|Home | About | Journals | Submit | Contact Us | Français|
Sequential expression of outer membrane protein antigenic variants is an evolutionarily convergent mechanism used by bacterial pathogens to escape host immune clearance and establish persistent infection. Variants must be sufficiently structurally distinct to escape existing immune effectors yet retain the core structural elements required for localization and function within the outer membrane. We examined this balance using Anaplasma marginale, which generates antigenic variants in the outer membrane protein Msp2 using gene conversion. The overwhelming majority of Msp2 variants expressed during long-term persistent infection are mosaics, derived by recombination of oligonucleotide segments from multiple alleles to form unique hypervariable regions (HVR). As a result, the mosaics are not under long-term selective pressure to encode a functional protein; consequently, we hypothesized that the Msp2 HVR is structurally permissive for mosaic expression. Using an integrated approach of predictive modeling with determination of the native Msp2 protein structure and function, we demonstrate that structured elements, most notably, β-sheets, are significantly concentrated in the highly conserved N- and C-terminal domains. In contrast, the HVR is overwhelmingly a random coil, with the structured α-helices and β-sheets being confined to the genomically defined structural tethers that separate the antigenically variable microdomains. This structure is supported by the surface exposure of the HVR microdomains and the slow diffusion-type porin function in native Msp2. Importantly, the predominance of the random coil provides plasticity for the formation of functional HVR mosaics and realization of the full potential of segmental gene conversion to dramatically expand the variant repertoire.
Highly antigenically variable microbes use a diversity of mechanisms to generate molecules that allow escape from host immune effectors and result in long-term persistence. Bacterial pathogens across multiple genera, including Anaplasma, Borrelia, Neisseria, and Treponema, generate outer membrane protein variants by gene conversion, resulting in novel expressed surface antigens (1,–3). The potential repertoire of antigenic variants is determined by the number of donor alleles in the pathogen genome combined with the occurrence of segmental recombinatorial events that can generate unique expression site mosaics (1,–3). Notably, segmental gene conversion results in expressed variants not represented by any one allele but generated in situ by recombination of donated oligonucleotides derived from different alleles. This generation of unique mosaics can tremendously expand the potential antigenic repertoire. The number of alleles encoding unique Msp2 variants in Anaplasma marginale is <10, and segmental recombination using oligonucleotide segments derived from these alleles allows the generation of >1,000 potential variants (4,–6). Similarly, segmental gene conversion in Treponema pallidum exponentially increases the potential number of TpK variants (7).
For the potential repertoire to translate into functional variation, two criteria must be met: the generated variant must be sufficiently antigenically distinct to evade immune recognition but must also maintain the outer membrane protein structure and function. Strict structural requirements would limit the number of potential variants that could be expressed by segmental gene conversion. In contrast, at least a minimal structure is required to ensure that the expressed variant domains remain functional and surface exposed. We and others have examined this relationship between antigenic variation and structural requirements using Anaplasma marginale Msp2 as a model (5, 8, 9).
Msp2 is highly conserved in the hydrophobic N-terminal and C-terminal domains among alleles within a strain and across strains, with essentially complete identity from amino acids 1 to 159 and 224/231 to 377/386 (4, 5, 10). In contrast, the identity of the surface-exposed domain encoded by alleles within the reference St. Maries strain ranges from 41 to 70% between any two alleles and 39% among all alleles (10,–12). Within the approximately 70 amino acids composing the hypervariable region (HVR), variation is highly concentrated in three microdomains (8). These hypervariable microdomains correspond to sites of segmental gene conversion, and analysis of expressed variants reveals that the expression site HVR is highly permissive in allowing combinations of microdomains from different donor alleles (5).
Importantly, these microdomains are separated by short stretches of amino acids shown to be conserved among >1,600 expression site variants (5, 8). These conserved residues have been designated “structural tethers” and are hypothesized to allow maintenance of the Msp2 structure and porin function in the face of overall HVR variation (8). Using the I-Tasser suite (13,–15), we generated in silico secondary structure models of A. marginale Msp2 variants encoded by the five unique alleles in the St. Maries strain (10) and by an expression site mosaic assembled from two distinct alleles, 1 and 9H1. The predicted Msp2 secondary structures for the HVR were predominantly a random coil with structured elements, α-helices, and β-sheets concentrated in the flanking N- and C-terminal conserved domains (Table 1). The predicted proportion of random coil in the HVR, 82%, was significantly greater than that in the structured conserved domains (Tukey 95% simultaneous confidence intervals). Based on this in silico prediction, we hypothesized that the Msp2 HVR is structurally permissive and that this permissiveness is maintained in Msp2 generated by segmental gene conversion. In this work, we tested this hypothesis using native Msp2 and discuss the results in the context of the antigenic variant repertoire.
To identify A. marginale strains expressing the native 9H1/1 mosaic variant, DNA was isolated from bacteremic calves using a DNeasy blood and tissue kit (Qiagen), and the msp2 expression site was amplified and sequenced. Protocols were approved by the Washington State University Institutional Animal Care and Use Committee. The expression site was first amplified (forward primer, 5′-GCGGTTGCACCAATCCATT-3′; reverse primer, 5′-GCCCTGCTTGGTGACGACT-3′) using AccuPrime Pfx SuperMix (Thermo Fisher Scientific) with the following conditions: an initial hold at 95°C for 5 min, followed by 35 repetitive cycles (95°C for 40 s, 60°C for 40 s, 68°C for 60 s) and a final elongation step at 68°C for 10 min. The PCR product was purified using a Qiagen MinElute PCR cleanup kit, and a second PCR was performed (forward primer, 5′-TATGATACAGCAAGAGGTCAGGT-3′; reverse primer, 5′-CCCCAGCAACTATGGCCTT-3′) with Pacific Biosciences-designed 5′ sequence tags (see Table S1 in the supplemental material) to allow multiplexing. The product of the second PCR was purified using Agencourt AMPure XP magnetic beads (Beckman Coulter), used to construct libraries, and sequenced using the Pacific Biosciences platform following the manufacturer's instructions.
An initial screen to identify an appropriate detergent for Msp2 isolation was performed. All procedures were carried out on ice with centrifugations at 4°C. A. marginale-infected cells were pelleted by centrifugation in phosphate-buffered saline (PBS), and the pellet was solubilized in 10 mM Tris, pH 7.5, containing either 0.15% Triton X-100 (Sigma), 0.5% n-dodecyl-β-d-maltopyranoside (Anatrace), or 2% β-octylglucoside (Anatrace). Following centrifugation, 10 μl of supernatant was loaded into a precast Bolt 4 to 12% gradient polyacrylamide gel (Thermo Fisher Scientific), electrophoresed, and transferred to a polyvinylidene difluoride (PVDF) membrane for Western blot detection using the anti-Msp2 monoclonal antibody 115/362.17.19. Bound antibody was detected using peroxidase-labeled goat anti-mouse IgA, IgG, and IgM (H+L) antibody (KPL) with chemiluminescence using a Pierce ECL Plus Western blotting substrate (Thermo Fisher Scientific).
A. marginale isolates expressing the St. Maries strain 9H1/1 Msp2 mosaic were recovered from bacteremic blood previously collected from four individual infections (C-31024, C-31019, C-82198, C-1454). Following removal of hemoglobin by repeated washing with PBS and centrifugation at 30,000 × g for 15 min, the preparation was sonicated using four 30-s pulses with a Branson digital sonifier with a 400-W maximum output at 80% intensity. Following sonication, A. marginale isolates were recovered by centrifugation at 30,000 × g for 15 min, washed 2 to 3 times, and resuspended in 10 mM sodium phosphate buffer (pH 7.4) containing 0.1% sodium N-lauroyl sarcosine (sarcosyl; Sigma), and the suspension was incubated for 30 min at 37°C and 400 rpm in a ThermoMixer. The isolation and purification of Msp2 were performed following the protocol described for the isolation of Msp2/P44 from A. phagocytophilum (16) with modifications. The membranes were centrifuged at 10,000 × g for 1 h, and the pellet was washed twice in the same sarcosyl-containing buffer, followed by two washes with 10 mM Tris-HCl (pH 7.5) and centrifugation at 20,000 × g for 5 min to remove the sarcosyl. The pellet was resuspended in 2% β-octylglucoside and incubated in a 37°C water bath for 10 min, followed by centrifugation at 18,000 × g for 30 min at 4°C. The supernatant containing solubilized A. marginale outer membrane proteins was filtered using a 0.22-μm-pore-size filter, and Msp2 was purified by high-performance liquid chromatography (HPLC). HPLC was performed using an Äkta Avant system (GE Healthcare) coupled with an S-200 Increase size exclusion chromatography column that had previously been equilibrated with 50 mM sodium phosphate (pH 7.0), 0.5 M NaCl, and 1% β-octylglucoside. Elution was performed at a 500-μl/min flow rate with an equilibration phase for 1.2 column volumes followed by an elution step for 1.5 column volumes. The per run loading was limited to 150 μl and 250 μg of sample to achieve an optimal compromise between recovery and purity. Eluted proteins were detected by determination of the absorbance at 280 nm. Data collection and analysis were done using Unicorn (version 7.0) control software (GE Healthcare). Samples were analyzed by SDS-PAGE and Western blotting (see above for the procedure), and the fractions containing purified Msp2 were pooled, the protein concentration was determined by a bicinchoninic acid assay, and the protein was concentrated in a 30-kDa Centricon device to a target concentration of 1 mg/ml. Finally, samples were dialyzed, using a 20-kDa membrane, against a pH 7.5 10 mM Tris buffer with 1% β-octylglucoside, and the protein concentration was determined as explained above. The identity of Msp2 was confirmed by excising Coomassie-stained bands from the SDS-polyacrylamide gels and having them analyzed by liquid chromatography-tandem mass spectrometry at the Washington State University Molecular Biology and Genomics Core.
Porin function was assessed using an in vitro proteoliposome swelling assay as previously described in detail (16,–19) with the minor modifications described below. Liposomes were prepared by mixing 2.4 μmol egg phosphatidylcholine (Avanti Polar Lipids) with 0.2 μmol dicetylphosphate (Sigma), both of which were dissolved in chloroform-methanol (2:1, vol/vol), and the mixture was dried with rotation under a nitrogen stream. The lipid films were then further dried for 1 h in a vacuum desiccator. The dried film was resuspended in 300 μl of benzene, dried under a nitrogen stream and then for 1 h in a vacuum desiccator, resuspended in 500 μl of ethyl ether, and dried under a gentle nitrogen stream to make a homogeneous and smooth film, followed by an additional 1 h of drying in a vacuum desiccator. All drying steps in the vacuum desiccator were performed in the dark. The films were then resuspended in water containing up to 20 μl purified protein to a final volume of 200 μl or in water without protein as a negative control. The resuspended mixture was sonicated (Branson model 3510; Branson Ultrasonic) until the solution was translucent. The samples were then dried in a vacuum desiccator. Proteoliposomes were made by adding 300 μl of 10 mM Tris-HCl buffer, pH 7.5, containing 15% (wt/vol) dextran T-40 (Pharmacosmos) to the dried film, followed by incubation at room temperature for 30 min and a final resuspension. The proteoliposomes were mixed with stachyose, sucrose, glucose, or arabinose in a quartz cuvette. Swelling was determined by measuring the shift in the optical density at 400 nm every 2 s for 150 s. Changes in the optical density were recorded using a SpectraMax Plus 384 plate reader (Molecular Devices) at 25°C running SoftMax Pro software (version 6.0).
HPLC-purified Msp2 was either held at room temperature or heated at 95°C for 5 min and then loaded into a precast Bolt 4 to 12% gradient polyacrylamide gel (Thermo Fisher Scientific) and electrophoresed. The gels were either stained with Coomassie blue or used to transfer proteins to a PVDF membrane for Western blot detection as described above.
The circular dichroism (CD) spectra of proteins at 0.1 mg/ml in 10 mM Tris, 1% β-octylglucoside, pH 7.5, were measured at wavelengths of between 260 and 190 nm and 20°C in 1-mm cuvettes using an Aviv model 400 spectropolarimeter. The mean residue ellipticity was calculated using an average molecular mass of 39 kDa and an average residue number of 381, which represent the allelic average. Experimental CD spectrum data were analyzed with BeStSel software (http://bestsel.elte.hu/) for secondary structure determination (20).
The amino acid sequence of full-length Msp2 without the signal peptide encoded by the five unique conserved alleles and 9H1/1 mosaic was uploaded into the I-Tasser suite (http://zhanglab.ccmb.med.umich.edu/I-TASSER/) for detailed modeling of secondary and tertiary structures (13 – 15). The generated .pdb files were then uploaded into the Polyview-3D server for the generation of tridimensional models with emphases on the secondary structure and highlighting of the Msp2 HVR (http://polyview.cchmc.org/polyview3d.html) (21). Disorder disposition was calculated using data from the DisProt database (http://www.disprot.org/) (22).
The detergent used in the extraction and purification of outer membrane proteins is essential for the yield and stability required for structural analysis. We tested three detergents commonly used for isolation of outer membrane proteins: Triton X-100, a nonionic detergent; n-dodecyl-β-d-maltopyranoside, a nonionic maltoside; and β-octylglucoside, a nonionic glucoside. We tested for temperature-dependent gel mobility shift upon heating, as porins kept at room temperature migrate faster (due to incomplete unfolding) on SDS-polyacrylamide gels than boiled porins (18, 19). All three detergents properly solubilized Msp2 and demonstrated the temperature-dependent shift upon boiling (Fig. 1). However, only β-octylglucoside was able to maintain the protein in the folded form. As a result, this detergent was used for all further extractions.
Msp2 purification was performed by HPLC coupled with a high-resolution size exclusion column. Only the central fraction of the main peak contained purified Msp2 (see Fig. S1 in the supplemental material); mass spectrometric analysis revealed that the leading shoulder from the main peak contained the 61-kDa Am 780 (Mascot score of 359 with six peptides detected and a coverage of 21%), previously annotated to be a hypothetical protein, and the trailing peak that overlaps the Msp2 peak contained the 30-kDa Msp4 (Mascot score of 246 with two peptides detected and a coverage of 13%). The overall shape and number of peaks observed in the chromatograms from four purified biological replicates (C-31024, C-31029, C-82198, C-1454) showed high concordance, indicating that the purification procedure is reproducible. Consistent with this purification, only a single band from the main peak was visible following SDS-PAGE and Coomassie blue staining (Fig. 2A). The band was excised and identified to be Msp2 by mass spectrometry (Mascot score of 2,703 with six peptides detected and a coverage of 39%). The Msp2 identity was further confirmed by Western blotting using anti-Msp2 monoclonal antibody 115/362.17.19, which recognizes an N-terminal domain conserved among all variants. The temperature-mediated mobility shift was also observed, with the room temperature sample showing a single band at approximately 32 kDa and the heated sample showing a single band with an apparent molecular mass of 41 kDa (Fig. 2B). By Western blotting we observed a difference in Msp2 band intensity between the folded 32-kDa protein and the unfolded 41-kDa protein at equal protein loadings. This likely reflects a lack of monoclonal antibody binding to the properly folded protein, consistent with the epitope being located in the hydrophobic N-terminal domain.
The spectral profile of purified Msp2 analyzed by circular dichroism analysis is indicative of a properly folded protein rich in β-sheets with a minimum UV spectrum at approximately 218 nm (Fig. 3). Detailed determination of the secondary structure was performed by analyzing the spectra using BeStSel software (20) and revealed that 30% of the content corresponded to β-sheets, with 13 to 14% being α-helices and 40% of the total secondary structure content being random coils (Table 2). Analysis of the four biological replicates showed both highly similar spectra and highly similar proportions in the content of secondary structures (Fig. 3 and Table 2). As the actual secondary structure content paralleled the I-Tasser suite-based prediction (13,–15), we determined the distribution of the Msp2 secondary structure within the HVR and the conserved N-terminal and C-terminal domains using I-Tasser. The central HVR is mostly random coils flanked immediately by α-helices and then by the N-terminal and C-terminal domains, which are composed predominantly of β-sheets intercalated by random coil segments (Fig. 4A). Most notably, the small β-sheet and α-helix regions within the HVR significantly associate (P = 0.02, two-tailed chi-square test) with the genetically defined conserved tethers that flank the immunologically relevant hypervariable microdomains within the HVR (Fig. 4B).
I-Tasser predicts five different possible tertiary structures on the basis of secondary structure analysis (see Fig. S2 in the supplemental material). Msp2 has been proposed to function as an outer membrane porin (10), which is supported by three Msp2 properties characteristic of Gram-negative bacterial outer membrane porins: (i) the presence of a C-terminal phenylalanine (10, 23), (ii) the temperature-dependent mobility shift (Fig. 1 and and2),2), and (iii) solubility in β-octylglucoside (Fig. 1 and and2).2). In order to be able to make a structure-function correlation and confirm the most probable tertiary structure, we performed functional porin assays on the purified and properly folded native Msp2. The results show the combined data for four biological replicates performed in triplicate (Fig. 5A). Msp2 demonstrated porin function with a statistically significantly different diffusion of arabinose (150 Da), glucose (180 Da), and sucrose (342 Da) compared with that for the liposome controls (Fig. 5B). There was an association of diffusion uptake with solute size: the smaller that the solutes were, the faster that the uptake was (Fig. 5B). There was no significant uptake of the stachyose, the largest solute (666 Da). Using the confirmed porin function, .pdb files for three-dimensional Msp2 structures were derived using the I-Tasser suite and processed using the Polyview-3D server to highlight the HVR. The structures consistent with porin function as well as surface exposure, previously demonstrated both by surface labeling and by antibody binding, for the HVR derived from each of the alleles and the 9H1/1 mosaic are depicted in Fig. 6 (24,–26).
The empirically determined native Msp2 structure supported the I-Tasser suite-predicted secondary structure. Importantly, the native HVR was confirmed to be highly enriched in random coils, with the sparse structured elements aligning with the conserved residues previously predicted to be structural tethers on the basis of sequence analysis of >1,600 variants (8). Although the predicted tertiary structure has not been confirmed by analysis of the Msp2 crystal structure, which would require stabilization by binding with an HVR-directed antibody or specific ligand, the accuracy of the models (Fig. 6) is supported by the concordance of the native Msp2 porin function with the predicted pore formation and of the HVR surface exposure with prior mapping using isotopic labeling and antibody binding (26,–30).
The conserved N- and C-terminal domains and the HVR tethers provide the overall Msp2 protein structure, allowing consistent surface exposure of the HVR microdomains but plasticity within the microdomains themselves. This is critically important, as the overwhelming majority of Msp2 variants during persistent infection are mosaics, derived by segmental gene conversion from different alleles, and thus have not been under long-term selection to encode a functional and structurally fit protein (4, 5, 10, 12, 31). The native Msp2 tested for porin function in the current study is a mosaic with microdomain 1 derived from allele 1 and microdomains 2 and 3 from allele 9H1. This retention of function while bringing in antigenically variant microdomains illustrates the important of permissiveness in the HVR.
The HVR microdomains can be classified as intrinsically disordered regions (Fig. 7) (22, 32, 33). A similar structure has been shown to allow the antigenically variant erythrocyte membrane protein-1 from Plasmodium falciparum to maintain effective receptor binding, despite the protein sequence variability that results from the 60 donor alleles. The P. falciparum-CD36 interaction is mediated by the presence of two disordered regions whose loose tertiary structure allows plasticity sufficient for receptor binding and cytoadhesion (34). This is consistent with the role of disordered protein regions in eukaryotic cell signaling, where they have been shown to mediate highly specific but low-affinity interactions (33, 35). This has clear relevance for the development of antibodies to Msp2 variant microdomains, which are present at high titers but transient (30, 36). The continuous generation of high numbers of low-affinity antibodies, due to the flexible tertiary structure of the microdomains, may preclude B-lymphocyte receptor affinity maturation and the generation of plasma cells producing high-affinity antibodies and high-affinity memory. This is consistent with two observations: (i) HVR microdomain-specific antibodies are short-lived (36), and (ii) individual Msp2 variants can be expressed more than once in persistent infection following the decline of specific antibody (30). Thus, the random coil structure of the Msp2 HVR may expand the potential variant repertoire both by allowing mosaic formation and by allowing reexpression of the same variant due to low-affinity binding.
Msp2 has been predicted to be a porin, in agreement with the function of its orthologue, A. phagocytophilum Msp2/P44 (16). In the present study, porin characteristics were demonstrated for the native protein, including heat modifiability and detergent solubility, and its function was confirmed using the liposome swelling assay (37,–39). The predicted tertiary structure indicates an eight-strand β-sheet barrel. This in silico prediction is in agreement with the results obtained in the liposome swelling assay, which are most consistent with a slow diffusion porin, as opposed to a general diffusion porin (17, 40,–42). Both the experimental results and the structural analysis suggest that Msp2 is similar to the proteins belonging to the large OmpA family: OmpA members are surface exposed, heat-modifiable outer membrane proteins with an eight-strand antiparallel β-sheet barrel (17, 43). In contrast, archetypical general diffusion porins, like Escherichia coli OmpF, have 16-strand β-sheet barrels and a high efficiency for the translocation of nutrients (37). The diffusion rates of E. coli OmpA are much lower than those observed for OmpF, in agreement with the results obtained for Msp2 (17, 42).
The fitness trade-off between a consistent outer membrane protein structure for optimal function and the capacity to alter a critical epitope conformation underlies the host-pathogen interaction. While a highly functional protein may facilitate rapid bacterial growth and high-level bacteremia, these characteristics constrain the ability to evade an effective immune response. In contrast, plasticity may limit bacterial growth fitness. Anaplasma, among other bacterial pathogens, has evolved to balance these competing pressures to establish long-term persistence in the host yet to do so at levels that allow successful onward transmission.
This research was supported by National Institutes of Health grant R37 AI44005. Telmo Graça was partially supported by scholarship SFRH/BD/68377/2010 through the Fundação para a Ciência e Tecnologia.
We thank Hiroshi Nikaido and Etsuko Sugawara at the University of California for their technical advice and critical interpretation of the liposome swelling assay and David Herndon at USDA, ARS, ADRU for technical support on sequencing.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00391-16.