Prions were demonstrated to play important roles in eukaryotes and prokaryotes, and they have been intensively studied during the last decade (Alberti et al., 2009). Prion proteins (PrPs) are characterized by self-propagation, undergoing a conformational switch from one conformational state to another, which leads to the creation of new prions (Bolton et al., 1982; Telling et al., 1995). Pathologically, prions are characterized by a process in which the infectious form of prion (PrPSc) interacts with the endogenous PrPs, catalyzing the transformation of the endogenous molecule into misfolded PrPSc aggregates (Ma and Lindquist, 2002). This was first observed in mammals, and the pathological deposition of insoluble protein aggregates was shown to be associated with the development of a number of human diseases, including scrapie, Creutzfeldt-Jakob disease, Alzheimer’s disease, Parkinson’s disease, amyloidosis, and other (Collinge, 2001; Wemheuer et al., 2017).

One of the best-known forms of PrPs are amyloid proteins that, in addition to the role of infectious agents, were shown to be an important component of many physiological processes in eukaryotes and prokaryotes (Westergard et al., 2007; Sabate et al., 2015; Walker et al., 2016).

Although the amino acid sequences and function of these evolutionarily unrelated proteins are diverse, all amyloid fibrils can be misfolded into distinct conformers, forming highly ordered cross-β structures (Cheng et al., 2012).

Recently, these highly ordered parallel or antiparallel amyloid β-sheets were identified in yeasts and bacteria, suggesting that they may play an important physiological role (de Groot et al., 2012; Angarica et al., 2013; Malinovska et al., 2013). Unique mechanical and biological properties of amyloid sheets have multiple functions in a variety of bacterial species, and the secretion of amyloid proteins represents an important step during biofilm formation (Chapman et al., 2002; Fowler et al., 2007; Blanco et al., 2012). To discriminate between the physiological and pathological roles of amyloid, these forms that are involved in physiological process have been termed the functional amyloid (Chiti and Dobson, 2006). Therefore, the propagation of prions was shown to enable environmental adaptation, the activation of stress response, and in yeasts these molecules play important roles in the preservation of long-term memory (Shorter and Lindquist, 2005; Suzuki et al., 2012). The aggregation of PrPs is an amino-acid sequence-specific process, but the mechanism underlying the de novo appearance of prions remains unclear. PrPs are rich in asparagine (Q) and glutamine (N) domains, which was exploited for the development of an algorithm aimed at the identification of candidate prionogenic domains (PrDs) (Michelitsch and Weissman, 2000; Alberti et al., 2009). Currently, the most widespread PrD prediction algorithm is based on the hidden Markov model (HMM), the probabilistic sequence model based on the maximum likelihood estimation (Eddy, 1998; Angarica et al., 2013).

The HMM can be found at the core of several bioinformatic approaches to the statistical representation of prion domains, which allow the scoring of protein sequences according to the likelihood that these proteins are prions (Toombs et al., 2010; Lancaster et al., 2014). To that end, the log-likelihood ratio (LLR) score is used to evaluate the sets of amino-acid interactions, with a large LLR indicating a high similarity between the examined interaction sets, whereas the LLR is nearing zero when comparing the sets of random interactions. Using these web-based algorithms, prion-forming domains were extensively studied in mammalian, bacterial, and fungal proteomes, demonstrating that prions are common in different organisms, and represent important global regulators (Iglesias et al., 2015; Yuan and Hochschild, 2017).

In prokaryotes, PrPs were shown to play important roles in molecular transport, secretion, cell wall development, and other processes (Blanco et al., 2012; Yuan et al., 2014). Moreover, prion determinants were shown to participate in the inter-kingdom communication between prokaryotes and eukaryotes, leading to the alterations in the amyloidogenesis in Caenorhabditis elegans following its colonization with amyloid-producing bacteria (Chen et al., 2016).

However, the prion-like domains in viruses, in particular the bacterial viruses, bacteriophages, have not been thoroughly examined. Bacteriophages are ubiquitous organisms that are found almost everywhere, from the oceans, with the estimated 10³¹ viral particles, to mammalians, with over 10¹⁵ bacteriophages in the adult human gut (Dupuis et al., 2013; Dalmasso et al., 2014). The life cycle of bacteriophages can be lytic, resulting in bacterial death and de novo assembly of virions, or lysogenic, in which bacteriophage DNA is integrated into the bacterial genome. Both bacteriophage life cycles are important in the regulation of bacterial population. Phage-mediated horizontal gene transfer, occurring during the lysogenic cycle, is especially important for the bacterial ecological adaptation and is a common mechanism underlying the spreading of antibiotic resistance (Ochman et al., 2000). Moreover, we previously demonstrated that the bacteriophages, by altering mammalian gut flora, can induce an increase in gut permeability and facilitate endotoxemia, indirectly affecting the vital activities of microorganisms (Tetz and Tetz, 2016; Tetz et al., 2017). Previous studies revealed a correlation between the increased gut permeability and the development of a variety of human pathologies, including diabetes, neurodegenerative, and cardiovascular diseases. Therefore, the role of bacteriophages as the potential indirect causative agent inducing the development of these diseases requires further investigation (Bischoff et al., 2014).

Here, we studied in detail the putative prion domains in bacteriophages. We employed HHM algorithm for the analysis of bacteriophage proteomes, using the UniProtKB database. To the best of our knowledge, this is the most extensive effort aimed at the identification of candidate PrD sequences among bacteriophages. Furthermore, we analyzed the trends in the distribution of PrDs in different protein families and the correlations between bacteriophage families and the host bacterial host. The present predictive approach for the first time uncovers a large set of putative prionogenic proteins whose further experimental characterization might contribute significantly to understanding bacteriophages biology.

To the best of our knowledge, this is the first analysis of PrDs among bacteriophages. The results of our study underline the necessity of characterizing bacteriophages and we propose here that the all bacteriophage genes should be known as the phagobiome.

Here, PrDs were more frequently found among archaeal phages, and the analysis of the three best-known Caudovirales families showed that these domains are more abundant among Podoviridae than Myoviridae or Siphoviridae.

Furthermore, our analyses demonstrated that approximately 50% of all sequenced bacteriophages available in public databases contain at least one PrD. The majority of bacteriophages contain less than five PrDs per proteome. Surprisingly, the vast majority of bacteriophages with more than five PrDs per proteome were shown to belong to the Myoviridae family. Since the total numbers of PrDs found in Myoviridae (2425) and Siphoviridae (1665) are of the same order of magnitude, these further supports the enrichment of PrDs in the Myoviridae proteomes. However, Myoviridae members have larger proteomes than the members of Siphoviridae family, which may explain the observed differences in the PrD enrichment (Sandaa, 2009). Further analyses indicated a direct correlation between the proteome size and the PrD enrichment, and therefore, we further explored whether the proteome size of bacteriophages in the Myoviridae family is related to the PrD enrichment. However, we have not observed a linear correlation between these parameters, suggesting that the PrD enrichment does not depend only on the size of a Myoviridae proteome.

Moreover, the majority of the top 100 scoring PrDs was also detected in the members of Myoviridae family, and among these, several short protein sequences were found to harbor a higher number of Q-rich domains than the longer ones. This indicates that the scores obtained in the analysis of protein sequences according to their prion-like characteristics do not correlate only with the size of a protein. However, the functional relevance of these findings remains unclear, since the majority of these proteins have not been characterized. The analysis of proteins with the known functions showed that the proteins involved in the replication of bacteriophage DNA and protein synthesis have the largest LLR scores.

Many PrD-containing proteins with known functions were identified among Bacillus phages that infect the spore-forming Bacillus spp. The identified PrDs were found in proteins responsible for attachment and penetration, replication of bacteriophage DNA, and release within the same Bacillus phages. Furthermore, the LLR scores of PrDs in the Bacillus phages were shown to have the highest meaning of the prion likelihood ratio compared with the phages associated with other bacterial genera. Previously, we reported a particular role of spore-forming bacteria in the global microbiota, which allowed their identification as the members of sporobiota (Tetz and Tetz, 2017). Taken together, previous results and the ones obtained in this study may suggest that the global distribution of sporobiota members and Bacillus spp., and their increased transfer between the ecological niches is associated with the presence of resistant transmissible spores due to their interactions with bacteriophages (von Wintersdorff et al., 2016).

However, the highest numbers of PrDs per proteome were found across Prochlorococcus and Synechococcus phages, whose hosts are represented by evolutionary ancient Cyanobacteria (Figure 4) (Schirrmeister et al., 2011). This finding along with the higher prevalence of PrDs in archaea suggests the possible association of the presence of PrDs and “evolutionary age,” however, this requires an independent analysis. Regardless, the enrichment of PrDs in the most abundant phages on Earth infecting marine Prochlorococcus and Synechococcus suggests the possible particular role of these elements (Sullivan et al., 2010).

To determine the potential correlations between the functions of PrD-containing proteins, bacteriophage families, and host organisms, we generated a heatmap that showed that the highest LLRs were found among Myoviridae and Siphoviridae members of Moraxellaceae, Listeriaceae, and Pseudoalteromonadaceae, which were shown to be associated with replication of phage DNA, protein synthesis, and bacteriophage assembly. However, these data do not correspond to the frequency of PrD in bacteriophage proteomes. A vast majority of the bacteriophage PrD-containing proteins are involved in the interactions between a bacteriophage and bacterial cell wall, their attachment and penetration, and the release of progeny phages.

Attachment and penetration represent crucial steps in the bacteriophage–bacterium interaction, since these processes allow the entry of the bacteriophage DNA inside the host. We identified PrDs in the proteins forming different structural components of bacteriophages, including head, neck, sheath, baseplate and fibers, for tailed viruses, receptor-binding proteins for tailless Tectiviridae, and in the coat proteins III and IV of the filamentous Inoviridae (Hulo et al., 2017). This indicates the role of the identified PrDs in the formation of surface structures and specific and non-specific attachments to bacterial cells (Spinelli et al., 2014). Moreover, a similar trend in the distribution of prion-like domains in proteins responsible for interactions and virulence was defined by Gene Ontology GO terms across bacteria (Iglesias et al., 2015). Notably, PrDs were common across different bacterial and bacteriophage families, indicating the universal regulatory role of these putative prion domains (Iglesias et al., 2015).

Furthermore, since we identified PrDs in the interacting proteins in different bacteriophage families infecting a variety of bacterial hosts, this implies that these PrDs are conserved across species. However, since the PrDs in the surface proteins were not identified in all analyzed bacteriophages, and in those where they were identified, PrDs were found in some, but not all surface proteins of the given phage, this suggests that the presence of PrD-containing proteins may be beneficial, but it is not obligatory. The attachment of bacteriophages is a complex process, accompanied by a number of bacteriophage-resistance responses initiated in a bacterial cell (Labrie et al., 2010). Our findings may indicate that the proteins with the PrDs can alter their conformational states, adding further complexity and specificity in the phage-bacterial interactions, and this may serve as a mechanism to overcome some bacterial resistance mechanisms. Moreover, it can be speculated that the PrDs in the interacting proteins may be involved not only in the interactions with host bacteria, but also in the interactions of bacteriophages with other components of microbial communities. These observations agree with our recent studies showing that the oral administration of bacteriophages induces an increase in the intestinal permeability and endotoxemia in mammals, significantly changing microbiota composition, beyond the possible direct effect of bacteriophages, and because of this, it was suggested that these effects are exerted through yet unknown mechanisms (Tetz and Tetz, 2016; Tetz et al., 2017).

Additionally, the results obtained in this study suggest that the presence of PrDs in bacteriophages may be associated with a number of mammalian diseases, in which the role of prions was demonstrated both in macroorganisms and in gut bacteria. Thus in the article by Chen et al. (2016), C. elegans fed with the prion-producing Escherichia coli was shown to have an enhanced prion aggregation in brain. It can be assumed that a similar process may be observed following the introduction of bacteriophages and that bacteriophages contribute to the observed processes.

Proteins that are involved in the bacterial cell wall interactions and the release of progeny phages from the host were shown to harbor PrDs as well. The release stage involves bacterial cell wall degradation from within the cells with bacteriophage-encoded peptidoglycan hydrolases, synthesized at the end of the multiplication cycle. We identified PrDs in phage-encoded endolysins, and the majority of these proteins was found in Siphoviridae family, with only a few detected among the members of Podoviridae and Myoviridae.

Some of the proteins enriched in the PrDs are involved in the replication of bacteriophage DNA and protein synthesis. In eukaryotes, PrD-containing proteins were shown to play a role in the interactions and binding with the nucleic acids (King et al., 2012). Nucleic acids were shown to play a crucial role in the PrP conversion into PrPSc-like species as cofactors (Macedo and Cordeiro, 2017).

In bacteriophages, we determined that the presence of PrDs can also be associated with the single-strand binding proteins, known as essential components of bacteriophage DNA synthesis. Therefore, this indicates that these PrDs can be easily misfolded into PrPSc through the interactions with the DNA, which was shown to be an important component inducing this misfolding (Cordeiro et al., 2001). This type of interaction with the host cell may represent an additional process through which the bacteriophages control bacterial intracellular functions (Wagner and Waldor, 2002; Davies and Davies, 2010).

Bacteriophages are important vehicles facilitating the genetic exchange between microorganisms, including the spreading of virulence factors and the antibiotic resistant genes (Lina et al., 1999). Here, we identified a PrD in the LukS-PV protein of some Siphoviridae members and unclassified phages. This cytotoxin can be found among methicillin-resistant S. aureus and belongs to the group of β-pore-forming toxins (Vandenesch et al., 2003). The presence of LukS-PV is associated with the increased virulence of S. aureus, leading to abscess formation and severe necrotizing pneumonia. The detected PrDs located in LukS-PV indicates the role of these domains in the virulence potential of the cytotoxins and their role in the pathogenicity. Our results may indicate that the presence of PrDs in LukS-PV affects its interactions with the membranes of the targeted eukaryotic cells and induce the pore-forming potential, which is supported by previous studies (Iglesias et al., 2015).

Taken together, we identified numerous putative PrD-containing proteins in bacteriophages. We observed consistent patterns of the distribution of PrDs across different bacteriophage families. However, since the infectious agents of some bacteria were shown to lack PrDs, this may indicate that the diversity of phagobiota and phogobiome is underestimated. Although bacteriophage genomes are significantly smaller than bacterial genomes, currently less than 2,500 phage genomes are deposited in the NCBI database, compared with almost 90,000 bacterial whole genome sequences.

Although several trends in the frequency of PrDs across bacteriophage families have been detected, future studies are required (Pourcel et al., 2017), to elucidate further the functions and presence of PrP-containing proteins in bacteriophages. The predictive approach employed here revealed for the first time a large set of putative PrPs, and further experimental characterization of these proteins may contribute to the understanding of bacteriophage biology.

GT and VT designed and conducted the experiments and analyzed data, GT wrote the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.