Which Way In? The RalF Arf-GEF Orchestrates Host Cell Invasion

Summary

Phylogenomics analysis indicates divergent mechanisms for host cell invasion across diverse species of obligate intracellular Rickettsia. For instance, only some Rickettsia species carry RalF, the rare bacterial Arf-GEF effector utilized by Legionella pneumophila to facilitate fusion of ER-derived membranes with its host-derived vacuole. For R. prowazekii (Typhus Group, TG), prior in vitro studies suggested the Arf-GEF activity of RalF, which is absent from Spotted Fever Group species, might be spatially regulated at the host plasma membrane. Herein, we demonstrate RalF of R. typhi (TG) and R. felis (Transitional Group) localizes to the host plasma membrane, yet R. bellii (Ancestral Group) RalF shows perinuclear localization reminiscent of RalF-mediated recruitment of Arf1 by L. pneumophila to its vacuole. For R. typhi, RalF expression occurs early during infection, with RalF inactivation significantly reducing host cell invasion. Furthermore, RalF co-localization with Arf6 and the phosphoinositide PI(4,5)P2 at the host plasma membrane was determined to be critical for R. typhi invasion. Thus, our work illustrates that different intracellular lifestyles across species of Rickettsia and Legionella have driven divergent roles for RalF during host cell infection. Collectively, we identify lineage-specific Arf-GEF utilization across diverse rickettsial species, previously unappreciated mechanisms for host cell invasion and infection.

Introduction

Bacteria invading eukaryotic cells employ diverse strategies for successful entry, intracellular colonization and intercellular spread [1,2]. Whether facultative or obligate, intracellular species must either modify the phagocytic vacuole for survival or lyse the phagosome and live freely within the host cytoplasm (or invade other cellular organelles) [3–6]. Either strategy is delicately underpinned by bacterial secretion of effectors, which have a myriad of characterized functions: e.g., engaging host signaling pathways, rearranging the host cytoskeleton, polymerizing host actin, subverting host vesicular traffic, etc. [7–9]. It is well established that divergent effectors from distantly-related intracellular species can operate in similar processes [10]; e.g., actin nucleators from species of Shigella, Listeria and Rickettsia [11,12] and phospholipases from species of Pseudomonas and Legionella [13,14]. Conversely, the ability for highly similar effectors from distantly-related species to function differently in host cells is a phenomenon that is poorly known, probably reflective of effector repertoires being highly specific to bacterial genera [15–17].

Species of Rickettsia (Alphaproteobacteria: Rickettsiales) are Gram-negative obligate intracellular parasites of a wide range of eukaryotic species [18]. Rickettsiae bind to host cells and induce phagocytosis [19,20], with internalized bacteria released into the cytosol upon rapid escape from the phagocytic vacuole. Bacteria spread intercellularly upon death and lysis of host cells, though some species move intercellularly prior to host cell lysis via host actin polymerization [21–23]. Several surface proteins characterized for adhesion and/or entry of host cells (Sca5, Adr1, Adr2) [24–28] and activation of cytoskeletal vinculin (Sca4) [29] are conserved across sequenced Rickettsia genomes, as are several enzymes implicated in phagosomal lysis (TlyC, PLD, Pat1) [30–33]. In contrast, other characterized adhesins (Sca0, Sca1, Sca2) [34–38], proteins involved in Arp2/3-dependent (RickA) [39,40] and -independent (Sca2) [41,42] host actin polymerization, and another phospholipase (Pat2) [43,44] are sporadically encoded across rickettsial lineages. This suggests that, despite superficially similar infection strategies, diverse Rickettsia species employ distinct molecular mechanisms for successful colonization of host cells [45].

One such protein that is differentially encoded across Rickettsia genomes is a highly similar counterpart to the RalF protein of Legionella spp. Collectively, these proteins contain a Sec7-domain, which in eukaryotes functions as a guanine nucleotide exchange factor (GEF) of ADP-ribosylation factors (Arfs) [46]. Remarkably, bacterial Sec7-domain containing proteins are unknown from other bacteria [47]. Legionella RalF (RalF_L) is a secreted effector, with its proximal C-terminal sequence mediating secretion through the dot/icm type IV secretion system (T4SS) [48]. RalF_L activates and recruits host Arf1 to the Legionella-containing vacuole (LCV), which is a modification of the phagosome [49]. The structure of RalF_L contains two distinct domains: an N-terminal Sec7 domain (S7D) and a C-terminal Sec7-capping domain (SCD) that regulates active site access to Arfs [50]. The S7D and SCD across RalF_L and Rickettsia RalF (RalF_R) share ~45% aa identity, though an extended variable region flanks the SCD of RalF_R proteins at the C-terminus [51].

A comparative study of RalF_L and RalF_R determined similar GEF activities for both proteins, yet divergent subcellular localization patterns driven primarily by intrinsic characteristics of the SCD [52]. The RalF_L SCD positions the protein at the endoplasmic reticulum for interception of host secretory vesicles, while the RalF_R SCD targets the protein to the host plasma membrane. Furthermore, a proline-rich region within the extended variable region of RalF_R interacts with components of the host actin cytoskeleton. Subsequently, membrane sensor regions were identified within the SCDs of RalF_L and RalF_R, with differential enrichments in aromatic and positively charged residues determining divergent lipid substrates that regulate Arf-GEF activities [53]. Collectively, these studies suggest that these distinguishing features (divergent SCD sensor regions, RalF_R-specific cytoskeletal-binding domain) mediate the spatial regulation of RalF activity in two diverse intracellular species with very different lifestyles.

Despite tremendous insight on the possible function of RalF_R during rickettsial host cell infection, important questions are left unanswered. As previous studies were performed in vitro [52,53], it still remains unknown if those Rickettsia species that carry ralF genes actually express RalF_R during infection, and if so, at what time point. Furthermore, as GEFs confer the spatial regulation of different Arf classes at discrete cellular locales [54–57], the Arf(s) specificity of RalF_R needs to be determined in light of the different subcellular localization of the protein compared to Legionella spp. Our work presented here addresses these unknowns by demonstrating RalF expression by R. typhi early during host cell invasion. Across several Rickettsia species, we identified the domain requirements for positioning RalF at host membranes, and for R. typhi, determined that RalF co-localization with Arf6 and PI(4,5)P₂ at entry foci was critical for invasion. Altogether, our work identifies Arf-GEF utilization as a lineage-specific invasion mechanism, illuminating the variable strategies that drive Rickettsia infection of host cells.

Results

RalF_Rt interacts with the rvh T4SS coupling protein (RvhD4) via its proximal C-terminal sequence and is secreted during host cell infection

As predicted Arf-GEFs, we anticipated RalF_R proteins to be secreted extracellularly into the host cell. Prior to invasion, L. pneumophila utilizes its dot/icm I-T4SS to translocate RalF_L into host cells [48]. Like RalF_L, R. typhi RalF (RalF_Rt) lacks a predicted N-terminal Sec secretion signal [58], trans-membrane spanning regions [59] and a β-barrel structure [60], suggesting its secretion via a Sec-independent pathway, possibly the Rickettsiales vir homolog (rvh) T4SS [61]. Accordingly, in order to determine if RalF_Rt interacts with the rvh T4SS, we performed a bacterial two-hybrid assay with full length RalF_Rt (RalF_RtFL) and RvhD4, the rvh T4SS coupling protein. T4SS coupling proteins (VirD4 family) are ATPases that function as “gatekeepers” to regulate substrate entry into the T4SS channel [62,63]. Co-transformation of bait (encoding RvhD4) and prey (encoding RalF_RtFL) plasmids in BacterioMatch II reporter electrocompetent cells resulted in bacterial growth on selective media (Fig 1A), indicating RalF_RtFL and RvhD4 interact, and thus implicating RalF_Rt as an rvh T4SS effector.

Secretion of RalF_L is dependent on hydrophobic residues within its C-terminal tail [48], while many other T4SS protein substrates have enrichments of positively charged residues at their C-termini that are important for secretion [64–66]. Accordingly, we evaluated RalF_Rt for the presence of a T4SS signal sequence (T4S) within its C-terminus. A T4S RalF_Rt truncation (RalF_RtΔT4S) was generated and tested for its ability to bind RvhD4 via the bacterial two-hybrid assay (Fig 1A). The percent growth of colony forming units (CFUs) of reporter cells harboring recombinant plasmids on dual selective screening medium was calculated relative to percent growth of CFUs obtained on non-selective His dropout medium by drop plate method for counting. An approximately 77% decrease in CFUs on dual selective media was observed with RalF_RtΔT4S compared to RalF_RtFL, indicating that the RalF C-terminus is important for interacting with RvhD4 (Fig 1B).

The ATPase activity of T4SS coupling proteins is essential for substrate translocation [67]. To confirm functionality of RvhD4, recombinant RvhD4 was assayed for ATPase activity. RvhD4 was found to release inorganic phosphate (Pi) from ATP in a concentration dependent manner compared to a rickettsial protein that lacks predicted ATPase activity (RT0600) (Fig 1C and 1D). This indicates Rickettsia RvhD4 is a functional ATPase that likely regulates protein secretion through the rvh T4SS.

The interaction of RalF_Rt with machinery of the rvh T4SS implies extracellular secretion. Our prior report that characterized the R. typhi surface proteome demonstrated that RalF_Rt is expressed and surface exposed [68]. To further confirm RalF_Rt secretion, purified R. typhi was treated with proteinase K, on the premise that surface exposed protein would be degraded with proteinase K treatment while subsurface proteins would be protected. Protease treatment caused a dose-dependent degradation of RalF_Rt with respect to the R. typhi cytoplasmic control protein, elongation factor Ts (EF-Ts, Fig 1E).

RalF_Rt is expressed early during infection and is required for R. typhi invasion of host cells

To determine when RalF_Rt is expressed during R. typhi infection, a polyclonal antibody against RalF_Rt was generated, qualified (S1 Fig) and used for immunofluorescence assays (Fig 1F). During early infection of host cells (10 min), RalF_Rt expression is high and diminishes as internalization progresses (30 min). Given RalF_Rt expression during early infection, we assessed its role during R. typhi invasion of host cells. When R. typhi was pre-treated with the anti-RalF_Rt polyclonal antibody, the average number of R. typhi per host cell decreased by 52% from an average of 10 to 4.8 bacteria per host cell (Fig 1G), indicating a role for RalF during host cell invasion. To rule out possible steric hindrance induced by the Fc portion of the anti-RalF_Rt antibody inhibiting rickettsial-host cell interactions that promote entry, R. typhi was pre-absorbed with anti-RalF_Rt Fab fragments. The average number of R. typhi per host cell was significantly decreased by 45% from 10 to 5.5 bacteria per host cell (Fig 1G) further confirming the involvement of RalF in host cell invasion.

RalF_R is divergent from RalF_L within the SCD lipid sensor region and also contains a C-terminal extension that is highly variable across Rickettsia homologs

Utilizing over 60 Rickettsia genome sequences, phylogenomics analyses were carried out to provide further insight on the role of RalF in rickettsial biology and pathogenesis. While a key factor in R. typhi infection of host cells, RalF-mediated invasion is not a strategy employed by all Rickettsia species, as evident by ralF pseudogenization in all species of Spotted Fever Group (SFG) rickettsiae, as well as two other species (R. canadensis and R. helvetica) [45]. Still, the remaining species, including R. bellii and all species within the Typhus Group (TG) and Transitional Group (TRG) rickettsiae, contain genes encoding RalF_R proteins that are highly conserved within the S7D and SCD as compared to RalF_L (Fig 2). Specifically, and in agreement with previous studies [52,53], all RalF_L and RalF_R proteins contain a highly conserved Sec7 active site within the S7D (S2A and S2B Fig), with RalF_R proteins having an enrichment of positively charged residues in the lipid sensor region of the SCD relative to RalF_L proteins (S3C Fig). Thus, based on these characteristics, all RalF_R proteins are predicted to spatially regulate their Arf-GEF activities at the host plasma membrane, where concentrated negatively charged phospholipids attract the RalF_R SCD [53].

Relative to RalF_L, the major distinguishing factor of RalF_R proteins is the presence of a variable sequence with Pro-rich region (VPR) within the C-terminal domain (Fig 2). Pro-rich regions are a common characteristic of proteins that target the actin cytoskeleton [70], and are typically present in Arf-GEFs recruited to cytoskeletal/plasma membrane junctions [71,72]. Across RalF_R proteins, the VPR is flanked by the SCD and T4S and is extraordinarily variable in sequence length and number of Pro residues across RalF_R proteins (Fig 2B). Remarkably, many SFG rickettsiae species, e.g. R. montanensis, contain putative ORFs encoding complete VPRs. Alignment of these ORFs with VPRs from full-length RalF_R proteins illustrates that a gene encoding RalF_R was present in the Rickettsia ancestor, with pseudogenization purging the complete Arf-GEF from most Rickettsia genomes (S4A Fig). This conclusion is supported by genome synteny analysis across ralF_R loci, which indicates a conserved position for ralF flanking the maeB gene in all sequenced Rickettsia genomes (S5A Fig). Thus, RalF_L and RalF_R proteins diversified early upon their establishment in ancestral Legionella and Rickettsia genomes, with the retention of VPRs within full-length RalF_R proteins implying an important function.

The RalF_R SCD regulates membrane localization while the VPR targets the host cytoskeleton

In light of the variability across the VPR of RalF_R proteins, we determined the C-terminal domain (CTD) requirements for subcellular localization across RalF_R proteins from several species (R. typhi, R. felis, R. montanensis and R. bellii) (Fig 3). The SCD- and VPR-mediated targeting to the host plasma membrane and actin cytoskeleton, respectively, for RalF of R. prowazekii (RalF_Rp) was used as a reference [52,53]. R. typhi and R. felis full-length RalF (RalF_FL) proteins primarily had diffuse staining within the cytoplasm with some plasma membrane localization. However, RalF_CTD (SCD-VPR-T4S) localized strongly to the plasma membrane and disrupted actin stress fibers. Additionally, R. typhi and R. felis RalF_CTD induced membrane ruffling and microvilli-like protrusions suggesting that the CTD plays a role in cytoskeletal rearrangements, similar to the known Arf6-GEF, EFA6 [71,72]. Furthermore, RalF_VPR (VPR-T4S), as well as the full-length VPR-containing ORF of R. montanensis, did not uniformly localize to the host plasma membrane, but instead were found strongly associated with intact actin stress fibers. Collectively, these results indicate that R. typhi and R. felis RalF proteins are similar to RalF_Rp, with both the SCD and VPR required to spatially regulate Arf-GEF activities at plasma membrane/actin cytoskeletal junctions.

Remarkably, RalF_FL and RalF_CTD of R. bellii did not target the plasma membrane, yet instead showed perinuclear localization reminiscent of ectopically expressed RalF_L. As R. bellii RalF_VPR associated with intact actin stress fibers, these data collectively indicate that the SCD alone is sufficient to localize RalF_Rb to the host cytoplasm. Visualization of the SCD sequence alignment across all RalF_L and RalF_R proteins revealed that RalF_Rb lacks three separate insertions within the SCD that are conserved in all other RalF_R proteins (S3B Fig). Thus, from a structural perspective, the SCD of RalF_Rb is more similar to RalF_L proteins than RalF_R proteins, which could explain why the SCD of R. bellii localizes to the perinuclear region of the cytoplasm. This is consistent with R. bellii sharing more genomic attributes with Legionella spp. [73], as well as being able to grow in various amoeba species unlike most other Rickettsia spp. (see Discussion).

RalF membrane localization was further confirmed using two independent approaches. First, membrane fractionation of HeLa cells transfected with RalF-expressing plasmids revealed that all RalF_CTD proteins were predominately enriched in the membrane fraction, with RalF_FL and RalF_VPR proteins having less membrane enrichment (Fig 3 and S6 Fig). Second, RalF transfected cells were stained with Alexa Fluor 594 wheat germ agglutinin to detect plasma membrane or probed with anti-PDI (endoplasmic reticulum) or anti-GM130 (Golgi apparatus) antibodies (Fig 4 and S7 Fig), with the Pearson’s correlation coefficients calculated to measure co-localization with the respective membrane markers (S8 Fig). RalF_CTD of R. typhi, R. felis, R. montanensis indicate localization to the plasma membrane, while R. bellii RalF_CTD localized to the endoplasmic reticulum membrane recapitulating results observed with labeling host cell actin (Fig 3). Collectively, these data demonstrate the affinities of RalF_R proteins for host membranes, identifying the SCD as the major determinant for membrane localization, combined with the targeting of actin cytoskeleton by the VPR (Table 1).

Finally, for R. typhi, we monitored the subcellular localization of its RalF_CTD construct lacking the T4S (RalF_RtCTDΔT4S). We observed indistinguishable localization patterns between RalF_RtCTD and RalF_RtCTDΔT4S (S4C Fig), suggesting that the T4S has no effect on localization or stress fiber disruption. In conjunction with results above (Fig 1A and 1B), these data bolster the role of the T4S of RalF_R proteins as an rvh T4SS translocation signal.

Plasma membrane localization of RalF_Rt is dependent on PI(4,5)P₂

Previous studies showed the SCD of RalF_Rp has affinities for negatively-charged phospholipids; i.e., phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂) and phosphatidylinositol 3,4,5-trisphosphate (PI(3,4,5)P₃) [52,53]. Given the enrichment of PI(4,5)P₂ at host membranes during early stages of phagocytosis [74], we evaluated the role of PI(4,5)P₂ in RalF_R localization. As a baseline, we utilized the standard phospholipase C (PLC)-mediated catalyzation of PI(4,5)P₂ within the IP3/DAG pathway of host cells [75]. Specifically, in the presence of ionomycin and Ca²⁺, PI(4,5)P₂ is hydrolyzed to inositol 1,4,5-trisphosphate and diacylglycerol via PLC isozymes that regularly deplete the plasma membrane of PI(4,5)P₂ following its role as a substrate in many signaling pathways [76]. To test PI(4,5)P₂-dependent localization of RalF_Rt to the plasma membrane, HeLa cells ectopically expressing RalF_RtCTD were treated with ionomycin and Ca²⁺, with the distribution pattern of RalF_RtCTD monitored by immunofluorescence. With ionomycin and Ca²⁺ treatment, RalF_RtCTD becomes cytosolic compared to plasma membrane localization in the presence of ionomycin alone (Fig 5A). Upon treatment with EGTA, which chelates Ca²⁺, PI(4,5)P₂ accumulates and RalF_RtCTD returns to the plasma membrane. HeLa cells expressing GFP-C1-PLCδ-PH, a biosensor of PI(4,5)P₂, were used as a positive control to demonstrate the hydrolysis of PI(4,5)P₂ in the presence of ionomycin and Ca²⁺. Additionally, RalF_RbCTD subcellular localization was shown to be unaffected by PI(4,5)P₂ hydrolysis, consistent with its perinuclear distribution in host cells. Collectively, these data indicate that PI(4,5)P₂ enrichment at the host plasma membrane is a requirement for efficient recruitment of RalF_Rt, and probably also RalF_Rf, given its similar subcellular localization pattern.

PI(4,5)P₂ recruitment is critical for R. typhi invasion of host cells

Phosphatidylinositols enriched at the host plasma membrane often play a critical role in bacterial internalization [74]. Given that RalF_Rt is expressed early (Fig 1F) and required (Fig 1G) for host invasion, and its localization to the host plasma membrane requires PI(4,5)P₂ enrichment (Fig 5A), we sought to determine if PI(4,5)P₂ is recruited by RalF_Rt during R. typhi infection. PI(4,5)P₂ localization during R. typhi invasion was analyzed using immunofluorescence microscopy with GFP-C1-PLCδ-PH as a biosensor of PI(4,5)P₂ localization. During early infection (i.e. 5 and 10 min post infection), PI(4,5)P₂ was highly localized to pseudopodia at the R. typhi entry foci (Fig 5B). As internalization progressed, R. typhi was surrounded by a vacuole with diminished PI(4,5)P₂ localization. Once R. typhi detached from the membrane, it was no longer associated with PI(4,5)P₂. Furthermore, detection of RalF_Rt during the infection process revealed co-localization of PI(4,5)P₂ and RalF_Rt during early infection (Fig 5C). In agreement with RalF_Rt early expression, which diminished at later stages of infection (Fig 1F), PI(4,5)P₂ recruitment decreased as infection progressed.

Finally, we evaluated whether or not the recruitment of PI(4,5)P₂ to R. typhi entry foci is critical for R. typhi infection. Pretreatment of HeLa cells with ionomycin and Ca²⁺ to deplete PI(4,5)P₂ from the membrane prior to infection resulted in a significant decrease in R. typhi infection (Fig 5D), strengthening the evidence that PI(4,5)P₂ is a target molecule involved in RalF_Rt-associated host cell invasion.

RalF_Rt interacts with Arf6 but not Arf5

The Arf-GEF activity of RalF_L is activated upon membrane binding, with Arf1 the preferred substrate [49,52]. Arf1 is predominantly localized to the Golgi apparatus and plays a role in intra-Golgi transport [77]. Given the association of RalF_Rt with the plasma membrane, we hypothesized that it might instead recruit Arf6, which is predominantly localized to the plasma membrane where it is involved in endocytosis, endosomal recycling and exocytosis of secretory granules [78–81]. Using immunofluorescence, RalF_RtFL was found to recruit Arf6 but neither Arf5 (Fig 6A) nor Arf1 (S9 Fig) to the plasma membrane. Arf5 localizes primarily to the endoplasmic reticulum/Golgi intermediate compartment and the cis-Golgi, where it regulates endoplasmic reticulum to Golgi transport; therefore, Arf5 was used as a negative control [82]. Interestingly, RalF_RbFL was similarly found to co-localize with Arf6 but not Arf5 or Arf1 in the perinuclear space.

To further confirm a RalF_RtFL and Arf6 interaction, a protein pull-down assay was performed. Using rHis-RalF_RtFL as bait and mRFP-Arf5 or mRFP-Arf6 as the prey, we confirmed that RalF_RtFL interacted with Arf6 and not Arf5 (Fig 6B).

Arf6 recruitment is critical for R. typhi invasion of host cells

Activation of Arf6 at the plasma membrane drives the recruitment of phospholipase D and phosphatidylinositol 4-phosphate 5-kinase (PIP5K), which ultimately results in actin remodeling [83,84]. To determine if Arf6 is recruited during R. typhi entry, we used immunofluorescence microscopy to monitor Arf6 localization. As early as 10 min post infection, Arf6 was recruited to the plasma membrane at R. typhi entry foci, while Arf5 remained cytoplasmic (Fig 6C). Given that RalF_Rt localizes with Arf6 at plasma membrane (Fig 6A) and recruits Arf6 at the R. typhi entry foci (Fig 6C), we predicted that knockdown of Arf6 would decrease R. typhi infection. Indeed, siRNA-mediated Arf6 knockdown significantly decreased the number of R. typhi per cell, while Arf5 knockdown had no significant effect on R. typhi infection(Fig 6D and 6E). These results indicate that RalF_Rt recruits Arf6 at the plasma membrane during early infection, with spatially regulated Arf-GEF activity required for host cell invasion.

Discussion

Bacteria invading eukaryotic cells employ diverse strategies to subvert the host cellular actin cytoskeleton, allowing for internalization into normally non-phagocytic host cells [85]. For some bacterial species, surface proteins bind host cell receptors and trigger an “outside-in” signaling cascade, which induces cytoskeletal rearrangements and recruits the endocytic machinery to entry foci [86]. Such receptor-mediated induction of bacterial uptake is a strategy employed by Listeria monocytogenes, which utilizes two adhesins (InlA and InlB) to bind host proteins (E-cadherin, receptor gC1qR, proteoglycans) and activate the tyrosine kinase receptor Met [87]. Invasive species of Yersinia also employ two adhesins (invasin and YadA) to bind a subset of β1-integrin host receptors, facilitating invasion that is dependent on signaling from the Rho GTPase Rac1 and activation of the actin nucleating complex Arp2/3 [88,89]. Alternatively, other bacterial species translocate effectors into host cells to initiate actin remodeling and facilitate bacterial uptake. For example, Salmonella typhimurium utilizes its type III secretion system to inject host cells with the effector SopE, which stimulates GDP/GTP nucleotide exchange on Rho GTPases Rac1 and Cdc42, resulting in membrane ruffling and actin cytoskeleton rearrangement [90]. While a receptor-mediated process has been previously characterized for R. conorii invasion of mammalian cells (discussed below), to date no secreted effectors for any Rickettsia species have been characterized for their role in inducing uptake into host cells.

While most species of the order Rickettsiales encode the rvh T4SS [91], effectors have only been identified for some species of the family Anaplasmataceae. For Anaplasma phagocytophilum, rvh effectors are translocated to the mitochondria (Ats-1) and nucleus (AnkA) to inhibit etoposide-induced apoptosis and down-regulate host defense genes, respectively [92–94]. AM185, AM470, AM705 (AnkA), and AM1141 have been identified as putative rvh T4SS effectors of Anaplasma marginale using a heterologous T4SS (L. pneumophila dot/icm), yet none have been characterized for their roles in invasion [95]. Ehrlichia chaffeensis utilizes the rvh T4SS to translocate the effector ECH0825 into host mitochondria, resulting in inhibition of Bax-induced apoptosis [96]. Herein, we identified RalF as the first rvh T4SS effector for species in the family Rickettsiaceae. We provide evidence that R. typhi RalF interacts with RvhD4, the rvh T4SS coupling protein that presumably recognizes effectors and regulates their translocation similar to VirD4 proteins of other P-type T4SSs. Treatment of purified R. typhi with proteinase K degraded surface exposed RalF, providing further evidence that RalF is secreted. Furthermore, using immunofluorescence we show that RalF_Rt is expressed early during host cell invasion. Because RalF is expressed early during invasion, we hypothesized that RalF is critical for invasion, which was confirmed using antibody pretreatment assays.

Prior studies comparing the subcellular localization of ectopic RalF_L and RalF_Rp determined that, despite strong conservation in the S7D and SCD across these proteins, cryptic signatures within the SCD targeted these proteins to different host membranes [52,53]. RalF_Rp localization to the plasma membrane, mediated by elevated positively charged residues within the lipid sensor of the SCD, was anticipated to be true for other RalF_R proteins, given the strong sequence conservation within the SCD across RalF_R homologs. Furthermore, despite extensive variation within the VPR across RalF_R proteins, the presence of proline-rich regions in all proteins suggested that this region likely encodes a conserved motif that facilitates interaction with the host cytoskeleton, as was shown for RalF_Rp [52,53]. Indeed, our co-localization assays confirmed that, for RalF of R. typhi and R. felis, the SCD mediates interaction with the host plasma membrane, with the VPR facilitating interaction with the host cytoskeleton.

In contrast, the perinuclear localization of RalF of R. bellii, reminiscent of the localization of ectopic expressed RalF_L proteins at the host secretory network, was unexpected. The VPR of RalF_Rb is similar in length to VPRs of RalF proteins from R. felis, R. akari and R. australis, with all of these proteins predicted to encode a coiled-coil motif typical of some eukaryotic Arf-GEFs; e.g., EFA6 [71]. While containing an extraordinary number of Pro residues, the VPR of RalF_Rb nonetheless targets the host cytoskeleton, suggesting that other characteristics of the protein mediate its localization to the cytoplasm. Indeed, in silico analyses revealed three conserved insertions within the SCD of RalF_R proteins that are absent from RalF_Rb. Furthermore, relative to all RalF proteins, the S7D of RalF_Rb contains an odd insertion as well as a slightly less hydrophobic active site (S2 Fig), the significance of which is unknown. It is tempting to speculate that the perinuclear localization of RalF_Rb reflects a unique cytosolic lifestyle of R. bellii, an ancestral lineage with a different genomic repertoire relative to other Rickettsia species [51] and the unique ability to grow and survive in several species of amoeba [73]. As R. bellii has been observed invading nuclei of mammalian cells in vitro [73], RalF may play a role in this process, though other Rickettsia species that lack RalF also are known to invade host cell nuclei [97]. Notwithstanding, the SCD-driven perinuclear localization of RalF_Rb might thus be considered the retention of an ancestral role for RalF_R proteins in targeting host vesicular trafficking, similar to RalF_L proteins. Collectively, our detailed dissection of the domain requirements for subcellular localization strongly implies differential utilization of Arf-GEF activities for those species of Rickettsia that encode RalF.

RalF_Rb aside, the subcellular localization of other RalF_R proteins to the plasma membrane suggested Arf6 might be their preferred host target, given the predominant localization of Arf6 to the plasma membrane [98] and a previous study showing that RalF_Rp can catalyze nucleotide exchange on Arf6 [52]. Arf6 activation by some intracellular pathogens (e.g., species of Salmonella, Yersinia and Chlamydia) is known to induce actin remodeling and mediate bacterial invasion via unique pathways. Salmonella enterica activates Arf6 to recruit the Arf-GEF ARNO, which in turn activates Arf1 to enable WASP family veroprolin homolog (WAVE) regulatory complex-dependent actin assembly [99]. Arf6 activation by species of Yersinia and Chlamydia leads to activated PIP5K, which converts PI(4)P to PI(4,5)P₂ at the plasma membrane [100,101]. As PI(4,5)P₂ enrichment at the host plasma membrane modulates many actin-binding proteins, including α-actinin, talin, vinculin, gelsolin, and the WASP-Arp2/3 complex [102–107], effector-driven accumulation of this phosphatidylinositide can be considered a strategy for induction of phagocytosis.

Given that R. typhi secretes RalF early during host cell invasion, we hypothesized that this Arf-GEF recruits Arf6 to entry foci, precipitating the enrichment of PI(4,5)P₂ at the plasma membrane to facilitate bacterial invasion. Indeed, our in vitro and in vivo results confirm that Arf6 co-localizes with RalF and R. typhi at entry foci. Additionally, during R. typhi infection, PI(4,5)P₂ noticeably accumulated in the membranes of pseudopodia, with a decreased concentration at the base of the phagocytic cup as internalization progressed. Furthermore, the role of PI(4,5)P₂ in bacterial internalization was bolstered by the significant reduction in R. typhi invasion upon PI(4,5)P₂ hydrolysis. Thus, an increase in PI(4,5)P₂ induced by rickettsial RalF activation of Arf6 is predicted to initiate actin remodeling and ultimately facilitate bacterial invasion (Fig 7). Remarkably, this function for RalF_R is markedly different than RalF_L, which is utilized by L. pneumophila to recruit Arf1 to the LCV [49]. Unlike species of Legionella, Rickettsia species do not reside in vacuoles but rather lyse the phagosome and replicate within the host cytoplasm. Thus, despite carrying a similar Arf-GEF that is unknown from any other bacteria, different intracellular lifestyles across species of Rickettsia and Legionella have driven divergent roles for RalF during bacterial infection.

Currently, the predominant knowledge of rickettsia entry and invasion of host cells is based on studies of species from SFG rickettsiae, whereby the surface antigen Sca5 binds host receptor Ku70 to activate a signaling cascade leading to Arp2/3 activation and ultimately actin polymerization, membrane rearrangement and bacterial invasion [26,108,109]. The conservation of Sca5 across all Rickettsia species implies that this receptor-mediated mechanism for entry is likely conserved (Fig 8) [45]. However, depletion of host Rho family GTPases and nucleation-promoting factors that activate Arp2/3 has only a modest effect on rickettsial invasion, suggesting there are other bacterial or host proteins that activate Arp2/3 during infection [108]. Most species of SFG rickettsiae encode an Arp2/3 activating protein, RickA, which could potentially play this role, although its secretion during infection has yet to be demonstrated. Interestingly, genes encoding RickA are absent from species of TG rickettsiae (R. typhi and R. prowazekii); thus, if bacterial Arp2/3 activators are a requirement for invasion, factors other than RickA must be utilized for species of TG rickettsiae. Accordingly, we propose that RalF plays a role in host actin rearrangement and bacterial invasion.

Aside from the Sca5-Ku70 interaction and subsequent downstream signaling cascade, other rickettsial adhesins have been characterized for facilitating host cell invasion (Fig 8). However, the lack of conservation of these adhesins (e.g., Sca0 and Sca2) across diverse Rickettsia species implies the existence of multiple mechanisms for rickettsial host cell invasion [45]. It is also probable that each species likely encodes redundancy for factors that facilitate entry, and that some factors may selectively operate for invasion of specific cells (arthropod versus mammalian) throughout the complex rickettsial lifecycle. Thus, it is probable that lineage-specific factors are employed by different species of Rickettsia to successfully invade and colonize diverse eukaryotic cells. Our identification of lineage-specific Arf-GEF utilization across diverse rickettsial species exemplifies this, and illuminates previously unappreciated mechanisms for host cell invasion and infection.

Materials and Methods

Bacterial strains, cell culture, and infection

Vero76 (African green monkey kidney, ATCC: CRL-1587), HEK293T and HeLa (ATCC: CCL-2) cells were maintained in minimal Dulbecco’s Modified Eagle’s Medium (DMEM with 4.5 gram/liter glucose and 480 L-glutamine; Mediatech, Inc.) supplemented with 10% heat inactivated fetal bovine serum (FBS) at 37°C with 5% CO₂. R. typhi strain Wilmington (ATCC: VR-144) was propagated in Vero76 cells grown in DMEM supplemented with 5% heat inactivated fetal bovine serum at 34°C with 5% CO₂. Rickettsiae were partially purified as previously described [111]. Infections with R. typhi were performed 18–24 hrs post transfection with a multiplicity of infection (MOI) of ~100:1. For antibody pretreatment experiments, partially purified R. typhi was incubated with 20 μg Melon Gel IgG (Thermo Scientific) purified rabbit pre-immune serum or anti-RalF_Rt polyclonal antibody or 20 μg purified rabbit pre-immune serum or anti-RalF_Rt Fab fragments for 30 min prior to infection. Fab fragments were purified using the Fab Purification Kit (Thermo Scientific) according to manufacture’s protocol.

Recombinant protein purification and antibody production

The expression and purification of recombinant proteins were performed as previously described [111]. Codon optimized (Life Technologies) R. typhi rvhD4 (RT0284) was cloned into pTrcHis2-TOPO vector under the control of the trc promoter (Life Technologies). Full-length R. typhi ralF was cloned into pEXP5-NT/TOPO (Life Technologies) and transformed into E. coli strain bl21-codonplus(de3)-ril (Stratagene). Primers used for cloning can be found in S1 Table. The expression of recombinant proteins was induced with 1 mM IPTG and recombinant proteins were purified by affinity chromatography under native conditions using nickel-nitrilotriacetic acid resin (Ni-NTA) superflow columns (Qiagen) according to manufacturer’s instructions. Polyclonal antibody was generated in rabbit using recombinant RalF_RtFL (Alpha Diagnostic Intl. Inc).

Bacterial two-hybrid assay

R. typhi gene sequences (RT0362, GenBank accession no. YP_067323) encoding full-length RalF (RalF_RtFL) and the rvh T4SS signal truncation (RalF_RtΔT4S) were cloned into the pTRG “prey” plasmid (BacterioMatch II two-hybrid system; Stratagene). A codon optimized (Life Technologies) R. typhi rvhD4 gene (RT0284, YP_067246) was cloned into the pBT “bait” plasmid. Primers used for cloning can be found in S1 Table. The bait (pBT-RvhD4) and prey (pTRG-RalF_RtFL or pTRG-RalF_RtΔT4S) plasmids (100ng each) were co-transformed into BacterioMatch II reporter electrocompetent cells according to the manufacturer’s instruction (GenePulser Xcell, BioRad). The percent growth of CFUs of reporter cells harboring recombinant plasmids on dual selective screening medium were calculated relative to CFUs obtained on non-selective His dropout medium by a drop plate method for counting CFUs [112].

VirD4 ATPase assay

RvhD4 ATPase activity was monitored using a Quantichrom ATPase/GTPase assay kit (Bioassay Systems), according to the manufacturer’s instructions and as described previously [113]. Briefly, 200–12.5 ng/well of purified recombinant RvhD4 protein was incubated in the presence of 1 mM ATP for 30 min at 37°C. Generated free phosphate was quantified by measuring absorbance at OD 620 nm. All of the samples were measured in triplicate wells, and data are given as averages ± S.D. of three independent experiments.

Protease treatment of R. typhi

R. typhi was purified from heavily infected Vero76 cells. Briefly, infected cells were scrapped into media and spun at 12,000 x g for 10 min at 4°C. Cells were resuspended in ice cold PBS, pH 7.2 containing MgCl₂ (PBS-Mg) and sonicated for 10 sec on ice using output 6 of a Sonic Dismembrator (Fisher Scientific). The lysate was filtered through a 5.0 μm filter (Millipore). The filtrate containing R. typhi was layered onto a 20% sucrose cushion at a 1:1 ratio and centrifuged at 16,000 x g for 15 min at 4°C to pellet R. typhi. R. typhi was resuspended in PBS-Mg and again purified with a 20% sucrose cushion. Purified R. typhi was treated with 400 μg/mL or 800 μg/mL Proteinase K (Sigma-Aldrich) for 1 hr at room temperature in PBS-Mg buffer as previously described [114]. Following incubation, Halt Protease and Phosphatase Inhibitor Cocktail (Thermo Scientific) was added to the reaction, and bacteria centrifuged at 16,000 x g for 10 min at 4°C. R. typhi were washed with PBS-Mg and resuspended in PBS-Mg and NuPAGE LDS sample buffer and reducing reagent (Life Technologies). Lysates were separated on a NuPAGE Bis-Tris SDS-gel (Life Technologies) and immunoblotted with rabbit anti-RalF or anti-EF-Ts as the R. typhi cytoplasmic marker [43,115]. Densitometry was performed using ImageJ (NIH) and RalF intensity was normalized to EF-Ts.

Bioinformatics and phylogenomics analyses

Using RalF_Rt as a query, BLASTP searches were performed against the NCBI ‘Rickettsia’ database (taxid:780). Full length RalF_R homologs were aligned with MUSCLE v3.6 [116] using default parameters. Initial domain characterization of RalF_R proteins followed that previously described for R. prowazekii [52]. Using Phyre v.2.0 [69], RalF_Rt was modeled to the crystal structures of Legionella pneumophila RalF (PDB 1XSZ, 4C7P) [50,53] to confirm the boundaries of the Sec7 domain (S7D) and Sec7-capping domain (SCD). The S7D and SCD of Rickettsia and Legionella RalF homologs were aligned with MUSCLE, superimposing the secondary structure of RalF_L over the alignment.

The divergent C-terminal domain (CTD) of RalF_R proteins was further described based on distinct characteristics, i.e. a sequence of variable length that includes a Pro-rich tract, as well as a putative secretion signal sequence within the terminal 40 aa. Additional Rickettsia proteins that lack the S7D and SCD, mostly from SFG rickettsiae, were utilized to characterize this region. All full length and partial RalF_R homologs were used to assess the synteny of the ralF locus across select Rickettsia genomes. For these genomes, gene neighborhood models were constructed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database [117], with manual adjustment to gene annotations. Additional bioinformatics/phylogenomics methodologies are described in S2–S5 Figs.

Mammalian expression plasmids and transfections

Genomic DNA from R. bellii str. OSU 85–1299, R. felis str. Pedreira, R. montanensis str. M5/6, and R. typhi str. Wilmington was purified using DNeasy Blood and Tissue Kit (Qiagen). RalF constructs were amplified as EcoRI/BamHI fragments using primers in S1 Table, with the exception of RalF_RbFL, which was cloned using Clontech InFusion technology. Amplicons were cloned into the pGEMT-Easy vector (Promega) and confirmed by sequencing (The Biopolymer/Genomics Core Facility, University of Maryland School of Medicine). Plasmids were digested with EcoRI and BamHI, with ralF fragments subcloned into the pEYFP-C1 vector (Clontech). All plasmids were transformed into Mix & Go Competent Cells—Strain Zymo 5α (Zymo Research). Plasmids pCDNA3-mRFP-Arf1, pCDNA3-mRFP-Arf5 and pCDNA3-mRFP-Arf6 were generous gifts from Prof. Vassilis Koronakis (University of Cambridge, UK). GFP-C1-PLCδ-PH (Addgene plasmid # 21179) was kindly gifted by Tobias Meyer [118]. All plasmids were purified using PerfectPrep EndoFree Plasmid Maxi Kit (5 Prime).

For transfections, HeLa cells seeded in 8-well chamber slides were transfected with 200ng plasmid per well using Xfect (Clontech) and HEK293T cells in T-75 flasks were transfected with 10μg plasmid using Lipofectamine 2000 (Life Technologies) according to manufactures’ protocols.

Immunofluorescence

Twenty-four hours post transfection or at indicated times post infection, cells were fixed with 4% PFA for 10 min at room temperature. Cells were washed three times with PBS and permeabilized in Blocking Buffer (0.2% saponin, 5% FBS in PBS) for 30 min. Primary antibodies mouse anti-PDI (clone RL90, BD Transduction Laboratories, diluted 1:200), mouse anti-GM130 (clone 610822, BD Transduction Laboratories, diluted 1:200), rat anti–R. typhi serum (1:500), rabbit anti-RalF_Rt (1:100), and rabbit anti-GFP (Life Technologies, diluted 1:1000) were diluted in blocking buffer and incubated with cells for 1 h. Cells were then washed with PBS and incubated with Alexa Fluor 594 or Alexa Fluor 488 secondary antibodies (Life Technologies) diluted 1:2000 in Blocking Buffer for 1 h or 30 min. Finally, cells were washed three times with PBS and mounted using ProLong Gold Anti-Fade mounting media with DAPI (Life Technologies). Actin was stained with Alexa Fluor 594 phalloidin (Life Technologies) and the plasma membrane stained with Alexa Fluor 594 wheat germ agglutinin (WGA, Life Technologies) according to manufacturer’s protocol. For confocal microscopy, cells were viewed under a Zeiss LSM 510 Meta Confocal Microscope (University of Maryland Baltimore Confocal Core Facility). For conventional fluorescence microscopy a Nikon Eclipse E600 fluorescent microscope with a Q Imaging Retiga 2000R camera was used to capture images with QCapture Pro software. Images were processed using ImageJ software (NIH). Co-localization analysis was performed using the CoLoc2 plugin in the ImageJ software program [119]. The Pearson’s correlation coefficient was calculated for 5–10 cells per condition from two independent experiments to measure the strength of association between each RalF protein and the cell organelle (i.e., plasma membrane, endoplasmic reticulum, or Golgi apparatus). Two-sided Student’s t-tests were performed to determine statistical significance for co-localization coefficients compared to control eYFP.

Cell fractionation

Cellular fractionation was completed as previously described [52]. Briefly, at 24 hrs post transfection, HEK293T cells were washed once with PBS and collected in 500 μL homogenization buffer (150 mM KCl, 20 mM HEPES pH 7.4, 2 mM EDTA) containing protease inhibitors and passed 30 times through a 27G-needle. The lysate was centrifuged at 2000 x g for 5 min at 4°C to remove the nuclear fraction. The supernatant was subsequently centrifuged at 100,000 x g for 1 h at 4°C to pellet the membrane fraction. The supernatant was removed (cytoplasmic fraction) and the pellet (membrane fraction) was resuspended in 80 μL of homogenization buffer. Twenty micrograms of cytoplasmic and membrane fractions were separated by SDS-PAGE and blotted with anti-GFP rabbit serum (Life Technologies). Rabbit anti-calnexin [clone ab13505] and mouse anti-GAPDH [clone 6C5] antibodies (Abcam) were used as markers of the membrane and cytosol fractions, respectively.

Ionomycin treatment

Transfected HeLa cells were washed with PBS and incubated in 100 μL of either phosphate buffered saline (PBS) or Krebs-Ringer solution (120 mM NaCl, 4.7 mM KCl, 1.1 mM CaCl₂, 0.7 mM MgSO₄, 10 mM glucose, 10 mM Na-HEPES, pH 7.4). Ionomycin (Sigma Aldrich) was added to a final concentration of 5 μM and cells were incubated for 10 min. EGTA was added to a final concentration of 2 mM and cells were incubated for 10 min. Cells were then infected with R. typhi (described above) or fixed and stained as described above.

Protein pull-down

RalF_Rt was cloned into the pTrcHisA vector (Life Technologies, see S1 Table for primer sequences) and transformed into Top10 E. coli cells (Life Technologies). Protein expression was induced with 1 mM IPTG overnight at 30°C. E. coli were lysed using Pierce Lysis Buffer in the presence of HALT Protease Inhibitors (Thermo Scientific) and imidazole added to a final concentration of 10 mM. Lysates were sonicated three times for 20 sec each using setting 6 of a Sonic Dismembranator (Fisher Scientific). mRFP-Arf5 and –Arf6 were expressed in HEK293T cells as described above. HEK293T cells were lysed using Pierce Lysis Buffer in the presence of HALT Protease Inhibitors (Thermo Scientific) and imidazole added to a final concentration of 10 mM.

Pull-down assays were performed using the Pierce Pull-Down PolyHis Protein:Protein Interaction kit according to manufacture’s protocol. Briefly, HisPur Cobalt Resin was incubated with rHis-RalF_Rt E. coli lysate or buffer alone for 1 hr. The resin was washed 5 times and then incubated with either mRFP-Arf5 or -Arf6 HEK293T lysate for 2 hr. Resin was again washed 5 times and bound proteins eluted with 290 mM imidazole elution buffer. Eluted proteins and 10% of the input protein were analyzed by protein immunoblot using the primary antibodies rabbit anti-RalF_Rt, rabbit anti-Arf5 (1:1000, Thermo Scientific, PA5-31432) and rabbit anti-Arf6 (1:1000, Thermo Scientific, PA1-093) and the secondary antibody HRP anti-rabbit IgG (1:2000, BioLegend, clone 6B9G9).

Gene knockdown

Negative and MISSION siRNAs against human Arf5 (SASI_Hs01_00024789) and Arf6 (SASI_Hs02_0033275) were obtained from Sigma Aldrich. All siRNA knockdowns were performed in HeLa cells using Lipofectamine 2000 (Life Technologies). Cells were used 24 hrs post transfection. Knockdowns were verified by western blot analysis using 1:1000 dilution of primary antibodies rabbit anti-Arf5 or anti-Arf6 (Thermo Scientific). As a loading control, membranes were re-probed with rabbit anti-GAPDH antibody (1:1000, Abcam).

Data analysis

Graphs show the mean ± SD of three independent experiments; 100 cells were counted for each condition in every experiment. Statistical analyses were performed using two-tailed equal variance Student’s t-test.

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6

Attachment 7

Attachment 8

Attachment 9

Attachment 10