(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Mechanism of Borrelia immune evasion by FhbA-related proteins

['Konstantin Kogan', 'Hilife Institute Of Biotechnology', 'University Of Helsinki', 'Helsinki', 'Karita Haapasalo', 'Department Of Bacteriology', 'Immunology', 'Tommi Kotila', 'Robin Moore', 'Faculty Of Biological']

Date: 2022-05

Immune evasion facilitates survival of Borrelia, leading to infections like relapsing fever and Lyme disease. Important mechanism for complement evasion is acquisition of the main host complement inhibitor, factor H (FH). By determining the 2.2 Å crystal structure of Factor H binding protein A (FhbA) from Borrelia hermsii in complex with FH domains 19–20, combined with extensive mutagenesis, we identified the structural mechanism by which B. hermsii utilizes FhbA in immune evasion. Moreover, structure-guided sequence database analysis identified a new family of FhbA-related immune evasion molecules from Lyme disease and relapsing fever Borrelia. Conserved FH-binding mechanism within the FhbA-family was verified by analysis of a novel FH-binding protein from B. duttonii. By sequence analysis, we were able to group FH-binding proteins of Borrelia into four distinct phyletic types and identified novel putative FH-binding proteins. The conserved FH-binding mechanism of the FhbA-related proteins could aid in developing new approaches to inhibit virulence and complement resistance in Borrelia.

Relapsing fever and Lyme Disease are infectious diseases caused by borrelia bacteria. Relapsing fever occurs sporadically worldwide, whereas distribution of Lyme Disease is restricted to the Northern Hemisphere. Both infections are transmitted to humans by blood eating ticks or lice. These infections are often difficult to diagnose due to nonspecific symptoms. To be able to cause infection, borrelia must circumvent the human immune responses. Here we describe a mechanism, how borrelia bacteria protect themselves in the human host by utilizing host proteins. By using X-ray crystallography, we solved the structure of an outer membrane protein FhbA from a relapsing fever causing borreliae, Borrelia hermsii, in complex with human complement regulator factor H. FhbA has a unique alpha-helical fold that has not been reported earlier. The structure of the complex revealed how FhbA binds factor H in a very specific manner. Factor H bound to FhbA on the surface of borrelia protects bacteria from the complement system and lysis. Based on the structure, we performed structure-guided sequence database analysis, which suggests that similar proteins are present in all relapsing fever causing borrelia and possibly in some Lyme disease agents.

Funding: This study was supported by grants to TM from the Biomedicum Foundation and Academy of Finland (grant 324236), to AG from the Sigrid Juselius Foundation, the BBSRC (grant number 104399), the Wellcome Trust (grant number 478571), to PL from Academy of Finland (grant 272130). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability: The crystal structure has been deposited to Protein Data Bank (PDB) under access code 6ZH1. Crystallographic experimental raw data are available from the PDB. All other data are within the manuscript and in Supporting Information files.

Copyright: © 2022 Kogan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

By applying X-ray crystallography, we determined the structure of FhbA from B. hermsii in complex with its main ligand, FH domains 19–20, revealing a novel fold. The binding mechanism obtained from our co-crystal structure was confirmed by structure-guided mutagenesis and subsequent binding assays. We show that serum sensitive E. coli expressing functional wild type FhbA protein on the outer membrane surface were protected against complement, whereas bacteria expressing mutant FhbA protein unable to bind FH were as sensitive to complement as E. coli. We also demonstrated that the ability of FhbA to bind FH through this structural mechanism is important in complement evasion of live B. hermsii in a whole blood assay. Finally, by sequence comparisons and secondary structure predictions, we identified a putative new family of FH-binding proteins in both RF and LD borreliae. By analysing a previously uncharacterized member of this family from B. duttonii (BdFhbA), we verified the conservation of the unique FH-binding fold.

Binding of FH has also been described for RF agents [ 28 ], and a ligand for FH in B. hermsii was shown by Hovis et al. to be a plasmid-encoded, surface-exposed Factor H binding protein A (FhbA)[ 29 ]. FhbA from B. hermsii (BhFhbA) was previously shown to interact with FH via domain 20 [ 30 ] and it was also reported to bind FHL-1 [ 31 ]. Other FH-binding proteins described from RF spirochetes include BpcA from B. parkeri [ 32 ], HcpA from B. recurrentis [ 33 ] and CbiA from B. miyamotoi [ 34 ]. However, the mechanisms by which these proteins bind FH have remained elusive.

In addition to binding FH, several other complement avoidance mechanisms have been described in borreliae (reviewed e.g., in [ 1 ]). For example, Bo. burgdorferi proteins BBK32 [ 24 , 25 ] and OspC [ 26 ] prevent classical pathway activation, and BGA66 and BGA71 of Bo. bavarensis inhibit the terminal pathway [ 27 ].

One of the best characterized complement evasion mechanisms, in spirochetes as in other microbes, is the utilization of FH. In LD-causing borreliae, several FH-binding proteins have been described (reviewed in [ 20 ]). Proteins can be grouped into three different classes by sequence similarity. First class is CspA-related proteins, where the canonical protein is CspA (or BBA68, Complement regulator-acquiring surface protein, BbCRASP-1) from Bo. burgdorferi. The second entity is the CspZ-related proteins, where the canonical protein is CspZ (BBH06/BbCRASP-2) from Bo. burgdorferi. Both groups of proteins bind FH and FHL-1. The third group is very large, as it contains several proteins from the OspE/OspF-related family. All proteins, which have been analysed in detail from the third group bind FH via the common microbial binding site on domain 20 of FH. Structures of CspA [ 21 ], CspZ [ 2 ], as well as OspE [ 22 ] and its homologues ErpC and ErpP [ 23 ] have been determined. The crystal structure of the OspE:FH19-20 complex showed in detail how OspeE bound FH [ 22 ].

For extracellular microbes, evasion of complement is a prerequisite for infectivity. Spirochetes as diderm bacteria are structurally vulnerable to complement mediated lysis [ 19 ], thus for them complement evasion is a fundamental requirement. During a blood meal in the vector’s midgut, and during the dissemination phase via skin and blood in humans, borreliae are unavoidably in direct contact with the complement system.

In the complement system, there are also FH-related proteins. Factor H-like protein 1 (FHL-1) is an alternatively spliced transcript of the cfh gene. FHL-1 consists of FH domains 1–7 with a unique C-terminal tail of four amino acids, and it has similar regulatory functions as FH [ 14 ]. In addition, there are five Factor H-related proteins (FHR1-5) encoded by their own genes. They contain four to nine SCR-domains, some with high homology to FH-domains [ 17 ]. The exact functions of FHRs are presently not known, but they may regulate FH by preventing its binding to ligands [ 18 ].

The alternative pathway (AP) of the complement system is the first innate immune defense mechanism targeting microbes. Activation begins by attachment of C3b proteins to the surface of a microbe. Surface-bound C3b-molecules form the basis for complement amplification leading to formation of the lytic membrane attack complexes. Complement activation needs to be tightly controlled to prevent harmful effects on the host. Many complement proteins are regulators of the system on surfaces and in the fluid phase (for a review, see [ 13 ]). The main AP regulator in serum is Factor H (FH), which consists of 20 globular short consensus repeat domains called SCR or FH-domains. FH regulates AP efficiently by three different mechanisms, all of which require that domains 1–4 of FH are bound to C3b [ 14 – 16 ].

Relapsing fever is caused by many species of borreliae that are present worldwide in warm and temperate regions [ 7 ]. Individual species have restricted geographical distributions. In the US, RF outbreaks are mainly caused by B. hermsii and B. turicatae [ 8 ]; whereas in Africa they are caused by B. duttonii and B. crocidurae [ 9 ]. Clinically, RF is characterized by recurring high fever episodes accompanied by nonspecific symptoms. The reason for recurrent fever episodes is antigenic variation: the ability of RF borrelia to change variable outer surface proteins to evade antibody recognition. Neurological manifestations, myocarditis and development of organ failure in kidneys, spleen and lungs in severe cases may follow [ 10 , 11 ]. Without adequate treatment, mortalities of up to 5% in epidemics of tick-borne RF have been described [ 12 ].

Lyme disease is mainly caused by three different genospecies of Borreliella (Bo. burgdorferi, Bo. afzelii and Bo. garinii). Bo. burgdorferi is found in North America, whereas all three exist in Europe [ 4 ]. In the US there are 42,000 ( https://www.cdc.gov/lyme/ ) and in Europe 70,000 cases of LD reported each year [ 5 ]. In LD, bacteria initially affect skin around the tick bite causing the most common sign of the infection, a ring-like rash called erythema migrans. Systemic manifestations of disseminated infection include neuroborreliosis, carditis and arthritis [ 6 ].

Lyme disease (LD) and relapsing fever (RF) are infections caused by borreliae spirochetes. Bacteria are transmitted to humans from vertebrate animals by blood feeding arthropods. RF and LD borreliae have different vectors. All LD borreliae are transmitted by Ixodes ticks and the majority of RF borreliae by soft bodied Ornithodoros ticks. Borreliae differ also in their outer membrane composition, genomic features, selection of vertebrate hosts, and immune evasion mechanisms [ 1 ]. Due to differences between the two groups, the division of genus Borrelia into two was recently suggested: RF-causing species would constitute the genus Borrelia, and LD-causing species the genus Borreliella [ 2 ]. In this publication, we chose to use the most recent nomenclature, even though the scientific community has not reached full consensus on the subject [ 3 ].

Taken together, FH binding proteins clearly fall into four groups, which also coincide with the phylogenetic classification of the borreliae species. All LD clade species with FH-binding proteins have CspA/CRASP-1 proteins, and the majority also possess CspZ/CRASP-2 and OspE/CRASP-5 -proteins. RF clade species exclusively have FhbA proteins and lack the other three, suggesting a significant role for FhbA proteins in complement evasion of RF borreliae.

There were two species that lacked all four classes of FH-binding proteins ( Table 3 ). The first is Bo. chilensis, which was originally isolated from ticks [ 47 ], and currently there is no data about its vertebrate host. The second is B. anserina, a bird isolate. Although another bird infecting species, Bo. garinii, was shown to bind avian FH [ 51 ], it is not known whether Bo. anserina binds FH at all.

We delineated the full spectrum of potential FH-binding proteins in LD and RF clades using the same set of genomes as for the FhbA searches to identify homologues for FH-binding proteins from Bo. burgdorferi. We used sequences from the known structures of CspA/CRASP-1[ 49 ], CspZ/CRASP-2[ 50 ] and OspE [ 22 ] in the search. Ten of the eleven genomes analyzed in the LD clade have proteins homologues to CspA/CRASP-1, six of eleven to CspZ/CRASP-2 and ten of eleven to OspE/CRASP-5 (in Bo. bavariensis the protein has an N-terminal region displaying 30% identity to OspE (Tables 3 and S3 ). In contrast, the members of the RF clade have neither CspA/CRASP-1 nor CspZ/CRASP-2 homologues.

The most extensively studied species from the LD clade, Bo. burgdorferi, has five different FH-binding proteins: CspA/BBA68/BbCRASP-1, CspZ/BBH06/CRASP-2, ErpP/BBN38/OspE/CRASP-3, ErpC/CRASP-4, ErpA/BBP38/BBL39/OspE/CRASP-5, with the last three being highly homologous to each other [ 27 ]. Below we use gene name as the protein name, as it is the accepted practice for microbial proteins, along with the well-known general name (complement regulatory acquiring surface protein: CRASP).

Together, these results clearly show that two homologous proteins, BhFhbA and BdFhbA, bind FH19-20 via a novel, conserved mechanism. Based on structure-assisted multiple-sequence alignments, we predict that other members of the family utilize the same mechanism.

We next measured binding of FH19-20 to BdFhbA. Gel filtration experiment showed that the proteins form a stable complex (Figs 5C and S9 ). Moreover, MST experiments revealed that FH19-20 bound BdFhbA ( Fig 5D ), although with lower affinity (1.06 ± 0.38 μM) than to BhFhbA ( Fig 2 and Table 2 ). The lower affinity might be due to small differences in the hydrophobic binding pocket. For example, BhFhbA Met137, which sits in the hydrophobic cavity, is replaced by Thr in BdFhbA. We also mutated the key phenylalanine (corresponding to Phe154 in BhFhbA) in BdFhbA to alanine (Phe130Ala). As expected, both the gel filtration (Figs 5C and S9 ) and MST assays ( Fig 5D ) showed a drastic decrease in the binding of FH19-20 to BdFhbA/Phe130Ala. We also showed that full length FH binds wild type BdFhbA but not the BdFhbA/Phe130Ala mutant ( S10 Fig ) and retains its cofactor activity in FI-mediated cleavage of C3b ( Fig 5E ). Finally, BdFhbA binds FH predominantly via domain 20 and possesses only weak affinity towards FH5-7 ( Fig 5F ), like BhFhbA. CD spectrometry showed that both wild-type BdFhbA and BdFhbA/Phe130Ala were correctly folded ( S7 Fig ).

(a) Sequence alignment of FhbA proteins from B. duttonii and B. hermsii shows a difference in the region between helices 3 and 4 (dashed red rectangle). (b) Structure-based homology model of BdFhbA shows almost identical overall structure to BhFhbA except shortening of helices 3 and 4. (c) Elution profiles of the proteins alone and in combination with each other on the size exclusion column. FH19-20 protein elution profile in magenta, wild type and Phe130A variant BdFhbA in light blue, and a combination of the two proteins in black curves. In each chromatography run, 100 μl of sample containing 20 nmole of each of the tested protein(s) was injected. (d) Binding of FH19-20 to BdFhbA and BdFhbA/Phe130A variant as detected by MST. K d -values with SEM (μM) calculated from three technical replicates (normalized fluorescence values) are shown for the wild type BdFhbA. For the BdFhbA/Phe130A variant, the MST software could not reliably determine a value for K d . (e) Cofactor activity of FH in factor I mediated cleavage of C3b. FH from serum or after purification was bound to BdFhbA coated on microwell plate. After washing, C3b and/or Factor I were added. C3b and its cleavage fragments (α’46 and α’43) were detected using anti-C3d antibody. (f) Binding of FH19-20 and FH5-7 to surface coated BdFhbA. Individual data points from ELISA are shown, and the bars represent mean values.

To examine if the BhFhbA binding mechanism we identified occurs in other members of this protein family, we cloned, expressed and purified a novel FhbA homologue from B. duttonii (BdFhbA). We chose this protein, because B. duttonii is known to bind FH [ 28 ]. BdFhbA protein has high conservation (41.8% identity) to BhFhbA ( Fig 5A ). We modelled the structure of BdFhbA with Phyre2 using our crystallized BhFhbA as a template, and observed an almost identical structure, except for shorter helix 3 and its connection to helix 4, as expected from the sequence alignment ( Fig 5A and 5B ). The location of the helices, the positioning of the hinge area, and the orientation of the critical residue Phe130 (corresponding to BhFhbA Phe154 in B. hermsii) were conserved.

Overall, it appears that FhbA-like genes are present in all RF clade species, except for B. anserina, and that the key binding loop containing Phe154 is conserved. However, although FhbA-like genes are identifiable at the DNA level also in the LD clade as well, it is not known if they encode functional proteins.

The translated genomic regions of three species from the LD borreliae clade aligned well with other BhFhbA proteins, although with some changes ( Fig 4A ). In Bo. valaisiana, the protein is two helices shorter, but the hinge region and the key Phe154 residue mediating FH-binding are conserved. Bo. bissettiae (NC_015916.1) has a Leu instead of conserved key Phe, several stop codons, and a single nucleotide deletion that leads to a frame shift (marked by ‘?’ in Fig 4A ). The only sequence that seems not to be altered at the DNA level is from Bo. afzelii. Potentially it can be translated into functional protein, though it has Leu instead of conserved Phe, as in Bo. bissettiae. Further studies are required to test whether any of these genes are expressed and provide similar protection as FhbA.

(a) Multiple sequence alignment of the FhbA-related proteins. The signal sequence is marked in magenta and secondary structure elements derived from the crystal structure of BhFhbA:FH19-20 are shown in blue above the sequence. The conserved hinge region is marked with a red rectangle, and an arrowhead points at the key Phe residue important for tight binding to FH20. The sequence alignment consists of protein sequences, predicted protein sequences and translated genomic regions that match the BhFhbA used as the search sequence (see Materials and Methods for details). Asterisk ‘*’ represents a stop codon and question mark ‘?’ stands for an incomplete codon, where a frameshift appears to have occurred. In generating the protein alignment, the frameshift was ignored, and the translation frame was preserved to allow further protein alignment after that problematic codon. Sequence parts with grey background are translated genomic sequences, which most probably are not present in derived proteins due to stop codons or frameshifts. Accession numbers and references for the sequences used in the alignment: B. hermsii YOR (W5SB08), B. coriaceae Co53 (W5T1N6), B. duttonii CR2A (W6TXL9), B. parkeri (D5GU46), B. turicatae (B0L8C8), B. miyamotoi (A0A075BUA1), B. recurrentis (C1L349), B. persica No12 Bp4780 (contig: NZ_AYOT01000066.1, nucleotides: 4020 -> 4611), B. hispanica CRI Bhis_2727 (contig: NZ_AYOU01000105.1, nucleotides: 3668 -> 4235), B. crocidurae str. Achema (contig: NC_017778.1, nucleotides: 46553 -> 45974), B. valaisiana VS116 (C0R979), Bo. afzelii (WP_011703930.1), B. bissettiae DN127 (contig: NC_015916.1, nucleotides: 3284 -> 2685). (b) Phylogenetic tree of the FhbA-family proteins constructed from the sequences shown in (a). The numbers represent the substitutions per position and the length of the lines is equivalent to these numbers. The three colours emphasize that the sequences cluster into three distinct groups.

Our structure-guided sequence database analysis approach allowed us to reliably identify 10 species with sequences homologous to BhFhbA in RF borreliae clade, and 3 species in the LD group (Figs 4A and S8 ). A phylogenetic tree of these homologous proteins ( Fig 4B ) shows that they form three different clusters. Nonetheless, in all 10 species of RF borreliae, the residue corresponding to Phe154, which is essential for FH-binding in BhFhbA, is conserved. In B. crocidurae, both sequenced strains (Achema and DOU) have stop codons in the signal sequence that alter the amino terminal regions of the proteins. It would be interesting to see if these truncated genes translate into functional proteins and provide protection like FhbA protein from B. hermsii.

‘+’ means that the gene has been identified at least in one source.’–‘means that the sequence was not found in any of the sources used. Data were acquired from the sequenced genomes, published reports and individual sequences deposited to the databases. Relevant genes were searched as described in Materials and Methods. Accession numbers for FhbA-related proteins are listed in the legend to Fig 4 .

The crystal structure of BhFhbA inspired us to study the distribution of FhbA-like and other FH-binding proteins within the whole Borreliaceae family. First, we performed a thorough search for homologous proteins within all available whole genomes (in total 154) and separately deposited sequences from the family Borreliaceae ( Table 3 ). Analysing sequence data from borrelia is demanding, as Borreliaceae have very complex genomes with both linear and circular plasmids [ 45 ] where length, diversity and composition vary between different species. It is thus possible that some FH-binding proteins might be absent from the databases due to plasmid loss, which has been reported in the Borreliella clade.

We then expressed BhFhbA wild type and BhFhbA/Phe154Ala mutant on the outer membrane of E. coli to study how binding of FH affects bacterial survival. We chose E. coli as it is a Gram-negative bacterium, and laboratory strains lacking any evasion mechanisms are efficiently killed by complement [ 43 ]. We utilized the autotransporter adhesin involved in diffuse adherence-I (AIDA-I) system [ 44 ] ( Fig 3F ) to deliver the BhFhbA protein to the outer surface of E. coli and showed by immunoblot analysis that His-tagged BhFhbA was present in the outer membrane fraction ( Fig 3G ). We then performed serum sensitivity assays by utilizing BhFhbA wild type and mutant proteins (see Materials and Methods ). Survival of the strain expressing wild-type BhFhbA was significantly (p<0.05) higher when compared to the control strain and to the strain expressing BhFhbA/Phe154Ala mutant ( Fig 3H ). However, in the absence of complement, all three strains showed similar survival ( Fig 3I ). These results show that binding of functional BhFhbA to FH is necessary and sufficient for increasing the survival of the bacteria.

After entering the body, RF Borrelia can survive and multiply in the blood, and cause massive spirochetemia, where bacteria are present in the blood at high densities (10 5 −10 6 bacteria/ml). We mimicked natural conditions by incubating 50,000 live B. hermsii bacteria in 100 μl of whole blood treated with hirudin to prevent coagulation, and measured the amount of terminal complement complexes (TCC) as indicators of complement activation [ 42 ]. We first confirmed that, as expected, the presence of bacteria in blood increases complement activation, which is seen as an increase in the amount of TCC in the sample. Adding purified BhFhbA to the reaction to inhibit binding of FH to B. hermsii led to even higher levels of TCC compared to bacteria alone. Conversely, adding of the FH-binding defective mutant BhFhbA/Phe154Ala had no effect on the levels of TCC, which were similar to bacteria alone ( Fig 3E ). Some enhancement of complement activation was also detected when BhFhbA was incubated in the absence of bacteria, whereas BhFhbA/Phe154Ala had no effect. We hypothesize that high-affinity binding of BhFhbA to FH may affect its ability to regulate complement in the fluid phase and/or can lead to formation of complement activating immune complexes.

These results suggest that BhFhbA binds FH predominantly via domain 20. We cannot exclude that under some circumstances, BhFhbA also interacts with FHL-1 or FH via domains other than 20, but the major binding site for BhFhbA is clearly FH domain 20.

(a) Binding of FH19-20 and FH5-7 to immobilized BhFhbA detected by ELISA. Individual data points are shown with bars indicating mean values. Values shown in panels a-e are absorbances measured by ELISA. (b) Effect of FH5-7 on binding of FH19-20 to immobilized BhFhbA detected by ELISA. n = 5 or more. Error bars represent S.D. (c) Concentration-dependent binding of FH19-20 and FH5-7 to BhFhbA. Individual data points from several ELISA assays are shown. (d) Controls for FH5-7 fragment confirmed by ELISA. Binding to ApoE [ 65 ] protein and heparin were used as positive controls, and BSA as negative control. Individual data points from assays are shown, and bars indicate mean values. (e) Complement activation measured by formation of soluble terminal complement complexes (TCC) in whole blood. Proteins added to reactions are shown below the graph. Data are presented as relative TCC-amount (%) compared to the sample where bacteria alone were incubated in blood (n = 4, error bars represent S.D.) Difference between bacteria alone to bacteria incubated with wild type FhbA (* p<0.05) calculated by paired t-test (Mann-Whitney U-test for independent samples). (f) Schematic representation of the survival assay using the AIDA-1 transport system 44 , where binding of membrane expressed FhbA to serum FH protects E. coli against complement attack. (g) Western blot showing the presence of His-tagged BhFhbA on the outer surface of E. coli. Outer membrane sample of E. coli expressing an empty AIDA1 system shows a band present at about 63 kDa, consistent with the expected molecular weight (lane 2). BhFhbA wt and Fhe154Ala mutant proteins are present only in outer membrane (out) at the expected molecular weight of 83 kDa (lanes 5 and 8). The anti-His signal was not detected in the supernatant fraction (sup) or in the inner membrane (in). (h) Serum survival of E. coli clones expressing BhFhbA, BhFhbA mutant F154A and empty vector (AIDA) as a control. Result was calculated as a percentage of bacteria that survived after 15 min incubation in serum as compared to the number of bacteria at time point zero (n = 4, error bars represent S.D.) Difference between BhFhbA clone to control significant (* p<0.05) calculated by one-way ANOVA supplemented with Dunnet’s test for unequal variances. (i) Survival of E. coli clones expressing BhFhbA, BhFhbA mutant F154A and empty vector as a control in the presence of inactivated (with 10 mM EDTA) serum without complement. Result was calculated as a percentage of colonies surviving after 15 min incubation in media as compared to the number of bacteria at time point zero (n = 3, error bars represent S.D.).

BhFhbA was immobilized to ELISA-plates, and purified FH5-7 or FH19-20 fragments were added. After washing steps, the binding of FH5-7 and FH19-20 was detected by a polyclonal anti-FH antibody. When compared to FH19-20, the FH5-7 fragment displayed very modest binding to BhFhbA ( Fig 3A–3D ). Moreover, FH5-7 did not affect the binding of FH19-20 to BhFhbA in a competition assay ( Fig 3B ). To analyze the binding interaction in another setup, we used MST to test if FH5-7 or FH19-20 affects the binding of full-length FH to BhFhbA in a competition assay. FH5-7 slightly decreased the binding of full-length FH to BhFhbA, but FH19-20 abolished it completely ( S6B Fig ).

B. hermsii was originally reported to acquire complement regulator FHL-1 from serum, as well as to bind a cloned fragment of FH containing domains 1–7 [ 31 ]. Domains 6–7 are identical in FH and FHL-1 and contain a binding site for several microbial proteins, like fHbp of Neisseria, CspA of Lyme disease borrelia and streptococcal M-protein [ 41 ]. We therefore decided to test whether BhFhbA binds this region by using an FH fragment of domains 5–7 (FH5-7).

To exclude the possibility that the critical effect of the Phe154Ala mutation on binding to FH is due to protein misfolding, we performed CD-spectroscopy and demonstrated that wild type and Phe154Ala mutant proteins have identical secondary structure profiles, indicating that both are correctly folded ( S7 Fig ).

In the binding data, we observed a 12-fold and a 5-fold decrease in the affinities of two other mutants, Met137Ala and Ile171Ala for FH19-20, respectively ( Table 2 ). In the complex, Trp1183 from FH20 is elegantly slide in between the coordinated aromatic binding stack of four hydrophobic residues Met137, Leu146, Phe154, and Ile171 ( Fig 1B ), providing a structural explanation for these biochemical results. However, most single alanine substitutions of charged residues had no effect on binding, suggesting that hydrophobic interactions dominate. To confirm the dominance of hydrophobic interactions further, we compared complex formation in high-salt conditions (PBS + 500 mM NaCl) and physiological environment (PBS) by gel filtration ( S5 Fig ). High ionic strength did not disrupt the binding of BhFhbA to FH19-20, supporting our proposal that the interaction is mainly hydrophobic, although involvement of other charged residues in the interaction cannot be excluded.

Mean binding affinities (K d in μM concentration ± SEM) calculated from three individual experiments ( S4 Fig ). Fold decrease in binding was calculated by normalizing the K d values relative to wild type BhFhbA (assigned a value of 1). *Designates p-value < 0.05 from an unpaired t-test in comparison to the wild type. n = 3 for each protein.

We used affinity ligand binding immunoblot as an initial robust screen for the effects of mutations on the binding to FH19-20 ( S3 Fig ). Some mutations decreased binding, and one (Phe154Ala) completely abolished it. Next, we used fluorescence-based MicroScale Thermophoresis (MST) to determine the binding affinities of the BhFhbA mutants to FH19-20 ( Table 2 and S4 Fig ). MST measurements showed ~70-fold decrease in binding of the Phe154Ala mutant (K d = 6.0 μM) to FH19-20 compared to the wild type protein (K d = 0.082 μM) ( Fig 2C and 2D ). This was also confirmed by gel filtration chromatography, where no complex formed between the Phe154Ala mutant and FH19-20 ( S5 Fig ). In addition, we compared the binding of wild type BhFhbA and the Phe154Ala mutant to full length FH using MST. As with FH19-20, wild type FhbA bound FH with high affinity (K d = 30 nM) and Phe154Ala mutant showed no binding ( S6A Fig ).

(a) The sequence of BhFhbA from Borrelia hermsii YOR (UniprotID: W5SB08) with mutated residues marked in red. (b) Positions of the mutations mapped on the structure of the BhFhbA:FH19-20 complex. BhFhbA α-helices 1–5 are shown in green, 6–9 in teal (cartoon), and FH19-20 is shown as an electrostatic surface model. (c) Binding of FH19-20 to wild type BhFhbA and (d) Phe154Ala mutant BhFhbA detected by MST. K d -values ± SEM (μM) calculated from three technical replicates (normalized fluorescence values) and individual data points are shown.

Next, we designed ten alanine point mutations located on the interface of BhFhbA that binds FH19-20. A group of mutations (Asn153Ala, Phe154Ala, Met137Ala, Ile171Ala) targeted at the hydrophobic binding pocket, whereas the rest (Phe85Ala, Asn88Ala, Lys91Ala, Glu178Ala, Phe181Ala, Glu198Ala) were targeted to the interface below the hydrophobic cavity ( Fig 2A and 2B ). The single alanine mutants of BhFhbA were expressed in E. coli as 6x-His fusion proteins and purified using Ni-NTA and size-exclusion chromatography.

There are two crystal structures of microbial proteins complexed with FH19-20: OspE from Bo. Burgdorferi [ 22 ] and BhFhbA. We compared the binding mechanisms between the two proteins. Previous binding inhibition assays using 15 mutants of FH20 showed that the binding-sites of BhFhbA and OspE overlap [ 30 ] , and that BhFhbA inhibited binding of OspE to FH19-20 and vice versa. We superimposed the FH19-20s of the two complex structures to analyze the similarities and differences in the binding. We identified five key amino acids in FH20 that form hydrogen bonds to both OspE and BhFhbA ( S1 Table and S2A Fig ). Thus, BhFhbA and OspE have completely unrelated folds, but nevertheless utilize the same surface and partially similar contacts to bind FH20 ( S2A Fig ). Comparison to the structure of FH19-20 in complex with sialic acid and C3d (PDBID: 4ONT [ 40 ]) revealed that the same surface on FH20 is also occupied by sialic acid and, moreover, that FH20 W1183 makes similar interactions with the sialic acid moiety as with the F154 of FhbA ( S2B Fig ). This supports our conclusion that the primary interface seen in the crystal structure is the biologically relevant one.

A careful analysis of the primary interface reveals that α-helices 6–9 of BhFhbA form a hydrophobic cavity in which FH20 sits ( Fig 1B ). Within the cavity, Trp1183 of FH20 surrounded by a sandwich like stack of aromatic residues sits between two BhFhbA residues: Phe154 from the hinge region and Tyr170 from helix α8. This tightly constrained binding pocket is further coordinated by the hydrophobic sidechains of α6 Met137 and Leu146 and α8 Ile171. Moreover, towards the C-terminus of FH20, Tyr199 from BhFhbA α9 forms van der Waals interactions with FH20 Val1200, with the closest approach being 3.8 Å ( Fig 1B ).

There are two possible FH19-20-BhFhbA interfaces in the crystal structure ( S1 Fig ). The primary and biologically relevant interface in the FH20 domain has a buried surface area of 1066 Å 2 , as determined by the PISA server [ 39 ]. The other interface, located at the tip of domain 20 of FH, arises from crystal packing. Further, the primary, biologically relevant interface contains more specific contacts between FH and FhbA, as demonstrated by the increased number of hydrogen bonds when compared to the other interface (18 vs. 6, S1 Table ).

We performed a structure-guided search from the Protein Data Bank to analyze the conservation of the protein. Interestingly, we could not find any proteins with a similar fold either from PDBeFold or the Dali server [ 38 ]. The structure of BhFhbA is thus a previously uncharacterized fold and further represents a novel FH-binding scaffold.

(a) A cartoon representation of the BhFhbA:FH19-20 complex. BhFhbA forms a bundle of nine α-helices resembling an arc-like or L-shaped structure. α-helices 1–5 (in green) wrap around the core formed by α-helices 6–9 (in teal). The connecting loop of α-helices 7 and 8 create a hinge-like structure close to C-terminus of BhFhbA. Together the L-shaped cavity (α-helices 3, 8 and 9) and the hinge-loop coordinate the binding of FH20 (in blue). (b) Close-up of the binding site with key sidechains represented as sticks. FH20 Trp1183 and Val1200 (in blue, indicated by *) coordinate binding by a group of hydrophobic residues in BhFhbA. Trp1183 of FH20 is inserted to the hydrophobic pocket formed by BhFhbA Phe154, Met137, I171 and Tyr170. Further, FH20 Val1200 is within van der Waals distance of BhFhbA Tyr170 and Tyr199. 2Fo-Fc density, contoured at 1 σ, is shown around the residues.

The structure of BhFhbA (residues 44–202) reveals a compact, single-domain fold, which is composed solely of a bundle of nine α-helices ( Fig 1A ). The first five α-helices (α1-α5) wrap around the core formed by α-helices 6–9 with the overall architecture of the fold resembling a curved arc or capital letter ‘L’. Helices 6–9 are essentially perpendicular to the rest of the helices. α3, α6, α8 and α9 compose the inside face of the L, creating a cavity with a hinge-like loop on top, formed by the connecting region between α-helices 6 and 7.

To understand how FH binds to the microbial surface protein BhFhbA, we determined the crystal structure of BhFhbA in complex with the FH19-20 fragment at 2.2 Å resolution. The initial solution was obtained by molecular replacement with the published structure of FH19-20 (PDB 2G7I [ 35 ]) as a search model. We identified a single molecule of FH19-20 in the asymmetric unit with clear density for another polypeptide. Next, we assigned BhFhbA to the density by multiple rounds of manual building in Coot [ 36 ] and refinement with BUSTER [ 37 ]. The final solution showed a 1:1 complex of FH19-20 and BhFhbA in the asymmetric unit. The model was refined to a good overall geometry, and water molecules and ions were introduced to clear unassigned densities. The final model fits the observed diffraction data with the final R work /R free values of 19.7%/23.7% ( Table 1 ), and all residues are in the most favored (98.0%) or allowed (1.65%) regions of the Ramachandran plot.

Discussion

By determining a 2.2 Å resolution crystal structure of B. hermsii surface protein BhFhbA in complex with FH19-20, we revealed the molecular mechanism by these two proteins interact with each other. The structure, combined with mutagenesis and binding studies, led to the identification of a conserved aromatic residue, Phe154, that is central in binding FH. Using structure-guided sequence analysis with the structure of BhFhbA as a search model, we identified several putative homologous proteins in relapsing fever and Lyme disease borreliae. To confirm the common binding mechanism between the FhbA-related proteins and FH, we also expressed a novel FH19-20 binding protein (BdFhbA) from B. duttonii, which causes relapsing fever, and mutated the key residue, Phe130. MST and gel filtration demonstrated that these two proteins share the same FH-binding mechanism. We also delineated four different FH-binding proteins families of Borrelia/Borreliella and show that the LD clade of borreliae has four different groups of FH-binding proteins, but the RF borreliae clade appears to possess only one.

BhFhbA was originally identified as an FH binding protein from B. hermsii, which causes relapsing fever [29]. The protein was predicted to be composed of four α-helices flanked by three loops. When compared to our crystal structure, the locations of predicted coiled coils and loop regions match poorly. Thus, earlier random [31] and site-directed mutagenesis [52] studies aimed at the predicted loop regions of the protein also targeted α-helical and core regions of the protein, affecting secondary structure elements and protein folding. We mapped six previously published mutants with reduced or no FH-binding activity [31] to our structure (S11 Fig). Detailed inspection of the environment of each mutated position explains decreased binding. For example, the Asn172Thr mutation disrupts two hydrogen bonds that keep helices 2 and 3 together. Though Asn172 is located far from the active hinge region, such a mutation is likely to affect overall folding or stability of the protein. Nevertheless, these earlier studies support the importance of the hinge region in FH binding.

BhFhbA binds to domain 20 of FH (Figs 1A and S1). Structural analysis of FH19-20 complexes with two microbial proteins, FhbA and OspE from Bo. Burgdorferi [22] suggests that the general microbial binding site on FH20 mediates interaction (S1 Table). However, the interactions are different: a hydrophobic binding pocket is formed between FH20 and BhFhbA, whereas the FH20:OspE interaction is mainly electrostatic in nature and mediated by hydrogen bonds. Binding of FH to sialic acid on erythrocytes, endothelial cells and platelets has been shown to protect host cells from complement [53]. Interestingly, the structure of the FH19-20:sialic acid:C3d [40] complex revealed that the same general microbial binding site in FH20 is involved in binding to sialic acid. This is a rare and interesting example of convergent evolution of a binding site utilization; the binding ligands and mechanisms are different, but the binding patch site on FH largely overlaps.

The other important interaction site in FH is in domains 6–7, part of which is present in full-length FH as well as in FHL-1. Typically, microbes bind FH via domain 20, or FH and FHL-1 via domain 7. BhFhbA is a rare example, as it was first reported to bind FHL-1 and FH fragment 1–7 [54] and later FH via domain 20 [30]. Here, we examined if BhFhbA has two binding sites on FH by comparing the interactions of FH19-20 and FH5-7 to FhbA (Figs 3A–3D and S6). Our results suggest that FhbA has only weak affinity for FH5-7 and it cannot compete with FH19-20 in binding to BhFhbA (Figs 3B and S6). When all FhbA-related proteins analyzed so far are considered, this result is not surprising. No binding of FH1-7 to FhbA-related protein HcpA from B. recurrentis [33], to BpcA of B. parkeri [32] or to CbiA of B. miyamotoi [40] was observed. It cannot, however, be excluded that under certain circumstances, e.g., in specific tissue locations, RF borrelia acquire FHL-1 as well, but the most important binding site for this protein is on FH20.

Microbes bind domains other than N-terminal domains FH1-4 so that this region can bind to and downregulate C3b. It has been previously shown that FH bound to BhFhbA retains its cofactor-activity in cleaving C3b both using purified proteins [30] and when BhFhbA is expressed on the bacterial cell surface [55]. We confirmed that FH bound to BdFhbA retains its cofactor-activity (Fig 5E). Furthermore, we previously showed that FH20 bound to BhFhbA or other microbial proteins enhances the cofactor-activity of FH in cleaving C3b [30]. The mechanism of this enhanced regulatory function is not clear, but we speculate that simultaneous binding to target via domain 20 and to C3b via domain 19 facilitates enhanced activity. Indeed, we previously solved the structure of a tripartite complex between microbial protein OspE, C3d and FH19-20 [56]. Our model of FhbA on the surface of Borrelia is based on the hypothesis that BhFhbA acts similarly to OspE and binds simultaneously to FH20 and C3d (Fig 6).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. A working model for how BhFhbA recruits FH to mediate immune evasion. Membrane-bound BhFhbA (bright green) recruits FH (cartoon representation) of the host through binding site on domain 20 of FH (in red). When bound to BhFhbA, FH domains 1–4 are free to bind to C3b (yellow) and inhibit complement. When bound to microbial protein via domain 20, FH can also bind C3d fragment (in pink) via domain 19. Full-length FH model was manually constructed in Pymol by combining existing structures of different subcomplexes. Model for FH1-4 bound to C3b was from PDB2wii [15]. Structures of FH19-20 (in red) bound to C3d were obtained from PDB entries 5nbq [56] and 2xqw [66]. https://doi.org/10.1371/journal.ppat.1010338.g006

Using BhFhbA as a search model in structure-guided sequence database analysis enabled us to identify 10 homologous DNA loci from the RF and three from the LD clades (Fig 3). All ten identified proteins in the RF borreliae group are very similar within the FH20 binding region, as both the key phenylalanine residue and the surrounding hinge region are highly conserved. Moreover, for BhFhbA and BdFhbA, mutations of the key phenylalanine dramatically affected binding to both FH19-20 and full-length FH (Figs 2, 4, 5, S4, S6 and S10). The only exception is B. turicatae, which has an asparagine instead of aspartic acid in the +2 position after Phe154. Consistent with our predicted binding mechanism, B. turicatae BtcA is the only FhbA-related protein of RF borreliaea that does not bind FH [32] (Table 4). Interestingly, there are three FhbA-related sequences in the borrelia from the LD clade, which cluster separately from the RF group (Fig 4B). Sequence data show that there are many deleterious alternations at the DNA level. It is not yet known, if these FhbA-related sequences are functional on a protein level.

Analysis of all available sequence data from borreliae demonstrates that the LD clade has evolved to have three to four different classes of FH-binding proteins, whereas the RF clade has just one, which we name the FhbA-related protein family (Table 4). FhbA-related proteins may be able to inhibit complement more efficiently than the other FH-binding proteins, thus compensating for the lack of other FH-binding proteins. The interactions of FhbA-related proteins with other complement proteins (Table 4) might also affect overall regulation of complement. It cannot be excluded that other, yet unknown, FH-binding proteins exist in relapsing fever spirochetes, or that some other complement evasion mechanisms, like binding of C4BP [57] or C1-inhibitor [58] provide enhanced protection.

Five FH-binding FhbA-related proteins have been shown to mediate serum resistance. B. hermsii strain YOR, which expresses BhFhbA, is more resistant to serum and causes more persistent infections in mice when compared with strain REN, which naturally lacks FhbA [52]. Expression of three FhbA-related proteins (BhCRASP-1[59], HcpA [33], BpcA [32]) in the serum-sensitive strain Bo. burgdorferi B313 led to increased serum survival of the mutant strain. In addition, B. miyamotoi CbiA established serum resistance when expressed in serum sensitive Bo. garinii strain G1[34]. Surprisingly, an fhbA knockout strain created from B. hermsii strain YOR retained resistance to complement in vitro and in mice, even though the strain did not express FhbA nor bound FH [55].

To analyze the effect of BhFhbA on serum mediated killing, we expressed BhFhbA and BhFhbA mutant Phe154Ala on the surface of a serum-sensitive laboratory strain of E. coli. In that environment wild-type BhFhbA, but not the mutant, protected bacteria from complement killing (Fig 3E). The binding mechanism suggested by our structure thus appears to be important also in a more physiological context. Similar results were obtained from the assay, where we incubated live B. hermsii borrelia in whole blood and measured complement activation (Fig 3E). Wild-type BhFhbA competed with B. hermsii for FH whereas the Phe154Ala mutant did not.

Our results thus demonstrate that BhFhbA is also functional on the surface of E. coli and can provide protection from complement in a natural environment. Previous results from the fhbA knockout strain suggest that other yet unidentified mechanism(s) to prevent formation of membrane-attack complexes may exist in B. hermsii strain YOR. This is not, perhaps, surprising because pathogens typically have several mechanisms that act alone or in tandem to help the bacteria evade innate immunity.

In summary, we present here a high-resolution structure of BhFhbA, an outer-surface complement evasion mediating protein from Borrelia hermsii, in complex with FH19-20. We found a dozen highly homologous proteins from Lyme disease and relapsing fever spirochetes, thus identifying a new family of immune evasion proteins, which we name the FhbA-related protein family. We propose that FhbA-related proteins are important complement evasion molecules in RF borreliae, and thus represent important targets to develop tools to prevent infections caused by borreliae.

[END]

[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1010338

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.