaped gp157 brace. Therefore, each star lobe anchors two fibers, and the entire outer peripheral region supports twelve fibers in total (Fig. S18a–c).

Discussion

In this study, we present a comprehensive structural atlas of the intact jumbo phage phiKZ virion, generated by high-resolution cryo-EM and multiple rounds of local reconstruction. The findings of this study establish that phiKZ represents one of the most structurally intricate bacteriophage assemblies resolved to date, incorporating at least 40 distinct structural proteins, including 28 that comprise its highly intricate tail machinery. This remarkable complexity is manifest in the modular organization and intricate interconnections among the portal-neck, tail sheath and tube, baseplate, and tail fiber, highlighting the extensive structural and assembly innovations that have evolved in jumbo phages.

A defining feature revealed by our structure is the intricate portal-neck region, which contains a sophisticated connector complex composed of the portal (gp129), adaptor (gp42), stopper (gp99), whisker (gp34), transglycosylase (gp144), and terminator (gp98) proteins. It is imperative to note that the configuration of two concentric arrays of six whisker-like gp34 assemblies in phiKZ, which are directly interconnected and further interact with the tail sheath via the protein gp144 acting as a molecular wedge, serves to reinforce the neck architecture. This arrangement is presumably pivotal for the stability of the connection between the large head and the massive tail. This organization differs markedly from smaller phages; for example, although phage Sf14 possesses similar whisker-like structures³³, they are only connected to the tail sheath through a loop and lack the direct interconnection observed in phiKZ. The tail sheath (gp29) and tube (gp30) display remarkable structural conservation with those of other jumbo phages, such as phiKp24¹⁵, suggesting a conserved mechanism for tail contraction and DNA delivery. In contrast, the phiKZ baseplate complex exhibits unique architectural innovations. The baseplate is organized into three regions: a central region, an inner peripheral region, and an outer peripheral region. The central region, which houses proteins such as gp174, acts as a central hub. The hub is encircled by the sophisticated inner peripheral region, which is built with a barrel-like core of six heterotrimeric wedges. This core is further reinforced by two interlocking, horseshoe-shaped hexagonal rings (HR1 and HR2). The stability of this entire assembly is critically ensured by the sheath initiator gp62, which anchors the wedge complex to the central region and the tail sheath. Furthermore, the integration between these two regions is direct and extensive; for instance, the central domains of the gp161 horns in the outer HR2 ring connect downward to the central hub protein gp174. This multi-layered, concentric organization, ranging from the central hub to the interlocking rings of the inner periphery, provides a robust mechanical foundation for the entire baseplate.

Most strikingly, the outer peripheral region of phiKZ features an elaborate arrangement in a hexagonal star configuration. This arrangement is maintained by a specialized protein network with distinct roles, providing an extended platform for fiber attachment. The protein gp27 constitutes the core scaffold, while gp26 cross-links internal components and gp139 mediates the critical attachment to the baseplate wedges. This star-shaped assembly is further stabilized by gp130, which connects adjacent star lobes, and a surrounding ring of gp157 dimers. The assembly of this massive network employs sophisticated strategies, including the use of flexible linkers and domain extensions, allowing for both structural rigidity and adaptability during virus assembly. Collectively, these features underscore the extraordinary architectural and assembly sophistication of jumbo phage phiKZ, and provide a valuable framework for understanding the construction and evolution of large viral machines.

Our high-resolution structure elucidates the functional aspects previously inferred from low-resolution studies. Despite the indications from earlier studies that phiKZ possesses six baseplate-attached fibers that function analogously to T4 short tail fibers^24,38, our structure reveals a more intricate organization with twelve peripheral fibers, six located at the unit1 positions and six at the unit2 positions of the star lobe in the outer peripheral region. These twelve flexible fibers function analogously to the short tail fibers of T4 in mediating host adsorption^17,24,38. Notably, previous studies had proposed that the C-terminal seven-bladed propeller fold of gp131 and its paralogs play a role in host cell binding³⁷. Our observation of low-resolution density, fitted with AlphaFold3-predicted models of this domain, lends support to its positioning for initial attachment to the bacterial surface.

Recent cryo-electron tomographic studies of phiKZ infection have directly visualized the process of tail attachment and genomic transfer³⁹. Subsequent to the adsorption of the fibers to the bacterial outer membrane, the tail undergoes a contraction of approximately 100 nm. This contraction drives the spike complex and tail tube through both the outer and inner membranes, delivering the phage genome into the cytoplasm. In phiKZ-like jumbo phages, the ejected genome is then rapidly encapsulated within a specialized compartment known as the early phage infection (EPI) vesicle, which is derived from the host’s inner membrane^39–41. In contrast, for phages like T4¹⁷, the genome is released directly into the cytoplasm without the formation of such a protective structure. The portion of the tail tube remaining outside the cell is about 50 nm in length, matching the length of the phiKZ fibers. This observation suggests that the fibers, similar to the short tail fibers in T4, play a pivotal role in anchoring the phage vertically to the bacterial surface during infection, thereby stabilizing the phage particle throughout the infection process.

Integration of these functional observations with our high-resolution structure provides a molecular framework for understanding the conformational changes during infection. Once the fibers have stably anchored phiKZ to the host cell surface, a series of conformational changes is triggered in the baseplate and tail fiber network. Based on structural changes observed in myophages T4³⁵ and E217⁴² before and after contraction, we propose an analogous structural transition for phiKZ: the peripheral region of the baseplate undergoes outward expansion, opening a channel, concurrent with contraction of the tail sheath. A key step likely preceding this transition is the disengagement of the central spike (gp164) from the pin ring (gp174). Our analysis reveals these components are connected by weak electrostatic interactions (Fig. S14), a feature reminiscent of the weak contact between the gp44 pin domain and the tail tip in phage E217⁴², which serves to retain the tip prior to infection. It is plausible that the conformational changes in the phiKZ baseplate disrupt this weak interaction, leading to the release of the spike complex. This coordinated process drives the central region of the baseplate, including the spike complex (composed of spike gp164, tip gp163.1, gp101, and gp182), which is structurally analogous to the T4 cell-puncturing device^17,35,43, together with the tail tube, to penetrate the host cell envelope. Similar to the dissociation of T4’s central spike proteins gp5/gp5.4, after traversing the outer membrane¹⁷, it is hypothesized that the homologous proteins gp164 and gp163.1 in phiKZ are also likely to dissociate following membrane penetration. Collectively, these structures provide a comprehensive molecular model that elucidates the manner in which phiKZ orchestrates its intricate apparatus to facilitate a successful host infection. This detailed structural framework will guide future mechanistic studies and may facilitate the rational engineering of these giant phages for potential therapeutic and biotechnological applications.

Methods

Purification of bacteriophage phiKZ

PhiKZ virions were prepared from infected P. aeruginosa PAO1 lysates using an established method⁴⁴. The P. aeruginosa strain PAO1 (ATCC 15692) was cultured in Luria-Bertani broth for 12 h at 37 °C. The P. aeruginosa PAO1 cells were then infected with phiKZ phages and incubated for 6 h at 37 °C. After incubation, cell debris was removed by centrifugation at 9000 × g for 60 min at 4 °C. The resulting supernatant was enriched by precipitation with polyethylene glycol (PEG) 8000 (10% w/v PEG in 1 M NaCl) overnight at 4 °C. The precipitated phages were resuspended in phage buffer (50 mM Tris-HCl and 10 mM MgCl₂, pH 7.4) and further purified by cesium chloride density gradient centrifugation using densities ranging from 1.2 to 1.6 g/ml. Centrifugation was performed at 120,000 × g for 4 h at 8 °C. After centrifugation, the bands containing phage particles were carefully collected and dialyzed in phage buffer (50 mM Tris-HCl and 10 mM MgCl₂, pH 7.4) overnight. Finally, the purified phiKZ phages were stored on ice prior to cryo-sample preparation.

Cryo-EM and data collection

3 μL of the purified phiKZ virion was carefully pipetted onto 300-mesh quantifoil R2/1 copper grids that had been glow-discharged for 30 s. The grid was blotted with filter paper for 3.0-4.0 s at 8 °C and 100% relative humidity, and then rapidly plunge-frozen in liquid ethane using a Thermo Fisher Scientific (TFS) Vitrobot Mark IV. Subsequently, the grids were imaged on a TFS Titan Krios G3i transmission electron microscope operated at a voltage of 300 kV, which was equipped with a Gatan BioQuantum energy filter and a K3 direct electron detector. The energy filter was set to a slit width of 20 eV.

Automated data acquisition was performed using TFS EPU software at a nominal magnification of 53,000×, corresponding to a pixel size of 1.36 Å. Finally, 8018 movies were recorded, with each movie fractionated into 32 frames and a total electron dose of 30e⁻/Ų. Defocus values ranged from −1.6 to −2.2 μm.

Image processing

A total of 8018 micrographs were generated after motion correction using MotionCor2⁴⁵. These micrographs were imported into cryoSPARC v4.6.0²⁸, and Patch CTF was employed for Contrast Transfer Function (CTF) estimation, which accounts for local defocus variations to enhance data quality. Following this, the Manually Curate Exposures tool was utilized to select micrographs, yielding a final dataset of 7367 high-quality micrographs for further analysis.

The phiKZ capsid particles were boxed using ETHAN software⁴⁶. The icosahedral capsid of phiKZ was reconstructed by using our program icosprocess²⁵ based on the common-line algorithm^47,48. To further improve the resolution of the structures, we applied local refinement and reconstruction to the phiKZ capsid by using local reconstruction. The local refinement and reconstruction were performed iteratively to improve the resolution until the orientations and centers in all datasets were stabilized, and the structural resolution could not be further improved (~3 Å).

To reconstruct the asymmetric structure of phage phiKZ head, we utilized our CryoIM²⁷, starting with the orientation parameters from the initial icosahedral reconstruction. Briefly, for each image of the phiKZ particle, we initially localized the unique vertex with the portal–neck by searching the 12 icosahedral vertices that were localized during the icosahedral reconstruction step. Subsequently, a low-resolution structure of the phage incorporating the portal–neck complex was obtained without imposing symmetry. This model was then iteratively refined to enhance resolution. In each cycle, the best matching asymmetric orientation was determined by correlating each particle image against 60 icosahedrally-equivalent projections. The iteration continued until convergence, yielding a final map of the phage structure with the portal–neck at approximately 10 Å resolution. However, no discernible density features corresponding to the inner body were observed within the asymmetric structure of the capsid.

All subsequent tail data processing steps were performed using cryoSPARC v4.6.0²⁸. For the overall tail reconstruction, 8843 particles were manually picked. These particles were extracted using a smaller Fourier-cropped box size (1800 pixels to 800 pixels). An initial volume of the tail was generated by imposing C6 symmetry during ab-initio reconstruction. The best classes were then selected and subjected to non-uniform refinement, resulting in a final 3D reconstruction of the overall tail structure at a resolution of 7.61 Å.

For the neck region, 32,911 particles were manually picked and extracted at a box size of 480 pixels. After 2D classification, 32,554 particles were retained. An initial volume of the neck was generated using a subset of these particles through ab-initio reconstruction. The particles from 2D classification and the initial model were then subjected to homogeneous refinement and non-uniform refinement, resulting in a final neck structure at a resolution of 3.52 Å (Fig. S3).

To resolve the C12-symmetric portal and adapter, the particles from the final neck reconstruction were shifted by 61 pixels along the Z-axis to generate an initial model. This model was subjected to homogeneous refinement and non-uniform refinement with C12 symmetry imposed, resulting in a refined initial model. Using the orientations from this refined model, 32,896 particles were re-extracted at a box size of 360 pixels. These particles were processed through 2D classification, followed by two rounds of homogeneous refinement and non-uniform refinement with C12 symmetry imposed, ultimately yielding a portal structure at a resolution of 3.37 Å (Fig. S3).

For the tail tube, an initial set of templates was generated using Create Templates, and the generated templates were then used by the tail tube template picker to automatically select particles. Perform 2D classification on the picked particles and remove classes containing poor-quality particles. Based on these classes, particles were re-extracted at a box size of 400 pixels, and 26,028 particles were selected through 2D classification. These particles were subjected to ab-initio reconstruction and homogeneous refinement with C6 symmetry imposed, resulting in a tail tube structure at a resolution of 3.15 Å, after applying helical refinement.

For the baseplate, 29,126 particles were manually picked and extracted at a box size of 400 pixels in cryoSPARC. The initial volume of the baseplate was generated using a subset of these particles through ab-initio reconstruction, followed by multiple rounds of 2D classification, homogeneous refinement, and non-uniform refinement. Given the unique morphological features of the phiKZ baseplate, 22,539 particles exhibiting clear baseplate characteristics were identified from the full dataset through 2D classification. In cryoSPARC, the particles selected from 2D classification and the initial volume were subjected to non-uniform refinement with C3 symmetry imposed. The resulting map was further refined using non-uniform refinement with C6 symmetry applied, and the “Optimize Per-Particle Defocus” option was enabled. This process yielded a final baseplate structure at a resolution of 3.65 Å (Fig. S4).

However, we observed that the resolution of the central region was notably lower after applying C6 symmetry during the baseplate reconstruction. To further improve the resolution of the central region in the baseplate, we started from the C3 symmetry-imposed baseplate structure obtained through non-uniform refinement and subjected it to homogeneous refinement, followed by C6 symmetry expansion. The expanded particles, along with a local mask generated around the tail tip using Chimera⁴⁹, were subjected to 3D classification. The resulting classes were then refined locally. The centers of the locally refined particles were shifted by 70 pixels along the Z-axis, and 27,796 particles were re-extracted at a box size of 300 pixels after removing duplicates using the Remove Duplicate Particles tool. These particles were processed sequentially through Reconstruct Only, Local Refinement, and Homogeneous Refinement, ultimately yielding a tail tip structure at a resolution of 4.02 Å (Fig. S4).

For the outer peripheral region in the baseplate and fiber region, C6 symmetry expansion was initially applied to the 3.65 Å baseplate complex and subsequently relocated all particle centers to the baseplate peripheral junction before extracting 135,070 particles (400 pixel box). These particles were then subjected to both Reconstruction Only and Local Refinement processes, producing a map of the outer peripheral attachment site at 4.14 Å resolution. To further resolve the full outer periphery, a mask was generated in ChimeraX⁵⁰ and applied in local refinement, resulting in a structure of the entire star lobe complex at 4.7 Å resolution. Subsequently, two additional masks were generated in ChimeraX⁵⁰ for the unit1 and unit2 of the star lobe for separate Local Refinement, achieving final resolutions of 4.39 Å and 4.55 Å, respectively (Fig. S4).

Model building

For the neck, tail tube, sheath, and baseplate regions of phiKZ (excluding outer peripheral region and tail fibers), we used ModelAngelo²⁹ to automatically build atomic models for 20 structural proteins based on the cryo-EM density maps (Table. S1). The resulting models were then manually refined and adjusted using Coot software⁵¹. For the outer peripheral region and fiber components, we first utilized AlphaFold3³⁰ to predict in silico structures of all putative tail-associated proteins encoded by phiKZ^22,23 (Fig. S19). After excluding proteins that had been previously identified in the tail structure, we obtained 26 potential candidates for phiKZ proteins. ModelAngelo²⁹ was then used to automatically construct atomic models for the outer peripheral region and fiber components at resolutions of 4.39 Å and 4.55 Å, respectively. Through systematic comparison between the 26 AlphaFold3-predicted structures and the ModelAngelo-generated atomic models, we docked the predicted structures of gp27, gp26, gp127, gp130, gp157, gp131-N, gp133-N, gp134-N, and gp135-N into the cryo-EM density maps (Fig. S6). The fitting process was initiated using EMBuild³¹ for gp26 and gp157, which exhibited substantial structural flexibility, followed by meticulous manual adjustment in Coot to optimize the structural alignment with the experimental density (Fig. S6). This hybrid approach, combining AI-based prediction, automated modeling, and manual refinement, enabled comprehensive structural determination of the phiKZ tail apparatus. All atomic models were further refined using Phenix real-space refinement⁵². The refinement and validation statistics are summarized in Table S2. All structural figures were generated using UCSF Chimera⁴⁹ and ChimeraX⁵⁰.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

H.L. and H.X. conceived, initiated, and coordinated the project; H.X. produced the phiKZ sample and prepared cryo-EM grids; H.X. and Y.P. collected the cryo-EM data; H.X., Z.P., J.Z., Y.C., Y.T., W.C., and H.L. processed the cryo-EM data; H.X., Z.P., J.Z., T.L., and S.H. built and refined the structure model; H.X., Z.P., and H.L. wrote the manuscript. All authors discussed the experiments and results and read and approved the manuscript.

Peer review

Peer review information

Nature Communications thanks Sarah Doore, who co-reviewed with Hailey Kerns, and the other, anonymous, reviewer for their contribution to the peer review of this work. A peer review file is available.

Data availability

The cryo-EM maps have been deposited at the Electron Microscopy Data Bank under accession codes EMD-67125 (overall tail), EMD-67123 (unit1), EMD-67124 (unit2), EMD-67095 (portal-adaptor), EMD-67102 (neck), EMD-67096 (sheath-tube), EMD-67121 (baseplate C6) and EMD-67098 (baseplate C3), EMD-67097 (docker), and EMD-67111 (outer peripheral). The atomic models have been deposited at the Protein Data Bank under accession codes 9XPE (portal-adaptor), 9XPS (neck), 9XQS (baseplate C6), 9XPF (sheath-tube), 9XPH (baseplate C3), 9XPG (docker), and 9XQD (outer peripheral). All other data are available in the main text or the supplementary materials.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-71561-2.