The bacteria strains, plasmids, and primers used in this study are described in Table 1. The Aa D7S-1 wildtype strain [21, 22], designated AaD7S, was used to construct the mutants in omp29 and omp29 paralogue (omp29^par) genes. The Aa HK1651 wildtype strain (Genbank Access AY 262734) was used to obtain the OMP29 recombinant protein. Aa strains were grown in a humidified 5% CO₂ atmosphere at 37°C (Shel Lab, OR, United States), in modified Trypticase Soy Broth (mTSB) containing 3% trypticase soy broth and 0.6% yeast extract, or on mTSB agar [mTSB with 1.5% agar (Oxoid Ltd, Basingstoke, Hampshire, England)].

The plasmids used in this study were replicated in Escherichia coli strains DH5-α and E. coli BL21(DE3) (Invitrogen, Carlsbad, CA, United States) were cultured by standard methods [26].

Spectinomycin (50 μg/ml, Sigma–Aldrich, St. Louis, MO, United States), tetracycline (4 μg/ml, Sigma–Aldrich), ampicillin (100 μg/ml, Sigma-Aldrich), kanamycin, (50/100 μg/ml, Sigma–Aldrich), and chloramphenicol (50 μg/ml, Sigma–Aldrich) were added to the media when selection of mutants or transformants were needed.

The growth curve was determined in mTSB with the addition of mineral oil to create a microaerophilic condition in 100-well HoneyComb plates (Bioscreen Automation for Microbiology, NJ, United States). The optical density (OD) at 420-580 nm was monitored in a spectrophotometer (Growth Reader Bioscreen C, Oy Growth CurvesAb Ltd., Bioscreen C Type FP-1100-C, FIN-21280 Raisio, Finland) in 1 h intervals for 16 h at 37°C, under agitation. The growth rate was calculated according to the formula: μ = LnNN0t, where N and N₀, respectively, correspond to the initial and final OD_420−580 at the exponential growth phase and t is the time-course of the growth curve.

Genomic DNA of Aa D7S-1 and Aa HK1651 was obtained by using QIAamp DNA mini kit (Qiagen, Valencia, CA, United States). Plasmids DNA was obtained by using QIAprep Spin Miniprep kit (Qiagen). E. coli DH5-α and BL21(DE3) strains (Invitrogen, Carlsbad, CA, United States) were transformed with pCR 4-TOPO (Invitrogen) and vector pET28B vectors (pET Expression System 28, Novagen), respectively, by electroporation (MicroPulser, BioRad, Hercules, CA, United States). Restriction enzymes, T4 ligase, and Taq DNA polymerase were purchased from New England BioLabs Inc. (Ipswich, MA, United States). PCR was performed with Taq high-fidelity polymerase (Invitrogen), and its products were purified using QIAquick PCR purification and QIAquick gel extraction kits (Qiagen).

Deletion mutants in omp29, omp29^par, and both genes were obtained by using the loxP/Cre system [21, 26]. Briefly, the spectinomycin cassette loxP-Specr-loxP was obtained by restriction digestion of plox2/Specr vector [24] with DraIII. The omp29 or omp29^par flanking regions were obtained by amplification with primer Omp29Up-F, Omp29Up-R, Omp29Dw-F, Omp29Dw-R, and Omp29ParalUp-F, Omp29ParalUp-R, Omp29ParalDw-F, and Omp29ParalDw-R, respectively, and digested with DraIII. This product was then ligated to the loxP-Specr-loxP fragment and transformed into the competent AaD7S strain.

Transformants were selected in TSYE agar with spectinomycin. The allelic replacement of the target gene was confirmed by PCR and sequencing. The removal of loxP-Specr-loxP of strains was done by transformation with pAT/Cre vector [25], which encodes Cre recombinase.

The omp29 and omp29^par deficient mutants of AaD7S were named AaΔ29 and AaΔ29P. The double deletion mutant (referred to as AaΔ29Δ29P) was constructed by the deletion of omp29^par from the AaΔ29 strain following the procedure described above, with the exception that the spectinomycin resistance cassette in omp29^par was not removed.

The success of the mutation strategy was confirmed by PCR and sequencing (data not shown) using the following primers: Omp29Up-R, Omp29Dw-F, Omp29ParalUp-R, and Omp29ParalDw-F, which annealed downstream and upstream of omp29 and omp29^par, respectively.

The Signal 3.0 Server program was used to predict the sequence that encodes the signal peptide of omp29. Based on this prediction, the pET OMP29 F and R primers with BamHI and XhoI restriction sites were used to amplify omp29 from Aa HK1651 (Genbank accession number AY 262734), constructed previously in pCR4-TOPO vector (designated as pTOPOomp29SPS). After digestion, the omp29 amplicon (without the signal peptide) was ligated to pET28B vector (pET Expression System 28, Novagen) and the resulting pET28B/omp29 vector was transformed into E. coli BL21 (DE3). The recombinant plasmids were confirmed by T7 Universal and Terminator primer sets for sequencing.

The OMP29 was expressed by pET28B/omp29 E. coli BL21 (DE3) following the manufacturer's instructions (New England Biolabs, Ipswich, MA, United States). A single colony of pET28B/omp29 E. coli BL21 (DE3) was inoculated into LB broth with kanamycin (50 μg/ml) (Sigma–Aldrich, St. Louis, MO, United States) and incubated under agitation at 120 rpm for 16 h at 37°C. Cells were transferred to fresh LB broths, and incubated under agitation at 250 rpm at 37°C until O.D. 600 nm ≈ 1 was reached. Expression of Aa OMP29^His was induced by adding 1 mM of isopropyl β-D-thiogalactoside (IPTG, Sigma–Aldrich, St. Louis, MO, United States) at 37°C and the recombinant protein was observed after 2 h in culture precipitate. Cells were then lysed [27] and Aa OMP29^His in cells lysate supernatant was purified by affinity chromatography using nickel chelating resin (Ni-NDA, Invitrogen).

Balb/c mice were immunized subcutaneously with 50 μg of purified Aa OMP29^His with adjuvant aluminum hydroxide (1:1) in a volume of 200 μl. Booster injections were performed after 7 days, 3, 4, 6, and 7 weeks of the initial challenge. The titers of anti-Aa OMP29^His antibodies were determined by immunoenzymatic assay (ELISA). Statistical analysis was performed in GraphPad Prism version 6.0 (GraphPad Software). This assay was approved by the Ethics and Research Committee on Animals of the Institute of Biomedical Sciences of the University of São Paulo under the protocol number 007/11.

The bacterial outer membrane extracts (OMEs) of AaD7S and its mutants AaΔ29, AaΔ29P, and AaΔ29Δ29P were obtained as previously described [28], and outer membrane proteins (OMPs) were resolved by standard SDS-PAGE (10% acrylamide) [29]. Purified Aa OMP29^His was also detected after SDS-PAGE using the Bolt Bis-Tris Plus gel (Life Technologies, Carlsbad, CA, United States) and the Bolt Mini Gel Tank (Life Technologies). The gels were stained with Colloidal blue (Life Technologies). OMP29 protein has the approximate size of 29 kDa, however, appears as a 34 kDa band when denatured [14].

After SDS-PAGE, the OMPs corresponding bands of AaD7S and its mutants AaΔ29, AaΔ29P, and AaΔ29Δ29P were excised, eluted with 1% acetic acid/H₂O, and analyzed in the mass spectrometer (LTQ Orbitrap Hybrid Mass Spectrometers), and raw data were processed using Proteome Discoverer software (Thermo Scientific). Twelve samples were analyzed. Individual proteins were identified and ranked by protein scores (calculated using SEQUEST algorithm on Proteome Discoverer), which is the sum of all peptide Xcorr values above the specified score threshold. The score threshold is calculated as follows: 0.8 + peptide charge × peptide relevance factor where peptide relevance factor is a parameter with a default value of 0.4. The peptide spectrum matches were used as a surrogate for the relative amounts of individual proteins in each sample.

Following electrophoresis, Aa OMP29^His was transferred to a nitrocellulose membrane using an iBlot 2 Dry Blotting System kit (Life Technologies). The membrane was blocked with 5% skimmed milk in TBS-T (20 mM Tris-HCl pH 7.6, 0.8% NaCl and 0.1% Tween 20). Anti-OMP29^His polyclonal serum obtained in Balb/c mice and serum of aggressive periodontitis patients colonized by Aa [15] were used as primary antibodies at 1:2,000 dilution in separate membranes. Secondary antibodies consisted of goat anti-mice IgG peroxidase conjugated (Amershan Bioscience, Amershan, Little Chalfont, United Kingdom) or goat anti-human IgG peroxidase conjugated (ThermoFisher, Carlsbad, CA, United States) diluted to 1:10,000. Aa OMP29^His bands were detected by using the SuperSignal West Pico Chemiluminescent Substrate (ThermoFisher, Carlsbad, CA, United States).

OBA-09 cells [30], immortalized GEC lineage, were grown at 37°C in humidified 5% CO₂ atmosphere in collagen type I pre-coated bottles (BD Bioscience, San Jose, CA, United States) containing complete Keratinocyte Serum-Free Medium (KSFM) (ThermoFisher, Carlsbad, CA, United States) supplemented with insulin, epidermal growth factor, and fibroblast growth factor (complete KSFM) according to the manufacturer's recommendation (ThermoFisher, Carlsbad, CA, United States). After confluent growth, OBA-09 cells were scraped, centrifuged at 593 × g for 10 min at 4°C, resuspended in complete KSFM, added to wells of plates at 1 × 10⁵ cell/ml, and incubated for 16 h.

The interaction assay was performed as previously described [11]. Log phase cultures of AaD7S and its mutants AaΔ29, AaΔ29P, and AaΔ29Δ29P were adjusted to 1 × 10⁸ cell/ml in a complete KSFM and added to OBA-09 cells at a multiplicity of infection of 1:1,000 (OBA-09:bacteria).

After 4 h of interaction, total RNA was isolated using the RNeasy Mini Kit (Qiagen; Valencia, CA, United States) and converted into cDNA using QuantiTect Reverse Transcription Kit (Qiagen).

Quantitative Real-Time Reverse Transcription-Polymerase Chain Reaction (qRT-PCR) analysis was carried out to evaluate cxcl-8 relative expression. The reaction was performed using the StepOnePlus Real-Time PCR System (Thermo Fisher Scientific, Rockford, United States) using the QuantiTect SYBR Green PCR and QuantiTect Primer Assay Cat # QT02108190 (Qiagen; Valencia, CA, United States).

The gene expression of inflammatory response was analyzed by Microarray using the Prime PCR Pathway Plate/Acute Inflammation Response H96 system (Bio-Rad; Hercules, CA, United States), according to the manufacturer's instructions and performed in CFX96 Touch Real-Time PCR Detection Systems (Bio-Rad).

Both Microarray and qRT-PCR data resulted in CT values for each sample. Gene dosages were obtained using the 2^−ΔΔCT Livak method [31, 32], considering glyceraldehyde-3-phosphate dehydrogenase (gapdh) as a reference gene and co-culture OBA-09 cell with AaD7S strain as the control. Statistical analysis was performed in GraphPad Prism version 6.0 (GraphPad Software).

The confluent monolayers of OBA-09 cell, as previously described, were added to 24 well/plates (ThermoFisher, Carlsbad, CA, United States) and challenged with E. coli serotype O111:B4 LPS (500 ng/ml; Sigma–Aldrich). Then, aliquots of 1 μg/ml and 10 μg/ml of Aa OMP29^His were added to each well. After 4 h of interaction, levels of CXCL-8 in free cell supernatant were established by Bio-Plex Pro Human Chemokine assay kit (Bio-Rad, United States), following the manufacturer's instructions. Those levels were determined by comparison to the standard curve (mean of fluorescence intensity vs. pg/ml), using Bio-Plex software manager 4.0. The results were shown in percentage with CXCL-8 concentration (pg/ml) of the 4 h cultured with OBA-09 being 100% and statistical analysis was performed in GraphPad Prism version 6.0 (GraphPad Software).

Multiple sequence alignment was performed with Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) using the default setting. The amino acid sequences of Aa OMP29 and OMP29^par of Aa strains HK1651, ANH9381, D7S-1, H5P-1, D11S-1, and D17P-2 (Table 1) were identified and downloaded from our annotated database at http://expression.washington.edu/genetable/script/gene_table_viewer [23].

The sequence alignments of OMP29 and OMP29^par of representative strains of serotype a, b, and c are provided in Supplementary Figure 1. The OMP29 is 346 aa long, and the OMP29^par is 356 aa long. Each orthologous protein demonstrates 99% amino acid sequence identity among strains. The OMP29 and OMP29par also share approximately 75% amino acid sequence identify.

The recombinant protein Aa OMP29^His was obtained, purified, and confirmed by its reaction with mice anti-Aa Omp29^His IgG polyclonal and by human aggressive periodontitis sera in a Western-blot assay (Supplementary Figure 2).

Aa mutants in omp29, omp29^par and in both genes were constructed and characterized. All strains presented similar growth rates in mTSB medium (Supplementary Figure 3). Analysis of OME revealed differences in the protein profile between the wildtype and the omp29 and omp29^par deletion mutants (Figure 1). To better assess the profile, the distinct bands were excised from the gel and evaluated by mass spectrophotometry. The identity and the relative amounts of individual proteins differentially expressed were determined (Supplementary Table 1).

Proteomic analysis indicated that OMP29 and OMP29^par were simultaneously expressed by the wildtype strain AaD7S, whereas the target proteins were absent in AaΔ29, AaΔ29P, and AaΔ29Δ29P. The deletion of omp29 led to overexpression of at least 3 times of both OMP29^par and OMP39. The deletion of OMP29^par resulted in a 2 times increase in OMP29 levels and 4 times increase in OMP39 levels (Figure 1). Furthermore, the double deletion in strain AaΔ29Δ29P resulted in the overexpression of OMP39 (11 times increase compared to the AaD7S). The relative expression levels of other proteins also changed in the mutants, although not as pronounced as seen for OMP39. The physiological significance of these changes is unknown but may represent a compensatory mechanism.

CXCL-8 plays a significant role in the pathogenesis of periodontitis. In this study, we aimed to confirm the effect of OMP29 and OMP29^par on the inhibition of cxcl-8 transcription. The effect of the 4 h interaction of Aa wildtype and its mutants with GECs on cxcl-8 transcription was evaluated by RT-qPCR. The AaΔ29Δ29P mutant promoted a significant increase in cxcl-8 transcripts levels in GECs compared to the wildtype. However, the deletion of omp29 or omp29^par did not result in cxcl-8 regulation, suggesting functional redundancy of OMP29 and OMP29^par in the regulation of cxcl-8 expression (Figure 2A).

To determine whether the effect on cxcl-8 expression was due to the compensatory expression of other proteins such as OMP39 in the mutants, we investigated the interaction between OBA-09 cell and Aa OMP29^His (Figure 2B). E. coli LPS was used to trigger CXCL-8 expression in OBA-09 cells [33, 34], and cells were challenged with 1 or 10 μg/ml of Aa OMP29^His. After 4 h of interaction, the expression of CXCL-8 was determined at the protein level. The analyses indicated that both concentrations of Aa OMP29^His could inhibit the transcription of cxcl-8, resulting in reduced CXCL-8 levels in OBA-9 cells supernatants compared to control cells (Figure 2B).

After observing the effect of OMP29 on CXCL-8 expression, we explored the gene expression profile of host acute inflammatory response to AaD7S and its mutants. The results showed modulatory effects (at least two-fold) of 21 genes (Table 2; for gene function, see Supplementary Table 2). Sixteen genes had increased (bcl2, birc3, casp3, c3, ep300, fas, fosb, grb2, il-1α, il-1β, il-6, cxcl-8, nr3c1, prkcq, socs3, and tnfrsf1β) and five had genes decreased (cd74, crp, faslg, tlr1, and vcam1) expression in OBA-09 cells co-cultured with AaΔ29Δ29P strain compared to AaD7S. The data suggested a potential role of OMP29 and its paralogue in the host response of GECs. Nevertheless, these findings suggested a possible influence of OMP29 and its paralogue in anti-apoptotic (e.g., bcl2, birc3, grb2, prkcq) and pro-apoptotic (e.g., casp3, c3, ep300, fosb, nr3c1, crp, and socs3) pathways. Furthermore, the interaction with the strain AaΔ29Δ29P down-regulated transcription of cd74, faslg, tlr1, and vcam1 and up-regulated fas and tnfrsf1β. The faslg and fas also participate in pro-apoptotic pathways. Finally, interaction with the strain AaΔ29Δ29P upregulated transcription of il-1α, il-1β, il-6, cxcl-8, and tnfrsf1β, encoding cytokines and chemokines involved in the inflammatory response. Notably, the effect promoted by the absence of OMP29 or OMP29^par was not as intense as observed for the double mutant, possibly due to the compensation mechanism induced by their absence (Table 2). Alternatively, the effects described above may be due to yet to be characterized pleiotropic effects of the deletion of omp29 and omp29^par.