Proteomic Analysis of Marine Bacteriophages: Structural Conservation, Post‐Translational Modifications, and Phage–Host Interactions

ABSTRACT

Marine bacteriophages, the most abundant biological entities in marine ecosystems, are essential in biogeochemical cycling. Despite extensive genomic data, many phage genes remain uncharacterised, creating a gap between genomic diversity and gene function knowledge. This gap limits our understanding of phage life cycles, assembly, and host interactions. In this study, we used mass spectrometry to profile the proteomes of 13 marine phages from diverse lifestyles and hosts. The analysis accurately annotated hypothetical genes, mapped virion protein arrangements, and revealed structural similarities among phages infecting the same host, particularly in tail fibre proteins. Protein structure comparisons showed conservation and variability in head and tail proteins, particularly in key domains involved in virion stabilisation and host recognition. For the first time, we identified post‐translational modifications (PTMs) in marine phage proteins, which may enhance phage adaptability and help evade host immune systems. These findings suggest that phages optimise their infection strategies through structural variations and PTM modifications, improving their adaptability and host interactions.

This study used proteomics to determine the relative abundance of proteins in 13 marine phages, constructing detailed phage architecture models. Protein structure comparisons revealed evolutionary traits, while identified post‐translational modifications (PTMs) suggest that phages utilise PTMs to evade host immune defences, providing new insights into phage adaptability and infection strategies.

1. Introduction

In the vast expanses of surface seawater, there is a dense microbial landscape dominated by viruses, with an average of 10⁷ particles per millilitre (Fuhrman 1999). These viruses, which mostly infect bacteria, play crucial roles in shaping bacterial population dynamics, altering community structures, facilitating the release of dissolved organic matter, and thereby influencing global biogeochemical cycles (Weinbauer 2004; Wommack and Colwell 2000; Suttle 2005). Despite their modest genome sizes, bacterial viruses (phages) display remarkable genomic diversity and intricate evolutionary relationships (Dion et al. 2020). Advancements in high‐throughput genome sequencing have significantly deepened our understanding of the diversity of marine viruses. Nevertheless, a substantial portion of marine viral genomes, accounting for approximately 82% of the marine virome (Gregory et al. 2019), remains poorly annotated and functionally enigmatic and thus is often described as ‘viral dark matter’.

While genomic analysis can reveal the diversity of viral genes and deduce their potential function, relying solely on genomic data cannot fully explain the biological roles and functionality of virus proteins. Complementary to genomic analysis, proteomics offers a robust approach to improve the annotation of viral genes, especially the structural components of virions (i.e., the complete infectious virus particle) (Kunec 2013) and has been used in studies of virus isolates (Kropinski et al. 2011; Frampton et al. 2015; Chan et al. 2015; Ahamed et al. 2019; Yuan and Gao 2016; Allen et al. 2008). In addition, structural biology has shed light on various facets of phages, including overall virion structure (Lander et al. 2012; Chen et al. 2017; Liu et al. 2019), the structures of essential viral proteins (Wang et al. 2018; Seul et al. 2014; Flayhan et al. 2014), and the mechanisms of adsorption and DNA injection (Bárdy et al. 2020; Cuervo et al. 2019; Arnaud et al. 2017), significantly enhancing our knowledge of virus functionality and host interactions. Interestingly, structural conservation among viral proteins is often maintained even in the absence of nucleotide and amino acid homology (Duda and Teschke 2019), highlighting an evolutionary trajectory that prioritises functional conservation over sequence similarity. For example, structural domains in virus proteins, such as the double jelly roll (DJR)‐fold and HK97‐fold found in phage major capsid proteins, are conserved across diverse sequences (Dion et al. 2020), underscoring their evolutionary significance. Although large‐scale experimental protein structure determination remains challenging, the rapid advancement of protein structure prediction methods (Georjon and Bernheim 2023) enables us to further elucidate the structural and functional implications of proteins identified through proteomic analysis.

Understanding the distribution of proteins within the virion is crucial for deciphering virion assembly and initiating the functional characterisation of structural proteins (Owen et al. 2015). Proteomics not only facilitates the identification of protein compositions and molecular structures (Angel et al. 2012) but also helps with predicting protein domains and structures (Johnson et al. 2017) and identifying post‐translational modifications (PTMs). PTMs play critical roles in regulating protein function, stability, and interactions. Studies on HIV, SARS‐CoV‐2, and influenza virus have shown that PTMs of viral structural proteins can help these viruses evade host immune responses (Li et al. 2021). This underscores the importance of PTMs in studying virus–host interactions. However, no research has yet reported PTMs in marine phages. Integrating bioinformatics analyses of the proteome can help to bridge the gap between genomic annotations and structural realities (Johnson et al. 2017). A few studies (Brum et al. 2016; Tschitschko et al. 2015, 2016; Xie et al. 2018) have attempted to identify viral dark matter in environmental samples using metaproteomics, significantly enhancing our understanding by revealing viral structural proteins, improving the functional annotations of viral dark matter, and supporting ecological research. However, no research has been conducted on viral isolates from different hosts, and much more work is required to fully uncover and understand the complexities of viral genomes and their functions, especially regarding the PTMs of marine viral proteins.

In this study, we employed proteomic techniques, specifically mass spectrometry, to analyse the proteomes of 13 diverse marine phages, including both virulent and temperate phages that infect different bacterial hosts. The comprehensive annotation of these phage proteins using proteomic methods was used to enhance the gene annotations. We generated predictive architectural models of the phages to illustrate the spatial distribution of structural proteins and provide visual insights into the viruses' assembly mechanisms and interactions with host cells. Our findings reveal conservation within phage structural proteins, even in the context of low sequence similarity, while variable domains likely contribute to evolutionary adaptability. Furthermore, our analysis revealed significant PTMs such as phosphorylation and acetylation in marine phage structural proteins. This discovery has not been previously reported in marine phages. These PTMs, which reflect the complex regulatory mechanisms phages use to thrive within their environments, may be crucial for phage adaptability and their potential evasion of host immune responses.

2. Materials and Methods

2.1. Preparation and Purification of Phages

The studied phages included vB_EliS‐L02 (Li et al. 2022) (OL955261.1) and vB_EliS‐R6L (Lu et al. 2017) (KY006853.1) siphoviruses isolated from Erythrobacter litoralis DSM 8509 that belong to Eliscbkvirus and Timquatrovirus, respectively; vB_DshP‐R2C (Cai et al. 2015) (KJ803031), a podovirus belonging to Baltimorevirus isolated from Dinoroseobacter shibae DFL12; vB_DshS_R4C (Cai et al. 2019) (MK882925) and vB_DshS_R5C (Yang et al. 2017) (NC_041921) siphoviruses belonging to Cronusvirus and Nanhaivirus, respectively, isolated from Dinoroseobacter shibae DFL12; and other siphoviruses, such as vB_EauS‐R34L2 (PQ394075), vB_EauS‐R34L1 (PQ394074), XM05‐1, ZS04, B2‐phi0088, A1‐phi0434A, and A3‐phi2201A (uploaded to https://doi.org/10.6084/m9.figshare.28444040) isolated from Erythrobacter sp. JL475 and Citromicrobium bathyomarinum MCCC1K00088, MCCC1A10434, JL2201; as well as the myovirus D2‐phi0434B isolated from Citromicrobium bathyomarinum MCCC1A10434 (Figure S1).

The hosts of the lytic phages were cultured at 28°C in RO medium (1 g/L yeast extract, 1 g/L peptone, and 1 g/L sodium acetate at pH 7.5) with shaking at 160 rpm/min. Phage stocks in SM buffer (50 mM Tris–HCl pH 7.5, 100 mM NaCl, and 8 mM MgSO₄) were added to the respective host cultures when the OD₆₀₀ of the cultures reached approximately 0.2–0.3. Phage lysates were obtained after the cultures became almost transparent.

Temperate phages were induced from their hosts. Citromicrobium bathyomarinum MCCC1A10434, MCCC1A07757, JL2201, MCCC1K00088, MCCC1A09357 were grown at 28°C in RO medium with shaking at 160 rpm/min. After 24 h of cultivation, 1 mL of mitomycin C (5 mg/L, Sigma, Germany) was added to each bacterial culture, which was incubated for 30 min. The cultures were then centrifuged at 7000 × g for 8 min at 4°C to remove the supernatant, and the bacterial pellet was resuspended in an equal volume of sterilised artificial seawater. This step was repeated to remove residual mitomycin C, after which the bacterial pellet was resuspended in RO medium and incubated on a shaker for 24 h to induce prophages to begin the lytic cycle, resulting in the production of phage lysate.

Each lysate was centrifuged at 10,000 × g for 10 min and filtered through a 0.2‐μm pore‐size polycarbonate membrane filter (Merck Millipore, Germany). The filtrate was then precipitated with 10% (w/v) polyethylene glycol 8000 in 1 M NaCl at 4°C for 30 h. The precipitated phages were centrifuged at 10,000 × g for 50 min at 4°C, and the pellet was gently resuspended in 5 mL of SM buffer. The suspension was further purified using a CsCl density gradient (1.3 mg/mL, 1.5 mg/mL, 1.7 mg/mL) at 34,100 × g for 24 h at 4°C. Purified phage particles were collected, dialyzed twice against SM buffer, and stored overnight in the dark at 4°C.

Phage DNA was isolated using a phenol–chloroform–isoamyl alcohol method and subsequently sequenced on the Illumina MiSeq platform with 2 × 250 bp paired‐end reads. Genome assembly was performed using CLC Genome Workbench software with a coverage of 43×. The temperate phage sequence of Citromicrobium bathyomarinum was identified from the C. bathyomarinum bacterial genome using the online phage search tool PHASTER (http://phaster.ca/).

2.2. Viral Protein Identification by Mass Spectrometry

Viral samples were treated with twice their volume of SDT (4% w/v SDS, 0.1 M DTT, 100 mM Tris–HCl, pH 7.6) and heated at 100°C for 10 min to denature phage proteins. The denatured proteins were separated using standard SDS‐PAGE. The gels containing separated proteins were shredded and treated with 200 μL of deinking solution (30 mM acetonitrile, 100 mM NH₄HCO₃) until the gels turned colourless, which was followed by the removal of the supernatant.

An equal volume of NH₄HCO₃ (100 mM) was added, and the mixture was incubated at room temperature for 15 min. The protein samples were then aspirated, the supernatant removed, and the residues lyophilized. The samples were digested with 2.5–10 ng/μL trypsin (Promega, USA) for 20 h. Tryptic peptides were fragmented using ultrasound and subsequently lyophilized.

The dried tryptic peptides were loaded onto an EASY column (Thermo Scientific, USA) for capillary high‐performance liquid chromatography (Thermo Scientific, USA). A 70‐min linear gradient of acetonitrile in 0.1% formic acid was applied as follows to separate and elute the peptides: 0.1% for 40 min, 5%–28% over 2 min, and 28%–90% over 28 min. The peptides were detected using an Orbitrap mass spectrometer (Thermo Q Exactive Plus, USA) in data‐dependent acquisition mode with dynamic exclusion. The master scan event (m/z range: 375–1500 m/z) was followed by an MS2 scan with the following parameters: HCD activation type, automatic gain control target of 105, 3 s per scan cycle, 50,000 m/z, maximum IT of 105 s. The raw data were analysed using MaxQuant software (Tyanova, Temu, and Cox 2016) with the following parameters: peptide charge of +2 and +4; maximum missed cleavages, 2; dynamic modifications, cysteine carbamidomethylation and methionine oxidation; fragment ions maximum error, 0.1 Da; precursor ion maximum error, 20 ppm.

Peptide fragments were analysed for mass‐to‐charge ratio (m/z) shifts, with phosphorylation increasing mass by 79.97 Da and acetylation by 42.01 Da. In tandem MS (MS/MS) using collision‐induced dissociation (CID), b ions (N‐terminus) and y ions (C‐terminus) were generated. Modifications altered the m/z values of these ions, allowing precise localisation of modification sites. Phosphorylation was identified by a 79.97 Da increase. Acetylation was detected by a 42.01 Da mass increase.

2.3. Proteomic Analysis

The raw mass spectrometry data were obtained as RAW files and analysed using MaxQuant software for database searching and protein identification. A custom‐built database derived from predicted protein sequences from the genome sequences of the phage under investigation was employed for peptide identification. A protein was considered reliably identified if at least two unique peptides were detected. The search parameters were set as follows: single isotopic mass, non‐specific enzyme digestion with a maximum of two missed cleavage sites, and precursor ion charge states of 2+, 3+ and 4+. Carbamidomethylation (C) was set as a fixed modification, while oxidation (M) was included as a dynamic modification. The mass tolerance was set to 20 ppm for precursor ions and 0.1 Da for fragment ions.

2.4. Proteome Bioinformatics Analysis

Amino acid sequences were compared using the UniProt database (https://www.uniprot.org/) by BLASTp. Alignment of amino acid sequences was performed using the Align tool of UniProt. Visualisation of comparative proteomic analysis of structural proteins was conducted using the gggnomes R package. The structure‐based annotation was performed by first predicting and aligning protein sequences using HHpred (Söding et al. 2005), with known protein structures from the PDB_mmCIF70 database. Proteins were annotated if the alignment had an E‐value less than 10⁻³ and a probability greater than 90%. The HHpred alignment results were then cross‐validated with domain predictions from the Pfam Database (Mistry et al. 2020) and the Conserved Domain Database (Marchler‐Bauer et al. 2013). Tertiary structures were predicted by Alphafold2 (Schrödinger, LLC 2024). PyMOL software (Schrödinger, LLC 2024) was used to superimpose and map the tertiary structures of pdb files.

In this study, several phage assembly proteins (e.g., TerL and TerS) or replication‐transcription‐related proteins (e.g., hydrolase and DNA polymerase) were identified (Figure 2A), which are not typically involved in the structures of virions (Rossmann and Rao 2012). Ideally, we aimed to purify and enrich phage particles for protein detection via mass spectrometry, expecting only those proteins present on phage particles to be identified. However, assembly or replication‐transcription‐related proteins were also detected, indicating that during the purification and enrichment of phage particles, proteins required for viral replication may also be included. Therefore, we excluded them from our analysis.

FIGURE 2.

Proteomic characterisation and predicted architecture of phages. (A) Comprehensive depiction of proteomic outcomes for 13 bacteriophages, where the stacked bar graph illustrates the annotated composition and relative abundance of various proteins within each phage, with distinct colour blocks denoting different proteins. (B) Delineation of the predictive architecture of the 13 bacteriophages constructed based on the proteomic data and corroborated by extant literature.

2.5. Prediction of Antiphage Systems

Anti‐phage systems were identified from bacterial genomes using DefenseFinder (Tesson et al. 2022) with default settings.

2.6. Data Available

MS data, and the re‐annotated GenBank files for all phage genomes, have been uploaded to the public repository: https://doi.org/10.6084/m9.figshare.28444040.

3. Results and Discussion

3.1. Mass Spectrometry Improves Phage Protein Annotation

In this study, we employed mass spectrometry to characterise the proteomes of 13 phages, with details provided in Table 1. These 13 phages were isolated or induced from diverse hosts of alphaproteobacteria and represented various morphological variations, including 11 siphoviruses, one myovirus, and one podovirus. Among them, the phages vB_EliS‐L02 (Li et al. 2022) and vB_EliS‐R6L (Lu et al. 2017), as well as vB_DshP‐R2C (Cai et al. 2015), vB_DshS_R4C (Cai et al. 2019), and vB_DshS_R5C (Yang et al. 2017), have already been published with their whole‐genome sequences and physiological characteristics. The proteomic profiles of these 13 phages are delineated in Tables S1–S13. We identified 12–24 proteins within the siphoviruses, 18 proteins within myovirus D2‐phi0434B, and 17 proteins within the N4‐like podovirus vB_DshP_R2C. Prior investigations have demonstrated that siphoviruses, exemplified by λ, are endowed with 12 structural proteins (Rajagopala et al. 2011), whereas myovirus T4 encompasses a larger array, amounting to 37 structural proteins (Leiman et al. 2003). Similarly, the podovirus N4 has been shown to possess 10 structural proteins (Choi et al. 2008).

TABLE 1.

Basic information on the 13 bacteriophages analysed for proteomic identification.

Phage	Original host (accession)	Phage type	ICTV genus	Accession	Prophage site	References
vB_EliS‐L02	Erythrobacter litoralis DSM 8509	Temperate	Eliscbkvirus	OL955261.1	—	Li et al. (2022)
vB_EliS‐R6L	Erythrobacter litoralis DSM 8509	Temperate	Timquatrovirus	KY006853.1	—	Lu et al. (2017)
ZS04	Erythrobacter litoralis DSM 8509	Virulent	NA	PV416405	—	This study
XM05‐1	Erythrobacter sp. JL475	Virulent	NA	PV416404	—	This study
vB_EauS‐R34L2	Erythrobacter sp. JL475	Virulent	NA	PQ394075	—	This study
vB_EauS‐R34L1	Erythrobacter sp. JL475	Virulent	NA	PQ394074	—	This study
vB_DshP‐R2C	Dinoroseobacter shibae DFL12	Virulent	Baltimorevirus	KJ803031	—	Cai et al. (2015)
vB_DshS_R4C	Dinoroseobacter shibae DFL12	Virulent	Cronusvirus	MK882925	—	Cai et al. (2019)
vB_DshS_R5C	Dinoroseobacter shibae DFL12	Virulent	Nanhaivirus	NC_041921	—	Yang et al. (2017)
B2‐phi0088	Citromicrobium bathyomarinum MCCC1K00088	Temperate	NA	GCA_038594445.1	438483‐476408	This study
A1‐phi0434A	Citromicrobium bathyomarinum MCCC1A10434	Temperate	NA	GCA_038594405.1	1978083‐ 2016706	This study
D2‐phi0434B	Citromicrobium bathyomarinum MCCC1A10434	Temperate	NA	GCA_038594405.1	2053083‐2090964	This study
A3‐phi2201A	Citromicrobium bathyomarinum JL2201	Temperate	NA	CP155577.1	1747309‐1786176	This study

Abbreviation: NA, not available.

Conventional approaches used to annotate viral genomes typically entail using servers or software platforms such as RAST (Overbeek et al. 2014) and BLASTp. These methods, however, may yield annotations devoid of experimental validation. Consequently, the resulting annotations often feature numerous hypothetical proteins and occasionally include ambiguously described structural protein annotations (Lu et al. 2017; Cai et al. 2015, 2019; Guo et al. 2019; Turner et al. 2017; Montso et al. 2023). For example, vB_EliS‐L02 has previously been annotated with only the portal protein, major capsid protein (MCP), tail tube protein (TTP), and tape measure protein (TMP) (Li et al. 2022); and similarly, vB_DshS_R4C (Cai et al. 2019), vB_DshS_R5C (Yang et al. 2017) and vB_EliS‐R6L (Lu et al. 2017) are also annotated with only three of these proteins. Other structural proteins in the sequence‐based annotations of these phages are either hypothetical proteins or generally labelled as structural proteins. We re‐annotated the genomes of these 13 phages using the more advanced annotation tools (Table S15), Pharokka (Bouras et al. 2022) and PHOLD (https://github.com/gbouras13/phold). Compared to RAST and BLASTp (based on the NCBI database), these tools demonstrated significant improvements in annotating phage structural proteins, particularly for neck proteins. However, their annotation of phage tail proteins, such as tail tube protein, hub protein, distal protein, and tail fibre protein, remained insufficient. These sequence‐based annotated proteins are indicated by red arrow boxes in Figure 1 and Table S15.

FIGURE 1.

Comparative proteomic analysis of structural proteins. A detailed comparison of the structural protein sequences identified by mass spectrometry for 13 bacteriophages. The red arrow boxes indicate proteins annotated based on sequence data, while the black arrow boxes represent supplementary annotations derived from structure‐based analysis. The structural proteins for each phage are arranged based on their order in the genome. The sequences of structural proteins are depicted using coloured blocks of varying lengths to represent the sequence lengths. The shaded areas between protein sequences indicate the degree of similarity, with darker shades denoting higher levels of similarity.

By integrating mass spectrometry data, which provided detailed protein amino acid sequences, abundance, and composition profiles, we are able to validate and refine the structural predictions, resulting in a more comprehensive and accurate structure‐based annotation using tools such as the Pfam Database (Mistry et al. 2020), HHpred (Söding et al. 2005), Phyre2 (Kelley et al. 2015), and I‐TASSER (Yang et al. 2015). This structure‐based annotation enabled the more accurate identification of numerous structural proteins, such as the head decoration protein, capsid fibre protein, stopper protein, as well as the tail distal tail protein, hub protein, megatron protein, tail fibre protein, and central fibre protein. The structural proteins of each phage, annotated based on the proteomic results using this structure‐based approach, are indicated by black arrow boxes in Figure 1, with different proteins represented by various colours. These annotations provide a foundation for future studies on phage stability (e.g., analysing protein–protein interactions within phage particles or performing molecular dynamics simulations (Huang et al. 2023)) and interactions with their hosts by integrating characterisation information. Thus, integrating proteomic evidence with structure‐based annotation is proving to be a valuable tool in advancing our understanding of phages and their interactions with host organisms.

Mass spectrometry identification also provided the relative abundance of each protein in the phage proteome, as shown in Figure 2A. The compositions of these phages were relatively similar because they predominantly consisted of structural proteins, with some functional proteins, metabolism‐related proteins, and unknown proteins constituting a smaller proportion of the relative abundance. The relative abundance results show a consistent pattern, with MCP consistently exhibiting the highest prevalence across these phages, followed by portal proteins and tail tube proteins.

3.2. Predictive Architectural Models

On the basis of the structural protein annotations derived from the proteomic characterisation of each phage, we generated predictive architectural models for each phage (Figure 2B). This process involved integrating the positional information on structural proteins obtained from transmission electron microscopy images of these phages (Figure S1), with composite structures derived from cryo‐electron microscopy (Cryo‐EM) analyses of phages exhibiting comparable morphologies (Arnaud et al. 2017; Zinke et al. 2020; Büttner et al. 2016; Bhardwaj et al. 2014). For example, the architecture of vB_DshS_R4C corresponds well with its cryo‐EM–resolved complex structure (Huang et al. 2023). Notably, comparative analysis revealed no significant differences in protein composition, relative abundance, or overall architecture between virulent and temperate phages, suggesting a shared structural framework regardless of lifestyle. The subsequent sections will describe our thorough examination and elucidation of the distinct characteristics displayed by proteins identified across diverse regions of phages.

These models offer visual representations of phage structures, helping to elucidate the spatial arrangement of proteins and highlight the architectural complexity of phages, thereby facilitating a deeper understanding of their assembly mechanisms and interactions with host cells. Thus, the integration of proteomic data with structure‐based annotation not only confirmed the presence and functional relevance of key structural proteins but also revealed the spatial organisation and functional relevance of these proteins within the phage architecture.

3.3. Head and Neck Proteins

The head structure of phages belonging to Caudoviricetes is primarily constructed using MCP, which forms the shell of phages (Rossmann and Rao 2012). In addition to MCP, some viruses may feature decoration proteins that contribute to capsid structure stabilisation (Wang et al. 2018), as well as capsid fibre proteins that facilitate host attachment (Xu et al. 2019). In our proteomic results, phage MCPs and portal proteins were often found to exhibit significantly higher relative abundances (Figure 2A). This phenomenon can be attributed to the higher copy numbers of these proteins in virions. For instance, Cryo‐EM analysis of phage vB_DshS_R4C revealed the presence of 415 copies of the MCP, as well as 12 copies of the portal protein (Huang et al. 2023). Such high copy numbers are indicative of the critical role of these proteins in maintaining the structural integrity and functionality of the phage capsid. In this investigation, it was observed that all phages exhibited MCP and portal protein. Specifically, vB_EliS‐L02, XM05‐1, vB_EauS‐R34L1, A1‐phi0434A, and A3‐phi2201A were found to contain head decoration proteins, whereas ZS04, vB_EliS‐R6L, and vB_DshP‐R2C were found to harbour capsid fibre proteins. The presence of decoration proteins in these phages suggests that they have evolved mechanisms for enhanced capsid stability, which is likely an instrument for phage–host interactions (Wang et al. 2018; Xu et al. 2019).

During the process of annotating structural proteins using HHpred, it was observed that, despite variations in viral morphology and low sequence similarity (Figure 1), the structures of MCP all conformed to the HK97‐like folding pattern, conforming to the typical scenario observed in tailed phages. We predicted the structures of these proteins for 13 phages in this study and conducted structural alignments, revealing their high degree of structural similarity (Figure 3A). Comparative structural analyses indicated that the primary structural differences between the MCPs of phage vB_DshS_R4C and those of the other analysed phages are predominantly localised to the E‐loop, as indicated in the cyan part in Figure 3A. In‐depth studies of the structures of MCPs from vB_DshS_R4C (Huang et al. 2023) and other prototype phages such as HK97, TW1 and P47‐26 (Wang et al. 2018; Zhang et al. 2013; Helgstrand et al. 2003; Zhao et al. 2017) have revealed that the E‐loop primarily contributes to the stability of the outer layer of the head capsid. Variability in this structural domain may account for the differences in outer capsid stability among these phages (Figure 3A).

FIGURE 3.

Structural conservation analysis of phage proteins. (A and B) Showcase the superimposed structures of major capsid and portal proteins from 13 phages, with conserved domains in pink and variable domains in cyan, where shaded areas indicate the results of the superimposition. (C–E) Extend this methodology to distal, hub, and megatron proteins of 11 siphoviruses, maintaining consistent colour coding. Each panel employs cartoon representation models, with the structure of R4C proteins as a reference, to effectively illustrate the structural conservation and variability among these viral proteins.

In the portal proteins, regions of structural variability are primarily found in the wing and clip domains (Figure 3B). The wing domain forms the central part of the portal protein and is likely involved in the packing and release of DNA (Huang et al. 2023), suggesting that variability in this domain could influence the mechanics or magnitude of force exerted during DNA packaging and release. The clip domain, which was thought to be in close contact with the capsid, participates in anchoring the portal to the capsid (Dedeo et al. 2019). Variations in this domain could affect the interaction forces or stability between the portal proteins and MCPs across these different phages. Research has shown that conformational changes to the portal protein create an undulating belt that tightly embraces the helical DNA, facilitating its translocation through sequential positional adjustments (Lebedev et al. 2007). Thus, such differences in structural domains could impact the efficiency of DNA translocation, potentially affecting the overall infectivity of the phage. The MCP and portal proteins of phages are structurally conserved, irrespective of phage morphology or life cycle type, and no clear correlation was observed between variable structural domains and these characteristics.

Three phages—vB_DshP‐R2C, ZS04, and vB_EliS‐R6L—possess capsid fibre proteins structurally similar to those of Bacillus subtilis phage phi29 (Tables S1, S5, and S7). Cryo‐EM studies of phi29 have shown that the base of the fibre provides a stable connection point that is essential for phage capsid integrity and function during the infection process. The distal end of the capsid fibre is posited to facilitate attachment or interactions with other phage particles or, potentially, the host cell during infection (Morais et al. 2005). The structural similarity of the capsid fibre proteins to those in phi29 indicates a similar mechanism of phage–host interaction is used. Here, we propose that the presence of capsid fibre proteins in these three phages enhances the stability of the capsid and potentially facilitates viral recognition and attachment to host cells (Wang et al. 2018; Morais et al. 2005).

The junction between the head and tail in siphoviruses and myoviruses typically includes portal, stopper, and adaptor proteins. Whereas podoviruses exhibit a simpler composition in their neck region, consisting only of portal and adaptor proteins (Rossmann and Rao 2012). vB_EauS‐R34L2, XM05‐1, vB_EliS‐L02, vB_EliS‐R6L, vB_DshS_R5C, B2‐phi0088, and A3‐phi2201A were observed to possess adaptor and stopper proteins (Figure 2B, Tables S1–S13). The detection of these proteins in multiple phages underscores their critical role in phage assembly and infection, ensuring efficient DNA transfer by facilitating correct tail‐to‐head attachment (Iwasaki et al. 2018; Fokine et al. 2013).

3.4. Tail Proteins

The tail of phages is primarily composed of multiple copies of tail tube proteins. Particularly, siphoviruses and myoviruses also harbour various baseplate proteins that stabilise the tail structure, in which they connect to tail fibre proteins and may be involved in viral tail contraction (Rossmann and Rao 2012). In this study, tail tube proteins were detected in siphoviruses and myoviruses with relatively high abundance, indicative of a high copy number of this protein within the virus. Additionally, in all siphoviruses and myoviruses, except for vB_DshS_R4C, terminator proteins were identified, which serve as termination proteins for tail tube assembly. However, both the genome and Cryo‐EM structure of vB_DshS_R4C demonstrated the presence of terminator proteins. The absence of these proteins in the proteome may be attributed to a higher degree of damage to the tails of vB_DshS_R4C during the experiment of viral enrichment, resulting in a relatively lower abundance of these tail proteins in the samples. This was supported by the lower relative abundance of tail tube proteins in the proteome of vB_DshS_R4C compared to those of other phages, though vB_DshS_R4C was demonstrated to possess 120 copies of the tail tube protein (Huang et al. 2023).

In the myovirus D2‐phi0434B, multiple baseplate proteins were detected and annotated (Figure 1 and Table S13). The structures of the tail tube protein and each baseplate protein in this phage resemble the structure of phage T4 (Yap et al. 2016) and R‐type pyocins, a type of bacteriocin considered akin to myoviruses that act as nanomachines (Ge et al. 2020). Bacteria normally secrete such protein complexes to identify and eliminate competitors in order to better monopolise resources (Ge et al. 2020). However, the evolutionary relationship between R‐type pyocins and myoviruses remains elusive.

Baseplate proteins, consisting of distal, hub, and megatron proteins, were detected in all siphoviruses in this study (Figure 1). The baseplate complex usually serves as the initial point of contact with host cells, making it particularly crucial to viral infection (Mateu 2013). Despite the low sequence similarity among these proteins in siphoviruses that target different hosts, especially citrophages that infect Citromicrobium bathyomarinum, in which the proteins share no sequence similarity with those of the other groups (Figure S2), HHpred search results (Tables S2‑S12) indicate that they have structural similarity to those of Rhodobacter capsulatus GTA (RcGTA) (Bárdy et al. 2020) and vB_DshS_R4C (Huang et al. 2023). Therefore, we predicted the structures of these proteins and compared them with the structure of the vB_DshS_R4C proteins from existing Cryo‐EM structural results, which revealed the highly conserved structural domains in these proteins (Figure 3C–E). Notably, megatron proteins are typically found only in α‐proteobacteria phages (Shakya et al. 2017). In α‐proteobacteria phages and gene transfer agents (GTAs), these three baseplate proteins also exhibit higher rates of horizontal gene transfer (Lang et al. 2012). The megatron protein's fibre‐binding domain and ris/penetration domain show similarity across these siphoviruses (citrosiphoviruses and the roseosiphovirus vB_DshS_R4C), suggesting that these long‐tailed phages displayed overall similar mechanisms of baseplate attachment to tail fibres and baseplate transmembrane processes (Huang et al. 2023). The conserved structural domains in the distal tail proteins, hub proteins, and megatron proteins of both RcGTA (Bárdy et al. 2020) and vB_DshS_R4C (Huang et al. 2023) imply that an overall similar mechanism of host attachment is employed. These proteins function in concert to recognise and bind to host cell membranes during attachment and prepare the phage for DNA injection. This conserved mechanism ensures that both phages can efficiently attach to and infect their desired hosts. Importantly, these findings appear to be independent of the lifestyle of the phages, as no significant differences were observed between virulent and temperate phages in terms of the structural conservation of baseplate proteins and their functional roles. Consequently, the 11 siphoviruses examined in this study may employ analogous strategies for attachment, adsorption, penetration, and DNA translocation that are akin to those of RcGTA and vB_DshS_R4C. Such functional consistency implies that these phages and phage‐like particles may have evolved similar strategies to optimise their infection processes. This evolutionary convergence highlights the importance of structural conservation in phage evolution and host interactions, enabling these phages to efficiently exploit their hosts, despite having genetic diversity. Understanding these conserved mechanisms holds promise for predicting phage behaviour and their interactions with bacterial hosts, offering insights into methods for combating bacterial infections and harnessing phages for biotechnological applications.

Tail fibre proteins were also identified and annotated in these phages. These proteins are crucial for phages to recognise and bind to host receptor proteins (Bebeacua et al. 2013; Swanson and Cingolani 2018) and function as a determinant in the host range of phages (de Jonge et al. 2019; Tétart et al. 1996; Yehl et al. 2019). Despite the high diversity of tail fibre proteins in this study, those infecting the same host exhibited similar structures (Figure 4). The tail fibres of citrosiphoviruses B2‐phi0088, A1‐phi0434A, and A3‐phi2201A (Figure 4A) exhibited significant structural similarities. Likewise, the tail fibres of two phages targeting Dinoroseobacter shibae DFL12, vB_DshS‐R4C and vB_DshS‐R5C (Figure 4B), were more similar in structure. While R2C infects the same host as phages R4C and R5C, its tail fibre was not predicted from the proteome data. This may be due to the fact that the tail fibre of this phage is relatively novel in sequence and structure, leading to an inability to annotate the protein. The tail fibre structures of two phages infecting Erythrobacter litoralis DSM 8509, vB_EliS‐R6L and ZS04, also showed some resemblance (Figure 4C). Additionally, the tail fibre structures of vB_EauS‐R34L2 and vB_EauS‐R34L1 (Figure 4D), invaders of Erythrobacter sp. JL475, were similar. However, the tail fibre structures of XM05‐1 and vB_EliS‐L02 (Figure 4E), which infect Erythrobacter sp. JL475 and Erythrobacter litoralis DSM 8509, respectively, exhibited only partial structural similarity. Numerous studies have substantiated that mutations in the tail fibre proteins of phages can significantly alter their host range, highlighting the essential function these structures serve in determining host specificity (Tétart et al. 1996; Yehl et al. 2019; Golomidova et al. 2016). However, despite the variability introduced by mutations, certain structural similarities in tail fibres can still be observed among phages with analogous host ranges. Building upon this foundation, the current study introduced a perspective by suggesting that both mutations and structural similarities in tail fibres serve as indicators of analogous host ranges. This observation inspires us to propose that tail fibre proteins with comparable structures might possess the capability to recognise and bind to similar host species, a potential evolutionary adaptation aimed at optimising host infection efficiency. By analysing the structural characteristics of tail fibres, researchers may be able to predict the potential host spectrum of a phage more effectively. Such insights are pivotal because they advance our understanding of phage–host interactions and enhance the targeted application of phages in therapeutic contexts. The implications of this research are significant because they reveal a new dimension to the predictive modelling of phage behaviour and their practical applications in biocontrol and medicine.

FIGURE 4.

Comparative structural analysis of tail fibre proteins. (A) Superimposed structures of tail fibre proteins from B2‐phi0088, A1‐phi0434A, and A3‐phi2201A, all of which infect Citromicrobium bathyomarinum. (B) Tail fibre structures of vB_DshS‐R4C and vB_DshS‐R5, which infect Dinoroseobacter shibae DFL12. (C) Comparison of tail fibre structures of vB_EliS‐R6L and ZS04, which both infect Erythrobacter litoralis DSM 8509. (D) Comparison of tail fibre structures of vB_EauS‐R34L2 and vB_EauS‐R34L1, which have the common host Erythrobacter sp. JL475. (E) Comparison of the tail fibre structures of XM05‐1 and vB_EliS‐L02, which infect Erythrobacter sp. JL475 and Erythrobacter litoralis DSM 8509, respectively.

3.5. Post‐Translational Modifications of Phage Structural Proteins

In this study, mass spectrometry spectra revealed a diverse array of PTMs on phage structural proteins, notably including phosphorylation and acetylation. PTMs encompassing phosphorylation, acetylation, and ubiquitination, among others, are crucial for regulating protein activity, stability, and cellular localization (Narita et al. 2019; Huiting et al. 2023). While prior research has predominantly focused on elucidating how viral PTMs, particularly acetylation, suppress host cell antiviral immune responses by altering host cellular proteins (Liu et al. 2021; Dong et al. 2019; Alawneh et al. 2016) no studies have reported the presence of PTMs in bacteriophages. These PTMs were observed across several phages in our study, with notable occurrences on various proteins such as MCP of vB_DshS_R4C and B2‐phi0088; the portal protein, MCP, and tail tube proteins of A1‐phi0434A; the decoration protein and MCP of A3‐phi2201A; and the terminator protein of D2‐phi0434B (Figure 5A). As an example, the acetylation and phosphorylation modification sites of MCP from vB_DshS_R4C are illustrated in Figure S3. The findings of this study have significantly broadened our understanding of the diversity and functional significance of phage PTMs.

FIGURE 5.

Post‐translational modifications of phage structural proteins and host antiphage systems. (A) Structural proteins identified by mass spectrometry as being post‐translationally modified, with different arrow blocks representing various structural proteins. The modifications are indicated by symbols, with small triangles (green) for phosphorylation and circles (yellow) for acetylation. (B) Specific sites of phosphorylation (green nodes) and acetylation (yellow nodes) modifications on the structural proteins; these modifications are showcased within the context of each protein's structure. (C) Situation of antiphage systems in the hosts corresponding to phage PTMs that are capable of recognising phage structural proteins. Grey blocks indicate the absence of these systems, while orange blocks denote their presence.

The identification of PTM sites, as depicted in Figure 5B and Table S14, represents a crucial advancement in our comprehension of protein functionality and interaction dynamics. From the illustration, it is apparent that, in the MCP of vB_DshS_R4C, B2‐phi0088, and A1‐phi0434A, the modification sites appear to be dispersed rather than confined to specific structural domains (Figure 5B). These observations raise an intriguing possibility: these modifications may not serve to stabilise the capsid structure. If that was the case, the PTMs would likely bind to specific structural domains to complement the interactions among those domains. The HK97‐like capsid is stabilised through interactions between its structural domains; specifically, the A‐domain forms central wedges in hexons and pentons, the P‐domain provides the outer framework, the E‐loop connects adjacent subunits, and the N‐arm and I‐domain facilitate intra‐capsomer interactions (Duda and Teschke 2019). However, no discernible pattern of modification sites was observed in key structural elements, including the portal, tail tube, decoration, or terminator (Figure 5B). Thus, these modifications appear to play more complex roles in other biological processes. For instance, they could influence the interaction of MCP with other proteins or molecules, thereby modulating the infection, replication, or assembly processes of the bacteriophages. Additionally, the distribution of these modification sites may be associated with the adaptability and survival strategies of the phages in different environments. This discovery serves as a noteworthy starting point for exploring the biological characteristics and adaptability of these phages.

Phages are targeted by complex bacterial antiphage defence systems that include mechanisms such as PifA‐mediated defence (Schmitt et al. 1991), CBASS or Pycsar (Tal et al. 2021), BREX (Goldfarb et al. 2015), DSR2 (Garb et al. 2022), CapRel (Zhang et al. 2022), Septu (Doron et al. 2018), SEFIR, and Dodola (Millman et al. 2022). These systems are adept at recognising specific viral structural proteins, including MCP, tail tube, and portal proteins. Upon detection, they initiate an immune response that impedes viral gene transcription by activating various effector mechanisms. These mechanisms can involve the direct cleavage of phage DNA, the modification of host cell processes, or other responses designed to effectively halt phage replication (Georjon and Bernheim 2023). Although these bacterial hosts possess sophisticated immune systems capable of recognising phage structural proteins (Figure 5C), the phages studied are still able to infect and lyse their hosts, indicating the presence of unknown evasion strategies. Similar to these potential mechanisms, previous research has demonstrated that viruses can evade host immunity through PTMs of their proteins, as seen in other viral systems (Li et al. 2021).

Viruses utilise PTMs of their key envelope proteins to conceal antigenic epitopes, thereby evading recognition by neutralising antibodies and successfully escaping the host immune system's detection and attack. For instance, HIV‐1 glycosylates its gp120 protein, shielding antigenic sites and preventing most individuals from producing broadly neutralising antibodies (Moore et al. 2012). In the case of influenza viruses, such as H1N1, glycosylation of the hemagglutinin protein, particularly in the globular head region, helps the virus evade immune responses, especially during antigenic drift (Job et al. 2013). Ebola virus also employs glycosylation of its GP2 protein, reducing its antigenicity and immunogenicity, which aids in immune escape (Ou et al. 2012). Likewise, SARS‐CoV‐2 heavily glycosylates their spike proteins, masking key antigenic regions to avoid neutralisation and facilitate infection (Zhou et al. 2020). Additionally, the functional role of phosphorylation of viral proteins remains unclear, but studies have shown that Dengue virus leverages phosphorylation of host cell proteins to regulate key processes such as RNA splicing, protein biosynthesis, and immune responses (Miao et al. 2019). By manipulating host protein phosphorylation, the virus enhances its replication, immune evasion, and overall infectivity. Similarly, in SARS‐CoV‐2, recent research has identified several novel PTMs on viral proteins, including phosphorylation, ubiquitination, and S‐nitrosylation. These modifications are believed to play critical roles in modulating the virus's replication, assembly, and evasion of host immune responses. For example, phosphorylation of viral proteins by host kinases was shown to be essential for their interaction with host proteins, influencing viral infectivity. Additionally, ubiquitination and S‐nitrosylation were found to regulate the stability and activity of viral proteins, further promoting viral survival and propagation within the host (Adams et al. 2022).

Thus, we considered that PTMs on phage structural proteins may serve roles within mechanisms used to evade host antiviral defences. Specifically, we propose that PTMs alter the structural properties or antigenicity of the phage proteins, thereby reducing the efficacy of the host's recognition pathways. As illustrated in Figure 6, the mechanism by which phages evade antiphage systems via PTMs involves structural proteins. In the absence of PTMs, phage structural proteins may be recognised by the host's antiphage systems that target these proteins, leading to the termination of translation and transcription processes and preventing the synthesis of progeny phages. Alternatively, when structural proteins undergo PTMs, they may evade detection by the host's antiphage systems, facilitating the successful synthesis of progeny phages. This hypothesis warrants further investigation to determine the exact impact of PTMs on the interaction between phages and host defences.

FIGURE 6.

Mechanism by which phages evade bacterial antiphage systems via post‐translational modifications of structural proteins. In the absence of PTMs, phage structural proteins are recognised by the host's antiphage systems that target these proteins, leading to the termination of translation and transcription processes and preventing the synthesis of progeny phages. Conversely, when structural proteins undergo PTMs, they evade detection by the host's antiphage systems, facilitating the successful synthesis of progeny phages.

4. Conclusions

This study used mass spectrometry to thoroughly characterise the proteomes of 13 phages, which exhibit diverse morphologies, employ various life strategies, and were isolated from bacteria belonging to different families. The analysis identified multiple proteins for each phage and employed structure‐based annotation methods to accurately determine the functions of several structural proteins, including head decoration and tail proteins. Predictive structural models were generated for each phage, showcasing the spatial arrangement of proteins and providing deeper insights into phage assembly mechanisms and interactions with host cells. These findings advance our understanding of phage biology and the intricate architecture of phages. Additionally, the study provided information on the relative abundance patterns of structural proteins across different phages, with MCPs and portal proteins showing higher relative abundance, underscoring their crucial roles in phage structural stability and functionality. Mass spectrometry was also used to identify various PTMs of phage structural proteins, such as phosphorylation and acetylation, which were identified for the first time in marine phages. Furthermore, the research suggests that these PTMs may aid phages in evading host antiphage defence systems. However, further experimental validation and functional studies are required to confirm this hypothesis, such as mutating specific PTM sites and conducting functional assays to assess the role of these modifications in phage–host interactions. Additionally, a more comprehensive approach using viral metaproteomics could help investigate the broader prevalence of PTMs across different viruses. Such studies, along with mechanistic validation experiments in diverse viral systems, are necessary to better understand how PTMs influence viral biology and immune evasion strategies.

Author Contributions

Shuzhen Wei: conceptualization, investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, data curation. Anan Wang: conceptualization, methodology, investigation, validation, formal analysis. Lanlan Cai: writing – review and editing, conceptualization. Ruijie Ma: methodology, writing – review and editing. Longfei Lu: writing – review and editing. Jiangtao Li: funding acquisition, project administration, supervision, writing – review and editing, resources. Rui Zhang: conceptualization, funding acquisition, project administration, resources, supervision, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1. Supporting Information.

Data Availability Statement

The data that supports the findings of this study are available in the Supporting Information of this article.