Characterization of two putative Dichelobacter nodosus footrot vaccine antigens identifies the first lysozyme inhibitor in the genus

The Gram-negative anaerobic bacterium Dichelobacter nodosus (Dn) causes footrot in ruminants, a debilitating and highly contagious disease that results in necrotic hooves and significant economic losses in agriculture. Vaccination with crude whole-cell vaccine mixed with multiple recombinant fimbrial proteins can provide protection during species-specific outbreaks, but subunit vaccines containing broadly cross-protective antigens are desirable. We have investigated two D. nodosus candidate vaccine antigens. Macrophage Infectivity Potentiator Dn-MIP (DNO_0012, DNO_RS00050) and Adhesin Complex Protein Dn-ACP (DNO_0725, DNO_RS06795) are highly conserved amongst ~170 D. nodosus isolates in the https://pubmlst.org/dnodosus/ database. We describe the presence of two homologous ACP domains in Dn-ACP with potent C-type lysozyme inhibitor function, and homology of Dn-MIP to other putative cell-surface and membrane-anchored MIP virulence factors. Immunization of mice with recombinant proteins with a variety of adjuvants induced antibodies that recognised both proteins in D. nodosus. Notably, immunization with fimbrial-whole-cell Footvax vaccine induced anti-Dn-ACP and anti-Dn-MIP antibodies. Although all adjuvants induced high titre antibody responses, only antisera to rDn-ACP-QuilA and rDn-ACP-Al(OH)3 significantly prevented rDn-ACP protein from inhibiting lysozyme activity in vitro. Therefore, a vaccine incorporating rDn-ACP in particular could contribute to protection by enabling normal innate immune lysozyme function to aid bacterial clearance.

structural analyses of Dichelobacter ACp. Dn-ACP contains two repetitive segments that both share homology to MliC/PliC and ACP class proteins (Fig. 1A, Table 1). The alignment was done with Clustal software 32 , using the following protein sequences: ACP proteins from Neisseria meningitidis (Nm), Neisseria gonorrhoeae (Ng), Neisseria animaloris (Na), Comamonadaceae bacterium (Cb) and Eikenella corrodens (Eik); and lysozyme inhibitors or MliC/PliC proteins from Brucella abortus (Ba), Pseudomonas aeruginosa (Pa), Yersinia enterocolitica (Ye), Salmonella typhimurium (St) and Escherichia coli (Ec). To find the closest relatives, we displayed the Clustal phylogenetic analysis using FigTree, which placed the N-terminal Dn-ACP domain in close relationship with members of the ACP group, while the C-terminal domain is placed between the MliC/PliC and ACP groups (Fig. 1B).
The ACP and MliC/PliC proteins are overall structurally similar ( Fig. 1C blue and red, respectively). However, ACP proteins are distinguished from MliC/PliC proteins by a sequence insertion, occurring after strand β4. Our earlier structure determinations of ACP proteins from the Neisseria species Nm-ACP (PDB accession code 5MY7, 29 ) and Ng-ACP (PDB accession code, 6GQ4 30 ) show this structural feature, annotated as Loop 4 in Fig. 1C. The loop insertion in ACP proteins occurs at a position where the binding interface with lysozyme has been identified for the MliC/PliC proteins 33 . The structural feature of the Loop 4 region allowed us to describe these ACP proteins as novel lysozyme inhibitors 29,30,34 , predicting a different lysozyme interaction mode.
The sequence alignment in Fig. 1A confirms presence of the ACP characteristic sequence insertion in both Dn-ACP domains. The homology model of Dn-ACP domains based on Ng-ACP consequently shows the same feature (green structures in Fig. 1C). The predicted structures for N-and C-terminal Dn-ACP are highly similar (box in Fig. 1C), and both show the Loop 4 insertion, predicting a lysozyme binding mode similar to ACP, but different from MliC/PliC proteins. This allows us to resolve the ambiguity of the phylogenetic prediction, assigning both Dn-ACP domains as bona fide ACPs. The predicted structural similarity between Dn-ACP and Neisseriae ACP proteins was also supported by the observation that murine antisera to rDn-ACP cross-reacted in Western blotting with rNm-ACP ( Supplementary Fig. 3A). Similarly, murine antisera to rNm-ACP 29 4). A crystal structure is available for the Legionella pneumophila (Lp)-MIP protein (PDB: 1FD9) 25,35 . Lp-MIP forms as a V-shaped homo-dimer in the bacterial outer membrane (OM), with the lipidated N-terminal dimerization α-helix anchoring the protein in the OM, and the globular C-terminal domain containing the FKBP-type PPIase function 25,35 . Using the same methods as for Dn-ACP, we have created a homology model for Dn-MIP to gain functional insights into dimerisation and activity (Fig. 2). To understand whether dimerisation would indeed occur, we investigated the dimerisation interface. Using the PISA software, we mapped a very similar interface of nearly 1800 Å 2 and entropy gain for dimer formation of nearly 23 kcal/mol for template and homology model. Details of the interface are seen in the zoom in Fig. 2. We further investigated conservation of the catalytic domain. Structures of inhibitor complexes to FKBP type proteins are known, e.g. the human FKBP protein in complex with the immunosuppressant FK506 (PDB:1BKF), shown in Fig. 2, or the Trypanosoma cruzi MIP protein in complex with the same inhibitor (PDB:1JVW). From comparison with these proteins it is clear that while there are changes in structure due to amino acid exchanges, the binding pocket creating the active site is conserved.

Antigenicity of rDn-MIP and rDn-ACP proteins in different vaccine formulations.
Recombinant rDn-ACP encoded by Allele 1 and rDn-MIP encoded by Allele 1 in serotype B strain VPI 2340 [11342] were used in mouse immunisation experiments, using a variety of adjuvant and delivery vehicles that have been shown to enhance the production of antibodies to recombinant proteins: (i) adsorption on Al(OH) 3 was used as standard, as it is a licensed adjuvant routinely used in veterinary vaccines; (ii) mixing with the saponin Quil A has been used successfully for veterinary applications, including in sheep 36 ; (iii) incorporation into liposomes, which provide an intrinsic adjuvant effect and can permit folding of membrane proteins, and are extensively used as a non-toxic delivery vehicle for veterinary vaccine antigens and therapeutic molecules [37][38][39] ; (iv) incorporation into liposomes with additional immunomodulator monophosphoryl lipid A (MPLA) 39 to increase antigenicity; (v) solubilisation with the zwitterionic detergent Zwittergent (Zw) 3-14, which provides a micellar structure 40 ; (vi) solubilisation with Zw 3-14 containing MPLA; (vii) emulsifying in Freund's water-in oil emulsion and in the (viii) oil emulsion MONTANIDE ISA 71 VG, as used in cattle, poultry and sheep 41,42 , which enable slow antigen release, thus stimulating high and long-lasting antibody responses. Controls included immunisation with Footvax ® , adjuvant mixtures without antigen and normal mouse serum (NMS).
Individual antisera were also tested in a D. nodosus whole cell ELISA. Serum reactivity was expressed as mean OD λ 450 nm value and standard deviation for each immunization group. All rDn-ACP-adjuvant formulations, except QuilA, elicited similar weak but specific reactivities against D. nodosus whole cells, which was statistically significant when compared to the corresponding sham-immunized groups (OD λ 450 nm ~0.2, P < 0.05) (Fig. 3C). All rDn-MIP formulations, except MONTANIDE, elicited weak but specific reactivity (OD λ450 nm ~0.2) against D. nodosus whole cells (Fig. 3D), which were also statistically significant, compared with the corresponding sham-immunized groups (P < 0.05). The mean reactivity of Footvax antisera against whole cells was significantly higher (P < 0.05) compared with any of the recombinant protein formulations, with values of OD λ450 nm > 0.8. Reactivity with NMS or any of the sham-immunized sera was not significant (P > 0.05) (Fig. 3C,D).
Testing antisera against whole cell lysates in Western blots showed weak but specific responses against rDn-ACP vaccine formulations (Fig. 4A), and strong and specific immune responses against rDn-MIP vaccine formulations (Fig. 4B). No significant reactivity was observed with any of the respective sham-immunized sera or NMS (Fig. 4A,B). Murine antisera to Footvax reacted strongly against multiple targets in the D. nodosus whole cell lysate (Fig. 4A,B) and also recognised both purified recombinant Dn-ACP and Dn-MIP proteins (Fig. 4C).
Dn-ACp is a lysozyme inhibitor. The predicted structural similarity of Dn-ACP with Neisseria ACP prompted us to test for in vitro lysozyme inhibition. Dose-dependent kinetics of purified rDn-ACP inhibition of Hen egg-white lysozyme (Hewl) was determined by measuring lytic absorbance over time of a suspension of Micrococcus lysodiekticus, a Gram-positive bacterium that is intrinsically sensitive to lysozyme. Hewl (10 U) induced ~55% lysis of M. lysodiekticus by 2 h, which increased to ~85% by 24 h (Fig. 5, Supplementary Fig. 5A). Addition of doses ≤ 3 pmol rDn-ACP was not protective against Hewl lysis (P > 0.05). However, addition of 15 pmol rDn-ACP reduced Hewl-mediated lysis of M. lysodeikticus cells by ~45% and ~ 20% after 2 h and 24 h (P < 0.05), respectively (Fig. 5, Supplementary Fig. 5A). Addition of ≥ 30 pmol rDn-ACP reduced Hewl-mediated lysis by ~100% after 2 h (P < 0.05) (Fig. 5, Supplementary Fig. 5A). In comparison, higher amounts of C-terminal domains of Dn-ACP were predicted using MODELLER and are shown in dark and light green respectively. The boxed overlay highlights subtle structural differences as predicted. For comparison, known structures of Nm-ACP (PDB:5MY7, light blue, with labelled beta strands), Ng-ACP (PDB:6GQ4, blue), St-PliC (PDB:3OE3, pink) and Ba-PliC (PDB:4ML7, dark red) are shown in identical orientation. In all structures, the functional Loop 4 region has been annotated. Additionally, the site of the lysozyme binding interface determined for Ba-PliC has been indicated.
www.nature.com/scientificreports www.nature.com/scientificreports/ meningococcal rNm-ACP protein were required to produce similar protection, with the addition of 180 pmol rNm-ACP reducing M. lysodiekticus lysis by ~80% after 2 h (Supplementary Fig. 5B) and ~59% after 24 h (Fig. 5), compared with the control Hewl. Addition of rDn-MIP did not inhibit Hewl-induced lysis (Fig. 5, Supplementary  Fig. 5C), and none of the tested proteins (rDn-ACP, rNm-ACP or rDn-MIP) on their own showed any lytic effect (Fig. 5, Supplementary Fig. 5). The possibility that the trace presence of SDS used to solubilize rDn-ACP and rNm-ACP proteins affected the enzymatic activity of Hewl was also tested. Addition of SDS at the same final concentration in the highest dose of Nm-ACP tested (2.25 × 10 −3 % w/v SDS) to a suspension of the bacteria in the presence or absence of Hewl had no significant effect (P > 0.05) (Fig. 5, Supplementary Fig. 5), confirming that lysis of M. lysodeikticus was due to Hewl lytic activity and that inhibition of Hewl in the presence of either rDn-ACP or rNm-ACP was due to the inhibitory activity of these proteins.
Lysozyme can kill bacteria through its hydrolase activity and through an enzyme-independent mechanism that relies on its cation state 43,44 . To exclude the possibility that lysozyme antimicrobial activity was responsible for the observed killing of M. lysodiekticus, bacterial suspensions were treated with Hewl that had been boiled for 1 h, a treatment that has been reported to destroy peptidoglycan hydrolase activity, whilst maintaining antimicrobial activity 45,46 . Addition of boiled lysozyme did not induce lysis of M. lysodiekticus cells compared to native Hewl ( Supplementary Fig. 6), demonstrating that inherent Hewl antimicrobial activity did not induce M. lysodiekticus cell death.

Discussion
The data presented here support proposing rDn-ACP and rDn-MIP antigens as novel antigens in combined vaccine strategies. Based on homology models and structural similarity, we derived structural and functional properties for both proteins. We presented data also on the function of rDn-ACP as a lysozyme inhibitor. Both Dn-ACP and Dn-MIP proteins are highly conserved in the http://pubmlst.org/dnodosus collection of isolates. In our study, we immunized with rDn-ACP encoded by Allele 1 (serotype B strain VPI2340), expressed by 86% of the isolates (146 isolates/170 total), and with rDn-MIP from Allele 1 of the same strain, expressed in 50% of the isolates (83 isolates/166 total). Assuming no cross-reactivity, a monovalent rDn-ACP Allele 1 protein vaccine would cover the majority of isolates, whereas a monovalent rDn-MIP Allele 1 protein vaccine would protect against 50% of disease-causing strains. Given the high sequence identity between non-redundant Dn-ACP and Dn-MIP proteins, levels of coverage might be even higher, assuming protein cross-reactivity.
Modelling of Dn-MIP structure demonstrated similarities with the Legionella-MIP, which has been extensively characterised as a FKBP-type PPIase 25 . PPIases are a superfamily of proteins ubiquitously distributed among living organisms, which function primarily to assist the folding and structuring of unfolded and partially folded polypeptide chains and proteins 25 . Lp-MIP and other MIP proteins exist as homo-dimers in the bacterial OM. Each monomer can be structurally divided into three regions: 1) a N-terminal α-helix that participates in a dimerization domain and is probably lipidated to allow anchoring into the OM; 2) a long connecting α-helix in the centre that ends with 3) a globular C-terminal domain containing the FKBP-type PPIase function 25 . MIP-like PPIases have been reported as essential cell-surface and membrane-anchored virulence factors. Lp-MIP was the first genetically identified FKBP-type PPIase that was shown to be required for virulence and intracellular infection 24,[47][48][49][50] . A Neisseria gonorrhoeae MIP homologue was required for intramacrophage survival 51,52 and a Neisseria meningitidis MIP homologue was important for meningococcal survival in the blood 53 . Furthermore, inhibition of Nm-MIP prevented meningococci from adhering, invading and/or surviving in epithelial cells 54 . Several bacteria that cause animal infections, including Chlamydia, Francisella, Aeromonas and Pseudoalteromonas, amongst many others, also produce MIP-like PPIases essential for virulence 25 . Furthermore, recombinant MIP proteins from Chlamydia spp. and Coxiella burnetti have been examined as potential vaccine antigens 25 . Based on sequence and structural resemblance and on our previous studies demonstrating that rNm-MIP protein was immunogenic and capable of inducing antibodies bactericidal for meningococci and cross-protective against gonococci 26-28 , we hypothesize that Dn-MIP protein is probably a virulence-associated PPIase and therefore merits investigating as a putative antigen for subunit footrot vaccines.
Sequence analysis revealed duplication of ACP domains in Dn-ACP as a striking observation. We showed that both domains fall structurally into the novel class of lysozyme inhibitors that we identified earlier in the Neisseriae 29,30,34 , and these domains had a characteristic loop insertion that alters the lysozyme binding interface. While lysozyme inhibition can be shown for these proteins 29,30,34 , the lysozyme binding mode would be predicted to differ from classic MliC/PliC proteins. Furthermore, antibodies to rDn-ACP cross-reacted with rNm-ACP and vice versa, which demonstrated the conservation and close relationship with Neisseria ACP proteins.
To our knowledge, there are no reports on lysozyme inhibition activity for Dichelobacter and our study is the first to show that Dn-ACP is a C-type lysozyme inhibitor. We demonstrated that the rDn-ACP protein prevented www.nature.com/scientificreports www.nature.com/scientificreports/ lysis of M. lysodiekticus cells in the presence of Hewl in vitro by inhibiting the enzymatic activity of lysozyme. Comparing the effectiveness of lysis with the rNm-ACP protein, six-fold less rDn-ACP was required and it is interesting to speculate that the strong Hewl inhibition by Dn-ACP may be rooted in the duplication of the ACP domains. Expression of Dn-ACP in vivo may enable Dichelobacter to avoid a key, ubiquitous vertebrate innate immune defence mechanism, and possibly contribute to establishing bacterial colonization and infection.
Inclusion of rDn-ACP within a defined antigen footrot vaccine or perhaps to existing vaccines including Footvax, might led to the potential induction of ovine antibody responses that inhibit the ability of Dichelobacter to avoid lysozyme activity and thus contribute to preventing infection by allowing host innate immune clearance by lysozyme. However, choice of adjuvant is critical for new vaccine(s) that might include rDn-ACP. A variety of different adjuvants were used with both rDn-ACP and rDn-MIP antigens. In general, all of the adjuvants helped stimulate similar high levels of antibodies to both proteins. However, we did observe differences in the ability of rDn-ACP antibodies to inhibit rDn-ACP in vitro, which was influenced by the nature of the adjuvant used. Only antisera generated by rDn-ACP-QuilA and rDn-ACP-Al(OH) 3 , and to a lesser extent to rDn-ACP-MONTANIDE and rDn-ACP-Freunds' mixtures, significantly prevented rDn-ACP protein from inhibiting Hewl activity in vitro. It is possible that these adjuvants induce antibodies with higher avidity that bind to the, as yet, uncharacterized rDn-ACP epitopes that are involved in binding to and inhibiting lysozyme. Moreover, despite the fact that immunization with Footvax induced murine antibodies that reacted with rDn-ACP, these antibodies were unable to prevent rDn-ACP inhibition of Hewl activity. Several possible reasons may account for this unexpected lack of antibody function, e.g. the oil adjuvant in Footvax is inappropriate; relatively lower levels of antibody were induced, perhaps as a consequence of different molar ratios of recombinant protein and Dn-ACP present in the  Fig. 1. www.nature.com/scientificreports www.nature.com/scientificreports/ Footvax dose used for immunization; and/or antibodies induced to Footvax Dn-ACP have lower avidity to functional and/or other non-functional epitopes.
A defined measure of protection would allow comparisons between studies and reduce the need for ovine protection trials, in which sheep would be vaccinated with these antigens in an appropriate adjuvant and the feet then challenged with D. nodosus. However, the mechanism of protection against footrot is not clear. Parenteral vaccination with Footvax confers protection, and generating specific interdigital mucosal immune responses may be important. There are in vitro assays that provide correlates for protection with other pathogens that express homologue ACP and MIP proteins, e.g. the complement-mediated serum bactericidal assays (SBA) described previously for Nm-ACP, Ng-ACP and Nm-MIP 26-28,30,55 and opsonophagocytosis assays using immortalized murine macrophage cell lines e.g. J774A.1 56 . However, no correlates are available for ovine protection. Nonetheless, we attempted to develop an SBA assay, but a functional bactericidal antibody response induced by antigen compared to sham-immunization could not be quantified, due principally to the sensitivity of Dichelobacter to animal sera used as exogenous complement sources. This same issue hindered our attempts to develop an opsonophagocytosis assay using J774A.1 murine macrophages, based on a method described by Romero-Steiner et al. 57 . We attempted also a slide agglutination test as originally described by Claxton et al. 14 , since previous studies have shown agglutination could be a potential biological marker, e.g. the level of serum K-agglutinating antibody stimulated by whole-cell vaccination was associated with protection against footrot 10 . There was visual evidence of partial agglutination of D. nodosus bacteria mixed with antisera to Dn-ACP-MONTANIDE, Dn-MIP-(Lip + M) and Dn-MIP-Freund's in comparison to treatment with NMS, suggesting possible interaction of antibodies to the surface antigen (data not shown). However, this agglutination assay was generally non-discriminatory and relatively insensitive. Recently, a three-dimensional skin explant model was described for studying anaerobic bacterial infections such as footrot 58 , but this has not been developed, and might not be appropriate for, vaccine efficacy testing.  A,B). The columns represent the geometric mean reciprocal ELISA titres (n = 5 animals per group) and the error bars represent the 95% confidence limits. No significant reactivity was observed with sera from sham-immunised animals or with NMS (absorbance values OD λ450nm < 0.1 for serum dilutions of 1/10; data not shown). All same antisera to (C) rDn-ACP or (D) rDn-MIP protein were reacted in whole-cell ELISA against D. nodosus whole cells (1 µg cells/well). The columns represent the average absorbance value (OD λ450nm ) (n = 5 animals per group) and the error bars represent the standard deviation (SD). Significant reactivity in comparison to the corresponding sham sera or NMS is marked with an asterisk (*) (P < 0.05).
www.nature.com/scientificreports www.nature.com/scientificreports/ In this study, our rationale for using an inbred mouse species was that it allowed us to pre-screen the antigenicity of both proteins, identify a suitable range of adjuvants and examine the mechanisms of lysozyme inhibition, in a more affordable way. However, the limitation of our study is that mice are maintained in a specific-pathogen free environment and their immune responses may not be similar to those of outbred sheep with immune systems that are exposed to a range of different microbes in their environment. Thus, despite the promising structural/ functional features of rDn-ACP and rDn-MIP as putative vaccine antigens, and in the absence of reliable in vitro as well as antisera to Footvax and normal mouse serum (NMS) as a control, were reacted against wild type D. nodosus whole cell lysate preparation (50 μg/well or 15 μg/well, respectively) in strip Western blot. Dn-ACP and Dn-MIP proteins were specifically recognised as a single band of Mr ~25.7 kDa or ~35.6 kDa, respectively, with all murine antisera to the corresponding single antigen (identified by the arrow). Antisera to Footvax, instead, reacted against multiple proteins in the lysate. All sham immunisation sera and NMS were non-reactive. Samples were tested on strips prepared from the same western blots and processed in parallel. (C) Pooled mice antisera (1/100 dilution; n = 5 animals) raised against footrot vaccine Footvax was reacted in Western blot against purified rDn-ACP and rDn-MIP proteins (1 μg each), which were both recognized as a single band of Mr ~30.2 kDa or ~40.1 kDa, respectively (identified by the arrow). Samples were tested and processed in parallel. Images were captured on a HP ScanJet flatbed scanner. Representative full-length blots are presented (n ≥ 3 experiments for each western blot). (2019) 9:10055 | https://doi.org/10.1038/s41598-019-46506-z www.nature.com/scientificreports www.nature.com/scientificreports/ assays for correlates of protection, the vaccine potential of both proteins ultimately requires expensive ovine protection trials.
In summary and to our knowledge, this is the first report of the function of Dn-ACP as a novel lysozyme inhibitor produced by Dichelobacter, and increases our knowledge of the distribution of this class of inhibitors amongst human and animal pathogens. A vaccine containing rDn-ACP could contribute to protection by enabling normal lysozyme function and further studies are needed to examine the role of Dn-ACP in particular during infection in vivo. Furthermore, MIP proteins have been acknowledged as potential targets for drug therapies during infection 54 and for vaccine development 25 . Thus, both antigens deserve consideration for next-generation footrot vaccines.
Bioinformatics. D. nodosus DNO_0012 peptidyl prolyl-cis-trans-isomerase FKBP-type, now DNO_ RS00050, and DNO_0725 potential Adhesin Complex Protein, now DNO_RS06795, were used as allele search terms in the https://pubmlst.org/bigsdb?db=pubmlst_dnodosus_isolates database 31 . Isolates number 172 (accessed April, 2019) and amino acid sequences for alleles were produced with Clustal (https://www.ebi.ac.uk/ Tools/msa/clustalo/) and dendrograms showing relationships between non-redundant alleles were generated with Jalview 2.8 (http://www.jalview.org/development/release-history/Version-282). sequence alignment and phylogenetic analysis for ACp proteins. Sequences alignment was performed using ClustalX 2.1 32 . Initially, sequences of all ACP were aligned using multiple sequence alignment with maximum iterations and all MliC/PliC were aligned using the same parameters separately. These were each saved as separate alignment profiles. The same method was used to align the N-and C-terminal sequences of Dn-ACP and this was also saved as a profile. Profile alignment was first performed using MliC/PliC and ACP, this alignment was then saved as a separate profile. The Dn-ACP profile was then aligned to the MliC/PliC/ACP profile to produce the final alignment. The aligned sequences were output in a phylogenetic tree format (.ph), which was displayed by FigTree v1.4.3 (http://tree.bio.ed.ac.uk/software/figtree) 60 . Sequence alignments presented were coloured using ESPript (ESPript -http://espript.ibcp.fr) 61,62 . The secondary structure predictions for Dn-ACP sequences were made using Jpred4 63 and the secondary structure for St-PliC and Ng-ACP were assigned by DSSP (http://swift.cmbi.ru.nl/gv/dssp/) 64 and based on the crystal structures 3OE3 and 6GQ4 respectively.

Modelling of Dn-ACp domains.
After splitting the sequence of Dn-ACP into two segments based on their similarity, each sequence was loaded into Chimera (http://www.rbvi.ucsf.edu/chimera/) 65 and a BLAST (https:// www.ncbi.nlm.nih.gov/BLAST/) search performed on each sequence. The BLAST search indicated PDB codes 6GQ4 and 5MY7 had sequence similarity. These PDBs were loaded into Chimera and the sequences aligned to Dn-ACP sequences. MODELLER 9.21 (https://salilab.org/modeller/) 66 was then run locally to generate 250 models of each Dn-ACP sequence (N-and C-term). These were then scored using the web server and the best model

Cloning, expression, and purification of recombinant Dn-ACP and Dn-MIP proteins. DNO_0725
and DNO_0012 were cloned, expressed and purified using methods previously described for rNm-ACP and rNm-MIP proteins, respectively 28,55 . Briefly, the DNO_0725 (dn-acp) gene sequence, encoding the amino acid sequence (AA 1-230) for Dichelobacter nodosus (Dn)-ACP protein, was amplified by PCR using primers DNO0725 Fw 5′-TTACTCGAGATGAAAAAAGAACAATTTTTAGCGGC-3′ and DNO0725 Rv 5′-TAT AAGCTTTTAACGTGCTTTACACATTTTATAAAGG-3′. The DNO_0012 (dn-mip) gene sequence, encoding the amino acid sequence (AA 1-329) for Dn-MIP protein, was amplified by PCR using primers DNO0012 Fw 5′-TAACTCGAGATGATGAAAAAAACTTCTTTAC-3′ and DNO0012 Rv 5′-TATAAGCTTCTATTTTTTAGCCGCTTCATCAGG-3′. The restriction sites for XhoI and HindIII are underlined. Both the 711 bp (dn-acp) and the 1008 bp (dn-mip) PCR products were purified using a Wizard ® SV Gel and PCR Clean-up System (Promega), double digested with XhoI and HindIII restriction enzymes (Promega) at 37 °C for 3 h, and purified from an agarose gel band. The purified digested PCR fragments were then ligated overnight at 4 °C to pRSETA vector, which was previously linearized with both same restriction enzymes. The recombinant vectors (pRSETA-dn-acp and pRSETA-dn-mip), encoding for Dn-ACP and Dn-MIP proteins fused to a N-terminal hexa-histidine tag, were transformed into E.coli DH5α competent cells for plasmid amplification and subsequently into competent E. coli BL21 (DE3) pLysS cells for protein expression. Recombinant (r)Dn-ACP protein was purified by Ni-NTA affinity chromatography under denaturing conditions. Bound protein was eluted using 100 mM NaH 2 PO 4 , 10 mM Tris-HCl, 6 M GuHCl and 250 mM imidazole buffer, pH 8.0, precipitated with 5% (v/v) Trichloroacetic acid (TCA), and subsequently resuspended in phosphate buffered saline (PBS), pH 7.4 with 0.5% (w/v) sodium dodecyl sulphate (SDS) for solubilization. rDn-MIP protein was purified by Ni-NTA affinity chromatography under non-denaturing conditions. Bound protein was eluted using 50 mM NaH 2 PO 4 , 300 mM NaCl and 250 mM imidazole buffer, pH 8.0, and subsequently dialyzed against PBS, pH 7.4 for 48 h. Protein concentration was determined using the BCA TM Protein Assay Kit (Pierce). The molecular mass (Mr) of rDn-ACP and rDn-MIP proteins, with a 39 AA-long N-terminal hexa-histidine tag (MRGSHHHHHHGMASMTGGQQMGRDLYDDDDKDRWGSELE), was 30.2 kDa and 40.1 kDa, respectively, as shown by SDS-PAGE ( Supplementary Fig. 8). The purified recombinant proteins were stored at −20 °C until needed.
Immunization of mice with rDn-ACp and rDn-MIp proteins. BALB/c mice (H-2 d haplotype) were bred within the animal facilities of the university under standard conditions of temperature and humidity with a 12 h lighting cycle and with food and water available ad libitum. Groups of five BALB/c mice of approximate equal sizes and weights (6-7 weeks of age) were immunized with purified rDn-ACP or rDn-MIP protein prepared separately with the following adjuvants.
rDn-ACP and rDn-MIP stock solutions were diluted in sterile saline and were adsorbed to (i) AL(OH) 3 , (ii) incorporated into Liposomes, (iii) Liposomes plus monophosphoryl lipid A (MPLA), (iv) Zwitterion (Zw) detergent 3-14 and (v) Zw 3-14 plus MPLA, as described previously 30 . Control preparations without protein were prepared similarly. The columns represent the mean % lysis (from n = 3 independent experiments) and the error bars represent the corresponding SEM. Data were compared with a paired t-Test and the asterisks denote significant inhibition (P < 0.05) of rDn-ACP function by anti-rDn-ACP sera, compared to treatment without antisera, after 2 h (and continuing after 24 h incubation) (*) or after 24 h only (**).
www.nature.com/scientificreports www.nature.com/scientificreports/ (vi) Freund's adjuvant. For the primary immunization, a volume of 0.35 mL of a solution of 400 µg/mL of rDn-ACP or rDn-MIP in sterile saline was mixed with 0.35 mL of Freund's Complete Adjuvant (FCA) and an emulsion prepared by repeated passage through a 21G syringe needle. The second and third doses were prepared similarly in Freund's Incomplete Adjuvant (FIA). Control FCA and FIA preparations were prepared similarly without protein.
(vii) Quil A (Sigma). A solution of 1 mg/mL Quil A was prepared in saline and sterilized by filtration through a 0.22 µm filter. A volume of 0.105 mL of sterile Quil A (1 mg/mL) was added to 0.35 mL of a solution of 400 µg/ mL of rDn-ACP or rDn-MIP in sterile saline and the volume adjusted to 0.7 mL with sterile saline. Control Quil A preparation was prepared similarly without protein.
(viii) MONTANIDE ISA 71 VG (Batch T35031). This oil adjuvant was kindly provided by Seppic (France) for the production of water-in-oil emulsions. MONTANIDE was prepared in a 70:30 ratio with protein in saline, following the manufacturer's recommendations. A volume of 0.49 mL of MONTANIDE was mixed with 0.21 mL of saline containing 140 µg of rDn-ACP or rDn-MIP in sterile saline and an emulsion prepared as described for Freund's adjuvants. Control MONTANIDE emulsion was prepared with 0.49 mL of adjuvant with 0.21 mL of saline alone.
Each mouse was immunized with a volume of 100 µL containing 20 µg of protein and adjuvant per dose on days 0, 14 and 28. Control mice received 100 µL of adjuvant alone. One group (n = 5) was kept as a source of normal mouse serum (NMS).
All mice were terminally bled by cardiac puncture under anaesthesia on day 42. All sera were stored at −20 °C until required and were decomplemented by heating at 56 °C for 30 min where stated. ethics statement. This study complied with the animal experimentation guidelines of the Home Office (HO), with approval granted under the Animals (Scientific Procedures Act, 1986) with a HO project licence number PPL 30/3126. The study was approved by the Animal Welfare and Ethics Review Board (AWERB) at the authors' institution (University of Southampton, no number assigned). Animal health and welfare was assessed daily by qualified animal technicians and no animals suffered significant adverse effects.

Characterization of biological and functional properties of antibodies to rDn-ACp and rDn-MIp. (i) Recombinant protein eLIsA.
Individual murine antisera were reacted in ELISA against rDn-ACP and rDn-MIP proteins (0.1 µg protein/well), as described previously 69 . Briefly, flat-bottomed polystyrene microtitre plates were coated overnight at 37 °C with recombinant protein in 0.05 M sodium carbonate buffer, pH 9.6. Wells were washed with PBS containing Tween-20 (0.05% v/v, PBST) and antisera were diluted serially in PBST containing 1% (w/v) BSA and the plates incubated for 1 h at 37 °C. After washing, antibody binding was detected using the appropriate HRP conjugate with 3,3′,5,5′-tetramethylbenzidine and H 2 0 2 as enzyme substrate. Absorbance was measured at λ450 nm after 10 min of incubation with enzyme substrate, and the ELISA titre, extrapolated from the linear portion of the serum titration curve, was taken as the reciprocal dilution which gave an increase in absorbance of 0.1 U after 10 min.
(ii) Whole cell ELISA. The wells of flat-bottomed polystyrene microtitre plates were coated with 1 µg of live D. nodosus cells/well in buffer carbonate buffer, pH 9.6, overnight at 4 °C. For this assay, individual antisera were tested at a single dilution of 1/100 on whole cells, which was the lowest dilution determined as optimal for avoiding non-specific reactivity of control, sham sera or normal mouse serum. Washing steps and antibody binding detection were done as described above. Absorbance measurements at λ450 nm were made after 10 min of incubation with enzyme substrate, as described above. For the recombinant protein and whole cell ELISA, data were compared using a two-sample t-Test, with P values < 0.05 considered significant.
(iii) Western immunoblotting. A D. nodosus whole cell lysate preparation (15 µg/well and 50 µg/well for testing antisera to rDn-MIP and antisera rDn-ACP antisera, respectively) and purified recombinant proteins (1 μg each), were separated on SDS-PAGE and then transferred to nitrocellulose by semi-dry blotting. After incubation with pooled murine sera (1/100 dilution; n = 5 animals), immunological reactivity was detected by using anti-mouse immunoglobulin-alkaline phosphatase conjugate (Bio-Rad, UK) as described previously 69 .
(iv) Hen egg-white lysozyme (Hewl) inhibition assay. Lysis kinetics of freeze-dried M. lysodeikticus cells (ATCC4698 -Sigma Aldrich) suspended at 1 mg/mL in 10 mM potassium phosphate buffer (PPB) pH 7.0 supplemented with a protease inhibition cocktail (Roche) were monitored over time (every 5 min for 2 h and a final end point of 24 h incubation) by measuring the change in optical density (OD) at λ 595 nm in the presence or absence of 10 U/well of Hen egg-white lysozyme (Hewl) at 25 °C. rDn-ACP protein inhibition activity of Hewl was determined by dose response (3-90 pmol/well) experiments, for which rNeisseria meningitidis (Nm)-ACP (6-180 pmol/well) and rDn-MIP (60-180 pmol/well) proteins were tested as positive and negative controls for Hewl inhibition, respectively. Possible detrimental effect of SDS on the integrity of M. lysodeikticus cells and/or on the enzymatic activity of Hewl was tested by adding the same final concentration of SDS (2.25 × 10 -3 % (w/v)) to a suspension of the bacteria with or without Hewl and compared their kinetic curves with the same samples without SDS. Adjuvant effect of rDn-ACP antisera on preventing rDn-ACP protein from inhibiting Hewl was demonstrated by adding to the reaction (for the condition of 30 pmol/well of rDn-ACP only) a 1/16 dilution of pooled (n = 5 mice per group), decomplemented murine antisera raised against the recombinant protein delivered in the different formulations, as well as the corresponding sham sera, Footvax antisera and NMS. (2019) 9:10055 | https://doi.org/10.1038/s41598-019-46506-z www.nature.com/scientificreports www.nature.com/scientificreports/ For testing if Hewl antimicrobial activity was responsible for M.lysodiekticus cell death, Hewl was boiled for 1 h and lysis kinetic experiments were done using the same conditions described above, comparing boiled Hewl with untreated Hewl and control (no Hewl).

Data Availability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files).