Many prokaryotes utilize type IV secretion systems (T4SSs) to translocate substrates (e.g. nucleoprotein, DNA, protein) across the cell envelope, and/or to elaborate surface structures (i.e. pili or adhesins). Among eight distinct T4SS classes, P-T4SSs are typified by the Agrobacterium tumefaciens vir T4SS, which is comprised of 12 scaffold components (VirB1–VirB11, VirD4). While most P-T4SSs include all 12 Vir proteins, some differ from the vir archetype by either containing additional scaffold components not analogous to Vir proteins or lacking one or more of the Vir proteins. In a special case, the Rickettsiales vir homolog (rvh) P-T4SS comprises unprecedented gene family expansion. rvh contains three families of gene duplications (rvhB9, rvhB8, rvhB4): RvhB9,8,4-I are conserved relative to equivalents in other P-T4SSs, while RvhB9,8,4-II have evolved atypical features that deviate substantially from other homologs. Furthermore, rvh contains five VirB6-like genes (rvhB6a–e), which are tandemly arrayed and contain large N- and C-terminal extensions. Our work herein focuses on the complexity underpinned by rvh gene family expansion. Furthermore, we describe an RvhB10 insertion, which occurs in a region that forms the T4SS pore. The significance of these curious properties to rvh structure and function is evaluated, shedding light on a highly complex T4SS.
Type IV secretion systems (T4SSs) are a diverse group of prokaryotic transporters, with eight distinct classes currently recognized (Guglielmini, de la Cruz and Rocha 2013; Guglielmini et al.2014). The MPF-T class, hereafter referred to as P-T4SSs (related to systems of the IncP group conjugative plasmids), are perhaps the best studied group, typified by decades of research on the vir T4SS of the phytopathogen Agrobacterium tumefaciens. Agrobacterium tumefaciens uses its vir T4SS, which is encoded by 12 genes (virB1-virB11, virD4), to transfer oncogenic DNA into plant cells to facilitate colonization (Crown Gall disease) (Kado 2014; Nester 2015). Studies on P-T4SSs encoded by the pKM101 and R388 conjugative plasmids of Escherichia coli have yielded tremendous insight on secretion machine architecture, revealing a massive 3-MDa nanomachine that spans the bacterial cell envelope (Chandran et al.2009; Fronzes et al.2009; Rivera-Calzada et al.2013; Low et al.2014). Conservation of gene composition and synteny across many species harboring these systems indicates that the P-T4SS architecture is likely similar despite the diversity of known translocated substrates, including plasmid DNA (Llosa et al.2002; Lawley et al.2003), peptidoglycan (Viala et al.2004), nucleoprotein such as T-DNA (Soo et al.2015), and protein effectors (Cascales and Christie 2003; Siamer and Dehio 2015).
Notwithstanding this genetic and structural conservation, studies on several P-T4SSs have revealed specializations that deviate from the ‘vir paradigm’ (Alvarez-Martinez and Christie 2009). The P-T4SSs of Brucella spp. (vir) and Bordetella pertussis (ptl) lack genes encoding VirD4, the T4SS coupling protein (T4CP), and thus rely on another mechanism for substrate translocation across the inner membrane (IM) (Locht and Keith 1986; Nicosia et al.1986; Delrue et al.2001; Ke et al.2015). ptl is also pilus-less, a consequence of a lack of the minor pilus subunit gene virB5 (Verma and Burns 2007). Aside from analogs to the 12 vir genes, the Helicobacter pylori cag P-T4SS contains an additional five scaffold components and 10 other cag-associated proteins, all of which facilitate translocation of the cytotoxin CagA, as well as peptidoglycan, into gastric epithelial cells (Terradot and Waksman 2011). The trw P-T4SS of Bartonella species is not at all involved in protein translocation, yet instead elaborates variable surface pili comprised of different combinations of proliferated VirB2 (TrwL) and VirB5 (TrwJ) proteins (Seubert et al.2003), with such structures interacting with host erythrocytes (Vayssier-Taussat et al.2010; Deng et al.2012). Remarkably, the vir P-T4SS of Xanthomonas citri, which has recently been characterized for ‘bacterial killing’ similar to T6SSs (Souza et al.2015), has VirB7 fused to an N0 domain (found in secretins from T2SSs, T3SSs, pIV filamentous phage and type IV pili) resulting in a ring-like structure around the outer membrane (OM) pore (Souza et al.2011). Collectively, these P-T4SS structural adaptations illustrate that characterization of unique systems is warranted to understand the manner by which P-T4SSs recognize host cells and translocate substrates.
Another highly specialized P-T4SS, Rickettsiales vir homolog (rvh) (Gillespie et al.2009), is encoded by all species of Rickettsia (Fig. 1). These Gram-negative obligate intracellular bacteria are parasites of a wide range of eukaryotic hosts, with some pathogenic species garnering special interest both as agents of emerging infectious diseases and for potential as weapons of bioterrorism (Azad 2007; Gillespie et al.2012a). Like ptl, the rvh T4SS lacks a virB5 gene, which is consistent with no observable pili on Rickettsia cell surfaces (Gillespie et al.2015a). Oddly, the rvh loci are found in scattered islands throughout Rickettsia genomes (Gillespie et al.2008). However, the most remarkable feature of the rvh T4SS is its unprecedented gene family expansion (Gillespie et al.2009). rvh contains three families of gene duplications (rvhB9, rvhB8, rvhB4): RvhB9,8,4-I are conserved relative to equivalents in other P-T4SSs, while RvhB9,8,4-II have evolved atypical features that deviate substantially from other homologs (Gillespie et al.2010). Furthermore, rvh contains a staggering proliferation of five VirB6-like genes (rvhB6a–e), which are tandemly arrayed and contain large N- and C-terminal extensions (Gillespie et al.2009). Curiously, we previously identified one of these proteins (RvhB6a) on the surface of Rickettsia typhi cells (Sears et al.2012). Aside from our very recent study that characterized the first rvh effector secreted by R. typhi during host cell infection (Rennoll-Bankert et al.2015), virtually nothing is known about how the anomalous rvh machine assembles to translocate effectors into host cells.
This study focuses on the complexity underpinned by rvh gene family expansion, including recent structural and expression data that highlight the oddity of rvh relative to other P-T4SSs. Additionally, we explore the relevance of a large insertion within RvhB10, which can be localized to a region of the OM that forms the T4SS pore. This new insight is discussed in light of an emerging role for the rvh T4SS in translocating protein effectors into host cells. While our focus is on Rickettsia species, several comparisons are made with other species of Rickettsiales in the families Midichloriaceae and Anaplasmataceae, which all harbor a version of the rvh T4SSs (Gillespie et al.2010, 2015a). For more information on the rvh T4SSs of species of Anaplasmataceae, the reader is referred to excellent reviews by Rikihisa and colleagues (Rikihisa et al.2009; Rikihisa and Lin 2010; Rikihisa, Lin and Niu 2010). Our major goal here is to illuminate the oddity of the rvh T4SS; thus, while we focus on Rickettsia species, we anticipate that this synopsis will provide valuable information to guide future studies that characterize the structure and function of this atypical secretion system for any species of Rickettsiales.
Three peculiar gene duplications define the rvh T4SS
rvhencodes two analogs each for VirB9, VirB8 and VirB4
Many bacterial genomes encode multiple, divergent T4SSs (e.g. dot/icm and lvh of Legionella spp.) and some genomes encode multiple copies of highly similar T4SSs (e.g. vir and vbh of Bartonella spp.). Rickettsial genomes are unique in that they encode a single T4SS enriched with gene duplication (RvhB9, RvhB8, RvhB4) and proliferation (3–5 copies of RvhB6) (Gillespie et al.2009, 2010). Regarding duplicate families, one paralog is conserved relative to equivalents in other P-T4SSs (rvhB9-I, rvhB8-I, rvhB4-I), while the other paralog has evolved atypical features that deviate substantially from other homologs (rvhB9-II, rvhB8-II, rvhB4-II). The significance of rvh paralogy remains unknown. At the heart of the matter is whether all six paralogs assemble into a single secretion machine, or alternatively, if different subsets of paralogs (e.g. RvhB9,8,4-I versus RvhB9,8,4-Il) differentially contribute to a T4SS that is dynamic throughout the complex rickettsial lifecycle. A better understanding of the ambiguity associated with rvh paralogy can be gained by considering the characteristics of RvhB9,8,4 paralogs, which importantly, are highly conserved in all species of Rickettsiales but unknown from other P-T4SSs (Gillespie et al.2010).
The long and short of RvhB9 proteins
VirB9 proteins contain N-terminal Sec signals and migrate to the OM after secretion. VirB9 interacts with the small lipoprotein VirB7 (Bayliss et al.2007), and the VirB7-VirB9-VirB10 heterotrimer forms the functional unit of the tetradecameric core complex (CC) (Chandran et al.2009; Fronzes et al.2009; Rivera-Calzada et al.2013). Prior to formation of the CC, VirB9 interacts with both VirB8 and VirB10 (Sivanesan et al.2010), as well as with VirB1 and VirB11 and probably VirB4 (see below). Unlike RvhB8 and RvhB4 paralogs, Rickettsia RvhB9 paralogs differ in architecture: while the entire N-terminal domains (NTDs) of both proteins (40% aa identity) are comparable to other VirB9 proteins, the C-terminal domain (CTD) of RvhB9-II is highly truncated (Fig. 2A). Thus, RvhB9-II lacks the necessary domain to interact with RvhB7 and RvhB10 in the CC, as well as the additional transient interactions mediated by the VirB9-CTD (VirB1, VirB8-VirB11). Despite this, we previously showed a functional Sec signal for R. typhi RvhB9-II (Ammerman, Rahman and Azad 2008). While CTDs are absent from RvhB9-II proteins of all Rickettsiaceae (species of Rickettsia, Occidentia, Orientia and Arcanobacter), they are present in RvhB9-II proteins of Anaplasmataceae (species of Neorickettsia, Xenolissoclinum, Wolbachia, Anaplasma and Ehrlichia) and Midichloriaceae (species of Midichloria and Jidaibacter), though these regions show variability in several residues that are highly conserved in their RvhB9-I counterparts and VirB9 proteins from other P-T4SSs (data not shown). Nonetheless, the functions of truncated RvB9-II proteins are likely limited to interactions described for the NTD of VirB9 proteins, namely interactions with VirB4 in the periplasm near the IM (Walldén et al.2012).
RvhB8 paralogs are structurally divergent
As a bitopic IM protein, VirB8 is a critical factor in assembly of the IM channel (IMC) of T4SSs (Sivanesan et al.2010). Specifically, VirB8 dimerizes via its CTD, with this globular structure mediating interactions with other T4SS proteins in the periplasm (Paschos et al.2006; Andrieux et al.2011). Aside from a few conserved residues, most of which underpin dimerization, VirB8 proteins are very divergent at the sequence level across diverse P-T4SSs (Sivanesan and Baron 2011). Despite the lack of sequence similarity, initial VirB8-CTD structures from Brucella suis (Terradot et al.2005) and A. tumefaciens (Bailey et al.2006) were highly similar, adopting a large extended β sheet (β1–β4) juxtaposed against five α-helices (α1–α5). Structures solved more recently for three VirB8 and two TrwG proteins from several Bartonella species confirmed the structural conservation of this globular VirB8-CTD fold (Gillespie et al.2015b). Furthermore, despite minimal sequence conservation across Bartonella VirB8 and TrwG proteins (∼28%ID), heterodimers could form in vitro, illustrating the importance of the globular fold in mediating dimerization. Remarkably, structures for other VirB8-like proteins from Gram-positive T4SSs, namely TcpC of Clostridium perfringens (Porter et al.2012) and TraM of Enterococcus faecalis (Goessweiner-Mohr et al.2013), revealed trimers as functional units. Additional structures solved recently for DotI (Legionella pneumophila) and TraM (IncI plasmid R64) of I-T4SSs revealed octomers and hexamers, respectively, with DotI also oligomerizing with its truncated paralog DotJ (Kuroda et al.2015). Collectively, these studies illustrate the importance of VirB8 oligomerization in the formation of the IMC of T4SSs.
Rickettsia RvhB8 paralogs are highly divergent from one another (21% aa identity) (Fig. 2B, top); yet, both proteins are similar in size and domain architecture to VirB8 proteins from other P-T4SSs (Gillespie et al.2009). However, the highly conserved NPXG motif, a loop formed between helix α5 and strand β4 that contacts helix α1 across the VirB8 dimerization interface, is not conserved in RvhB8-II proteins. The recently solved crystal structure of R. typhi RvhB8-II (PDB 4O3V) revealed an atypical dimer relative to the other solved VirB8 and TrwG structures, while structural modeling of R. typhi RvhB8-I suggests this protein adopts the typical VirB8 globular fold (Gillespie et al.2015b) (Fig. 2B, middle). Despite both RvhB8 paralogs forming dimers in solution, the divergent structures indicate that RvhB8-I/RvhB8-II heterodimers do not form. Remarkably, deviant NPXG motifs are present within RvhB8-II proteins from all Rickettsiales species (Fig. 2B, bottom). Evaluation of 1239 non-redundant proteobacterial VirB8 proteins revealed a small subset (n = 131) of NPXG variants, but all of these proteins could be modeled with high confidence to monomeric and dimeric VirB8 structures (Gillespie et al.2015b). Thus, the rvh T4SS is unique in that two structurally divergent VirB8 dimers potentially contribute to IMC assembly, though RvhB8-II would certainly assemble in a non-canonical manner. Similar to the structures of DotI, TraM (R64) TcpC, and TraM (E. faecalis), RvhB8-II lacks helix α5, indicating this protein might oligomerize differently than the conserved dimerization of VirB8 proteins of P-T4SSs. Future determination of rvh IMC composition will help resolve the structural and functional significance of the two divergent RvhB8 proteins.
Only one RvhB4 paralog contains a conserved NTPase active site
The VirB4 ATPase is the most conserved component of T4SSs (Alvarez-Martinez and Christie 2009; Guglielmini, de la Cruz and Rocha 2013), functioning in both substrate transfer, pilus biogenesis and VirB2 dislocation from the IM (Fullner, Lara and Nester 1996; Christie et al.2005; Kerr and Christie 2010). Like VirD4 and VirB11, the VirB4 CTD harbors a highly conserved nucleotide-binding domain (NBD) that is critical for ATPase activity (Berger and Christie 1993; Fullner, Stephens and Nester 1994; Watarai, Makino and Shirahata 2002). The NBD includes Walker A and B boxes, as well as three other conserved motifs (C, D and E), of which motif D (a critical arginine finger) limits ATPase activity to oligomers (Walldén et al.2012). T4SS macro-structures place the VirB4 NTD bound (Walldén et al.2012) or in close proximity (Low et al.2014) to the NTD of VirB9 at the IM/periplasm, supporting a previous hypothesis that VirB4 regulates secretion (Cascales and Christie 2004b).
While similar in size (∼800 aa) and domain architecture relative to VirB4 proteins of other P-T4SSs, Rickettsia RvhB4 paralogs are quite divergent from one another (25% aa identity) (Fig. 2C, top). RvhB4-I active site motifs are conserved relative to other VirB4 proteins; however, RvhB4-II proteins contain mutations in critical residues within motifs WB and C–E (Fig. 2C, middle). Curiously, atypical RvhB4-II proteins are conserved across all Rickettsiales genomes (Fig. 2C, bottom). Moreover, in silico analysis of hundreds of VirB4-like proteins did not uncover proteins containing mutations across all five conserved motifs (data not shown). Thus, ‘active site mutant’ VirB4 proteins appear to be unique to the rvh T4SS. Accordingly, we hypothesize that RvhB4-II proteins are incapable of ATP hydrolysis, yet might still assemble into the rvh T4SS since the proteins are full length. Such incorporation into the IM may provide an antagonistic role to regulate secretion, particularly by preventing RvhB4 oligomerization since arginine fingers (motif D) are absent from RvhB4-II proteins. Deciphering the interacting partners of RvhB4-II proteins will be necessary to understand why rvh encodes a second VirB4 molecule seemingly incapable of providing energy for substrate transfer and RvhB2 dislocation from the IM.
rvhstructure and function in light of simultaneously expressedrvhB9,8,4paralogs
We previously demonstrated that both rvhB8 genes are expressed by R. typhi during infection of HeLa cells (Gillespie et al.2015b). Herein, we show that R. typhi constitutively expresses all six genes encoding RvhB9,8,4 paralogs throughout 1 day of HeLa cell infection (Fig. 2D). Whether all six RvhB9,8,4 proteins are simultaneously translated remains to be determined. For R. typhi (Fig. 1) and other Rickettsia genomes (Gillespie et al.2009), rvh gene arrangement is muddled. In particular, rvhB8-I and rvhB9-II loci are adjacent within a large rvh island that also includes rvhB7, rvhB10, rvhB11 and rvhD4, while rvhB8-II and rvhB9-I are found in a separate operon. Thus rvhB9 and rvhB8 paralogs are inextricably co-expressed. And while rvhB4 genes are well separated in Rickettsia genomes, with only rvhB4-I associated with other rvh genes (rvhB3 and rvhB6a–e), both paralogs are highly expressed throughout infection. Thus, if Rvh-I (RvhB9,8,4-I) and Rvh-II (RvhB9,8,4-II) paralogs spatiotemporally assemble into an rvh secretion machine at different stages of the rickettsial lifecycle (e.g. early versus late infection, or arthropod versus vertebrate host), posttranscriptional regulation would be required to purge those rvh transcripts that are unnecessary at specific time points.
In light of the simultaneous expression of rvhB9,8,4 paralogs, we offer three possible scenarios for concurrent incorporation of Rvh-I and RvhB-II proteins into functional secretion machines. First, Rvh-I and Rvh-II paralogs could assemble into discrete machines, with Rvh-I functioning in effector translocation and RvhB-II forming a transporter with unknown function (Fig. 2E, left and right). Such a scenario would entail a remarkably precise assembly mechanism with both machines in close proximity, since RvhB2 proteins are predominantly unipolar on R. typhi cells (Gillespie et al.2015a). However, a function for Rvh-II is difficult to envision, since such a machine would lack an OM CC (no VirB9 CTD), contain an atypical IMC (RvhB8-II) and lack energy from a mutant ATPase (RvhB4-II). Second, Rvh-I and Rvh-II paralogs could cycle ‘on-and-off’ of a core rvh scaffold, regulating the secretion of substrates throughout the rickettsial lifecycle (Fig. 2E, center). Such an autoregulatory function is appealing since the Rvh-II paralogs all contain characteristics that would seemingly block substrate translocation; however, a mechanism for orchestrating such autoregulation would be extraordinarily intricate, particularly given the entangled expression of rvhB9,8,4 paralogs. Finally, Rvh-I and Rvh-II paralogs could collectively assemble into a single rvh T4SS. Considering the stoichiometry of machine components determined in the electron microscopic structure of the R388 P-T4SS VirB3-VirB10 complex (Low et al.2014), a ratio of 14:12:12 can be expected for RvhB9, RvhB8 and RvhB4 proteins within the rvh T4SS. Exactly how the Rvh-I and Rvh-II paralogs would account for these copy numbers, or whether such a complex machine would deviate from the R388 P-T4SS architecture, is unclear.
rvh encodes five aberrant VirB6-like proteins
Proteins of the VirB6/TrbL family are defined by their pronounced hydrophobicity, which facilitates their insertion into the IM where they interact with other IMC T4SS components (Fig. 1). These highly polytopic proteins, which typically contain 4–5 predicted transmembrane-spanning (TMS) regions, are extraordinarily divergent across systems, with conservation limited to a central TMS-cytoplasmic loop that contains a highly conserved Trp residue (Christie, Whitaker and González-Rivera 2014). This Trp residue is essential (Lawley et al.2003) and required for VirB6 polar localization (Judd, Mahli and Das 2005), and in conjunction with selective constraints on overall hydrophobicity, functions to assist the transfer of substrates to VirB8 in the IMC (Jakubowski et al.2004; Judd, Kumar and Das 2005). While typically ∼300 aa in size, larger VirB6/TrbL proteins are known from several diverse T4SSs, with extensions found at either the N- or C-terminus (Alvarez-Martinez and Christie 2009). In F-T4SSs, these ‘VirB6 extensions’ function in mating pair formation/entry exclusion (Marrero and Waldor 2005, 2007; Audette et al.2007) as well as DNA release prior to transformation (Kohler et al.2013). For P-T4SSs, larger VirB6 proteins have not been functionally characterized.
Proliferated VirB6-like proteins are a hallmark of the rvh T4SS, with copy number of tandemly arrayed RvhB6 proteins ranging from 3 to 5 across diverse Rickettsiales species (Gillespie et al.2010). In addition to this oddity, all RvhB6 proteins contain either an extended N-terminal sequence (ENS) or an extended C-terminal sequence (ECS), or both. Specific protein domains have not been identified within each ENS and ECS, though they tend to be laden with repeat regions (RRs) that often exhibit extraordinary variability across species and even closely related strains; e.g. RvhB6-4 (NCBI accession no. ABD44094) of Anaplasma phagocytophilum (Al-Khedery et al.2012). For Ehrlichia chaffeensis, all four of its rvhB6 genes are co-expressed during growth in tick and human cells, with various RvhB6–RvhB6 and RvhB6–RvhB9 interactions suggesting that multiple RvhB6 proteins assemble at the rvh IMC (Bao et al.2008). Furthermore, at least one RvhB6 protein (YP_507312) was shown to be extracellular within the E. chaffeensis vacuole, indicating that RvhB6 proteins may have additional functions beyond facilitating substrate translocation at the IMC.
Aside from a few minor exceptions (Gillespie et al.2009), all Rickettsia species encode five tandemly arrayed VirB6-like proteins, RvhB6a–e (Fig. 3A). For R. typhi, we previously characterized the N-terminal secretion signals for three of these RvhB6 proteins, though all five proteins contain predicted signal sequences (Ammerman, Rahman and Azad 2008). While all five proteins are strongly conserved across Rickettsia genomes (Gillespie et al.2009), each RvhB6 analog is highly divergent from one another (Fig. 3A, shading across schemas). For TMS region predictions, placement of the conserved Trp residue within the cytoplasm resulted in highly divergent IM topologies for each RvhB6 protein (Fig. 3B). Nearly every TMS region occurs within the VirB6/TrbL domain, suggesting large regions of the ENS and ECS for each protein are not IM-associated. A phylogeny estimation of the VirB6/TrbL domains from all five RvhB6 proteins indicates that an ancestral duplication gave rise to RvhB6a/b and RvhB6c/d/e progenitors, with subsequent duplications within each lineage yielding the five modern proteins (Fig. 3C). The ENS and VirB6/TrbL domains of all five proteins are rich in Cys residues, which may serve as scaffolds for cross-linking these domains at the IM/periplasm (Fig. 3D). Curiously, the large ECS of both RvhB6a and RvhB6e lack Cys residues, indicating that these domains may associate differently at the IMC or localize elsewhere.
How five divergent VirB6-like proteins collectively assemble at the IMC to facilitate substrate transfer is currently unknown. There are 24 copies of VirB6 within the R388 P-T4SS VirB3-VirB10 complex (Low et al.2014); assuming all five RvhB6 proteins concurrently assemble at the rvh IMC, five copies of each protein would yield a copy number (n = 25) similar to the stoichiometry of VirB6 within the R388 P-T4SS. In addition, the assortment of large ENS and ECS from the various RvhB6 proteins would almost certainly form an additional structure unknown in other P-T4SSs. Perhaps, this additional structure is necessary to accommodate the duplicate components (RvhB9,8,4) that might result in a bulkier IMC relative to other P-T4SSs. Alternatively, the ENS and ECS of each RvhB6 protein may recognize specific sets of substrates and facilitate their delivery to the IMC.
Remarkably, we previously identified RvhB6a on the surface of R. typhi cells, though the exact portion of the protein that was exposed remains unknown (Sears et al.2012). As the ECS of RvhB6a contains three RRs characteristic of surface antigens (Fig. 3E), it is likely that this is the region of RvhB6a that is surface exposed. These RRs (shown only for Typhus Group rickettsiae in Fig. 3E) are very divergent across species and strains of Rickettsia (data not shown). In particular, the entire RR-3a is absent in several pathogenic species of spotted fever group Rickettsia, such as R. conorii and R. rickettsii. This indicates the possibility of a variable surface structure associated with the rvh T4SS. Such a structure may have evolved to accommodate the lack of a canonical P-T4SS pilus, facilitating contact with host membranes prior to invasion. Whether the RvhB6a ECS is exposed via elongation to the rickettsial surface, or by cleavage and secretion to the rickettsial surface, remains to be determined.
Rickettsia VirB10 proteins contain insertions near the T4SS pore
Along with VirB7 and VirB9, 14 copies of VirB10 are present within P-T4SSs (Chandran et al.2009; Fronzes et al.2009; Rivera-Calzada et al.2013). Specifically, VirB7 and the NTDs of VirB9 and VirB10 comprise a heterotrimeric subunit of the tetradecameric CC (Fig. 4A). The NTD of VirB10 extends through the periplasm (and seemingly the IMC), traversing the IM and forming contacts with the IM-embedded T4CP (VirD4) (de Paz et al.2005; Jakubowski et al.2009; Cascales et al.2013; Segura et al.2013). Given this topology, VirB10 is thought to become activated by VirD4 (and VirB11) ATP binding and hydrolysis, undergoing a structural transition that enlarges the T4SS channel to support the translocation of substrates (Cascales and Christie 2004a; Garza and Christie 2013). This conformational switch is seemingly dependent on a conserved Gly near the base of the CC pore (Banta et al.2011). Extending from this region is an α-helical domain, termed the antenna projection (AP), which forms a cap structure atop the CC (Fig. 4A, red shading). Crystal structures of H. pylori ComB10 and E. coli TraF (VirB10) differ in the region encompassing the CC pore (Fig. 4B), probably a consequence of the former being crystallized alone (Terradot et al.2005) versus the latter being crystallized together with TraO (VirB9) and TraN (VirB7) (Chandran et al.2009). Comparison of the region encompassing the CC pore across diverse VirB10 family proteins reveals two strongly conserved regions flanking highly variable sequences (Fig. 4C). As per the TraF structure (Chandran et al.2009) and gating defects resulting from mutagenesis of the conserved Gly (Banta et al.2011), it is probable these conserved regions form the base of the CC pore, with the variable sequences comprising surface-exposed APs that are unique to each protein.
Selecting two conserved regions (Fig. 4C, red boxes) flanking the predicted APs across 2408 VirB10 family proteins from five classes of Proteobacteria (excluding Rickettsiales), we determined that the APs are generally conserved in sequence length (average length = 57 residues) despite a lack of sequence conservation (Fig. 4D). Remarkably, the average length of this same region across 115 Rickettsiales VirB10 family proteins is significantly larger (127 residues), and when broken down across major rickettsial lineages, indicates only two groups (Rickettsiaceae and Wolbachia species) contain exceedingly large insertions within the RvhB10 AP (Fig. 4E). Analysis of these insertion-containing RvhB10 proteins from four divergent species indicates minimal sequence conservation, though multiple α-helices are predicted within each protein (Fig. 4F). For species within the genera Rickettsia, Orientia and Wolbachia, these insertions within the RvhB10 AP are highly conserved (data not shown). Thus, if the insertion was present in the ancestral RvhB10 protein, all but two lineages retained it, albeit with substantial diversification occurring relative to the remaining regions of the RvhB10 architecture.
The architecture of the OM CC is known to be variable in at least two other P-T4SSs. For the vir T4SS of certain Xanthomonas species (Souza et al.2015), VirB7 is fused to a secretin-like N0 domain that adds an additional layer around the outer perimeter of the tetradecameric CC (Souza et al.2011). In the H. pylori cag T4SS, VirB7 (CagT) and VirB10 (CagY) contain additional variable regions that are surface exposed, with the large repeat-laden region of CagY associated with the cag pilus (Fischer 2011; Terradot and Waksman 2011; Frick-Cheng et al.2016). However, to our knowledge, the insertions within certain rickettsial RvhB10 proteins are the first reported expansions of the APs within P-T4SSs. The significance of these large insertions is not known, nor is it clear why only two lineages of Rickettsiales RvhB10 proteins contain them. No functional similarities unique to species of Rickettsiaceae and Wolbachia that would link to a modified OM CC are evident, and analysis of the other Rvh proteins within these species does not reveal other shared features relative to rvh T4SSs from the remaining Rickettsiales. More clarity on these curious insertions awaits structural analysis of the rvh CC and determining if these extended APs are surface exposed, and if so, what targets (e.g. host lipids and proteins, other rickettsial surface proteins) they interact with.
rvh secretion: an emerging role in protein translocation
The ability of the rvh T4SS to translocate a range of different substrate types is currently unknown. For Rickettsia species, we hypothesized that the rvh T4SS may be used to transfer plasmids, provided that the first described Rickettsia plasmids harbored degraded conjugative F-T4SSs (Gillespie et al.2007, 2008). Given that a diverse assortment of plasmids are now known to be widespread across Rickettsia species (Baldridge et al.2008, 2010; Gillespie et al.2012b, 2014), yet complete F-T4SSs are rare (Gillespie et al.2015a), plasmid transfer via the rvh T4SS remains an important area for future investigation. A more universal role for the rvh T4SS in protein translocation also seems likely, given the strong conservation of the rvh T4SS (see above) and a lack of plasmids in other rickettsial genera (Gillespie et al.2012a). For several species of Anaplasma, rvh protein substrates have been proposed based on their secretion by T4SSs in heterologous expression systems: e.g. AnkA of A. phagocytophilum secreted via the vir T4SS of A. tumefaciens (Lin et al.2007), and AnkA plus three other proteins of A. marginale secreted via the dot/icm I-T4SS of L. pneumophila (Lockwood et al.2011). Other studies have identified rvh effectors, such as Ats-1 of A. phagocytophilum (Niu et al.2010) and its homolog from E. chaffeensis (Liu et al.2012), which directly interact with their cognate T4CP (RvhD4) in bacterial two-hybrid (B2H) assays. Recently, we also utilized the B2H assay to demonstrate that an R. typhi secreted protein, RalF, interacts with its cognate RvhD4, and that this interaction is abolished when 45 residues from the RalF C-terminus are deleted (Rennoll-Bankert et al.2015). Thus, it appears that the RalF proximal C-terminus contains a T4SS signal, consistent with other studies demonstrating that C-terminal tails of some T4SS protein substrates contain requisite characteristics that mediate interactions with T4CPs (Atmakuri, Ding and Christie 2003; Nagai et al.2005; Schulein et al.2005; Vergunst et al.2005). The identification of additional rvh protein substrates will be necessary to determine if this mechanism for RvhD4 recognition of effector C-terminal tails is conserved.
Inactivation of RalF via antibody blocking significantly reduces host cell invasion, indicating RalF is expressed early during R. typhi host cell infection (Rennoll-Bankert et al.2015) and suggesting that rvh-mediated translocation of RalF deposits this effector on the rickettsial cell surface. This is consistent with our prior identification of RalF as a component of the R. typhi surface proteome (Sears et al.2012). As a Sec 7-domain-containing protein, R. typhi RalF likely functions as a guanine nucleotide exchange factor of host ADP-ribosylation factors (Arfs). Indeed, we demonstrated that RalF co-localization with Arf6 and the phosphoinositide PI(4,5)P2 at entry foci on the host plasma membrane (PM) was critical for invasion. This led to our hypothesis that RalF activation of Arf6 amplifies PI(4,5)P2 at entry foci, which in turn initiates actin remodeling and other host signaling cascades that collectively facilitate rickettsial invasion (Rennoll-Bankert et al.2015). While these data indicate that RalF is delivered into host cells prior to R. typhi invasion, the mechanism by which the rvh T4SS engages host cells to facilitate RalF translocation remains unknown.
If rvh-mediated translocation of RalF involves contact with host cells, a canonical P-T4SS pilus does not play a role in this process. As discussed above, the rvh T4SS lacks a virB5 gene, which in other P-T4SSs encodes the minor pilin subunit that is essential for pilus formation (Schmidt-Eisenlohr et al.1999; Lai et al.2000). While VirB5 is dispensable for substrate transfer, VirB2 (the major pilin subunit) is required for this process (Cascales and Christie 2004b; Jakubowski et al.2005); thus, it can be anticipated that T4SSs lacking virB5 would only manifest VirB2 in close association with the secretion channel. Indeed, using immunogold EM, we previously observed RvhB2 localized at the cell envelope of bacterial poles but did not label any extracellular structures that would indicate the presence of a pilus (Gillespie et al.2015a). Thus, for Rickettsia species, RvhB2 does not seem to be playing a role in elaborating surface pili. Remarkably, for other Rickettsiales species, RvhB2 is usually encoded by duplicated or proliferated divergent genes (Gillespie et al.2010; Sutten et al.2010; Al-Khedery et al.2012). For these species, it is likely that at least one RvhB2 protein functions in substrate transfer, while other paralogs might instead equip bacteria with surface antigenicity to facilitate host immune evasion. This expansion of genes encoding pilin subunits is reminiscent of the trw P-T4SS of some Bartonella species, wherein proliferated VirB2 (TrwL) and VirB5 (TrwJ) proteins interchangeably assemble into variable surface pili (Seubert et al.2003), with these structures mediating bacterial interactions with host erythrocytes (Vayssier-Taussat et al.2010; Deng et al.2012). The significance of variable RvhB2 proteins and more so why rvhB2 gene number varies so extensively across rickettsial genomes, remains to be determined.
A precise role for RvhB2 (or other Rvh proteins) in facilitating effector delivery into host cells (i.e. away from the bacterial surface) is in need of investigation, considering that this process is a prime drug target. We previously hypothesized that an rvh pilus would be unnecessary for rickettsiae, as direct translocation of substrates into host cells occurs as a consequence of an obligate intracellular lifestyle (Gillespie et al.2010). However, it is timely to reconsider how rickettsiae might translocate rvh effectors across host membranes to ensure their effective delivery to host cell targets. For effectors that facilitate host invasion (i.e. induction of phagocytosis by RalF), translocation across the host PM might require a structure to ‘inject’ these effectors. For Rickettsia species, secretion during early infection of effectors that may be rvh substrates (e.g. patatin phospholipases (Rahman et al.2010, 2013)) probably facilitates phagosomal lysis, mediating escape to the host cytosol. If these effectors traverse the phagosomal membrane, they may require a structure to facilitate translocation. Once in the host cytosol, secretion of rvh effectors during late infection probably does not require a surface elaboration of the rvh channel. For vacuolar rickettsial species (e.g. Anaplasmataceae), a host membrane always presents a barrier to rvh effector translocation. As effectors from A. phagocytophilum are known to localize to the nucleus (Garcia-Garcia et al.2009) and mitochondria (Niu et al.2010), translocation across the vacuolar membrane is critical for positioning these effectors in the host cytosol, wherein their inherent import signals mediate their effective delivery to host organelles. How these rvh effectors traverse the vacuolar membrane is unknown, though it is likely that a structure, possibly an elaboration of the rvh channel, exists to facilitate translocation. Since RvhB6-2 of E. chaffeensis is entirely secreted from the bacterial surface yet contained within the vacuole, it is tempting to speculate that extracellular RvhB6 proteins may form channels across the vacuolar membrane to facilitate effector delivery to the host cytoplasm.
Despite the lack of a canonical P-T4SS pilus, it is possible that an atypical surface elaboration of the rvh channel facilitates effector translocation across host membranes. For Rickettsia species, two Rvh proteins (RvhB6 and RvhB10) provide the best candidates for formation of such a surface structure. Based on its characteristics described above, RvhB6a-ECS may directly engage translocated effectors upon their exit from the rvh CC and mediate their secretion across host cell membranes. The ENS and ECS of other RvhB6 proteins, if surface localized, may also participate in such a process. For RvhB10, insertions within the AP may form a slight elaboration that is able to contact, and possibly penetrate, host cell membranes to facilitate effector delivery. Whether RvhB2 is able to make contacts with the RvhB6 surface-exposed regions and/or the APs of RvhB10 (or both) remains to be determined. This region encompassing the rvh OM CC pore is potentially the most variable feature across different rvh T4SSs; elucidation of its structure and function will most certainly reveal system-specific rvh surface elaborations that mediate host cell contact, aid host immune evasion via surface antigenicity and facilitate substrate translocation across host membranes.
CONCLUSION AND OUTLOOK
Rickettsial genomes are honed for parasitism, with all known Rickettsiales species relying on host metabolites to compensate for incomplete metabolic pathways (Andersson et al.1998). Many pseudogenes that scar modern rickettsial genomes reflect the end products of reductive evolution. We balk at the notion that the RvhB9,8,4-II paralogs are pseudogenes, since the peculiar characteristics of these proteins are conserved in all rvh T4SSs (Gillespie et al.2010, 2015b). Thus, the odd assortment of rvh T4SS genes indicates a complex secretion machine that has been retained despite intense evolutionary pressure for genome minimization (Gillespie et al.2009). Neofunctionalization of rvhB9,8,4 gene duplications has yielded three families containing highly divergent paralogs, all seemingly inextricably tied to rvh structure and function. If the recently solved structure of R. typhi RvhB8-II is any indication (Gillespie et al.2015b), we can anticipate that RvhB9,8,4-II proteins equip the rvh T4SS with an architecture unknown to other P-T4SSs. The incorporation of multiple divergent RvhB6 proteins, each with their own unique TrbL/VirB6 domain and bizarre ECS and ENS, further increases the complexity of potential rvh T4SS assemblies.
A superficial assessment of the composition of the rvh T4SS might lead some to consider its gene family expansion as a redundant feature. However, a careful appreciation of the complexity underpinning rvh gene family expansion indicates a system characteristic of a Rube Goldberg machine. In other words, the assortment of conserved duplications (RvhB9,8,4) and proliferations (RvhB6) might be regarded as a result of evolutionary overengineering, with the rvh T4SS performing a simple task (secretion) in a seemingly complicated fashion. Characterization of any single protein within the expanded rvh families will tell us little about the overall function of the rvh machine; therefore, the challenge moving forward will be to determine how these various parts come together to operate the rvh machine. Despite the difficulty in utilizing conventional tools for genetic characterization of obligate intracellular bacteria, future mutagenesis studies will be paramount for understanding the significance of multiple RvhB9, RvhB8, RvhB4, RvhB6 proteins to rvh structure and function. Ultimately, an rvh mutant (if obtainable) will be of substantial importance for definitively characterizing rvh substrates, moving Rickettsiology away from relying on surrogate expression systems for tenuously identifying these substrates.
In conjunction with this work, it will be necessary to characterize rvh components using tools and approaches that best model the complex rickettsial intracellular lifecycle, which often involves multiple eukaryotic hosts (e.g. arthropod and vertebrate). This will help clarify whether one static T4SS is the operational architecture, or if rvh assembly is dynamic throughout the rickettsial lifecycle. While obtained mutants may only allow the characterization of substrates under specific conditions, nonetheless, our analysis provided here lays the groundwork for dissecting out the intricate nature of the rvh T4SS. A desirable outcome of this work will be the identification of components of the atypical rvh architecture that are amenable to drug targeting, provided that inhibitory molecules of T4SSs from several bacterial species have been identified (Fernandez-Lopez et al.2005; Hilleringmann et al.2006; Paschos et al.2011; Smith et al.2012).
This work was supported by grants awarded to AFA and ), JJG and MSR ), and PJM (contract nos. and ). M.L.G., K.E.R-B. and S.S.L. were supported in part by the and from the NIAID. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.
Conflict of interest. None declared.