Introduction

Microbial selectives

Most immune components have been discovered within the context of pathogenesis (Tanji and Ip, 2005; Kawai and Akira, 2010). This emphasis has led to the implicit assumption that immunology is the study of host versus pathogen (Casadevall and Pirofski, 2014). This two-dimensional bias is exemplified through the pervasive use of antimicrobial when describing host–microbe interactions (Brogden, 2005; Casadevall and Pirofski, 2014). In nature, host–pathogen interactions occur within the context of an ecological community, that is, a host in symbiosis with its microbial partners, which is called the holobiont (Casadevall and Pirofski, 2014; Bordenstein and Theis, 2015). These symbioses run the gamut of mutualistic to parasitic/pathogenic. Niche exclusion is an essential dynamic for maintaining the holobiont; any microbe, compound or entity that removes a microbe from a particular ecosystem creates novel niche space for another microbe to occupy (Rodriguez-Brito et al., 2010). In this perspective, we will utilize the term microbial selective to describe mechanisms that maintain specific microbes associated with a metazoan host.

The ever-changing, ubiquitous surface mucus layer (SML)

Mucosal environments coat the surfaces of specific epithelial cell types across the spectrum of metazoan life (Bäckhed et al., 2005; Brown and Bythell, 2005). These environments are constructed by mucin macromolecules, which consist of a peptide backbone covalently bonded to variable oligosaccaride side chains (Ferez-Vilar and Hill, 1999; Hang and Bertozzi, 2005; Corfield, 2013). The process of glycosylation is controlled by secondary structural motifs (Julenius et al., 2005), the cellular repertoire of glycosyltransferases and their localization within the Golgi apparatus resulting in distinct cellular profiles (Hanisch, 2001). O-linked glycosylation has been shown to be integral in immune protection across the animal phyla (Tsuboi and Fukuda, 2001; Bond et al., 2014). Following posttranslational modification, mucins are either tethered to the epithelial cell surface or secreted into the surrounding environment forming the SML. Molecular interactions between mucin molecules via hydrophobic cysteine-rich domains (Silberberg and Meyer, 1982; Bansil et al., 1995) and the formation of disulfide bonds (Roberts, 1976) results in a viscoeleastic material that provides the host with a physiochemical barrier from the surrounding environment (Gendler and Spicer, 1995; Johansson et al., 2013). In addition to providing the host with protection, the SML also concentrates particles from the environment by providing a smooth layer that encourages laminar (versus turbulent) flow, thereby creating an effective particle trap (Wild et al., 2004; Yang et al., 2012; Hill et al., 2014). There is significant turnover of the SML; the mouse gastrointestinal tract is capable of replacing its entire mucin pool in a single day (Faure et al., 2002) and corals release up to 4.8 l of mucus per square meter of reef per day (Wild et al., 2004).

Microbes and viruses in the SML

Despite the high turnover, the SML is inhabited by a diverse and stable assemblage of microbes and their associated viruses, forming the SML microbiome (Bäckhed et al., 2005; Lozupone et al., 2012; Schluter and Foster, 2012; Closek et al., 2014). Individual members of the SML microbiome gain access to energy-rich mucins (Derrien et al., 2010) while providing the metazoan host with a variety of benefits, including immune protection (Cash and Hooper, 2005; Sun and Chang, 2014) and nutrient production (Thompson et al., 2015). Here we focus on SML-associated bacteria and their predators, bacteriophage (a.k.a. phage). To ensure retention within the SML, bacteria and phage have evolved mucus-binding proteins capable of responding to rapid environmental change. For example, Lactobacillus sp. express a range of proteins containing mucus-binding domains that exhibit high genetic heterogeneity between strains, suggesting they are adaptive (MacKenzie et al., 2010). Similarly, T4 phage use the immunoglobulin-like (Ig-like) domains of their capsid proteins to promote mucus adherence. (Fraser et al., 2006; Barr et al., 2013). Ig-like domains and related protein folds, such as C-type lectins, contain variable regions, potentially allowing phage to adapt to changes in the mucin pool and maintain specific phage–metazoan associations (Minot et al., 2012; Barr et al., 2013).

Phage drive bacterial evolution

Within the SML, phage outnumber their bacterial hosts by roughly an order of magnitude (Barr et al., 2013). Upon infection, phage replicate via either lytic or lysogenic life cycles. The lytic cycle involves the production of new virus particles, ultimately leading to cell lysis and viral release. Alternatively, in the lysogenic cycle the phage genome integrates into the host genome and becomes a prophage. Temperate phage utilize both lytic and lysogenic strategies and are important drivers of evolution in the SML (De Paepe et al., 2016). Depending on the genetic repertoire of the newly acquired prophage, bacterial physiology can be directly affected through the donation of novel genes, disruption of host genes and manipulation of cellular metabolism (Brüssow et al., 2004). In addition, some temperate phage provide their host with immune protection by preventing the attachment of other phage particles (superinfection exclusion) (Soller and Epstein, 1965) or preventing phage propagation of a secondary infection (superinfection immunity) (West and Scott, 1977; Fogg et al., 2010; Abedon, 2015). These mutualistic temperate phage enhance the competitive fitness of their hosts (Bossi et al., 2003) and drive bacterial evolution (Obeng et al., 2016).

Discussion

Colonization of the SML in the last common metazoan ancestor (LCMA)

Fossil evidence and molecular data suggest the LCMA emerged sometime between the Cryogenian and Ediacaran periods approximately 542–720 million years ago (Davidson and Erwin, 2009). The LCMA most likely consisted of two cell layers: an ectoderm with a SML and an internally facing endoderm (Müller, 2003; Lang et al., 2007). We propose that the first bacteria arrived to the SML through active chemotaxis toward energy-rich mucins (Bansil et al., 1995; Stocker and Seymour, 2012), random sequestration by SML fluid dynamics (Wild et al., 2004; Yang et al., 2012; Hill et al., 2014) or both. The dynamic properties of the SML would have selected for bacteria that could be maintained through the expression of mucus-binding proteins or similar mucus-binding mechanisms (MacKenzie et al., 2010). Once established within the SML, bacteria that provided the metazoan host with a fitness advantage via competitive exclusion of potential pathogens or nutrient production would have been further selected. The arrival of the first phage may have occurred in conjunction with the first bacteria as an integrated prophage or from the environment as a temperate/lytic phage. Prophage associated with the first bacterial colonizers increased host fitness through superinfection exclusion and superinfection immunity mechanisms (Soller and Epstein, 1965; Abedon, 2015). As the first bacterial species continued to propogate, phage capable of binding to mucins (for example, Ig-like domains, among others) were favored by natural selection. Colonization by additional bacterial species and their associated phage continued in the SML until niche space was filled and a community was formed. Mutualistic bacteria continued to protect the metazoan host through competitive exclusion of potential pathogens (Hibbing et al., 2010) while phage provided immune protection as proposed in the bacteriophage adherence to mucus (BAM) model (Barr et al., 2013, 2015). In the BAM model, phage adhering to mucus provide the host with immune protection against invading pathogens (Figure 1a). For an in depth discussion of where competitive exclusion and lytic dynamics are operating within the SML, the authors point the reader to the following article (Silveira and Rohwer, 2016).

Figure 1
figure 1

Microbial selective mechanisms in the LCMA. (a) BAM. Energy-rich mucin macromolecules secreted from the ectoderm formed a SML that was colonized by bacteria and associated phage. Mutualistic phage provided the LCMA with immune protection from invading bacteria via (1) competitive exclusion and (2) lytic attack. (b) TNF-mediated apoptosis. Mucins also increased the rate of contact between eukaryotic viruses and their metazoan hosts leading to the formation of mutualistic relationships. Viruses that provided the LCMA host with immune protection were maintained. If a beneficial virus became pathogenic or if a parasitic virus invaded, the infected cell was removed via TNF-mediated apoptosis and other apoptotic mechanisms.

Full size image

Tumor necrosis factor (TNF)-induced apoptosis—an internal microbial selective

Although the colonization of the SML by specific phage species provided the metazoan host with immune protection, the same fluid dynamics also increased the retention of eukaryotic viruses and subsequent adsorption to the LCMA host. As with the phage, these eukaryotic viruses developed mechanisms to bind to mucins and infect host cells (for example, influenza virus today; Wild et al., 2004). Upon infection of a multicellular host, the canonical response is programmed cell death or apoptosis of the infected cell, thus preventing viral dissemination to neighboring cells (Barber, 2001). Apoptosis has been observed across the spectrum of life from bacteria to animals (Lewis, 2000; Bidle et al., 2007). Many versions of apoptosis exist (Holler et al., 2000; Berg et al., 2001; Bratosin et al., 2001) and the general process likely emerged with the origin of multicellularity (Ellis and Horvitz, 1986; Raff, 1992; Steller, 1995; Aravind et al., 2001). However, metazoan apoptosis appears to be unique through its use of TNF receptors (Quistad and Traylor-Knowles, 2016), which are activated by TNF ligands (Aggarwal, 2003). Many of the domains involved with apoptotic signaling via TNF receptors are also present and functional in cnidarians, considered to be among the oldest animal phyla (Lasi et al., 2010; Quistad et al., 2014; Sakamaki et al., 2014, 2015; Lu et al., 2016; Moya et al., 2016).

The targeted destruction of a virally infected cell is the most conservative approach to maintain organismal integrity; however, viruses can also provide the host with a selective advantage. For example, Herpesviruses provide mice with protection from bacterial infection (Barton et al., 2007) and latent dynamics with Herpesviruses and their metazoan hosts have been described from cnidarians (Vega Thurber et al., 2008; Grasis et al., 2014) to humans, suggesting an ancient origin (Steiner, 1996). In addition, similar to temperate phage, metazoan viruses provide their hosts with immune protection via superinfection exclusion (Tscherne et al., 2007; Zou et al., 2009). If the beneficial virus or virally encoded element is transferred to the germline through reverse transcription (RNA viruses) or recombination (DNA viruses), then the trait could be inherited by future generations and drive evolutionary processes (for example, endogenous retroviruses) (Grow et al., 2015). Evidence for past viral co-option events can be found throughout the modern metazoan immune system (Villarreal, 2011) including the canonical response to viral infection: interferon production (Chuong et al., 2016).

We propose that competition between viruses for host niche space led to the formation of mutualistic relationships between the LCMA and its resident viruses. Those associations that provided the host with immune protection were maintained. In those cases where resident viruses were not protective or they developed into a pathogenic infection, the LCMA removed the infected cell via TNF-mediated apoptosis, among other apoptotic mechanisms (Figure 1b). Based on extant animal phyla, the LCMA possessed a large and dynamic stem cell population (Bosch, 2009), therefore, it could rapidly replace any cells deemed to be a risk to organismal integrity without incurring a major fitness cost. Taken together, TNF receptors served as the viral gatekeepers to the LCMA, promoting beneficial chronic infections and eliminating destructive interactions.

Conclusions

Here we have proposed a model for the development of metazoan immunity via external (phage) and internal (TNF-mediated apoptosis) microbial selective mechanisms. The LCMA secreted mucins from epithelial tissue, generating an SML that selected for bacteria and phage with mucin-binding properties. Phage provided the LCMA with an external microbial selective in which phage bound to mucus via hypervariable domains protect the metazoan host from invading bacteria (BAM) (Barr et al., 2013, 2015). In addition to attracting bacteria and phage, mucins also increased the rate of contact with eukaryotic viruses resulting in the development of mutualistic symbiosis that provided the LCMA with immune protection. If a new virus was pathogenic or if a resident virus became parasitic, those cells were eliminated via TNF-mediated apoptosis and other versions of apoptosis. We hypothesize that both microbial selective mechanisms described here evolved during the Precambrian era and continue to drive the evolution of metazoan immunity in modern day phyla.