Rickettsia conorii, a member of the spotted fever group (SFG) of the genus Rickettsia and causative agent of Mediterranean spotted fever, is an obligate intracellular pathogen capable of infecting various mammalian cell types. SFG rickettsiae express two major immunodominant surface cell antigen (Sca) proteins, OmpB (Sca5) and OmpA (Sca0). While OmpB-mediated entry has been characterized, the contribution of OmpA has not been well defined. Here we show OmpA expression in Escherichia coli is sufficient to mediate adherence to and invasion of non-phagocytic human endothelial cells. A recombinant soluble C-terminal OmpA protein domain (954–1735) with predicted structural homology to the Bordetella pertussis pertactin protein binds mammalian cells and perturbs R. conorii invasion by interacting with several mammalian proteins including β1 integrin. Using functional blocking antibodies, small interfering RNA transfection, and mouse embryonic fibroblast cell lines, we illustrate the contribution of α2β1 integrin as a mammalian ligand involved in R. conorii invasion of primary endothelial cells. We further demonstrate that OmpA-mediated attachment to mammalian cells is in part dependent on a conserved non-continuous RGD motif present in a predicted C-terminal ‘pertactin’ domain in OmpA.Our results demonstrate that multiple adhesin–receptor pairs are sufficient in mediating efficient bacterial invasion of R. conorii.
Members of the genus Rickettsia are Gram-negative, obligate intracellular bacteria that are transmitted to a human host via an arthropod vector. Rickettsial species are categorized as either a member of the typhus group (TG) or the spotted fever group (SFG), based on differences in antigenicity to lipopolysaccharide, the presence of certain outer membrane proteins, and in part on the diseases they cause (Anacker et al., 1987; Vishwanath, 1991; Walker et al., 1995; Feng and Walker, 2003). Members of both groups are the aetiological agents of severe emerging infectious diseases throughout the world, with Rickettsia prowazekii (TG) and Rickettsia rickettsii (SFG) being classified as select agents by the United States Centers for Disease Control and Prevention (CDC) due to the severity of the disease, the existence of antibiotic-resistant strains (Weiss and Dressler, 1962a,1962b), and the potential for aerosol transmission (Oster et al., 1977).
Transmission of SFG rickettsiae, including Rickettsia conorii, to the human host occurs primarily via tick-bite inoculation when the salivary contents of the infected vector are transferred during a blood meal (Hackstadt, 1996). Expansion of the bacterial population and horizontal cell–cell transmission near the inoculation site results in a localized dermal and epidermal necrosis referred to as an eschar or tache noir (Walker et al., 1988). Once an infection is established in the host, SFG rickettsiae primarily target the endothelial lining of the vasculature, causing injury to the vascular endothelium and infiltration of perivascular mononuclear cells, which leads to vasodilation, an increase in fluid in the interstitial space. These symptoms are often accompanied by a characteristic ‘spotted fever’ dermal rash in some infected patients (Hand et al., 1970; Walker et al., 1988). Damage to target endothelial cells, particularly in the lung and brain, can result in severe pathology such as pulmonary oedema, interstitial pneumonia and neurological and other multi-organ manifestations (Raoult and Roux, 1997). Because initial clinical manifestations mimic flulike symptoms, disease is often misdiagnosed or proper diagnosis is delayed. Some broad-spectrum antibiotics, such as doxycycline, are capable of eradicating an infection; however, if left untreated, mortality rates are estimated to be as high as 20% (Yagupsky and Wolach, 1993; Yagupsky, 2000).
Previous studies have identified a family of at least 19 genes termed surface cell antigen (sca) proteins that encode either predicted secreted proteins or outer membrane proteins (Blanc et al., 2005). While most of these genes are split, fragmented or absent in the majority of rickettsial species, there are five sca genes that appear to have evolved under positive selection and are present in the genomes of nearly all rickettsial species: ompA (sca0), ompB (sca5), sca1, sca2 and sca4 (Blanc et al., 2005). The predicted proteins encoded by these genes, excluding sca4, share homology with a family of proteins in Gram-negative bacteria termed autotransporters, many of which are known virulence factors (Henderson and Nataro, 2001). In R. conorii, Sca1 has been shown to mediate bacterial adherence, while Sca2 has been proven sufficient to mediate both adherence to and invasion of mammalian cells (Cardwell and Martinez, 2009; Riley et al., 2010). OmpB has also been shown to participate in both the adherence and invasion processes (Uchiyama et al., 2006; Chan et al., 2009) and to play a role in humoral and cellular protective immune responses (Feng et al., 2004a,2004b; Chan et al., 2011). Previous studies have also identified plasma membrane-associated Ku70 as a receptor for OmpB (Martinez et al., 2005), the only reported mammalian host receptor to date. While OmpA has also been shown to elicit humoral immune responses (Vishwanath et al., 1990; Diaz-Montero et al., 2001; Feng and Walker, 2003; Feng et al., 2004a) and to mediate adherence in a related rickettsial species, R. rickettsii (Li and Walker, 1998), its function in R. conorii remains largely unknown.
The presence of ompA/sca0 in the R. conorii genome and other SFG rickettsiae would suggest that, similar to other related Sca proteins, OmpA in R. conorii plays a critical role in mediating interactions with mammalian cells. To elucidate the role of OmpA in rickettsial–host cell interactions, we have adapted a heterologous expression system in a surrogate Gram-negative species, Escherichia coli, that has been previously utilized to successfully study Sca1, Sca2 and OmpB protein function (Uchiyama et al., 2006; Cardwell and Martinez, 2009; Chan et al., 2009; Riley et al., 2010). Here, we show that R. conorii OmpA, when expressed at the outer membrane of E. coli cells, is sufficient to mediate adherence to and invasion of mammalian cells in vitro in the absence of additional virulence factors. Our results also demonstrate that OmpA-mediated cellular invasion is dependent on α2β1 integrin expression and is independent from the OmpB-Ku70-mediated invasion pathway of R. conorii.
Expression of recombinant R. conoriiOmpA in E. coli
Previously, an E. coli heterologous expression system was used to study the function of other individual R. conorii Sca proteins (Uchiyama et al., 2006; Cardwell and Martinez, 2009; Chan et al., 2009; Riley et al., 2010). We were able to adapt this system to express OmpA at the E. coli outer membrane by cloning the full-length R. conorii ompA open reading frame into the IPTG-inducible vector, pET-22b, resulting in pMC022. Western immunoblot analysis of E. coli outer membrane fractions confirmed that BL21 (DE3)/pMC022 (ompA) expresses a high molecular weight His6-reactive protein which corresponds to the predicted mobility of the full-length R. conorii OmpA protein (Fig. 1A, left panel). We confirmed that the expressed protein was OmpA by using specific monoclonal antibodies (mAbs) against the protein (Fig. 1A, right panel). Furthermore, using flow cytometry analysis, we demonstrated the presence of OmpA protein on the surface of induced BL21(DE3)/pMC022 cells (Fig. S1). In addition, as has been observed for other Sca proteins in this heterologous expression system (Cardwell and Martinez, 2009; Chan et al., 2009; Riley et al., 2010), OmpA does not appear to be proteolytically processed by E. coli.
OmpA is sufficient to mediate adherence to and invasion of host cells
To assay for the putative contribution of R. conorii OmpA during the initial host cell interaction, E. coli expressing OmpA at the outer membrane were initially assessed for the capacity to associate with various cultured mammalian cells by a colony-forming unit (CFU) adherence assay. As shown in Fig. 1B, expression of OmpA is sufficient to mediate cell association of E. coli to cultured HeLa and human microvascular lung endothelial (HMVEC-L) cells when compared to E. coli harbouring a control vector, pET-22b. Similar results were obtained using the human endothelial cell line, EA.hy 926 (data not shown). We confirmed the sufficiency of OmpA to mediate cell association to HMVEC-L cells using a fluorescence-based adherence assay (Fig. 1C). A standard gentamicin protection assay was used to determine whether OmpA was also sufficient to mediate host cell invasion, as has been previously demonstrated for a cohort of Sca proteins in R. conorii (Uchiyama et al., 2006; Cardwell and Martinez, 2009; Chan et al., 2009). Confluent monolayers of mammalian cells were infected with IPTG-induced E. coli harbouring either pET-22b, pMC022 or pYC9, a OmpB-expressing plasmid used as a positive control (Chan et al., 2009). Extracellular bacteria were killed by gentamicin, and internalized bacteria were quantified by CFU enumeration (Fig. 1B, right panels). These data illustrate that, in the absence of any additional R. conorii surface antigens, OmpA is sufficient to facilitate adherence to and internalization into non-phagocytic human epithelial cells and primary human endothelial cells.
Soluble OmpA protein binds mammalian cells and inhibits rickettsial invasion
We then sought to determine whether recombinant purified OmpA protein was able to associate with mammalian cells and competitively inhibit R. conorii infection. Attempts to express and purify the full-length OmpA passenger domain (aa 35–1735) as a soluble protein were unsuccessful, likely due to instability caused by several tandem repeats in the N-terminus of the protein. However, as shown in Fig. 2A, we were able to express and purify a portion of the R. conorii OmpA passenger domain comprised of two N-terminal tandem repeats and the remainder of the C-terminus of the passenger domain (aa 954–1735) fused to an N-terminal glutathione-S-transferase moiety (GST-OmpA954–1735). We incubated HeLa and HMVEC-L cells in suspension with GST-OmpA954–1735 or GST alone, and processed the samples for flow cytometric analysis. As shown in Fig. 2B, GST-OmpA954–1735, but not GST alone, is able to bind to mammalian host cells.
For the evaluation of the specific contribution of OmpA to a rickettsial infection, we pre-incubated confluent monolayers of HMVEC-L cells with GST-OmpA954–1735 or GST and then infected with R. conorii. In a concentration-dependent manner, incubation with GST-OmpA954–1735 is able to inhibit R. conorii association at levels up to approximately 50% at the highest concentration used (Fig. 2C). To assess for the specificity of inhibition against another conserved Sca protein, we pre-incubated the EA.hy 926 human endothelial cell line with GST or GST-OmpA954–1735 and then infected these cells with E. coli BL21 (DE3) harbouring either pMC022 (ompA) or pYC9 (ompB). As shown in Fig. 2D, GST-OmpA954–1735 competitively inhibits OmpA-mediated cellular association, but does not have any effect on the ability of OmpB-expressing bacteria to associate with endothelial cells. These data taken together demonstrate that recombinant OmpA protein is capable of binding mammalian cells, and is sufficient to competitively inhibit R. conorii and OmpA-mediated adherence to target mammalian cells.
R. conoriiOmpA interacts with specific mammalian surface protein complexes
We next sought to identify mammalian interacting proteins that could potentially serve as receptors for OmpA. We performed an in vitro ‘GST pull-down’ assay using soluble HeLa cell proteins incubated with either GST-OmpA954–1735 or GST alone. The resulting complexes were captured with glutathione-sepharose and eluted with excess glutathione. Unique putative interacting proteins that were exclusive to GST-OmpA954–1735, but not GST, were analysed by tandem mass spectrometry. This analysis revealed that GST-OmpA954–1735 appears to interact with several high molecular weight mammalian proteins including filamin-A, epiplakin and dynein (Fig. 3B). Each of these cytoskeletal linker proteins are involved in sub-membranous complexes, but none are directly associated with the plasma membrane (Margadant et al., 2011) suggesting that they are not viable candidates for R. conorii receptors. Interestingly, the most abundant species found in the analysis, filamin-A, has been shown to form an adhesion complex with β1 integrin, and integrins have been demonstrated to be involved both as receptors and as downstream signalling effectors linked specifically to bacterial and viral invasion processes (Isberg and Barnes, 2001; Eto et al., 2007; Stewart and Nemerow, 2007; Cantor et al., 2008). We therefore hypothesized that β1 integrin may function as a mammalian ligand for OmpA. To test this, we analysed the samples sent for mass spectrometry by Western immunoblot analysis with antibodies specific to several β and α integrin subunits. We revealed the presence of both β1 and α2 integrin, but not β4 and α1 integrin in the GST-OmpA954–1735 eluates (Fig. 3C). Taken together, these results suggest that α2β1 integrin may be a mammalian interacting protein with R. conorii OmpA.
α2β1 integrin is involved in R. conorii invasion of mammalian cells
We then investigated whether β1 integrin colocalized with invading R. conorii in infected endothelial cells. HMVEC-L cells were infected with R. conorii and then processed for immunofluorescence microscopy. Analysis of single cross sections from confocal Z-stacks reveals that β1 integrin colocalizes with several invading R. conorii (Fig. 4A and B, arrows). These results suggest that infection of the mammalian host cell with R. conorii leads to the recruitment of β1 integrin to entry foci and that some β1 integrin-containing receptors at the plasma membrane are involved in the uptake process.
We sought to further investigate whether β1 integrin or other integrin subunits could serve as a mammalian ligand for R. conorii. We utilized a panel of functional blocking mAbs specific to β integrin subunits (Eto et al., 2007) to confirm the involvement of β1 integrin in the R. conorii invasion pathway. Pre-incubation of HMVEC-L cells with functional blocking β integrin mAbs had no significant effect on total R. conorii cell association (data not shown); however, pre-incubation with a mAb against β1 integrin (MAB1951) significantly perturbed invasion compared to other β integrin mAbs and an isotype-matched control mAb (9E10) (Fig. 5A). These results are consistent with previous findings using functional blocking antibody against Ku70, a previously identified R. conorii receptor (Martinez et al., 2005). Small interfering RNA (siRNA) transfection was then used to validate the phenotype illustrated by the blocking mAbs. We transfected HMVEC-L cells with scrambled siRNA or siRNAs against β1 integrin and infected these cells with R. conorii. As shown in Fig. 5B, reduction of endogenous β1 integrin expression leads to a corresponding significant inhibition of R. conorii invasion, similar to levels observed in the mAb functional blocking assay, but does not affect the R. conorii adherence (data not shown). To further confirm the role of β1 integrin as a ligand involved in rickettsial host cell internalization, we performed similar cell association and internalization assays using a mouse embryonic fibroblast cell line, GD25, in which the gene encoding for β1 integrin has been inactivated and in GD25β1A cells, a stably transfected derivative of GD25 that encodes the widely expressed wild-type β1 integrin splice variant, β1A (Fèassler and Meyer, 1995). As shown in Fig. 5C, rickettsial invasion of the β1 integrin-null cell line was significantly decreased relative to cells in which β1 integrin expression has been restored.
Integrins are heterodimeric proteins consisting of α and β subunits (Margadant et al., 2011). Prior analysis of affinity chromatography eluates by Western immunoblot analysis showed the presence of α2 integrin; therefore, we sought to then elucidate whether this was the α subunit that pairs with β1 integrin to form the functional OmpA ligand. Using a panel of functional blocking mAbs against α integrin subunits, we demonstrated that inhibition of α2 integrin function, but not other α integrins, significantly diminished R. conorii invasion of HMVEC-L cells (Fig. 5D). As shown in Fig. 5E, we observed a similar inhibition of R. conorii invasion using siRNA transfection against α2 integrin, suggesting that α2β1 integrin is a specific R. conorii mammalian internalization factor.
OmpA is the R. conorii surface protein that interacts with α2β1 integrin
OmpB is a related rickettsial autotransporter protein and was previously demonstrated to be sufficient to interact with plasma membrane-associated Ku70 and lead to bacterial internalization (Martinez et al., 2005; Chan et al., 2009). We wanted to confirm that R. conorii OmpA, and not another rickettsial outer membrane autotransporter protein, was the corresponding rickettsial ligand that interacted with β1 integrin. We initially infected GD25 and GD25β1A cells with E. coli harbouring pET-22b, pMC022 (ompA) and pYC9 (ompB) and then determined the ability of bacteria to associate with these cells. As shown in Fig. 6A, OmpA-mediated adherence is diminished to levels similar to the non-adherent control (pET-22b) in cells that are genetically depleted for β1 integrin (GD25 cells). In contrast, OmpB-mediated adherence is not affected. To further support these findings, we then reduced endogenous α2 and β1 integrin expression in EA.hy 926 cells using siRNA transfection, and infected these cells with E. coli BL21 (DE3) strains harbouring the control plasmid pET-22b, pMC022 (ompA) or pYC9 (ompB). Reduction of endogenous β1 and α2 integrin expression by siRNA transfection, both individually and in tandem, perturbs OmpA-mediated association with human endothelial cells, but it has no effect on OmpB-mediated bacterial adherence. Taken together, these results demonstrate that OmpA and α2β1 integrin represent a bona fide adhesin–mammalian ligand pair involved in cellular attachment to and ultimately entry into non-phagocytic mammalian cells in a mechanism that is independent from OmpB-Ku70-mediated rickettsial entry pathway.
The OmpA ‘pertactin domain’ contains a non-continuous RGD motif involved in cellular adherence
We next sought to characterize the molecular details governing OmpA–integrin interactions. A Blastp-based bioinformatic analysis of the functional blocking R. conorii OmpA domain (aa 954–1735) did not reveal any significant amino acid sequence homology to known adhesins from other pathogenic Gram-negative bacteria. However, using a web-based structural homology prediction software suite, Phyre (Kelley and Sternberg, 2009), we determined that R. conorii OmpA954–1735 is predicted to adopt a right-handed β-helix fold (Fig. 7A) similar to a fold found in the passenger domain of the Bordetella pertussis pertactin autotransporter protein (Junker et al., 2006). Pertactin binds host mammalian cells in an integrin-dependent manner through a canonical RGD integrin-binding motif (Everest et al., 1996). Proteins containing linear RGD motifs are sufficient to mediate β integrin-dependent interactions, but are not absolutely required for this process as has been demonstrated for the FimH protein of type 1 piliated uropathogenic E. coli (UPEC) (Eto et al., 2007) and the invasin protein from Yersinia pseudotuberculosis (Leong et al., 1990; 1991). A canonical linear RGD sequence is not present within the OmpA passenger domain (aa 35–1735); however, OmpA does contain a RNIGD sequence (aa 1106–1110, Fig. 7B), which is invariant in OmpA proteins from other pathogenic SFG rickettsiae (Fig. 7C). This motif is predicted to be surface-exposed and positioned so as to interact with mammalian cells (Fig. 7B). We therefore sought to determine whether OmpA-mediated cellular association is dependent on an RGD sequence. We pre-incubated EA.hy 926 cells with a control peptide and a peptide containing the RGD sequence and then infected these cells with E. coli BL21 (DE3) harbouring pMC022 (ompA), pYC9 (ompB) and pRI203, a plasmid encoding for the Y. pseudotuberculosis invasin protein (Isberg et al., 1987). As shown in Fig. 7D, the RGD containing peptide is sufficient to inhibit OmpA- and invasin-mediated cellular association, but had no effect on OmpB-mediated adherence. We then determined whether the OmpA ‘pertactin domain’ is sufficient to inhibit cellular adherence mediated by an unrelated antigen whose function is also blocked by RGD containing proteins. As shown in Fig. 7E, GST-OmpA954–1735, but not GST alone, is capable to competitively inhibiting Y. pseudotuberculosis invasin-mediated adherence to mammalian cells. Taken together, these results suggest that the R. conorii OmpA protein contains a non-linear, but functional RGD-like motif that is likely involved in mediating association with plasma membrane β1 integrin-containing receptors.
Invasion of target host cells is essential for the subsequent intracellular survival and proliferation of obligate intracellular pathogens like R. conorii. Previous work has demonstrated that R. conorii specifically binds to plasma membrane-associated Ku70 in a OmpB-dependent manner (Martinez et al., 2005; Chan et al., 2009), and that invasion is dependent on Arp2/3, an actin-nucleating protein complex, as well as a multitude of host signalling events mediated by c-Cbl, clathrin, caveolin 2, Cdc-42, phosphoinositide 3-kinase, c-Src and other kinases (Martinez and Cossart, 2004; Chan et al., 2009). Here we demonstrate that R. conorii invasion is also mediated by an interaction between OmpA and the mammalian interacting factor, α2β1 integrin.
The α2β1 integrin heterodimer has been previously shown to be a receptor for a variety of matrix and non-matrix ligands including collagen type I, laminins, E-cadherin and several viruses (Languino et al., 1989; Mizuno et al., 2000; Graham et al., 2003). This receptor is expressed on various cell types, including epithelial and endothelial cells, and interactions between extracellular ligands and α2β1 integrin have been implicated in several biological processes such as caveolin- and clathrin-dependent endocytosis (Shi and Sottile, 2008; Ezratty et al., 2009), inflammation and immunity (Gahmberg et al., 1998). Engagement of extracellular ligands at the cell surface is known to induce integrin clustering, and this can have positive feedback both on the binding avidity of the integrin receptor itself as well as the activation of an array of downstream signalling events (Huveneers and Danen, 2009). Our immunofluorescence results presented here show that R. conorii recruits β1 integrin to sites of bacterial entry in primary endothelial cells, suggesting that engagement of a rickettsial ligand (OmpA) can induce integrin receptor clustering and activation. Of note, several bona fide integrin ligands initiate signalling cascades, leading to the phosphorylation of conserved tyrosine resides on the β integrin cytoplasmic domain and activation of c-Src and focal adhesion kinase (FAK) (Cantor et al., 2008). R. conorii invasion into non-phagocytic mammalian cells involves the activation of src-family tyrosine kinases and tyrosine phosphorylation of FAK (Martinez and Cossart, 2004), two signalling events strongly associated with β1 integrin activation (Parsons, 2003). In addition, F-actin dynamics are altered downstream of integrin clustering by recruitment and activation of several proteins, GTPases and lipid kinases, including Cdc42, and PI 3-kinase (Huveneers and Danen, 2009). It is possible that, by engaging integrin receptors, R. conorii may stimulate these signalling pathways ultimately leading to the observed F-actin polymerization required for efficient bacterial internalization (Martinez and Cossart, 2004). Whether OmpA-expressing bacteria can directly trigger signalling cascades known to be associated with β1 integrin activation is currently under investigation.
An intact ompA gene is present in other pathogenic SFG rickettsiae such as R. rickettsii, Rickettsia japonica and Rickettsia africae; however, ompA is split, fragmented or otherwise absent in species belonging to the TG including R. prowazekii and Rickettsia typhi (Blanc et al., 2005), suggesting that different autotransporter–mammalian ligand pairs govern the invasion of this class of obligate intracellular pathogens. Interestingly, a mutation in ompA in the SFG rickettsial strain R. rickettsii Iowa results in the lack of observable OmpA protein and is correlated with a defect in virulence in a guinea pig model of infectivity (Ellison et al., 2008), further highlighting the importance of OmpA for infections mediated by SFG rickettsiae. While the ompA gene products expressed in SFG rickettsiae vary in predicted molecular weight, each OmpA protein contains a conserved putative integrin interaction motif, RNIGD. Proteins containing non-continuous RGD motifs of varying lengths including R … D, KQAGDV, LDV/IDS and RLD/KRLDGS sequences have been shown to be involved in mediating interactions with various integrin receptors (Ruoslahti, 1996). Our results demonstrate that R. conorii OmpA contains a functional non-continuous RGD domain that is involved in mediating interactions with β1 integrins. Whether the RNIGD motif present in the OmpA C-terminal ‘pertactin’ domain is required for interactions with α2β1 integrin is currently under active investigation.
Our results also demonstrate that similar to other invasive ‘zipper’ mechanism pathogens such as Listeria monocytogenes and UPEC, R. conorii expresses several antigens conserved in other SFG rickettsiae that are sufficient to mediate bacterial adherence to and internalization of mammalian cells independent of one another. We have also demonstrated that purified, recombinant proteins containing portions of Sca1, Sca2, OmpB (Cardwell and Martinez, 2009; Chan et al., 2009; Riley et al., 2010) and now OmpA are capable of competitively inhibiting R. conorii adherence and invasion of mammalian cells, suggesting that these Sca proteins are interacting with putative mammalian receptors. Interestingly, Ku70 and α2β1 integrin are expressed at the plasma membrane of endothelial cells (Senger et al., 1997; Muller et al., 2005) suggesting R. conorii may utilize these receptors to initially gain access to an intracellular niche after being introduced into the vasculature of a host by a feeding tick. It is possible that other antigens, such as Sca1 and Sca2, may also contribute to this initial attachment and invasion of endothelial cells respectively. Alternatively, these proteins may be involved in allowing R. conorii and other SFG rickettsiae to target organs such as the lungs, liver and spleen where pathology is prevalent in models of disseminated disease (Walker et al., 1994; Chan et al., 2011). While several Sca proteins have now been demonstrated to be sufficient for invasion of mammalian cells, we propose that effective entry of SFG rickettsiae into target cells in vivo likely involves OmpA and other Sca proteins working in concert to generate the signal transduction cascades necessary to trigger the entry process.
The results presented here highlight the involvement of integrins and integrin-dependent signalling events in the initial rickettsia–host cell interaction. However, we do not exclude the possibility that β integrins may play additional roles during an infection. R. conorii infections of endothelial cells have been correlated with the disruption of cellular adherens and tight junctions leading to an increase in cellular permeability (Valbuena and Walker, 2005). While the molecular details of pathology associated with rickettsial disease are not well understood, studies have shown that neutrophil-mediated vascular permeability is integrin-dependent (DiStasi and Ley, 2009). Understanding the mechanisms that lead to alterations in endothelial junctions may prove useful in efforts to identify host and bacterial targets that may be exploited for the development of anti-rickettsial therapies.
Cell lines and bacterial strains
The human cervical adenocarcinoma cell line HeLa and the human endothelial cell line EA.hy 926 were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum and 1× non-essential amino acids (cDMEM). Human lung microvascular endothelial (HMVEC-L) cells were grown in endothelial growth medium supplemented with a -2-MV bullet kit (Lonza). The β1 integrin-null cell line GD25 and its stably transfected β1 integrin-expressing derivative GD25β1A (kindly provided by Dr D.F. Mosher, the University of Wisconsin-Madison) were grown in cDMEM. Puromycin was supplemented at 10 μg ml−1 to maintain the stable transfection. All mammalian cells were grown at 37°C/5% CO2. Bacterial genetic manipulations were performed in E. coli One Shot®TOP10 chemically competent cells (Invitrogen) grown at 37°C in LB broth supplemented with carbenicillin (50 μg ml−1) where appropriate. E. coli BL21 (DE3) cells were transformed with the indicated plasmids and plated on selective LB agar containing carbenicillin (50 μg ml−1). R. conorii Malish 7 was propagated in Vero cells and isolated and stored as previously described (Martinez and Cossart, 2004). Titres were determined by a limiting-dilution infectivity assay and calculated using the Reed and Muench formula.
Antibodies and other reagents
Anti-GST polyclonal rabbit antibody Z-5, polyclonal antibodies against β1 integrin (sc-6622, sc-8978) and β4 integrin (sc-9090) were purchased from Santa Cruz Biotechnology. The RGD containing peptide (GRGDNP) and the control peptide (GRGESP) were also purchased from Santa Cruz Biotechnology. Alexa Fluor 488/546-conjugated goat anti-rabbit IgG and Alexa Fluor 488/546-conjugated goat anti-mouse IgG, Texas Red-conjugated phalloidin and DAPI (4′,6′-diamidino-2-phenylindole) were purchased from Molecular Probes. Rabbit polyclonal antisera against the R. conorii OmpA passenger domain were generated using recombinant protein purified from BL21 (DE3)/pBOB001 (described below) essentially as described in Chan et al. (2011). Rabbit polyclonal anti-R. conorii antiserum (RCPFA) generated against paraformaldehyde (PFA)-fixed bacteria has been described (Chan et al., 2011). Rabbit polyclonal anti-E. coli antiserum was generated as described (Chan et al., 2009). IRDye 800CW-conjugated donkey anti-rabbit IgG and IRDye 680LT-conjugated donkey anti-mouse IgG were obtained from LI-COR. mAb against OmpA (clone 13-3) (Anacker et al., 1985; 1987) was kindly provided by R. Heinzen, PhD (NIAID, Rocky Mountain Laboratories). mAbs against β1 integrin (MAB1951F, FCMAB375A4, 217648) and polyclonal antibody against α2 integrin (AB1936) were purchased from Millipore. β (ECM440) and α (ECM430) integrin antibody screening kits were also purchased from Millipore. Complete protease inhibitor tablets were purchased from Roche. Annealed siRNAs against β1 integrin sense strand, α2 integrin sense strand and Silencer negative control siRNA were purchased from Ambion. RNAiMAX was purchased from Invitrogen.
Plasmid DNA constructs
The full-length ompA open reading frame was amplified by PCR from a chromosomal preparation of R. conorii Malish 7 using forward and reverse primers 5′-AGGATCCAGCGAATATTTCTCCAAAATTATTT-3′ and 5′-AACTCGAGAAATTAACACGAACTTTCACACT-3′ respectively. The resulting PCR product contained the restriction sites BamHI and XhoI incorporated 5′ and 3′ of the ompA gene. The PCR product was initially TOPA TA-cloned into pCR2.1 (Invitrogen), resulting in pMC003, then digested with BamHI and XhoI for insertion into the expression vector pET-22b (Novagen), resulting in plasmid pMC022. The sequence coding for the entire OmpA passenger domain (aa 35–1735) was PCR amplified from pMC022 as a template using primers 5′-AAGGATCCATTGCTGTTTCAGGTGTTATTG-3′ and 5′-AACTCGAGTTACATATCTTCATCACCAGAAGAA-3′. The PCR product was digested with BamHI and XhoI and directionally cloned into pYC55 (Chan et al., 2011) resulting in pBOB001. The C-terminal half of the OmpA passenger domain encompassing amino acids 954–1735 was PCR amplified from pMC022 using the forward and reverse primers 5′-AAGGATCCACATTACAAGCTGGAGGAAG-3′ and 5′-AACTCGAGTTACATATCTTCATCACCAGAAGAA-3′ respectively. The resulting PCR product was digested with BamHI and XhoI and ligated into pGEX-2TKP (a generous gift from T. Kouzarides, Gurdon Institute, UK), resulting in the plasmid pBOB002. Construction of pYC9 is described elsewhere (Chan et al., 2009). The pRI203 plasmid containing the Y. pseudotuberculosis inv gene has been described (Isberg et al., 1987) and was kindly provided by M. Mulvey, PhD (The University of Utah).
E. coli outer membrane protein fractionation
Outer membrane protein fractions from E. coli were prepared essentially as described in Nikaido (1994). Briefly, 10 ml of induced E. coli BL21(DE3) culture was pelleted and resuspended in 1 ml of lysis buffer (20 mM Tris, pH 8.0, plus protease inhibitor). Cells were lysed by sonication on ice at amplitude 4, 15 s on and 15 s off, repeated four times or until lysates became translucent. Unbroken cells were spun down by centrifugation at 8000 g for 2 min at 4°C. The supernatant containing the whole cell lysate was added to a clean tube, and 50 μl of 10% sarkosyl was added (final concentration 0.5%) at room temperature to extract inner membrane proteins. Outer membrane proteins were pelleted by high-speed centrifugation (100 000 g) at 4°C for 30 min. Membranes were resuspended in 0.5 ml of 20 mM Tris, pH 8.0, 0.3 M NaCl, resolved by SDS-PAGE, and transferred to nitrocellulose membrane. Membranes were analysed by immunoblot with anti-His6 rabbit polyclonal serum and anti-OmpA mouse mAbs. Immunoreactive species were revealed with chemiluminescence and film exposure or on a LI-COR Odyssey CLx imaging system. The specificity of the OmpA serum was verified by Western immunoblotting using purified OmpA35–1735-His, purified GST-OmpB36–1334 (Chan et al., 2009), the isolated OmpA serum and a mAb (mAb 6B.6) against OmpB (Chan et al., 2011).
Protein expression and purification
Overnight cultures of E. coli BL21 (DE3) harbouring pBOB001 were diluted 1:10 into fresh media containing carbenicillin and grown at 37°C to mid-exponential phase (OD600 = 0.6). Protein expression was induced by the addition of 0.1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) for 4 h at 30°C. Bacteria were harvested by centrifugation and resuspended in denaturing lysis buffer (100 mM NaH2PO4, 10 mM Tris-Cl, 8 M urea pH 8.0 containing protease inhibitor cocktail) and lysed by sonication (40% amplitude, 5 s on/10 s off, 20 min – Fisher Scientific sonic dismembrator model 500). Lysates were cleared by centrifugation at 13 000 g for 30 min at 4°C and applied to a Ni-NTA FF column. The column was washed extensively with wash buffer (100 mM NaH2PO4, 10 mM Tris-Cl, 8 M urea pH 6.3) and then proteins were eluted (100 mM NaH2PO4, 10 mM Tris-Cl, 8 M urea, pH 5.9). Fractions containing protein (OmpA35–1735-His) were analysed by SDS-PAGE and then dialysed into phosphate-buffered saline (PBS) before being stored at −80°C until use.
Overnight bacterial cultures of E. coli BL21 (DE3) transformed with pBOB002 were diluted 1:10 into 1 l of fresh LB medium with carbenicillin and grown at 37°C to mid-exponential phase (OD600 = 0.6). The expression of GST-OmpA954–1735 was induced by the addition of 0.1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) for 3.5 h at 30°C. Bacteria were harvested by centrifugation, resuspended in PBS, pH 7.4, containing protease inhibitor, and lysed by sonication as described above. Lysates were cleared by centrifugation at 13 000 g for 30 min at 4°C and applied to a GST-TRAP FF column on an Åkta fast protein liquid chromatography (FPLC). Protein was eluted with 30 mM glutathione and fractions were dialysed into PBS before being stored at −80°C until use.
In vitro pull-down assay
Fifty micrograms of soluble GST or GST-OmpA954–1735 protein was incubated with cleared HeLa whole cell lysate in NP40 lysis buffer [20 mM Tris (pH 8.0), 150 mM NaCl, 1% NP40, 10% glycerol] for 2 h at room temperature. To this suspension, 50 μl of a 50% slurry of immobilized glutathione-agarose in PBS was added and incubated overnight at 4°C. The agarose beads were then spun down 5 min at 600 g and washed five times in 1 ml of a 1:10 NP40 buffer dilution in PBS, rocking each time for 4 min. To elute bound protein, 30 μl of 30 mM glutathione in PBS was added and samples were incubated with gentle rocking for 30 min at room temperature. The beads were again spun down at 600 g and the supernatant was transferred to a clean tube. 6× SDS-PAGE sample buffer was added to each sample; proteins were resolved by SDS-PAGE and then analysed by silver stain or Western immunoblot analysis. Protein identification by micro-capillary LC/MS/MS analysis was performed at the Taplin Mass Spectrometry Facility (https://taplin.med.harvard.edu/)
Cell association and invasion assays
Cell association and invasion assays for R. conorii and E. coli expressing R. conorii antigens in HeLa cells, EA.hy 926 cells and HMVEC-L cells were performed as previously described (Chan et al., 2009). BL21 (DE3)/pMC022 (ompA) was grown overnight at 37°C, diluted 1:10 into fresh media, grown to mid-logarithmic phase and induced with 0.1 mM IPTG for 3 h at 30°C, while BL21 (DE3)/pYC9 (ompB) was induced with 0.1 mM IPTG for 3 h at 37°C. The BL21 (DE3)/pRI203 (inv) strain was grown overnight at 37°C, diluted 1:10 into fresh media and then grown to mid-logarithmic phase prior to use. To further demonstrate adherence of the indicated E. coli strains to mammalian cells, HMVEC-L cells were seeded onto sterile coverslips in 24-well plates and then infected with 50 μl of an OD600 = 1.0 dilution of BL21 (DE3) containing pET-22b (empty vector) or pMC022 (ompA). Coverslips were processed for immunofluorescence analysis as described in Chan et al. (2009) using Texas Red phalloidin (1:200), DAPI (1:10 000), rabbit anti-E. coli antiserum (1:500) and Alexa Fluor 488-conjugated goat anti-rabbit IgG (1:1000).
For protein blocking experiments, HMVEC-L and EA.hy 926 cells were pre-incubated with the indicated concentrations of GST [100 μg ml−1 (3.84 μM)] and GST-OmpA954–1735 [100 μg ml−1 (1.3 μM)] for 30 min prior to the addition of bacteria. Inhibition assays using the RGD peptide (GRGDNP, 100 μM) and the control peptide (GRGESP, 100 μM) in EA.hy 926 cells were performed essentially as described in Leininger et al. (1991). For antibody inhibition experiments, HMVEC-L cells were incubated with 10 μg ml−1 mAbs against β integrins and α integrins for 30 min in serum-free DMEM (SF DMEM) prior to the addition of R. conorii at an MOI of 10. For experiments using BL21 (DE3) derivatives, bacteria were induced as described above with diluted to an OD600 of 1.0 and then 50 μl of each strain was used to infect the indicated mammalian cell line. Data are representative of at least three independent experiments. P-values were determined using a standard Student's t-test.
RNA interference (RNAi)
HMVEC-L and EA.hy 926 cells were plated onto six-well plates at 2.0 × 105 cells per well and transfected using 30 pmol of the described siRNAs and 6 μl of RNAiMAX. Forty-eight hours post transfection, cells were harvested by trypsinization and replated for an additional 24 h onto sterile glass coverslips in 24-well plates for association and invasion assays as described above, or six-well tissue culture plates for subsequent Western immunoblot analysis of protein expression. Actin was used as a loading control to demonstrate equal loading. The efficacy of transfection was determined by densitometry analysis of immunoblot images captured on the LI-COR Odyssey CLx Imager.
To illustrate colocalization of β1 integrin with R. conorii, HMVEC-L cells were seeded onto sterile glass coverslips in 24-well plates, infected for 30 min with R. conorii, and then processed for immunofluorescence as described (Martinez and Cossart, 2004) using anti-RCPFA, and Alexa Fluor 488 goat anti-rabbit IgG antibodies to label R. conorii. β1 integrin was visualized in infected non-permeabilized cells with 2 μg ml−1 anti-β1 integrin mAb (clone 4B7R) and Alexa Fluor 546-conjugated anti-mouse IgG. Images were captured on a Leica TCS SP2 AOBS Laser Scanning Confocal microscope. Analysis of colocalization was performed on individual confocal images using the ‘Measure RGB’ function in the public domain NIH ImageJ 1.46r analysis software package (http://imagej.nih.giv/ij) with colocalization depicted as overlapping histograms of the green and red channels.
HMVEC-L and HeLa cells in a 10 cm culture dish were washed thoroughly with PBS and dislodged using 5 ml of PBS + 1 mM EDTA with incubation at 37°C/5% CO2. Cells were collected and pelleted by centrifugation, 200 g for 7 min. Cells were resuspended in serum-free DMEM (SF-DMEM) and aliquoted in samples of 1.0 × 106 cells in 1 ml volume. Purified recombinant GST (5 μM) or GST-OmpA954–1735 (5 μM) was added to cells and incubated for 30 min at room temperature. Cells were pelleted as described above, washed with 500 μl of PBS, and fixed in 2% PFA for 20 min. Samples were subsequently washed, blocked in 2% bovine serum albumin (BSA)/PBS and processed for analysis using rabbit anti-GST (Z-5) and Alexa Fluor 488 goat anti-rabbit IgG antibodies. Samples were analysed on a BD LSR-II flow cytometer using fluorescein isothiocyanate (FITC) parameters and the FloJo software package. A minimum of 50 000 events were counted for each sample.
For localization of OmpA at the surface of E. coli, BL21 (DE3) containing either pET22-b (empty vector) or pMC022 (ompA) were grown under non-inducing and inducing conditions as described above. Bacteria were washed three times in PBS and then diluted to an OD600 = 0.5. Five hundred microlitres of this bacterial suspension was incubated using PBS/2% BSA containing anti-OmpA antisera (1:500), Alexa Fluor 488 goat anti-rabbit IgG antibodies and DAPI (5 μg ml−1). Samples were analysed on a BD LSR-II flow cytometer using fluorescein isothiocyanate (FITC) and DAPI parameters and the FloJo software package. A minimum of 100 000 events were counted for each sample.
Bioinformatic analysis and protein modelling
OmpA sequences from SFG rickettsiae were aligned using the ClustalW protein analysis function in the MacVector software package. Accession numbers for OmpA sequences in the analysis are as follows: R. conorii Malish 7 (AAA17405.1), R. rickettsii ‘Sheila Smith’ (ABV76839.1), R. japonica YH (BAK97125.1) and R. africae (AAC35172.2). Amino acids 954–1735 from R. conorii Malish 7 OmpA were submitted online to the Phyre2 server for analysis (Kelley and Sternberg, 2009). The highest scoring model was based on the B. pertussis p62 pertactin protein structure (pdb1DAB) (Emsley et al., 1996).
We would like to thank M.M. Cardwell for constructing the pMC022 plasmid. We also thank members of the Martinez lab for critical review of the manuscript. An award to J. J. M. from the National Institute of Allergy and Infectious Diseases (NIAID), Infectious Diseases Branch (AI 072606) supported this work. We also wish to acknowledge membership within the Region V Great Lakes Regional Center of Excellence (GLRCE) in Biodefense and Emerging Infectious Diseases Consortium (NIAID Award U54-AI-057153).