The RagA and RagB proteins of Porphyromonas gingivalis.

Summary

RagA and RagB proteins are major components of the outer membrane of the oral pathogen Porphyromonas gingivalis and, while recently suggested to represent a novel peptide uptake system, their full function is still under investigation. Herein, we (a) discuss the evidence that the rag locus contributes to P. gingivalis virulence; (b) provide insight to Rag protein potential biological function in macromolecular transport and other aspects of bacterial physiology; (c) address the host response to Rag proteins which are immunodominant and immunomodulatory; and (d) review the potential of Rag-focused therapeutic strategies for the control of periodontal diseases.

Destructive periodontal diseases are highly prevalent infectious diseases of the tissues surrounding the teeth that, as well as being an important health issue, come with significant global financial costs. Periodontitis results from an imbalance in the immune response to a dysbiotic microbial community. Porphyromonas gingivalis is a Gram-negative, anaerobic, asaccharolytic bacterium, and a model periodontal pathobiont. P. gingivalis has also been associated with a number of systemic disease sequelae, such as pre-term birth (Miyauchi et al., 2018; Vanterpool et al., 2016), vascular dysfunction (Carter, France, Crean, & Singhrao, 2017; Xuan et al., 2017), oral and other cancers (Gao et al., 2016; Lafuente Ibanez de Mendoza, Maritxalar Mendia, Garcia de la Fuente, Quindos Andres, & Aguirre Urizar, 2019) and Alzheimer’s disease (Carter et al., 2017; Dominy et al., 2019).

The virulence factors of P. gingivalis include the gene products of the rag locus, which represent major, closely associated proteinaceous components of the outer membrane (Bonass et al., 2000; Curtis, Hanley, & Aduse-Opoku, 1999; Masuda, Murakami, Noguchi, & Yoshimura, 2006; Nagano et al., 2007). Initial interest in RagA and RagB was stimulated by the observation that RagA proteins are immunodominant antigens in subjects with periodontal diseases (Curtis, Slaney, Carman, & Johnson, 1991), a phenomenon confirmed in independent studies (Imai, Murakami, Nagano, Nakamura, & Yoshimura, 2005; Zeller et al., 2014). RagA and RagB are cotranscribed (Hanley, Aduse-Opoku, & Curtis, 1999) and translate to proteins of 55 kDa and 115kDa, respectively (Curtis et al., 1999; Curtis et al., 1991; Shi et al., 2007), as determined in W83, W50 and WPH35. Shi et al reported that insertional inactivation of ragA in strain W50 resulted in abrogated expression of both ragA and ragB gene products (Shi et al., 2007). Additionally, individual ragA and ragB deletion mutants in strain WPH35 were generatedand shown to be phenotypically negative for both proteins (Shi et al., 2007).

RagA and RagB are predicted to occur as the receptor pair on the bacterial cell surface with the former being an integral outer membrane β-barrel protein and the latter RagA-associated lipoprotein. A number of studies showed that RagA and RagB are not enriched in outer membrane vesicles, yet a recent analysis (Haurat et al., 2011; Murakami, Imai, Nakamura, & Yoshimura, 2002) yet recent analysis showed relative abundance of these proteins in vesicular membrane at least in the W50 strain (Veith et al., 2014). Regardless of this discrepancy the pair has been traditionally considered to exhibit typical features of a TonB-dependent outer membrane receptor (Bonass et al., 2000; Curtis et al., 1999; Hall et al., 2005). While the importance of the rag locus in P. gingivalis virulence in vitro and in animal models was established soon thereafter (Dolgilevich, Rafferty, Luchinskaya, & Kozarov, 2011; Nagano et al., 2007; Shi et al., 2007), the biological function of RagA and RagB has received only limited attention beyond their assumed role in transmembrane transport. Recent structure-function analyses and host-pathogen interaction studies, however, have begun to elucidate the roles played by RagA and RagB in bacterial physiology and immune activation (Goulas et al., 2016; Hutcherson, Bagaitkar, et al., 2015; Madej et al., 2020), as addressed herein.

Rag proteins and P. gingivalis virulence

In a study of 132 P. gingivalis strains, ragB was shown to be located at seemingly randomly distributed sites within the chromosome of 17 isolates (13%), with the authors suggesting that ragB may be one of several genes that help determine pathogenic potential in humans (Frandsen, Poulsen, Curtis, & Kilian, 2001). Earlier, Hanley et al had suggested RagAB expression was more likely in P. gingivalis in deeper periodontal pockets, the same group being the first to show that ragA and ragB represented a horizontally-transferred pathogenicity island (Curtis et al., 1999; Hanley et al., 1999).

The contribution of ragAB to the pathogenic potential of P. gingivalis has now been reported in various studies, as summarized in Figure 1. A ragB and, particularly, a ragA mutant generated in strain WPH35 each exhibited reduced virulence following subcutaneous injection, compared to wild-type P. gingivalis, as assessed by either murine lethality or lesion size that necessitated a humane experimental endpoint (Shi et al., 2007) (Figure 1A and B). Attenuated virulence was also noted in a P. gingivalis W83 ragB mutant, as determined by enumeration of viable bacteria disseminated to murine spleens following an initial dorsal subcutaneous inoculum (Nagano et al., 2007) (Figure 1C). In the same study, further evidence for RagA and RagB physical associations was provided in the finding that anti-RagA antibodies co-immunoprecipitated RagB and vice versa for anti-RagB antibodies (Nagano et al., 2007). P. gingivalis W83 ragA and ragB mutants exhibit significantly reduced capacity to enter epithelial cells, as determined in antibiotic protection assays, compared to the parental strain (Dolgilevich et al., 2011) (Figure 1D). In our own work, epithelial invasion by a ragB P. gingivalis 33277 mutant is, essentially, abrogated, as determined in telomerase immortalized gingival keratinocytes (unpublished data). In keeping, P. gingivalis AJW4, and several other strains that are inefficient in epithelial invasion. This was initially attributed to the lack of ragA and ragB orthologs (Dolgilevich et al., 2011) but analysis of this strain genome clearly indicates that a different variant of the RagAB operon is present. Therefore, it remains to be determined if differences in epithelial cell invasions by P. gingivalis strains are related to expression of different variants of RagA and RagB proteins.

Figure 1. Virulence traits of the ragAB locus of P. gingivalis.

Subcutaneous injection of P. gingivalis WPH35 resulted in (A) lesion formation and/or (B) murine lethality that was attenuated upon ragA or ragB insertional inactivation (Shi et al., 2007). (C) In P. gingivalis W83, mutants of ragB were recovered from mouse spleens at significantly reduced levels compared to parental strain, despite doubled infectious inocula (Nagano et al., 2007). (D) The ability of AJW4, a strain expressing different variants of ragA and ragB, to invade KB epithelial cells was dramatically compromised, relative to W83. Mutation of either rag gene in W83 resulted in a similarly suppressed epithelial colonization phenotype (Dolgilevich et al., 2011).

Several studies have provided further indirect evidence that RagA and RagB may be relevant to human periodontal diseases, in that expression levels have been associated with important disease risk factors or other features. For example, ragAB activity has been shown to be temperature-, but not pH-, osmolarity- or oxygen-regulated (Bonass et al., 2000; Masuda et al., 2006; Murakami et al., 2002; Murakami et al., 2004), suggesting that the rag locus may be responsive to subgingival inflammatory fluxes. Indeed, Liu, Zhang et al (2013) reported a correlation between the detection of P. gingivalis rag gene transcripts in gingival crevicular fluid and periodontal clinical indices. Stress has been associated with periodontal diseases risk (Hilgert, Hugo, Bandeira, & Bozzetti, 2006), while adrenaline has been shown to increase the activity of ragA and several other virulence-related genes (Graziano et al., 2014). Interestingly, destructive periodontal diseases in humans are considerably exacerbated by tobacco consumption, accompanied with increased P. gingivalis infectivity and recalcitrance (Zeller et al., 2014). RagA and RagB expression is considerably enhanced in cigarette smoke extract-exposed P. gingivalis W83 (Bagaitkar et al., 2009). However, the etiological pertinence of enhanced Rag protein expression in active inflammatory episodes, during stress, or in smokers has yet to be determined.

RagA and RagB localization to the P. gingivalis surface occurs independently of the type IX secretion system (T9SS) that is engaged in translocation across the outer membrane, posttranslational maturation and surface anchorage of proteins possessing the conserved C-terminal domain. In stark contrast to T9SS cargo proteins, such as gingipains, which are highly enriched in outer membrane vesicles (OMV) RagA is not trafficked into these secretory structures of either W50 or ATCC 33277. Interestingly, however, both Rag proteins are abundant in vesicles derived from porS and waaL W83 mutants impaired in synthesis of polysaccharide component of LPS. These mutants release gingipain into the medium in soluble forms and do not pack them into vesicles (Haurat et al., 2011; Murakami et al., 2004). Apparently, deficiency of LPS affects mechanism of protein sorting onto vesicles and wild-type strains of P. gingivalis are devoid of RagA. Therefore, their role in P. gingivalis pathogenesis would appear to be limited to whole cells. Next, then, we will consider the physiological role(s) played by RagB in intact bacteria.

Rag protein function in P. gingivalis

While definitive functions for the Rag proteins of P. gingivalis have yet to be elucidated, several clues are, nevertheless, available. Rag proteins do not appear to play a significant role in susceptibility or resistance to a bank of common antibiotics nor to exhibit hemagglutinin capabilities (Nagano et al., 2007). Nagano et al (2007) noted that ragA and ragB W83 mutants grew normally in rich medium, but that growth was slower in medium in which P. gingivalis was required to metabolize native proteins to meet nutritional needs. The authors also confirmed the requirement for a fully intact operon for the translation of either protein component of the RagAB complex, and its essentiality for P. gingivalis growth on proteins despite unaffected gingipain activity. Therefore, it was hypothesized that cell surface-associated RagA and RagB were involved in the translocation of protein degradation products. As comparisons of the orthologous rag genes suggest a saccharide transport function, a potential role in peptide acquisition was intriguing. It is of particular interest that, while a broad assortment of secreted peptidases and proteases and pathways for amino acid metabolism are important virulence components, the genome of P. gingivalis appears to encode only two predicted peptide uptake systems, PG0444 and, PG2082, as proposed following interpretation of the complete genome sequence of strain W83 (Nelson et al., 2003).

To expand, RagA shows substantial homology to SusC of Bacteroides thetaiotaomicron, an outer membrane protein considered to be involved in starch uptake, and similarity to a number of TonB-dependent outer membrane transporter proteins, such as the Escherichia coli proteins FepA, FhuA, and BtuB (Hall et al., 2005). Limited homology between RagB and the SusC partner protein, SusD, is also apparent (Hall et al., 2005) with RagB bearing hallmarks of a signal peptidase II-cleavable lipoprotein (Nagano et al., 2007). RagB also exhibits homology to the Tannerella forsythia protein, NanU (Goulas et al., 2016), a component of a SusC/SusD-like neuraminate uptake system, NanO/NanU (Phansopa et al., 2013). SusC/SusD and NanO/NanU are involved in malto-oligosaccharide/starch (Reeves, D'Elia, Frias, & Salyers, 1996; Reeves, Wang, & Salyers, 1997) and sialic acid (Phansopa et al., 2013) binding and acquisition, respectively. Therefore, RagA and B have been traditionally thought to be involved in sugar binding and translocation in P. gingivalis. Indeed, the solved crystal structure of purified RagB putatively identified three potential saccharide-binding sites on the RagB molecular surface. It should be noted that while P. gingivalis is asaccharolytic, this does not mean that this bacterium does not have a need for sugar intake. However, saccharide transport must be for reasons other than energy acquisition, e.g., capsule generation and other structural requisites (Hutcherson et al., 2015) or self-glycosylation of proteins, including important surface molecules such as gingipains (Curtis et al., 1999b), peptidylargine deiminase (PPAD) (Glew et al., 2014) and the minor fimbrial antigen, Mfa1 (Zeituni, McCaig, Scisci, Thanassi, & Cutler, 2010).

Recent structure-function analyses have revealed more specific macromolecular transport possibilities. Both RagB (Goulas et al., 2016) and the RagAB complex (Madej et al., 2020) have now been crystalized. We have shown that RagB consists of four tetratrico peptide repeats (TPR1-4), each arranged as two helices connected by a linker, plus two capping helices found downstream. In essence, substantial RagB structural similarity with B. thetaiotaomicron SusD and T. forsythia NanU, was confirmed. In keeping with these phenomenon, three potential monosaccharide-binding sites were tentatively assigned to the RagB molecular surface (Figure 2).

Figure 2. Structural similarities of the P. gingivalis RagB and Bacteroides thetaiotaomicron SusD proteins.

(A) Superposition in cross-eye stereo of PgRagB (blue ribbon, tentative sugars in turquoise) and BtSusD in its complex with maltoheptaose (red ribbon, sugar moieties in pink; PDB 3CK9; (Koropatkin, Martens, Gordon, & Smith, 2008). (B) Orthogonal view of (A). The figure is reproduced, with permission, from (Goulas et al., 2016).

Intriguingly, however, analysis of the combined RagAB complex has returned contemporary thought to Rag proteins functioning as peptide transporters. The x-ray crystal structure of RagAB purified from P. gingivalis W83 revealed a dimeric RagA2B2 complex with an architecture resembling that of SusCD (Madej et al., 2020), a TonB-dependent carbohydrate transporter in B. thetaiotaomicron (Glenwright et al., 2017). In the heterotetrameric complex, each RagA molecule, with the structure of a 22 stranded β-barrel similar to SusC, is tightly capped on the extracellular side by RagB (equivalent of SusD) with peptide 13 residues in length visible in the large, closed cavity between subunits. The dimeric structure of RagAB was confirmed by single-particle cryo-EM structural analysis of a detergent solubilized complex. Three distinct conformations of the complex were identified, corresponding to (a) both RagA barrels closed by their RagB (closed-closed, CC), (b) one RagA barrel open, the other closed (open-closed) and (c) both open (open-open) (Figure 3). This suggests that RagB molecules work as lids that can open and close separately in the dimeric complex. A peptide was found in each structure occupying the same position as in the crystal structure of RagAB complex in the CC conformation. These results, together with functional characterization of the RagAB system, provide direct evidence for “pedal bin” mechanism of peptide uptake by P. gingivalis, which is essential for the effective consumption of proteinaceous nutrients (Madej et al., 2020).

Figure 3. Differential conformations of the RagA2B2 transporter of P. gingivalis.

Representations of the (A) Closed-Closed (B), Open-Closed and (C) Open-Open conformations of RagA2B2, as determined by cryo-EM. Views are from the plane of the outer membrane (left and middle) and from the extracellular space (right). Bound peptide is shown in magenta while the plug domains of RagA are coloured dark blue. The figure is reproduced, with permission, from (Madej et al., 2020).

It is important to note that a post-translational modification system, lysine residue succinylation, has recently been reported that influences multiple proteinaceous virulence factors of P. gingivalis (Wu et al., 2019). These include gingipains and fimbriae as well as RagB. While it is known that such succinylation can adjust protein function through structural and charge alterations, the pathogenic implications of this and other potential post-translational modifications for RagAB and P. gingivalis are currently unclear. Although earlier analysis suggested that only a subset of P. gingivalis strains contain the rag locus (Curtis et al., 1999; Frandsen et al., 2001; Su et al., 2010) whole genome sequencing clearly indicates that the ragAB operon is present in all type strains and clinical isolates analyzed to date, as determined by interrogation of the NCBI data base. It appears that up to 30% differences in the RagA sequence between different strains of was responsible for failure to detect the ragA gene in some strains. The same is true for ragB. Four rag variant types have been described or detected in oral samples (Dashper et al., 2017; Hall et al., 2005; Y. Liu, Zhang, Wang, Guo, & Xiao, 2013; Su et al., 2010; Wang, Zhang, & Pan, 2009). In a comparison of the genomes of 23 P. gingivalis strains Dashper et al noted that of the ragA and ragB variant alleles described by Hall et al, types 2 and 4 were the most prevalent (Dashper et al., 2017). Hall et al, however, reported that rag type 1 was the most prevalent allele detected amongst 168 clinical isolates (Hall et al., 2005). In a related study, rag types-3 and rag-4 were the predominant genotypes in the patients of orthodontic gingivitis and mild-to-moderate periodontitis (Y. Liu et al., 2013). Only a single ragAB operon has been noted in individual strains, while ragA is not found without ragB or vice versa (Dashper et al., 2017; Hall et al., 2005). Any potential for functional variation of alternate rag loci has yet to be ascertained but, based on Madej et al (2020), a new concept is emerging that different variants of the ragAB operon allow P. gingivalis strains to forage on different sets of peptides generated by degradation of host proteins. The contribution of this variation to P. gingivalis fitness in vivo is an open question. We will next consider the immune response to rag gene products.

Host response to Rag proteins

As noted above, the rag locus encodes immunodominant P. gingivalis antigens (Curtis et al., 1991; Imai et al., 2005; Zeller et al., 2014), is temperature-responsive (Bonass et al., 2000), while Rag protein expression profiles are related to periodontal pocket depth (Curtis et al., 1999; Hanley et al., 1999) and the detection of rag genes has been associated with the clinical index of gingival tissue (Y. Liu et al., 2013).

Early studies demonstrated that the IgG response to RagB, as determined by serum antibody recognition of the W83 RagB protein, was higher in adult periodontitis subjects than matched controls (Curtis et al., 1991). More recently, Imai et al (2005) and Zeller et al (2014) reported a strong systemic RagB-cognizant antibody response in a subset of subjects with periradicular lesions as well as chronic and advanced periodontitis, according to the disease classification system in use at the time. However, a deeper understanding of Rag-immune interactions remains largely elusive.

To this end, Hutcherson et al (2015a) exposed professional, human inflammatory cells to purified, recombinant RagB (Hutcherson, Bagaitkar, et al., 2015). In this experimental system, RagB induced the expression of multiple genes encoding proinflammatory mediators in monocytes, including IL-1α, IL-1β, IL-6 and IL-8 and CCL2 in a dose-related manner. RagB was shown to act as a pro-inflammatory mediator by acting as Toll-like receptor 2 (TLR2) and TLR4 agonist that activates STAT4 and NF-κB signaling. In keeping with the in vitro rRAGB data, a ragB mutant exhibited reduced inflammatory capacity in comparison with wildtype P. gingivalis W83, in a manner that was rescued upon ragB complementation. Therefore, RagB could play an important role in the etiology of P. gingivalis-associated periodontal inflammation. To the best of our knowledge, there are no equivalent studies of potential host response engagement of RagA or composite RagAB.

Rag proteins as therapeutic targets for periodontal disease

In addition to the identification of Rag proteins as an immunoactive virulence factor, as discussed above, a number of studies suggest that the rag locus may represent an appropriate therapeutic target for P. gingivalis-associated diseases. Melatonin receptor agonists, N-acetyl-5-methoxytryptamine (melatonin) and its derivative, ramelteon, have been shown to inhibit planktonic and biofilm growth of P. gingivalis 33277 and, at sub-optimal antimicrobial concentrations, to suppress the activity of key virulence factors, including gingipains and ragA (Zhou et al., 2016). Zhao et al. (2020) employed a metal organic framework, empirically C₃₅H₂₁Br₄N₆O₄Zn, which abrogated the growth of an unspecified P. gingivalis strain. The authors argued that an affiliated reduction in ragA and ragB gene activity implicates RagA and B as essential for growth. It should be noted, however, that transposon sequencing interrogation by several laboratories has not identified ragA or ragB as inherently or conditionally essential (Hutcherson, Gogeneni, et al., 2015; Klein et al., 2017; Klein et al., 2012; Miller et al., 2017; Miller & Scott, 2021; Naito, Tominaga, Shoji, & Nakayama, 2019).

As RagB predominates the P. gingivalis-cognizant antibody response in humans and is an important bacterial virulence factor, this particular surface protein been considered as an attractive potential vaccine target. Kong et al (2015) developed a RagB B cell antigen epitope vaccine and utilized it to generate an efficient RagB immunoglobulin response that was accompanied by adaptive immune cell expansion and reduced abdominal lesion size in W83 inoculated mice. This work built on the prior publication were Zheng et al (2013) immunized mice with a W83 ragB DNA-based vaccine and observed efficient RagB-specific IgG production. In a similar manner to Kong et al 2015), an enhanced follicular Th cell population and reduced lesion size following delivery of P. gingivalis was noted. Interestingly, TonB-dependent outer membrane receptors have been studied as potential vaccine targets for a variety of other bacteria, not limited to Neisseria meningitidis (Stork et al., 2010), Burkholderia mallei and Burkholderia pseudomallei, where TonB mutants exhibit reduced virulence, Actinobacillus pleuropneumoniae (Khakhum et al., 2019; Mott, Vijayakumar, Sbrana, Endsley, & Torres, 2015), the aquaculture pathogen, Pseudomonas fluorescens (Hu, Dang, & Sun, 2012), and the porcine pathogen, Actinobacillus pleuropneumoniae (J. Liu et al., 2011).

Conclusions

To summarize, RagB has been associated with P. gingivalis virulence by contributing to efficient growth, subcutaneous lesion development and epithelial cell invasion. While rag-based vaccine development has received less attention of late, the recent advances in our knowledge of RagB and RagAB composite atomic structures (Goulas et al., 2016; Madej et al., 2020) have the potential to drive renewed. This is due to the recognition of RagAB as a potential elusive oligopeptide acquisition machine and the identification of key functional domains within the RagAB heterotetramer.

Data availability statement:

Data sharing not applicable. This is a review article.